Psi Chi Journal of Psychological Research by Psi Chi, the International Honor Society in Psychology

Psi Chi Journal of

Psychological Research SUMMER 2020 | VOLUME 25 | ISSUE 2

ISSN: 2325-7342 Published by Psi Chi, The International Honor Society in Psychology

Â®

® ®

PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH S U MM E R 2020 | VOLU ME 25, N U M BE R 2

EDITOR DEBI BRANNAN, PhD Western Oregon University Telephone: (503) 751-4200 E-mail: debi.brannan@psichi.org ASSOCIATE EDITORS ERIN AYALA, PhD St. Mary's University of Minnesota JENNIFER L. HUGHES, PhD Agnes Scott College TAMMY LOWERY ZACCHILLI, PhD Saint Leo University STEVEN V. ROUSE, PhD Pepperdine University ROBERT R. WRIGHT, PhD Brigham Young University-Idaho EDITOR EMERITUS MELANIE M. DOMENECH RODRIGUEZ, PhD Utah State University MANAGING EDITOR BRADLEY CANNON DESIGNER TAYLOR BROWN-STONE EDITORIAL ASSISTANT REBECCA STEMPEL ADVISORY EDITORIAL BOARD GLENA ANDREWS, PhD George Fox University AZENETT A. GARZA CABALLERO, PhD Weber State University MARTIN DOWNING, PhD NDRI ALLEN H. KENISTON, PhD University of Wisconsin–Eau Claire MARIANNE E. LLOYD, PhD Seton Hall University DONELLE C. POSEY, PhD Washington State University PAUL SMITH, PhD Alverno College

ABOUT PSI CHI Psi Chi is the International Honor Society in Psychology, founded in 1929. Its mission: "recognizing and promoting excellence in the science and application of psychology." Mem bership is open to undergraduates, graduate students, faculty, and alumni making the study of psychology one of their major interests and who meet Psi Chi’s minimum qualifications. Psi Chi is a member of the Association of College Honor Societies (ACHS), and is an affiliate of the American Psychological Association (APA) and the Association for Psychological Science (APS). Psi Chi’s sister honor society is Psi Beta, the national honor society in psychology for community and junior colleges. Psi Chi functions as a federation of chapters located at over 1,180 senior colleges and universities around the world. The Psi Chi Central Office is lo cat ed in Chatta nooga, Tennessee. A Board of Directors, composed of psychology faculty who are Psi Chi members and who are elected by the chapters, guides the affairs of the Organization and sets policy with the approval of the chapters. Psi Chi membership provides two major opportunities. The first of these is academic rec ognition to all inductees by the mere fact of membership. The second is the opportunity of each of the Society’s local chapters to nourish and stimulate the professional growth of all members through fellowship and activities designed to augment and enhance the regular curriculum. In addition, the Organization provides programs to help achieve these goals including con ven tions, research awards and grants competitions, and publication opportunities.

JOURNAL PURPOSE STATEMENT The twofold purpose of the Psi Chi Journal of Psychological Research is to foster and reward the scholarly efforts of Psi Chi members, whether students or faculty, as well as to provide them with a valuable learning experience. The articles published in the Journal represent the work of undergraduates, graduate students, and faculty; the Journal is dedicated to increas ing its scope and relevance by accepting and involving diverse people of varied racial, ethnic, gender identity, sexual orientation, religious, and social class backgrounds, among many others. To further support authors and enhance Journal visibility, articles are now available in the PsycINFO®, EBSCO®, Crossref®, and Google Scholar databases. In 2016, the Journal also became open access (i.e., free online to all readers and authors) to broad en the dissemination of research across the psychological science community. JOURNAL INFORMATION The Psi Chi Journal of Psychological Research (ISSN 2325-7342) is published quarterly in one volume per year by Psi Chi, Inc., The International Honor Society in Psychology. For more information, contact Psi Chi Central Office, Publication and Subscriptions, 651 East 4th Street, Suite 600, Chattanooga, TN 37403, (423) 756-2044. www.psichi.org; psichijournal@psichi.org. Statements of fact or opinion are the responsibility of the authors alone and do not imply an opinion on the part of the ofﬁcers or members of Psi Chi. dvertisements that appear in Psi Chi Journal do not represent endorsement by Psi Chi of the A advertiser or the product. Psi Chi neither endorses nor is responsible for the content of third-party promotions. Learn about advertising with Psi Chi at http://www.psichi.org/ Advertise COPYRIGHT

Permission must be obtained from Psi Chi to reprint or adapt a table or figure; to reprint quotations exceeding the limits of fair use from one source, and/or to reprint any portion of poetry, prose, or song lyrics. All persons wishing to utilize any of the above materials must write to the publisher to request nonexclusive world rights in all languages to use copyrighted material in the present article and in future print and nonprint editions. All persons wishing to utilize any of the above materials are responsible for obtaining proper permission from copyright owners and are liable for any and all licensing fees required. All persons wishing to utilize any of the above materials must include copies of all permissions and credit lines with the article submission.

SUMMER 2020 | VOLUME 25 | ISSUE 2

Does Life Satisfaction Mediate the Relationship Between Mood and Daydreaming Frequency?

151

Addictive Technology: Prevalence and Potential Implications of Problematic Social Media Use

Ryan F. Tudino, Nicole L. Mowry, and William A. Jellison* Quinnipiac University

Chloe Tanega and Andrew Downs* University of Portland

The Effect of Competitive, Cooperative, and Solo Play on Subjective Vitality and Negative Affect

162

Context Effects on Recognition Memory for Words

Abhilasha Vishwanath and Joshua Shive* Tennessee State University

Doug Maynard* , Leah J. Mancini, and Vania Rolón State University of New York at New Paltz

172

Predicting Student-Athlete Mental Health: Coach–Athlete Relationship

110

Disentangling the Effects of Study Time and Study Strategy on Undergraduate Test Performance

Megan Powers , Jana Fogaca , Regan A. R. Gurung* , and Callan M. Jackman Oregon State University

Zachary J. Cole and Darrell L. Butler* Ball State University

121

An Examination of the Influence of Serial Position on False Memory and Recognition

181

Do Hugs and Their Constituent Components Reduce Self-Reported Anxiety, Stress, and Negative Affect?

Preman Koshar and Megan L. Knowles* Franklin & Marshall College

Josephine Audiffred and Carissa L. Broadbridge* Saint Xavier University

192

Who Am I? Identity Development During the First Year of College

130

Puberty, Parents, and Depression: An EMA Study in Adolescent Girls

Madelynn D. Shell* , David Shears, and Zoe Millard The University of Virginia’s College at Wise

Danielle Apple and Stefanie Sequiera University of Pittsburgh

203

Planning to Practice: Action and Coping Plans Increase Days of Meditation Practiced

142

Relationships Between Self-Leadership, Psychological Symptoms, and Self-Related Thought in an Undergraduate Sample

Jonathan N. Cloughesy , Alissa J. Mrazek* , Michael D. Mrazek*, and Jonathan W. Schooler* University of California Santa Barbara

Sarah A. Myers , Carissa L. Philippi* Leah Reyna, and Gregory Dahl University of Missouri–St. Louis

, SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

*Faculty mentor

https://doi.org/10.24839/2325-7342.JN25.2.90

Does Life Satisfaction Mediate the Relationship Between Mood and Daydreaming Frequency? Ryan F. Tudino, Nicole L. Mowry, and William A. Jellison* Quinnipiac University

ABSTRACT. The current study tested to see if life satisfaction mediates the relationship between positive mood and daydreaming frequency. Two nonexperimental studies were conducted in which participants completed a questionnaire that assessed their positive mood, life satisfaction, and frequency of daydreaming. In Study 1, including 170 collegiate students, statistically significant results supported our correlational hypotheses between positive mood and life satisfaction (r = .67, p < .001), life satisfaction and daydreaming frequency (r = -.22, p = .002), and positive mood and daydreaming frequency (r = -.27, p < .001). However, the proposed mediational model, that life satisfaction mediated the relationship between positive mood and daydreaming frequency, was not supported because the negative relationship between positive mood and daydreaming frequency remained statistically significant when life satisfaction was added to the regression equation (ß = -.22, SE = .07; 95% CI [-.30, -.01]; p = .033). Study 2, including 102 adults, aimed to increase the reliability and generalizability of Study 1. Results from Study 2 also did not support the mediational role of life satisfaction. However, the results of Study 2 also demonstrated a negative relationship between positive mood and daydreaming frequency (r = -.35, p < .001), even when life satisfaction was included in the regression equation (ß = -.39, SE = .08; 95% CI [-.45, -.13]; p < .001). We discuss possible age and gender effects, and future directions of research. Keywords: positive mood, life satisfaction, daydreaming frequency

R SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

ecent research has highlighted the significance of daydreaming, a temporary separation from reality that focuses on an individual’s thoughts (e.g., Giambra & Traynor, 1978; Golding & Singer, 1983; Mar, Mason, & Litvak, 2012). Killingsworth and Gilbert (2010) found that people’s minds wander 46.9% of the time. However, the potential relationship between daydreaming frequency, life satisfaction, and mood has received little attention in the literature. Although some researchers have suggested that daydreaming precedes a negative affect (Killingsworth & Gilbert, 2010), others have demonstrated that one’s mood leads the mind to wander (Smallwood, Fitzgerald, Miles, & Phillips, 2009). The current study further explored the relationship between positive mood

and daydreaming frequency by assessing whether life satisfaction may serve as a mediator. Although daydreaming and mind-wandering may be considered two different constructs, the current literature has often used these terms inter changeably. For instance, theorists have defined daydreaming in three predominant ways—as fanciful desired wishes, as thinking unrelated to an ongoing activity (mind-wandering), or as unintended mental content that comes to mind effortlessly (Klinger, 2009). Rather than defining daydreaming as a unitary construct that either includes or excludes mind-wandering, Klinger (2009) suggested that daydreaming be defined as nonworking thought that is spontaneous or fanciful. In this way, daydreaming includes mind-wandering.

*Faculty mentor

Tudino, Mowry, and Jellison | Mood, Life Satisfaction, and Daydreaming Frequency

This appears to be the rule of thumb, as several studies such as Killingsworth and Gilbert (2010) use the terms daydreaming and mind-wandering conversely. Therefore, for the purpose of this article, the two constructs may be used interchange ably as well. The relationship between daydreaming fre quency, life satisfaction, and mood may be seen in the rationale for why daydreams occur. Although some researchers have found daydreaming to be an effective way for people to think about and solve situations that concern them in everyday life (Giambra & Traynor, 1978), others have high lighted the negative impact that daydreams may have on an individual. For instance, Singer (1975) described how daydreaming may be character ized by obsessive negative ruminations and the inability to concentrate on an external task. This description of daydreaming can then be translated to people’s emotions (Marchetti, Van de Putte, & Koster, 2014). Furthermore, Stawarczyk, Majerus, and D’Argembeau (2013) found that negative affect was associated with higher daydreaming frequency when participants were concerned about giving a speech in the near future, compared to a control group in which no speech had to be given. Therefore, daydreaming may provide a way to ruminate about past or future events that cause sig nificant distress and negative affect in individuals. Another plausible explanation is that day dreaming may act as an escape for people with problems concerning them in everyday life (Singer, 1975). For example, one form of daydreaming, maladaptive daydreaming, is characterized by absorption into fantasy that involves far-fetched dreams such as having relationships with celebri ties and having an idealized version of self (e.g., Abu-Rayya, Somer, & Meari-Amir, 2019; Bigelsen, Lehrfeld, Jopp, & Somer, 2016). This ability to daydream about exotic situations has been found to be involved in a negative reinforcement loop, in which emotionally distressed individuals alleviate their anguish by participating in compensatory fantasies (Somer, Somer, & Jopp, 2016). Therefore, an individual’s current daily affect and life situation may influence the frequency with which a person engages in daydreaming. Hence, the present study explored whether life satisfaction mediates, or explains, the relationship between positive mood and daydreaming frequency. Although mood and life satisfaction are similar constructs, there is a key difference between the two. For the purpose of this article, mood is

operationally defined as one’s general daily affect. That is, one’s emotion in a given moment. This contrasts with life satisfaction, which is defined as the subjective well-being associated with the assessment of one’s life. People may be satisfied with their lives, but unhappy in a given moment or situation. For instance, successful students may be satisfied with their overall ability to graduate college and find a job but may be unhappy because of an argument they had with a friend earlier in the day. Conversely, other students may be dissatisfied with their lives for not being able to graduate on time but may be happy because they received a monetary loan for another semester. Thus, given that these two constructs are distinct from one another, both mood and life satisfaction may be related to daydreaming in different ways. Enhancing one’s mood, or general daily affect, may decrease one’s frequency of daydreams. Klinger, Murphy, Ostrem, and Stark-Wroblewski (2005) found that individuals with high positive emotionality disclosed fewer daydreams and more real-life experiences. Different studies have found that the more depressive, or negative, mood an individual experienced, the more daydreams they had (e.g., Giambra & Traynor, 1978; Golding & Singer, 1983). Therefore, these findings suggest that the more individuals are concerned and anxious about their current situation, the more daydreams they will experience. Together, these studies suggest that mood is negatively correlated with daydreaming frequency, and, hence, by enhancing an individual’s mood, one can decrease the number of daydreams. Additional research has suggested that an individual’s life satisfaction may also decrease the person’s frequency of daydreams (Mar et al., 2012). Brannigan, Schaller, and McGarva (1993) determined that high life satisfaction, in regard to one’s current status of “fitting in” with others, was negatively correlated with the frequency of daydreams. Furthermore, Killingsworth and Gilbert (2010) proposed that mind-wandering reflects one’s happiness in surrounding situations, suggesting that life satisfaction is not only deter mined by the acceptance of others, but also with one’s physical surroundings. Additionally, Ruby, Smallwood, Engen, and Singer (2013) proposed that daydreaming happens because unfulfilled goals are more important to individuals than their current surroundings. Therefore, these findings suggest that the more individuals are concerned

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

Mood, Life Satisfaction, and Daydreaming Frequency | Tudino, Mowry, and Jellison

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

and anxious about life goals, the more daydreams they will experience. Taken together, these stud ies suggest that, as life satisfaction increases, daydreaming decreases. In addition to finding a relationship between life satisfaction and daydreaming frequency, several studies have found a correlation between mood and life satisfaction. Golding and Singer (1983) found that individuals who had a positive outlook had an increased satisfaction with life. Additionally, other studies have found that individuals who had a positive mood about themselves had greater life satisfaction (e.g., Flynn & MacLeod, 2015; Mar et al., 2012). Furthermore, Greenwald and Harder (1997) found a negative correlation between an individual’s self-derogation and mood such that, as mood decreased, self-derogation increased. These studies suggest that mood not only influences life satisfaction, but also influences an individual’s self-image. However, the findings of Giambra and Traynor (1978) suggest that mood and daydream ing frequency are negatively related. Moreover, Golding and Singer (1983), taken together with Brannigan and colleagues (1993), suggest that life satisfaction may impact the frequency of daydreams. Therefore, because mood tends to influence life satisfaction, an individual’s satisfac tion with life may mediate the negative relationship between mood and daydreaming frequency. In the current nonexperimental study, we tested the following hypotheses. First, we expected that mood would be positively correlated with life satisfaction such that, as positive mood increased, life satisfaction would increase. Second, it was expected that mood would negatively correlate with daydreaming frequency such that, as positive mood increased, daydreaming frequency would decrease. Third, we expected that life satisfaction would negatively correlate with daydreaming frequency such that, as life satisfaction increased, daydreaming frequency would decrease. Finally, we also anticipated that life satisfaction would mediate the relationship between mood and daydreaming frequency. To test our hypotheses, participants completed the Positive and Negative Affect Schedule (PANAS), which measures overall positive mood by subtracting the negative affect score mean from the positive affect score mean, the Satisfaction With Life Scale (SWLS), which measures life satisfaction, and a shortened version of the Imaginal Processes Inventory (IPI), which measures daydreaming frequency.

Study 1 Method Participants. One hundred seventy college students participated in this study.1 Participants were selected through cluster sampling of students enrolled in an introductory psychology class.2 Although par ticipation was voluntary, participants received class credit for participating. Alternatives to research participation for credit (e.g., completing an article summary) were also available. Procedure. Approval was gained from the institutional review board, and participants were selected through the cluster sample. Participants gave informed consent before beginning the study. Next, each participant completed a series of measures (see Measures section below)3, which included filler and distractor items from the Rosenberg Self-Esteem Scale (Rosenberg, 1965; 10 items) and Locus of Control Scale (Rotter, 1966; 7 items) that were not used for these analyses. Each participant took roughly 15–20 minutes to complete the survey either in a private workspace in the psychology research lab (n = 45) or online (n = 125) during the middle of the academic semester. After completing the survey, participants were given more information about the study and thanked for their participation. Measures. Positive mood. An individual’s general daily affect was assessed using the PANAS (Watson, Clark, & Tellegen, 1988), with a positive affect section (α = .85; 10 items; e.g., “Interested” and “Enthusiastic”) and a negative affect section (α = .86; 10 items; e.g., “Scared” and “Distressed”). Participants responded to each item, or word in this instance, on a Likert-type scale ranging from 1 (very slightly) to 5 (extremely). Items were scored such that, on the positive affect scale, greater positive values reflect more positive mood, whereas on the negative affect scale, greater positive values reflect Data were collected over two separate academic semesters (Time 1: 45 participants; Time 2: 125 participants). The two samples did not significantly differ in their level of positive mood, F(1, 168) = 0.73, p = .39, d = .14, life satisfaction, F(1, 168) = 1.47, p = .23, d = .21, or daydreaming frequency, F(1, 168) = 0.03, p = .85, d = .03. 2 Although demographic data were not collected on this sample, the population from which the sample was drawn consisted of predominantly traditional first- and second-year students (age 18–19), European American, and majority women (70%). The total number of students enrolled in PS101 courses in which the samples were drawn were 555 at Time 1 and 465 at Time 2. 3 Measures for all participants were completed in the following order: (a) daydreaming frequency, (b) locus of control filler, (c) satisfaction with life, (d) self-worth filler, and (e) mood. 1

Tudino, Mowry, and Jellison | Mood, Life Satisfaction, and Daydreaming Frequency

more negative mood. The total PANAS score was calculated by subtracting the mean negative score from the mean positive score, such that greater positive values reflect more positive, compared to negative, mood. Life satisfaction. How satisfied participants were with their life was assessed using the SWLS (Diener, Emmons, Larsen, & Griffin, 1985; α = .82; 5 items; e.g., “If I could live my life over, I would change almost nothing.”). Participants responded to each item on a Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). Items were scored such that greater positive values reflected greater satisfaction with life. Daydreaming frequency. The amount an indi vidual daydreams was assessed using a shortened version of the IPI (Singer & Antrobus, 1970; α = .92; 12 items; e.g., “I daydream…”). Participants responded to each item on a scale from 1 (never or rarely) to 5 (always or most of the time). Items were scored such that greater positive values reflected a greater frequency of daydreams. Results Correlations. A series of bivariate correlations were conducted to explore the relationship between mood, life satisfaction, and daydreaming frequency. Descriptive statistics and all bivariate correlations are displayed in Table 1. As displayed in Table 1, a positive correlation was found between mood and life satisfaction such that, as positive mood increased, life satisfaction also increased. Additionally, a negative correla tion was found between mood and daydreaming frequency such that, as positive mood increased, daydreaming frequency decreased. Similarly, a negative correlation was found between life satis faction and daydreaming frequency such that, as life satisfaction increased, daydreaming frequency decreased. Thus, our first three hypotheses were supported. Multiple regression. A multiple regression was conducted with mood as the antecedent vari able, life satisfaction as the mediator variable, and daydreaming frequency as the outcome variable. As displayed in Figure 1, once an individual’s satisfaction with life was included in the regression equation, the relationship between positive mood and daydreaming frequency remained statistically significant. However, the relationship between life satisfaction and daydreaming frequency was no longer statistically significant. Thus, the proposed mediational model with life satisfaction as the mediator variable was not supported.

Discussion Results of Study 1 failed to fully support the proposed mediational model that life satisfac tion mediates the relationship between positive mood and daydreaming frequency. Nevertheless, our first three hypotheses were supported, which demonstrated clear links between mood, life satisfaction, and daydreaming frequency (observed power = .98; Faul, Erdfelder, Buchner, & Lang, 2009; Faul, Erdfelder, Lang, & Buchner, 2007). However, emotional intelligence, defined as the capability to regulate the emotions of oneself and others and to distinguish one’s own emotions from others in order to think and act, is greater in older adults compared to younger adults (Chen, Peng, & Fang, 2016). This increase was especially present in older adults for the components of understanding and regulating emotions, a key aspect needed to reliably report one’s own emotions (Tsaousis & Kazi, 2013). Given that the population of participants included in Study 1 were young adults aged 18–19, Study 2 aimed to generalize the findings outside of the collegiate student population in order to obtain a more reliable and representative sample. TABLE 1 Descriptive Statistics and Zero Order Correlations Between Positive Mood, Life Satisfaction, and Daydreaming Frequency Among College Students for Study 1 Measures

1. Positive Mood

1.44

1.08

2. Life Satisfaction

5.04

1.10

.67**

3. Daydreaming Frequency

2.73

0.78

-.27**

-.22**

Note. N = 170. Larger positive values on all scales reflect stronger or more positive endorsements of those constructs. ** p < .01.

FIGURE 1 β = -.22* Daydreaming Frequency

Positive Mood

β = .67**

β = -.08 Life Satisfaction

Figure 1. Life satisfaction does not mediate the relationship between positive mood and daydreaming frequency among college students. N = 170. Larger positive values on all scales reflect stronger or more positive endorsements of those constructs. * p < .05. **p < .01.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

Mood, Life Satisfaction, and Daydreaming Frequency | Tudino, Mowry, and Jellison

Study 2 Method Participants. One hundred two participants (Mage = 36.0, SD = 12.3)4 were recruited through Amazon Mechanical Turk (MTurk). MTurk samples are a reliable and demographically diverse sample of the population (Buhrmester, Kwang, & Gosling, 2011; Huff & Tingley, 2015; Kees, Berry, Burton, & Sheehan, 2017). Participants consisted of 37 women and 63 men (two participants did not report their gender); racial/ethnic breakdown consisted of 82 European American, 7 African American, 5 Asian, 3 multiracial, and 5 not reporting their race or ethnicity. Procedure. Approval for the revised follow-up study was gained from the institutional review board. Participants completed the survey online using SurveyMonkey as in Study 1. The measures (see below) were identical to those used in Study 1. Measures. Positive mood. The total PANAS (Watson et al., 1988) score was calculated by subtracting the mean negative score (α = .92) from the mean positive score (α = .95). Therefore, greater positive values reflect more positive, compared to negative, mood. Life satisfaction. The SWLS (Diener et al., 1985; α= .92) was used. Items were scored such that higher positive values reflected greater satisfaction with life. Daydreaming frequency. The IPI (Singer & Antrobus, 1970; α = .95) was used. Items were scored such that greater positive values reflected a greater frequency of daydreams. Results Correlations. A series of bivariate correlations were conducted to explore the relationship between mood, life satisfaction, and daydreaming frequency. Descriptive statistics and all bivariate correlations are displayed in Table 2. As displayed in Table 2, a statistically sig nificant positive correlation was found between mood and life satisfaction such that, as positive mood increased, life satisfaction also increased. Additionally, a statistically significant negative cor relation was found between mood and daydreaming frequency such that, as positive mood increased, daydreaming frequency decreased. However, no sta tistically significant correlation was found between life satisfaction and daydreaming frequency. SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

M age and SD were calculated based on the available data through Amazon's Mechanical Turk. Participant ages included those in their 20s (27%), 30s (41%), 40s (16%), 50s (8%), 60s (5%) and 70s (2%). 4

Multiple regression. A multiple regression was conducted with positive mood as the antecedent variable, life satisfaction as the mediator variable, and daydreaming frequency as the outcome vari able. As displayed in Figure 2, when life satisfaction was included in the regression equation, the relationship between positive mood and daydream ing frequency remained statistically significant. In addition, because the correlation between life satisfaction and daydreaming frequency was not statistically significant, the proposed mediational model with life satisfaction as the mediator variable was not supported. Discussion The results of Study 2 did not support the proposed mediational model or replicate the results found in Study 1, given that life satisfaction was not cor related with daydreaming frequency. However, positive mood was positively correlated with life satisfaction. In addition, positive mood was negatively correlated with daydreaming frequency. These results suggest that the link between life satisfaction and daydreaming frequency may not be as strong for a general sample as with college students (observed power = .97; Faul et al., 2009; Faul et al., 2007).

General Discussion Based on prior research regarding the individual relationships between positive mood, life satisfac tion, and daydreaming frequency (e.g., Brannigan, Schaller, & McGarva, 1993; Flynn & MacLeod, 2015; Giambra & Traynor, 1978; Golding & Singer, 1983; Klinger et al., 2005; Mar et al., 2012), we hypothesized that there would be a positive correla tion between positive mood and life satisfaction, a negative correlation between positive mood and daydreaming frequency, and a negative cor relation between life satisfaction and daydreaming frequency. In addition, we hypothesized that life satisfaction would mediate the relationship between positive mood and daydreaming frequency. The two studies conducted did not support this mediational model. The results of Study 1 supported our cor relational hypotheses but did not support life satisfaction as the mediator variable. Although all three pairs of variables were significantly correlated, positive mood remained statistically significant when life satisfaction was included in the regression equation. Given that post-collegiate adults score higher on emotional intelligence scales

Tudino, Mowry, and Jellison | Mood, Life Satisfaction, and Daydreaming Frequency

for understanding and regulating (Tsaousis & Kazi, 2013), Study 2, then, aimed to test the hypothesized model using a more representative sample. The results of Study 2 only partially supported our correlational hypotheses and did not support life satisfaction as the mediator variable. Specifically, Study 2 found a positive correlation between posi tive mood and life satisfaction as well as a negative correlation between positive mood and daydream ing frequency. However, no statistically significant relationship was found between life satisfaction and daydreaming frequency. Additionally, when life satisfaction was included in the regression equation, the relationship between positive mood and daydreaming frequency remained statistically significant. Taken together, these two studies suggest that life satisfaction does not mediate the relationship between positive mood and daydream ing frequency, and that the relationship between life satisfaction and daydreaming frequency may vary across samples. The pattern of results of Study 1 and Study 2, although not definitive, may suggest an alternative purpose to daydreaming for collegiate students and postcollegiate adults. Although the media tional model remained unsupported for both a collegiate and general sample, there was a slight positive trend in the relationship for postcol legiate adults in Study 2 once life satisfaction was added to the regression equation. This contrasts with the slight negative trend in the relationship for collegiate students in Study 1. As a result, this may suggest that the association between life satisfaction and daydreaming frequency may be affected by cohort or age. Specifically, this may indicate that collegiate students daydream in order to ruminate about a past or future event (Stawarczyk et al., 2013) or escape from the present situation (Bigelsen et al., 2016) and that postcollegiate adults daydream to proactively solve problems and plan for the future (e.g., Giambra & Traynor, 1978; Ruby et al., 2013). The first explanation for collegiate students is consistent with the guilty-dysphoric daydream ing theory whereas the second explanation for postcollegiate adults is consistent with the positive constructive daydreaming theory, both proposed by Singer (1975). The latter of these two theories takes on a more positive perspective, whereas the first theory is from a more negative perspective. Thus, the relationship between life satisfaction and daydreaming frequency may become more positive as individuals age.

Although the current studies focused on daydreaming as a method to escape from negative scenarios in one’s life or to ruminate about them, the results of Study 2 may indicate a change in the use of daydreaming as individuals age. In fact, several studies have found that postcollegiate adults show a preference toward positive information compared to negative information. This may be due to greater emotional regulation in postcol legiate adults that utilize cognitive mechanisms that enhance positive information and decrease negative information (e.g., Mather & Carstensen, 2005; Reed & Chan, 2014). Thus, generalizing outside of the collegiate population may introduce the positivity bias of postcollegiate adults into the mediational model. Moreover, the potential for age effects is fur ther supported by Zhiyan and Singer (1997), who found that positive constructive daydreaming was positively correlated with the Big Five personality dimension of openness to experience, guilty-dys phoric daydreaming was positively associated with neuroticism and negative emotionality, and poor attentional-control-associated daydreaming was negatively correlated with conscientiousness and TABLE 2 Descriptive Statistics and Zero Order Correlations Between Positive Mood, Life Satisfaction, and Daydreaming Frequency Among General Sample of Adults for Study 2 Measures

1. Positive Mood

1.24

1.15

2. Life Satisfaction

4.01

1.63

.46**

3. Daydreaming Frequency

2.61

0.87

-.35**

-.09**

Note. N = 170. Larger positive values on all scales reflect stronger or more positive endorsements of those constructs. ** p < .01.

FIGURE 2 β = -.39** Daydreaming Frequency

Positive Mood

β = .46**

β = .09 Life Satisfaction

Figure 2. Life satisfaction does not mediate the relationship between positive mood and daydreaming frequency among general sample of adults. N = 102. Larger positive values on all scales reflect stronger or more positive endorsements of those constructs. ** p < .01.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

Mood, Life Satisfaction, and Daydreaming Frequency | Tudino, Mowry, and Jellison

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

positive emotionality. Other researchers have found that conscientiousness and openness to experience generally increases with age and that neuroticism decreases (e.g., Roberts, Walton, & Viechtbauer, 2006; Soto, John, Gosling, & Potter, 2011). Thus, it is possible that the postcollegiate adults, who were included in the participant pool for Study 2, may alter the mediational model due to their increase in positive constructive daydreaming rather than guilty-dysphoric daydreaming. Future studies may want to control for age in order to determine whether these age effects exist. Although the results found no support for the proposed mediational model, several limita tions in the study should be discussed. First, the operational definitions provided in the literature regarding daydreaming and mind-wandering are vague. It is possible that these related terms are actually two different constructs that are difficult to measure separately. Additionally, some research has suggested that mind-wandering may not be a unitary construct, but rather a construct comprised of intentional and unintentional daydreaming (Seli, Risko, & Smilek, 2016). Therefore, the two aspects of the construct may have different effects. Intentional daydreaming may be used when attempting to problem solve (positive constructive daydreaming) and unintentional daydreaming may be used to escape a situation or ruminate (guiltydysphoric daydreaming). Given that the measure used in this study to assess daydreaming frequency did not distinguish between intentional and unin tentional daydreaming, future studies may want to explore the qualitative difference between the two. Another limitation of the present study involves the study design. Mood was only assessed at one instance. Future studies may want to explore other ways to induce daydreaming, or mind-wandering, so that mood could be assessed before and after the daydreaming episode. If one’s general daily affect is negative after a daydreaming episode, this may suggest that the individual was ruminating about a negative stressor. On the other hand, if the daily affect of the individual is more positive or unchanged after the daydreaming episode, this may suggest that the daydreaming episode was used to proactively solve problems (Killingsworth & Gilbert, 2010). Second, given the direct relationship between positive mood and daydreaming frequency in both studies, it is possible that positive mood may mediate the relationship between life satisfaction and daydreaming frequency. Future studies may want to explore whether this new mediational model is better supported. Moreover, the present

studies also utilized two vastly different samples. The first study included individuals from a predomi nantly female participant pool (~70%) who were college-aged, whereas the second study included a majority of male participants (~63%) who were adults. Future studies may want to further explore how age and gender influence the effects of mood and life satisfaction on daydreaming frequency. For example, future research should explore whether gender may moderate the relationships among our variables. Recent research has found a relationship between gender and daydreaming frequency (e.g., Golding & Singer, 1983; Mar et al., 2012) and even gender and life satisfaction (e.g., Mar et al., 2012; Matud, Bethencourt, & Ibáñez, 2014). For instance, Mar et al. (2012) found that men tended to daydream about the past more than women did. Additionally, Matud et al. (2014) found that social support was positively correlated with life satisfaction among women and self-esteem was positively correlated with life satisfaction among men. Future studies should explore whether these variables play a role in explaining daydreaming frequency differently for men and women. Although the results of these studies are not definitive, given the strong association between positive mood and daydreaming frequency found in both studies and the trend in life satisfaction, several implications regarding the phenomenon of daydreaming frequency have the potential to arise. For instance, if certain types of daydreaming (inten tional or unintentional) are related to positive and negative outcomes, psychiatrists or therapists may be able to use daydreaming frequency as an additional method to aid in the treatment of mental disorders such as depression. For example, collegiate students may be able to alter their daydreaming style from guilty-dysphoric to positive and constructive. In fact, Ernst, Blanc, De Seze, and Manning (2015) found that induction of positive mental images increased autobiographical memory and episodic future thinking, factors that are dampened in individuals who are depressed or experience negative mood. Similarly, Linke and Wessa (2017) found that posi tive mental imagery training was positively associated with reward sensitivity and behavioral activation. Therefore, if the trends in these studies are found to be valid, clinicians may be able to implement a form of daydreaming manipulation in order to help patients to realistically strive and achieve a satisfac tory life. These hypotheses need to be examined further, but if they hold true, then this study might provide clinicians with new treatment options to better enhance the quality of life of their patients.

Tudino, Mowry, and Jellison | Mood, Life Satisfaction, and Daydreaming Frequency

References Abu-Rayya, H. M., Somer, E., & Meari-Amir, S. (2019). The psychometric properties of the Arab 16-item maladaptive daydreaming scale (MDS-16AR) in a multicountry Arab sample. Psychology of Consciousness: Theory. Research, and Practice, 6, 171–183. http://dx.doi.org/10.1037/cns0000183 Bigelsen, J., Lehrfeld, J. M., Jopp, D. S., & Somer, E. (2016). Maladaptive daydreaming: Evidence for an under-researched mental health disorder. Consciousness and Cognition, 42, 254–266. http://dx.doi.org/10.1016/j.concog.2016.03.017 Brannigan, G. G., Schaller, J. A., & McGarva, A. (1993). Approval motivation and sexual daydreaming. The Journal of Genetic Psychology, 154, 383–387. http://dx.doi.org/10.1080/00221325.1993.10532191 Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6, 3–5. http://dx.doi.org/10.1177/1745691610393980 Chen, Y., Peng, Y., & Fang, P. (2016). Emotional intelligence mediates the relationship between age and subjective well-being. The International Journal of Aging and Human Development, 83, 91–107. http://dx.doi.org/10.1177/0091415016648705 Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49, 71–75. http://dx.doi.org/10.1207/s15327752jpa4901_13 Ernst, A., Blanc, F., De Seze, J., & Manning, L. (2015). Using mental visual imagery to improve autobiographical memory and episodic future thinking in relapsing-remitting multiple sclerosis patients: A randomised-controlled trial study. Restorative Neurology and Neuroscience, 33, 621–638. http://dx.doi.org/10.3233/RNN-140461 Faul, F., Erdfelder, E., Buchner, A., & Lang, A-G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149–1160. https://doi.org/10.3758/brm.41.4.1149 Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. https://doi.org/10.3758/bf03193146 Flynn, D. M., & MacLeod, S. (2015). Determinants of happiness in undergraduate university students. College Student Journal, 49, 452–460. Giambra, L. M., & Traynor, T. D. (1978). Depression and daydreaming: An analysis based on self-ratings. Journal of Clinical Psychology, 34, 14–25. Golding, J. M., & Singer, J. L. (1983). Patterns of inner experience: Daydreaming styles, depressive moods, and sex roles. Journal of Personality and Social Psychology, 45, 663–675. http://dx.doi.org/10.1037/0022-3514.45.3.663 Greenwald, D. F., & Harder, D. W. (1997). Fantasies, coping behavior, and psychopathology. Journal of Clinical Psychology, 53, 91–97. http://dx.doi.org/10.1002/(SICI)1097-4679(199702)53:2<91::AID-JCLP1>3.0.CO;2-X Huff, C., & Tingley, D. (2015). ‘Who are these people?’ Evaluating the demographic characteristics and political preferences of MTurk survey respondents. Research and Politics, 2, 1–12. https://doi.org/10.1177/2053168015604648 Kees, J., Berry, C., Burton, S., & Sheehan, K. (2017). An analysis of data quality: Professional panels, student subject pools, and Amazon’s Mechanical Turk. Journal of Advertising, 46, 141–155. https://doi.org/10.1080/00913367.2016.1269304 Killingsworth, M. A., & Gilbert, D. T. (2010). A wandering mind is an unhappy mind. Science, 330, 932. https://doi.org/10.1126/science.1192439 Klinger, E. (2009). Daydreaming and fantasizing: Thought flow and motivation. In K. D. Markman, W. M. P. Klein, & J. A. Suhr (Eds.), Handbook of imagination and mental simulation (pp. 225–239). New York, NY: Psychology Press. Klinger, E., Murphy, M. D., Ostrem, J. L., & Stark-Wroblewski, K. (2005). Disclosing daydreams versus real experience: Attitudes, emotional reactions, and personality correlates. Imagination, Cognition and Personality, 24, 101–138. http://dx.doi.org/10.2190/FTRA-31CH-6A2W-HV3N Linke, J., & Wessa, M. (2017). Mental imagery training increases wanting of rewards and reward sensitivity and reduces depressive symptoms. Behavior Therapy, 48, 695–706. http://dx.doi.org/10.1016/j.beth.2017.04.002 Mar, R. A., Mason, M. F., & Litvack, A. (2012). How daydreaming relates to life satisfaction, loneliness, and social support: The importance of gender and daydream content. Consciousness and Cognition: An International Journal, 21, 401–407. http://dx.doi.org/10.1016/j.concog.2011.08.001 Marchetti, I., Van de Putte, E., & Koster, E. H. W. (2014). Self-generated thoughts and depression: From daydreaming to depressive symptoms. Frontiers in Human Neuroscience, 8, 1–10. https://doi.org/10.3389/fnhum.2014.00131

Mather, M., & Carstensen, L. L. (2005). Aging and motivated cognition: The positivity effect in attention and memory. Trends in Cognitive Sciences, 9, 496–502. http://dx.doi.org/10.1016/j.tics.2005.08.005 Matud, P. M., Bethencourt, J. M., & Ibáñez, I. (2014). Relevance of gender roles in life satisfaction in adult people. Personality and Individual Differences, 70, 206–211. http://dx.doi.org/10.1016/j.paid.2014.06.046 Reed, A. E., & Chan, L. (2014). Meta-analysis of the age-related positivity effect: Age differences in preferences for positive over negative information. Psychology and Aging, 29, 1–15. http://dx.doi.org/10.1037/a0035194 Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132, 1–25. http://dx.doi.org/10.1037/0033-2909.132.1.1 Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rotter, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs, 80, 1–28. http://dx.doi.org/10.1037/h0092976 Ruby, F. J. M., Smallwood, J., Engen, H., & Singer, T. (2013). How self-generated thought shapes mood: The relation between mind wandering and mood depends on the socio-temporal content of thoughts. PLOS ONE, 8, 1–8. http://dx.doi.org/10.1371/journal.pone.0077554 Seli, P., Risko, E. F., & Smilek, D. (2016). On the necessity of distinguishing between unintentional and intentional mind wandering. Psychological Science, 27, 685–691. https://doi.org/10.1177/0956797616634068 Singer, J. L. (1975). Navigating the stream of consciousness: Research in daydreaming and related inner experience. American Psychologist, 30, 727–738. http://dx.doi.org/10.1037/h0076928 Singer, J. L., & Antrobus, J. S. (1970). Imaginal Process Inventory. Center for Research in Cognition and Affect. Graduate Center, City University of New York: New York City, NY. Retrieved from http://www.slideshare.net/Solercanto/ipi-questionnaire Smallwood, J., Fitzgerald, A., Miles, L. K., & Phillips, L. H. (2009). Shifting moods, wandering minds: Negative moods lead the mind to wander. Emotion, 9, 271–276. http://dx.doi.org/10.1037/a0014855 Somer, E., Somer, L., Jopp, D. S. (2016). Parallel lives: A phenomenological study of the lived experience of maladaptive daydreaming. Journal of Trauma and Dissociation, 17, 561–576. http://dx.doi.org/10.1080/15299732.2016.1160463 Soto, C. J., John, O. P., Gosling, S. D., & Potter, J. (2011). Age differences in personality traits from 10 to 65: Big five domains and facets in a large cross-sectional sample. Journal of Personality and Social Psychology, 100, 330–348. http://dx.doi.org/10.1037/a0021717 Stawarczyk, D., Majerus, S., & D’Argembeau, A. (2013). Concern-induced negative affect is associated with the occurrence and content of mind wandering. Consciousness and Cognition, 22, 442–448. http://dx.doi.org/10.1016/j.concog.2013.01.012 Tsaousis, I., & Kazi, S. (2013). Factorial invariance and latent mean differences of scores on trait emotional intelligence across gender and age. Personality and Individual Differences, 54, 169–183. http://dx.doi.org/10.1016/j.paid.2012.08.016 2 Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. http://dx.doi.org/10.1037/0022-3514.54.6.1063 Zhiyan, T., & Singer, J. L. (1997). Daydreaming styles, emotionality, and the big five personality dimensions. Imagination, Cognition and Personality, 16, 399–414. http://dx.doi.org/10.2190/ATEH-96EV-EXYX-2ADB Author Note. Ryan F. Tudino, Quinnipiac University; Nicole Mowry, Quinnipiac University; and William A. Jellison, Quinnipiac University. This research was supported in part by a student research and experiential learning grant from the College of Arts and Sciences from Quinnipiac University. Portions of this work were presented at the 2017 New England Psychological Association regional convention. Special thanks to Psi Chi Journal reviewers for their support. Correspondence concerning this article should be addressed to William A. Jellison, Center for Psychological Science, Quinnipiac University, 275 Mount Carmel Ave., Hamden, CT, 06518. E-mail: William.Jellison@quinnipiac.edu

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

https://doi.org/10.24839/2325-7342.JN25.2.98

The Effect of Competitive, Cooperative, and Solo Play on Subjective Vitality and Negative Affect Doug Maynard* , Leah J. Mancini, and Vania Rolón State University of New York at New Paltz

ABSTRACT. In the current investigation, we examined the impact of game mode upon changes in subjective vitality, an indicator of well-being marked by feelings of aliveness and energy, as well as negative affect. In Study 1, college students (N = 106) in pairs were randomly assigned to play a dexteritybased card game in either a competitive or a cooperative mode. In Study 2, 54 college students played the same game alone in a solo mode. Participants in all 3 conditions experienced a significant increase in subjective vitality (d = .73, .59, and .47 for the competitive, cooperative, and solo conditions, respectively) and a significant decrease in negative affect (d = .59, .26, and .34, respectively) after playing the game. In Study 1, there was no significant difference in changes to emotional state between the competitive and cooperative modes of play. Finally, contrary to expectations, more competitive participants did not benefit more than less competitive participants from playing in the competitive mode.

Open Data badge earned for transparent research practices. Data available at https://osf.io/xh3es/

Keywords: play, subjective vitality, negative affect, competitive play, cooperative play, solo play

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

ith the advent of positive psychology at the turn of the 21st century, greater attention has been paid to the different ways in which adults might flourish and experience greater well-being. Researchers in the past two decades have demonstrated the benefits of engaging in a variety of positive intra- and interpersonal activities such as generosity (e.g., Dunn, Aknin & Norton, 2008), gratitude (e.g., Emmons & McCullough, 2003), and mindfulness (e.g., Brown & Ryan, 2003). It is somewhat surprising, then, that play in adults has been understudied as a possibly important facet of a life well-lived. As Van Vleet and Feeney (2015) noted, although many scholars have investigated the varieties and functions of play in children and nonhuman animals, they have paid much less attention to adult play and its effects. Because playing games is a particularly good way to satisfy the basic human needs identified by self-

determination theory (Deci & Ryan, 2000; Ryan & Deci, 2017)—namely autonomy, competence and relatedness (Rigby & Ryan, 2011)—more research examining the impact of game play upon the positive emotional states and well-being of adults is called for. The primary goal of the current study was to establish whether a brief game play experience increases a particular aspect of well-being: the feeling of being alive and full of energy, known as subjective vitality (Ryan & Frederick, 1997). We also explored whether this experience produces a decrease in negative affect. As interactive media, games can be exception ally diverse in their content, medium, structure and length. One of the features of a game session that has the potential to strongly affect the play experi ence is whether one is playing competitively against an opponent (as in chess and individual sports such as golf or tennis), together with others cooperatively

*Faculty mentor

Maynard, Mancini, and Rolón | Play and Subjectivity Vitality

against the challenges of the game (as in Dungeons and Dragons or massively multiplayer online roleplaying games [MMORPGs] such as World of Warcraft), or by oneself (as in solitaire or many single-player video games). Although researchers have compared cooperative versus competitive play, no consistent pattern of results has emerged. The literature on this topic is also currently limited to video games, where people frequently play together remotely, rather than in the same space. Given the growing popularity of modern board and card games (e.g., Wingfield, 2014) and other forms of nondigital play (e.g., tabletop roleplaying games, escape rooms), there is a need to extend existing research to incorporate these types of colocated play experiences. Therefore, the second goal of the current study was to explore possible differences in subjective vitality and negative affect across the two modes of social play (competitive and cooperative). Adult Play and Well-Being The vast majority of scholarly attention on the role of play in development and functioning has focused on animals (e.g., Burghardt, 2005) and children (e.g., Smith & Roopnarine, 2019). However, there is evidence that play is important for well-being in adults. For example, Csikszentmihalyi (2000) described an early study where he asked individuals to refrain from “doing anything that is ‘play’ or non-instrumental” (p. 162). After 48 hours, 80% of participants reported an increase in physical symptoms such as tiredness, irritability, and headaches, as well as decreases in creativity and reasonableness. Van Vleet and Feeney (2015) reviewed evidence that play among partners in romantic relationships is associated with feelings of trust and intimacy toward one’s partner as well as reductions in conflict. In addition, Russoniello, O’Brien, and Parks (2009) showed that playing casual video games can improve mood, increase relaxation, and reduce anxiety. Self-determination theory (Deci & Ryan, 2000; Ryan & Deci, 2017) provides a model for under standing why play may be vital to adult well-being. According to self-determination theory, all humans have three basic needs: autonomy (the need to freely choose one’s own actions), competence (the need for challenge and to demonstrate the skills to meet those challenges), and relatedness (the need to meaningfully and securely connect with others). People tend to be intrinsically motivated to engage in activities that satisfy these needs, and doing so fosters well-being. Game play is very

often associated with freedom of choice, mastery, and/or social connection. For example, a chess player not only decides to play, but makes many strategic decisions during each game, can see improvement with practice, and may connect with opponents and members of the broader play community. Ryan and colleagues (Rigby & Ryan, 2011; Ryan, Rigby, & Przybylski, 2006), through their Player Experience of Need Satisfaction model, demonstrated how self-determination theory applies specifically to video game play. Specifically, satisfaction of autonomy, competence, and relat edness needs through play resulted in increased enjoyment and motivation to play again in the future. Moreover, satisfaction of competence needs in particular was associated with positive changes in mood and state self-esteem. Subjective vitality. Constructs such as positive emotion, life satisfaction, and self-esteem have been heavily researched as indicators of well-being. Subjective vitality, defined as “one’s conscious experience of possessing energy and aliveness” (Ryan & Frederick, p. 530), has received relatively less attention. Ryan and Frederick argued that subjective vitality has both somatic and psychologi cal origins. For example, although the experience of pain or illness will tend to reduce one’s feeling of vitality, two individuals with the same physical condition may experience that condition differ ently (e.g., as a temporary challenge or a constant burden), resulting in differing perceived energy levels. More broadly, they argued that a feeling of energy and vitality can spring from any activity that satisfies autonomy, competence, and relatedness needs. For example, Nix, Ryan, Manly, and Deci (1999) experimentally varied whether participants engaged in a card-sorting activity in an autonomous or controlled fashion, and found that those in the former condition reported greater feelings of vitality. Ryan and Frederick (1997) argued that the experience of subjective vitality has the potential to spur personal agency and growth, foster resilience, and lead to behavioral choices that support physi cal health. Little research thus far has examined the role of vitality as a precursor to behavior, wellbeing, or physical health, however. Instead, most studies have focused upon subjective vitality as an outcome of positive and negative physical experi ences and activities (e.g., Niemiec, Ryan, Patrick, Deci, & Williams, 2010). Ryan and Frederick (1997) found that individuals experiencing more severe physical symptoms and more disabling pain

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

Play and Subjectivity Vitality | Maynard, Mancini, and Rolón

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

100

reported lower vitality, whereas physical self-esteem and adherence to an exercise regimen were both associated with greater subjective vitality. Others have shown that, for physical activities such as play ing soccer (Ommundsen, Lemyre, Abrahamsen & Roberts, 2010) and taking walks during work breaks (Kinnaflick, Thøgersen-Ntoumani, Duda, & Taylor, 2014), perceived autonomy within the activity is associated with greater subjective vitality. Play and vitality. We chose to focus on the effects of play upon subjective vitality for several reasons. The experience of subjective vitality is an indicator of flourishing (Ryan & Deci, 2001) and is positively associated with other aspects of well-being such as self-actualization and self-esteem (Ryan & Frederick, 1997) and life satisfaction (Uysal, Satici, Satici, & Akin, 2014). As noted above, games in particular can satisfy the human needs of autonomy, competence, and relatedness. In turn, satisfaction of these needs is expected to promote vitality (Deci & Ryan, 2000). Therefore, if play can be shown to foster a sense of aliveness in adults, it suggests that play might serve as a pleasurable form of “wellness exercise.” Another reason to expect game play to pro mote subjective vitality is that games provide the opportunity to experience fiero, an Italian word for pride in success, which game designers have adopted to refer to the emotional thrill of triumph ing over adversity (McGonigal, 2011). The rush of fiero from victory or a particularly well-played moment within a game (even if one ultimately loses the game) is likely to increase one’s sense of being alive and full of energy—that is, it is likely to increase subjective vitality. Research on the effect of play upon feelings of energy and vitality is mostly limited to sports and exercise games with a clear emphasis on physical activity (e.g., Mackintosh, Standage, Staiano, Lester, & McNarry, 2016; Ommundsen et al., 2010). In the only two published experimental studies on video game play and vitality, Ryan et al. (2006) found that vitality actually dropped after playing a 3D adventure video game for 40 minutes, although the change in vitality was more positive for par ticipants whose competence need was satisfied through play. Researchers have yet to examine the effects of playing a game that is neither physically demanding nor digital in nature upon subjective vitality. Because participants in our study played for a shorter period of time than in Ryan et al.’s research (15–20 minutes), and had the opportunity to improve their performance through multiple plays (thereby fostering competence), we expected

that the brief play session would energize rather than drain participants. Therefore, Hypothesis 1 was that, regardless of the mode of play, participants would experience an increase in subjective vitality from pregame to postgame. Negative affect. Whereas our focus in the cur rent investigation was on the impact of play upon vitality as a marker of a state of well-being, play also has the potential to reduce negative emotions and stress. For example, Wang, Rouse, and Mancuso (2017) found that participants who played the video game The Sims 2 for 25 minutes experienced no change in positive affect, but a significant decrease in negative affect. Therefore, we also examined changes in negative affect as a function of competi tive, cooperative, and solo play. Competitive Versus Cooperative Play Until the rise of video games in the latter part of the 20th century, most games were both social (rather than solitary) and competitive in nature (either one vs. one, one vs. many, or team vs. team). An increase in the availability of cooperative game experiences began with tabletop roleplaying games such as Dungeons and Dragons, followed by many modern video games, and more recently cooperative tabletop games (e.g., Pandemic) and escape rooms (Nicholson, 2015). Whether one is playing competitively or cooperatively may provide different experiences in terms of need satisfaction and effects on feelings of energy and aliveness. Competitive and cooperative play are both social in nature, but they do differ in several respects. The former (usually) has a goal structure where progress toward victory for one player tends to impede progress toward victory for other players, whereas in the latter, players’ goals are aligned, and they succeed or fail together (Bonta, 1997). Working together toward a game goal could either promote or inhibit a state of well-being. On the one hand, cooperation can lead to increased feelings of relatedness as the two (or more) players coordinate their actions to achieve a game goal together and hopefully enjoy a shared victory. Standage, Duda, and Pensgaard (2005) found that participants playing a dancing video game experienced greater need satisfaction, positive affect, and subjective vitality when playing together against another team, as opposed to playing against each other (1 vs. 1). Iizuka (1994) found that participants play ing a video bowling game cooperatively spent more time looking at and talking to play partners than others who played the same game competitively. Finally, cooperative play can also create moments of camaraderie and mutual storytelling even in the

Maynard, Mancini, and Rolón | Play and Subjectivity Vitality

face of failure (Maynard & Herron, 2016). On the other hand, the direct competition against another human opponent might promote physiological arousal and greater opportunities for fiero, both of which could lead to feelings of vitality. Kivikangas, Kätsyri, Järvelä, & Ravaja (2014) found that men (and, in one of two experiments, women) experienced greater positive emotion and arousal in response to competitive over cooperative video game play. Schmierbach, Xu, Oeldorf-Hirsch, and Dardis (2012) found that participants enjoyed play ing a video game of football more against another player than cooperatively or against a computer opponent, and suggested that the interdependence inherent in cooperative play may create frustration and detract from the enjoyment of the activity. In sum, there is not a consistent pattern of results on the effects of cooperative versus competitive play on outcomes such as enjoyment or emotional state. Therefore, we did not make specific predictions regarding the relative effect of competitive versus cooperative play upon subjec tive vitality. However, we examined this possible effect in order to contribute to the literature on potential differences in the experience of playing cooperatively versus competitively. Finally, individuals differ in the kinds of play they prefer, and some enjoy competitive situa tions more than others. For example, Yee (2006) surveyed 3,000 MMORPG players and identified 10 separate player motivations, two of which were competition and teamwork. A strong match between one’s play preferences and the play experi ence may promote particularly positive outcomes. Therefore, in the present study, we included a mea sure of trait competitiveness to test the hypothesis that a player’s inclination (or disinclination) for competition would interact with the game mode. With Hypothesis 2, we expected that trait competi tiveness would moderate the relationship between game mode and subjective vitality, such that the increase in vitality from pregame to postgame would be stronger for competitive participants in the competitive condition. Solo Play Video games on computers and consoles in the 1980s and beyond greatly increased the opportunity to play games alone, and in the past decade there has been a notable increase in board games that provide modes for solo play (Leorke, 2018). The key feature of solitary play is the lack of human play partners. In terms of need satisfaction, although solo play can satisfy autonomy and competence

needs as one makes in-game choices and develops their skills, it provides no inherent opportunity to relate to others or experience a sense of belong ingness during the game. In contrast, multiplayer games can provide opportunities to meet new people and establish or strengthen friendships while playing the game (e.g., Williams et al., 2006). Further, sharing an activity can enhance the experience relative to when it is undertaken alone (Boothby, Clark, & Bargh, 2014). Kaye and Bryce (2014) found that, in recalling solitary and social video game sessions, participants reported greater positive emotion as a result of social play. We followed up our investigation of social play in Study 1 with a second study utilizing the same research protocol but examining a solo play mode. We hypothesized that, although the benefits of solo play may be somewhat smaller as compared to competitive or cooperative play, it would still result in increased subjective vitality (Hypothesis 1) because it could still satisfy a player’s need for autonomy and competence.

Study 1 Method Participants. Participants were 108 college students who received psychology subject pool credit in exchange for their participation. We recruited participants by posting information about our study on the department’s online subject pool system account (Sona Systems, www.sona-systems.com), where participants can view available studies and reserve a time slot to participate. Two participants were excluded from analyses for skipping whole pages of the pre- or postgame questionnaire, result ing in a final sample of 106 participants (53 play pairs). Participants had a mean age of 20.64 years (SD = 1.69 years). There were 81 female participants (76.4%), 24 male participants (22.6%), and one participant who marked “other” as their gender. Because of difficulties associated with schedul ing two participants to play together (namely, when only one participant signed up or two signed up but one did not show), we adjusted the recruitment procedure so that the participant who signed up was asked to bring someone to play the game with them (such as a friend, classmate, or roommate). As a result, of the 106 participants, 34 participants were assigned to their partners (17 play pairs) and 72 played with a partner they had brought with them to the study (36 play pairs). Because participating with an assigned versus chosen partner may affect the play experience, we included this variable in all analyses, as described below.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

101

Play and Subjectivity Vitality | Maynard, Mancini, and Rolón

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

102

Design. Rhino Hero (Frisco & Strumpf, 2011), a dexterity card-stacking game, was chosen for this study because it is simple (making it easy to learn quickly), quick (playtime is roughly 5 minutes), and relatively unknown in the United States (as it is produced in Germany). The original Rhino Hero game is competitive, where players take turns plac ing cards to form levels of an apartment building. Some cards have particular abilities, such as forcing one’s opponent to move a wooden rhino figure up the building. A player wins by being the first to get rid of their hand of seven cards without knocking the tower over. We created a cooperative version of the game by giving each player a smaller hand of five cards. Their goal in this mode was to build the tower together, attempting to win by getting rid of both of their hands without knocking the tower over. Prior to their arrival, pairs of participants were randomly assigned to play either the competitive (30 pairs, n = 60) or cooperative (23 pairs, n = 46) version of the game based upon the results of a die roll. Measures. Subjective vitality. We measured participants’ current feeling of being full of energy and life with the Ryan and Frederick (1997) subjective vitality measure both before and after play. Based upon the results of validation work by Bostic, Rubio, and Hood (2000), we used the six-item version of the scale that omitted the single reverse-scored item from the original scale. A sample item is, “I feel energized right now.” For each item, respondents chose a response from 1 (not at all true) to 7 (very true). Bostic et al. reported coefficient alpha values above .80 across two samples for this scale. In Study 1, coefficient alpha values were very good (.89 and .93 for before and after play, respectively). Negative affect. The Negative Affect subscale of the Positive and Negative Affect Scale (PANAS; Watson, Clark, & Tellegen, 1988) was included in both the pregame and postgame questionnaire. It contains 10 single-word items (e.g., “distressed,” “irritable”). Items represent emotional states, and respondents indicated the extent to which they felt each of those states at the present moment on a Likert-type scale from 1 (very slightly or not at all) to 5 (extremely). Watson et al. (1988) reported a coeffi cient alpha of .85 when using the “present moment” version of the scale instructions. Coefficient alpha values for Study 1 before play (a = .81) and after play (a = .73) were fair to good. Competitiveness. Trait competitiveness was measured in the postgame questionnaire with the

Enjoyment of Competition subscale of the Revised Competitiveness Index (Houston, Harris, McIntire, & Francis, 2002), which contains 9 items (e.g., “I get satisfaction from competing with others.”). Respondents indicated agreement with each item on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). Houston et al. reported a coefficient alpha of .90 for this subscale, and in Study 1, inter nal reliability was very good (a = .92). Win percentage. In the postgame questionnaire, we asked participants to record the number of games that they played and how many of those they won (either against their opponent or together with their teammate, depending on the condition). We then calculated a win percentage by dividing the number of wins by the total number of games played. Open-ended responses. In addition to the quanti tative measures above, we asked participants three open-ended questions in the postgame question naire: “What was your favorite part about playing the game?”, “What was your least favorite part about playing the game?”, and “Is there anything else you would like to comment about in terms of your experience playing the game today?” Responses to these questions were not the primary focus of this study, but we will note some themes as they pertain to the quantitative findings in the Discussion. Procedure. Prior to beginning data collection, we received approval to conduct this study from the State University of New York at New Paltz Human Research Ethics Board. After completing an informed consent form, participants completed a pregame questionnaire including state measures of subjective vitality and negative affect. They were introduced to the game by a research assistant, who explained the rules of Rhino Hero for the mode they had been randomly assigned to and noted that they would have roughly 20 minutes to play while the assistant sat outside the room within earshot in case of any questions. The research assistant also mentioned that, because Rhino Hero is a short game, there would be time for at least two full plays of the game, and that participants should keep track of how many times they played on a card provided to them. A single page “cheat sheet” was also provided to participants as a reminder of the rules of the game. The assistant then left the participants in the room to play. After 15 minutes had elapsed, the assistant knocked on the door and indicated that the participants should complete the in-progress game. Finally, participants were given a postgame questionnaire to complete that included measures of subjective vitality, negative affect, and trait

Maynard, Mancini, and Rolón | Play and Subjectivity Vitality

competitiveness, as well as game results (how many won and lost), the three open-ended questions, and demographic questions (i.e., gender and age). Results All participants reported that they had never played Rhino Hero prior to their participation in the study. They played between one and five games in the time allotted, with nearly all pairs playing three (50.9%), two (21.7%), or four (21.7%) games. The percent of games participants won ranged from 0 to 100%; 79.7% of participants won at least once. Descriptive statistics, correlations, and coefficient alpha values are displayed in Table 1. Subjective vitality and negative affect were negatively related at pregame (r = -.40, p < .001), but not at postgame (r = -.18, p = .059). Interestingly, win percentage was not related to postgame vitality or negative affect (both ps above .50). Women scored lower on trait competitiveness (r = -.34, p < .001) than men did. Because participants played the game in pairs, their scores on dependent variables may be influenced by their play partner. We tested for this potential nonindependence by calculating a partial intraclass correlation on the data from the complete pairs across these two conditions, controlling for condition, and with a liberal alpha level of .20, as recommended by Kenny, Kashy, and Cook (2006). The partial intraclass correlation was significant at this alpha level, r = .23, F(33,35) = 1.59, p = .091. This result suggests that subjective vitality scores from play partners are indeed more similar than scores from two participants who did not play together, beyond the effect of the condi tion they were assigned to. Therefore, in testing our hypotheses, we utilized linear mixed effects (sometimes called mixed modeling; e.g., Baayen, Davidson, & Bates, 2008) using the lme4 function (Bates, Maechler, & Bolker, 2012) in the R statistical package (R Core Team, 2012). This approach is similar to standard multiple regression but allowed us to account for nonindependence by modeling random effects such as the idiosyncratic differences in pregame and postgame measures from play pair to play pair. Pregame vs. postgame subjective vitality. In Hypothesis 1, we predicted that participants would experience an increase in subjective vitality after playing the game, regardless of whether they played competitively or cooperatively. We also wished to examine whether the play mode (cooperative or competitive) affected the degree of change in vitality scores from pregame to postgame. Means and standard deviations for pregame and postgame

subjective vitality scores are presented in Table 2, along with effect sizes (Cohen’s d). As shown, participants in both conditions reported higher average levels of subjective vitality postgame as compared to pregame. We began by testing a random intercept model that predicted subjective vitality scores as a function of time (before versus after play), play mode, trait competitiveness, and recruitment method, while accounting for variance between participants (i.e., individual differences) and play pairs. As shown in Table 3, the increase in subjective vitality from pregame to postgame across both conditions was 0.64 on a 7-point scale, which was significant, t(106) = 6.92, p < .001, d = 0.66. There was also a significant effect of play mode upon subjective vitality, t(54.61) = -2.75, p = .008, with participants in the cooperative condition experiencing lower TABLE 1 Means, Standard Deviations, and Correlations Among Study 1 Variables Variable

1. Subjective Vitatlity

4.20

1.24

–

2. Negative Affect

1.51

0.55

-.40**

–

3. Win Percentage

0.43

0.33

-.03

-.11

–

4. Subjective Vitality

4.84

1.28

.71**

-.31**

-.06

–

5. Negative Affect

1.33

0.38

-.14

.57

.04

-.18

–

6. Competitiveness

3.49

0.82

.19

-.08

.12

.22*

.01

20.64

1.69

.16

-.09

.09

.13

-.16

.11

–

-.18

.05

-.12

.05

-.34**

-.31**

Pregame Measures

Postgame Measures

7. Age 8. Gender (men = 0; women = 1)

–

Note. N = 106. ** p < .05. ** p < .01.

TABLE 2 Means, Standard Deviations, and Effect Sizes for Pregame-Postgame Measures Subjective Vitality Condition

Negative Affect

Pregame

Postgame

Pregame

Postgame

M (SD)

Cohen's d

M (SD)

Cohen's d

Study 1 Competitive

4.45(1.27)

5.18(1.19)

0.73

1.48(0.47)

1.28(0.32)

0.59

Cooperative

3.87(1.11)

4.38(1.26)

0.59

1.54(0.64)

1.39(0.44)

0.26

Total

106

4.20(1.24)

4.83(1.28)

0.66

1.51(0.55)

1.33(0.38)

0.39

4.40(1.40)

4.79(1.39)

0.47

1.43(0.48)

1.31(0.38)

0.34

Study 2 Solo

103

Play and Subjectivity Vitality | Maynard, Mancini, and Rolón

subjective vitality than those in the competitive condition independent of time. To test for a possible effect of play mode upon the amount of change in subjective vital ity, we ran a second model that was identical to the first model, except for the inclusion of an interaction term between time and play mode. Based on a comparison between this model and the base model, the estimate for the interaction was not significant and did not improve model fit, χ(1) = 1.47, p = .225. Figure 1 shows that, despite random assignment, participants in the coop erative condition had lower pregame subjective vitality scores than participants in the competitive condition. However, the nonsignificant interac tion suggests that participants in both conditions experienced a similar increase in vitality after playing the game. The role of competitiveness. The results of the first model indicate that trait competitiveness predicted vitality after accounting for time, play mode, and recruitment method, t(101.91) = 2.23, p = .028; that is, overall, participants with higher competitiveness scores reported higher levels of vitality, independent of which condition they were TABLE 3 Linear Mixed Effects Model for Subjective Vitality (Study 1) Fixed Effect

Estimate

Intercept

4.71

Time

0.64

Play Mode

-0.63

Trait Competitiveness

0.28

Recruitment Method

-0.35

0.22

58.87

21.21

<.001

0.09

106.00

6.92

<.001

0.23

54.61

-2.75

.008

0.12

101.91

2.23

.028

0.25

54.22

-1.42

.162

Note. N = 106 (53 pairs). Random effects (participant and play pair) are not displayed for simpilicity.

FIGURE 1

Subjective Vitality

6.00

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

104

Play Mode Competitive Cooperative Solo

5.00

4.00 3.00 2.00

Pregame

Time

Postgame

Figure 1. Mean pre- and postgame subjective vitality scores for participants in the competitive and cooperative conditions in Study 1.

assigned to. In Hypothesis 2, we predicted that trait competitiveness would moderate the effect of play mode, such that increases in subjective vitality would be strongest for competitive participants in the competitive play mode condition. To test this hypothesis, we compared two models, which both used change scores in subjective vitality as the out come variable. The first included the fixed effects of play mode, competitiveness (which was mean centered prior to this analysis), and recruitment method, while accounting for the random effect of play pair. The second model was identical except that it also included a term for the interaction between play mode and trait competitiveness. None of the variables in the model significantly predicted change in subjective vitality, and the inclusion of the interaction term did not improve model fit, χ(1) = 0.03, p = .867. Therefore, Hypothesis 2 was not supported. Changes in negative affect. Table 2 shows that, in both conditions, participants reported a mean decrease in negative affect from pregame to postgame. To examine changes in negative affect, we ran a linear mixed-effects model with negative affect as the outcome variable, participant and play pair as random effects to be accounted for, and time, play mode, trait competitiveness, and recruitment method as the predictors (see Table 4). Time was the only significant predictor, with a mean decrease of -0.20 points (on a 5-point scale) from pregame to postgame, t(106) = -3.32, p = .001. An interaction term between time and play mode added to a second model was not significant (esti mate = 0.05, t(106) = 0.50, p = .615), suggesting that the decrease in negative affect did not depend on whether participants were playing the cooperative or competitive version of the game. Study 1 summary. Across the board, partici pants reported an increase in subjective vitality and a decrease in negative affect after playing Rhino Hero. The amount of change in either emotional state did not depend upon which mode of the game they were playing (cooperative versus competitive). In addition, contrary to expectations, participants higher in trait competitiveness did not experience a greater increase in subjective vitality when playing the competitive mode. One thing that participants had in common regardless of play mode was that they played the game together with a partner. We do not know whether players would experience similar changes in emotional state if they played the game by them selves. Therefore, in Study 2, we examined pre- and postgame subjective vitality and negative affect in a solo play mode of the same game.

Maynard, Mancini, and Rolón | Play and Subjectivity Vitality

Study 2 Method Participants. Participants were 56 college students who received psychology subject pool credit in exchange for their participation, recruited through the department’s online subject pool system account as in Study 1. Individuals who par ticipated in Study 1 did not have the opportunity to participate in Study 2. Two participants were excluded from analyses for a lack of understanding of study instructions, resulting in a final sample of 54 participants. Participants had an average age of 22.31 years (SD = 4.81 years); there were 40 women (74.1%), 12 men (22.2%), and two reporting either nonbinary or “other” gender (3.8%). Design, measures, and procedures. Participants in this study played a solo version of Rhino Hero cre ated for the purposes of the study. The solo version of the game was identical to the cooperative version described in Study 1, except that the participant was given all 10 cards and played alone; if they were able to use all 10 cards without knocking the tower over, they won. We calculated win percentage as the percentage of games they played in which they were able to use all 10 of their cards without knocking over the tower. All other measures in Study 2 were identical to those in Study 1. Coefficient alpha values for Study 2 were similar to those in Study 1 (a = .91 and .92 for subjective vitality before and after play, respectively, a = .83 and .69 for nega tive affect before and after play, respectively, and a = .91 for trait competitiveness) and were good to very good with the exception of negative affect measured after play, which was fair. We obtained State University of New York at New Paltz Human Ethics Review Board approval prior to data collec tion, and the procedure was otherwise identical to that in Study 1. Results Participants played between one and five games in the time allotted, with most playing three (38.9%) or four (29.6%) games. The percent of games par ticipants won ranged from 0 to 100%, and 79.2% of participants in each condition won at least once. Descriptive statistics, correlations, and coefficient alpha values are displayed in Table 5. Win percent age was not related to postgame vitality (r = .02, p = .889), but a higher percentage of games won was associated with lower levels of postgame nega tive affect (r = -.40, p = .003). We expected that participants playing the game alone would experience an increase in

subjective vitality from pre- to postgame. Consistent with this hypothesis, a paired-samples t test revealed a significant increase in subjective vitality, t(53) = 3.45, p = .001, d = 0.47, 95% CI [0.16, 0.62]. We also examined whether there were changes in negative affect. Participants playing alone expe rienced a significant decrease in negative affect from pregame to postgame, t(53) = -2.48, p = .016, d = 0.34, 95% CI [-.02, -.22]. Means and standard deviations for these measures are shown in the bottom row of Table 2.

Discussion The purpose of the present research was to investi gate the effect of a brief session of nondigital play upon feelings of energy and liveliness in adults. We modified a published dexterity card game in order to examine the effects of competitive and cooperative play (Study 1) and solo play (Study 2) TABLE 4 Linear Mixed Effects Model for Negative Affect (Study 1) Fixed Effect

Estimate

Intercept

1.40

0.08

55.05

16.65

<.001

Time

-0.18

0.04

-3.96

<.001

Play Mode

0.08

0.09

48.47

0.92

.363

Trait Competitiveness

-0.03

0.05

102.19

-0.70

.488

Recruitment Method

0.10

0.09

48.08

1.08

.285

106

Note. N = 106 (53 pairs). Random effects (participant and play pair) are not displayed for simpilicity.

TABLE 5 Means, Standard Deviations, and Correlations Among Study 2 Variables Variable

1. Subjective Vitatlity

4.40

1.40

–

2. Negative Affect

1.43

0.48

-.18

–

3. Win Percentage

0.43

0.31

.06

-.20

–

4. Subjective Vitality

4.79

1.39

.82

-.15

-.02

–

5. Negative Affect

1.31

0.38

-.15

.68**

-.40**

-.05

–

Pregame Measures

Postgame Measures

6. Competitiveness 7. Age 8. Gender (men = 0; women = 1)

3.66

0.78

-.10

-.05

.05

-.07

–

22.31

4.81

.05

-.22

.14

.08

-.24

-.15

–

-.08

.07

.00

.08

.12

-.26

-.27

Note. N = 106. ** p < .05. ** p < .01.

105

Play and Subjectivity Vitality | Maynard, Mancini, and Rolón

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

106

on subjective vitality and negative affect. All partici pants across both studies played with the identical game components, for roughly the same amount of time, and had the same ultimate goal (i.e., don’t knock the tower over). Consistent with Hypothesis 1, participants in all three play modes experienced an increase in vitality (as well as a decrease in negative affect) from pregame to postgame. There was no significant difference in the amount of change from pregame to postgame for the competitive versus the cooperative condition. A number of open-ended responses provided by participants in the two studies corroborated the observed changes in the quantitative measures. For example, 16 participants (10% of all participants) volunteered that the related emotions of excite ment, suspense, anticipation, and uncertainty comprised their favorite part of playing the game (e.g., “It was exciting and any little move could make it fall;” “[My favorite part was] Seeing how high we could build up the cards and the suspense of trying not to have it fall.”). More broadly, 17 (11%) participants mentioned a positive change in their mental or physical state after having had the play experience, and not a single participant mentioned a negative change due to play. The most common changes cited were an improve ment in mood (e.g., “I was a bit cranky before the game & it lifted my spirits.”), stress reduction or increased relaxation (e.g., “[I] left more relaxed than when I came in.”), and increased alertness and energy (e.g., “I felt awake and energized and forgot my concerns.”). Relatedly, 15 participants (9%) described being able to forget daily concerns such as school or work (e.g., “It was a fairly simple game but it kept my attention and allowed me to escape the world for a little while.”). The increases in vitality and decreases in negative affect, both of which appeared in some participants’ postgame comments, suggest that a brief session of play can simultaneously boost feelings of energy while reducing negative emotions. Future researchers should therefore explore the possibilities for brief play interventions for well-being, perhaps within the context of workplaces, schools, or care facilities. The increase in vitality across all conditions in this experiment is consistent with the notion that games are intrinsically motivating and can foster individual well-being. However, the results contradicted those of Ryan et al. (2006), who found an overall decrease in vitality in two experiments of video game play. There are at least two explanations for this discrepancy that merit further investigation.

First, in our experiment, participants played for 15–20 minutes as compared to the 40 minutes in Ryan et al.’s studies. It is possible that the impact of play upon vitality is nonlinear, initially energizing players but eventually draining them if the play continues for too long. Second, playing a video game may be more taxing physiologically than a card game in terms of visual fatigue and repetitive motor actions. Because our investigation of solo play was in a separate study from the other two conditions, we could not directly compare the emotional effects of play between social play versus solitary play. However, two tentative clues from the cur rent research suggest that people may experience greater subjective vitality as a result of social as compared to nonsocial play, at least with regards to a nondigital game such as the one we used in our research. First, the size of the increase in subjective vitality from pregame to postgame was somewhat stronger in Study 1 (d = 0.66 across the two conditions) than in Study 2 (d = 0.47). Second, in the open-ended postgame responses from Study 1, nearly half of participants in the competitive (45%) and cooperative (48%) condi tions mentioned their partner or interactions with them as their favorite part of playing the game. This was even true of participants who were assigned a play partner; over a third (n = 9; 36%) mentioned enjoying the opportunity to meet or make a connection with someone new (e.g., “I feel I built a relationship with the other par ticipant by talking and strategizing.”) or had positive things to say about their partner (e.g., “[My partner] was very engaged and talkative, which made our interactions meaningful and genuine.”). Another intriguing finding is that win percentage was related to post-game negative affect for solo play (r = -.40, p = .003) but not for social play (r = .04, p = .713); it is possible that without the social component of the game, the player’s perfor mance has a greater impact upon their emotional state after play. Taken together, these clues indicate that future research directly comparing the emo tional effects of social to solitary play may be fruitful. Contrary to Hypothesis 2, participants did not experience a greater boost in vitality when their preference (or lack thereof) for competitive situations matched the mode of game they played. One possibility for this lack of finding is that the participants in the cooperative condition were still competing, just against the game itself rather than against each other. There was also evidence from

Maynard, Mancini, and Rolón | Play and Subjectivity Vitality

the open-ended responses that at least a couple of participants in the cooperative condition added a competitive layer (e.g., “[My favorite part was] the fact that it wasn’t competitive but we still sort of made it that way.”), whereas some in the competitive condition described having a collaborative side-goal (e.g., “My partner and I were really focused on try ing to move Rhino hero up the tower. We wanted to get him as high as possible.”). These comments are a helpful reminder that players do not always limit the way they play to the stated rules and objectives. They often come to a play experience with their own social, cognitive, and emotional goals, which can affect their experience of the game (Yee, 2006). Study Limitations and Future Directions There were several limitations to the current research that should be considered. First, because all participants completed the subjective vitality and negative affect measures before and after their play session, it is possible that some participants formed demand characteristics about the purpose of the research. Although the open-ended feedback cor roborates the quantitative findings regarding the benefits of play on one’s emotional state, follow-up research with distractor measures to obscure the goals and hypotheses of the study would be helpful. Second, there are several important limits to the generalizability of the results. All participants were college students, and they all played different modes of a simple card-stacking dexterity game. The tactile (holding and moving the cards and the wooden rhino figure), and colocated (both players in the same physical location, for Study 1) nature of the game means that the current findings may not apply to games with limited physical actions or video games where social interaction occurs online. The emotions that a certain style of game tends to elicit may also matter. For example, a game with escalating tension and the possibility of a collapsing tower of cards like Rhino Hero might be expected to produce greater feelings of energy than more pensive, cerebral games (e.g., chess). Future research with different gaming experiences is crucial to test the boundary conditions of the current findings, given the great diversity of games and ways of playing them. Another area for future research relates to the duration and potential benefits of a boost in subjective vitality. Because the postgame measures were administered soon after the play session, it is unclear how long the immediate emotional ben efits might endure. For example, does time spent

playing actually shape how people feel and act for the remainder of the day, or do these states quickly recede as people move on to the next activity on their schedule? More broadly, does meeting one’s need for play support long-term well-being? There are hypothesized but as-yet untested positive out comes of feeling a sense of energy and vitality, such as engaging in behavior that promotes personal growth and health (Ryan & Frederick, 1997). If these relationships can be established, subjective vitality may emerge as an important mediator that helps explain how play in adults leads to positive outcomes. Third, most definitions of play note that it is, at its heart, a voluntary activity (Huizinga, 1938/2016; Suits, 1978). In this experiment (as in most game-based research), although participants did volunteer to be in a study where they would play a game, they did not get to choose the game, the game mode, or the length of play (indeed, some noted that they wished they could have played longer). It could be argued that having the freedom to choose how and when to play would increase satisfaction of one’s need for autonomy, and thus enhance the benefits of play (such as boosting feel ings of vitality) to an even greater degree. Finally, it should be noted that the competitive mode was the original mode of the game design, whereas the cooperative and solo modes were modifications we made for the purposes of the study in order to standardize the play experience in other ways (e.g., game theme, components, goals). As a result, there might be inherent differences in excitement level across the conditions. Conclusion These two studies represent a first attempt to examine changes in subjective vitality as a result of a brief play experience with the same game but under three different styles of play (competitive, cooperative, and solo). All three ways of playing resulted in an increased sense of vitality or aliveness, and a decrease in negative emotion. This suggests that even a brief play session with an assigned, unfamiliar game can boost experienced wellness, at least temporarily. Future research is needed to determine whether the encouraging results found in the current study persist for a longer period of time (e.g., for the remainder of the day), as well as to explore how features of the game (such as the actions involved in play) or situation (such as the number of players) impact a player’s emotional state.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

107

Play and Subjectivity Vitality | Maynard, Mancini, and Rolón

References

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

108

Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59, 390–412. https://doi.org/10.1016/j.jml.2007.12.005 Bates, D. M., Maechler, M., & Bolker, B. (2012). lme4: Linear mixed-effects models using S4 classes. R package version 0.999999-0. Bonta, B. D. (1997). Cooperation and competition in peaceful societies. Psychological Bulletin, 121, 299–320. https://doi.org/10.1037/0033-2909.121.2.299 Boothby, E. J., Clark, M. S., & Bargh, J. A. (2014). Shared experiences are amplified. Psychological Science, 25, 2209–2216. https://doi.org/10.1177/0956797614551162 Bostic, T. J., Rubio, D. M., & Hood, M. (2000). A validation of the subjective vitality scale using structural equation modeling. Social Indicators Research, 52, 313–324. https://doi.org/10.1023/A:1007136110218 Brown, K. W., & Ryan, R. M. (2003). The benefits of being present: Mindfulness and its role in psychological well-being. Journal of Personality and Social Psychology, 84, 822–848. https://doi.org/10.1037/0022-3514.84.4.822 Burghardt, G. (2005). The genesis of animal play: Testing the limits. Cambridge, MA: The MIT Press. Csikszentmihalyi, M. (2000). Beyond boredom and anxiety. San Francisco, CA: Jossey-Bass. Deci, E. L., & Ryan, R. M. (2000). The ‘what’ and ‘why’ of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11, 227–268. https://doi.org/10.1207/S15327965PLI1104_01 Dunn, E. W., Aknin, L. B., & Norton, M. I. (2008). Spending money on others promotes happiness. Science, 319(5870), 1687–1688. https://doi.org/10.1126/science.1150952 Emmons, R. A., & McCullough, M. E. (2003). Counting blessings versus burdens: An experimental investigation of gratitude and subjective well-being in daily life. Journal of Personality and Social Psychology, 84, 377–389. https://doi.org/10.1037/0022-3514.84.2.377 Field, A. (2018). Discovering statistics using IBM SPSS Statistics (5th ed.). Thousand Oaks, CA: Sage. Frisco, S., & Strumpf, S. (2011). Rhino hero [card game]. Bad Rodach, Germany: HABA. Houston, J., Harris, P., McIntire, S., & Francis, D. (2002). Revising the competitiveness index using factor analysis. Psychological Reports, 90, 31–34. https://doi.org/10.2466/PR0.90.1.31-34 Huizinga, J. (1938/2016). Homo ludens: A study of the play-element in culture. Kettering, OH: Angelico Press. Iizuka, Y. (1994). Gaze in cooperative and competitive games. Japanese Journal of Experimental Social Psychology, 33, 237–242. https://doi.org/10.2130/jjesp.33.237 Kaye, L. K., & Bryce, J. (2014). Go with the flow: The experience and affective outcomes of solo versus social gameplay. Journal of Gaming and Virtual Worlds, 6, 49–60. https://doi.org/10.1386/jgvw.6.1.49_1 Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York, NY: Guilford Press. Kinnafick, F., Thøgersen-Ntoumani, C., Duda, J. L., & Taylor, I. (2014). Sources of autonomy support, subjective vitality and physical activity behaviour associated with participation in a lunchtime walking intervention for physically inactive adults. Psychology of Sport and Exercise, 15, 190–197. https://doi.org/10.1016/j.psychsport.2013.10.009 Kivikangas, J. M., Kätsyri, J., Järvelä, S., & Ravaja, N. (2014). Gender differences in emotional responses to cooperative and competitive game play. PLoS ONE, 9, 1–16. https://doi.org/10.1371/journal.pone.0100318 Leorke, D. (2018). Solo board gaming: An analysis of player motivations. Analog Game Studies, 5(4). Retrieved from http://analoggamestudies.org/2018/12/ solo-board-gaming-an-analysis-of-player-motivations/ Maynard, D., & Herron, J. (2016). The allure of struggle and failure in cooperative board games. Analog Game Studies, 3(3). Retrieved from http://analoggamestudies.org/2016/05/the-allure-of-struggle-and-failurein-cooperative-board-games/ Mackintosh, K. A., Standage, M., Staiano, A. E., Lester, L., & McNarry, M. A. (2016). Investigating the physiological and psychosocial responses of single- and dual-player exergaming in young adults. Games for Health, 5, 375–381. https://doi.org/10.1089/g4h.2016.0015 McGonigal, J. (2011). Reality is broken: Why games make us better and how they

can change the world. New York, NY: Penguin Press. Nicholson, S. (2015). Peeking behind the locked door: A survey of escape room facilities. White Paper available at http://scottnicholson.com/pubs/erfacwhite.pdf Niemiec, C. P., Ryan, R. M., Patrick, H., Deci, E. L., & Williams, G. C. (2010). The energization of health-behavior change: Examining the associations among autonomous self-regulation, subjective vitality, depressive symptoms, and tobacco abstinence. The Journal of Positive Psychology, 5, 122–138. https://doi.org/10.1080/17439760903569162 Nix, G. A., Ryan, R. M., Manly, J. B., & Deci, E. L. (1999). Revitalization through self-regulation: The effects of autonomous and controlled motivation on happiness and vitality. Journal of Experimental Social Psychology, 35, 266–284. https://doi.org/10.1006/jesp.1999.1382 Ommundsen, Y., Lemyre, P., Abrahamsen, F., & Roberts, G. C. (2010). Motivational climate, need satisfaction, regulation of motivation and subjective vitality: A study of young soccer players. International Journal of Sport Psychology, 41, 216–242. R Core Team. (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Rigby, S., & Ryan, R. M. (2011). Glued to games: How video games draw us in and hold us spellbound. Santa Barbara, CA: Praeger/ABC-CLIO. Russoniello, C. V., O’Brien, K., & Parks, J. M. (2009). The effectiveness of casual video games in improving mood and decreasing stress. Journal of CyberTherapy & Rehabilitation, 2, 53–66. Ryan, R. M., & Deci, E. L. (2001). On happiness and human potentials: A review of research on hedonic and eudaimonic well-being. Annual Review of Psychology, 52, 141–166. https://doi.org/10.1146/annurev.psych.52.1.141 Ryan, R. M., & Deci, E. L. (2017). Self-determination theory. New York, NY: The Guilford Press. Ryan, R. M., & Frederick, C. M. (1997). On energy, personality and health: Subjective vitality as a dynamic reflection of well-being. Journal of Personality, 65, 529–565. https://doi.org/10.1111/j.1467-6494.1997.tb00326.x Ryan, R. M., Rigby, C. S., & Przybylski, A. (2006). The motivational pull of video games: A self-determination theory approach. Motivation and Emotion, 30, 344–360. https://doi.org/10.1007/s11031-006-9051-8 Schmierbach, M., Xu, Q., Oeldorf-Hirsch, A., & Dardis, F. E. (2012). Electronic friend or virtual foe: Exploring the role of competitive and cooperative multiplayer video game modes in fostering enjoyment. Media Psychology, 15, 356–371. https://doi.org/10.1080/15213269.2012.702603 Smith, P. K., & Roopnarine, J. L. (Eds.) (2019). The Cambridge handbook of play: Developmental and disciplinary perspectives. Cambridge, UK: The Cambridge University Press. Standage, M., Duda, J. L., & Pensgaard, A. M. (2005). The effect of competitive outcome and task-involving, ego-involving, and cooperative structures on the psychological well-being of individuals engaged in a co-ordination task: A self-determination approach. Motivation and Emotion, 29, 41–68. https://doi.org/10.1007/s11031-005-4415-z Suits, B. (1978). The grasshopper: Games, life and utopia. Toronto, Ontario: University of Toronto Press. Terr, L. (1999). Beyond love and work: Why adults need to play. New York, NY: Touchstone. Uysal, R., Satici, S. A., Satici, B., & Akin, A. (2014). Subjective vitality as mediator and moderator of the relationship between life satisfaction and subjective happiness. Educational Sciences: Theory & Practice, 14, 489–497. https://doi.org/10.12738/estp.2014.2.1828 Van Vleet, M., & Feeney, B. C. (2015). Young at heart: A perspective for advancing research on play in adulthood. Perspectives in Psychological Science, 10, 639–645. https://doi.org/10.1177/1745691615596789 Wang, I., Rouse, S. V., & Mancuso, E. K. (2017). The virtual self: Avatar and individual determinants of mood. Psi Chi Journal of Psychological Research, 22, 29–38. https://doi.org/10.24839/2325-7342.JN22.1.29 Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. https://doi.org/10.1037/0022-3514.54.6.1063 Williams, D., Ducheneaut, N., Xiong, L., Zhang, Y., Yee, N., & Nickell, E. (2006). From tree house to barracks: The social life of guilds in World of Warcraft. Games and Culture: A Journal of Interactive Media, 1, 338–361. https://doi.org/10.1177/1555412006292616

Maynard, Mancini, and Rolón | Play and Subjectivity Vitality

Wingfield, N. (2014, May 5). High-tech push has board games rolling again. New York Times. Retrieved from http://www.nytimes.com/ Yee, N. (2006). Motivations for play in online games. CyberPsychology & Behavior, 9, 772–774. https://doi.org/10.1089/cpb.2006.9.772 Author Note. Doug Maynard, https://orcid.org/0000-00024128-9985, Leah J. Mancini, and Vania Rolón, Psychology Department, State University of New York at New Paltz. Leah J. Mancini is now at New York State Department of Labor - Division of Research and Statistics. Vania Rolón is now at Brunel University London.

We wish to thank Egamaria Alacam, Ariel Barter, Kian Betancourt, Joanna Herron, Alessandra Moss, Steven O’Rourke, Andrew Perry, Usman Shakil, Sydney Shepard, and Allison Vaughn for their assistance with different phases of this research. Special thanks to Psi Chi Journal reviewers for their support. Correspondence concerning this article should be addressed to Doug Maynard, Psychology Department, State University of New York at New Paltz, 1 Hawk Drive, New Paltz, NY, 12561, USA. E-mail: maynardd@newpaltz.edu

109

https://doi.org/10.24839/2325-7342.JN25.2.110

Disentangling the Effects of Study Time and Study Strategy on Undergraduate Test Performance Zachary J. Cole and Darrell L. Butler* Ball State University

ABSTRACT. Increased study time is associated with improved memory. Students tend to use study time as a benchmark for gauging how prepared they are for a test. While studying, students tend to rely on rote memorization. This has led to students using judgments of processing fluency to determine their level of understanding for the study material. Elaboration, or active learning, is also associated with improved memory. The effects of elaboration appear to be confounded with study time. Two experiments were conducted to disentangle the effects of study time and study strategy on test performance. For both experiments, participants read an article, were randomly assigned to study elaboration or memorization flashcards, and took a test. In Experiment 1, study time was not controlled. Experiment 2 followed the same procedure as Experiment 1 except participants were randomly assigned to study for 7.5 or 15 minutes. For Experiment 1, the elaboration group studied longer (they had more to study), but were actually more efficient than the memorization group. The elaboration and memorization groups scored better on the test than the control group. For Experiment 2, the extended study condition scored better than the brief study condition, and the elaboration condition scored better than the memorization condition. There was no interaction between the study time and study strategy conditions. These findings suggest that study time and study strategy act independently to affect test performance. Keywords: study time, processing time, elaboration, memorization, test performance

S SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

110

tudy time is frequently positively associated with memory performance (e.g., d’Ydewalle, Swerts, & Corte, 1983), a principle known as the total time hypothesis (first reported by Ebbinghaus, 1885). When provided with a finite amount of study material, students who study longer score better on tests (Cooper & Pantle, 1967). As a result, study time and academic success are often considered complementary (Karpicke, Butler, & Roediger, 2009; Kornell & Bjork, 2007; Landrum, Turrisi, & Brandel, 2006). In a series of studies observing junior high, high school, and undergraduate students, Christopoulos, Rohwer, and Thomas (1987) and Dellucchi, Rohwer, and Thomas (1987) found that study time increased with grade level. This increase

in study time was attributed to an increase in the workload demands of classwork at higher grade levels (Christopoulos et al., 1987). Alternatively, in a similar set of studies, Curley, Estrin, Thomas, & Rohwer (1983) and Thomas, Iventosch, and Rohwer (1987) suggested that changes in study time can be attributed to the demands and char acteristics of the coursework. In an effort to explore the effects of coursework on study time, d’Ydewalle et al. (1983) found that students who were anticipating a more difficult test studied longer, and might have used more elaborative study strategies. Additionally, the students expecting a more difficult test performed better, regardless of the actual difficulty of the test (d’Ydewalle et al., 1983). These findings, in

*Faculty mentor

Cole and Butler | Disentangling Study Time and Study Strategy

conjunction with the understanding that more elaborate study methods take longer to carry out (Entwistle & McCune, 2004; Hilgard, Irvine, & Whipple, 1953), have led to the belief that students who study longer must be achieving a more mean ingful understanding of the studied material (i.e., d’Ydewalle et al., 1983). A meaningful understanding for the studied material is characterized by a holistic conceptual knowledge for the newly learned material (Ausubel, 2012; Mayer, 2002), which has been integrated with previously understood concepts (Ausubel, 2012; Baddeley, 2000) and is likely well-organized (Bower, 1970). One form of meaningful learning is known as elaboration (Mayer, 2002; Novak, 2002), typi cally defined as thinking about the material rather than just repeating the information over and over. Similarly, some have argued that active reading is fundamental to meaningful learning. Adler and Van Doren (1972) originated and defined the concept of active learning as applying specific strate gies, such as summarizing, criticizing, or developing and using study guides or other artifacts in an effort to comprehend, memorize, and synthesize informa tion. Since then, a number of frameworks have been offered to help learners develop good active reading approaches (e.g., Artis, 2008; Carlston, 2011; Pugh, 1978; Zhang et al., 2002). Compared to elaborate study methods, memorization results in an atomistic conceptual understanding by solidifying memory for the stud ied material (Ausubel, 2012; Mayer, 2002; Roediger & Karpicke, 2006). The atomistic nature of memo rization results in relatively isolated memories for the studied information (Novak, 2002) that are comparatively less robust than memories created using elaboration (Craik & Lockhart, 1972). The belief that meaningful learning methods require more study time (e.g., Entwistle & McCune, 2004; Hilgard et al., 1953) is at odds with more recent evidence suggesting that elaboration is actu ally more efficient than memorization (Karpicke et al., 2009; Roediger & Karpicke, 2006). Claims of elaborative efficiency are supported on two theoretical fronts: (a) The association between study time and performance can be attributed to more material being studied, but not elaborated (Christopoulos et al., 1987, Dunlosky et al., 2013); and (b) elaboration requires less repetitive mainte nance than memorization (Bobrow & Bower, 1969; Dunlonsky, Rawson, Marsh, Nathan, & Willingham, 2013; Roediger & Karpicke, 2006). The redundant nature of reading and rereading notes or other

materials is inherently lacking efficiency, and can lead to longer total study times and worse perfor mance than studying to achieve a more meaningful understanding of the material (e.g., Bower, Clark, Lesgold, & Wiznenz, 1969). Karpicke et al. (2009) has suggested that the misplaced sense of efficacy in unproductive study strategies is supported by judgments of preparedness that are based on processing fluency, rather than actual indicators of preparedness (e.g., ability to retrieve studied information). In other words, students believe they have a strong understanding of the study material when it becomes easier to read. There is no objective threshold that can be used to identify whether a concept is understood meaningfully (Bradshaw & Anderson, 1982). According to Karpicke et al. (2009), the lack of understanding among college students of what constitutes effective learning has led to a culture of students largely dependent on the use of study strategies that contribute to an overwhelming “illusion of competence” (p. 478). Purpose Contrary to the evidence described above, the prevailing zeitgeist appears to presume that increased study times lead to a more meaningful understanding of the studied material (Karpicke et al., 2009). Although increased study time is associ ated with better academic performance (Cooper & Pantle, 1967; Keith, 1982), this effect appears to be confounded with study strategy selection (e.g., Bower et al., 1969, Karpicke & Roediger, 2008). The goal of the experiments reported below was to disentangle the effects of study time and study strategy by demonstrating their respective influence on test performance.

Experiment 1 Participants read a research article then studied elaboration or memorization flashcards in prepara tion for a test. The amount of study material might not have been equivalent for elaboration and memorization groups, thus confounding study time and type of study. For example, students studying to memorize concepts and their definitions might have read fewer words than students studying materials designed to encourage elaboration. For this reason, Experiment 1 was designed to measure processing time per word (efficiency) in addition to total study time. Study time was not controlled in this experiment.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

111

Disentangling Study Time and Study Strategy | Cole and Butler

We hypothesized that: (H1) Students would study longer when the amount of study materials was greater. Thus, participants in the elaboration group would study for a longer total period of time than the participants in the memorization group. We also expected that (H 2) elaboration would be more efficient than memorization. Thus, the elaboration group was anticipated to exhibit a shorter word processing time than the memoriza tion group. Third, (H3) the elaboration group was hypothesized to exhibit a deeper understanding for the material by performing better on the test than participants in the memorization and control groups.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

112

Method Participants. The sample consisted of 97 under graduates attending Ball State University. The sample represented a diverse cross section of majors offered at the university. Participants were mostly White (84.5%), women (n = 67), between 18 and 50 years old (M age = 21.6, SD = 5.7). The rest of the sample consisted of 30 men. The ethnicities making up the rest of the sample were 2.1% African American, 1% Hispanic, 1% Native American, and 1% Middle Eastern. All participants were awarded class research credit in exchange for participating in the experiment. Materials and procedure. This experiment was conducted entirely over the Internet. Responses to online research studies have been found to be reasonably equivalent to those conducted in a laboratory setting (Butler, 1986; Whitley & Kite, 2013). For this reason, the expediency of online data collection led us to conduct the entire experi ment within an online Qualtrics survey. Participants were randomly assigned to one of three groups: memorization, elaboration, or control. Data collec tion continued until each group reached at least 30 participants. The experiment consisted of four phases: (a) read, (b) study, (c) distraction, and (d) testing. The memorization and elaboration groups carried out all four phases of the experiment, but studied different flashcards during the study phase. The control group did not participate in the study phase of the experiment. At the start of the survey, participants were presented with the informed consent. After agree ing to participate, participants were presented with a concise and detailed briefing explaining the procedure. The briefing explained that a minimum score of 75% on the test was necessary

to demonstrate a sufficient level of comprehension for the studied material. Furthermore, participants were led to believe that if they were to score below the threshold of 75% on the test, they would be asked to restudy the material and take another test. Participants were not actually held to this standard. The purpose of this deception was to provide a standard performance goal for all participants, and to motivate them to take the task seriously. The performance threshold was set at 75% because this was deemed a level that most college students believe they can achieve with a reasonable effort. After being briefed, participants began the read phase of the experiment. They were presented with an article by Mathews (2014) titled, “Hoarding Disorder: More Than Just a Problem of Too Much Stuff.” This article was chosen because there was enough detailed content to make the test difficult, but not an excessive number of technical terms that might be difficult for participants to understand. Definitions that were likely beyond a colloquial level of understanding were accompanied with a definition to improve readability. The article was 55 sentences long, totaling 1,926 words in length. The provided definitions were included in the overall wordcount. The edited article was submit ted to a readability analysis using Microsoft Word’s built-in tools. The readability analysis provided the following readability statistics: Flesch Reading Ease = 14.4; Flesch-Kinkaid Grade Level = 19.7; Passive Sentences = 21.4%. After 10 minutes, participants in the memoriza tion and elaboration groups were automatically redirected to the study phase of the experiment. After being redirected, participants were not able to navigate back to the reading material. The study material consisted of 20 flashcards covering information taken directly from the article. All flashcards consisted of two sides. Side A posed a question, and Side B provided an answer. All flashcards studied by the memorization and elaboration groups were matched for content. For the memorization group, the answer on Side B of the flashcard consisted of factual or definitive information described in the article (see Figure 1). For the elaboration group, the answer on Side B of the flashcard consisted of the same factual information provided to the memorization group, and an additional applied example (see Figure 2). The flashcards were presented one side at a time in chronological order to match the presenta tion of associated concepts in the reading material. Participants were only able to navigate forward

Cole and Butler | Disentangling Study Time and Study Strategy

through the flashcards. They read and proceeded through the flashcards at a self-determined pace. The flashcards presented to the memorization group consisted of 383 words total. The flashcards presented to the elaboration group consisted of 796 words. Once participants had been presented with each flashcard one time, they were directed to the distraction phase of the experiment. This phase of the experiment consisted of a task designed to clear working memory. The control group completed the distraction task immediately after reading the article. The distraction task consisted of 10 questions. Participants were asked to complete as many ques tions as possible in 1 minute. They were able to see the timer count down from 60 seconds while com pleting the task. Although most participants were able to complete about seven of the demographics questions, none of the participants were able to fully complete the questionnaire. After 1 minute exactly, participants were automatically redirected to the testing phase of the experiment. The test consisted of 14 questions designed to test comprehension for the informa tion provided in the reading material. Half of the test questions were designed to assess shallow (factual) understanding of the material (e.g., “What symptoms are commonly exhibited in people with hoarding disorder [HD]?”). The other half of the test questions were designed to assess deep (meaningful) understanding of the material (e.g., “Do people with obsessive compulsive disorder [OCD] respond better to SSRI medications or SNRI medications?”). Questions 1, 3, 7, 9, 11, and 13 assessed deep level knowledge. Each question was presented along with four possible answers. Participants were instructed to choose the best answer available. All questions were presented in chronological order matching how the associated concepts were introduced in the read and study phases of the experiment. Participants were given an unlimited amount of time to complete the test. Upon completing the test, participants were directed to a screen declaring that they had completed the research study. Participants were informed that there actually was no minimum score requirement. No participants were required to restudy the material or retake the test, regardless of their score on the test. At the end of the survey, participants were presented with their test score, and were able to review their performance for each question.

Results The items were initially inspected to insure reliability of the test. A binomial test showed that items 2 (p correct = .072, p < .001) and 12 (p correct = .124, p = .002) were significantly below FIGURE 1 1a. What are typical symptoms of HD?

Abbreviations ADHD: Attention Deficit Hyperactivity Disorder CBT: Cognitive Behavioral Therapy HD: Hoarding Disorder OCD: Obsessive Compulsive Disorder

1b. A desire to save and difficulty or indecision discarding items.

Abbreviations ADHD: Attention Deficit Hyperactivity Disorder CBT: Cognitive Behavioral Therapy HD: Hoarding Disorder OCD: Obsessive Compulsive Disorder

Figure 1. Memorization flashcards. Side A presented a question. Side B presented a fact. Every flashcard had all of the abbreviations listed at the bottom of the flashcard.

FIGURE 2 1a. What are typical symptoms of HD?

Abbreviations ADHD: Attention Deficit Hyperactivity Disorder CBT: Cognitive Behavioral Therapy HD: Hoarding Disorder OCD: Obsessive Compulsive Disorder

1b. A desire to save and difficulty or indecision discarding items. For example, if a person with HD received a pair of shoes that did not fit, that individual would be inclined to store that pair of shoes. For example, if a person with HD had a pair of shoes that no longer fit, that individual would have difficulty giving away or disposing of those shoes. Abbreviations ADHD: Attention Deficit Hyperactivity Disorder CBT: Cognitive Behavioral Therapy HD: Hoarding Disorder OCD: Obsessive Compulsive Disorder Figure 2. Elaboration flashcards. Side A presented the same question that was presented to the memorization group. Side B presented the same fact presented to the memorization group, and an applied example for each concept.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

113

Disentangling Study Time and Study Strategy | Cole and Butler

chance. For both questions, participants reliably selected a wrong answer that was very similar to the correct option. These two shallow items were removed from further analysis. Additionally, because all test questions were matched with specific flashcards, the flashcards associated with the removed questions were also removed from all proceeding analyses of study time and processing time. Test score. A one-way between-subjects Analysis of Variance (ANOVA) was used to assess the differ ence in the mean proportion of correct answers on the test between the groups. All assumptions of normality were met. The test indicated a difference between groups: F(2, 94) = 17.09, p < .001, ɳ2 = .267. Tukey’s pairwise comparisons showed that there was no difference between the elaboration (M = 0.58, SD = 0.17) and the memorization groups (M = 0.68, SD = 0.14), t(62) = 1.33, p = .56, Cohen’s d = 0.34, but the control group (M = 0.40, SD = 0.17) scored worse than the elaboration group, t(68) = 4.50, p < .001, Cohen’s d = 1.08, and the memorization group, t(58) = 5.62, p < .001, Cohen’s d = 1.45. Furthermore, the question types were inspected for differences. A mixed-design 2 (question type) x 3 (study strategy) factorial ANOVA was used to examine the differences between questions meant to assess shallow and deep levels of understanding, and how these differences were moderated by the assigned study method. All of the assumptions for this test were met. The interaction between ques tion type and study method yielded an F ratio of F(2, 94) = 3.55, p = .033, ɳ2p = .07. Post-hoc tests were conducted for the data presented in Figure 3 using Scheffe’s method for complex comparisons (Klockars & Hancock, 2000; Scheffe, 1970). For the deep questions, F(1, 68) = 1.95, p = .167, ɳ2 = .03, the elaboration group

Proportion Correct

FIGURE 3

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

114

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Elaboration Memorization Control

Shallow Questions

Deep Questions

Figure 3. Study strategy by question type. Both study groups did better than the control for both question types. The elaboration group did better than the memorization group on the shallow questions. Error bars are 95% confidence intervals.

(M = 0.48, SD = 0.16) was not able to outperform the memorization group (M = 0.51, SD = 0.14). However, for shallow questions, F(1, 68) = 4.85, p = .03, ɳ2 = .07, the elaboration group (M = 0.8, SD = 0.2) outperformed the memorization group (M = 0.62, SD = 0.26). There was no difference between the control group for the shallow ques tions (M = 0.45, SD = 0.25) and the control group for the deep questions (M = 0.32, SD = 0.18): F(1, 68) = 1.25, p = .268, ɳ2 = .02. Study time. All study time calculations consisted only of data from the memorization and elabora tion groups (the control group did not study). Study time was defined as the total amount of time spent reviewing the flashcards. The study time data was collected automatically using the meta data function in Qualtrics. Study time was measured in seconds (s). A between-subjects t test was used to assess the differences in study time between the two groups. Levene’s test was not significant, F(1, 68) = 3.19, p = .08, indicating that the assumption of homogeneity of variances was met, but there was a positive skew in the distribution. A natural log transformation of the data resulted in a normal distribution of the results. The t test indicated that the elaboration group (M = 370.49 s, SD = 228.80 s) studied longer than the memorization group (M = 249.52 s, SD = 154.84 s), t(68) = 2.52, p = .014, Cohen’s d = 0.603. Processing time was measured in seconds per word (spw). Processing time was calculated by dividing study time by the number of words in the study material. A between-subjects t test was used to assess the difference in processing time between the two groups. Levene’s test was not sig nificant, F(1, 68) = 2.74, p = .10, indicating that the assumption of homogeneity of variances was met, but there was a positive skew in the distribution. A natural log transformation of the data resulted in a normal distribution. The t test indicated that the memorization group (M = 0.78 spw, SD = 0.47 spw) spent more time processing the information than did the elaboration group (M = 0.52 spw, SD = 0.34): t(68) = 2.96, p = .004, Cohen’s d = 0.71. Although the elaboration group spent more total time studying, the memorization group spent more time per word processing the study material. See Figure 4 for a graphical comparison of study time and processing time measurements for the two groups.

Cole and Butler | Disentangling Study Time and Study Strategy

In Experiment 1, the effects of total time and pro cessing time were confounded with study strategy. However, based on the results in Experiment 1, we could estimate reasonable amounts of study time students would use to prepare for a test of the materials. With that information, Experiment 2 controlled study time and processing time in order to examine them independently. Experiment 2 used the same reading material, flashcards, distraction task, and test questions as used in Experiment 1. However, study time was controlled. Participants in the elaboration and memorization conditions were assigned to a brief (7.5 minutes) or extended (15 minutes) study period. Also, processing time was controlled by timing the presentation of flashcard slides. We hypothesized that (H1) participants in the elaboration condition would perform better on the test than participants assigned to the memorization condition. We also expected that (H2) participants

Method Participants. The sample consisted of 121 under graduates from the same university as Experiment 1. Two participants in this experiment were removed because their scores on the comprehen sion test were significantly below chance (12.5%). These were the only participants in either of the experiments to score below 25%, indicating that they might not have studied, or might not have understood the task. The sample represented a cross section of majors offered at the university. Participants were mostly White (84.6%) women (n = 92) between 18 and 37 years old (M age = 20.8, SD = 2.9). The rest of the sample consisted of 29 men, and the ethnicities making up the rest of the sample were 4.1% African American, 1.7% Middle Eastern, and 0.8% Hispanic. Anyone who participated in Experiment 1 was not eligible to participate in Experiment 2. Participants were awarded class credit for participation. Materials and procedure. As with Experiment 1, the entire procedure was conducted entirely within an online Qualtrics survey. Experiment 2 consisted of the same four phases completed in Experiment 1: (a) read, (b) study, (c) distraction, and (d) testing. The same materials used in Experiment 1 were used in Experiment 2. Participants were randomly assigned to study flashcards meant to promote memorization or elaboration for a brief (7.5 minutes) or extended (15 minutes) study period. Data collection continued until each group contained at least 30 participants. FIGURE 4 400

Study Time

Processing Time

360 320 280 240 200

Elaboration

Memorization

0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40

Processing Time (spw)

Experiment 2

in the extended study condition would score bet ter than participants in the brief study condition. Third, we hypothesized that (H3) there would be no interaction between study time and study method.

Study Time(s)

Discussion The results supported our hypotheses that the elaboration group would study longer and process the study materials more efficiently than the memorization group. These results uphold the general understanding that, although elaboration takes more time to carry out than memorization (Hilgard et al., 1953), elaboration is more efficient (Karpicke et al., 2009; Roediger & Karpicke, 2006). The elaboration group outperformed the control group, but not the memorization group. The elabo ration group scored better than the memorization group on the shallow questions, but surprisingly, not the deep questions. The findings that the elaboration group did not perform better than the memorization group on deep questions is in contrast with a plethora of literature that has demonstrated the superior ity of meaningful learning when compared to memorization (Ausubel, 2012; Bower et al., 1969; Bradshaw & Anderson, 1982; Craik & Tulving, 1975; Karpicke & Blunt, 2011; Mayer, 2002). Because students tend to study until they reach a level of understanding that they believe will allow them to achieve a desired level of performance (LaPorte & Nath, 1976), the lack of control over study time could have contributed to the lack of a difference between study strategies. It is also worth noting that study time and processing time exhibited an inverse relationship. A second experiment was needed to further delineate the individual effects of study time and processing time on test performance.

Figure 4. Study time vs. processing time. The elaboration group studied for a longer period of time than the memorization group and spent less time processing the study materials. Error bars are 95% confidence intervals.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

115

Disentangling Study Time and Study Strategy | Cole and Butler

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

116

The read phase of Experiment 2 did not change from Experiment 1. The study phase for Experiment 1 differed from Experiment 2 in that the flashcards were presented for a controlled amount of time by the Qualtrics program. To make sure that participants were attending to the flashcards, they were required to check in at random intervals during the presentation of the study material. A check-in trial simply consisted of a flashcard explaining that participants must click forward to continue studying. The length of time each flashcard was pre sented was determined by the number of words on the flashcard (i.e., processing time). Processing time was determined by calculating the amount of time it took the elaboration group in Experiment 1 to read the flashcards. The elaboration group was used as the baseline for total study time because they had the shorter processing time in Experiment 1. Hence, this group provided a more accurate measure of what was likely the amount of time necessary for participants to successfully read and comprehend the flashcards. A mean processing time of .50 spw was used as a baseline for determining the study time for the brief study condition. To ensure that none of the participants would have trouble reading the mate rial, a 1-minute buffer was added to the total study time. This method resulted in a study time of 7.5 minutes (0.57 spw) for the brief study condition. A reading time of 0.57 spw amounts to a read ing pace of 105.2 words per minute (wpm). Given that college students generally read at a pace of about 300 wpm (Carver, 1992; Taylor, 1965), the 7.5-minute study time was not expected to affect participants’ ability to read and comprehend the flashcards. Although the brief study condition was likely able to read all of the flashcards without dif ficulty, the 7.5-minute study time limit was expected to limit additional processing of the material that might occur beyond control of the experiment. The study time for the extended study condi tion was determined by doubling the study time allowed for the brief study condition. The extended study condition was given 15 total minutes to study the flashcards. This amounted to a processing time of 1.14 spw. The flashcards for the elaboration condition contained a total of 413 more words than the flashcards for the memorization condition. For the brief study condition, this difference in volume of study material resulted in a total study time for the elaboration condition that was 235.41 s longer than the total study time for the memorization

condition. For the extended study condition, this difference in volume of study material resulted in a total study time for the elaboration condition that was 470.82 s longer than the total study time for the memorization condition. At the beginning of the study phase, partici pants in the memorization condition completed an additional puzzle task to account for the total time difference between conditions. This allowed for both of the study strategy conditions to have the same amount of time between (a) reading the article and taking the test and (b) studying the flashcards and taking the test. The puzzle task required participants to complete a difficult word search entitled, “Getting to Know Indiana.” The word search had 50 hidden words all centered around the state of Indiana. This theme was chosen because there was no information that would likely interfere with the information presented in the article. When the time ran out on the puzzle task, participants were automatically redirected to the memorization flashcards. None of the participants were able to complete the entire word search before the time ran out. Upon completion of the study phase, par ticipants were automatically redirected to the distraction phase of the experiment. The distrac tion task consisted of the same questions used in the distraction task for Experiment 1. After participants completed as many questions as possible in 1 minute, they were automatically directed to the testing phase. The testing phase consisted of the same test that was administered to participants in Experiment 1. As in Experiment 1, participants were informed during the briefing that if they did not score a 75% on the test they would be required to restudy the material and retake the test. No participants were actually held to this standard. When participants completed the test, they were redirected to a screen that informed them that they had successfully completed the study and thanked them for their participation. Participants were able to see their test score and review their performance after complet ing the entire survey. Results As with Experiment 1, each test item was inspected for reliability. The binomial test followed the same pattern as Experiment 1: Items 2 (p correct = .125, p < .001) and 12 (p correct = .133, p = .001) were significantly below chance. Once again these items were removed from the proceeding analyses of test performance.

Cole and Butler | Disentangling Study Time and Study Strategy

Discussion The results supported our hypothesis that, when processing time was controlled, increasing total study time would result in better test scores. The results also supported our hypothesis that the elaboration condition would score better on the test than the memorization condition. Additionally,

the results supported our hypothesis that there would be no interaction between study time and study method. The findings from this study support a large literature indicating that elaboration leads to better test performance than memorization (e.g., Adler & Van Doren, 1972; Artis, 2008; Bower, 1970; Bower et al., 1969; Carlston, 2011; Karpicke & Roediger, 2008), and that more study time leads to better test performance (e.g., Cooper & Pantle, 1967; d’Ydewalle, Swerts, & Corte, 1983; Ebbinghaus, 1885; Karpicke, Butler, & Roediger, 2009; Keith, 1982; Kornell & Bjork, 2007; Landrum et al., 2006). The specific contribution in this experiment is that the study time and study strategy factors operate independently and are about equal in impact on test performance.

General Discussion Previous research has concluded that study time is only effective to the extent that elaborative study methods are utilized (Cooper & Pantle, 1967). This sentiment has been echoed by Bower et al. (1969), Craik and Lockhart (1972), and Roediger and Karpicke (2006), all of whom concluded that test performance is largely based on the applied FIGURE 5

Proportion Correct

0.8

Extended

Brief

0.7 0.6 0.5 0.4

Elaboration

Memorization

Figure 5. The effect of study strategy on test performance acted independently of study time. Error bars are 95% confidence intervals.

FIGURE 6

Proportion Correct

Test scores. A between-subjects 2 (study time) x 2 (study strategy) factorial ANOVA was used to assess the test performance of the four different conditions. Because study time was dictated by the predetermined processing time coefficient, and the total number of words varied systematically based on study strategy, a Weighted Least Squares (WLS) correction was used to control for the difference in total study time between the elaboration and memo rization conditions. The WLS correction applied weights to the data based on total study time. All of the assumptions of the ANOVA were met. The main effect for study time yielded an F ratio of F(1, 117) = 13.42, p < .001, ɳ2p = .103, indicat ing that the extended study condition (M = 0.63, SD = 0.13) outperformed the brief study condition (M = 0.53, SD = 0.13) on the test. The main effect for study strategy yielded an F ratio of F(1, 117) = 13.48, p < .001, ɳ2p = .103, indicating that the elaboration condition (M = 0.62, SD = 0.14) out performed the memorization condition (M = 0.53, SD = 0.13) on the test. As can be seen in Figure 5, there clearly was no interaction between study time and study strategy: F(1, 117) = 0.06, p = .812, ɳ2p = .000). As in Experiment 1, the question types were inspected for differences. A mixed-design 2 (ques tion type) x 2 (study strategy) factorial ANOVA was used to assess the differences between the deep and shallow questions, and how these differences were moderated by the assigned study method. Levene’s test indicated that the assumption for homogene ity of variances was not met, F(7, 1432) = 52.64, p < .001. To account for the lack of normality, the main effects were assessed using the KruskalWallis test. The main effect for question type, H(1) = 44.03, p < .001, indicated that the shallow questions (M = 0.68, SD = 0.46) were easier than the deep questions (M = 0.5, SD = 0.5). As can be seen in Figure 6, there were no interactions between the question type and study strategy, F(1, 116) = 1.26, p = .26, ɳ2p = .01, study time, F (1, 116) = 1.77, p = .186, ɳ2p = .02, or for the interaction between study time, study method, and test question type, F(1, 116) = 0.13, p = .72, ɳ2p = .00.

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Elaboration Memorization

Extended Brief Shallow Questions

Extended Brief Deep Questions

Figure 6. Study strategy by study time by question time. The shallow questions were easier than the deep questions. Error bars are 95% confidence intervals.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

117

Disentangling Study Time and Study Strategy | Cole and Butler

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

118

depth of processing. The current study posits that processing time is a third variable bridging the effects of study time and study strategy. When pro cessing time is not controlled, the effects of study time and study strategy become entangled. When processing time is controlled, study time and study strategy are individuated. Given the systematic differences in total study time and processing time between the two study methods, the adjustments made by the students were likely made in reference to judgments of processing fluency (Karpicke et al., 2009). In this light, the results from Experiment 1 suggest that participants found the elaboration flashcards easier to read. The improved readability of elaborative study materials appears to provide meaningful learning strategies with an efficiency advantage over memorization (e.g., Bower et al., 1969; Karpicke & Roediger, 2008; Kember et al., 1995). The shallow questions were easier than the deep questions, but the elaboration condition did not outperform the memorization condition on either of the question types. These findings sug gest that studying the elaboration flashcards did not necessarily result in a more meaningful level of understanding than studying the memorization flashcards. If the participants were using the same strategy to study both sets of flashcards, information gained from the applied examples provided in the elaboration condition might have failed to translate to the conceptual comparisons assessed in the deep questions. In Experiment 1, the provided examples promoted the elaboration condition’s memory for factual information. Although the results for Experiment 2 followed this general trend, the elaborative advantage was not related to question type. From this, we can conclude that the elabora tions in this study contributed to strengthening the memory for the studied information, but might not have provided a platform for exhibiting a more holistic understanding of the studied material (e.g., Ausubel, 2012; Mayer, 2002). Additionally, the study materials for the elabo ration groups contained only one elaboration per studied concept. In Experiment 1, this proved to be too weak of a manipulation to differentiate the effects of the two study strategies. An increase in the difference of the number of elaborations between the groups might have improved the power of the manipulation. Alternatively, the lack of control over study time might have contributed variance to the results that could not be accounted for by controlling the study strategy.

Although this study observed the differences in elaborative and memorization study strategies, there was no measure of what mental strategies the participants were actually implementing. As such, it is possible that participants in either group could have developed their own elaborations, or simply attempted to memorize all of the material. The differences between the processing times for elaboration and memorization groups suggest that the study methods were reliably different. Future research could clarify these differences by having participants record their study activity, or report the strategies they use. Another important factor to consider is control over the volume of the provided study material and processing time. Experiment 2 was only able to apply a secondary statistical control to account for differences in processing time between conditions. For the manipulation to control for processing time, the amount of material being studied by the groups being compared must be equal. This can be difficult to accomplish without diluting the less meaningful material with unrelated information, or making the material difficult to read. In other paradigms, this issue of control has been managed by having participants study qualitatively different word lists (e.g., Bower, 1969; Karpicke & Roediger, 2008). This method serves as a natural control for the third variable of processing time, but also does not allow for the differentiation between study time and processing time. That being said, although it may be difficult to accomplish, future studies could attempt to develop study materials that are equiva lent in the amount of information being processed. Although a variety of study methods can be considered elaborative or meaningful (e.g., Ausubel, 2012; Mayer, 2002), this study only focused on one form of elaboration (i.e., applied examples). Further confirmation is necessary to establish that the findings associated with one type of elaboration can be generalized to every type of elaboration. One issue that the current research does not address is when it may be best to implement the implications of this study with students. Previous research (e.g., Christopoulos et al., 1987; Dellucchi et al., 1987) has reported that students increase study time as they progress in their formal educa tion from middle school into college. However, a number of researchers have found that the strate gies used by these students do not seem to evolve much. The research here suggests that at some point(s) students should be strongly encouraged to adopt more effective learning/study strategies.

Cole and Butler | Disentangling Study Time and Study Strategy

This may require both explicit instruction and the use of assigned tasks that are difficult for students to complete without using elaborative strategies. The current research did not address when this should be done. Conclusions The purpose of this study was to determine the independent roles of study time and study method on test performance. Our results showed that elaboration was a more efficient study strategy than memorization. Furthermore, this study dem onstrated that the effects of study time and study strategy exhibited a separate, but relatively equal, influence on test performance. Study time is an important factor relating to academic performance, but how that time is spent is generally considered more important for predict ing academic performance (Delucchi, Rohwer, & Thomas, 1987; Kember et al., 1995; Schuman et al., 1985). Although there appears to be some awareness among college students that some study strategies are more effective than others (Tang, 1994), students continue to rely heavily on the use of inefficient and unproductive memorization strategies (Christopoulos et al., 1987; Entwistle & McCune, 2004; Karpicke et al., 2009). In some cases, this may be a product of the inability of undergraduates to effectively execute these study strategies (Tang, 1994). In other cases, the exhib ited naivety of these students may be propagated by a widespread “illusion of competence” (Karpicke et al., 2009, p. 478) when memorizing study material. The most effective study strategies take advan tage of the effects of study time and study method. To maximize efficiency and develop a more in-depth understanding of the learning material, students should focus on implementing meaningful learning strategies such as elaboration. What is still unclear is at what point students should be strongly encouraged to adopt more effective learning/study strategies.

References Adler, M. J., & Van Doren, C. (1972). How to read a book: The classic guide to intelligent reading. New York, NY: Simon and Schuster. Artis, A. B. (2008). Improving marketing students’ reading comprehension with the SQ3R method. Journal of Marketing Education, 30, 130–137. https://doi.org/10.1177/0273475308318070 Ausubel, D. P. (2012). The acquisition and retention of knowledge: A cognitive view. https://doi.org/10.1007/978-94-015-9454-7 Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 11, 417–423. https://doi.org/10.1016/s1364-6613(00)01538-2 Bobrow, S. A., & Bower, G. H. (1969). Comprehension and recall of sentences. Journal of Experimental Psychology, 80, 455–461. https://doi.org/10.1037/h0027461

Bower, G. H. (1970). Organizational factors in memory. Cognitive Psychology, 1, 18–46. https://doi.org/10.1016/0010-0285(70)90003-4 Bower, G. H., Clark, M. C., Lesgold, A. M., & Wiznenz, D. (1969). Hierarchical retrieval schemes in recall of categorized word lists. Journal of Verbal Learning and Behavior, 8(3), 323–343. https://doi.org/10.1016/s0022-5371(69)80124-6 Bradshaw, G. L., & Anderson, J. R. (1982). Elaborative encoding as an explanation of levels of processing. Journal of Verbal Learning and Verbal Behavior, 21, 165–174. https://doi.org/10.1016/s0022-5371(82)90531-x Butler, D. L. (1986). Automation of instructions in human experiments. Perceptual and Motor Skills, 63, 435–440. https://doi.org/10.2466/pms.1986.63.2.435 Carlston, D. L. (2011). Benefits of student-generated note packets a preliminary investigation of SQ3R implementation. Teaching of Psychology, 38, 142–146. https://doi.org/10.1177/0098628311411786 Christopoulos, J. P., Rohwer, W. D., & Thomas, J. W. (1987). Grade level differences in students’ study activities as a function of course characteristics. Contemporary Educational Psychology, 12, 303–323. https://doi.org/10.1016/s0361-476x(87)80003-6 Cooper, E. H., & Pantle, A. J. (1967). The total-time hypothesis in verbal learning. Psychological Bulletin, 68, 221–234. https://doi.org/10.1037/h0025052 Carver, R. P. (1992). Reading rate: Theory, research, and practical implications. Journal of Reading, 36, 84–95. Retrieved from https://www.jstor.org/stable/pdf/40016440.pdf. Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671–684. https://doi.org/10.1016/s0022-5371(72)80001-x Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology, 104, 268–294. https://doi.org/10.1037/0096-3445.104.3.268 Curley, R. G., Estrin, E. T., Thomas, J. W., & Rohwer, W. D. (1987). Relationships between study activities and achievement as a function of grade level and course characteristics. Contemporary Educational Psychology, 12, 324–343. https://doi.org/10.1016/s0361-476x(87)80004-8 Delucchi, J. J., Rohwer, W. D., & Thomas, J. W. (1987). Study time allocation as a function of grade level and course characteristics. Contemporary Educational Psychology, 12, 365–380. https://doi.org/10.1016/s0361-476x(87)80006-1 d’Ydewalle, G., Swerts, A., & Corte, E. D. (1983). Study time and test performance as a function of test expectations. Contemporary Educational Psychology, 8, 55–67. https://doi.org/10.1016/0361-476x(83)90034-6 Dunlonsky, J. Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14, 4–58. https://doi.org/10.1177/1529100612453266 Ebbinghaus, H. (1964). Memory: A contribution to experimental psychology. https://doi.org/10.1037/10011-000 Entwistle, N., & McCune, V. (2004). The conceptual bases of study strategy inventories. Educational Psychology Review, 16, 330–345. https://doi.org/10.1007/s10648-004-0003-0 Hilgard, E. R., Irvine, R. P., & Whipple, J. E. (1953). Rote memorization, understanding, and transfer: An extension of Katona’s card-trick experiments. Journal of Experimental Psychology, 46, 288–292. https://doi.org/10.1037/h0062072 Karpicke, J. D., & Blunt, J. R. (2011). Retrieval practice produces more learning than elaborative studying with concept mapping. Science, 331, 772–775. https://doi.org/10.1126/science.1199327 Karpicke, J. D., Butler, A. C., & Roediger, H. L. (2009). Metacognitive strategies in student learning: Do students practice retrieval when they study on their own? Memory, 17, 471–479. https://doi.org/10.1080/09658210802647009 Karpicke, J. D., & Roediger, H. L. (2008). The critical importance of retrieval for learning. Science, 319, 966–968. https://doi.org/10.1126/science.1152408 Keith, T. Z. (1982). Time spent on homework and high school grades: A largesample path analysis. Journal of Educational Psychology, 74, 248–303. https://doi.org/10.1037/0022-0663.74.2.248 Kember, D., Jamieson, Q. W., Pomfret, M., & Wong, E. T. T. (1995). Learning approaches, study time and academic performance. Higher Education, 29, 329–343. https://doi.org/10.1007/bf01384497 Klockars, A. J., & Hancock, G. R. (2000). Scheffe’s more powerful F-protected post hoc procedure. Journal of Educational and Behavioral Statistics, 25, 13–19. https://doi.org/10.2307/1165310

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

119

Disentangling Study Time and Study Strategy | Cole and Butler

Kornell, N., & Bjork, R. A. (2007). The promise and perils of self-regulated study. Psychonomic Bulletin and Review, 14, 219–224. https://doi.org/10.3758/bf03194055 Landrum, R. E., Turrisi, R., & Brandel, J. M. (2006). College students’ study time: Course level, time of semester, and grade earned. Psychological Reports, 98, 675–682. https://doi.org/10.2466/pr0.98.3.675-682 LaPorte, R. E., & Nath, R. (1976). Role of performance goals in prose learning. Journal of Educational Psychology, 68, 260–264. https://doi.org/10.1037/0022-0663.68.3.260 Mathews, C. A. (2014). Hoarding disorder: More than just a problem of too much stuff. Journal of Clinical Psychiatry, 75, 893–894. https://doi.org/10.4088/jcp.14ac09325 Mayer, R. E. (2002). Rote versus meaningful learning. Theory Into Practice, 41, 226–232. https://doi.org/10.1207/s15430421tip4104_4 Novak, J. D. (2002). Meaningful learning: The essential factor for conceptual change in limited or inappropriate propositional hierarchies leading to empowerment of learners. Science Education, 86, 548–571. https://doi.org/10.1002/sce.10032 Pugh, A. K. (1978). Silent reading: An introduction to its study and teaching. Heinemann Educational. Retrieved from https://www.jstor.org/stable/pdf/40033218.pdf Roediger, H. L., & Karpicke, J. D. (2006). The power of testing memory: Basic research and implications for educational practice. Perspectives on Psychological Science, 1, 181–201. https://doi.org/10.1111/j.1745-6916.2006.00012.x Scheffe, H. (1970). Multiple comparison testing versus multiple estimation. Improper confidence sets. Estimation of directions and ratios. The Annals of Mathematical Statistics, 41, 1–29. https://doi.org/10.1214/aoms/1177697184 Schuman, H., Walsh, E., Olson, C., & Etheridge, B. (1985). Effort and reward: The

assumption that college grades are affected by quantity of study. Social Forces, 63, 945–966. https://doi.org/10.2307/2578600 Tang, C. (1994). Effects of modes of assessment on students’ preparation strategies. In G. Gibbs (Ed.), Improving student learning: Theory and practice (pp. 151–170). Oxford: Oxford Centre for Staff Development. Taylor, S. E. (1965). Eye movements in reading: Facts and fallacies. American Educational Research Journal, 2, 187–202. https://doi.org/10.3102/00028312002004187 Thomas, J. W., Iventosch, L., & Rohwer, W. D. (1987). Relationships among student characteristics, study activities, and achievement as a function of course characteristics. Contemporary Educational Psychology, 12, 344–364. https://doi.org/10.1016/s0361-476x(87)80005-x Whitley, B. E., & Kite, M. E. (2013). Principles of research in behavioral science (Vol. 3). https://doi.org/10.4324/9781315450087 Zhang, G., Cheng, Z., Huang, T., He, A., & Koyama, A. (2002). Design of an effective learning method SQ3R based distance learning system. In Cyber Worlds, 2002. Proceedings. First International Symposium (pp. 318–322). https://doi.org/10.1109/CW.2002.1180896 Author Note. Zachary J. Cole, https://orcid.org/0000-00020692-5739, Ball State University; and Darrell L. Butler, https://orcid.org/0000-0001-7688-2861, Ball State University. Special thanks to Ball State University professors Dr. Kristin Ritchey and Dr. Michael Tagler for the helpful comments and reviews. Correspondence concerning this article should be addressed to Zachary Cole, Department of Psychology, University of Nebraska-Lincoln, 325 Burnett Hall, Lincoln, NE 68508.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

120

https://doi.org/10.24839/2325-7342.JN25.2.121

An Examination of the Influence of Serial Position on False Memory and Recognition Josephine Audiffred and Carissa L. Broadbridge* Saint Xavier University

ABSTRACT. The Deese-Roediger-McDermott paradigm has long shown that individuals will falsely recall a critical lure (Deese, 1959; Roediger & McDermott, 1995). Roediger and McDermott (1995) attributed this false recall to activation and monitoring. During encoding, individuals activate information related to the critical lure, then monitoring processes during retrieval fail to identify that the critical lure was activated internally, leading to recall of the critical lure. In the present study, we examined the effect of the serial position of the strongest associates on false recall and recognition of the critical lure. Using a within-subjects design, participants were exposed to 10 lists of words via MediaLab. Each list placed highly associated words at either the beginning, middle, or end of the list. After each list, participants completed a recall task followed by a recognition task. Results showed that participants recalled the critical lure significantly more often when high associates were at the beginning of the list (p = .002, η 2 = .07), but serial position had no effect on recognition (p = .13, η2 = .03). Recognition levels showed that, regardless of the serial position of the strongest associates, participants recognized the critical lure at similar proportions. High recall of the critical lure when strong associates were at the beginning of the list can be attributed to the higher likelihood of activation spreading to the lure when the participant is studying the list. Furthermore, there is a decreased likelihood of accurate source monitoring due to the early activation of the lure during study. Keywords: false memory, DRM paradigm, serial position, recognition, recall

alse memory is a widely studied aspect of memory. Individuals could remember an event very differently than how it actually occurred, they may recall an experience that did not truly happen, or they may remember words from a list that were not there at all. The most common method used to study false memory for words from a list is the Deese-Roediger-McDermott (DRM) paradigm. This method was first used by Deese (1959) who developed 12 word lists with items that were strongly associated with an unpresented item—the critical lure—and presented the lists to undergraduate students. The objective was to understand why individuals were

*Faculty mentor

likely to remember an item that was not presented in the learned list, and he found that different lists produced different rates of false recall. His study provided the groundwork for the DRM paradigm that would evolve and spark a plethora of further research (e.g., Arndt, 2012; Beato, Cavadid, Pulido, & Pinho, 2012; Gallo & Roediger, 2002; Roediger & McDermott, 1995; Roediger, Watson, McDermott & Gallo, 2001). Intrigued by the results of Deese’s study, Roediger and McDermott (1995) designed two experiments to replicate and expand on the work of Deese. In their first study, they used six lists from Deese (1959) that had the highest false recall rates.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

121

Serial Position and False Memory | Audiffred and Broadbridge

Researchers read each list to participants, then asked participants to recall the last words they heard first, followed by whichever other words they remembered in any order. Once they had recalled all six lists, participants were given a recognition test containing all six critical lures as well as studied and unstudied words. They were also asked to rate how confident they were that they had seen each word during the study period. Results showed that participants recalled the critical lure 40% of the time and reported being sure that the critical lures had been studied (Roediger & McDermott, 1995). In their second experiment, Roediger and McDermott (1995) had participants view 16 lists. Participants completed a recall test immediately following half the lists and completed a distractor task following the other half. After viewing all of the lists, each participant was given a recognition test that included words from the 16 studied lists and from an additional 8 lists that were not studied. They were asked to rate words as old or new and as remembered or known (Roediger & McDermott, 1995). This remember-know distinction was first recognized by Tulving (1985). He argued that remembering involves autonoetic consciousness, or a true sense of having experienced the event. This is a more complex type of recall than knowing. Knowing involves only the belief that something was experienced (Rajaram, 1993). This can be exemplified by people’s experience of their birth day. Most people know when their birthday is, but very few individuals claim to remember being born. Roediger and McDermott (1995) found that false recall of the critical lure was present 55% of the time, an increase from the previous experiment. During the recognition test, participants also showed a higher probability of remembering the critical lure compared to knowing it was on the list (Roediger & McDermott, 1995). The results of their experiment were so surprising that the procedure became popular among false memory researchers, and work in this area has flourished.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

122

Factors Affecting False Memory Backward associative strength. Numerous factors (e.g., providing participants with information about list-learning paradigms, manipulating the position of strong associates within the study list, manipulating participants’ attention, and indicat ing that there is a critical lure) have been found to influence false memory (Prohaska, DelValle, Toglia, & Pittman, 2016). One of the most common factors examined across the literature is the backward associative strength (BAS), the level of association

between list items and the critical lure in a DRM list (e.g., Hicks & Hancock, 2002; Roediger et al., 2001; Stadler, Roediger, & McDermott, 1999). BAS plays a vital role in the false memory of a critical lure. Research has suggested that false recall of a critical lure is more prevalent when lists contain associates with higher BAS values (e.g., Gallo & Roediger, 2002; Hicks & Hancock, 2002). For instance, Hicks and Hancock (2002) conducted a study where DRM lists were split in half and the average BAS values for one list was higher than the other. Results showed that the lists that had the higher BAS average also had a higher rate of false recall compared to the lower BAS averaged lists. Gallo and Roediger (2002) also examined the BAS of words in the DRM lists. They found that participants had lower false recall rates for lists with lower mean BAS values than those with higher mean BAS values (Gallo & Roediger, 2002). Serial position. Serial position of the words with the highest BAS values may also affect the likelihood of the critical lure being recalled. Serial position has been shown to affect recall of words in simple list learning tasks (Ebbinghaus, 1948; Murdock, 1962). This serial position effect is the tendency to recall the first and last items in the list the best, whereas the items presented in the middle are least likely to be recalled. Ebbinghaus (1948) was the first researcher to discover the serial position effect. He referred to the improved recall for the last words in a list as the recency effect and the improved recall of the words at the beginning of the list as the primacy effect (Ebbinghaus, 1948). Other researchers have replicated and expanded on the work of Ebbinghaus. For instance, Murdock (1962) investigated the serial position effect using lists that varied in length and presentation time. His results showed that, regardless of list length or presentation time, participants were more likely to show a primacy effect for the first 3–4 words in a list, and a distinctive s-shaped curve depicted the recency effect over the last 8 words. Items in the middle of the list were not remembered as well. Murdock suggested that this may be due to proac tive and retroactive inhibition occurring within the list. Interestingly, not many studies have explicitly examined the effect of serial position on false recall in the DRM paradigm. However, Roediger and McDermott (1995) did note that a strong primacy effect was present when the words with the strongest BAS values had occurred early in the lists. In the present study, we sought to explicitly examine the influence of serial position on false memory for a critical lure.

Audiffred and Broadbridge | Serial Position and False Memory

Theories of False Memory Although various theories have tried to explain the phenomenon of false memory, the two most promi nent theories are the activation monitoring theory and fuzzy trace theory. The activation monitoring theory argues that the probability of false recall is determined by two processes, spreading activation and monitoring, that occur during encoding and retrieval of information (Roediger & McDermott, 1995). On the other hand, the fuzzy trace theory (Brainerd & Reyna, 2002) argues that memory traces are stored through gist traces (commonali ties among events) and verbatim traces (specific perceptual details of an experience used to help differentiate memories). The following sections describe each of these theories. Activation monitoring theory. The activation monitoring theory is one of the primary theories explaining the false memory phenomenon (Gallo & Roediger, 2002; Roediger & McDermott 1995; Roediger et al., 2001). This theory focuses on the processes of spreading activation and monitor ing (Roediger & McDermott, 1995). Spreading activation occurs when semantic networks become activated during search processes. When one concept is activated, this activation spreads to other, related concepts, with the most highly associated concepts being activated first followed by more weakly associated concepts (Roediger & McDermott, 1995). Then, during search for previously viewed items, the monitoring process determines the origin of activation in order to determine the authenticity of the memory, and subsequently acceptance or rejection of the lure (Ardnt, 2012; Gallo, 2010; Roediger, Balota & Watson, 2001). This theory helps explain the recognition of the critical lure during the DRM paradigm because an implicit associative response is created, and source monitoring fails to declare it as originating internally. This then leads to the lure being falsely remembered as present in the studied list (Roediger & McDermott, 1995). Fuzzy trace theory. The other prominent theory of false memory, fuzzy trace theory, was first used as a model to explain the relationship between the validity of solutions to reasoning problems and memory for background events (Brainerd & Reyna, 2002). Brainerd and Reyna explained the theory using five distinct principles, the first three of which are relevant to false memory. The first principle, parallel storage, describes each memory trace spe cifically with verbatim representations of the surface form of experienced items and gist representations

that are interpretations of the experienced items. For example, the word peach would have the surface form of how the word is spelled and the gist form of fruit. Principle two, differential retrieval, dictates that verbatim and gist traces are retrieved through different mechanisms. Verbatim traces are better retrieved when items are experienced and if the situation shows verbatim traces to be stronger than the gist traces. For gist traces, retrieval is easier for nonexperienced, highly meaningful items and in situations where gist traces are instead stronger than verbatim. The third principle, dual-opponent processes, explains how verbatim and gist traces work together to support true memories, but work against each other in the case of false memories. According to this principle, the memory for the critical lure would be supported by the gist trace, but the verbatim trace would work against the gist trace by suppressing the critical lure. Verbatim traces, on the other hand, would only single out the exact words that were shown, thereby inhibiting false memory (Brainerd & Reyna, 2002). The Present Study In the present study, we manipulated the order of items in each list based on the BAS values in order to compare the two prominent theories for false memory. We used 10 of Roediger et al.’s (2001) word lists along with their critical lures and BAS val ues. Highly associated items were grouped together and placed either at the beginning, middle, or end of a list. Our objective was to investigate whether serial position of highly associated words affects the probability of false recall and recognition of a critical lure. In line with the activation monitoring theory, we hypothesized that participants would falsely recall and recognize the critical lure more often if the high BAS words were positioned at the begin ning or end of the list, rather than in the middle. Activation monitoring theory states that false recall is caused by items activating a semantic network of associations, thus activating the target word, and by the individual’s failure to accurately monitor the source of the information, which then leads them to believe that they viewed the target word before (Roediger & McDermott, 1995). Additionally, words at the beginning (primacy) and end (recency) of a list have been shown to have higher recall rates (Ebbinghaus, 1948; Murdock, 1962). Activating the critical lure at these high recall points in the list through placement of high BAS value words increases the likelihood that the source monitoring process will fail and that the critical lure will be

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

123

Serial Position and False Memory | Audiffred and Broadbridge

recalled and/or recognized. Alternatively, recall/recognition of the criti cal lure could be greater when high associates are placed at the beginning of the list than at the end of the list. Such results would support the fuzzy trace theory because this would suggest that verbatim traces are stronger than gist traces when the high associates are presented last but that gist traces are stronger when the high associates are presented first.

Method Participants Participants were 87 undergraduate students between the ages of 18 and 26 (M = 18.90, SD = 1.12; 83% women, 17% men) who were enrolled in psychology classes at a private Midwestern university. Participants were ethnically diverse (33% European American, 43% Hispanic/ Latinx, 12% Arab/Middle Eastern, and 13% other) and received 2 credits of research participation for completing the study. There were no exclusion criteria for participants.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

124

Materials and Procedures Following approval from the Institutional Review Board at Saint Xavier University (approval #FA18014AP0910), we recruited participants through SONA, an online research participation system. At their scheduled session, participants read and signed the consent form. Participants then completed a demographic questionnaire that asked for their age, sex, and ethnicity. Finally, they were presented with the word lists followed by the recall and recognition tasks. Each word list, taken from the compiled lists used by Roediger et al. (2001) [three of which were also used by Stadler et al. (1999)], was based on a specific critical lure. A total of 10 lists, each contain ing 15 words, were chosen. The critical lures for these lists were king, doctor, bread, city, mountain, trash, sweet, needle, slow, and window. These lists were chosen based on their inclusion of words varying in their level of association to the critical lure. Specifically, we chose lists that contained 5 high, 5 moderate, and 5 weak associates in order to evenly vary the serial position of the strength of the association. For example, for the critical lure King, the list included the highly associated items throne, queen, crown, reign, monarch, moderately associated items royal, palace, prince, chess, leader, and the weakly associated items dictator, George, rule,

England, and subject. Following the choice of the appropriate lists, we used a Latin square to create three possible word orders for each list, high associates–moderate associates–low associates (HML), moderate associ ates–low associates–high associates (MLH), and low associates–high associates–moderate associates (LHM). A random number generator was then used to create 18 conditions for the order of presenta tion of the word lists. First, we created six orders in which the lists would be presented (e.g., king, doctor, bread, city, mountain, trash, sweet, needle, slow, window). Next, we randomized the order of the serial position within each list. The final set of 18 presentation orders can be seen in Table 1. All stimuli and tests were displayed using MediaLab (Version 2014.1.127; Jarvis, 2014). Items were presented at the center of the computer screen against a light blue background. Each word was 7 mm tall and presented in all capital letters in black Arial font at a rate of 2 seconds per word. After the presentation of each list, participants were given a recall test and then a recognition test. For the recall test, a text box appeared asking participants to “Please recall any words you just saw in the most recent word list.” This is an open-ended question, similar to an essay question on an exam, for which participants were given 90 seconds to recall as many of the words as possible. For the recognition test, 13 words (6 studied items, 6 unstudied items, and the critical lure) were presented. Participants were asked, “Did the word ____ appear in the list you just saw?”, and told to either answer yes or no. This is comparable to a multiple-choice question on an exam because participants only have to recognize the words they have seen before from the options provided. When all 10 lists were completed, partici pants were given a debriefing form and thanked for their time. Participants took about 45 minutes to complete the entire process.

Results The Effect of Serial Position on Memory for Studied Words The researchers coded the recall of studied words and critical lures for each list (recalled = 1, not recalled = 0). Only the exact word was counted (e.g., a participant who recalled king was scored with a 1, whereas a participant who recalled leader was scored with a 0). We then calculated the proportion of studied words that participants recalled by posi tion (1–15) and by presentation order (HML, MLH, or LHM). These data were used to plot the serial

Audiffred and Broadbridge | Serial Position and False Memory

position curve for each of the three presentation orders (see Figure 1). Finally, for each presentation order, we averaged the recall scores for each part of the curve (primacy, middle words, and recency). We hypothesized the typical serial position effect for the presented words regardless of whether highly associated items were placed at the beginning, middle, or end of the list. To examine this hypothesis, a 3 (HML, MLH, LHM) x 3 (first words, middle words, last words) within-subjects Analysis of Variance (ANOVA) with post-hoc tests using the Bonferroni correction for Type I error was conducted. The main effect of presentation order was not significant, F(2, 164) = 0.61, p = .54, ηp2 = .01. Recall of presented words did not differ based on which types of associates were presented first. The main effect of position was significant, F(2, 164) = 70.28, p < .001, ηp2 = .46. As can be seen in Figure 1, the serial position effect was shown for all three presentation order conditions. Figure 2 shows that words presented first (M = 0.58, SE = 0.02) or last (M = 0.59, SE = 0.02) were

remembered significantly better than words pre sented in the middle (M = 0.45, SE = 0.02) of the list (ps < .001), whereas words presented first were recalled equally as well as the words presented last (p = .94). The interaction of order and position was also significant, F(4, 328) = 17.54, p < .001, ηp2 = .18. Simple effects tests revealed that the serial position effect was present for all three presentation orders, but there were some variations in the pattern of this effect. For the HML order, the effect of position on recall was significant, F(2, 164) = 29.98, p < .001, η 2 = .27 (see Figure 2). Post-hoc tests with the Bonferroni correction for familywise error revealed that participants remembered the words that were presented first significantly better than the words presented in the middle (p < .001) or the words presented at the end (p = .02) of the list. They also remembered the words presented last significantly better than the words presented in the middle (p < .001) of the list. When high associates were presented at the end of the list (MLH), the effect of position

TABLE 1 List and Presentation Orders 1

Needle LHM

Window MLH

City HML

Trash LHM

Sweet HML

Mountain LHM

Doctor LHM

Trash LHM

City MLH

Window MLH

King HML

Needle LHM

Needle HML

City HML

Slow HML

Mountain HML

King HML

Needle HML

City HML

Sweet LHM

Doctor HML

Bread MLH

Mountain LHM

Sweet HML

Slow LHM

Mountain HML

Sweet MLH

Bread LHM

Needle LHM

Trash MLH

Slow HML

King MLH

Doctor LHM

Sweet MLH

Bread MLH

Slow LHM

King HML

Mountain MLH

Window MLH

King MLH

Trash LHM

City HML

Needle HML

Doctor LHM

King HML

Trash MLH

Doctor HML

King HML

Mountain LHM

Slow HML

Needle HML

Bread MLH

Needle HML

Mountain HML

Sweet LHM

Slow HML

Sweet LHM

Doctor LHM

Bread MLH

Window MLH

City MLH

Window LHM

Doctor LHM

Doctor HML

Bread LHM

Slow HML

Window MLH

Needle HML

Bread MLH

Window LHM

City MLH

Trash LHM

Slow HML

Trash MLH

Mountain MLH

Sweet HML

Doctor LHM

Trash LHM

Slow LHM

Bread MLH

Mountain MLH

City HML

Bread LHM

City HML

Window MLH

King MLH

King HML

Sweet MLH

Window LHM

Mountain LHM

Trash MLH

Doctor HML

Trash HML

Sweet LHM

Mountain MLH

Doctor MLH

Bread LHM

Window HML

Slow LHM

Window LHM

Sweet HML

Slow MLH

Mountain MLH

Trash HML

Mountain HML

Needle LHM

King LHM

Sweet HML

Needle MLH

Window LHM

Doctor MLH

City LHM

Sweet LHM

Needle LHM

Mountain HML

Sweet MLH

Trash MLH

Sweet HML

Bread HML

Slow MLH

Needle MLH

Slow MLH

Trash HML

City LHM

Needle MLH

King MLH

City MLH

Needle MLH

Doctor MLH

Bread HML

King LHM

Window HML

Mountain HML

Doctor HML

Bread HML

Mountain LHM

Mountain MLH

King LHM

City LHM

Window HML

Sweet MLH

Doctor MLH

Bread HML

Slow LHM

King MLH

Window HML

Bread HML

Doctor MLH

Sweet MLH

King LHM

Trash HML

City MLH

Doctor HML

Trash MLH

Sweet LHM

Needle MLH

Window HML

City LHM

Slow MLH

Bread LHM

Window LHM

Mountain LHM

Slow LHM

King LHM

Trash HML

Slow MLH

Trash HML

Doctor MLH

Needle LHM

King MLH

City MLH

Needle MLH

City LHM

Window HML

King LHM

Bread LHM

City LHM

Slow MLH

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

125

Serial Position and False Memory | Audiffred and Broadbridge

on recall was also significant, F(2, 164) = 87.26, p < .001, η2 = .52 (see Figure 2). Post-hoc tests with the Bonferroni correction for familywise error revealed that participants remembered the words that were presented last significantly better than the words presented in the middle (p < .001) or the words presented first (p = .002) in the list. They also remembered the words presented first significantly better than the words presented in the middle (p < .001) of the list. Finally, when high associates were presented in the middle of the list (LHM), the effect of position on recall was significant, F(2, 164) = 8.48, p < .001, η2 = .09 (see Figure 2). Post-hoc tests with the Bonferroni correction for familywise error revealed that participants remembered the words that were presented in the middle of the list

Proportion of Words Recalled

FIGURE 1 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30

HML

MLH

LHM

10 11 12 13 14 15

Serial Position of Presented Words Figure 1. The proportion of presented words recalled by list order and serial position. This graph depicts the proportion of words recalled based on the position of the word and the order of high, moderate, and low associated words within the list. For each list order, the data approximate a typical serial position curve. HML = high associates–moderate associates– low associates. MLH = moderate associates–low associates–high associates. LHM = low associates–high associates–moderate associates.

FIGURE 2

Proportion of Words Recalled

0.70 0.60 0.50 0.40

0.20 0.10 0.00

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

126

First Middle Last

0.30

HML

MLH List Order

LHM

Figure 2. The effect of list order on the serial position effect. This graph depicts the differences in recall of first, middle, and last words based on the list order condition. For the high associates–moderate associates–low associates (HML) order, the typical serial position effect is seen. However, when high associates were presented in the middle or at the end of the list, the serial position effect was modified. For the moderate associates–low associates–high associates (MLH) order, the recency effect was stronger than the primacy effect, whereas in the low associates–high associates–moderate associates (LHM) order, the middle words were remembered more frequently than is typical for a list learning task.

significantly worse than the words presented at the beginning (p = .02) or at the end (p < .001), but they remembered the words at the beginning and end of the list equally well (p = .32). Contrary to our hypothesis, the presentation order did have an effect on the serial position effect. When highly associated items were presented in the beginning of a list, we found the typical serial position effect, but when the high associates were presented at the end, we found a greater recency effect than primacy effect. Finally, when high associ ates were presented in the middle of the list, the strength of the serial position effect was reduced (i.e., middle words were remembered better than expected). This is evidenced by the smaller effect size for the serial position effect in this condition. The Effect of Serial Position on Memory for the Critical Lure Next, we calculated the proportion of critical lures recalled separately for each list type (i.e., when high, moderate, and low associates were presented first) by averaging the codes across lists with the same order-of-word presentation. This resulted in three outcome variables, one for each orderof-word presentation (i.e., HML, MLH, & LHM). We followed the same procedure to calculate the proportion of critical lures recognized for each order of associates. A one-way within-subjects ANOVA with posthoc tests using the Bonferroni correction for Type I error was conducted to examine the effect of serial position on recall. Our hypothesis was that participants would be more likely to falsely recall the critical lure when high associates were placed in the beginning of a list, rather than when moderate or weak associates were placed at the beginning of the list. Results showed that there was a significant effect of serial position on the proportion of critical lures recalled, F(2, 164) = 6.32, p = .002, η2 = .07. When high associates (M = 0.39, SD = 0.34) were placed in the beginning of the list, participants falsely recalled a greater proportion of critical lures than when moderate associates (M = 0.28, SD = 0.26) and weak associates (M = 0.26, SD = 0.24) were placed at the beginning of the list. There was no significant difference in the proportion of criti cal lures recalled between MLH and LHM orders (see Figure 3). Results for recall supported our hypothesis that serial position would have an effect on false recall. Power analysis revealed that there was adequate power to find this effect (power = .89). A second one-way within-subjects ANOVA was

Audiffred and Broadbridge | Serial Position and False Memory

The main finding of this study was that the serial position of the high associates influenced the likelihood of recalling a critical lure but did not influence the likelihood of recognizing it. Below, we discuss the theoretical implications of this study and examine limitations to our study that could have suppressed results. In this study, we used the BAS values from Roediger and colleagues (2001), and we divided words into high, moderate, and weak associates to the critical lure. We then randomized the serial position of these groupings. The placement of highly associated items within a list did affect the likelihood of a participant recalling the critical lure, such that participants recalled the critical lure more often when high associates were placed first. These results can be readily explained by the activation monitoring theory (Roediger & McDermott, 1995). When the highly associated items were placed first in the list, it might have caused activation of the critical lure earlier in the list learning process. Later, when recalling the list, participants might have found it more difficult to determine the source of activation because of the time that passed from the activation during learning. It could also be the case that placing the highly associated items first in the list led to stronger activation of the semantic network. Thus, activation spreads to include the critical lure, along with other highly related items during encoding. When the brain fails to accurately monitor which items are truly seen, false memory occurs. That monitoring error leads participants to believe they saw the nonrepresented items in the list viewed. However, the lower rate of false recall when

FIGURE 3 Proportion of Recalled Critical Lures

Discussion

highly related items were placed at the end of the list is more readily explained by fuzzy trace theory (Brainerd & Reyna, 2002). Such results suggest that, when highly associated items were placed at the end of the list, the verbatim traces were stronger than the gist traces at recall. This would have led to suppression of the critical lure and the lower false recall rates. In our findings, there was not a difference in false recognition based on serial position. Participants recognized critical lures regardless of whether high, medium, or low associates were pre sented first. Fuzzy trace theory better explains this finding. According to fuzzy trace theory, memory leaves both gist and verbatim traces (Brainerd & Reyna, 2002). When a gist trace is strong, it can cause a person to falsely remember something (Brainerd & Reyna, 2002). In our study, each list was made of associates of the critical lure, and

0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

HML

MLH List Order

LHM

Figure 3. The effect of the serial position of highly associated items on recall of the critical lure. This graph depicts the proportion of recalled critical lures as a function of the associative strength of words that were presented first (serial position). There was a significant effect of serial position on the number of critical lures recalled. When high associates were presented first, more critical lures were recalled than when medium associates (p = .01) or low associates (p = .002) were presented first. There was no significant difference in the proportion of critical lures recalled when medium associates vs. low associates were presented first (p = .43). HML = high associates–moderate associates–low associates. MLH = moderate associates– low associates–high associates. LHM = low associates–high associates–moderate associates.

FIGURE 4 Proportion of Critical Lures Recognized

conducted to examine the effect of serial position on recognition. We hypothesized that there would be an effect of serial position on recognition of the critical lure. There was not, however, a significant effect of serial position on the number of critical lures recognized, F(2, 130) = 2.10, p = .13, η2 = .03. Although slightly more participants recognized the critical lure when high associates (M = 0.77, SD = 0.27) were presented first, the results were not significantly different from either MLH (M = 0.72, SD = 0.25) or LHM orders (M = 0.70, SD = 0.26; see Figure 4). This result did not support our hypothesis. Power analysis revealed that there was not adequate power to find this small effect (power = .43); however, additional analyses revealed that a sample of 322 participants would have been needed to find an effect of this size.

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

HML

MLH List Order

LHM

Figure 4. The effect of the serial position of highly associated items on recognition of the critical lure. This graph depicts the proportion of recognized critical lures as a function of the associative strength of words that were presented first. There was not a significant effect of serial position on the number of critical lures recognized. HML = high associates–moderate associates–low associates. MLH = moderate associates–low associates–high associates. LHM = low associates– high associates–moderate associates.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

127

Serial Position and False Memory | Audiffred and Broadbridge

the specific associates did not change, only their position in the list changed. This should result in the same gist trace regardless of the position of the words, thus there was no difference in false recognition based on serial position. Participants recollected strong gist traces, which likely caused the critical lure to seem familiar, and they therefore recognized it as part of the studied list. The gist trace is a fuzzy representation of what was encoded, so viewing the critical lure that the list they just saw was derived from will cause participants to think that they saw it during the study period. Ebbinghaus (1948), and many others since, found that individuals can remember items better when presented at the beginning of a list. This primacy effect has long been thought to be due to rehearsal (Murdock, 1962), but our results suggest that other factors may be at work as well. The activation of the semantic networks associated with the words individuals are rehearsing may also contribute to this primacy effect. The combination of rehearsal and activation, particularly when the high associates are presented first, seems to increase the likelihood of false recall of the critical lure. Participants recalled the critical lure more because having high associates in the beginning not only allows them to process and store them thoroughly, it also adds stronger semantic connections to the criti cal lure. Our results showed no recency effects for recall or recognition. Recency effects are typically thought to be due to the recent words remaining in working memory without maintenance (e.g., Fiore, Borella, Mammarella, & Beni, 2012; Wiswede, Rüsseler, & Münte, 2007). Our results seem to sup port this. We found no difference in the recall or recognition of the critical lure when high associates were presented in the middle of the list versus at the end of the list. This suggests that the words that remain salient in working memory are the actual presented words instead of the critical lure. Activation of the semantic network does not change what participants saw most recently, so there is not a failure of source memory for the recency effect.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

128

Limitations This study does have some limitations regarding the word lists that were chosen. First, some lists may hold stronger overall connections to the critical lure, and thus, result in higher false memory rates than other lists. This could have contributed to the high rates of recognition of the critical lures that might have masked the effect of serial position on false recognition. Second, participants were asked

to complete the recall test before the recognition test for every word list. This could have resulted in the recall test influencing performance on the recognition test. This could be why recognition of the critical lures was so high in our sample. Finally, we did not test the words for emotional relevance. If some lists were less neutral than others, this could have affected participants recall of presented words and their recall of the critical lure. Future Research There is a plethora of future research directions in the realm of false memory. Based on the find ings of the current study, future work could focus on the effect of instruction wording on recall and recognition. In the current study, we did not explain to participants what a critical lure was nor did we discuss information about false recollection. Warning participants about the potential for false recall of a critical lure could impact their attention to the word lists and could decrease their recall of the critical lure. Indeed, multiple studies have supported this (e.g., Carneiro & Fernandez, 2010; Newstead & Newstead, 1998; Watson, McDermott, & Balota, 2004). Another direction that future research could take would be to examine lists of emotion-inducing words. Emotional stimuli are often better remem bered in controlled environments (Charles, Mather, & Carstensen, 2003). It would also be interesting to examine whether stimuli other than words (e.g., pictures) could induce false memory. We could examine this on its own or in combination with emotional relevance. This could be a way to connect false memory in controlled laboratory environ ments to false memory in real-world settings such as witness statements or perpetrator identification in a line-up. These possible directions in research are important to explore so that people can apply these findings to real-world problems and techniques. In education, researchers and teachers are constantly developing and updating strategies to help students remember what they learn such as creating strong associations or mental imagery that students can connect to the information they are learning. Additionally, participants in this study showed higher false memory rates in the recognition task than in the recall task. This result could help stu dents with their test-taking strategies for multiple choice exams by teaching them to focus on recall first. Encoding specificity research would support this idea as well. Wiseman and Tulving (1976)

Audiffred and Broadbridge | Serial Position and False Memory

have shown recognition failures in the absence of cues that were present at study. In the present study, no cues were presented for the words during study, recall, or recognition, which may be why participants showed higher rates of recognition than recall. In their personal lives, most people have experienced situations where they heard some information and could not remember where they heard it. This is a source monitoring error that prevents individuals from interpreting the validity of what was heard, similar to the source monitoring failures found in this study. This research connects strongly to situations people encounter on a dayto-day basis and is an important topic that deserves further analysis. Conclusion In conclusion, our study contributed to the body of research on false memory by examining the serial position effect. The serial position effect has not been thoroughly examined in relation to false recall and recognition, so the results of this study contributed to the knowledge about how the place ment of items varying in associative strength can affect what someone remembers. We have shown that the serial position of highly associated items did influence false recall, but not false recognition. These results are very positive and worthy of further study by others looking to delve deeper into how serial position, or other manipulations, can affect an individual’s memory.

References Arndt, J. (2012). The influence of forward and backward associative strength on false memories for encoding context. Journal of Experimental Psychology: Learning, Memory, & Cognition, 38, 747–756. https://doi.org/10.1037/a0026375 Beato, M. S., Cavadid, S., Pulido, R. F., & Pinho, M. S. (2012). No effect of stress on false recognition. Psicothema, 25, 25–30. https://doi.org/10.7334/psicothema2012.158 Brainerd, C. J., & Reyna, V. F. (2002). Fuzzy-trace theory and false memory. Current Directions in Psychological Science, 11, 164–169. https://doi.org/10.1111/1467-8721.00192 Carneiro, P., & Fernandez, A. (2010). Age differences in the rejection of false memories: The effects of giving warning instructions and slowing the presentation rate. Journal of Experimental Child Psychology, 105, 81–97. https://doi.org/10.1016/j.jecp.2009.09.004 Charles, S. T., Mather, M., & Carstensen, L. L. (2003). Aging and emotional memory: The forgettable nature of negative images for older adults. Journal of Experimental Psychology: General, 132, 310–324. http://dx.doi.org/10.1037/0096-3445.132.2.310 Deese, J. (1959). Influence of inter-item associative strength upon immediate free recall. Psychological Reports, 58, 305–312. https://doi.org/10.2466/pr0.1959.5.3.305 Ebbinghaus, H. (1948). Concerning memory, 1885. In W. Dennis (Ed.), Century psychology series. Readings in the history of psychology (pp. 304–313). https://doi.org/10.1037/11304-034

Fiore, F., Borella, E., Mammarella, I. C., & De Beni, R. (2012). Age differences in verbal and visuo-spatial working memory updating: Evidence from analysis of serial position curves. Memory, 20, 14–27. https://doi.org/10.1080/09658211.2011.628320 Gallo, D. A. (2010). False memories and fantastic beliefs: 15 years of the DRM illusion. Memory & Cognition, 38, 833–848. https://doi.org/10.3758/mc.38.7.833 Gallo, D. A., & Roediger, H. L. (2002). Variability among word lists in eliciting memory illusions: Evidence for associative activation and monitoring. Journal of Memory and Language, 47, 469–497. https://doi.org/10.1016/s0749-596x(02)00013-x Hicks, J. L., & Hancock, T. W. (2002). Backward associative strength determines source attributions given to false memories. Psychonomic Bulletin & Review, 9, 807–815. https://doi.org/10.3758/bf03196339 Jarvis, B. G. (2014). MediaLab (Version 2014.1.127) [Computer Software]. New York, NY: Empirisoft Corporation. Murdock, B. B. (1962). The serial position effect of free recall. Journal of Experimental Psychology, 64, 482–488. https://doi.org/10.1037/h0045106 Newstead, B. A., & Newstead, S. E. (1998). False recall and false memory: The effects of instructions on memory errors. Applied Cognitive Psychology, 12, 67–79. https://doi.org/10.1002/(SICI)1099-0720(199802)12:1<67::AIDACP492>3.0.CO;2-1 Prohaska, V., DelValle, D., Toglia, M. P., & Pittman, A. E. (2016). Reported serial positions of true and illusory memories in the Deese/Roediger/McDermott paradigm. Memory, 24, 865–883. https://doi.org/10.1080/09658211.2015.1059455 Rajaram, S. (1993). Remembering and knowing: Two means of access to the personal past. Memory & Cognition, 21, 89–102. https://doi.org/10.3758/BF03211168 Roediger, H. L., Balota, D. A., & Watson, J. M. (2001). Spreading activation and arousal of false memories. In H. L. Roediger, J. S. Nairne, I. Neath, & A. M. Surprenant (Eds.), Science conference series. The nature of remembering: Essays in honor of Robert G. Crowder (pp. 95–115). https://doi.org/10.1037/10394-006 Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, & Cognition, 21, 803–814. https://doi.org/10.1037/0278-7393.21.4.803 Roediger, H. L., Watson, J. M., McDermott, K. B., & Gallo, D. A. (2001). Factors that determine false recall: A multiple regression analysis. Psychonomic Bulletin & Review, 8, 385–407. https://doi.org/10.3758/bf03196177 Stadler, M. A., Roediger, H. L., & McDermott, K. B. (1999). Norms for word lists that create false memories. Memory & Cognition, 27, 494–500. https://doi.org/10.3758/bf03211543 Tulving, E. (1985). How many memory-systems are there? American Psychologist, 40, 385–398. http://dx.doi.org/10.1037/0003-066X.40.4.385 Watson, J. M., McDermott, K. B., & Balota, D. A. (2004). Attempting to avoid false memories in the Deese/Roediger–McDermott paradigm: Assessing the combined influence of practice and warnings in young and old adults. Memory & Cognition, 32, 135–141. https://doi.org/10.3758/BF03195826 Wiseman, S., & Tulving, E. (1976). Encoding specificity: Relation between recall superiority and recognition failure. Journal of Experimental Psychology: Human Learning and Memory, 2, 349–361. https://doi.org/10.3758/BF03197394 Wiswede, D., Rüsseler, J., & Münte, T. F. (2007). Serial position effects in free memory recall – An ERP-study. Biological Psychology, 75, 185–193. https://doi.org/10.1016/j.biopsycho.2007.02.002 Author Note. Josephine Audiffred, https://orcid.org/0000-0001-8276-8299, Department of Psychology, Saint Xavier University; Carissa L. Broadbridge, https://orcid.org/0000-0002-1041-4771, Department of Psychology, Saint Xavier University. This research study was supported by the Provost’s Student-Faculty Collaboration Grant provided by Saint Xavier University. Correspondence concerning this article should be addressed to Josephine Audiffred, 5649 S Trumbull Avenue, Chicago, Illinois, 60629. E-mail: audiffred.j01@mymail.sxu.edu

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

129

https://doi.org/10.24839/2325-7342.JN25.2.130

Puberty, Parents, and Depression: An EMA Study in Adolescent Girls Danielle Apple and Stefanie Sequiera University of Pittsburgh

ABSTRACT. Pubertal onset has frequently been shown to predict depressive symptoms in adolescent girls. Research has also suggested that parents play an important role during this developmental period. The goal of the present study was to test whether high parent–child (PC) closeness might buffer the link between pubertal status and depressive symptoms in adolescent girls. To do this, we used ecological momentary assessment (EMA) to assess adolescents’ perceptions of their closeness to their parents in daily life. One-hundred-fifteen 11–13-year-old female adolescents reported on their parental closeness using a smartphone for 16 days. They also self-reported on their depressive symptoms, pubertal status, and parental closeness using questionnaires. Surprisingly, findings failed to support an association between pubertal status and depressive symptoms, and no interaction between pubertal status and closeness on depressive symptoms was found. Interestingly, however, the EMA measure of closeness was significantly correlated with pubertal status for closeness to dad and closeness to mom at trend level but not with depressive symptoms. Alternatively, the questionnaire measure of closeness was significantly correlated with depressive symptoms but not with pubertal status. These results suggest that how adolescent girls report on relationships may be important for understanding associations between pubertal status, depressive symptoms, and parenting. The findings have implications for the future investigation of the relationship between puberty and depressive symptoms and for the use of EMA and questionnaires to study PC relationship quality. Keywords: depression, puberty, parent–child relationships, ecological momentary assessment

O SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

130

ne trend that emerges in adolescence is a significant increase in rates of major depressive disorder (Avenevoli, Swendsen, He, Burstein, & Merikangas, 2015) and depressive symptoms (Gardner & Lambert, 2019; Vannucci, Flannery, & McCauley Ohannessian, 2018) among female adolescents. Existing research has suggested that these increases are explained in part by the onset of puberty (Angold, Costello, & Worthman, 1998; Lewis et al., 2018). A separate body of literature has also suggested that positive parent– child (PC) relationship qualities may buffer the effects of stress on adolescent depressive symptoms (Anderson, Salk, & Hyde, 2015; Ge, Lorenz,

Conger, Elder, & Simons, 1994; Ge, Natsuaki, Neiderhiser, & Reiss, 2009; Hazel, Oppenheimer, Technow, Young, & Hankin, 2014). However, few studies have examined how PC relationships may moderate the association between puberty and depressive symptoms. Because puberty can be a significant source of stress in an adolescent girl’s life (Conley & Rudolph, 2009; Steinberg, 2017), a warm parental relationship may help buffer the effects of puberty on depressive symptoms. Further, the work that has been done has focused on self-reported PC relationship quality but has not examined PC relationship variables as experienced by adolescents in daily life (Benoit, Lacourse, & Claes, 2013;

*Faculty mentor

Apple and Sequiera | Puberty, Parents, and Depression

Booth, Johnson, Granger, Crouter, & McHale, 2003; Rudolph & Troop-Gordon, 2010). Thus, the primary goal of the current study was to use the Ecological Momentary Assessment (EMA) to test the hypothesis that PC closeness would moderate the relationship between pubertal status and depressive symptoms in adolescent girls. Socioemotional Effects of Puberty Effects of puberty on depressive symptoms. More advanced pubertal status (i.e., the level of devel opment based on visible physical changes) is associated with more depressive symptoms in girls both concurrently (Conley & Rudolph, 2009) and longitudinally (Conley, Rudolph, & Bryant, 2012; Ge, Elder, Regnerus, & Cox, 2001; McGuire, McCormick, Koch, & Mendle, 2019; Trépanier et al., 2013). Studies using the Tanner stages (measure of secondary sex characteristics; Angold et al., 1998) and the onset of menarche (Joinson, Heron, Lewis, Croudace, & Araya, 2011; Trépanier et al., 2013) as measures of pubertal development found that mature girls reported higher levels of depressive symptoms than both immature girls and boys in general. Notably, pubertal status has been shown to be a better predictor of depressive symptoms than age (Angold et al., 1998; Conley & Rudolph, 2009). Additionally, puberty partially explains the gender differences in depression rates that begin to emerge around age 13, with girls showing greater increases in depressive symptoms (Angold et al., 1998; Lewis et al., 2018; Salk, Hyde, & Abramson, 2017). A major biological change that occurs dur ing puberty is the increase in sex hormones. Elevated estrogen and testosterone are associated with increases in depressive symptoms (Angold, Costello, Erkanli, & Worthman, 1999). These hormones stimulate the development of second ary sex characteristics like pubic and body hair growth and breast development, as well as gains in height and weight (Steinberg, 2017). Adolescent girls may not be prepared for these potentially stressful physical changes, and these changes may affect their self-esteem and mood beyond the effects of the hormones themselves (Ge et al., 2003). Additionally, increased activation of the hypothalamic-pituitary-adrenal axis leads to increased release of cortisol in response to stress during puberty, which may heighten an adolescent girl’s risk for developing depressive symptoms (Gunnar, Wewerka, Frenn, Long, & Griggs, 2009; Trépanier et al., 2013).

One proposed theory for the association between pubertal status and depression in girls is the stressful change hypothesis, which posits that the changes associated with puberty, particularly menarche, are inherently stressful and contribute to increased vulnerability to depression (Joinson et al., 2011). In one study, researchers found that girls who had already reached menarche had higher levels of salivary cortisol and depressive symptoms than both girls who had not reached menarche and boys, thus lending support to the stressful change hypothesis (Trépanier et al., 2013). Effects of puberty on PC relationships. Socially, pubertal development has been linked to important changes in interpersonal relationships during adolescence. Due to the physical changes associated with puberty, young girls may be perceived as more mature or sexually active (Compian, Gowen, & Hayward, 2009), and consistent with a biosocial model, these changes can impact an adolescent’s social relationships (Booth et al., 2003). For example, the emergence of secondary sex charac teristics signals the ascent to reproductive maturity, which may elicit strong emotions from both parent and child and lead to changes in PC interactions (Paikoff & Brooks-Gunn, 1991). In addition to adapting to the physical changes in their pubescent child, parents must adjust to adolescents becoming more invested in the peer group and less interested in spending time with family (Laursen & DeLay, 2011; Suleiman & Dahl, 2019). As adolescents undergo changes during puberty, they typically experience diminished feelings of closeness with their parents (Steinberg, 1988; Suleiman & Dahl, 2019), which partially results from increases in adolescent desire for privacy and less parent–adolescent physical affec tion (Keijsers, Branje, Frijns, Finkenauer, & Meeus, 2010; Steinberg, 2017). Puberty appears to affect both maternal and paternal relationships; when examined separately, decreases in both maternal and paternal closeness were associated with more advanced pubertal status (Steinberg, 1987). Alternatively, a later study found decreases in father–daughter closeness but not mother– daughter closeness with increasing pubertal status (Steinberg, 1988). Notably, girls generally report more overall closeness with mothers than with fathers (Steinberg, 2017). Therefore, it is important to examine maternal and paternal relationships separately when studying their influences on adolescent adjustment.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

131

Puberty, Parents, and Depression | Apple and Sequiera

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

132

Effects of PC Relationships on Depression The link between PC relationships and adolescent depression has been extensively studied, with research suggesting that positive PC relation ship qualities may serve as a protective factors against adolescent depressive symptoms (Finan, O h a n n e s s i a n , & G o r d o n , 2 0 1 8 ; G a r i é p y, Honkaniemi, & Quesnel-Vallée, 2016; Ge et al., 2009). For example, studies have found that high parental warmth and support are associated with fewer depressive symptoms in adolescent girls (Cumsille, Martínez, Rodríguez, & Darling, 2015; Greenberger, Chen, Tally, & Dong, 2000), and a recent meta-analysis found that strong parental relationships predict better depressive symptom trajectories (Shore, Toumbourou, Lewis, & Kremer, 2018). Thus, these studies suggest that parents still play an important role in protecting against the onset of depression in their adolescents, despite a decline in closeness (Suleiman & Dahl, 2019) and time spent together (Steinberg, 2017). Many of these studies consider mothers and fathers as a single parenting dimension (i.e., parental), but other researchers have emphasized using separate measures of maternal and paternal relationship qualities (Bean, Bush, McKenry, & Wilson, 2003; Vazsonyi & Belliston, 2006). In one study, researchers compared maternal and paternal relationship qualities and found maternal support and closeness accounted for more variance in adolescent depressive symptoms than paternal support and closeness (Vazsonyi & Belliston, 2006). However, other studies have found that, even when measured separately, maternal and paternal close ness (Ge et al., 2009; Vazsonyi & Belliston, 2006) and support (Needham, 2007; Vazsonyi & Belliston, 2006) are both negatively associated with adolescent depressive symptoms. These results suggest that both mothers and fathers play a critical role in protecting against adolescent depressive symptoms and should be measured separately when examin ing parental relationships. The buffering hypothesis posits that social support can buffer the effects of stressful life events on risk for depression (Cohen & Wills, 1985). The literature surrounding the buffering hypothesis as it relates to PC relationships and risk for depression is mixed. In one study, higher parent-reported PC relationship quality buffered the effects of peer stress on depressive symptoms (Hazel et al., 2014). Other studies found that maternal, but not paternal, warmth and support (Ge et al., 1994) and closeness (Ge et al., 2009) moderated the

association between stress and depressive symptoms in adolescent girls. Despite evidence for the buffer ing hypothesis found in these studies, others did not find similar support (Burton, Stice, & Seeley, 2004; Greenberger et al., 2000). One explanation for contrary findings may be the use of parent-report versus child-report measures of closeness (Ge et al., 2009). Further, relying on self-report measures of PC relationship quality may be subject to recall bias. An alternative data collection method that assesses PC relationship quality in a natural setting, such as EMA, may help address these recall bias issues. EMA allows researchers to collect data in real-time, thus maximizing ecological validity and minimizing recall bias often seen in retrospective reporting (Shiffman, Stone, & Hufford, 2007). EMA can provide information on an adolescent’s current emotional state and potential contribut ing factors to that state, such as the presence of a parent or peer (Silk et al., 2011). For example, in a study of clinically depressed adolescents, Forbes and colleagues (2012) used EMA to assess factors contributing to treatment response. They found that time spent with fathers as measured by EMA predicted better treatment response. In a study using the experience sampling method (similar to EMA; Csikszentmihalyi & Larson, 1987), research ers found that although the overall time spent with family decreased from early to late adolescence, both time spent alone with mother and with father did not significantly decrease (Larson, Richards, Moneta, Holmbeck, & Duckett, 1996). Furthermore, they found that girls reported an increase in talking with their mothers alone, and the topics discussed became more interpersonal in nature from early to late adolescence. Although some work has been done with EMA assessing a child’s time spent with parents (Forbes et al., 2012; Larson et al., 1996; Silk et al., 2011), no study to our knowledge has used EMA to examine child’s report of closeness with parents and how that affects the relationship between pubertal status and depressive symptoms. Interactions Between Puberty, Depression, and PC Relationships Although there is robust evidence supporting the link between puberty and depressive symptoms, puberty and PC relationships, and depressive symp toms and PC relationships, there is less research on the association among all three variables. One pos sible way to conceptualize the relationship among these variables is the contextual-amplification

Apple and Sequiera | Puberty, Parents, and Depression

model, which suggests that pubertal maturation is related to depression risk in certain interpersonal contexts that exacerbate depressive symptoms (Benoit et al., 2013; Rudolph & Troop-Gordon, 2010; Winer, Parent, Forehand, & Breslend, 2016). Rudolph and colleagues (2010) examined the moderating role of family contextual factors on the association between pubertal timing and depression in 167 adolescents. Whereas pubertal status is an objective measure of an adolescent’s pubertal stage, pubertal timing is an adolescent’s pubertal status relative to their age (Rudolph & Troop-Gordon, 2010). Earlier pubertal timing predicted depressive symptoms only in girls who reported high or moder ate levels of family stress and in girls whose mothers endorsed high or moderate levels of depression. Researchers have suggested that these contextual risk factors decrease the available parental warmth and support needed for early-maturing adolescents to handle the challenges of puberty (Rudolph & Troop-Gordon, 2010). Other studies have also found support for the contextual amplification model with respect to pubertal timing (Benoit et al., 2013). In a longitudinal study of 1,431 adolescents, Benoit and colleagues (2013) measured pubertal timing at ages 12–13 and perceived parental rejection at ages 14–15 to determine if early pubertal timing predicted depressive symptoms in late adolescence (ages 16–17) and if parental rejection moder ated this association. Early pubertal timing was associated with more depressive symptoms in adolescent girls, and this relationship was moder ated by perceived parental rejection, such that more depressive symptoms were reported in early developing girls who had higher perceived parental rejection. Although authors found support for the contextual-amplification model, the current study was interested in measures of pubertal status rather than pubertal timing. Although there is evidence that both pubertal timing and status are significant predictors of depression (Conley & Rudolph, 2009), studies comparing the two have found that pubertal status is a better predictor (Joinson et al., 2012; Lewis et al., 2018). Given these findings and the increased risk of depression in girls at the onset of puberty, regardless of timing (Angold et al., 1998), pubertal status will be measured as opposed to pubertal timing. Additionally, these studies focus on negative relationships qualities such as high paren tal rejection (Benoit et al., 2013) and family stress (Rudolph & Troop-Gordon, 2010). The current study sought to assess how a positive relationship quality, high PC closeness, might buffer the link

between pubertal status and depressive symptoms. The Present Study PC relationship quality appears to play an impor tant role in the association between pubertal timing and adolescent depressive symptoms (e.g., Benoit et al., 2013; Rudolph & Troop-Gordon, 2010), but more research is needed to better understand how PC relationship quality affects this relationship. Furthermore, we sought to investigate the role of parental closeness as a potential buffer on the link between puberty and depressive symptoms. This is important because puberty appears to influence perceptions of PC closeness (e.g., Steinberg, 1988), and high parental closeness has been directly linked to lower depressive symptoms in adolescents (Ge et al., 2009). In addition, few studies have examined maternal and paternal close ness separately, which is a gap in the literature this study sought to address (Bean et al., 2003; Vazsonyi & Belliston, 2006). Finally, most of the previous literature relied on retrospective questionnaires conducted in the lab to assess PC relationship quality. The current study utilized EMA in order to address the issue of retrospective reporting while providing insight on PC relationships in natural, real-time conditions (Wilson, Smyth, & MacLean, 2013). There were two main goals to this study. The first was to examine the link between pubertal status and depressive symptoms in a sample of adolescent girls. The second goal was to examine the separate effects of mother–daughter relation ship quality and father–daughter relationship quality on the link between pubertal status and depression. In accordance with prior literature (Conley & Rudolph, 2009; Conley et al., 2012; Ge et al., 2001; Lewis et al., 2018), we hypothesized first that more advanced pubertal status would be associated with more depressive symptoms. Second, we hypothesized that both maternal and paternal closeness would moderate the associa tion, such that the relationship between pubertal status and depressive symptoms would be weaker for girls who report more closeness and stronger for girls who report less closeness. An exploratory aim was to examine whether these models are significant for both mothers and fathers. Because using EMA to measure parental closeness is novel, a second exploratory aim was to test whether the EMA findings replicate with the Network of Relationships Inventory-Relationships Quality

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

133

Puberty, Parents, and Depression | Apple and Sequiera

Version (NRI-RQV), a questionnaire assessing parental closeness (Furman & Buhrmester, 1985).

Method Participants One-hundred-fifteen adolescent girls between the ages of 11 and 13 (M = 12.3, SD = 0.8) were recruited from the community through advertise ments and announcements. Participants were oversampled for shy/fearful temperament, which has been shown to predict depression in later adolescence and adulthood (Murberg, 2009). Shy/ fearful temperament was assessed using the Early Adolescent Temperament Questionnaire-Revised (EATQ- R), and about two-thirds of participants were considered at high-risk for depression while one-third was considered at low-risk for depression (Ellis & Rothbart, 2001). Participants could not meet criteria for a DSM-5 current or lifetime diagnosis of major depres sive disorder, anxiety disorder (except specific phobia), or any psychotic or autism spectrum disorder, as measured by the Kiddie-Schedule for Affective Disorders and Schizophrenia-Present and Lifetime version (K-SADS-PL; Kaufman et al., 1997). Additional exclusionary criteria included an IQ < 70 as determined by the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 2011), presence of head injury or congenital neurological anomalies, lifetime presence of a neurological or serious medical condition, acute suicidality, pregnancy, uncorrected visual disturbance, or medications that affect the central nervous system.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

134

Procedures The study, which was approved by the University of Pittsburgh Institutional Review Board, assessed participants at three time points. At Time 1 (ages 11–13) and Time 2 (ages 13–15) participants completed several lab visits consisting of clinical evaluations, questionnaires, and behavioral and neuroimaging tasks. At Time 3 (ages 14–16) participants completed clinical evaluations and questionnaires. The current study will only use data from Time 1, which consisted of three visits and a 2-week, home-based EMA protocol that occured between Visit 2 and 3. During Visit 1, occurring in the lab, participants gave informed consent and were told their eligibil ity for Visit 2 would be determined at the end of Visit 1. A research assistant administered the WASI and a trained graduate student or doctoral-level clinical interviewer administered the K-SADS-PL to

the participant and their primary caregiver to assess current and past DSM-5 diagnoses. Participants were eligible to continue based on their WASI score (IQ > 70) and no exclusionary current or past diagnoses. During Visit 2 at the lab, participants com pleted a variety of behavioral tasks that are not addressed in the current study. At the end of Visit 2, participants were given an Android smartphone and provided with details about the EMA protocol and how to use the smartphone. The EMA protocol lasts 16 days, during which participants were asked questions about recent social interactions and their behavioral and emotional responses to these interactions. Measures Pubertal Development Scale (PDS). Pubertal status was measured using the female PDS (Petersen, Crockett, Richards, & Boxer, 1988), a self-report measure used to assess physical development. Correlations between self-reported PDS and physician ratings of physical development range between .61 and .67 (Brooks-Gunn, Warren, Rosso, & Gargiulo, 1987). The current study utilized the coding system developed by Shirtcliff, Dahl, and Pollak (2009) that converts the PDS to a 5-point scale reflecting the Tanner stages. The scale reflects hormonal signs of development: pubic/body hair and skin changes are associated with adrenarcheal hormones (i.e. DHEA), while menarche, growth spurt, and body changes are associated with gonadal hormones (Shirtcliff et al., 2009). In this sample, the PDS demonstrated acceptable reliability (Cronbach’s α = .76; Cortina, 1993). Mood and Feelings Questionnaire (MFQ). Current depressive symptoms were measured using the child-reported MFQ (Costello & Angold, 1988). The scale consists of 33 items that describe depres sive symptoms for children and adolescents ages 8–18. Participants rated each statement as “True,” “Sometimes,” or “Not True.” The MFQ was filled out by both parent and child; however, only the child-report scale was used for the current study because parents may underestimate depressive symptoms in their children (Wood, Knoll, Moore, & Harrington, 1995), and the curent study was interested in the child’s own perceptions of their depressive symptoms. In this sample, the MFQ demonstrated high reliability (Cronbach’s α = .87). Ecological Momentary Assessment (EMA). PC closeness was reported by participants in real-time using EMA. Using a smartphone app

Apple and Sequiera | Puberty, Parents, and Depression

for Web Data Express developed by the Office of Academic Computing in the University of Pittsburgh Department of Psychiatry, partici pants entered their responses to questions about their daily interactions with parents and peers into an Android smartphone. Participants were sampled over 16 consecutive days by receiving an electronic notification to respond to questions about their social interactions and responses to those interactions. A maximum of 54 completed EMA samples was possible because participants received notifications three times per day on weekdays (once in the morning between 7 a.m. and 8 a.m. and twice between 4 p.m. and 9:30 p.m.) and four times per day on the weekends between 10 a.m. and 9:30 p.m. Participants received notifications at random points throughout these prespecified time intervals. In our sample, the average number of contacts completed was 42, and five participants completed all 54 EMA samples. After receiving a notification, participants were led through a series of questions about their recent social interactions. Specifically, participants were asked to report all the people they were with when they received the notification (e.g., mother, father, friend) and indicate how close/connected they feel with that person on a scale from “Not at all” to “Extremely” by sliding a bar on the application. Network of Relationships InventoryRelationship Qualities Version. The NRI-RQV was tested in the moderation model separately from EMA as an alternative measure of PC closeness. The questionnaire consists of 30 items with 10 scales; five assess positive relationship qualities, and five assess negative qualities (Buhrmester & Furman, 2008). A closeness factor can be computed by averaging the scores from the five positive scales. The child can fill out the NRI-RQV about any family member or friend. For the current study, they were only asked to rate their mother; therefore, we only tested the NRI-RVQ in the moderation model for closeness to mothers. In this sample, the NRI-RQV demonstrated high reliability (Cronbach’s α = .93). Data Analysis EMA data analysis. PC closeness was assessed by analyzing the closeness question if participants endorsed being with either their mother or father at least three times over the 16-day period. Participants who identified being with a parent fewer than three times were excluded from the EMA analyses for that parent. If a participant endorsed being with her mom more than three times but not with her

dad more than three times, data on her closeness to mom was still used. Similarly, we used paternal data for girls who only reported being with their father more than three times. Closeness to mom ratings (calculated on a 0–100 scale) were summed and divided by the total number of time points the participant endorsed being with mom to achieve an average closeness rating. This analysis was repeated identically but separately for closeness to dad ratings. This analysis was sensitive to the fact that participants were not with either parent at every sampling. Main analysis. Pubertal status, depressive symptoms, and closeness ratings were included as continuous variables. All variables were centered for ease of interpretation prior to analyses. First, the direct effect of pubertal status on depressive symp toms was tested by running a multiple regression analysis using SPSS version 24.0 while controlling for participant age. Second, the moderating role of parental closeness on the link between pubertal status and depressive symptoms was tested in a moderation model using the PROCESS macro for SPSS (Hayes, 2017). The moderation model was first tested using EMA data while controlling for participant age and frequency of interactions with parents over the 16-day period. This model was evaluated separately for reported closeness to moms and dads. Second, the model was tested using the NRI-RQV questionnaire as an alternative measure of maternal closeness only.

Results Preliminary Analyses The average age of participants was 12.3 years (SD = 0.82). Our sample was relatively diverse, with 67% identifying as White, Nonhispanic, 25% Black, Nonhispanic, 1% Asian, 1% Native American, and 10% biracial or multiracial. Sixty-seven percent of participants reported currently living with both biological parents, 17% lived with their biological mother, 9% lived with their biological mother and stepfather, and the remaining 8% reported other living arrangements (i.e., biological parentsjoint custody, biological mother and boyfriend/ girlfriend, adoptive parent, and grandparent). Analyses of Variance revealed that there were no significant differences based on racial group and living arrangements on the primary study variables (i.e., depressive symptoms, pubertal status, and average closeness). Examination of the variables included in the model revealed that all variables appeared

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

135

Puberty, Parents, and Depression | Apple and Sequiera

normally distributed. However, we did find low variability in higher levels of depressive symptoms. Scores on the MFQ in this sample ranged from 0 to 30 (out of a possible 66), and although overall variability in depressive symptoms was moderate (variance = 47.79), only two participants reported MFQ scores above what is considered the clinical cutoff signaling the potential presence of major depressive disorder (total score > 26). Table 1 presents means and standard devia tions for the variables included in the model as well as preliminary analyses conducted to determine the relationships between various study variables. Bivariate correlations revealed that participant age was positively related to pubertal status (r = .46, p < .001), so participant age was included as a covariate to isolate effects of pubertal status. Average closeness scores were significantly associ ated with frequency of contacts between the participant and her mother (r = .21, p = .043) and between the participant and her father (r = .11, p = .086), so frequency of contacts was controlled for in further analyses. Bivariate correlations revealed that the EMA measure of closeness was significantly correlated with the NRI-RQV closeness factor (r = .50, p < .001); therefore, the NRI-RQV was tested in the moderation model as an alternate measure of mother–child closeness. TABLE 1 Pearson Correlations of Study Variables and Measures Variables

Age

–

PDS

.46*

–

MRQ-C

-.12

-.06

Parental MDD

.002

-.09

.17 †

EMA Maternal Cloesness

-.13

-.21 †

-.03

–

EMA Paternal Closness

-.18

-.29*

-.005

-.10

.88**

–

Frequency Contact Mom

-.21*

-.05

-.16

-.28**

.21*

Frequency Contact Dad

.007

.13

-.23*

-.21

NRI-RQV

.03

-.01

-.22*

-.11

3.54

8.88

0.29

68.3

64.3

1.06

6.91

0.46

23.2

23.03

M SD

12.3 0.82

.27*

–

.03

.19 †

.58**

–

.50**

.46**

.29**

.16

–

8.94

7.04

3.89

4.41

3.89

0.72

–

Note. PDS = Pubertal Development Scale, MFQ-C = Mood and Feelings Questionnaire, child report, MDD = Major Depressive Disorder, EMA = Ecological Momentary Assessment, NRI-RVQ = Network of Relationships Inventory, Relationship Qualities Version. † p < .10. * p < .05. ** p < .010.

136

Interestingly, preliminary analyses revealed that the EMA measure of closeness had a small yet significant correlation with pubertal status for dads (r = -.29, p = .011) and moms at trend level (r = -.21, p = .051) but was not significantly corre lated with child depressive symptoms. Alternatively, the questionnaire measure of closeness was signifi cantly correlated with child depressive symptoms (r = -.22, p = .021) but was not significantly cor related with pubertal status. Girls with more depressive symptoms reported less closeness to their mom retrospectively using the NRI-RVQ, and girls with more advanced pubertal status reported less closeness to their moms and dads in real-time using EMA. Effect of Pubertal Status on Depression A simple linear regression was run to examine the effect of pubertal status on child depressive symptoms, controlling for child age and parental history of major depressive disorder. This overall model was not significant, F(2, 97) = 0.30, p = .745, R2 = .014. Fourteen participants were missing data on pubertal status. These participants did not differ from participants with pubertal status data on age or depressive symptoms (ps > .39). Three partici pants were missing data on depressive symptoms. These participants did not differ from participants with data on age or pubertal status (ps > .09). PC Closeness as a Moderator Between Pubertal Status and Depressive Symptoms The PROCESS macro for SPSS (Hayes, 2017) was used to test whether PC closeness (centered) moderated the relationship between pubertal status (centered) and child depressive symptoms. Participant age and frequency of contacts (with either mom or dad) were included as covariates. Four participants were excluded from the mater nal model and nine participants were excluded from the paternal model due to having less than three contacts. No significant interactions were found for closeness to mom or closeness to dad. We ran follow-up analyses including participants with less than three contacts with either parent, and the results did not change. As an exploratory aim, the model was also tested using the NRI-RQV as an alternate measure of maternal closeness. Again, no significant interaction was found. All analyses reported above were repeated while controlling for parental history of major depressive disorder (yes/no), and there were no significant changes in results.

Apple and Sequiera | Puberty, Parents, and Depression

Discussion The goals of the current study were to test the relationship between pubertal status and depres sive symptoms in a sample of adolescent girls and examine the potential moderating role of PC closeness on this association. To test this model, we gathered self-report data of pubertal status and depressive symptoms and used EMA to col lect data on participants’ daily contacts with their parents and momentary feelings of closeness to their parent. Retrospective questionnaire data on parental closeness were also collected to serve as a comparison to EMA. Inconsistent with our first hypothesis, we did not find a significant relationship between partici pants’ pubertal status and depressive symptoms. We thus failed to replicate prior research suggesting that more advanced pubertal status is associated with more depressive symptoms in adolescent girls (Conley & Rudolph, 2009; Conley et al., 2012; Ge et al., 2001). One potential explanation for these findings could be the use of a self-report measure of pubertal development. As previously discussed, we used the PDS to measure perceived pubertal development according to the Tanner stages, which reflect different secondary sex characteristics that are driven by different hormones (Shirtcliff et al., 2009). Pubertal onset is associated with an increase in both gonadal (i.e., estrogen and testosterone) and adrenal (i.e., cortisol) hormones, which in turn are associated with increased risk for depressive symptoms (Angold et al., 1999; Gunnar et al., 2009). Angold and colleagues (1999) compared two mea sures of puberty, the Tanner stages and hormonal assays (testosterone and estrogen) on their effects on depressive symptoms. Both measures predicted depressive symptoms; however, the effect of puber tal changes as measured by the Tanner stages on depressive symptoms was no longer significant when hormones were added to the model. Conversely, the effects of hormones on depressive symptoms remained significant even when controlling for the Tanner stages. This finding suggests that using hormones as a measure of pubertal status may be more sensitive in predicting depressive symptoms than the Tanner stages. Indeed, a recent study using Tanner stages recommended that future studies use direct hormonal measures to examine effects of puberty on depressive symptoms (Lewis et al., 2018). Furthermore, the PDS is a composite measure of pubertal development; it measures changes that are driven by different hormones, such as breast development (estrogen) and pubic hair growth

(androgens; Shirtcliff et al., 2009). Some research ers have argued that indicators of pubertal status should be separated because different hormones appear to be differentially related to depressive symptoms (Angold et al., 1999; Joinson et al., 2012; Lewis et al., 2018). For example, Lewis and colleagues (2018) examined breast and pubic hair growth separately and found that breast develop ment, but not pubic hair growth, was associated with more depressive symptoms both concurrently and prospectively. Thus, a relationship between pubertal status and depressive symptoms might not have been detected in our study because we used a composite measure of pubertal status rather than looking at specific changes. The stressful change hypothesis suggests that puberty is an inherently stressful event that increases risk for depressive symptoms in girls. The experience of pubertal stress is often transient and therefore might not have been captured in our measure of depressive symptoms (Joinson et al., 2011). Perhaps specific changes associated with puberty are more important in predicting depressive symptoms than puberty itself, such as physical changes in body shape. Interestingly, Vogt Yuan (2007) found that adolescent girls’ perceptions of being overweight explained the relationship between pubertal status and depressive symptoms; pubertal status alone was not significant in this model. In other studies, researchers found that pubertal development predicted depressive symptoms only in adolescent girls who had poor body image (Marcotte, Fortin, Potvin, & Papillon, 2002) or lower body esteem (Hamlat et al., 2015). Taken together, these studies suggest that, although puberty leads to physical changes, the perception of these changes is more important in predicting depressive symptoms than pubertal status alone. Another stressful change in adolescence that was not measured in the current study is the increasing importance and influence of peers (Laursen & DeLay, 2011; Steinberg, 2017). For example, researchers have found that puberty interacts with peer stress (Conley et al., 2012) and peer victimization (Hamlat et al., 2015) to predict depressive symptoms in adolescent girls. A more recent study found that both pubertal status and peer rejection sensitivity prospectively predicted more depressive symptoms (McGuire et al., 2019), and another found that peer support predicted fewer depressive symptoms (Shore et al., 2018). Taken together, these results suggest that earlier pubertal development may negatively affect girls’

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

137

Puberty, Parents, and Depression | Apple and Sequiera

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

138

social relationships, and this peer stress contributes to depressive symptoms during a developmental time when peer approval is extremely important. Thus, similar to the physical changes occurring during puberty, the peer stress associated with puberty may be more important in predicting depressive symptoms during adolescence than pubertal status itself. Current findings also did not support the hypothesized moderation model, such that neither mother-daughter closeness nor father-daughter closeness moderated the relationship between pubertal status and depressive symptoms. We did not find support for this moderation with either the EMA measure of maternal and paternal closeness or the questionnaire measure of maternal close ness. Prior research has suggested that a close PC relationship may serve as a protective factor against developing depressive symptoms in adolescence in the presence of adverse/stressful life events (Cohen & Wills, 1985; Ge et al., 1994; Ge et al., 2009), family and school stress (Greenberger et al., 2000), romantic stress (Anderson et al., 2015), and peer stress (Hazel et al., 2014). Although puberty can contribute to these different stressors, a measure of puberty itself as an indicator of stress may not be strong enough to detect the potential buffering effects of positive PC relationship quali ties on depressive symptoms. Future studies should examine specific stressors influenced by puberty and how parents can serve as a protective factor against developing depressive symptoms in the presence of these stressors. Furthermore, the current study drew from the contextual-amplification model, which posits that pubertal maturation is related to depression risk in certain interpersonal contexts that exacerbate depressive symptoms (Benoit et al., 2013; Rudolph & Troop-Gordon, 2010). However, these studies examined more negative relationship quality factors such as high family stress, maternal depres sion, and parental rejection. Positive relationship qualities such as high closeness may not act in a way to decrease risk for depressive symptoms as negative qualities do to increase risk. Although low parental closeness can also serve as a negative relationship quality factor, it may not be as salient as harsh parenting or overt parental rejection and thus may not exacerbate depressive symptoms to the extent that other negative relationship qualities do. In addition, prior studies showing support for the contextual-amplification model (Benoit et al., 2013; Rudolph & Troop-Gordon, 2010) not only

focused on negative interpersonal factors, but they also measured pubertal timing rather than pubertal status. Perhaps a measure of pubertal timing, which assesses pubertal status within the context of age and the status of peers, would more strongly predict depressive symptoms. Future work should examine whether high parental closeness is a more salient protective factor against the negative effects associated with early pubertal timing rather than pubertal status. Although we did not find support for the pro posed model, we did find an interesting difference between the EMA and questionnaire measures of closeness in our preliminary analyses. Both measures were highly correlated; however, there were differences in how these measures related to depressive symptoms and pubertal status. The NRI-RVQ was correlated with adolescent depres sive symptoms, but the EMA measure of closeness was not correlated with depressive symptoms. Girls who reported more depressive symptoms retrospectively reported less closeness to their mom, but when asked to report on closeness in real time via EMA, there was no difference compared to girls who reported fewer depressive symptoms. One explanation for this finding may be related to autobiographical memory bias. Youth with more depressive symptoms often exhibit overgeneralized negative memory biases; that is, when they recall events, they remember them as more negative than they actually experienced in the moment (Silk et al., 2011). Our study shows that EMA may be used to circumvent this autobiographical memory bias and may provide a more accurate picture of their perceived closeness with their parents as it is experienced in daily life. This may have important implications for understanding the mechanisms through which depression influences perceptions of PC relationship quality. Alternatively, we found that only the EMA measure of closeness was correlated with pubertal status. Girls with more advanced pubertal status reported less closeness to their parents in daily life using EMA compared to girls who were less advanced, but there was no difference based on pubertal status when girls were asked to report retrospectively on their parental relationships. Given that puberty is associated with increased PC distance and decreased feelings of closeness (Suleiman & Dahl, 2019), girls with more advanced pubertal status may experience less momentary closeness with their parents but may not recall less closeness when asked to reflect retrospectively on

Apple and Sequiera | Puberty, Parents, and Depression

their relationships. EMA appears to be detecting differences in experienced parental closeness based on pubertal status that questionnaire mea sures may not detect. There are several study limitations that should be addressed. First, the study only included ado lescent girls ages 11–13. Gender differences in depression begin to emerge around age 13 (Salk et al., 2017), and only two participants in the study reached the clinical cutoff for potential presence of depression (Costello & Angold, 1988). Although two-thirds of participants were considered at risk for depression due to having shy/fearful tem perament, the present study was a cross-sectional secondary analysis of a longitudinal project. Examining how pubertal status at Time 1 predicted depressive symptoms at Time 2 or 3 might have allowed us to see an association between pubertal status and depressive symptoms. Another limitation is our use of the PDS as a composite measure of pubertal status; research has suggested that specific pubertal changes may be associated with depressive symptoms (Joinson et al., 2012; Lewis et al., 2018). Additionally, the PDS does not directly measure hormonal changes, which may be more predictive of depressive symptoms (Angold et al., 1999). Future studies should investigate this relationship using an alternative measure of pubertal status. It is important to note that participants only rated their mother on the questionnaire measure of closeness because most participants completed the dyadic visits with their mother, and because they completed a battery of questionnaires and tasks during this visit, they were only asked to rate their mother to reduce burden for participants. When comparing EMA to questionnaire measures of closeness, it would be useful to rate both parents separately. Finally, our sample, although somewhat diverse, was primarily White and of a restricted age range. Further, few participants reported clinically significant depressive symptoms; thus, our results may not generalize to other female adolescent populations. There are also potential limitations to using EMA. EMA is an intensive process that requires participants to continuously report on their emo tions and behavior, so sampling may be biased such that only some participants will respond, or certain times during the day may be reported on more than others (Shiffman & Stone, 1998). Additionally, the average number of completed EMA contacts was 42, and only five participants completed all 54 contacts.

There is a potential response bias, such that par ticipants who completed more EMA contacts might have been more motivated to accurately and actively report on their experiences. Researchers must also consider the issue of reactivity, where asking participants to observe and report on their thoughts and behavior may influence those thoughts and behaviors (Shiffman et al., 2007). Furthermore, our EMA measure of closeness was a single item asking how close or connected the adolescent feels to her parent. Future work should incorporate more ques tions to capture the nuances of PC relationships; for example, perhaps an adolescent feels close to a parent in the moment but not supported in their recent decisions, potentially affecting how they make their closeness rating. It is also important to note that participants who reported being with a parent less than three times were excluded from analyses, which could skew the data toward adolescents with positive parental relationships. However, only four participants were excluded in the maternal model and nine participants in the paternal model, and the inclusion of these participants in follow-up analyses did not change the results. Despite the limitations of EMA, it also reflects a strength of the current study. EMA is an eco logically valid measure that allowed us to measure participants’ closeness to their parents in real-time and in naturalistic settings, which indicates how adolescent girls experience and perceive close ness to their parents in everyday life (Shiffman et al., 2007). Furthermore, the EMA measure of closeness was significantly correlated with the NRI-RQV, a widely used questionnaire measure of PC relationship quality. In support of validity, our EMA measure of parental closeness appears to be tapping into positive PC relationship qualities. This study contributed to the growing body of literature that advocates for the use of ecologically valid measures when studying adolescent emotions and social interactions (Forbes et al., 2012; Silk et al., 2011). Another strength of the study is the separate examination of maternal and paternal closeness. Although we did not find support for either model of parental closeness, it is important to continue looking at moms and dads separately to capture the differences in these relationships and how they independently contribute to adolescent adjustment (Bean et al., 2003; Vazsonyi & Belliston, 2006).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

139

Puberty, Parents, and Depression | Apple and Sequiera

Conclusion The results of the current study have important implications for future investigations. First, when examining the relationship between pubertal status and depressive symptoms in adolescent girls, researchers should consider using measures of pubertal status other than the PDS (Angold et al., 1999; Joinson et al., 2012). Future studies should also examine the influence of puberty-related changes on depressive symptoms, such as changes in body image (Marcotte et al., 2002; Vogt Yuan, 2007) and peer relationships (Conley et al., 2012). Second, the results of this study have important implications for the use of EMA and questionnaire measures when studying adolescent girls and their parental relationships. EMA may be a useful tool for avoiding autobiographical memory bias in adolescents with some depressive symptoms (Silk et al., 2011), and it may detect differences in PC relationship quality in daily life that questionnaire measures may not detect. Future work should examine the factors that contribute to diminished closeness to target these issues in PC relationships, which have the potential to protect against an adolescent developing depressive symptoms.

References

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

140

Anderson, S. F., Salk, R. H., & Hyde, J. S. (2015). Stress in romantic relationships and adolescent depressive symptoms: Influence of parental support. Journal of Family Psychology, 29, 339–348. https://doi.org/10.1037/fam0000089 Angold, A., Costello, A., Erkanli, A., & Worthman, C. (1999). Pubertal changes in hormone levels and depression in girls. Psychological Medicine, 29, 1043–1053. https://doi.org/10.1017/S0033291799008946 Angold, A., Costello, E. J., & Worthman, C. M. (1998). Puberty and depression: The roles of age, pubertal status and pubertal timing. Psychological Medicine, 28, 51–61. https://doi.org/10.1017/S003329179700593X Avenevoli, S., Swendsen, J., He, J. P., Burstein, M., & Merikangas, K. R. (2015). Major depression in the national comorbidity survey-adolescent supplement: Prevalence, correlates, and treatment. Journal of the American Academy of Child & Adolescent Psychiatry, 54, 37–44. https://doi.org/10.1016/j.jaac.2014.10.010 Bean, R. A., Bush, K. R., McKenry, P. C., & Wilson, S. M. (2003). The impact of parental support, behavioral control, and psychological control on the academic achievement and self-esteem of African American and European American adolescents. Journal of Adolescent Research, 18, 523–541. https://doi.org/10.1177/0743558403255070 Benoit, A., Lacourse, E., & Claes, M. (2013). Pubertal timing and depressive symptoms in late adolescence: The moderating role of individual, peer, and parental factors. Developmental Psychopathology, 25, 455–471. https://doi.org/10.1017/S0954579412001174 Booth, A., Johnson, D. R., Granger, D. A., Crouter, A. C., & McHale, S. (2003). Testosterone and child and adolescent adjustment: The moderating role of parent–child relationships. Developmental Psychology, 39, 85–98. https://doi.org/10.1037/0012-1649.39.1.85 Brooks-Gunn, J., Warren, M. P., Rosso, J., & Gargiulo, J. (1987). Validity of selfreport measures of girls’ pubertal status. Child Development, 829–841. https://doi.org/10.2307/1130220 Buhrmester, D., & Furman, W. (2008). The Network of Relationships Inventory: Relationship Qualities Version. Burton, E., Stice, E., & Seeley, J. R. (2004). A prospective test of the stressbuffering model of depression in adolescent girls: No support once again. Journal of Consulting and Clinical Psychology, 72, 689–697. https://doi.org/10.1037/0022-006X.72.4.689 Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering

hypothesis. Psychological Bulletin, 98, 310–357. https://doi.org/10.1037/0033-2909.98.2.310 Compian, L. J., Gowen, L. K., & Hayward, C. (2009). The interactive effects of puberty and peer victimization on weight concerns and depression symptoms among early adolescent girls. The Journal of Early Adolescence, 29, 357–375. https://doi.org/10.1177/0272431608323656 Conley, C. S., & Rudolph, K. D. (2009). The emerging sex difference in adolescent depression: Interacting contributions of puberty and peer stress. Developmental Psychopathology, 21, 593–620. https://doi.org/10.1017/S0954579409000327 Conley, C. S., Rudolph, K. D., & Bryant, F. B. (2012). Explaining the longitudinal association between puberty and depression: Sex differences in the mediating effects of peer stress. Developmental Psychopathology, 24, 691–701. https://doi.org/10.1017/S0954579412000259 Cortina, J. M. (1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78, 98–104. https://doi.org/10.1037/0021-9010.78.1.98 Costello, E. J., & Angold, A. (1988). Scales to assess child and adolescent depression: Checklists, screens, and nets. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 726–737. https://doi.org/10.1097/00004583-198811000-00011 Csikszentmihalyi, M., & Larson, R. (1987). Validity and reliability of the experience sampling method. Journal of Nervous and Mental Disease, 175, 526–536. Cumsille, P., Martínez, M. L., Rodríguez, V., & Darling, N. (2015). Parental and individual predictors of trajectories of depressive symptoms in Chilean adolescents. International Journal of Clinical Health Psychology, 15, 208–216. https://doi.org/10.1016/j.ijchp.2015.06.001 Ellis, L. K., & Rothbart, M. K. (2001). Revision of the Early Adolescent Temperament Questionnaire. Paper presented at the 2001 Biennial Meeting of the Society for Research in Child Development. Minneapolis, MN. Finan, L. J., Ohannessian, C. M., & Gordon, M. S. (2018). Trajectories of depressive symptoms from adolescence to emerging adulthood: The influence of parents, peers, and siblings. Developmental Psychology, 54, 1555–1567. https://doi.org/10.1037/dev0000543 Forbes, E. E., Stepp, S. D., Dahl, R. E., Ryan, N. D., Whalen, D., Axelson, D. A., . . . Silk, J. S. (2012). Real-world affect and social context as predictors of treatment response in child and adolescent depression and anxiety: An ecological momentary assessment study. Journal of Child and Adolescent Psychopharmacology, 22, 37–47. https://doi.org/10.1089/cap.2011.0085 Furman, W., & Buhrmester, D. (1985). Children’s perceptions of the personal relationships in their social networks. Developmental Psychology, 21, 1016–1024. Gardner, A. A., & Lambert, C. A. (2019). Examining the interplay of self-esteem, traitemotional intelligence, and age with depression across adolescence. Journal of Adolescence, 71, 162–166. https://doi.org/10.1016/j.adolescence.2019.01.008 Gariépy, G., Honkaniemi, H., & Quesnel-Vallée, A. (2016). Social support and protection from depression: Systematic review of current findings in Western countries. British Journal of Psychiatry, 209, 284–293. https://doi.org/10.1192/bjp.bp.115.169094 Ge, X., Elder, G. H., Regnerus, M. C., C., & Cox, C. (2001). Pubertal transitions, perceptions of being overweight, and adolescents’ psychological maladjustment: Gender and ethnic differences. Social Psychology Quarterly, 64, 363–375. https://doi.org/10.2307/3090160 Ge, X., Kim, I. J., Brody, G. H., Conger, R. D., Simons, R. L., Gibbons, F. X., & Cutrona, C. E. (2003). It’s about timing and change: Pubertal transition effects on symptoms of major depression among African American youths. Developmental Psychology, 39, 430–439. https://doi.org/10.1037/0012-1649.39.3.430 Ge, X., Lorenz, F. O., Conger, R. D., Elder, G. H., & Simons, R. L. (1994). Trajectories of stressful life events and depressive symptoms during adolescence. Developmental Psychology, 30, 467–483. https://doi.org/10.1037/0012-1649.30.4.467 Ge, X., Natsuaki, M. N., Neiderhiser, J. M., & Reiss, D. (2009). The longitudinal effects of stressful life events on adolescent depression are buffered by parent-child closeness. Development and Psychopathology, 21, 621–635. https://doi.org/10.1017/S0954579409000339 Greenberger, E., Chen, C., Tally, S. R., & Dong, Q. (2000). Family, peer, and individual correlates of depressive symptomatology among U.S. and Chinese adolescents. Journal of Consulting and Clinical Psychology, 68, 209–219. https://doi.org/10.1037/0022-006x.68.2.209 Gunnar, M. R., Wewerka, S., Frenn, K., Long, J. D., & Griggs, C. (2009). Developmental changes in hypothalamus-pituitary-adrenal activity over the transition to adolescence: Normative changes and associations with puberty. Developmental Psychopathology, 21, 69–85.

Apple and Sequiera | Puberty, Parents, and Depression

https://doi.org/10.1017/S0954579409000054 Hamlat, E. J., Shapero, B. G., Hamilton, J. L., Stange, J. P., Abramson, L. Y., & Alloy, L. B. (2015). Pubertal timing, peer victimization, and body esteem differentially predict depressive symptoms in African American and Caucasian girls. Journal of Early Adolescence, 35, 378–402. https://doi.org/10.1177/0272431614534071 Hayes, A. F. (2017). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (2nd ed.). New York NY: Guilford Press. Hazel, N. A., Oppenheimer, C. W., Technow, J. R., Young, J. F., & Hankin, B. L. (2014). Parent relationship quality buffers against the effect of peer stressors on depressive symptoms from middle childhood to adolescence. Developmental Psychology, 50, 2115–2123. https://doi.org/10.1037/a0037192 Joinson, C., Heron, J., Araya, R., Paus, T., Croudace, T., Rubin, C., . . . Lewis, G. (2012). Association between pubertal development and depressive symptoms in girls from a UK cohort. Psychological Medicine, 42, 2579– 2589. https://doi.org/10.1017/S003329171200061X Joinson, C., Heron, J., Lewis, G., Croudace, T., & Araya, R. (2011). Timing of menarche and depressive symptoms in adolescent girls from a UK cohort. British Journal of Psychiatry, 198, 17–23. https://doi.org/10.1192/bjp.bp.110.080861 Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., . . . Ryan, N. (1997). Schedule for affective disorders and schizophrenia for school-age childrenpresent and lifetime version (K-SADS-PL): Initial reliability and validity data. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 980–988. https://doi.org/10.1097/00004583-199707000-00021 Keijsers, L., Branje, S. J., Frijns, T., Finkenauer, C., & Meeus, W. (2010). Gender differences in keeping secrets from parents in adolescence. Developmental Psychology, 46, 293–298. https://doi.org/10.1037/a0018115 Larson, R. W., Richards, M. H., Moneta, G., Holmbeck, G., & Duckett, E. (1996). Changes in adolescents’ daily interactions with their families from ages 10 to 18: Disengagement and transformation. Developmental Psychology, 32, 744–754. https://doi.org/10.1037/0012-1649.32.4.744 Laursen, B., & DeLay, D. (2011). Parent-child relationship. In B. Brown & M. Prinstein (Eds.), Encyclopedia of adolescence (Vol. 2, pp. 233–240). New York, NY: Academic Press. Lewis, G., Ioannidis, K., van Harmelen, A. L., Neufeld, S., Stochl, J., Lewis, G., . . . Goodyer, I. (2018). The association between pubertal status and depressive symptoms and diagnoses in adolescent females: A population-based cohort study. PLoS One, 13(6), https://doi.org/10.1371/journal.pone.0198804 Marcotte, D., Fortin, L., Potvin, P., & Papillon, M. (2002). Gender differences in depressive symptoms during adolescence: Role of gender-typed characteristics, self-esteem, body image, stressful life events, and pubertal status. Journal of Emotional and Behavioral Disorders, 10, 29–42. https://doi.org/10.1177/106342660201000104 McGuire, T. C., McCormick, K. C., Koch, M. K., & Mendle, J. (2019). Pubertal maturation and trajectories of depression during early adolescence. Frontiers in Psychology, 10, 1362. https://doi.org/10.3389/fpsyg.2019.01362 Murberg, T. A. (2009). Shyness predicts depressive symptoms among adolescents. School Psychology International, 30, 507–519. https://doi.org/10.1177/0143034309107065 Needham, B. L. (2007). Reciprocal relationships between symptoms of depression and parental support during the transition from adolescence to young adulthood. Journal of Youth and Adolescence, 37, 893–905. https://doi.org/10.1007/s10964-007-9181-7 Paikoff, R. L., & Brooks-Gunn, J. (1991). Do parent-child relationships change during puberty? Psychological Bulletin, 110, 47–66. https://doi.org/10.1037/0033-2909.110.1.47 Petersen, A. C., Crockett, L., Richards, M., & Boxer, A. (1988). A self-report measure of pubertal status: Reliability, validity, and initial norms. Journal of Youth and Adolescence, 17, 117–133. Rudolph, K. D., & Troop-Gordon, W. (2010). Personal-accentuation and contextual-amplification models of pubertal timing: Predicting youth depression. Developmental Psychopathology, 22, 433–451. https://doi.org/10.1017/S0954579410000167 Salk, R. H., Hyde, J. S., & Abramson, L. Y. (2017). Gender differences in depression in representative national samples: Meta-analyses of diagnoses and symptoms. Psychological Bulletin, 143, 783–822. https://doi.org/10.1037/bul0000102 Shiffman, S., & Stone, A. A. (1998). Introduction to the special section: Ecological momentary assessment in health psychology. Health Psychology, 17, 3–5. https://doi.org/10.1037/h0092706

Shiffman, S., Stone, A. A., & Hufford, M. R. (2007). Ecological momentary assessment. Annual Review of Clinical Psychology, 4, 1–32. https://doi.org/10.1146/annurev.clinpsy.3.022806.091415 Shirtcliff, E. A., Dahl, R. E., & Pollak, S. D. (2009). Pubertal development: Correspondence between hormonal and physical development. Child Development, 80, 327–337. https://doi.org/10.1111/j.1467-8624.2009.01263.x Shore, L., Toumbourou, J. W., Lewis, A. J., & Kremer, P. (2018). Longitudinal trajectories of child and adolescent depressive symptoms and their predictors–A systematic review and meta‐analysis. Child and Adolescent Mental Health, 23, 107–120. https://doi.org/10.1111/camh.12220 Silk, J. S., Forbes, E. E., Whalen, D. J., Jakubcak, J. L., Thompson, W. K., Ryan, N. D., . . . Dahl, R. E. (2011). Daily emotional dynamics in depressed youth: A cell phone ecological momentary assessment study. Journal of Experimental Child Psychology, 110, 241–257. https://doi.org/10.1016/j.jecp.2010.10.007 Steinberg, L. (1987). Impact of puberty on family relations: Effects of pubertal status and pubertal timing. Developmental Psychology, 23, 451–460. https://doi.org/10.1037/0012-1649.23.3.451 Steinberg, L. (1988). Reciprocal relation between parent-child distance and pubertal maturation. Developmental Psychology, 24, 122–128. https://doi.org/10.1037/0012-1649.24.1.122 Steinberg, L. (2017). Adolescence (11 ed.). New York, NY: McGraw-Hill Education. Suleiman, A. B., & Dahl, R. (2019). Parent–child relationships in the puberty years: Insights from developmental neuroscience. Family Relations, 68, 279–287. https://doi.org/10.1111/fare.12360 Tréepanier, L., Juster, R. P., Marin, M. F., Plusquellec, P., Francois, N., Sindi, S., . . . Lupien, S. (2013). Early menarche predicts increased depressive symptoms and cortisol levels in Quebec girls ages 11 to 13. Developmental Psychopathology, 25, 1017–1027. https://doi.org/10.1017/S0954579413000345 Vannucci, A., Flannery, K. M., & McCauley Ohannessian, C. (2018). Age-varying associations between coping and depressive symptoms throughout adolescence and emerging adulthood. Development Psychopathology, 30, 665–681. https://doi.org/10.1017/S0954579417001183 Vazsonyi, A. T., & Belliston, L. M. (2006). The cultural and developmental significance of parenting processes in adolescent anxiety and depression symptoms. Journal of Youth and Adolescence, 35, 491–505. https://doi.org/10.1007/s10964-006-9064-3 Vogt Yuan, A. S. (2007). Gender differences in the relationship of puberty with adolescents’ depressive symptoms: Do body perceptions matter? Sex Roles, 57, 69–80. https://doi.org/10.1007/s11199-007-9212-6 Wechsler, D. (2011). Wechsler Abbreviated Scale of Intelligence–Second Edition (WASI-II). San Antonio, TX: NCS Pearson. Wilson, S. J., Smyth, J. M., & MacLean, R. R. (2013). Integrating ecological momentary assessment and functional brain imaging methods: New avenues for studying and treating tobacco dependence. Nicotine & Tobacco Research, 16, 102–110. https://dx.doi.org/10.1093%2Fntr%2Fntt129 Winer, J. P., Parent, J., Forehand, R., & Breslend, N. L. (2016). Interactive effects of psychosocial stress and early pubertal timing on youth depression and anxiety: Contextual amplification in family and peer environments. Journal of Child and Family Studies, 25, 1375–1384. https://doi.org/10.1007/s10826-015-0318-0 Wood, A., Knoll, L., Moore, A., & Harrington, R. (1995). Properties of the mood and feelings questionnaire in adolescent psychiatric outpatients: A research note. Journal of Child Psychology and Psychiatry, 36, 327–334. https://doi.org/10.1111/j.1469-7610.1995.tb01828.x Author Note. Danielle Apple, https://orcid.org/00000003-0278-311X, Department of Psychology, University of Pittsburgh; Stefanie Sequiera, https://orcid.org/00000001-8622-8652, Department of Psychology, University of Pittsburgh. Danielle Apple is now at the Department of Psychology at Drexel University, Philadelphia, PA. This study was supported by the National Institute of Mental Health (project number: 5R01MH103241-05) Special thanks to Dr. Jennifer Silk and Psi Chi Journal reviewers for their support. Correspondence concerning this article should be addressed to Danielle Apple, Drexel University, Philadelphia, PA. E-mail: dea49@drexel.edu

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

141

https://doi.org/10.24839/2325-7342.JN25.2.142

Relationships Between Self-Leadership, Psychological Symptoms, and Self-Related Thought in an Undergraduate Sample Sarah A. Myers , Carissa L. Philippi* University of Missouri–St. Louis

, Leah Reyna, and Gregory Dahl

ABSTRACT. The concept of self-leadership was developed in the 1980s as an integral component of Internal Family Systems therapy (IFS). According to IFS, a self-led person can more effectively manage stress and difficult life events. These enhanced coping abilities associated with self-leadership are thought to reduce symptoms of psychiatric disorders. Several studies have investigated self-leadership in connection with workplace behavior, psychological health, and physical health. Few studies have examined the relationship between self-leadership and psychological symptoms in a community sample. Another important question is whether self-leadership is related to other established measures of self-related thought. In the current study, adult undergraduate participants (n = 166) completed self-report measures assessing psychological symptoms and self-related thought. We predicted that there would be a relationship between (H1) self-leadership and fewer psychological symptoms and (H2) self-leadership and lower levels of negative self-reference, but higher positive self-regard. Our findings revealed significant negative relationships between self-leadership and symptoms of depression (p < .001, f 2 = .55), posttraumatic stress disorder (p < .05, f 2 = .03), and social anxiety (p < .001, f 2 = .21). Enhanced self-leadership was also associated with distinct correlations with measures of negative versus positive aspects of self-related cognition. Specifically, self-leadership was associated with lower levels of self-rumination (p = .005, f 2 = .05) and negative self-related responses (p = .002, f 2 = .06), but greater levels of self-reflection (p < .001, f 2 = .12) and positive self-related responses (p = .021, f 2 = .03). These findings provide novel empirical support for a relationship between self-leadership, symptoms of psychological conditions, and measures of self-related thought. Keywords: depression, anxiety, posttraumatic stress disorder, self-leadership, internal family systems

S SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

142

elf-leadership was developed as a tool for clients to moderate their own psychiatric symptoms within Internal Family Systems (IFS) therapy (Schwartz, 2013). IFS therapy defines the self as the central leader and moderator of psychiatric symptoms, whereas self-leadership refers to the ability to comfort, regulate, and improve one’s psychiatric symptoms by achieving a state

of mindfulness and nonjudgmental awareness. Similar to a family dynamic, parts of a personality may cause conflict and extremes in the family system of a person’s inner psyche, and the self cultivates cooperation within the internal ecosystem so as to be a caregiver to traumatized, sad, or resentful parts that result from psychiatric symptoms. The purpose of the present study was to investigate

*Faculty mentor

Myers, Philippi, Reyna, and Dahl | Self-Leadership in an Undergraduate Sample

the relationship between self-leadership and psychiatric symptoms. Foundational to IFS therapy is the notion that self-leadership can reduce the symptoms experienced by many of those patients suffering from mental illness (Schwartz, 1995). Self-leadership is characterized by the traits of compassion, objectivity, nonjudgment, creativity, and calmness. Schwartz described working with patients who experienced severe childhood trauma who, during the course of IFS therapy, were able to instantly access a mindful state that is a hallmark trait associated with the self. This mindful state is an indicator of the presence of self-leadership and is necessary for a client to develop the skills needed to generate their own ability to calm the psychiatric symptoms (Schwartz, 1995; Sweezy & Ziskind, 2013). Schwartz observed in clinical sessions that traits of self-leadership are lower before beginning therapy and associated with greater psychiatric symptoms. Based on these clinical observations, it has been proposed that the enhanced coping abilities associated with adequate self-leadership would lead to a reduction in symptoms of prevalent psychiatric disorders such as anxiety, depression, and posttraumatic stress disorder (PTSD; Schwartz, 1995). These enhanced coping abilities can take the form of mindful observation of extreme emotions, being present with sadness and anger, developing independent internal conflict resolution strategies, and learning to self-soothe (e.g., baths, quiet time, walks). Although IFS therapy is currently used in sev eral countries to treat different psychiatric disorders (e.g., anxiety, depression, PTSD; Anderson, 2013; Schwartz, 2013; Sweezy & Ziskind, 2013; Twombly, 2013; Wonder, 2013), empirical research investigat ing associations between self-leadership and mental health is limited. Previous correlational studies have examined self-leadership in connection with stress, coping styles, physical health, and workplace out comes (Dolbier et al., 2001, 2010), marital problems (Green, 2008), and body dissatisfaction (Bezner et al., 1997). For example, one study found that higher self-leadership was correlated with reduced stress, healthier coping styles, increased optimism, better health, and improved work outcomes (Dolbier et al., 2001). Another study reported positive cor relations between self-leadership and stress-related growth (Dolbier et al., 2010). However, to our knowledge, no studies outside of clinical observa tions have yet examined the relationship between self-leadership and psychological symptoms in a community sample. Given that specific case studies

suggest that IFS therapy may reduce symptoms associated with different psychiatric conditions (Schwartz, 1995; Sweezy & Ziskind, 2013), there is reason to hypothesize that higher levels of selfleadership would be associated with fewer symptoms of anxiety, depression, and PTSD. The self-report questionnaires assessing psychological symptoms used in the present study are directly related to Schwartz’s clinical observations of the presence of anxiety, depression, and PTSD symptoms. In the current study, we tested this hypothesis in a nonclinical undergraduate sample of adults with primarily subclinical levels of self-reported psychi atric symptoms. It is also unknown whether self-leadership is associated with other well-established psychological measures of self-related thought. Several studies have demonstrated relationships between selfreferential cognition and psychiatric conditions, in particular when the self-focused thoughts are con centrated on negative aspects about oneself (e.g., “I am worthless”). Using a variety of different mea sures, a greater bias toward negative self-focused cognition has been consistently documented in depression (Hards et al., 2020; Ingram et al., 1987; Ingram & Smith, 1984; Joormann et al., 2006; Kaiser et al., 2018; Nolen-Hoeksema et al., 2008; Siegle et al., 2004; Smith & Greenberg, 1981) as well as in social anxiety/social phobia in both clinical and nonclinical populations. Consequently, in the pres ent study, we included several self-report scales and an open-ended sentence completion task to assess both negative and positive aspects of self-related thought, including self-consciousness, rumination, self-reflection, and self-focused thought that is either positively or negatively valenced (Ingram et al., 1987; Ingram & Smith, 1984; Woodruff-Borden et al., 2001). Characterizing the relationship between self-leadership and other measures of self-referential cognition could help support the convergent validity of self-leadership and may further refine the construct of self-leadership as defined within IFS. In the current study, we first examined how self-leadership relates to mental health based on self-report measures of depression, anxiety, PTSD, and social anxiety symptoms in a sample of under graduates. Second, we investigated the relationship between self-leadership and positive and negative features of self-related thought using self-report measures and a sentence completion task. The hypotheses were twofold: (a) Self-leadership was expected to negatively correlate with psychological

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

143

Self-Leadership in an Undergraduate Sample | Myers, Philippi, Reyna, and Dahl

symptoms, and (b) self-leadership was expected to be associated with lower levels of negative self-focus (e.g., rumination), but higher levels of positive self-reference (e.g., self-reflection).

Method Participants and Procedure We collected data from undergraduate students ages 18 to 61 years of age (n = 166; 27 male/131 female) enrolled in psychology courses at a Midwestern University. The average (SD) age of participants was 23 years old (7.1). Participants had varying levels of education including some college (20.2%), associate’s degree (16.5%), bachelor’s degree (18.6%), professional degree (17.6%), and doctoral degree (2.1%). Greater than half of par ticipants were European American/White (56%), approximately one third of participants were African American/Black (32%), and the remain ing participants were Hispanic/Latino (4.8%), Asian/Pacific Islander (3%), and other (6.6%). All participants gave informed consent according to a protocol approved by the Institutional Review Board. All students received course credit for their participation in the study. See Table 1 for all study variables (described below). Measures Self-leadership scale. This scale (Steinhardt et al., 2003) was used to measure self-leadership as TABLE 1 Study Variables Variable Self-Leadership

M(SD)

71.7(15.8)

.445

Depression

12.1(11.3)

.387

Anxiety

13.2(11.7)

.001

PTSD

22.2(18.6)

.512

Social Anxiety

10.2 (3.4)

.003

Private Self-Consciousness (SCSR)

17.9 (4.8)

.151

Public Self-Consciousness (SCSR)

14.0 (3.4)

.114

Self-Rumination (RRQ)

43.3 (9.4)

.041

Self-Reflection (RRQ)

Psychological Symptom Measures

Self-Related Thought Measures

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

144

41.0 (8.8)

.016

Negative Self-Related Responses (SCT)

0.17(0.15)

.408

Positive Self-Related Response (SCT)

0.34(0.17)

.128

Note. PTSD = posttraumatic stress disorder; SCSR = Self-Consciousness Scale-Revised; RRQ = Rumination-Reflection Questionnaire; SCT = Self-Focus Sentence Completion Task. a Differences between female (n = 131) and male (n = 27) participants were evaluated for all variables using one-sample t tests and p values are reported.

defined by IFS. This scale has 20 items measuring qualities that are considered hallmarks of self-lead ership including calmness, confidence, creativity, courage, and compassion. Participants rate each item on a scale from 1 (almost never/never occurs) to 5 (almost always/always occurs). Some example items include, “I feel a sense of inner peace” or “I treat myself with kindness.” Higher scores on this scale indicate higher levels of self-leadership (e.g., optimal psychological well-being, mindfulness, nonjudgmental awareness). The internal consistency of the self-leadership scale in the present sample was high (Cronbach’s α = .94). The Beck Depression Inventory-II. This scale (BDI-II; Beck et al., 1996) was chosen to measure self-reported symptoms of depression. The BDI-II is one of the most widely used tools to assess depres sion symptoms (Beck et al., 1996). Psychometric studies have provided support for good convergent and discriminant validity of the BDI-II (Schotte et al., 1997; Steer et al., 1997). The BDI-II consists of 21 items that assess depressive symptoms such as worthlessness, loss of energy, and fatigue. Each item is rated on a scale from 0 (an absence of symptoms) to 3 (maximum severity). Example options for worthless ness item range from “I do not feel I am worthless” to “I feel utterly worthless.” The scores across all items were summed up to calculate the total depres sion score for each participant. Total depression scores can range from 0 to 63, with higher scores indicating greater severity of depressive symptoms. The internal consistency of the BDI-II in the present sample was high (Cronbach’s α = .94). The Beck Anxiety Inventory. This scale (BAI; Beck & Steer, 1990) was selected to measure selfreported anxiety symptoms because it has evidence for adequate convergent and discriminant validity (Fydrich et al., 1992; Steer et al., 1993). The BAI consists of 21 items that measure situations, sensa tions, and thoughts that are associated with anxiety. Items on the BAI are rated on a scale from 0 (not at all) to 3 (severely). Some example items include “numbness or tingling” and “difficulty in breath ing.” Scores across all items were then summed up to compute a total anxiety score. The anxiety scores can range from 0 to 63, and higher scores indicate greater anxiety. The internal consistency of the BAI in the present sample was high (Cronbach’s α = .93). PTSD Checklist for DSM-5. The checklist (PCL-5; Weathers et al., 2013) is a 20-item question naire used to assess PTSD symptoms (Blevins et al., 2015), including re-experiencing, avoidance, negative alterations in cognition and mood, and

Myers, Philippi, Reyna, and Dahl | Self-Leadership in an Undergraduate Sample

hyperarousal. Participants rate how much they were bothered by their symptoms in the past month from 1 (not at all) to 5 (extremely). Some examples include “repeated, disturbing dreams of the stress ful experience” and “avoiding memories, thoughts, or feelings related to the stressful experience.” Psychometric research suggests that the PCL-5 has good internal consistency (Cronbach’s α = .96), test-retest reliability (r = .84), and convergent and discriminant validity (Bovin et al., 2016). For the present study, we focused on total PTSD severity scores, which were calculated by summing the scores across all items. The internal consistency of the PTSD checklist for DSM-5 in the present sample was high (Cronbach’s α = .96). Self-Consciousness Scale-Revised. This scale (SCSR; Scheier & Carver, 1985) is a 23-item selfreport questionnaire designed to measure thoughts and behaviors related to personal insight and self-focused attention. Psychometric analysis of this questionnaire has revealed adequate internal consistency (Cronbach’s α = .75-.84), test-retest reliability (.74-.77), and evidence for convergent and discriminant validity (Carver & Glass, 1976; Scheier & Carver, 1985; Turner et al., 1978). Some example items include “I’m quick to notice changes in my mood” and “I think about myself a lot.” For each item, participants used a 4-point rating scale ranging from 0 (not at all like me) to 3 (a lot like me). Separate scores were computed for the three subscales of the self-consciousness scale used most frequently in clinical and subclinical populations (Hope & Heimberg, 1988; Ingram & Smith, 1984; Jostes et al., 1999): Public Self-Consciousness, Private Self-Consciousness, and Social Anxiety. Higher scores on the SCSR indicate greater levels of public self-consciousness, private self-consciousness, and social anxiety. The internal consistency of the SCSR in this sample was adequate for public self-consciousness (Cronbach’s α = .80), private self-consciousness (Cronbach’s α = .75), and social anxiety (Cronbach’s α = .85) and comparable to previous studies (Scheier & Carver, 1985). The Public and Private Self-Consciousness subscales were included in self-related thought measures whereas the Social Anxiety subscale was included in the psychopathology measures. Rumination-Reflection Questionnaire. This questionnaire (Trapnell & Campbell, 1999) is a 28-item self-report questionnaire assessing aspects of ruminative thought or negative self-related thought patterns (e.g., “Often I’m playing back over in my mind how I acted in a past situation”) and selfreflection (e.g., “I love exploring my inner self”).

Participants rated each item on a scale ranging from 1 (strongly disagree) to 5 (strongly agree). As in previous studies, scores were calculated separately for Self-Rumination and Self-Reflection subscales (Trapnell & Campbell, 1999). Psychometric research indicates that these subscales have good convergent and discriminant validity (Trapnell & Campbell, 1999). In the current sample, the inter nal consistency was high for the Self-rumination (Cronbach’s α = .92) and Self-reflection subscales (Cronbach’s α = .87). Self-Focus Sentence Completion Task. This measure assessed self-focus and researchers have provided evidence for its reliability and validity based on six validation studies using large normative and clinical samples (Exner, 1973). For the sentence completion task, participants were given 30 different sentence stems to complete (e.g., “I think…” or “My father…”). Participants were instructed to complete each sentence as they wished, with no other instruction regarding the content of their responses. As in previous research (Exner, 1973), each participant’s response was scored based on four focus categories: (a) self-focused (e.g., I think… “therefore I am.”), (b) other-focused (e.g., It upsets me when… “the Cardinals do not win.”), (c) self- and other-focused (e.g., If only I would… “have enough money to support myself and my family.”), or (d) non-person-focused (e.g., When I look in the mirror… “I see a reflection.”). We also coded each response for overall valence (as in Ingram & Smith, 1984), including positive, negative, and neutral, which resulted in focusby-valence categories. Total scores corresponded to the sum of all responses for each response category. Four pairs of two raters, blinded to the other behavioral data, were trained in sentence completion task coding in two steps. The raters first separately scored all responses for focus and valence. Using a two-way random effects intraclass correlation coefficient, interrater reliability was calculated for these initial ratings for this group and adequate interrater reliability was found for each response category included in the analyses (self-negative responses = .85; self-positive responses = .87; reliability was averaged across raters), which was within acceptable limits (Exner, 1973). Second, the raters conferred and agreed on a final code for each response (e.g., self-focused and positive valence). The final codes for the total proportion of self-negative and self-positive responses were used in the present study (e.g., total number of self-focused negative responses / total self-focused responses).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

145

Self-Leadership in an Undergraduate Sample | Myers, Philippi, Reyna, and Dahl

Data Analysis We used SPSS 25 for all data analysis. We performed two separate multiple regression analyses to examine the association between self-leadership and measures of psychopathology and self-related thought. For each regression, we performed outlier tests using established tests of influence in regres sion with the following criteria: Cooks distance > 1 (Cook & Weisberg, 1982), leverage > .13 (Stevens, 2012), and Mahalanobis distance > 15 (Barnett & Lewis, 1994). Based on these tests, number of outli ers varied between 0, 1, 4, 5, or 6 depending on the regression. We deleted the participant outlier score only for that specific regression analysis as opposed to using listwise deletion for outliers. Demographic information was not available for all participants (missing age for n = 25; gender for n = 8; education for n = 15; ethnicity for n = 15), therefore analyses reported below controlling for demographic variables did not include these partici pants. A few participants (n = 5) were also missing for depression, anxiety, and PTSD measures. In this case, these participants were not included in the analyses with these psychopathology measures, but they were included in the other analyses.

Results Demographic Information Age was significantly correlated with self-reflection, r = .31, p < .001, but not with any other psycho logical symptoms or self-related thought variables (ps = .28–.95). Differences between female and male participants were present for anxiety, t(154) = -3.87, p = .001, social anxiety, t(156) = -3.07, p = .003, self-rumination, t(153) = -2.06, p = .041, and self-reflection variables, t(155) = -3.87, p = .001. There was a significant difference in education level for positive, F(6,142) = 4.33, p < .001, and negative, F(6,142) = 7.48, p < .001, self-related responses on the sentence completion task, but not for any other variables (ps = .11–.98). There was a sig nificant difference in ethnicity for self-leadership, F(4, 144) = 2.88, p = .025, but not for any other variables (ps = .13–.75). As a result of these analyses and outlier tests, we also report the results of two follow-up multiple regression analyses controlling for age, gender, education, and ethnicity and excluding outliers in all analyses below. SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

146

Self-Leadership and Psychological Symptoms Consistent with our first hypothesis, participants who scored higher on self-leadership had signifi cantly fewer psychological symptoms across almost

all measures including self-report symptoms of depression, = -.58, t(156) = -9.34, p < .001, f 2 = .55, PTSD, = -.15, t(156) = -2.16, p < .05, f 2 = .03, and social anxiety, = -.30, t(156) = -5.73, p < .001, f 2 = .21. Self-leadership was not significantly related to anxiety symptoms, = .06, t(156) = 1.00, p = .32. Collinearity statistics for all variables were within normal limits: VIF = 1.1–1.9, tolerance = .5–.9. All results remained significant after controlling for age, gender, education, and ethnicity and excluding outliers (see Table 2). Self-Leadership and Measures of Self-Related Thought In line with our second hypothesis, self-leadership was inversely correlated with almost all measures assessing negative aspects of self-related thought, and positively correlated with measures assessing positive aspects of self-related thought. Specifically, higher levels of self-leadership were associated with lower levels of self-rumination, = -.24, t(159) = -2.83, p = .005, f 2 = .05, and negative self-related responses on the sentence completion task, = -.21, t(159) = -3.09, p = .002, f 2 = .06. On the other hand, higher levels of self-leadership were associated with greater levels of self-reflection, = .33, t(159) = 4.29, p < .001, f 2 = .12, and positive self-related responses on the sentence completion task, = .16, t(159) = 2.34, p = .021, f 2 = .03. There was no significant relationship between self-leadership and private self-consciousness, = -.06, t(159) = -0.67, p = .51, or public self-consciousness, = -.04, t(159) = -0.51, p = .61. Collinearity statistics for all variables were within normal limits: VIF = 1.1-1.9, tolerance = .5-.9. After controlling for all demo graphic variables and removing outliers, all results remained significant (see Table 3).

Discussion In the present study, we investigated the con nections between self-leadership, psychological symptoms, and self-related thought in a nonclinical setting. Our results supported our hypotheses. First, we found that higher levels of self-leadership were associated with less severe self-reported symptoms of depression, PTSD, and social anxiety even after controlling for age, gender, education, and ethnic ity. Second, we also demonstrated that higher levels of self-leadership were correlated with reduced negative self-related thought as well as greater positive self-reference. Our results indicating that higher levels of

Myers, Philippi, Reyna, and Dahl | Self-Leadership in an Undergraduate Sample

self-leadership were correlated with diminished psychological symptoms are consistent with past research on self-leadership in nonclinical samples showing enhanced psychological functioning (e.g., healthy coping styles), reduced stress, greater perceived wellness, and stress-related resilience (Bezner et al., 1997; Dolbier et al., 2001, 2010). To our knowledge, this was the first empirical study to reveal a (negative) correlation between the self-leadership scale and psychological symptoms of depression, PTSD, and social anxiety in a nonclinical sample. Importantly, these findings remained significant after controlling for age, gender, education, and ethnicity. Broadly, our results are consistent with the IFS model, in which appropriate self-leadership as cultivated through IFS therapy is thought to be associated with reduced symptoms in patients with mental health conditions (Anderson, 2013; Schwartz, 1995, 2013; Sweezy & Ziskind, 2013; Twombly, 2013; Wonder, 2013). However, the present findings were correlational and based on a nonclinical sample of individuals with lower levels of psychological symptoms. As a result, replication in a clinical sample will be neces sary. A recent randomized control trial suggested that IFS may be associated with sustained increases in self-compassion and decreases in depressive symptoms (Shadick et al., 2013). This is relevant to our findings because the largest effect size we found was for the analysis relating self-leadership to depressive symptoms. However, Shadick et al. did not assess changes in self-leadership. Longitudinal treatment studies will be necessary to investigate whether IFS therapy increases self-leadership and diminishes psychiatric symptoms in clinical samples and how IFS differs from other forms of therapy in promoting self-leadership. When comparing self-leadership with other measures of self-related thought, we identified distinct relationships with both negative and posi tive aspects of self-referential cognition. Specifically, self-leadership was inversely correlated with lower levels of self-rumination after covarying for demo graphic factors. These findings are consistent with a previous study in undergraduate students reporting a negative correlation between self-leadership and ineffectiveness or the tendency to perceive oneself as inadequate, worthless, and insecure (Dolbier et al., 2001). By contrast, greater self-leadership was positively correlated with increased positive self-ref erence on the sentence completion task. Our results mirror prior work demonstrating relationships between self-leadership and increased optimism

(Dolbier et al., 2001), physical self-esteem (Bezner et al., 1997), and stress-related growth (Dolbier et al., 2010). Given that self-leadership is comprised of qualities such as compassion and confidence, it is not surprising that greater self-leadership would be related to both diminished negative self-reference and increased positive self-perception. Our findings are also relevant to compassion-focused therapy, which includes training individuals to cultivate self-compassion (Gilbert, 2009; Gilbert & Irons, 2018). Similar to research on self-leadership, greater self-compassion has been associated with several benefits to psychological and physical health (see Bluth & Neff, 2018, for review). Together, these findings also provide support for convergent validity of the self-leadership scale as a measure of positive aspects of self-referential cognition. This is particularly relevant to IFS therapy, which seeks TABLE 2 Multiple Regression Analyses Between Self-Leadership and Psychological Symptoms Controlling for Demographic Variables Psychological Symptom Measures

SE B

-.58

.10

-.45

<.001

Anxiety

.06

.08

.06

.46

PTSD

-.12

.06

-.19

.03

-1.27

.26

-.34

<.001

Depression

Social Anxiety

Note. PTSD = posttraumatic stress disorder. a In a follow-up multiple regression analysis controlling for age, gender, education, and ethnicity and excluding outliers, relationships between self-leadership and all psychological symptom measures remained significant, except for anxiety. Collinearity statistics were within the normal limits: VIF = 1.1–2.0, tolerance = .5–.9. Participants without age, gender, education, or ethnicity information were not included in the multiple regression analyses.

TABLE 3 Multiple Regression Analyses Between Self-Leadership and Self-Related Thought Controlling for Demographic Variables Self-Related Thought Measures

SE B

Private Self-Consciousness (SCSR)

-.04

.31

-.01

.89

Public Self-Consciousness (SCSR)

-.20

.31

-.07

.52

Self-Rumination (RRQ)

-.48

.14

-.35

.001

Self-Reflection (RRQ)

.40

.16

.25

.01

Negative Self-Related Responses (SCT)

-16.58

7.83

-.19

.04

Positive Self-Related Responses (SCT)

20.93

6.75

.26

.002

Note. SCSR = Self-Consciousness Scale-Revised; RRQ = Rumination-Reflection Questionnaire; SCT = Sentence Completion Task. a In a follow-up multiple regression analysis controlling for age, gender, education, and ethnicity and excluding outliers, all relationships remained significant, except for private and public selfconsciousness. Collinearity statistics were within the normal limits: VIF = 1.2–2.3, tolerance = .4–.8. Participants without age, gender, education, or ethnicity information were not included in the multiple regression analyses.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

147

Self-Leadership in an Undergraduate Sample | Myers, Philippi, Reyna, and Dahl

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

148

to cultivate a healthy and mindful self (Schwartz, 1995; Sweezy & Ziskind, 2013). In future studies, it would be interesting to explore whether established measures of self-related thought also change after successful treatment with IFS therapy. Our results suggest that self-leadership is not entirely explained by the self-referential cognition measures. The associations between self-leadership and self-rumination and positive self-related responses on the sentence completion task revealed small effect sizes (f 2 < .05). More generally, these measures of self-related thought do not appear to be overlapping with all of the characteristics of self-leadership. Given that selfleadership is defined by calmness, confidence, creativity, courage, and compassion, it is possible that self-leadership may be more highly correlated with measures that specifically assess those traits. Relatedly, cultivating self-leadership relies in part on engaging cognitive skills associated with mindfulness, including the ability for the self to control the more undesirable aspects of one’s personality and psychiatric symptoms (Schwartz, 1995). Further, mindfulness based stress reduction training and greater trait mindfulness have been associated with reduced psychological symptoms across different psychiatric disorders, including depression, anxiety, PTSD, and social anxiety (Boyd et al., 2018; Chi et al., 2018; Hjeltnes et al., 2018; Hoge et al., 2013; Paul et al., 2012). As such, measures of trait mindfulness may provide stronger support for the convergent validity of the self-leadership scale. However, additional research will be needed to further establish the convergent and discriminant validity of this scale. There are limitations to this study that should be mentioned. First, our study used an under graduate sample with self-reported psychological symptoms not assessed by a clinical psychologist and with relatively low levels of symptoms, possibly limiting the generalizability of the findings. Further studies will be critical to determine whether these findings replicate in other populations such as middle-aged or older adults as well as individuals diagnosed with psychiatric conditions. Second, there are several possible confounding variables such as sleep quality, exercise, and drug use that were not measured in the present study. Given that our sample was comprised of undergraduate students, these factors might have influenced our findings (Fox, 2000; Singleton & Wolfson, 2009). For example, given that increased alcohol

consumption and poor sleep quality have been associated with worse academic performance in undergraduate samples (e.g., Singleton & Wolfson), these same factors may have a general impact on ratings for the self-report measures. It is also possible that regular physical exercise may increase positive self-regard because exercise has been associated with greater self-esteem (Fox, 2000). Third, we had fewer male participants in our sample, which may limit our ability to generalize these results. Fourth, we did not examine measures that were entirely distinct from self-leadership, which could more precisely assess discriminant validity. Thus, additional research should inves tigate discriminant validity of the self-leadership scale in more detail. Fifth, we focused only on self-leadership and not the various components of the IFS theory, such as the distinction between self and parts. Additional research is warranted to examine whether measurement of the parts or subpersonalities described in IFS theory also align with psychological research related to self and personality. Future studies may also help clarify the similarities and differences between self-leadership defined within IFS and the concept of self-leadership defined in industrial organi zational psychology as a form of leadership that challenges common notions of how followers take after workplace setting supervisors (Shek et al., 2015). Self-leadership as defined in industrial organizational psychology involves leading oneself to perform work in a timely manner and to provide internal motivation to carry out workplace respon sibilities (Manz, 1986; cf. Stewart et al., 2019). Effective self-leadership has been associated with better workplace outcomes, enhanced socializa tion for new employees, and favorable personality characteristics (Amundsen & Martinsen, 2015; Cranmer et al., 2019; Ho & Nesbit, 2018; Stewart et al., 2019). Conclusion In sum, we provided novel empirical evidence for a connection between the concept of selfleadership in IFS and symptoms of depression, PTSD, and social anxiety in a nonclinical sample. We also found evidence for convergent validity between self-leadership and traditional measures of self-related thought. Although these results await replication in a clinical sample, these find ings suggest that self-leadership may be useful as a dimensional measure of more positive aspects of self-referential cognition.

Myers, Philippi, Reyna, and Dahl | Self-Leadership in an Undergraduate Sample

References Amundsen, S., & Martinsen, Ø. L. (2015). Linking empowering leadership to job satisfaction, work effort, and creativity: The role of self-leadership and psychological empowerment. Journal of Leadership & Organizational Studies, 22, 304–323. https://doi.org/10.1177/1548051814565819 Anderson, F. G. (2013). ‘Who’s taking what?’ Connecting neuroscience, psychopharmacology and internal family systems for trauma. In M. Sweezy & E. L. Ziskind (Eds.), Internal Family Systems Therapy (pp. 107–126). New York, NY: Routledge. Barnett, V., & Lewis, T. (1994). Outliers in statistical data (3rd ed.). Wiley. Beck, A. T., & Steer, R. A. (1990). Manual for the Beck Anxiety Inventory. Psychological Corporation. Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck Depression Inventory-II. Psychological Corporation. Bezner, J. R., Adams, T. B., & Steinhardt, M. A. (1997). Relationship of body dissatisfaction to physical health and wellness. American Journal of Health Behavior, 21, 147–155. Blevins, C. A., Weathers, F. W., Davis, M. T., Witte, T. K., & Domino, J. L. (2015). The Posttraumatic Stress Disorder Checklist for DSM‐5 (PCL‐5): Development and initial psychometric evaluation. Journal of Traumatic Stress, 28, 489–498. https://doi.org/10.1002/jts.22059 Bluth, K., & Neff, K. D. (2018). New frontiers in understanding the benefits of self-compassion. Self and Identity, 17, 605–608. https://doi.org/10.1080/15298868.2018.1508494 Bovin, M. J., Marx, B. P., Weathers, F. W., Gallagher, M. W., Rodriguez, P., Schnurr, P. P., & Keane, T. M. (2016). Psychometric properties of the PTSD checklist for diagnostic and statistical manual of mental disorders–fifth edition (PCL-5) in veterans. Psychological Assessment, 28, 1379. https://doi.org/10.1037/pas0000254 Boyd, J. E., Lanius, R. A., & McKinnon, M. C. (2018). Mindfulness-based treatments for posttraumatic stress disorder: A review of the treatment literature and neurobiological evidence. Journal of Psychiatry & Neuroscience, 43, 7–25. https://doi.org/10.1503/jpn.170021 Carver, C. S., & Glass, D. C. (1976). The Self-Consciousness Scale: A discriminant validity study. Journal of Personality Assessment, 40, 169–172. https://doi.org/10.1207/s15327752jpa4002_8 Chi, X., Bo, A., Liu, T., Zhang, P., & Chi, I. (2018). Effects of mindfulness-based stress reduction on depression in adolescents and young adults: A systematic review and meta-analysis. Frontiers in Psychology, 9, 1034. https://doi.org/10.3389/fpsyg.2018.01034 Cook, R. D., & Weisberg, S. (1982). Residuals and influence in regression. New York, NY: Chapman and Hall. Cranmer, G. A., Goldman, Z. W., & Houghton, J. (2019). I’ll do it myself: Selfleadership, proactivity, and socialization. Leadership & Organization Development Journal, 40, 684–698. https://doi.org/10.1108/LODJ-11-2018-0389 Dolbier, C. L., Jaggars, S. S., & Steinhardt, M. A. (2010). Stress-related growth: Pre-intervention correlates and change following a resilience intervention. Stress and Health, 26, 135–147. https://doi.org/10.1002/smi.1275 Dolbier, C. L., Soderstrom, M., & Steinhardt, M. A. (2001). The relationships between self-leadership and enhanced psychological, health, and work outcomes. The Journal of Psychology, 135, 469–485. https://doi.org/10.1080/00223980109603713 Exner, J. E. (1973). The self focus sentence completion: A study of egocentricity. Journal of Personality Assessment, 37, 437–455. https://doi.org/10.1080/00223891.1973.10119902 Fox, K. R. (2000). The effects of exercise on self‐perceptions and self‐esteem. In S. J. H. Biddle, K. R. Fox, & S. H. Boutcher (Eds.), Physical activity and psychological well-being. New York, NY: Routledge. Fydrich, T., Dowdall, D., & Chambless, D. L. (1992). Reliability and validity of the Beck Anxiety Inventory. Journal of Anxiety Disorders, 6, 55–61. https://doi.org/10.1016/0887-6185(92)90026-4 Gilbert, P. (2009). Introducing compassion-focused therapy. Advances in Psychiatric Treatment, 15, 199–208. Cambridge Core. https://doi.org/10.1192/apt.bp.107.005264 Gilbert, P., & Irons, C. (2018). Compassion focused therapy. In S. Palmer (Ed.), The beginner’s guide to counseling and psychotherapy (pp. 127–139). Thousand Oaks, CA: Sage. Green, E. J. (2008). Individuals in conflict: An internal family systems approach. The Family Journal, 16, 125–131. https://doi.org/10.1177/1066480707313789 Hards, E., Ellis, J., Fisk, J., & Reynolds, S. (2020). Negative view of the self and

symptoms of depression in adolescents. Journal of Affective Disorders, 262, 143–148. https://doi.org/10.1016/j.jad.2019.11.012 Hjeltnes, A., Moltu, C., Schanche, E., Jansen, Y., & Binder, P.-E. (2018). Both sides of the story: Exploring how improved and less-improved participants experience mindfulness-based stress reduction for social anxiety disorder. Psychotherapy Research, 28, 106–122. https://doi.org/10.1080/10503307.2016.1169330 Ho, J., & Nesbit, P. L. (2018). Personality and work outcomes: A moderated mediation model of self-leadership and gender. International Journal of Management Excellence, 10, 1292–1304. https://doi.org/10.17722/ijme.v10i2.416 Hoge, E. A., Bui, E., Marques, L., Metcalf, C. A., Morris, L. K., Robinaugh, D. J., . . . Simon, N. M. (2013). Randomized controlled trial of mindfulness meditation for generalized anxiety disorder: Effects on anxiety and stress reactivity. The Journal of Clinical Psychiatry, 74, 786–792. https://doi.org/10.4088/ JCP.12m08083 Hope, D. A., & Heimberg, R. G. (1988). Public and private self-consciousness and social phobia. Journal of Personality Assessment, 52, 626–639. https://doi.org/10.1207/s15327752jpa5204_3 Ingram, R. E., Lumry, A. E., Cruet, D., & Sieber, W. (1987). Attentional processes in depressive disorders. Cognitive Therapy and Research, 11, 351–360. https://doi.org/10.1007/BF01186286 Ingram, R. E., & Smith, T. W. (1984). Depression and internal versus external focus of attention. Cognitive Therapy and Research, 8, 139–151. https://doi.org/10.1007/BF01173040 Joormann, J., Dkane, M., & Gotlib, I. H. (2006). Adaptive and maladaptive components of rumination? Diagnostic specificity and relation to depressive biases. Behavior Therapy, 37, 269–280. https://doi.org/10.1016/j.beth.2006.01.002 Jostes, A., Pook, M., & Florin, I. (1999). Public and private self-consciousness as specific psychopathological features. Personality and Individual Differences, 27, 1285–1295. https://doi.org/10.1016/S0191-8869(99)00077-X Kaiser, R. H., Snyder, H. R., Goer, F., Clegg, R., Ironside, M., & Pizzagalli, D. A. (2018). Attention bias in rumination and depression: Cognitive mechanisms and brain networks. Clinical Psychological Science, 6, 765–782. https://doi.org/10.1177/2167702618797935 Manz, C. C. (1986). Self-leadership: Toward an expanded theory of self-influence processes in organizations. Academy of Management Review, 11, 585–600. https://doi.org/10.5465/amr.1986.4306232 Modini, M., Rapee, R. M., & Abbott, M. J. (2018). Processes and pathways mediating the experience of social anxiety and negative rumination. Behaviour Research and Therapy, 103, 24–32. https://doi.org/10.1016/j.brat.2018.01.009 Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). Rethinking rumination. Perspectives on Psychological Science, 3, 400–424. https://doi.org/10.1111/j.1745-6924.2008.00088.x Paul, N. A., Stanton, S. J., Greeson, J. M., Smoski, M. J., & Wang, L. (2012). Psychological and neural mechanisms of trait mindfulness in reducing depression vulnerability. Social Cognitive and Affective Neuroscience, 8, 56–64. https://doi.org/10.1093/scan/nss070 Saboonchi, F., Lundh, L.-G., & Öst, L.-G. (1999). Perfectionism and selfconsciousness in social phobia and panic disorder with agoraphobia. Behaviour Research and Therapy, 37, 799–808. https://doi.org/10.1016/S0005-7967(98)00183-1 Scheier, M. F., & Carver, C. S. (1985). The Self‐Consciousness Scale: A revised version for use with general populations. Journal of Applied Social Psychology, 15, 687–699. https://doi.org/10.1111/j.1559-1816.1985.tb02268.x Schotte, C. K. W., Maes, M., Cluydts, R., De Doncker, D., & Cosyns, P. (1997). Construct validity of the Beck Depression Inventory in a depressive population. Journal of Affective Disorders, 46, 115–125. https://doi.org/10.1016/S0165-0327(97)00094-3 Schwartz, R. C. (1995). Internal Family Systems Therapy. New York, NY: Guilford Press. Schwartz, R. C. (2013). Moving from acceptance toward transformation with Internal Family Systems Therapy (IFS). Journal of Clinical Psychology, 69, 805–816. https://doi.org/10.1002/jclp.22016 Shadick, N. A., Sowell, N. F., Frits, M. L., Hoffman, S. M., Hartz, S. A., Booth, F. D., . . . Schwartz, R. C. (2013). A randomized controlled trial of an internal family systems-based psychotherapeutic intervention on outcomes in rheumatoid arthritis: A proof-of-concept study. The Journal of Rheumatology, 40, 1831. https://doi.org/10.3899/jrheum.121465 Shek, D. T. L., Ma, C. M. S., Liu, T. T., & Siu, A. M. H. (2015). The role of selfleadership in service leadership. International Journal on Disability and

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

149

Self-Leadership in an Undergraduate Sample | Myers, Philippi, Reyna, and Dahl

Human Development, 14, 343. https://doi.org/10.1515/ijdhd-2015-0455 Siegle, G. J., Moore, P. M., & Thase, M. E. (2004). Rumination: One construct, many features in healthy individuals, depressed individuals, and individuals with lupus. Cognitive Therapy and Research, 28, 645–668. https://doi.org/10.1023/B:COTR.0000045570.62733.9f Singleton, R. A., & Wolfson, A. R. (2009). Alcohol consumption, sleep, and academic performance among college students. Journal of Studies on Alcohol and Drugs, 70, 355–363. https://doi.org/10.15288/jsad.2009.70.355 Smith, T. W., & Greenberg, J. (1981). Depression and self-focused attention. Motivation and Emotion, 5, 323–331. https://doi.org/10.1007/BF00992551 Steer, R. A., Ball, R., Ranieri, W. F., & Beck, A. T. (1997). Further evidence for the construct validity of the Beck Depression Inventory-II with psychiatric outpatients. Psychological Reports, 80, 443–446. https://doi.org/10.2466/pr0.1997.80.2.443 Steer, R. A., Ranieri, W. F., Beck, A. T., & Clark, D. A. (1993). Further evidence for the validity of the Beck Anxiety Inventory with psychiatric outpatients. Journal of Anxiety Disorders, 7, 195–205. https://doi.org/10.1016/0887-6185(93)90002-3 Steinhardt, M., Dolbier, C., Mallon, M., & Adams, T. (2003). The development and validation of a scale for measuring self-leadership. Journal of SelfLeadership, 1, 20–31. Stevens, J. P. (2012). Applied multivariate statistics for the social sciences (5th ed.). New York, NY: Routledge. Stewart, G. L., Courtright, S. H., & Manz, C. C. (2019). Self-leadership: A paradoxical core of organizational behavior. Annual Review of Organizational Psychology and Organizational Behavior, 6, 47–67. https://doi.org/10.1146/annurev-orgpsych-012218-015130 Sweezy, M., & Ziskind, E. L. (2013). Internal family systems therapy— New dimensions. New York, NY: Routledge. Trapnell, P. D., & Campbell, J. D. (1999). Private self-consciousness and the fivefactor model of personality: Distinguishing rumination from reflection. Journal of Personality and Social Psychology, 76, 284–304.

https://doi.org/10.1037//0022-3514.76.2.284 Turner, R. G., Carver, C. S., Scheier, M. F., & Ickes, W. (1978). Correlates of selfconsciousness. Journal of Personality Assessment, 42, 285–289. https://doi.org/10.1207/s15327752jpa4203_10 Twombly, J. H. (2013). Integrating IFS with phase-oriented treatment of clients with dissociative disorder. In M Sweezy & E. L. Ziskind (Eds.), Internal Family Systems Therapy-New dimensions (pp. 72–89). Routledge. Weathers, F. W., Litz, B. T., Keane, T. M., Palmieri, P. A., Marx, B. P., & Schnurr, P. P. (2013). The PTSD Checklist for DSM-5 (PCL-5). Scale available from the National Center for PTSD. www.ptsd.va.gov Wonder, N. (2013). Treating pornography addiction with IFS. In Martha Sweezy & E. L. Ziskind (Eds.), Internal Family Systems Therapy-New dimensions (pp. 159–165). New York, NY: Routledge. Woodruff-Borden, J., Brothers, A. J., & Lister, S. C. (2001). Self-focused attention: Commonalities across psychopathologies and predictors. Behavioural and Cognitive Psychotherapy, 29, 169–178. Cambridge Core. https://doi.org/10.1017/S1352465801002041 Author Note. Sarah A. Myers, https://orcid.org/0000-00032605-0877, Department of Psychological Sciences, University of Missouri–St. Louis; Carissa L. Philippi, https://orcid.org/0000-0003-3741-3661, Department of Psychological Sciences, University of Missouri–St. Louis; Leah Reyna, Department of Psychological Sciences, University of Missouri–St. Louis; Gregory Dahl, Center for Behavioral Health, University of Missouri–St. Louis. This study was supported by the University of Missouri– St. Louis Undergraduate Research Grant Program. Correspondence concerning this article should be addressed to Carissa L. Philippi, Department of Psychological Sciences, University of Missouri–St. Louis, St. Louis, MO 63121. E-mail: philippic@umsl.edu

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

150

https://doi.org/10.24839/2325-7342.JN25.2.151

Addictive Technology: Prevalence and Potential Implications of Problematic Social Media Use Chloe Tanega and Andrew Downs* University of Portland

ABSTRACT. College students spend a significant amount of time using social media, and there is evidence that at least some of the rapid growth in social media use may be driven by social media companies’ efforts to implement behavioral engineering strategies designed to maximize the amount of time individuals spend on their platforms. The purpose of the present study was to determine whether such behavioral engineering strategies are leading individuals to become addicted to social media and to examine whether those who report problematic social media use (PSMU) may be at risk for mental health problems. Two-hundred ninety-four college students completed an online survey asking about indicators of PSMU, mental health symptoms, and well-being. Depending on the cut score used, between 8.2% and 51.3% of college students may be at risk for PSMU. No matter which cut score was used, participants identified as being at risk for PSMU reported higher levels of mental health symptoms across several domains, as well as lower well-being. These results suggest that individuals who spend time using social media platforms may be at risk for PSMU and highlight the need for clinicians and researchers to establish empirically based diagnostic criteria, as well as effective treatments, for PSMU. Keywords: social media, assessment, addictive behavior

ocial media is fully integrated into society today, and the use of social media is currently the main activity of Internet-users (JassoMedrano & López-Rosales, 2018). With the advent of smartphones and the wide variety of platforms available, individuals can now seamlessly switch from site to site when browsing online. Indeed, Hardy and Castonguay (2018) found that millennials switch between media sites an average of 27 times per hour. Currently, the most popular social media platforms are Facebook and Instagram. As of 2017, Facebook had approximately 2 billion monthly users, which corresponds to over 25% of the world’s population (Østergaard, 2017). Facebook’s acquisition of Instagram, a highly visual social media platform, gave way to a broadened user base to include younger generations, in addition to the older population already found on Facebook. Instagram grew rapidly, beginning in 2010 and already had over 500 million users by 2016 (Sherlock & Wagstaff, 2018). *Faculty mentor

Facebook’s stated mission is “to give people the power to build community and bring the world closer together.” Consistent with that mission, Facebook was originally created to connect college students but has since expanded to connect people from around the world. With billions of users spend ing roughly 50 minutes per day on social media sites (Brailovskaia & Margraf, 2016), it seems as though these platforms may allow individuals the opportunity to express their innate need to belong (Hardy & Castonguay, 2018). Social connection is conceptualized by psychologists as a vital human need, and researchers have long known that social relationships decrease the risk for a host of prob lems such as depression, alcoholism, and lowered immune response (Cacioppo & Cacioppo, 2014; House, Landis, & Umberson, 1988). As such, social media companies could potentially be viewed as providing a valuable service that contributes to the well-being of a significant proportion of the world’s population.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

151

Problematic Social Media Use | Tanega and Downs

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

152

Consistent with such a view, some research studies have supported the idea that the connec tions made via social media may benefit users. Individuals’ social attractiveness increases with the number of Facebook friends they have and is associated with higher life satisfaction (Grieve, Indian, Witteveen, Tolan, & Marrington, 2013; Hardy & Castonguay, 2018; Utz, 2010). Such results support findings that the main motivation to use social networking sites is to facilitate and maintain social relationships in order to increase social capital (Johnston, Tanner, Lalla, & Kawalski, 2013). Social media sites can also serve as a platform for authentic self-presentation, which is associated with positive well-being in users (Berryman, Ferguson, & Negy, 2018). With its integration into daily life, many individuals find their social support online (Hardy & Castonguay, 2018), and college students report using social media as part of their daily rou tine for reasons such as escaping academic stress, having limited hobbies, and maintaining social relationships (Jasso-Medrano & López-Rosales, 2018). Several studies have also shown that college students frequently use social media sites such as Facebook as a “cry for help” and to openly discuss mental health (Berryman et al., 2018; Jelenchick, Eickhoff, & Moreno, 2013). Despite the potential benefits of social media use noted by researchers, several studies have also found that frequent social media use correlates with several indicators of psychological distress. For example, there is a significant relationship between Facebook use and depression (Steers, Wickham, & Acitelli, 2014), poor self-esteem (Kalpidou, Costin, & Morris, 2011), high anxiety (Labrague, 2014), high body dissatisfaction (Fardouly, & Vartanian, 2015), low self-perceived physical attractiveness (Haferkamp & Kramer, 2011), and lower over all well-being and life satisfaction (Shakya and Christakis, 2017; Tromholdt, 2016). Specifically, for highly visual social media sites such as Instagram, social comparison may play an important role, act ing as a mediator for poor psychological well-being, such that the more people use Instagram and the more people they follow, the higher their levels of depressive symptoms and body dissatisfaction and the lower their self-esteem (Sherlock & Wagstaff, 2018; Tiggemann, Hayden, Brown, & Veldhuis, 2018). Consistent with such results, Shensa, Sidani, Dew, Escobar-Viera, and Primack (2018) found that using multiple social platforms is associated with increased levels of anxiety and depression, and excessive social media use has been strongly linked with poor sleep quality (Xanidis & Bignell, 2015).

Although such correlational studies cannot demonstrate causation, some recent studies have revealed that social media use does indeed appear to cause negative effects for users. Using a longitu dinal design, Shakya and Christakis (2017) found that more active Facebook users showed worsening mental health and well-being over time, even when controlling for initial well-being. In another study, Tromholt (2016) utilized an experimental design and found that Facebook users who abstained for a week had higher life satisfaction and mental well-being than those who did not abstain. Several other recent experimental studies have found that viewing images and appearance-related comments on Instagram leads to increased negative mood, anxiety, body dissatisfaction, drive for thinness, and decreased self-compassion and self-esteem, particularly for female users (Brown & Tiggemann, 2016; Hendrickse, Arpan, Clayton, & Ridgeway, 2017; Kleemans, Daalmans, Carbaat, & Anschutz, 2018; Sherlock & Wagstaff, 2018; Slater, Varsani, & Diedrichs, 2017; Tiggemann & Barbato, 2018; Tiggemann et al., 2018). The undesirable outcomes associated with time spent on social media raise important questions about why the social connections touted by compa nies in their mission statements may have negative, rather than positive, effects on users. One possible explanation for this apparent paradox is that the kind of social connectivity accessed by social media use is not of sufficient quality to contribute to well-being. Because social media companies are almost entirely dependent on advertising revenue, their business model focuses on having the highest number of users possible spending as much time on their platform as possible. As recently stated by a former Facebook employee, “You have a business model designed to engage you and get you to basically suck as much time out of your life as pos sible and then selling that attention to advertisers” (Anderson, 2018). Theoretically, such companies could focus on maximizing the quality of the social connections they foster in order to achieve their desired quantity of user time and associated advertising revenue. Unfortunately, there is evidence that these companies are instead investing heavily in behav ioral engineering strategies designed solely to maximize user time on their platforms, and perhaps even entice users to become addicted. As noted by a former social media employee in a recent interview, “Behind every screen on your phone, there are generally like literally a thousand engineers that have worked on this thing to try to

Tanega and Downs | Problematic Social Media Use

make it maximally addicting” (Anderson, 2018). For example, Facebook and Instagram use artificial intelligence, that is informed by data based on user behavior, and a variable ratio schedule of reinforce ment to customize user news feeds. This may be done not to inform, but rather to appeal to strong emotions and intermittently surprise users in order to spark dopaminergic reward pathways and keep users looking, posting, and sharing at as high a rate as possible for as long as possible (Deibert, 2019; Turel, He, Xue, Xiao, & Bechara, 2014). Other examples of behaviorally engineered “features” designed to promote addictive behavior include notifications (i.e., discriminative stimuli that prime users to pay attention to social media throughout the day so they can access the rewards doled out on that variable ratio reinforcement schedule), “streaks” on SnapChat that reward users for send ing as many consecutive messages as possible, and auto-play on YouTube, which starts a new video as the one being watched ends so that users mindlessly keep watching as long as possible. All social media sites keep new content coming in an endless feed so that users have to opt out rather than opt in, which is a well-known way to increase compliance with desired behaviors due to the “status quo bias” (Samuelson & Zeckhauser, 1988). These habit-form ing features are intentionally integrated throughout all social media platforms, and their effects on users are well-known to the companies themselves. As noted by a former Facebook employee, “there was definitely an awareness of the fact that the product was habit-forming and addictive” (Anderson, 2018). There is a burgeoning consensus that such behavioral engineering efforts are successful in increasing user time on social media and may also be leading to addictive behavior in some individuals. For example, approximately 24% of U.S. teens report being online “almost constantly” with most of that time being spent on social media applications (Barry, Sidoti, Briggs, Reiter, & Lindsey, 2017). Zaremohzzabieh, Samah, Omar, Bolong, and Kamarudin (2014) argued that Facebook addiction is similar to other behavioral addictions such as gambling, shopping, or even abusing sub stances. Consistent with such a view, excessive social media use is believed to be associated with loss of control, negative repercussions (i.e., impairments to daily functioning), poor selective and sustained attention, reduction in physical activity, giving up other interests and activities, and an anxiety to remain connected (Echeburúa, 2013; Kim, Kim, & Jee, 2015). Additionally, researchers have found

neurobiological differences in individuals with social media addiction that are consistent with those found in individuals with substance use disorders such as reduced gray matter in the insula (Turel, He, Brevers, & Bechara, 2018). Further supporting the validity of the concept of social media addic tion are the findings of Beison and Rademacher (2017) that a family history of alcohol dependence accounts for a significant proportion of problematic smartphone use. If the behavioral engineering efforts of social media companies are indeed leading some users to become addicted, important questions need to be answered. First, it is necessary to specify the diagnostic criteria and cutoffs that would define an individual as having a social media use problem. To date, the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013) only provides validated diagnostic criteria for substance use disorders and gambling disorder. However, there is increasing recognition that other behavioral addictions (e.g., sex, shopping, Internet) may exist. Indeed, the DSM-5 includes proposed diagnostic criteria for Internet gaming disorder as a “condition for further study,” and the most recent version of the International Classification of Diseases (World Health Organization, 2018) includes gam ing disorder as an official diagnosis in its section on “Disorders due to addictive behaviours.” Expanding this work, several researchers have proposed that social media addiction is also a legitimate disorder, while also noting significant and problematic vari ability across studies in how social media addiction is defined and measured (see Ryan, Chester, Reece, & Xenos, 2014, for a review). In an attempt to develop a reliable and valid measure of social media addiction, van den Eijnden, Lemmons, and Valkenburg (2016) modified the proposed DSM-5 criteria for Internet gaming disorder to be applicable to social media use, thus creating the Social Media Disorder Scale (SMDS-9). The authors did so based on the belief that Internet gaming disorder and problematic social media use (PSMU) are specific forms of the broader construct of Internet addiction. Supporting that belief, van Eijnden and colleagues conducted a psychometric evaluation of the SMDS-9 and evaluated its test-retest and internal consistency reliability, factor structure, content, convergent, and criterion validity, and sensitivity and specific ity. The authors concluded that the SMDS-9 could provide a reliable and valid measure of PSMU (see van den Eijnden et al., 2016, for details).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

153

Problematic Social Media Use | Tanega and Downs

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

154

Another important consideration that needs to be addressed is the prevalence of PSMU. Because prevalence estimates are necessarily affected by the cut score that is used as part of the determination of whether an individual may meet diagnostic criteria for any disorder, it is critical to determine how many symptoms an individual must exhibit to potentially qualify for a diagnosis of PSMU. In their study, van den Eijnden and colleagues (2016) used a cut score of 5 out of 9 symptoms, which is the experimental cut score proposed in the DSM-5 for Internet gaming disorder, and found that between 7.3% and 11.6% of adolescents across three studies met criteria for PSMU. However, the DSM-5 notes that a cut score of 5 symptoms provides a conservative definition of Internet gaming disorder that may be adjusted as empirical evidence accumulates. Indeed, the DSM-5 criteria for substance use disorders requires the presence of only 2 symptoms, whereas the criteria for gambling disorder requires the presence of 4 symptoms. This uncertainty around cut scores raises the possibility that the actual prevalence of PSMU may increase if a cut score lower than 5 is eventually adopted. Finally, it is also necessary to determine whether PSMU has negative impacts aside from those included in the symptoms themselves. For example, do individuals who show PSMU have worse mental and/or physical health than those who do not? Van den Eijnden and colleagues (2016) found that SMD-9 scores were significantly correlated with depression, loneliness, attentional problems, impulsivity, and low self-esteem. However, it is important to note that the correlations were not particularly strong, ranging from .19 to .37, nor did the researchers determine whether there were significant differences on those variables between those who met their criteria for PSMU and those who did not. The current study sought to add to the research literature on social media addiction by examin ing the percentage of college students who may be at risk for PSMU. Expanding the work of van den Eijnden and colleagues (2016), we elected to evaluate two cut scores in the present study: the conservative cut score of 5 used in the DSM-5 experimental criteria for Internet gaming disorder, and the more liberal cut score of 2 used in the established and widely used DSM-5 substance use disorder diagnostic criteria (American Psychiatric Association, 2013). It was hypothesized that those who met either cut score for PSMU would report higher levels of mental health symptoms and lower well-being.

Method Participants Data was collected between September and December 2018 using a Qualtrics survey that was distributed to students at a private university in the northwest United States via Facebook posts and an online research participation system for students enrolled in an introductory psychology course (n = 294). The sample averaged 18.9 years of age (SD = 1.2) and was comprised of 61.1% first-year students, 18.2% second-year students, 8.1% thirdyear students, and 11.8% fourth-year students. Women comprised 77% of the sample. Reported ethnicities included White or European American (58.6%), Asian American (16.0%), Hispanic or Latino (14.9%), Hawaiian/Pacific Islander (4.4%), Black or African American (2.4%), and Native American or Alaska Native (1.0%). Ninety-five per cent of respondents reported being native English speakers and 94.3% reported the United States as their country of origin. Eighty-three percent of participants’ parents or grandparents attended col lege. Eighty-eight percent of participants reported being heterosexual, 1.4% were gay or lesbian, and 8.8% were bisexual. Procedures Participants completed an online survey comprised of the Social Media Disorder Scale-9 (SMDS-9), the Symptoms and Assets Screening Scale (SASS), and a demographic survey. Students currently enrolled in introductory psychology classes received class credit for survey completion. The University of Portland institutional review board approved all procedures and materials used in this study, and each participant provided informed consent prior to completing the study. Measures Social Media Disorder Scale-9 (SMDS-9). The SMDS-9 is a 9-item instrument designed to assess whether a respondent potentially displays disor dered social media use. The SMDS-9 items were modified versions of the proposed diagnostic criteria for Internet gaming disorder, which is included in the DSM-5 (American Psychiatric Association, 2013). Participants responded “yes” or “no” to nine items such as “During the past year, have you tried to spend less time on social media, but failed?” Affirmative responses were summed to create a score for each participant who indicated the number of PSMU symptoms they reported expe riencing in the past year. Psychometric evaluation of

Tanega and Downs | Problematic Social Media Use

the SMDS-9 revealed internal consistency reliability coefficients (Cronbach’s α) ranging from .76 to .83 for the 9-item scale across three studies, and a test-retest reliability coefficient of .66. Regarding validity, SMDS-9 scores were significantly and posi tively correlated with compulsive Internet use (r = .51) and self-declared social media addiction (r = .48; see van den Eijnden et al., 2016 for additional psychometric data). Because an empirically based cut score for PSMU has not yet been determined, we elected to examine cut scores of 5 (i.e., the experimental cut score for Internet gaming disorder) and 2 (i.e., the established cut score for substance use disorders) in the present study. Participants who endorsed 5 or more out of 9 symptoms were defined as at-risk for “High Cut Score PSMU” and those who endorsed 2 or more symptoms were defined as at-risk for “Low Cut Score PSMU.” Symptoms and Assets Screening Scale (SASS). The SASS is a 30-item self-report screening measure that was developed to assess mental health in the college student population. Participants rated each of the 30 items on a 4-point Likert-type scale ranging from 0 (not true) to 3 (certainly true). The SASS generates a 23-item overall psychological distress score ranging from 0 to 69, as well as the following 5-item subscale scores ranging from 0 to 15: Depressive Symptoms (e.g., “I feel hopeless”), Anxious Symptoms (e.g., “I get scared easily or often feel afraid”), Substance Problems (e.g., “I have difficulty limiting or cutting down on my use of alcohol or drugs”), Eating Problems (e.g., “I am very afraid of gaining weight or becoming fat”), and Well-Being/Assets (e.g., “I feel good about myself”). A psychometric evaluation of the SASS generated internal consistency reliability coefficients (Cronbach’s α) ranging from .73 to .81 for the five subscales and .86 for the 23-item overall distress measure, as well as test-retest coef ficients ranging from .75 to .83 for the subscales and .87 for the overall distress scale. In the same study, the SASS subscales and overall distress scale were significantly and positively correlated with well-established measures of their respective constructs with coefficients ranging from .68 to .83 (see Downs, Boucher, Campbell, & Dasse, 2013, for additional psychometric data). Demographics. The survey ended with demo graphic questions, including gender, age, ethnicity, country of origin, native language, sexual orienta tion, and year in college.

Results Prevalence of PSMU Table 1 shows the percentage of participants who endorsed each possible number of indicators of PSMU (0 through 9). Overall, 8.1% of participants (4.5% of men and 9.2% of women) qualified as at risk for High Cut Score PSMU, as measured by the conservative experimental criteria of 5 or more indicators used to identify Internet gaming disorder. Over 51% of participants (37.9% of men and 54.8% of women) qualified as at risk for Low Cut Score PSMU as measured by the more liberal criteria of 2 or more indicators used to identify substance use disorders. Men endorsed a mean of 1.26 (SD = 1.49) out of 9 PSMU indicators, whereas women endorsed a significantly higher mean of 1.85 (SD = 1.67) indicators, t(290) = 2.61, p = .009, d = 0.37. Table 2 shows the percentage of participants who endorsed each specific indicator of PSMU. As seen in the table, the most common indicators reported by participants were trying to spend less time on social media but failing, and using social media to escape from negative feelings. The least common indicator reported by participants was having serious conflicts with family, friends, or others because of social media use. PSMU and Mental Health Presented in Table 3 are bivariate Pearson correla tion coefficients for all study variables. As expected, the number of PSMU indicators endorsed by par ticipants was significantly correlated with all mental TABLE 1 Percentage of College Students Endorsing Each Possible Number of Problematic Social Media Use Indicators by Gender Number of Indicators

Percentage of All Participants n = 294

Percentage of Men n = 66

Percentage of Women n = 228

24.7

38.3

21.1

21.8

20.0

22.1

29.1

25.0

30.5

11.3

5.0

12.7

4.7

6.7

4.2

5.5

3.3

6.1

1.8

1.7

1.9

0.4

0.5

SUMMER 2020

0.4

0.5

0.4

0.5

PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

155

Problematic Social Media Use | Tanega and Downs

health symptoms. Specifically, PSMU indicators were significantly and positively correlated with eating problems, r (293) = .33, p < .001, anxiety symptoms, r (293) = .27, p < .001, depressive symp toms, r(293) = .22, p < .001, substance use problems, r (223) = .13, p = .03, and total symptoms, r (291) = .34, p < .001, and were significantly inversely cor related with well-being, r (291) = -.23, p < .001. Also as expected, all mental health symptom subscale scores on the SASS were significantly positively TABLE 2 Percentage of College Students Endorsing Each Indicator of Problematic Social Media Use by Gender Problematic Social Media Use Indicators

Percentage of All Participants n = 294

Percentage of Men n = 66

Percentage of Women n = 228

Regularly found that you can’t think of anything else but the moment that you will be able to use social media again

7.4

9.1

7.0

Regularly felt dissatisfied because you wanted to spend more time on social media

8.1

10.6

7.0

Often felt bad when you could not use social media

18.2

15.2

19.3

Tried to spend less time on social media, but failed

51.4

34.8

56.6

Regularly neglected other activities (e.g., hobbies, sports, etc.) because you wanted to use social media

15.9

9.1

17.5

6.4

1.5

7.9

Regularly lied to your family or friends about the amount of time you spend on social media

11.8

9.1

12.7

Often use social media to escape from negative feelings

51.4

37.9

54.8

Had serious conflict with your family, friends, or other people because of your social media use

1.7

0.0

2.2

Regularly had arguments with others because of your social media use

TABLE 3 Bivariate Correlations Between Problematic Social Media Use Indicators, Mental Health Symptoms, and Well-Being Variable

.33**

.13*

.27**

.22**

.34**

-.23**

–

.38**

.41**

.47**

.71**

-.36**

3. Substance Problems

–

.24**

.40**

.55**

-.21**

4. Anxiety Symptoms

–

.70**

.83**

-.53**

–

.88

-.66**

–

-.63**

1. Problematic Social Media Use Indicators 2. Eating Problems

5. Depressive Symptoms 6. Total Symptoms 7. Well-Being

-.31**

Note. p < .05. p < .001. **

–

156

correlated with each other, and were significantly inversely correlated with well-being. Following those correlational analyses, a series of independent-samples t tests were conducted to determine whether participants who were consid ered at risk for PSMU reported more mental health symptoms than those who were not classified as at risk for PSMU. Because there is not yet an estab lished cut score for a potential diagnosis of PSMU, we first compared participants who endorsed 5 or more indicators (High Cut Score PSMU) with those who endorsed 4 or fewer indicators, and then compared participants who endorsed 2 or more indicators (Low Cut Score PSMU) with those who endorsed 0 or 1 symptoms. High Cut Score PSMU. As seen in Table 4 and consistent with our hypothesis, those who were at risk for High Cut Score PSMU (5 or more symp toms) reported significantly higher levels of total mental health symptoms, depressive symptoms, anxiety symptoms, and problematic eating symp toms than did those who were not at risk for High Cut Score PSMU (4 or fewer symptoms). There were no differences between those at risk for High Cut Score PSMU and those not at risk for High Cut Score PSMU on reported levels of well-being or substance use problems. Low Cut Score PSMU. Table 5 shows the men tal health symptom and well-being scores for those categorized as at risk for and not at risk for PSMU when the cut score was dropped down to the more liberal criteria of two or more indicators. Those who were at risk for Low Cut Score PSMU criteria reported significantly higher levels of total mental health symptoms, problematic eating symptoms, anxiety symptoms, depressive symptoms, as well as lower well-being than did those who were not at risk for Low Cut Score PSMU (0 or 1 symptoms). There was no difference between those at risk for Low Cut Score PSMU and those who were not at risk for Low Cut Score PSMU on the level of substance use problems reported.

Discussion The primary purpose of this study was to examine the prevalence of PSMU in the college student population using two different cut scores: (a) the conservative experimental cut score of 5 indicators out of 9 proposed for further study in the DSM-5 to diagnose Internet gaming disorder; and (b) the more liberal established cut score of 2 indicators out of 9 used in the DSM-5 to diagnose substance use disorders. The results revealed that

Tanega and Downs | Problematic Social Media Use

the prevalence of PSMU may range from 8.2% to 51.3% among college students, depending on the cut score used. These findings raise interesting questions about just how successful the behavioral engineering efforts of social media companies are in foster ing addictive behavior in users by “exploiting a vulnerability in human psychology” in order to “consume as much user time as possible” as stated by Facebook’s founding president in a recent inter view (Anderson, 2018; Deibert, 2019). On the one hand, a prevalence estimate of 51.3% would seem to be quite high for any psychological disorder. However, there is some evidence that the actual prevalence of PSMU may be significantly higher than the estimates of 8.2% generated in this study and the 7.3% and 11.6% found by van den Eijnden and colleagues (2016) using the same conservative high cut score of 5 indicators. Specifically, previous studies have generated prevalence estimates ranging from 33% to 50% for psychological problems such as depression and anxiety in the college student population (Eisenberg, Gollust, Golberstein, & Hefner, 2007; Garlow, Rosenberg, Moore, Haas, Koestner, Hendin, & Nemeroff, 2008; Hunt & Eisenberg, 2010), and recent surveys from the American College Health Association (2018) found that 31.1% of students reported being diagnosed with or treated for a psychological disorder in the previous 12 months. In other words, it is quite common for college students to meet diagnostic criteria for at least one psychological disorder at any given time. In addition, research has revealed that the primary activity college students are using their cell phones for is to access social media (Barry et al., 2017; Jasso-Medrano & López-Rosales 2018). Indeed, researchers have found that social media use, rather than gaming or Internet use, is the primary driver of smartphone addiction in college students (Roberts, Petnji Yaya, & Manolis, 2014). As researchers, clinicians, and diagnostic systems such as the DSM-5 (American Psychiatric Association, 2013) and the International Classification of Diseases (World Health Organization, 2018) increasingly recognize the existence of behavioral addictions such as gambling disorder and Internet gaming disorder, it seems likely that PSMU (or a closely related variant) will likely be recognized as a legitimate behavioral addiction in the future. As such, it will be critically important to determine the exact diagnostic criteria for PSMU, including the proper DSM-5 cut score to be used

for diagnostic and research purposes. Doing so will require evaluating the sensitivity and specificity of different cut scores when compared to indepen dent measures of diagnostic status (e.g., clinician report) and clinically significant impairment or distress (e.g., serious problems at work, home, or with social relationships) in order to validate PSMU as a legitimate disorder and to guard against false positives (incorrectly pathologizing normal behavior) and false negatives (failing to diagnose actual pathological behavior). Alternatively, as clinicians and researchers increasingly conceptualize most forms of psycho pathology as dimensional, rather than categorical phenomena (American Psychiatric Association, 2013), it is quite possible that PSMU may be more TABLE 4 Mean Scores on Mental Health Symptoms and Well-Being by High Cut Score Problematic Social Media Use Status At Risk for High Cut Score Problematic Social Media Use (5 or More Indicators) n = 24 Variable

Not at Risk for Problematic Social Media Use (4 or Fewer Indicators) n = 270

Eating Problems

6.58

(3.34)

4.96

(3.26) 2.33

.021

0.49

Substance Problems

1.50

(2.40)

1.01

(2.02) 1.11

.269

0.22

Anxiety Symptoms

7.13

(3.94)

4.72

(3.74) 3.00

.003

0.63

Depressive Symptoms Total Symptoms Well-Being

5.08

(3.88)

3.58

(3.53) 1.98

.048

0.40

25.21

(12.60)

17.95

(11.15) 3.02

.003

0.61

9.29

(3.14)

10.27

(3.11) 1.47

.142

0.31

TABLE 5 Mean Scores on Mental Health Symptoms and Well-Being by Low Cut Score Problematic Social Media Use Status At Risk for Low Cut Score Problematic Social Media Use (2 or More Indicators) n = 151 Variable

Not At Risk for Problematic Social Media Use (0 or 1 Indicators) n = 143

Eating Problems

5.97

(3.45)

4.18

(2.85)

4.70

.000

0.57

Substance Problems

1.25

(2.53)

0.85

(1.37)

1.64

.101

0.20

Anxiety Symptoms

5.71

(3.80)

4.08

(3.65)

3.74

.000

0.44

Depressive Symptoms Total Symptoms Well-Being

4.44

(3.96)

2.94

(2.94)

3.68

.000

0.43

21.89

(12.27)

15.06

(9.31)

5.31

.000

0.58

9.51

(3.10)

10.90

(2.99)

3.89

.000

0.40

157

Problematic Social Media Use | Tanega and Downs

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

158

appropriately conceptualized as dimensional, thus potentially negating the need to identify a single specific cut score for diagnostic purposes. Rather, any person who displays one or more indicators could potentially be considered to be displaying PSMU, depending on the extent to which their social media use causes clinically significant impairment and/or distress. The results of this study provide some support for a dimensional con ceptualization of PSMU as all of the mental health problems assessed were significantly correlated with the number of PSMU indicators reported by partici pants. Further, even those who reported relatively mild levels of PSMU (i.e., two or more indicators) appeared to be at risk for experiencing distress or impairment, as they reported significantly higher levels of overall psychological distress, anxiety symptoms, depressive symptoms, and disordered eating symptoms, as well as lower well-being, than did those who reported one or zero indicators of PSMU. A third possibility is that a combined approach may emerge as most appropriate for assessing PSMU with cut scores denoting different levels of severity. Such a system would be analogous to how the DSM-5 currently classifies substance use disorders as mild, moderate, or severe, depending on the number of symptoms present (American Psychiatric Association, 2013). Of course, our study cannot prove that PSMU caused the mental health symptoms participants were experiencing because one could reasonably hypothesize that individuals who experience more mental health problems may subsequently be at risk for developing PSMU. However, there is accumulat ing evidence that problematic (or even “normal”) social media use may indeed cause impairment and/or distress. Using prospective experimental designs, researchers have demonstrated that active Facebook users show worsening mental health and well-being over time (Shakya & Christakis, 2017) and that Facebook users who abstain for a period of time as short as a week experience improvements in well-being (Tromholt, 2016). Individuals exposed to images on Instagram show immediate increases in a host of negative mental health variables such as body dissatisfaction, anxiety, depressive symptoms, and immediate decreases in self-esteem and self-compassion (Brown & Tiggemann, 2016; Hendrickse et al., 2017; Kleemans et al., 2018; Sherlock & Wagstaff, 2018; Slater, Varsani, & Diedrichs, 2017; Tiggemann & Barbato, 2018; Tiggemann et al., 2018). These experimental studies are consistent with our findings and provide

support for the notion that PSMU may have signifi cant negative impacts on individuals’ mental health and well-being. Such findings raise questions about why social media, which is purportedly designed to increase social connections, may paradoxically have negative effects on users. Researchers have pointed to the social comparisons that social media platforms fos ter such that, as users scroll their social media feeds and see others apparently doing well, they often try to boost their own self-image, which perpetuates a “self-enhancement envy spiral” (Krasnova et al., 2015). In other words, an ongoing competition may arise on social media sites as individuals compare themselves to the content they consume, with subsequent negative feelings arising if they perceive themselves as doing less well than others. Consistent with such an interpretation, studies have found that social comparison is an important variable that mediates the negative effects that exposure to Instagram has on body dissatisfaction and drive for thinness (Hendrickse et al., 2017; Kleemans et al., 2018). Similarly, Hanna and colleagues (2017) found that social comparison and self-objectifica tion mediated the relationship between Facebook use and body shame, symptoms of depression and anxiety, and low self-esteem. Other researchers have noted that the pressure to appear as perfect offline as a person appears online may further increase an individual’s social comparison and anxiety levels even when not engaging with social media (Rauch, Strobel, Bella, Odachowski, & Bloom, 2014). If some users experience negative emotions as a result of using social media platforms, it may be fair to ask why those users do not simply spend less time using social media. Continued use despite negative consequences is a pattern consistent with the behavior of many individuals with substance use disorders who continue to use substances despite negative impacts on their health, relationships, and ability to function at school and/or work. Recent research has suggested that these similarities may be due, at least in part, to neurobiological factors such as reduced gray matter in the posterior insula that may cause increases in delay discounting (Turel et al., 2018; Wood & Bechara, 2014), in which indi viduals show a stronger preference for immediate rewards despite the potential negative long-term consequences of their behavior. In other words, just as cues and the high associated with substance abuse are known to trigger the brain’s impulsive reward seeking system (e.g., dopamine pathways in the ventral striatum) and reduce activity in

Tanega and Downs | Problematic Social Media Use

inhibitory systems (e.g., prefrontal and orbitofron tal cortex; He et al., 2018), the cues and immediate rewards engineered into social media platforms may have a similar effect, causing some individu als to compulsively pick up their smartphone and get on social media even if doing so may have the long-term effect of worsening their quality of life. Given that research on social media addiction is relatively new, it is not currently known whether PSMU leads to the neurobiological differences reported by Turel and colleagues, or if pre-existing differences in brain structure and function may predispose certain individuals to be at higher risk for developing problematic social media use. Limitations, Conclusions, and Recommendations This study was limited by a cross-sectional design and a reliance on an online self-report survey that introduced the possibility of biased responding and did not allow for causal explanations. It is also important to note that our categorization of participants as being at risk for or not at risk for PSMU was based solely on participant report on a single checklist, which is not at all analogous to a valid psycho-diagnostic evaluation. Ideally, future studies would include a multimethod, multiinformant evaluation of PSMU symptoms in order to more accurately categorize participants and determine proper cut scores, and to evaluate whether a dimensional or combined approach to understanding PSMU may be more appropriate than a categorical approach. Another limitation was the disproportionately high percentage of women and younger college students in the sample and low percentage of African American participants, which limits the generalizability of the findings. Although women make up more than 60% of the student population where this study was conducted, women comprised 77% of the sample, and the average participant age of the sample was 18.9 years old, a reflection of the students enrolled in Introductory Psychology when the study was conducted. We considered analyzing our data separately by gender, however, it was not feasible to do so because of the small number of men (n = 3) who were at risk for High Cut Score PSMU. Future studies should seek to address whether there are gender differences in the extent to which PSMU is associated with psychological distress and/or impairment, as our results suggested that women may be at higher risk for PSMU than are men. Finally, our study did not ask participants to report their time using social

media, which would have allowed us to examine how time spent on social media related to indica tors of PSMU. Despite those limitations, this study generated some interesting results that raise legitimate concerns that the behavioral engineering efforts of social media companies may be fostering addictive behavior in a significant proportion of users. Thus, it is critically important that researchers, clinicians, and those interested in public health continue to determine exactly how PSMU can be accurately assessed, diagnosed, and treated. However PSMU may come to be defined in the future, our results suggest that those who display even mild levels of PSMU may be at risk for a range of negative mental health effects including increased symptoms of depression, anxiety, eating problems, and overall psychological distress. Moving forward, it will be important for researchers to examine both the short-term and long-term impacts of PSMU to determine the level of risk associated with social media use problems. Future studies should also seek to elucidate the specific mechanisms by which PSMU may impact mental health. Finally, it will be necessary to conduct longitudinal studies to examine the effects of PSMU on brain structure and function because such research is currently in its infancy. These are especially vital questions to answer because the current generation of young people will never know a world without social media, and the long-term effects of social media use have yet to be investigated.

References American College Health Association. (2018). National College Health Assessment. Retrieved from https://www.acha.org/documents/ncha/ NCHAII_Fall_2018_Undergraduate_Reference_Group_Executive_ Summary.pdf American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: Author. Anderson, H. (2018, July). Social media apps are ‘deliberately’ addictive to users. BBC News. Retrieved from https://www.bbc.com/news/technology-44640959 Barry, C. T., Sidoti, C. L., Briggs, S. M., Reiter, S. R., & Lindsey, R. A. (2017). Adolescent social media use and mental health from adolescent and parent perspectives. Journal of Adolescence, 61, 1–11. https://doi.org/10.1016/j.adolescence.2017.08.005 Beison, A., & Rademacher, D. J. (2017). Relationship between family history of alcohol addiction, parents’ education level, and smartphone problem use scale scores. Journal of Behavioral Addictions, 6, 84–91. https://doi.org/10.1556/2006.6.2017.016 Berryman, C., Ferguson, C. J., & Negy, C. (2018). Social media use and mental health among young adults. Psychiatric Quarterly, 89, 307–314. https://doi.org/10.1007/s11126-017-9535-6 Brown, Z., & Tiggemann, M. (2016). Attractive celebrity and peer images on Instagram: Effect on women’s mood and body image. Body Image, 19, 37–43. https://doi.org/10.1016/j.bodyim.2016.08.007 Cacioppo, J. T., & Cacioppo, S. (2014). Social relationships and health: The toxic

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

159

Problematic Social Media Use | Tanega and Downs

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

160

effects for perceived social isolation. Social and Personality Psychology Compass, 8, 58–72. https://doi.org/10.1111/spc3.12087 Deibert, R. J. (2019). The road to digital unfreedom: Three painful truths about social media. Journal of Democracy, 30, 25–39. https://doi.org/10.1353/jod.2019.0002 Downs, A., Boucher, L. A., Campbell, D. G., & Dasse, M. (2013). Development and initial validation of the Symptoms and Assets Screening Scale. Journal of American College Health, 61, 164–174. https://doi.org/10.1080/07448481.2013.773902 Echeburúa, E. (2013). Overuse of social networking. In P. M. Miller, S. A. Ball, M. E. Bates, A. W. Blume, K. M. Kampman, D. J. Kavanagh, . . . P. De Witte (Eds.), Principles of addiction: Comprehensive addictive behaviors and disorders, Vol. 1 (pp. 911–920). (pp. 911–920). San Diego, CA: Elsevier Academic Press. Eisenberg D., Gollust, S. E., Golberstein, E., & Hefner, J. L. (2007). Prevalence and correlates of depression, anxiety, and suicidality among university students. American Journal of Orthopsychiatry, 77, 534–542. https://doi.org/10.1037/0002-9432.77.4.534 Fardouly, J., & Vartanian, L. R. (2015). Negative comparisons about one’s appearance mediate the relationship between Facebook usage and body image concerns. Body Image, 12, 82–88. https://doi.org/10.1016/j.bodyim.2014.10.004 Garlow, S. J., Rosenberg, J., Moore, J. D., Haas, A. P., Koestner, B., Hendin, H., & Nemeroff, C. B. (2008). Depression, desperation, and suicidal ideation in college students: Results from the American Foundation for Suicide Prevention College Screening Project at Emory University. Depression and Anxiety, 25, 482–488. Grieve, R., Indian, M., Witteveen, K., Tolan, G. A., & Marrington, J. (2013). Faceto-face or Facebook: Can social connectedness be derived online? Computers in Human Behavior, 29, 604–609. https://doi.org/10.1016/j.chb.2012.11.017 Haferkamp, N., & Krämer, N. C. (2011). Social comparison 2.0: Examining the effects of online profiles on social-networking sites. Cyberpsychology, Behavior, and Social Networking, 14, 309–314. https://doi.org/10.1089/cyber.2010.0120 Hanna, E., Ward, L. M., Seabrook, R. C., Jerald, M., Reed, L., Giaccardi, S., & Lippman, J. R. (2017). Contributions of social comparison and self‐ objectification in mediating associations between Facebook use and emergent adults’ psychological well-being. Cyberpsychology, Behavior, and Social Networking, 20, 172–179. https://doi.org/10.1089/cyber.2016.0247 Hardy, B. W., & Castonguay, J. (2018). The moderating role of age in the relationship between social media use and mental well-being: An analysis of the 2016 General Social Survey. Computers in Human Behavior, 85, 282–290. https://doi.org/10.1016/j.chb.2018.04.005 He, Q., Huang, X., Turel, O., Schulte, M., Huang, D., Thames, A., . . . Hser, Y. (2018). Presumed structural and functional neural recovery after long-term abstinence from cocaine in male military veterans. Progress in NeuroPsychopharmacology & Biological Psychiatry, 84, 18–29. https://doi.org/10.1016%2Fj.pnpbp.2018.01.024 Hendrickse, J., Arpan, L. M., Clayton, R. B., & Ridgway, J. L. (2017). Instagram and college women’s body image: Investigating the roles of appearancerelated comparisons and intrasexual competition. Computers in Human Behavior, 74, 92–100. https://doi.org/10.1016/j.chb.2017.04.027 House, J. S., Landis, K. R., & Umberson, D. (1988). Social relationships and health. Science, 241, 540–545. https://doi.org/10.1126/science.3399889 Hunt J., & Eisenberg D. (2010). Mental health problems and help-seeking behavior among college students. Journal of Adolescent Health, 46, 3–10. https://doi.org/10.1016/j.jadohealth.2009.08.008 Jasso-Medrano, J. L., & Lopez-Rosales, F. (2018). Measuring the relationship between social media use and addictive behavior and depression and suicide ideation among university students. Computers in Human Behavior, 87, 183–191. https://doi.org/10.1016/j.chb.2018.05.003 Jelenchick, L. A., Eickhoff, J. C., Moreno, M. A. (2013). ‘Facebook depression?’ Social networking site use and depression in older adolescents. Journal of Adolescent Health, 52, 128–130. https://doi.org/10.1016/j.jadohealth.2012.05.008 Johnston, K., Tanner, M., Lalla, N., & Kawalski, D. (2013). Social capital: The benefit of Facebook ‘friends’.’ Behaviour & Information Technology, 32, 24–36. https://doi.org/10.1080/0144929X.2010.550063 Kalpidou, M., Costin, D., & Morris, J. (2011). The relationship between Facebook and the well-being of undergraduate college students. Cyberpsychology, Behavior, and Social Networking, 14, 183–189.

https://doi.org/10.1089/cyber.2010.0061 Kim, S.-E., Kim, J.-W., & Jee, Y.-S. (2015). Relationship between smartphone addiction and physical activity in Chinese international students in Korea. Journal of Behavioral Addictions, 4, 200–205. https://doi.org/10.1556/2006.4.2015.028 Kleemans, M., Daalmans, S., Carbaat, I., & Anschütz, D. (2018). Picture perfect: The direct effect of manipulated Instagram photos on body image in adolescent girls. Media Psychology, 21, 93–110. https://doi.org/10.1080/15213269.2016.1257392 Krasnova, H., Widjaja, T., Buxmann, P., Wenninger, H., & Benbasat, I. (2015). Why following friends can hurt you: An exploratory investigation of the effects of envy on social networking sites among college-age users. Information Systems Research, 26, 585–605. https://doi.org/10.1287/isre.2015.0588 Labrague, L. (2014). Facebook use and adolescents’ emotional states of depression, anxiety, and stress. Health Science Journal, 8, 80–89. Lee, E. J., & Kim, Y. W. (2014). Motivation for Twitter Use Measure [Database record]. PsycTESTS. https://doi.org/10.1037/t44002-000 Østergaard, S. D. (2017). Taking Facebook at face value: Why the use of social media may cause mental disorder. Acta Psychiatrica Scandinavica, 136, 439–440. https://doi.org/10.1111/acps.12819 Rauch, S. M., Strobel, C., Bella, M., Odachowski, Z., & Bloom, C. (2014). Face to face versus Facebook: Does exposure to social networking web sites augment or attenuate physiological arousal among the socially anxious? Cyberpsychology, Behavior, and Social Networking, 17, 187–190. https://doi.org/10.1089/cyber.2012.0498 Roberts, J. A., Pullig, C., & Manolis, C. (2015). I need my smartphone: A hierarchical model of personality and cell-phone addiction. Personality and Individual Differences, 79, 13–19. https://doi.org/10.1016/j.paid.2015.01.049 Ryan, T., Chester, A., Reece, J., & Xenos, S. (2014). The uses and abuses of Facebook: A review of Facebook addiction. Journal of Behavioral Addictions, 3, 133–148. https://doi.org/10.1556/JBA.3.2014.016 Samuelson, W., & Zeckhauser, R. (1988). Status quo bias in decision making. Journal of Risk and Uncertainty, 1, 7–59. Shakya, H. B., & Christakis, N. A. (2017). Association of Facebook use with compromised well-being: A longitudinal study. American Journal of Epidemiology, 185, 203–211. https://doi.org/10.1093/aje/kww189 Shensa, A., Sidani, J. E., Dew, M. A., Escobar-Viera, C. G., & Primack, B. A. (2018). Social media use and depression and anxiety symptoms: A cluster analysis. American Journal of Health Behavior, 42, 116–128. https://doi.org/10.5993/AJHB.42.2.11 Sherlock, M., & Wagstaff, D. L. (2018). Exploring the relationship between frequency of Instagram use, exposure to be idealized images, and psychological well-being in women. Psychology of Popular Media Culture. Advance online publication. https://doi.org/10.1037/ppm0000182 Slater, A., Varsani, N., & Diedrichs, P. C. (2017). #fitspo or #loveyourself? The impact of fitspiration and self-compassion Instagram images on women’s body image, self-compassion, and mood. Body Image, 22, 87–96. https://doi.org/10.1016/j.bodyim.2017.06.004 Steers, M.-L. N., Wickham, R. E., & Acitelli, L. K. (2014). Seeing everyone else’s highlight reels: How Facebook usage is linked to depressive symptoms. Journal of Social and Clinical Psychology, 33, 701–731. https://doi.org/10.1521/jscp.2014.33.8.701 Tiggemann, M., & Barbato, I. (2018). ‘You look great!’ The effect of viewing appearance-related Instagram comments on women’s body image. Body Image, 27, 61–66. https://doi.org/10.1016/j.bodyim.2018.08.009 Tiggemann, M., Hayden, S., Brown, Z., & Veldhuis, J. (2018). The effect of Instagram ‘likes’ on women’s social comparison and body dissatisfaction. Body Image, 26, 90–97. https://doi.org/10.1016/j.bodyim.2018.07.002 Tromholt, M. (2016). The Facebook experiment: Quitting Facebook leads to higher levels of well‐being. Cyberpsychology, Behavior and Social Networking, 19, 661–666. https://doi.org/10.1089/cyber.2016.0259 Turel, O., He, Q., Brevers, D., & Bechara, A. (2018). Delay discounting mediates the association between posterior insular cortex volume and social media addiction symptoms. Cognitive, Affective & Behavioral Neuroscience, 18, 694–704. https://doi.org/10.3758/s13415-018-0597-1 Turel, O., He, Q., Xue, G., Xiao, L., & Bechara, A. (2014). Examination of neural systems sub-serving Facebook ‘addiction.’ Psychological Reports, 115, 675–695. https://doi.org/10.2466/18.PR0.115c31z8 Utz, S. (2010). Show me your friends and I will tell you what type of person you

Tanega and Downs | Problematic Social Media Use

are: How one’s profile, number of friends, and type of friends influence impression formation on social network sites. Journal of Computer-Mediated Communication, 15, 314–335. https://doi.org/10.1111/j.1083-6101.2010.01522.x van den Eijnden, R. J. J. M., Lemmens, J. S., & Valkenburg, P. M. (2016). The Social Media Disorder Scale. Computers in Human Behavior, 61, 478–487. https://doi.org/10.1016/j.chb.2016.03.038 Wood, S. M. W., & Bechara, A. (2014). The neuroscience of dual (and triple) system in decision making. In V. F. Reyna & V. Zayas (Eds.), The neuroscience of risky decision making (pp. 177–202). https://doi.org/10.1037/14322-008 World Health Organization. (2018). International statistical classification of diseases and related health problems (11th Revision). Retrieved from https://icd.who.int/browse11/l-m/en Xanidis, N., & Brignell, C. M. (2016). The association between the use of social network sites, sleep quality and cognitive function during the day. Computers in Human Behavior, 55, 121–126. https://doi.org/10.1016/j.chb.2015.09.004

Zaremohzzabieh, Z., Samah, B. A., Omar, S. Z., Bolong, J., & Kamarudin, N. A. (2014). Addictive Facebook use among university students. Asian Social Science, 10, 107–116. https://doi.org/10.5539/ass.v10n6p107 Author Note. Chloe Tanega, https://orcid.org/0000-0002-1596-3737, Department of Psychological Sciences, University of Portland; and Andrew Downs, https://orcid.org/0000-0002-8611-7407, Department of Psychological Sciences, University of Portland. We would like to thank all of the reviewers who helped to make this a better manuscript. Correspondence concerning this article should be addressed to Andrew Downs, Department of Psychological Sciences, University of Portland, Portland, OR 97203. E-mail: downs@up.edu

161

https://doi.org/10.24839/2325-7342.JN25.2.162

Context Effects on Recognition Memory for Words Abhilasha Vishwanath and Joshua Shive* Tennessee State University

ABSTRACT. The present study examined whether word frequency, study context, and word repetition produce differences in word recognition and context recognition. We also tested a prediction of the dual process model where recognition memory has two individual processes, namely recollection and familiarity. Participants studied lists of words presented in contexts defined by background color, screen position of the word, and study question during encoding. Word frequency, study context, and word repetition were manipulated during the encoding phase. During the subsequent retrieval phase, participants performed two memory tasks: a word recognition task involving old judgments or new judgments and a context recognition task involving remember judgments and know judgments for words reported as an old judgment. We found effects of each of the manipulated factors on recognition memory. False alarms were higher for common words than uncommon words (η2p = .38, p < .001). Repeated words were remembered better than nonrepeated words (η2p = .76, p < .001). Words repeated across contexts during the study were recognized better than words repeated in the same context during encoding (η2p = .18, p = .03). We also found effects of repetition and word frequency on reaction time. Repeated words were recognized faster than nonrepeated words (η2p = .23, p = .01), and uncommon words were recognized faster than common words (η2p = .38, p = .001). However, we did not find evidence to support the dual process model’s predictions about the impact of context on remember and know judgments. Keywords: recognition memory, word frequency, repetition, context, dual process model

T SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

162

he process by which different kinds of information are encoded and retrieved by memory is still unclear, despite the current knowledge of distinct memory systems (Murnane, Phelps, & Malmberg, 1999; Oberauer, 2018). The dual process model of recognition memory states that two separate processes govern recognition: familiarity and recollection (Norman & O’Reilly, 2003; Opitz, 2010; Rugg & Yonelinas, 2003). Familiarity involves recognizing a piece of information or stimulus without accessing other information about the context in which the material was learned (Opitz, 2010). Recollection involves retrieving all the information about the stimulus

and the context in which it was presented (Opitz, 2010). The dual process model of recognition shows the complexity of recognition memory and the retrieval process (Jacoby & Dallas, 1981; Mulligan, Smith, & Spataro, 2016; Murnane, Phelps, & Malmberg, 1999). This study replicates and extends an experi ment that tested the dual process model’s predictions about the influence of context and repetition on recognition memory (Opitz, 2010). Our study replicates the behavioral portion of that study’s methodology, which measured reaction time and accuracy, and extends the previous work to examine the influence of word frequency on

*Faculty mentor

Vishwanath and Shive | Context Effects on Recognition Memory

recognition memory. Thus, this study has the poten tial to provide further evidence of the importance of context and repetition, while adding to the literature on how word frequency impacts memory. Repetition, Context, and Word Frequency Repeated presentations of a stimulus tend to improve memory encoding (Xue et al., 2010). When the conditions of encoding match each other across encoding sessions, similar patterns of neural activity associated with the episodic memory of the event are evoked. For example, an fMRI study of activation in cortical areas associated with object recognition and memory encoding showed similar patterns of activity across encoding trials for items that were subsequently recognized (Xue et al., 2010). However, simple repetition of materials does not have a significant influence on memory recall (Karpicke, 2012; Tulving, 1966) or recognition (Jacoby & Dallas, 1981). Contemporary researchers of memory have connected the depth of processing one uses to encode information with subsequent retrieval for the information (Baddeley & Hitch, 2017). For example, thinking about how you would use an item to survive on a desert island produces better free recall for the item than think ing about how pleasant it is (Nairne, Thompson, & Pandeirada, 2007). Furthermore, memory is better when the context during an encoding phase matches the context during retrieval than when there is a mis match between encoding and retrieval conditions (Godden & Baddeley, 1975). Context refers to the details of the surrounding environment in which information is learned or experienced. Contextual information can include low-level visual or auditory features such as background display (Chun, 2000), the visual position of a stimulus (Hollingworth, 2006), or background music (Coutinho & Scherer, 2017). It can also encompass higher-level features (i.e., features that rely on knowledge or previous experience) such as the emotional context of a scene (Finke, Zhang, Best, Lass-Hennemann, & Schächinger, 2018) or the entirety of the physical environment (Godden & Baddeley, 1975). The effects of contextual congruence can also occur when a study participant is unaware of the congru ence (Chun, 2000; Jiang & Sisk, 2019). Chun and Jiang (2003) compared visual searches for letters on displays that appeared only once during the course of the experiment and displays shown several times during the course of the experiment. Letters on these repeated displays always appeared in the same

location. Chun and Jiang showed that, although all visual searches got faster over the course of the experiment, searches were faster for items on the repeated displays. Furthermore, fewer than half of participants reported noticing the repeated displays. Although memory research has emphasized the importance of repetition and context, the effects of word frequency (i.e., how often an English word occurs in written and spoken language) are less clear. Word frequency has a significant effect on word recognition tasks (Brysbaert & New, 2009). In a lexical decision task, where a participant is shown a string of letters and asked to decide whether the letter string forms an English word, high frequency words are evaluated faster than low frequency words (Balota & Chumbley, 1984). This suggests easier lexical access to common words. However, word frequency has no significant effect on reaction time in tasks such as category verification (Andrews, 1992). Thus, there seem to be inconsistent effects of word frequency on different types of memory tasks. Dual Process Model of Recognition Memory The process of familiarity and recollection of the dual process model, which we introduced earlier, are suggested to retrieve different kinds of informa tion regarding a stimulus, where familiarity may be more characteristic of semantic memory and recollection is characteristic of episodic memory (Henson & Gagnepain, 2010; Oberauer, 2018). Some studies have proposed that familiarity is an automatic process whereas recollection is a controlled process (Jacoby, 1991; Oberauer, 2018; Yonelinas & Jacoby, 1995). Although there has not been consensus on a dual process or single process model, there is evidence that one distinct form of information can be retrieved without the other (Henson & Gagnepain, 2010; Jacoby, 1991; Oberauer, 2018; Opitz, 2010). The dual process model of recognition memory makes predictions about how context and rep etition impact memory for words. First, the dual process model predicts that participants should respond differently depending on whether words are learned within the same context or in differ ent contexts. Secondly, the model proposes that presenting a word in the same context multiple times facilitates stimulus binding, which associates a word with the contextual features that occurred during its presentation. The model predicts less stimulus binding when words are presented across contexts. Thus, the model predicts better memory

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

163

Context Effects on Recognition Memory | Vishwanath and Shive

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

164

for words presented in the same context at testing as encoding, as well as better memory for repeated words than non-repeated words. Stimulus binding can be tested by asking participants to make remember judgments and know judgments about words presented during the encoding phase. Remembering entails retrieving the contextual features in which a word appeared, while knowing does not. For example, if partici pants indicate that they remember the word in its context, as part of the recollection process, this corresponds to higher stimulus context binding, which is most likely to occur with words repeatedly presented in a single context (Norman & O’Reilly, 2003; Opitz, 2010). However, if participants indicate only knowing the word and not the contextual features in which it was presented, as part of the familiarity process, this would reflect lower stimulus context binding, which is likely to occur with words presented across all three contexts (Norman & O’Reilly, 2003; Opitz, 2010). Opitz (2010) tested the dual process model by examining recognition memory for pictures in different colored backgrounds and screen positions. During the learning phase of the study, several contextual features were manipulated: the number of presentations, position on the screen, screen background, and the encoding task question presented after a picture. The repeated pictures in this learning phase were divided into two groups: one group of pictures was repeated with the same contextual features and the other was repeated using multiple contextual features. During the retrieval phase, participants were asked to recognize which pictures they had seen before and whether they could remember seeing them in their context or merely knew that they had seen them before. Opitz (2010) collected both behavioral data and physiological data, using event-related poten tials, to determine the independence of familiarity and recollection processes. He found that repeated pictures were recognized faster than nonrepeated pictures, and participants were best at recognizing repeated pictures that occurred across different contexts and worst at recognizing pictures pre sented only once. In addition, remember and know judgments elicited different event-related poten tials. Specifically, remember judgments elicited stronger late responses (550–770 ms after stimulus presentation) in parietal recording sites than know judgments did. These results lend further support to the dual process model’s proposal that recol lection and familiarity are independent processes.

Present Study The current study expanded on the Opitz (2010) study in two ways. First, it examined memory for words rather than pictures. Second, the study examined whether word frequency affects recogni tion under repetition and context manipulations. The study evaluated four research questions about recognition memory: (a) are the reaction times faster for words when they are repeated than when presented once?, (b) is the number of correct responses greater for repeated words than nonrepeated words?, (c) are there differences in recognition for common and uncommon words?, and (d) are the know judgments greater for words presented across contexts and are the remember judgments greater for words presented within the same contexts? We predicted that our experimental manipula tions involving repetition, word frequency, and context would affect accuracy, reaction time, and remember/know judgments during recognition. We predicted greater accuracy (i.e., higher hit rates and lower false alarm rates) for words repeated three times compared to single presentations, as well as lower reaction times and greater accuracy for common words than uncommon words. We also predicted that know and remember judgments would differ based on within- and across-context presentations. Specifically, in line with the dual process model of recognition, we predicted a greater number of know judgments for words pre sented across different contexts during encoding and a greater number of remember judgments for words always presented in the same context during encoding.

Method Participants This study represents a replication with extension of the experiment described in Opitz (2010). Thus, we used the data reported in that article to estimate the required sample size for the current study. Because Opitz (2010) did not report effect sizes, we first used their reported F-statistic values and degrees of freedom values to calculate effect sizes for the reported statistical tests. The smallest of these effect sizes (for the hypothesis test comparing recognition across contexts versus within contexts) was η2p = .33. A power analysis using G*Power (Faul, Erdfelder, Lang, and Buchner, 2007) revealed that detecting a difference between two means for a within-groups design using α = .05, β = .20 and nonsphercity correction ε = 1 would require data from 19 participants.

Vishwanath and Shive | Context Effects on Recognition Memory

A total of 26 participants completed the experiment. All participants were recruited from Tennessee State University campus through recruit ment presentations in classrooms, recruitment across the campus, and through the SONA system, which is the subject pool management system of the university’s psychology department. Each partici pant was 18 years or older and enrolled as a student. One participant was excluded for not meeting the age criteria. Another participant had to be elimi nated from the analysis because the data collection program did not record the participant’s reaction time data. The average age of the remaining 24 participants was 20.91 years (age range: 18–38, SD = 4.28), and 17 of the participants were women. The students reported normal color vision and no light sensitivity. The students received extra credit for select courses for participation in the study. However, the extra credit offered for participation was only one of several extra credit opportunities that were available for the course so that participa tion in the study was not required for success in the course. Materials The stimuli were presented on an iMac with a 21.5-inch monitor with a display resolution of 1920 x 1080 pixels. We presented the experiment and collected the data using the Psychophysics Toolbox for MATLAB (Brainard, 1997; Kleiner, Brainard, Pelli, 2007; Pelli, 1997). The study used six lists of 42 words. We chose equal numbers of common and uncommon words from the Kučera and Francis written frequency scale, which measures word occurrences per million words (MRC, n.d.). All words had four to seven letters. The words were comprised of nouns, adjectives, and verbs as distinguished in the Medical Research Council Psycholinguistic Database. The words were not restricted to any single category to get enough unique uncommon and common list of words to use in the study that could be associ ated with each of the task questions. For each task question, approximately half of the words required a “yes” response. The uncommon words (e.g., her ring, benzene, cowhide, belfry, frown) had a mean Kučera and Francis frequency of 1.39 and a range of 1 to 3, and the common words (e.g., rifle, green, market, paper, writing) had a mean frequency of 147.07 and a range of 60 to 400. The concreteness and imageability of the words were between the scale of 1 and 7 as determined by the Medical Research Council Psycholinguistic Database.

The encoding and retrieval phases of the experiment were presented on varying background colors. The colors and RGB values are as follows: dark grey (RGB: 64, 64, 64), black (RGB: 0, 0, 0), white (RGB: 255, 255, 255), light grey (RGB: 128, 128, 128), and red (RGB: 255, 0, 0). In the encod ing phase, the words were presented at different positions on the screen aside from the center. The positions were left side of the screen, which was approximately 27° of visual angle to the left of the fixation point, and the right side of the screen, which was approximately 27° of visual angle to the right of the fixation point at a viewing distance of approximately 18 inches. Design The study used a 2 (frequency) x 2 (repetition) x 2 (context) experimental design to examine the effects of repetition of common and uncommon words in different contexts on recognition. The word frequency factor compared common and uncommon words, as indexed by the Kučera and Francis scale. The repetition factor had two levels: repeated words were presented three times during the encoding phase, whereas nonrepeated words were presented only once during the encoding phase. Study context referred to whether a word was presented in the same context each time it was presented during the encoding phase or whether it appeared across contexts during the encoding phase. Procedure The study was approved by Tennessee State University’s Institutional Review Board. After which, participants were recruited. The partici pants were first given an oral briefing about the purpose of the study and then directed to read the required consent form. After obtaining consent, participants completed a demographic question naire that asked about age, birth sex, gender identity, handedness, and whether the participant has normal or corrected to normal vision, normal color vision, and no light sensitivity. Next, participants began the experiment, which had two phases: an encoding phase, consist ing of three blocks, and a retrieval phase comprised in a single block. All responses were recorded using a keyboard. The participants were made aware that they could take breaks in between blocks, if and as needed.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

165

Context Effects on Recognition Memory | Vishwanath and Shive

Encoding phase. In the encoding phase, all words presented within a block had the same com bination of three contextual features: background color, screen position, and task question (see Figure 1). In one block, words were presented on the left side of the screen on a dark grey background, and participants were asked whether the item the word represented would fit inside a shoebox. In another block, words were presented on the right side of the screen on a black background, and participants were asked if the word was a noun. In the last block, words were presented at the center of the screen on a white background, and participants were asked if the item the word represented was a human creation. The three blocks were presented in a random order for each participant. The words presented in the experiment were chosen from six lists of 42 words from the Kučera and Francis written frequency scale. The six lists were first divided into two sets of three lists each. For each participant, one set was chosen to be presented in the encoding phase, and the other set was presented as distractors in the retrieval phase. The assignment of each list as target or distractor was randomized for each participant. We assigned words from the study lists to three blocks of trials to create different levels of

100 ms

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

166

radio Does it fit inside a shoebox?

forest Is it a human creation?

500 ms

100 ms

500 ms

100 ms

500 ms

FIGURE 1

balloon Is it a noun?

Figure 1. Illustration of the three blocks in the study phase, which consists of three combinations of the contextual features in each block. The contextual features are background colors (dark grey, white, and black), screen positions (left, center, and right), and task questions (Does it fit inside a shoebox? Is it a human creation? Is it a noun?). The words are presented for 500 ms, and the question appears after a 1000 ms interval. Note that the words are shown in white and black here for readability. During the experiment, words in all contexts were displayed using a red font.

repetition context. Words from the first list were only presented once during the encoding phase. To create a repetition-within-context level, one third of the words from the second list were pre sented three times within a particular block of the encoding phase (that is, in the same background color, at the same place on the screen, followed by the same encoding question). To create a repetition-across-contexts level, the words from the third list were presented once in each of the three blocks, so that they appeared once in each of the different contexts. The words from the three lists chosen for the encoding phase amounted to 98 trials in each block. The 98 trials of a block were comprised of 14 words from List 1, which were presented once (14 trials), 14 words from List 2, which were presented three times (42 trials), and all 42 words of List 3, which were presented once (42 trials). Each block took approximately five minutes to complete. At the start of each block, participants were asked to indicate a “yes” or “no” response using two marked keys on the keyboard to the task questions associated with the combination. Each word in the encoding phase was presented for 500 ms. After each word, there was a 1000 ms interval before the participant was asked to answer the task question with a yes or no response. Retrieval phase. In the retrieval phase, all words were presented at the center of the screen on a light grey background (see Figure 2). The words in the retrieval phase comprised of all the words from the set of three lists presented in the encoding phase (42 words from three lists, which add up to 126 words) and the remaining words from the set of three lists (42 words from three lists, which add up to 126 words), which functioned as the distractor words. All words were presented once in the retrieval phase. Thus, the retrieval phase comprised of a total of 252 trials that took approximately 25 minutes to complete. The words were presented for 500 ms each. Participants were asked to make an old judgment or a new judgment after 1000 ms of word presentation, using two marked keys on the keyboard. An old response indicated that they recognized the word from the encoding phase and a new response indicated that they did not recognize the word from the encoding phase. If participants made an old judgment, an additional question appeared ask ing them to make a know judgment or remember judgment using two different marked keys on the

Vishwanath and Shive | Context Effects on Recognition Memory

Results The data were analyzed to compare effects of word frequency, repetition, and contextual fea tures on accuracy and speed of recognition. All responses and reaction times for old words and new words were analyzed for overall effects and word frequency effects. However, only the old words could be analyzed for repetition and context effects. The new words were only presented in the retrieval phase and thus could not be assessed for repetition and context effects, because these fac tors were manipulated in the encoding phase. The repeated-measures Analysis of Variance (ANOVA) was used for analysis of all conditions, and multiple comparisons were corrected using the Bonferroni correction.

the above mentioned four conditions of responses are presented in Table 1. The average reaction times were assessed using these four response conditions as well (see Table 1). Old and new responses had high accuracy rates. To assess the overall and word frequency effects on accuracy, hit rates from all context and repetition conditions were compared to the false alarm rates. A repeated-measures ANOVA conducted for hit rates and false alarms for uncommon and common words revealed a main effect of response type, F(1, 23) = 204.40, p < .001, η2p = .89, frequency, F(1, 23) = 14.26, p = .001, η 2 p = .38, and an interaction of response type and frequency, F(1, 23) = 15.24, p = .001, η2p = .39. For old words, the pairwise comparison revealed that hits (M = 0.84, 95% CI [.81, .89]) were greater than false alarms (M = 0.21, 95% CI [.11, .31]). The significant FIGURE 2 500 ms

keyboard. The remember judgment indicated that participants remembered the contextual features in which the words were presented, for example, left side or dark grey background. The know judg ment indicated that participants remembered only the word and none of the other features.

Have you seen the word before or is it a new word?

100 ms

Accuracy To assess the accuracy of old responses, hit rates and false alarms were calculated and compared. Hit rate was defined as number of correct old responses for old words divided by the total number of old words. False alarm rate was defined as number of old responses to new words divided by the total number of new words. The data were also assessed for accuracy of new responses by comparing correct new response rate and false negative rate. Correct new response rate was all new responses to new words divided by the total number of new words, and false negative rate was all new responses to old words divided by all old words. The proportions for

radio

Old

New

Do you remember the context details of the word or just the word?

teeth

Figure 2. Simulation of a single trial in the test phase where words appeared at the center of the screen on a light grey background. The words are presented for 500 ms, and the question appears after a 1000 ms interval. A “yes” response to the question leads to another question. A “no” response to the question starts a new trial. Note that the words are shown in black here for readability. During the experiment, words in all contexts were displayed using a red font.

TABLE 1 Accuracy and Reacton Time Data Old Words Overall

New Words

Frequency Uncommon

Common

Repetition Single

Across Context

Overall Within Context

Frequency Uncommon

Common

Proportions Old Repsonses

0.85(0.02)

0.85(0.05)

0.85(0.02)

0.73(0.03)

0.92(0.02)

0.89(0.02)

0.21(0.02

0.16(0.05)

0.26(0.05)

New Responses

0.79(0.05)

0.84(0.05)

0.74(0.05)

0.27(0.03)

0.08(0.02)

0.11(0.02)

0.15(0.02

0.15(0.02)

Old Responses

443 (29.66)

396 (25.86)

489 (37.56)

503 (44.17)

402 (26.28)

441 (34.15)

609 (71.50)

565 (68.62)

624 (88.39)

SUMMER 2020

New Responses

894 (102.44)

709 (97)

984 (126.84)

885 (130.57)

1036 (366.39)

1046 (137.02)

565 (48.50)

514 (50.50)

578 (49.09)

PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

Reaction Time

Note. Reaction times (ms) and mean proportions (+SEM) for old and new responses to old and new words.

167

Context Effects on Recognition Memory | Vishwanath and Shive

interaction effect showed that common words were more likely to be false alarms than uncommon words, but common and uncommon words did not differ for hits. Figure 3 illustrates this interaction. Likewise, correct new responses (M = 0.79, 95% CI [.69, .89]) were higher than false negatives (M = 0.15, 95% CI [.11, .19]). The uncommon words were more accurately recognized as new words than common words, but uncommon and common words did not differ in the number of false negatives they produced. Reaction times were lower for hits and correct new responses. The repeated-measures ANOVA for hit rates, false alarms, and word frequency showed a main effect of response type, F(1, 23) = 9.15, p = .006, η 2p = .28, but no effect of frequency, F(1, 23) = 1.95, p = .17, η2p = .07, or interaction, F(1, 23) = 0.11, p = .73, η2p = .001. The pairwise compari sons showed that hits (M = 442ms, 95% CI [381, 504]) were recognized faster than false alarms (M = 594ms, 95% CI [467, 721]). The analysis was repeated for correct new responses, false negatives, and word frequency. The ANOVA revealed a main effect of response type, F(1, 23) = 17.11, p < .001, η2p = .42, but no effect of frequency, F(1, 23) = 0.84, p = .36, η2p = .03, or interaction, F(1, 23) = 1.57, p = .22, η2p = .06. The pairwise comparisons showed that correct new responses (M = 521 ms, 95% CI [448, 595]) were faster than false negatives (M = 805 ms, 95% CI [640, 970]). Thus, the correct responses for both old and new words were faster than incorrect responses. Repetition and Context To examine the effects of repetition, all old words were grouped into repeated and nonrepeated words. The context effects were analyzed by further FIGURE 3

Proportion of Responses

1.0

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

168

0.9 0.8 0.7 0.6 0.5

Uncommon Common

0.4 0.3 0.2 0.1 0.0

Fit Rate

False Alarm Response Type

Figure 3. Comparison of hit rates and false alarm rates and common and uncommon words using a repeated-measures Analysis of Variance. Error bars show SEM.

classifying the repeated words into across-context and within-context. The proportions of hits and false negatives for all repeated and nonrepeated conditions, along with their reaction times, are presented in Table 1. However, an ANOVA was conducted for the grouped conditions using the hit rate. The false negatives were excluded from the analysis because participants were highly accu rate in their new responses. The word frequency was also analyzed by subdividing all groups into uncommon and common words. The accuracy and reaction times of hit rates and word frequency were compared in all repetition and context groups. Old words were recognized best when repeated across contexts. The ANOVA for repeated and nonrepeated words using hit rate as the dependent variable showed a significant main effect of repeti tion, F(1, 23) = 74.05, p < .001, η2p = .76, but no effect of word frequency, F(1, 23) = 0.00, p = .98, η2p < .001, or an interaction, F(1, 23) = 0.02, p = .87, η2p = .001. The pairwise comparison revealed that hit rates for repeated words (M = 0.90, 95% CI [.87, .94]) were higher than for words presented only once (M = 0.72, 95% CI [.66, .79]), as shown in Figure 4. The context effects were assessed using the ANOVA for context repetitions and word frequency of hit rates. The results showed a main effect of context repetition, F(1, 23) = 5.06, p = .03, η2p = .18, but not for frequency, F(1, 23) = 0.10, p = .75, η2p = .004, or interaction, F(1, 23) = 1.53, p = .23, η2p = .06. Pairwise comparisons showed that words repeated across context had a higher hit rate (M = 0.92, 95% CI [.88, .96]) than within-context repetitions (M = 0.89, 95% CI [.85, .93]). Figure 4 depicts this effect of the two repetition contexts on the hit rate, along with hit rate for single presenta tions for comparison. Reaction times were lower for uncommon words but showed no effect for repetition or context. The repeated and nonrepeated groups, and word frequency ANOVA for hit rate reac tion times, showed a main effect of repetition, F(1, 23) = 6.89, p = .01, η2p = .23, and frequency, F(1, 23) = 14.40, p = .001, η2p = .38, but no interac tion effect, F(1, 23) = 1.28, p = .26, η2p = .05. The pairwise comparisons showed that repeated words (M = 418 ms, 95% CI [361, 475]) were recognized faster than nonrepeated words (M = 505 ms, 95% CI [411, 600]), and uncommon words (M = 409 ms, 95% CI [349, 469]) were recognized faster than common words (M = 514 ms, 95% CI [425, 603]) for both repeated and nonrepeated words.

Vishwanath and Shive | Context Effects on Recognition Memory

Evaluating the Dual Process Model In line with the dual process model, we predicted that participants should respond differently in the recognition task depending on if they saw the words multiple times within a single context or spread across the three contexts. To test this, we analyzed the proportion of remember judgments and calculated a corrected know proportion to reflect pure familiarity (Opitz, 2010). Corrected know responses were calculated by using the formula K/(1-R) where K refers to proportion of know responses for hits and R refers to propor tion of remember responses for hits. Corrected know response rate was calculated, as it assumes the independence paradigm by treating the two processes as being mutually exclusive (Yonelinas & Jacoby, 1995). Hit rates were used in this analysis because only old responses prompted the know or remember question to participants. The hits were divided into groups for across- and within-context repeated words, and word frequency. Remember response rates and corrected know response rates were calculated for all groups. Across- and within-context repetitions did not affect remember and know judgments. We performed a three-way ANOVA for word frequency, context, and memory judgment type (corrected know vs. remember). There was no main effect for frequency, F(1, 18) = 3.79, p = .16, η2p = .17, context, F(1, 18) = 1.52, p = .23, η2p = .78, judgment type, F(1, 18) = 5.04, p = .06, η 2 p = .22, or context and judgment type interaction, F(1, 18) = 0.00, p = .98, η2p < .001. There was, however, a significant three-way interaction of memory judgment, word frequency, and context, F(1, 18) = 5.04, p = .04, η2p = .22. Hit rate was highest for know responses for uncommon words presented across contexts. Overall, the study context could not predict remember or know responses. Repeated presenta tions within the same context did not produce a higher proportion of remember responses, and across-context repetitions did not correspond to higher know responses.

Discussion This project investigated the effects of stimulus repetition, context, and word frequency on rec ognition memory. We found that word frequency affected the speed and accuracy of recognition for both common and uncommon words. Contrary to our predictions, we found that uncommon words were recognized faster than common words, and common words were more likely to be false alarms. However, the effect size (η2p = .38) for this pattern was small and could indicate a possible random effect. It is also possible that the scale we used to select common and uncommon words (the Kučera and Francis frequency scale) did not produce lists of words that differed only in word frequency. More recent frequency scales use a larger corpus and include spoken language in determining the word frequency (Brysbaert & New, 2009). The study revealed that repetition and context have prominent effects on recognition. Repeated words were recognized both faster and more accurately than non-repeated word. In addition, we found that repeating words across the three contexts resulted in better recognition for words, whereas recognition for words repeated within a single context was less accurate. However, the repeated words within a single context were still more accurate than single presentations. The descending order of accuracy from across-context to within-context, and lastly to nonrepeated words was consistent with Opitz (2010). High accuracy for across-context presenta tions may indicate that participants were able to recognize words better when they saw the words in varying background colors and screen positions. This may speak to effects of extrinsic versus intrinsic context for recognition memory, mentioned by Godden and Baddeley (1980). They proposed that FIGURE 4 0.9 0.8 0.7 0.6 Hit Rate

The reaction times were further analyzed for context and word frequency effects. The ANOVA for hit rates showed a main effect of frequency, F(1, 23) = 7.88, p = .01, η2p = .25, but no main effects for context, F(1, 23) = 2.80, p = .10, η2p = .10, or interaction, F(1, 23) = 0.03, p = .85, η 2p < .001. Uncommon words (M = 381 ms, 95% CI [327, 435]) were recognized faster than common words (M = 457 ms, 95% CI [384, 531]) in both acrossand within-context repetition conditions.

0.5 0.4 0.3 0.2 0.1 0.0

Single Within Across Repetition Condition

Figure 4. Hit rates for repeated words, within and across contexts, and nonrepeated words. Error bars show SEM.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

169

Context Effects on Recognition Memory | Vishwanath and Shive

extrinsic context such as features of the experiment room or perhaps the background color of the screen in the present study may not be as important for recognition as the intrinsic context of the words such as semantics. Although the dual process model correctly predicted the effects of repetition, we did not find evidence to support its predictions about the effects of context on remember and know judgments, which other researchers were able to find with picture stimuli (Opitz, 2010). However, this may have been because of a repetition lag between repeated word presentations in our study. The repeated words within the same context were not repeated in continuous trials (that is, one after another), but rather there were several other words presented between the first, second, and third presentations of the same word. This lag in repetition might have produced weaker binding of the contextual features to the word. Future work examining recognition memory for words should consider the impact of the contextual congruence on memory performance. In the present study, the test conditions used a neutral context. In other words, the features of the test environment (word position and background color) were the same for all words, regardless of their context during learn ing. Future work could introduce congruent and incongruent test conditions. For example, if a word was presented on the left side then this aspect could be presented either in congruence (same side) or incongruence (opposite side) during the test recognition phase. Furthermore, context features and dual process of recognition and familiarity can be compared in different age groups to see if there is a decline in recognition abilities among young and old adults. Additionally, recognition processes can be examined using imaging and EEG to investigate the physiological bases for a single or dual process model.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

170

Conclusion We examined if word frequency, contextual fea tures, and repetition affect recognition memory for words, along with examining the dual process model of recognition memory. The results showed significant main effects of repetition and frequency on recognition. The study, however, did not find any evidence to support the dual process model’s prediction about independent processes of recogni tion and stimulus binding. Further experiments and models need to be explored to provide better clarity of the retrieval process of recognition memory.

References Andrews, S. (1992). Frequency and neighborhood effects on lexical access: Lexical similarity or orthographic redundancy? Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 234–254. http://dx.doi.org/10.1037/0278-7393.18.2.234 Baddeley, A. D., & Hitch, G. J. (2017). Is the levels of processing effect languagelimited? Journal of Memory and Language, 92, 1–13. https://doi.org/10.1016/j.jml.2016.05.001 Balota, D. A., & Chumbley, J. I. (1984). Are lexical decisions a good measure of lexical access? The role of word frequency in the neglected decision stage. Journal of Experimental Psychology: Human Perception and Performance, 10, 340–357. http://dx.doi.org/10.1037/0096-1523.10.3.340 Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10, 433–436. http://dx.doi.org/10.1163/156856897X00357 Brysbaert, M., & New, B. (2009). Moving beyond Kučera and Francis: A critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for American English. Behavior Research Methods, 41, 977–990. https://doi.org/10.3758/BRM.41.4.977 Chun, M. M. (2000). Contextual cueing of visual attention. Trends in Cognitive Sciences, 4, 170–178. https://doi.org/10.1016/S1364-6613(00)01476-5 Chun, M. M., & Jiang, Y. (2003). Implicit, long-term spatial contextual memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29, 224–234. https://doi.org/10.1037/0278-7393.29.2.224 Coutinho, E., & Scherer, K. R. (2017). The effect of context and audio-visual modality on emotions elicited by a musical performance. Psychology of Music, 45, 550–569. https://doi.org/10.1177/0305735616670496 Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. https://doi.org/10.3758/BF03193146 Finke, J. B., Zhang, X., Best, D. R., Lass-Hennemann, J., & Schächinger, H. (2018, April 10). Self-resemblance modulates processing of socio-emotional pictures in a context-sensitive manner: Evidence from startle modification and heart rate deceleration. Journal of Psychophysiology. Advance online publication. http://dx.doi.org/10.1027/0269-8803/a000216 Godden, D., & Baddeley, A. (1980). When does context influence recognition memory? British Journal of Psychology, 71, 99–104. https://doi.org/10.1111/j.2044-8295.1980.tb02735.x Godden, D. R., & Baddeley, A. D. (1975). Context‐dependent memory in two natural environments: On land and underwater. British Journal of Psychology, 66, 325–331. http://dx.doi.org/10.1111/j.2044-8295.1975.tb01468.x Henson, R. N., & Gagnepain, P. (2010). Predictive, interactive multiple memory systems. Hippocampus, 20, 1315–1326. https://doi.org/10.1002/hipo.20857 Hollingworth, A. (2006). Scene and position specificity in visual memory for objects. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, 58–69. https://doi.org/10.1037/0278-7393.32.1.58 Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30, 513–541. https://doi.org/10.1016/0749-596X(91)90025-F Jacoby, L. L., & Dallas, M. (1981). On the relationship between autobiographical memory and perceptual learning. Journal of Experimental Psychology: General, 110, 306–340. http://dx.doi.org/10.1037/0096-3445.110.3.306 Jiang, Y. V., & Sisk, C. A. (2019). Contextual cueing. In S. Pollman (Ed.), Springer neuromethods: Spatial learning and attentional guidance. New York, NY: Springer. Karpicke, J. D. (2012). Retrieval-based learning: Active retrieval promotes meaningful learning. Current Directions in Psychological Science, 21, 157–163. https://doi.org/10.1177/0963721412443552 Kleiner, M., Brainard, D., & Pelli, D. (2007, August). What’s new in psychtoolbox-3? In 30th European Conference on Visual Perception (ECVP 2007) (p. 14). Pion Ltd. MRC Psycholinguistic Database. (n.d.). Retrieved from http://websites. psychology.uwa.edu.au/school/MRCDataBase/uwa_mrc.htm Mulligan, N. W., Smith, S. A., & Spataro, P. (2016). The attentional boost effect and context memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 42, 598–607. http://dx.doi.org/10.1037/xlm0000183 Murnane, K., Phelps, M. P., & Malmberg, K. (1999). Context-dependent recognition memory: The ICE theory. Journal of Experimental Psychology: General, 128, 403–415. http://dx.doi.org/10.1037/0096-3445.128.4.403

Vishwanath and Shive | Context Effects on Recognition Memory

Nairne, J. S., Thompson, S. R., & Pandeirada, J. N. (2007). Adaptive memory: Survival processing enhances retention. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33, 263–273. http://dx.doi.org/10.1037/0278-7393.33.2.263 Norman, K. A., & O’Reilly, R. C. (2003). Modeling hippocampal and neocortical contributions to recognition memory: A complementary-learning-systems approach. Psychological Review, 110, 611–646. http://dx.doi.org/10.1037/0033-295X.110.4.611 Oberauer, K. (2018). On the automaticity of familiarity in short-term recognition: a test of the dual-process assumption with the PRP Paradigm. Journal of Cognition, 1, 20. http://doi.org/10.5334/joc.21 Opitz, B. (2010). Context-dependent repetition effects on recognition memory. Brain and Cognition, 73, 110–118. https://doi.org/10.1016/j.bandc.2010.04.003 Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics: Transforming numbers into movies. Spatial Vision, 10, 437–442. http://dx.doi.org/10.1163/156856897X00366 Rugg, M. D., & Yonelinas, A. P. (2003). Human recognition memory: A cognitive neuroscience perspective. Trends in Cognitive Sciences, 7, 313–319. https://doi.org/10.1016/S1364-6613(03)00131-1 Tulving, E. (1966). Subjective organization and effects of repetition in multi-trial

free-recall learning. Journal of Verbal Learning and Verbal Behavior, 5, 193–197. https://doi.org/10.1016/S0022-5371(66)80016-6 Xue, G., Dong, Q., Chen, C., Lu, Z., Mumford, J. A., & Poldrack, R. A. (2010). Greater neural pattern similarity across repetitions is associated with better memory. Science, 1193125. https://doi.org/10.1126/science.1193125 Yonelinas, A. P., & Jacoby, L. L. (1995). The relation between remembering and knowing as bases for recognition: Effects of size congruency. Journal of Memory and Language, 34, 622–643. https://doi.org/10.1006/jmla.1995.1028 Author Note. Abhilasha Vishwanath, https://orcid.org/0000-0002-8283-0141, Department of Psychology, Tennessee State University; Joshua Shive, Department of Psychology, Tennessee State University. Abhilasha Vishwanath is now at the Department of Psychology at University of Arizona, Tucson, AZ. The authors want to thank Ian Neath for helpful conversations regarding this project. Correspondence concerning this article should be addressed to Abhilasha Vishwanath. E-mail: avishwanath@email.arizona.edu

171

https://doi.org/10.24839/2325-7342.JN25.2.172

Predicting Student-Athlete Mental Health: Coach–Athlete Relationship Megan Powers , Jana Fogaca Oregon State University

, Regan A. R. Gurung*

, and Callan M. Jackman

ABSTRACT. Student athletes must balance numerous challenges as they work at their academics and their sport. In this context, social support, especially from coaches, could have the potential of contributing to these athletes’ well-being. The present study aimed to investigate if coach–athlete relationships could predict college student athletes’ mental health outcomes (i.e., well-being, depression, and anxiety) beyond the known effects of gender and personality on mental health. Student athletes (N = 79, 56 men, 23 women) between 18 and 23 years of age (M = 19.50, SD = 1.25) completed measures of depression, anxiety, psychological quality of life, and coach–athlete relationship online. Results showed that both personality and coach–athlete relationships were significantly correlated to mental health outcomes. Multiple regression analyses showed the predictive power of the coach–athlete relationship over gender and personal factors in the prediction of depression and psychological well-being, but not anxiety. Results provide support for the importance of the coach–athlete relationship for athletes’ well-being, although more research with larger and more diverse samples is necessary to confirm this relationship.

Open Data and Open Materials badges earned for transparent research practices. Data and materials are available at https://osf.io/efvb9/

Keywords: mental health, student athlete, depression, anxiety, coach–athlete relationship

M SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

172

ental health issues in college are rising precipitously. The proportion of college students diagnosed with mental health conditions increased from 21.9% in 2007 to 35.5% in 2017 (Lipson, Lattie, & Eisenberg, 2018). Collegiate athletes may be at even more risk for mental health issues than nonathlete students. Together with overcoming developmental challenges, such as becoming independent and coping with uncertainty found in college (Chickering, 1969; Pascarella & Terenzini, 2005), student athletes contend with extra pressures, such as competition and athletic lifestyles, that nonathletes do not endure (Etzel, Watson, Visek, & Maniar, 2006). Social support may play an important role in student-athlete mental health

(Hammen, 2005), especially the relationship with coaches (Lentz, Kerins, & Smith, 2018). In the present study, we examined major contributors to college athletes’ mental health with a special focus on the coach–athlete relationship. The National Collegiate Athletic Association (NCAA) has over 360,000 student athletes who attend and compete at universities around the United States. Meeting academic challenges with the additional rigors of athletic competition can unduly tax students. Not surprisingly, stress plays a significant role in the mental health of high-school and college athletes (Lentz et al., 2018). The average student athlete exhibits similar or higher rates of depressive disorders than nonstudent athletes (Etzel, 2009; Wolanin, Gross, & Hong,

*Faculty mentor

Powers, Fogaca, Gurung, and Jackman | Student-Athlete Mental Health

2015; Wolanin, Hong, Marks, Panchoo, & Gross, 2016). In a major study of college students, 28% of the female and 21% of the male student athletes reported symptoms of depression in the last 12 months; 31% of male and 48% of female student athletes reported symptoms of anxiety (Davorean & Hwang, 2014). Even more worrisome, a retrospec tive study from 2003 to 2012 showed that suicide represented the cause of 7.3% of NCAA student athletes’ deaths during these nine seasons (Rao, Asif, Drezner, Toresdahl, & Harmon, 2015). Mental health is related to personal, inter personal, and environmental factors, such as personality, social support, and life stressors (Hammen, 2005; Kotov, Gamez, Schimidt, & Watson, 2010; Oexle & Sheehan, 2020). Among personal factors, gender (Storch, Storch, Killiany, & Roberti, 2005; Wolanin et al., 2016) and personality (Kotov et al., 2010) have been linked to mental health in the past (Brewer, 1993). For example, a survey of 398 undergraduate students showed that female student athletes had higher social anxiety and depression than male athletes and nonathlete students (Storch et al., 2005). The authors noted that women may be exposed to a greater number of stressors during their collegiate careers, internalizing these stressors more than their male counterparts, and feeling less satisfied with their overall collegiate experience (Storch et al., 2005). Another survey of 465 NCAA Division I student athletes found that women showed a risk 1.84 times higher than men of reporting depressive symptoms (Wolanin et al., 2016). Specific personality traits such as conscientious ness, extroversion, agreeableness, neuroticism, and openness—or the Big Five traits—may also relate to mental health issues. Personality traits have been linked to competition anxiety (Binboga, Guven, Çatikkas, Bayazit, & Tok, 2012; Velikic, Knezevic, & Rodic, 2014) and social physique anxiety in athletes (Cangur, Yaman, Ercan, Yaman, & Tok, 2017). A meta-analysis has also shown a link between the Big Five personality traits and anxiety, depressive, and substance abuse disorders in the general population, especially high neuroticism and low conscientiousness (Kotov et al., 2010). Kotov et al. suggested that personality traits are strongly related to psychopathology and should be taken into consideration in research and clinical practice. Besides the importance of personality and gender, environmental factors can also have a significant effect on mental health. Specifically, mental health issues among college student athletes

can be linked to the number of daily stressors that they face, which in the long term could lead to mental illness, if not well managed (Davoren & Hwang, 2014; Hammen, 2005; Wenzel, Glanz, & Lerman, 2002). Student athletes must maintain a full course load, work out to meet the physical demands necessary to succeed, adapt to frequent traveling for competition, and cope with injuries, while experiencing public pressure to perform (Ferrante & Etzel, 2009). Student athletes’ yearround training also results in athlete-specific physical and psychosocial demands (Etzel et al., 2006). Overall, student athletes cope with various stressors while adapting to the college environ ment and finding a well-balanced life between sports, school, and social life (Beauchemin, 2014). Finding the right balance in student athletes’ lives is an added challenge because each individual is unique in the ways these aspects work together and the ways in which they do not (Whitehead & Senecal, 2019). Because of the significant impact of stress on mental health, interpersonal factors that help athletes cope with these stressors should also be considered when analyzing student athletes’ men tal health. A factor known to act as a buffer from stress is social support (Lu et al., 2016; Wenzel et al., 2002). Social support has been operationalized in different ways, and measures of perceptions, receipt, networks, and quality have all been related to health (Sarason, Sarason, & Gurung, 2001). Among college students, social support has been related to lower risk of presenting depressive and anxiety symptoms (Hefner & Eisenberg, 2009; Merianos, Nabors, Vidourek, & King, 2013). In addition, student athletes who show higher social connectedness tend to have fewer symptoms of depression (Armstrong & Oomen-Early, 2009). Teammates’ social support negatively correlates to depression in female student athletes (Hagiwara, Iwatsuki, Isogai, Van Raalte, & Brewer, 2017). Despite the importance of social support for better mental health outcomes, the influence of social support from the coach, and consequently the coach–athlete relationship, has not been satis factorily explored (Felton & Jowett, 2013b; Felton & Jowett, 2017). Coaches can be an important source of support and can instill confidence in their athletes (Lentz et al., 2018; Lu et al., 2016). Coaches can also be a source of stress with a poor coach–athlete relationship adding stress in the athletes’ lives (Chyi, Lu, Wang, Hsu, & Chang, 2018). In contrast, a good relationship could help

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

173

Student-Athlete Mental Health | Powers, Fogaca, Gurung, and Jackman

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

174

in the identification of pre-existing mental health issues and disclosure (Schary, 2019). Although the role of the coach has been studied, most research focuses on how coaches influence motivation and satisfaction (e.g., Langan, Toner, Blake, & Lonsdale, 2015; Raabe & Zakrajsek, 2017; Riley & Smith, 2011; Wu, Lai, & Chan, 2014). The coach– athlete relationship is strongly related to athletes’ basic psychological needs and has the potential to be helpful during physically, psychologically, and emotionally challenging times (Choi, Cho, & Huh, 2013). Consequently, our working definition of social support involves the provision of emotional, esteem, informational, and tangible aid by the coach (Koh, Kokkonen, & Law, 2019). Research on social support in general, and from the coach in particular, reinforces the idea that better understanding the coach–athlete rela tionship may help predict athletes’ mental health. For example, research on basic needs satisfaction provided by the coach has shown that support from a coach predicts well-being outcomes (Davis & Jowett, 2014; Felton & Jowett, 2017). In one survey of 215 athletes, perceived social support from the coach predicted athletes’ basic needs satisfaction (Felton & Jowett, 2013a). Support from the coach has also been related to other athlete well-being factors such as vitality, positive and negative affect, and physical self-concept (Felton & Jowett, 2013b). Coach–athlete attachment anxiety was related to difficulties in emotional regulation, which in turn was linked to aggression, alcohol use, and psycho logical distress (Hebard, 2015). More research is necessary to clarify the association between coach–athlete relationships and mental health. As illustrated, multiple variables may influence student athletes’ mental health, including gender, personality, and social support. Despite the appar ent importance of the athletes’ relationships with their coaches to their mental health, this variable has not been adequately studied in relation to important mental health outcomes, such as depres sion and anxiety. The aim of the present study was to investigate if the variable coach–athlete relation ship can predict student athletes’ mental health after controlling for the effect of personal factors. Specifically, we tested the extent to which coach– athlete relationships predicted student-athlete mental health beyond the influence of gender and personality. Identifying specific variables that are connected to student athletes’ mental health will help professionals develop more effective interven tions to prevent mental health issues.

Method Participants The sample consisted of 56 female and 23 male student athletes at a midsized Midwestern public university (Division I of the NCAA). Participants ranged between 18 and 23 years old (M = 19.50, SD = 1.25) and from college first-year students to college seniors. Student athletes iden tified themselves as European American (n = 77), African American (n = 1), and Asian (n = 1). Student athletes averaged 14.44 enrollment credits, which ranged from 12–18 credits (SD = 1.69). The average student athlete practiced 2.93 hours per day (SD = 2.14; range 0–5). The average number of days the student athlete practiced each week was 5.77 (SD = 0.74; range 3–7). Participants took part in 16 sports, and the highest represented sports were swimming and diving (n = 26), volleyball (n = 11), and soccer (n = 11). Measures Student athletes answered demographic questions concerning age, class, number of course credits, ethnicity, and the number of hours of participation in their specific sport each day and each week. Participants also completed questionnaires about their personality traits, coach–athlete relationship, and mental health. Personality. We measured personality with a questionnaire based on the Big Five factors, the Ten Item Personality Inventory (Gosling, Rentfrow, & Swann, 2003). It consists of 10 statements, two for each factor, that are rated on a 7-point Likert-type scale ranging from 1 (disagree strongly) to 7 (agree strongly). Sample statements included “extraverted, enthusiastic” and “open to new experiences, com plex.” Gosling et al. demonstrated that the Ten Item Personality Inventory has adequate validity and reliability, which makes it a good option to decrease the size of surveys involving Big Five traits while maintaining good psychometric standards. Given that each personality trait is only measured with two items, internal reliability measurements are not commonly calculated. Coach–athlete relationship. The Coach–Athlete Relationship Questionnaire (CART-Q) (Jowett & Ntoumanis, 2004) was used to assess the coach– athlete relationship. Jowett and Ntoumanis (2004) have shown that this scale has good validity and reli ability. This measure has 11 items and 3 subscales. The subscales consist of commitment (3 questions), closeness (4 questions), and complementarity (4 questions). The student athletes answered the

Powers, Fogaca, Gurung, and Jackman | Student-Athlete Mental Health

CART-Q questions based on their relationship with their head coach. Sample questions include “I feel close to my head coach” and “I respect my head coach.” The student athletes selected a response based on a 7-point Likert-type scale, ranging from 1 (strongly disagree) to 7 (strongly agree). The Cronbach α for CART-Q was .98. Mental health. To draw a comprehensive assessment of the athletes’ mental health, we used measurements of mental disorders (i.e., depression and anxiety) and of a positive element of mental health (i.e., psychological quality of life). Each scale has been validated in previous studies and shown to have convergent and divergent validity. Positive elements of health. WHO-Quality of Life (WHOQOL-BREF) is a brief version of the World Health Organization Quality of Life assessment (The WHOQOL Group, 1998). It has been tested in a sample of people across 23 countries, showing sound validity and reliability (Skevington, Lotfy, & O’Connell, 2004). The scale consists of 26 questions and 4 domains (physical health, psychological health, social relationships, and environmental health), but only psychologi cal health was used in the present study. Sample statements included, “How much do you enjoy life?” and “How satisfied are you with yourself?” For Questions 1 through 4, responses ranged from 1 (not at all) to 5 (an extreme amount). For Question 5, the statements ranged from 1 (very dissatisfied) to 5 (very satisfied). For Question 6, the statements ranged from 1 (never) to 5 (always). Reliability was high, Cronbach α = .83. Negative elements of health. The Center for Epidemiological Studies-Depression Scale Revised (CESD-R; Eaton, Muntaner, Smith, Tien, & Ybarra, 2004) consists of 20 statements on a 5-point Likerttype scale ranging from 1 (not at all or less than 1 day last week) to 5 (nearly every day for two weeks). Sample items are comprised of “Nothing made me happy” and “I felt like I was moving too slowly.” This scale has shown good validity and reliability and is vastly used in psychiatric epidemiology, with the strength of being atheoretical (Eaton et al., 2004; Van Dam & Earleywine, 2011). The CESD-R value that determines if an individual is considered clinically depressed is a score of 16 or higher out of the total 20 questions. The Cronbach α for the CESD-R was .86. The Beck Anxiety Inventory Manual consists of 21 statements regarding how the individual felt within the past month (Beck & Steer, 1993). Participants use a 4-point response scale, ranging

from 0 (not at all) to 3 (severely – it bothered me a lot). Sample statements included are “numbness or tingling” and “difficulty in breathing.” The Beck Anxiety Inventory Manual showed strong reli ability with a Cronbach’s α of 0.89. A score of 1–21 indicates low anxiety, a score of 21–35 indicates moderate anxiety, and a score greater than 36 indicates high anxiety. Individuals scoring in the range of moderate to high anxiety should follow up with a medical professional. Procedure Athletes completed the questionnaires on Qualtrics during the month of November. After approval of the Institutional Review Board, the university athletic department’s academic coordinator sent emails to every student athlete (n = 247) and included an invitation to participate in the online survey. Seventy-nine athletes completed surveys for a 32% response rate. Participants gave consent to participate in the survey, provided the background information requested, and completed each of the surveys listed above in this report. Participants filled in their first and last names at the end of the survey to allow clinical referrals based on clinical levels of depression or anxiety. Only the researchers had access to the data. Data Analysis We used the Statistical Package for the Social Sciences for three major data analyses. First, we conducted descriptive analyses of all variables, both for the complete sample and separated by gender. Next, we correlated the major predictor variables personality and coach–athlete relation ship subscales with measures of mental health (depression, anxiety, and psychological quality of life), controlling for gender. Finally, we used three hierarchical multiple regression analyses to assess if the coach–athlete relationship could predict each of the major mental health measures beyond gender and personality. We entered gender and personality in the first step and coach–athlete relationship in the second to test if it predicted variance in mental health variables over and above gender and personality.

Results Table 1 illustrates the mean and standard deviations of all variables presented in the survey. We used an Analysis of Variance (ANOVA) to assess if there were significant differences between genders on the mental health variables. All descriptive data

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

175

Student-Athlete Mental Health | Powers, Fogaca, Gurung, and Jackman

are shown in Table 1. Men and women varied on five measures in the study. Men showed higher scores on the WHOQOL scale, F(1, 72) = 4.08, p = .047, and three measures of the coach–athlete relat ionship, and lower scores on emotional stability, F(1, 72) = 11.18, p = .001. Consistent with our focus on this important relationship, our results show men reporting higher levels of complementar ity, F(1, 72) = 4.75, p = .033, and commitment, F(1, 72) = 5.88, p = .018, on the CART-Q. Associations Between Major Variables Table 2 shows partial correlations between each of the mental health variables and the predictor variables, controlling for gender. Results show many significant, moderate correlations between the variables examined. Consistent with the main focus of this article, the strongest associations are seen between well-being, depression, and the sub scales of coach–athlete relationship. In addition, emotional stability correlated significantly with all mental health measures. It was positively and strongly correlated with quality of life, and nega tively and moderately correlated to depression and anxiety. The correlations supported the importance of including measures of personality, especially emotional stability, and coach–athlete relationship measures in the hierarchical regression analyses. TABLE 1 Descriptive Characteristics for Individual Variables by Gender Variable

Male (n = 23) M(SD)

Female (n = 56) M(SD)

Ten Item Personality Inventory Extraversion

9.36 (2.97)

9.45 (2.69)

Agreeableness

8.73 (2.60)

9.00 (1.96)

Conscientiousness

11.55 (2.22)

11.59 (1.98)

Emotional Stability*

10.95 (2.36)

8.73 (2.71)

Openness

10.73 (2.62)

10.61 (1.88)

55.52(15.09)

45.66(17.24)

16.29 (5.47)

12.70 (5.77)

23.05 (6.33)

19.54 (7.23)

Coach–Athlete Relationship Questionnaire Commitment

Closeness

21.90 (5.33)

18.22 (6.92)

World Health Organization- Quality of Life*

Complementarity

23.95 (3.76)

21.88 (4.00)

Center for Epidemiological StudiesDepression Scale Revised

10.20(14.11)

13.79(11.18)

6.15 (9.50)

9.43 (7.31)

Beck Anxiety Inventory

Note. * Indicates statistically significant (p < .05) differences between gender.

176

A post-hoc power analysis showed that assuming a medium effect size and a standard significance level of p = .05, power was sufficient with our sample size = .74. Assuming a large effect between mental health and coach–athlete relationship shows our sample has adequate power. Predicting Mental Health Three hierarchical multiple regressions had gender and personality inserted on the first step and coach–athlete relationship on the second step. The first multiple regression had depression as the dependent variable. In predicting depres sion, only the model including coach–athlete relationship as a predictor was significant, with a medium effect size, F(7,61) = 2.22, p = .045, R2 = .21. The combination of gender, personality, and coach–athlete relationship predicted 21% of the variance in depression scores. Adding coach–athlete relationship on the second step of the hierarchical regression accounted for an additional 15% of the variance in depression, ΔF(1, 60) = 11.37, p = .001, ΔR2 = .15. Regarding specific predictors, only coach–athlete relation ship (β = -.43, p = .001) was a significant predictor of depression, where a stronger coach–athlete relationship was associated with lower depression scores. Table 3 shows the values for all predictors. The second hierarchical regression, which had psychological quality of life as the dependent vari able, also had only the second model as significant, with a medium effect size, F(7,61) = 2.78, p = .014, R2 = .24. The model accounted for 24% of the variance in psychological quality of life, with the addition of coach–athlete relationship con tributing to half of this variance (ΔR 2 = .12), ΔF(1, 61) = 9.39, p = .003. Regarding specific predictors, again, coach–athlete relationship was the only significant predictor (β = -.38, p = .003), where a stronger coach–athlete relationship was associated with higher quality of life scores. Table 3 shows the values for all predictors. In the third hierarchical regression, anxiety was the dependent variable. Neither model was signifi cant in predicting anxiety. Gender and personality had a statistically nonsignificant contribution of 5% to the anxiety scores with a small effect size, F(6,60) = 0.54, p = .777, R2 = .05, and the model including coach–athlete relationship had a non significant contribution of 11%, ΔF(7,59) = 1.05, p = .407, ΔR2 = .11. We summarize the regression analyses in Table 3.

Powers, Fogaca, Gurung, and Jackman | Student-Athlete Mental Health

Discussion This study aimed to assess the associations between coach–athlete relationships and student athletes’ mental health. It also aimed to analyze if this association would be significant after accounting for the influence of student athletes’ gender and personality. Our results provide a detailed picture of the relationships between different measures of mental health, a range of personality factors, and most importantly, measures of the coach–athlete relationship. This section will discuss the findings regarding the relationship of each of these variables with different aspects of student athletes’ mental health. The relationship between coaches and athletes presents itself as a major candidate for focus in the context of mental health of athletes (Lentz et al., 2018). Coaches’ relationship with their athletes showed a clear association with psychological quality of life and depression and contributed to the prediction of 15% of depression and 12% of psychological quality of life scores in this sample, after accounting for the effects of gender and personality variables. These are important findings, as they reinforce the need for coaches to invest on the improvement of their relationships with their athletes. Previous research has shown that the coach–athlete relationship influences important aspects of athletes’ performance and development (Prophet, Singer, Martin, & Coulter, 2017). In addi tion, the International Sport Coaching Framework (2013) includes building relationships in its list of coaching competencies. It is one of the few times, however, that this variable is related to aspects of athletes’ mental health. Even though the link between coach–athlete relationship and depression in athletes has not been studied extensively in the past, it is not surprising that this relationship may exist. Lentz et al. (2018) suggested that the athletes’ relationships with their coaches could represent a source of support amid the many stressors that college student athletes endure. On most teams, the coach is the main point of contact between the athlete and the university, and on some campuses the coach may be the gatekeeper for all interactions with a university. A coach and his or her approval or disapproval could also play a major role in the athlete’s psyche. Implicitly or explicitly the athletes’ prospects, play ing time, and career, can hinge on the coach. It is not surprising that the relationship with the coach can play a major role in the athlete’s mental health. Further, a poor relationship with the coach could be a stressor in itself, not only negating support,

but also generating more stress on top of the many pressures that student athletes already undergo (Lentz et al., 2018). Previous research (i.e., Davis & Jowett, 2014; Felton & Jowett, 2013b; Felton & Jowett, 2017) has shown an association of athlete’s attachment style and coach support with well-being outcomes such as affect, vitality, and performance self-concept. In the present study, coach–athlete relationship was corre lated to psychological quality of life and depression, besides predicting them over and above the effects of personality and gender. These findings indicate that the coaches’ behaviors and their relationship with their athletes may have a connection with their TABLE 2 Correlations Between Major Variables, Controlled by Gender WHO

CESD-R

BAI

Extraversion

.06

.08

.04

Agreeableness

.14

-.60

.07

Conscientiousness

.22

.00

-.21

Emotional Stability

.47***

-.31*

-.29*

Openness

.12

-.11

.03

Commitment*

.38**

-.36**

-.14

Closeness

.37**

-.44***

-.26*

Complementarity

.36**

-.41**

-.15

Note. WHO = World Health Organization. CESD-R = Center for Epidemiological StudiesDepression Scale Revised. BAI = Beck Anxiety Inventory Manual. * p < .05. ** p < .01. *** p < .001.

TABLE 3 Summary of Multiple Regression Analyses for the Prediction of Depression, Anxiety, and Psychological Quality of Life Depression

Anxiety

Psychological QOL

SE B

Gender

-.79

1.26

-.08

.44

1.12

.39

-.96

1.07

-.11

Extraversion

-.09

.36

-.03

.08

.32

.03

.47

.31

.18

Agreeableness

.13

.29

.06

-.14

.26

-.08

.11

.25

.06

Conscient.

.40

.32

.15

.29

.07

.06

.27

.03

Emotional St.

.07

.43

.02

.23

.38

.08

.03

.35

.01

Openness

.12

.30

.05

.-08

.27

.04

-.07

.26

-.03

-.12

.03

-.43**

-.06

.03

-.27

.09

.03

.38**

1st Block

2nd Block C–A Rel. R2

.21

2.22

.11

.24

1.05

2.78*

Note. QOL = Quality of Life. C–A Rel. = Coach–athlete relationship. * p < .05. **p < .01. ***p < .001.

177

Student-Athlete Mental Health | Powers, Fogaca, Gurung, and Jackman

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

178

athletes’ mental health, beyond the effect of the athletes’ personality on their own mental health. However, further research with larger and more varied samples (i.e., from various universities and divisions) are needed to confirm the relationship of these variables. In addition, future studies with longitudinal or experimental designs could help identify if this relationship is causal or if there is a third confounding or mediating variable affecting this association. Beyond the coach–athlete relationship, our results shed light on other aspects of student athletes’ mental health. We found significant gender differences in quality of life scores and that emotional stability was significantly correlated to all measures of student-athlete mental health. In a meta-analysis that included 137 personality traits as correlates of subjective happiness and well-being, DeNeve and Cooper (1998) found that among the Big Five factors, emotional stability was the strongest predictor of happiness, and that extraversion and agreeableness were good predic tors of positive affect. However, emotional stability was not a significant predictor in the hierarchical multiple regressions. These findings indicate that it may be important to consider emotional stability’s role in student athletes’ mental health in research and practice, but further research is necessary to establish this connection with confidence. Our results are tempered by some important limitations. First, we acknowledge that different teams have their seasons start and end at different times of the year. Our data was collected at one point in the year and, although some teams were already playing, others would only be practicing. The stress and related mental health issues would correspondingly be different. Second, teams vary in size. Whereas a swimming team may have many members and even many assistant coaches, a vol leyball team will have a smaller number of players. Although we measured team membership, we did not factor team size into our analyses. Both these factors are especially important when we note that this is a correlational study and we cannot measure changes in athlete or coach behavior and result ing changes in mental health. Third, the sample is relatively homogeneous, being predominantly women, almost all European American, and all at one university. Finally, although a substantial portion of the student athletes volunteered for our study, the sample is under 50% of the athlete population on campus. It is possible the athletes in the sample varied in some way from those who did not take part.

Conclusion Although the present study only demonstrates initial evidence that coach–athlete relationship and some aspects of college student athletes’ mental health may be connected, it is important to consider this finding within the context of previous studies that have found relationship of coaches’ behaviors and support with various aspects of athletes’ psy chological well-being, such as positive and negative affect, vitality, satisfaction, and performance selfconcept (Davis & Jowett, 2014; Felton & Jowett, 2013b; Felton & Jowett, 2017; Reinboth, Duda, & Ntoumanis, 2004). In addition, this relationship makes theoretical sense, because coaches are key sources of support for college student athletes and relationship issues represent crucial stressors in college students’ lives (Lentz et al., 2018). The coach–athlete relationship had never been con nected to depression and psychological quality of life, to our knowledge. In the present study, the coach–athlete relationship was not only correlated to depression and life satisfaction, but could also predict part of the athletes’ scores on these scales. More studies are necessary to confirm this con nection, but in light of the evidence that coaches’ behaviors are connected to various well-being outcomes, coach education that targets specific behaviors that could support student athletes’ psychological well-being and mental health should become a central part of coaching education. Although there are great examples of suggestions on how to improve coach–athlete relationships to improve performance (e.g., Ferrar et al., 2018; Prophet et al., 2017), the potential effect of this relationship on athletes’ mental health should become central in these trainings as well.

References Armstrong, S., & Oomen-Early, J. (2009). Social connectedness, self-esteem, and depression symptomatology among collegiate athletes versus nonathletes. Journal of American College Health, 57, 521–526. https://doi.org/10.3200/jach.57.5.521-526 Beauchemin, J. (2014). College student-athlete wellness: An integrated outreach model. College Student Journal, 48, 268–280. Beck, A. T., & Steer, R.A. (1993). Beck Anxiety Inventory Manual. San Antonio, TX: Psychological Corporation. Binboga, E., Guven, S., Çatikkas, F., Bayazit, O., & Tok, S. (2012). Psychophysiological responses to competition and the Big Five personality traits. Journal of Human Kinetics, 33, 187–194. https://doi.org/10.2478/v10078-012-0057-x Brewer, B. W. (1993). Self-identity and specific vulnerability to depressed mood. Journal of Personality, 61, 343–364. https://doi.org/10.1111/j.1467-6494.1993.tb00284.x Cangur, S., Yaman, C., Ercan, I., Yaman, M., & Tok, S. (2017). The relationship of anthropometric measurements with psychological criteria in female athletes. Psychology, Health, & Medicine, 22, 325–331. https://doi.org/10.1080/13548506.2016.1234714 Chickering, A. W., McDowell, J., & Campagna, D. (1969). Institutional differences

Powers, Fogaca, Gurung, and Jackman | Student-Athlete Mental Health

and student development. Journal of Educational Psychology, 60, 315–326. https://doi.org/10.1037/h0027840 Choi, H., Cho, S., & Huh, J. (2013). The association between the perceived coach–athlete relationship and athletes’ basic psychological needs. Social Behavior and Personality: An International Journal, 41, 1547–1556. https://doi.org/10.2224/sbp.2013.41.9.1547 Chyi, T., Lu, F. J. H., Wang, E. T. W., Hsu, Y. W., Chang, K. H. (2018). Prediction of life stress on athletes’ burnout: The dual role of perceived stress. PeerJ, 6, e4213. https://doi.org/10.7717/peerj.4213 Davis, L., & Jowett, S. (2014). Coach–athlete attachment and the quality of the coach–athlete relationship: Implications for athlete’s well-being. Journal of Sports Sciences, 32, 1454–1464. https://doi.org/10.1080/02640414.2014.898183 Davorean, A. K., & Hwang, S. (2014). Depression and anxiety prevalence in student-athletes. In G. T. Brown (Ed.), Mind, body and sport: Understanding and supporting student-athlete mental wellness (pp. 38–39). Indianapolis, IN: National Collegiate Athletic Association (NCAA). DeNeve, K. M., & Cooper, H. (1998). The happy personality: A meta-analysis of 137 personality traits and subjective well-being. Psychological Bulletin, 124, 197–229. https://doi.org/10.1037/0033-2909.124.2.197 Eaton, W. W., Muntaner, C., Smith, C., Tien, A., & Ybarra, M. (2004). Center for Epidemiologic Studies Depression Scale: Review and revision (CESD and CESD-R). In Maruish M. E. (Ed.), The use of psychological testing for treatment planning and outcomes assessment (3rd ed.) (pp. 363–377). Mahwah, NJ: Lawrence Erlbaum. Etzel, E. (Ed.) (2009). Counseling and psychological services for college studentathletes. Morgantown, WV: FIT. Etzel, E., Watson, J., Visek, A., & Maniar, S. (2006). Understanding and promoting college student-athlete health: Essential issues for student affairs professionals. NASPA Journal, 43, 518–546. Felton, L., & Jowett, S. (2013a). The mediating role of social environmental factors in the associations between attachment styles and basic needs satisfaction. Journal of Sport Sciences, 31, 618–628. https://doi.org/10.1080/02640414.2012.744078 Felton, L., & Jowett, S. (2013b). ‘What do coaches do’ and ‘how do they relate’: Their effects on athletes’ psychological needs and functioning. Scandinavian Journal of Medicine and Sports Sciences, 23, 130–139. https://doi.org/10.1111/sms.12029 Felton, L., & Jowett, S. (2017). A self-determination theory perspective on attachment, need satisfaction, and well-being in a sample of athletes: A longitudinal study. Journal of Clinical Sport Psychology, 11, 304–323. https://doi.org/10.1123/jcsp.2016-0013 Ferrante, A. P., & Etzel, E. (2009). College student-athlete and counseling services in the new millennium. In E. Etzel (Ed.), Counseling and psychological services for college student-athletes (pp. 1–49). Morgantown, WV: FIT. Ferrar, P., Hosea, L., Menson, M., Dubina, N., Krueger, G., Staff, J., & Gilbert, W. (2018). Building high performing coach–athlete relationships: The USOC’s National Teach Coach Leadership Education Program (NTCLEP). International Sport Coaching Journal, 5, 60–70. https://doi.org/10.1123/iscj.2017-0102 Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A very brief measure of the Big Five personality domains. Journal of Research in Personality, 37, 504–528. https://doi.org/10.1016/S0092-6566(03)00046-1 Hagiwara, G., Iwatsuki, T., Isogai, H., Van Raalte, J. L., & Brewer, B. W. (2017). Relationships among sports helplessness, depression, and social support in American college student-athletes. Journal of Physical Education & Sport, 17, 753–757. Hammen, C. (2005). Stress and depression. Annual Review of Clinical Psychology, 1, 293–314. https://doi.org/10.1146/annurev.clinpsy.1.102803.143938 Hebard, S. P. (2015). A predictive model of coach–athlete attachment and emotion regulation on student-athlete aggression, alcohol use, and psychological distress (Unpublished doctoral dissertation). University of North Carolina – Greensboro, Greensboro, NC. Hefner, J., & Eisenberg, D. (2009). Social support and mental health among college students. American Journal of Orthopsychiatry, 79, 491–499. https://doi.org/10.1037/a0016918 International Council for Coaching Excellence. Association of Summer Olympic International Federations, & Leeds Metropolitan University. (2013). International sport coaching framework (version 1.2). Champaign, IL: Human Kinetics.

Jowett, S., & Ntoumanis, N. (2004). The coach–athlete relationship questionnaire (CART-Q): Development and initial validation. Scandinavian Journal of Medicine & Science in Sports, 14, 245–257. https://doi.org/10.1111/j.1600-0838.2003.00338.x Koh, K. T., Kokkonen, M., & Law, H. R. B. (2019). Coaches’ implementation strategies in providing social support to Singaporean university athletes: A case study. International Journal of Sports Science & Coaching, 14, 681–693. https://doi.org/10.1177/1747954119876099 Kotov, R., Gamez, W., Schimidt, F., & Watson, D. (2010). Linking ‘Big’ personality traits to anxiety, depressive, and substance use disorders: A meta-analysis. Psychological Bulletin, 136, 768–821. https://doi.org/10.1037/a0020327 Langan, E., Toner, J., Blake, C., & Lonsdale, C. (2015). Testing the effects of a self-determination theory-based intervention with youth Gaelic football coaches on athlete motivation and burnout. The Sport Psychologist, 29, 293–301. https://doi.org/10.1123/tsp.2013-0107 Lentz, B., Kerins, M. L., & Smith, J. (2018). Stress, mental health, and the coach– athlete relationship: A literature review. The Applied Research in Coaching and Athletics Annual, 33, 214–238. Lipson, S. K., Lattie, E. G., & Eisenberg, D. (2018). Increased rates of mental health service utilization by U.S. college students: 10-year population-level trends (2007–2017). Psychiatric Services, 70, 60–63. https://doi.org/10.1176/appi.ps.201800332 Lu, F. J. H., Lee, W. P., Chang, Y.-K., Chou, C.-C., Hsu, Y.-W., Lin, J.-H., & Gill, D. L. (2016). Interaction of athletes’ resilience and coaches’ social support on the stress-burnout relationship: A conjunctive moderation perspective. Psychology of Sport and Exercise, 22, 202–209. https://doi.org/10.1016/j.psychsport.2015.08.005 Merianos, A. L., Nabors, L. A., Vidourek, R. A., & King, K. A. (2013). The impact of self-esteem and social support on college students’ mental health. American Journal of Health Studies, 28, 27–34. Oexle, N., & Sheehan, L. (2020). Perceived social support and mental health after suicide loss. Crisis: The Journal of Crisis Intervention and Suicide Prevention, 41, 65–69. https://doi.org/10.1027/0227-5910/a000594 Pascarella, E., & Terenzini, P. (2005). How college affects students: A third decade of research. San Francisco, CA: Jossey-Bass. Prophet, T., Singer, J., Martin, I., & Coulter, T. J., (2017). Getting to know your athletes: Strengthening the coach–athlete dyad using an integrative personality framework. International Sport Coaching Journal, 4, 291–304. https://doi.org/10.1123/iscj.2017-0009 Raabe, J. & Zakrajsek, R. A. (2017). Coaches and teammates as social agents for collegiate athletes’ basic psychological need satisfaction. Journal of Intercollegiate Sport, 10, 67–82. https://doi.org/10.1123/jis.2016-0033 Rao, A. L., Asif, I. M., Drezner, J. A., Toresdahl, B. G., & Harmon, K. G. (2015). Suicide in National Collegiate Athletic Association (NCAA) athletes: A 9-year analysis of the NCAA resolutions database. Sports Health, 7, 452–457. https://doi.org/10.1177/1941738115587675 Reinboth, M., Duda, J., & Ntoumanis, N. (2004). Dimensions of coaching behavior, need satisfaction, and the psychological and physical welfare of young athletes. Motivation and Emotion, 28, 297–313. https://doi.org/10.1023/b:moem.0000040156.81924.b8 Riley, A., & Smith, A. (2011). Perceived coach–athlete and peer relationships of young athletes and self-determined and motivation for sports. International Journal of Sport Psychology, 42, 115–133. Sarason, B. R., Sarason, I. G., & Gurung, R. A. R. (2001). Close personal relationships and health outcomes: A key to the role of social support. In B. R. Sarason & S. Duck (Eds.), Personal relationships: Implications for clinical and community psychology (p. 15–41). John Wiley & Sons Ltd. Schary, D. P. (2019). Servants in the weight room: Coaches using servant leadership to improve student-athlete well-being. Strength & Conditioning Journal, 41, 25–30. https://doi.org/10.1519/SSC.0000000000000347 Skevington, S. M., Lotfy, M., & O’Connell, K. A. (2004). The World Health Organization’s WHOQOL-BREF quality of life assessment: Psychometric properties and results of the international field trial. A report from the WHOQOL group. Quality of Life Research, 13, 299–310. https://doi.org/10.1023/b:qure.0000018486.91360.00 Storch, E. A., Storch, J. B., Killiany, E. M., & Roberti, J. W. (2005). Self-reported psychopathology in athletes: A comparison of intercollegiate studentathletes and non-athletes. Journal of Sport Behavior, 28, 86–98. The WHOQOL Group. (1998). Development of the World Health Organization WHOQOL-BREF quality of life assessment. Psychological Medicine, 28,

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

179

Student-Athlete Mental Health | Powers, Fogaca, Gurung, and Jackman

551–558. https://doi.org/10.1017/s0033291798006667 Van Dam, N. T., & Earleywine, M. (2011). Validation of the Center for Epidemiologic Studies Depression Scale-Revised (CESD-R): Pragmatic depression assessment in the general population. Psychiatry Research, 186, 128–132. https://doi.org/10.1016/j.psychres.2010.08.018 Velikic, D., Knezevic, J., & Rodic, N. (2014). Relations of some personality traits and characteristics of sportsmen with the level of sports anxiety. SportLogia, 10, 35–43. https://doi.org/10.5550/sgia.141001.en.005v Wenzel, L., Glanz, K., & Lerman, C. (2002). Stress, coping, and health behavior. In K. Glanz, B. K. Rimer, & F. M. Lewis (Eds.), Health Behavior and Health Education: Theory, Research, and Practice. San Francisco, CA: Jossey-Bass. Whitehead, P. M., & Senecal, G. (2019). Balance and mental health in NCAA Division I student-athletes: An existential humanistic view. The Humanistic Psychologist. Advance online publication. https://doi.org/10.1037/hum0000138 Wolanin, A., Gross, M., & Hong, E. (2015). Depression in athletes: Prevalence and risk factors. American College of Sports Medicine, 14, 56–60. https://doi.org/10.1249/JSR.0000000000000123 Wolanin, A., Hong, E., Marks, D., Panchoo, K., & Gross, M. (2016). Prevalence of clinically elevated depressive symptoms in college athletes and

differences by gender and sport. British Journal of Sports Medicine, 50, 167–171. https://doi.org/10.1136/bjsports-2015-095756 Wu, A. M. S., Lai, M. H. C., & Chan, I. T. (2014). Coaching behaviors, satisfaction of needs, and intrinsic motivation among Chinese university athletes. Journal of Applied Sport Psychology, 26, 334–348. https://doi.org/10.1080/10413200.2014.888107 Author Note. Megan Powers, https://orcid.org/0000-00031899-6291, University of Wisconsin, Green Bay; Jana Fogaca, https://orcid.org/0000-0002-2309-7541, University of Wisconsin, Green Bay; Regan A. R. Gurung, https://orcid.org/0000-0002-3542-4378, School of Psychological Sciences, Oregon State University; and Callan M. Jackman, https://orcid.org/0000-0003-3303-4293, School of Psychological Sciences, Oregon State University. Special thanks to Psi Chi Journal reviewers for their support. Correspondence concerning this article should be addressed to Callan Jackman, School of Psychological Science, Oregon State University, Corvallis, OR, 97331. E-mail: jackmanc@oregonstate.edu

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

180

https://doi.org/10.24839/2325-7342.JN25.2.181

Do Hugs and Their Constituent Components Reduce Self-Reported Anxiety, Stress, and Negative Affect? Preman Koshar and Megan L. Knowles* Franklin & Marshall College

ABSTRACT. Past research has suggested that touch and pressure can have antidepressant and anxiolytic properties. The present investigation hypothesized that brief interventions of hugging and its constituent components (pressure and the presence of a friend) during a stressful situation would reduce anxiety, social anxiety, stress, and negative affect while increasing social support, relative to control condition. Undergraduate participants (n = 155) completed the Trier Social Stress Test while either receiving hugs from a friend (hug), having a friend nearby (friend), wearing a weighted pressure vest (vest), or having nothing added (control). There was no significant effect of condition on state measures of anxiety (ηp2 = .01, p = .79), social anxiety (ηp2 = .01, p = .70), stress (ηp2 = .02, p = .58), negative affect (ηp2 = .01, p = .77), or social support (ηp2 = .03, p = .22). These findings suggest that brief interventions with hugs, weighted pressure vests, or the presence of a friend are not effective at increasing social support nor at reducing anxiety, social anxiety, stress, or negative affect. Alternative explanations for these results are discussed.

Open Data badge earned for transparent research practices. Data available at https://osf.io/g9z8k/

Keywords: hugging, weighted vests, anxiety, touch, pressure

ouch and, more broadly, pressure, have long been held to have medicinal properties, even though scientific research has only recently begun to put these beliefs to the test (Classen, 2012). Touch is defined here as “the tactile stimulation of one person by another,” whereas pressure is defined as tactile stimulation induced by any person or object (Mulaik et al., 1991, p. 308). Touch in particular has been underresearched, with no formal evaluation of its benefits for a variety of common psychological maladies, despite systematic reviews largely lauding its wide-ranging benefits across a variety of mediums (e.g., Field, 2010; Jakubiak & Feeney, 2017). Massage, for instance, has been found to reduce anxiety and cortisol levels (e.g., Field, 2010; Field et al., 1992). Massage and other forms of

*Faculty mentor

touch also improve sleep and decrease the produc tion of substance P, a chemical associated with pain production, in addition to increasing the release of oxytocin, serotonin, and dopamine in the brain (e.g., Field, 2010; Stock & Uvnäs‐Moberg, 1988). Oxytocin, serotonin, and dopamine increases have been linked to reductions in anxiety, stress, and depression (for a review, see Field, 2010; Field et al., 1992). According to the polyvagal theory, these neurochemical changes are likely due to vagus nerve stimulation from skin pressure (e.g., Field, 2010; Gamse, Lembeck, & Cuello, 1979; Porges, 2001; Stock & Uvnäs‐Moberg, 1988). These neurochemical changes may explain why touch decreases anxiety in cardiac patients (Weiss, 1990), as well as existential anxiety after a death reminder (Koole, Tjew A Sin, & Schneider, 2014).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

181

Hugs and Their Constituent Components | Koshar and Knowles

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

182

Touch is also often used as a demonstration of affection, and some forms of affection (though not always physical affection) have been associated with reduced cortisol levels and stress (e.g., Floyd et al., 2007; Floyd & Riforgiate, 2008). Touch and other forms of tactile stimulation have also been implemented in sensorimotor therapy, which is used to help traumatized individuals reconnect with their bodies and modulate their own arousal (e.g., Ogden & Minton, 2000; for an overview, see Van der Kolk, 2015). However, it is important to note that touch is not always beneficial—a wrist touch on highly socially anxious people was dem onstrated to increase anxiety (Wilhelm, Kochar, Roth, & Gross, 2001). Overall, this growing body of research indicates that touch can (across a wide range of modalities) improve a diverse set of variables related to mental health. The range of possible responses to touch, from reducing stress to increasing anxiety, suggests that these responses are, to some degree, dependent on context and individual perception (Field, 2010; Wilhelm et al., 2001). This is one of the many reasons why a well-researched stress test such as the Trier Social Stress Test (TSST) was selected for participants to undergo; that way the context of the interventions would be constant across conditions. Both reviews and original research have found that touch in close adult relationships increases oxytocin levels, and when that touch is paired with a positive interpretation of the touch, it leads to lowered stress and could cause cognitive changes that increase both relational and psychological well-being (e.g., Holt-Lunstad, Birmingham, & Light, 2008; for a systematic review, see Jakubiak & Feeney, 2017). Touch also stimulates the vagus nerve, which activates the pregenual anterior cingulate cortex (related to rewarding pleasant stimulations) even more strongly than pressure, suggesting that interpreting skin pressure as touch may enhance any neurochemical benefits (Jakubiak & Feeney, 2017; Lindgren et al., 2012). These benefits, although most pronounced in response to touch, have their roots in pressure. Pressure has been described as “relaxing” and “calming” ever since Temple Grandin described her “Squeeze Machine,” now termed the “Hug Machine” (Edelson, Edelson, Kerr, & Grandin, 1999; Grandin, 1992, pp. 63, 67). There have been several designs of this machine, but usually the individual using the machine lies on their stomach in it and pulls a lever that tightens cushioned pads

and boards all around the user’s body (Edelson et al., 1999; Grandin, 1992). Grandin stated that her Hug Machine “calm[ed] down [her] anxiety” (Grandin, 1992, p. 66). However, this has only ever been investigated in a single study evaluating 12 children diagnosed with autism, which found that the Hug Machine significantly reduced tension and marginally reduced anxiety (Edelson et al., 1999). It is worth noting that this study consisted of a small sample of children and focused entirely on nonnormative populations. One other study evaluated the effects of a similar contraption on college students, and although Krauss (1987) found no clear effect on state anxiety, subjective relaxation did increase. These conflicting results may be due to the unusual contraptions employed by Krauss and Grandin, the small samples used in each study, or the different qualities of the samples. The vast majority of the research, from Grandin onward, has been aimed at identifying the sensory benefits of pressure (particularly with small, nonnormative samples of children), despite inconsistent and rarely significant results (Losinski, Sanders, & Wiseman, 2016; Stephenson & Carter, 2009). The mixed results for pressure as an effective remedy for autism and other sensory disorders (as well as the promising preliminary results highlighting pressure’s anxiolytic proper ties) suggest that this focus on utilizing pressure to improve sensory and behavioral issues may be misguided. Sensory and behavioral improvements may be caused by nothing more than a reduction in anxiety, as there is some evidence that anxiety is related to sensory overresponsivity (Green & Ben-Sasson, 2010; Mazurek et al., 2013). Pressure might also disproportionately benefit individuals diagnosed with autism because their elevated anxiety (compared to a normative population) may be reduced by the pressure (White, Oswald, Ollendick, & Scahill, 2009). Previous research ers might have been measuring uninformative variables in their attempts to ascertain why pres sure—particularly weighted pressure vests—seemed to help some children diagnosed with autism. Research on weighted pressure vests has suf fered from the same issues as research on pressure as a whole, with the majority of studies on the vests only addressing behavioral or sensory issues in autism, attention deficit disorders, or perva sive development disorder (e.g., Kane, Luiselli, Dearborn, & Young, 2004; Lin, Lee, Chang, & Hong, 2014). Weighted pressure vests are defined

Koshar and Knowles | Hugs and Their Constituent Components

here as adjustable vests that can be tightened and are filled with small metal or sand-filled weights that in total weigh anywhere from 1 to 10 pounds. Most of these studies were focused on children and relied on very small sample sizes (e.g., Fertel-Daly, Bedell, & Hinojosa, 2001; Losinski, Cook, Hirsch, & Sanders, 2017). Systematic reviews have revealed that weighted pressure vests are not effective at treating autism, among other disabilities (e.g., Stephenson & Carter, 2009). Very little literature exists on the effect of weighted pressure vests on anxiety despite the vests being the focus of a substantial amount of research on nonnormative populations and animals. To date, this potential effect has been most clearly described in a handful of studies on dogs where it has been effective at reducing dogs’ fear of thunder (Cottam, Dodman, & Ha, 2013; Fish, Foster, Gruen, Sherman, & Dorman, 2017; King, Buffington, Smith, & Grandin, 2014). Weighted blankets, or blankets that have been filled with metal weights or a heavy material such as sand and often weigh 12 pounds or more, have more clear results than the weighted pressure vests (e.g., Champagne, Mullen, Dickson, & Krishnamurty, 2015). However, many of the studies have the same issues: small sample sizes, samples that only include children, and samples that consist only of nonnormative individuals (i.e., individuals diagnosed with autism, individuals with insomnia, or patients in a psychiatric inpatient facility; e.g., Champagne et al., 2015; Gee, Peterson, Buck, & Lloyd, 2016; Gringras et al., 2014). Although the results have been mixed, there is some agreement that weighted blankets reduce anxiety, at least in some specific populations and contexts such as inpatients at a psychiatric facility and individuals undergoing dental care (e.g., Chen, Yang, Chi, & Chen, 2013; Novak, Scanlan, McCaul, MacDonald, & Clarke, 2012). Weighted vests were chosen over weighted blankets for this study primarily due to the need for participants to be sitting up for the TSST; it is likely that a weighted blanket would have fallen off if not actively held in place for the dura tion. The potential anxiolytic effects of other types of pressure have yet to be confirmed empirically. One common, perhaps ubiquitous form of both pressure and touch is hugging (e.g. Fromme et al., 1989). Hugging has also been overlooked in the literature. One of the few hugging studies found that the number of daily perceived hugs correlated positively with increased social support,

which then buffered stress and improved resistance to infection (Cohen, Janicki-Deverts, Turner, & Doyle, 2015). This finding supports the hypothesis that hugging can reduce stress while increasing social support. Also, the frequency of hugs has been associated with lower blood pressure and higher oxytocin, suggesting that hugs have a wide array of positive benefits that are all linked to vagal stimula tion (Light, Grewen, & Amico, 2005; Porges, 2001; Stock & Uvnäs‐Moberg, 1988). However, most of the experimental studies on hugging actually assessed warm contact, which consists of positive social and physical interaction with a partner (usually a romantic partner) that often culminates in a 20-second hug. Such “warm contact” appears to increase oxytocin and self-reported happiness while reducing alpha amylase and blood pressure (Grewen, Girdler, Amico, & Light, 2005; HoltLunstad, Birmingham, & Light, 2008; Matsunaga et al., 2011). No study has analyzed whether hugging alone can reduce psychopathological symptomol ogy, and the effects of hugging on anxiety has yet to be the subject of published research. It is also unclear whether any benefits that might arise from hugging are due to the touch and pressure involved in the act, or if they are due simply to the presence of a supportive friend, as the presence of a friend alone can reduce some measures of stress such as cardiovascular reactivity, which is “the response of physiological parameters, often blood pressure or heart rate, to a laboratory stressor” (e.g., Christenfeld et al., 1997, p. 388; Grewen, Girdler, Amico, & Light, 2005). Cardiovascular reactivity is related to vagal stimulation, just like touch and pressure, implying that similar or connected processes may be at the root of these benefits (e.g., Huang, Webb, Zourdos, & Acevedo, 2013). In sum, pressure, touch, and the presence of a friend all show promise as symptom reduction strategies for anxiety, stress, and depression (e.g., Cohen et al., 2015; Field, 2010). However, much of the work on touch and pressure has substantial limitations, often featuring populations that are not generalizable or measuring inconsistent vari ables of interest. Anxiety, stress, and depression are all exceedingly common, both as various disorders and as subthreshold nuisances; these issues incur profound amounts of pain, suffering, economic costs, and death every year (DuPont et al., 1996; Kessler et al., 2005). New strategies for managing these disorders and the corresponding suffering these maladies cause could have wide-ranging benefits.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

183

Hugs and Their Constituent Components | Koshar and Knowles

Present Investigation In light of past research demonstrating that warm interpersonal contact, physical pressure, and the presence of friends produce positive outcomes, we tentatively expected hugs, weighted pressure vests, and friends to mitigate anxiety, stress, and depression to some degree. More specifically, we hypothesized that brief interventions of hug ging and its two primary components, pressure (wearing a weighted pressure vest), and social presence (being in the presence of a friend) throughout a stressful situation would reduce state anxiety, state social anxiety, state stress, and state negative affect while increasing state social support. Negative affect was included as an exploratory variable of interest despite the weaker evidence supporting its relationship to touch and pressure due to its close relationship to anxiety and stress. Participants completed the TSST (Kirschbaum, Pirke, & Hellhammer, 1993), which consisted of a speech task and a math task performed in front of a camera monitored live by a judge, in order to determine if any of the experimental conditions had a protective effect on participants’ otherwise elevated stress and anxiety levels. Elevating stress and anxiety also made the corresponding variables more easily measurable and any differences more easily detectable. Hugging, and its two primary components, pressure and social presence, were examined as strategies for reducing anxiety, stress, and negative affect for the first time in this study. These strategies could potentially be used as affordable supplements to common treatments such as cognitive-behavioral therapy or medication.

Method

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

184

Participants One hundred fifty-five (110 women, 43 men, 2 other; age: M = 19.55 years, SD = 1.24) undergradu ates from Franklin & Marshall College participated in return for either partial course credit or $5 from January to May 2018. After institutional review board approval from the Franklin & Marshall College IRB was obtained, participants signed up for the study online either through Sona (www. sona-systems.com) or via a publicly available Google Sheets spreadsheet. Participants were recruited from an introductory psychology class, via word-ofmouth (including paid recruiters who received $3 per pair of participants), Facebook posts to a subset of college class pages, and emails to a randomized sample of 400 students.

Prior to arrival at the lab, participants signed up for the study, entitled “Social Behavior and Emotions” in order to minimize demand charac teristics. It was made clear to all participants in the description of the study that they needed to sign up with a friend. This was necessary to ensure that all conditions had equivalent groups of participants; otherwise the conditions that do not require a friend to come along (control and vest) would have more participants who cannot bring a friend, while the friend-bringing conditions (hug and friend) would have more participants who can bring a friend, thus creating inequivalent groups with differing char acteristics. However, it would also be problematic for all participants to bring a friend, as the Friend condition is specifically trying to measure the effect of the presence of a friend, and this variable would be very difficult to differentiate if all participants brought a friend. It would also make it impossible to ascertain if any benefits in the vest condition were from the weighted pressure vest alone if a friend was also present. As such, after signing up, participants were randomly assigned without replacement to a condition based on when they signed up and their assumed gender, and were then either emailed a reminder to bring a friend to the study (hug or friend conditions) or an update informing them to come alone (control or vest conditions). Measures We first assessed general anxiety, social anxiety, depression, stress, and social support at the trait level to determine whether they moderated the impact of condition on our outcomes of interest. Immediately after participants completed the TSST, we assessed the same variables at the state level as our outcomes of interest. All state measures were modified from their original forms to reference the TSST. Demographic information. All participants provided their gender and their age in addition to completing the following scales. Generalized Anxiety Disorder Scale (GAD-7). We measured trait anxiety using the GAD-7, a 7-item self-report scale developed by Spitzer, Kroenke, Williams, and Löwe (2006). The GAD-7 measured trait anxiety on a 4-point Likert-type scale ranging from 0 (not at all) to 3 (nearly every day) and included items such as “Feeling nervous, anxious, or on edge.” This scale demonstrated adequate validity and reliability (α = .92) when taken by adults in a clinical setting (Spitzer et al., 2006). The scale was internally reliable using a student sample in the current study, α = .88.

Koshar and Knowles | Hugs and Their Constituent Components

Social Interaction Anxiety Scale (SIAS-6) and Social Phobia Scale (SPS-6). We measured trait social anxiety by combining the 6-item SIAS-6 and the 6-item SPS-6 scales, both developed by Peters, Sunderland, Andrews, Rapee, and Mattick (2012), into one self-report scale. The SIAS-6 and SPS-6 measured trait social anxiety on a 5-point Likerttype scale ranging from 0 (not at all characteristic or true of me) to 4 (extremely characteristic or true of me) and included items such as “I can feel conspicuous standing in a line.” The individual scales demon strated adequate validity and reliability (αs > .90) in clinical samples (Peters et al., 2012), and the composite scale demonstrated adequate internal reliability in the present study, α = .84 Patient Health Questionnaire (PHQ-2). We measured trait depression using the PHQ-2, a two-item self-report scale developed by Löwe, Kroenke, and Gräfe (2005). The PHQ-2 measured trait depression on a 4-point Likert-type scale rang ing from 0 (not at all) to 3 (nearly every day) and included items such as “Little interest or pleasure in doing things.” The scale demonstrated adequate validity and reliability (α = .83) in a sample of medical outpatients (Löwe et al., 2005). The scale demonstrated adequate internal reliability in the present study using the Spearman-Brown coefficient (α = .73) despite the limitations of two-item scales (Eisinga, Te Grotenhuis, & Pelzer, 2013). Perceived Stress Scale (PSS). We measured trait stress using the PSS, a 10-item self-report scale developed by Cohen, Kamarck, and Mermelstein (1983). The PSS measured trait stress on a 5-point Likert-type scale ranging from 0 (never) to 4 (very often) and included items such as “In the last month, how often have you been upset because of something that happened unexpectedly?” This scale has been found to be internally reliable in clinical and student populations, α = .84 to .86, as well as in the present study, α = .84 (Cohen et al., 1983). Multidimensional Scale of Perceived Social Support (MSPSS). We measured trait social sup port using the MSPSS, a 12-item self-report scale developed by Zimet, Dahlem, Zimet, and Farley (1988). The MSPSS measured trait social support on a 7-point Likert-type scale ranging from 1 (very strongly disagree) to 7 (very strongly agree) and included items such as “There is a special person who is around when I am in need.” This scale has been reliable in student samples (α = .84 to .92; Zimet et al., 1988). It was also internally reliable in the present study, α = .87.

Spielberger State-Trait Anxiety Inventory (STAI). We measured state anxiety using the shortform STAI, a 6-item self-report scale developed by Marteau and Bekker (1992). The STAI measured state anxiety on a 4-point Likert-type scale ranging from 1 (not at all) to 4 (very much) and included items such as “I feel calm.” This scale has been found to be internally reliable in a combined sample including students and pregnant women, α = .82, as well as in the present study, α = .81 (Marteau & Bekker, 1992). State Social Anxiety Questionnaire (SSAQ). We measured state social anxiety using the SSAQ, a 6-item self-report scale developed by Kashdan and Steger (2006). The SSAQ measured state social anxiety on a 5-point Likert-type scale ranging from 1 (not at all) to 5 (extremely) and included items such as “I worried about what the judges thought of me.” This scale originally had seven items, but one item (“I found it hard to interact with people”) was removed due to most participants not interacting with anyone but the experimenter. This scale has been found to be internally reliable with a student sample, α = .91, as well as in the present study, α = .93 (Kashdan & Steger, 2006). State Stress Scale. We measured state stress using the State Stress Scale, a single-item self-report scale based on the work of Park, Armeli, and Tennen (2004). The State Stress Scale consists of a single item, “How stressed did you feel during the speech and math tasks you just completed?”, with responses on a 5-point Likert-type scale ranging from 1 (not at all) to 5 (extremely). Positive Affect and Negative Affect Schedule (PANAS). We measured state negative affect using the 10-item negative affect subset of the PANAS, a self-report scale developed by Watson, Clark, and Tellegen (1988). The PANAS measured state negative affect on a 5-point Likert-type scale rang ing from 1 (not at all) to 5 (extremely) and including items such as “distressed.” This scale has been found to be internally reliable in student samples, α = .84 to .90, as well as in the present study, α = .89 (Watson et al., 1988). State Social Support Scale. We measured state social support using the State Social Support Scale, a single-item self-report scale based on the MSPSS (Zimet et al., 1988). It consisted of the item “To what extent did you feel emotionally supported during the speech and math tasks you just completed?” The state social support measure was evaluated on a 5-point Likert-type scale ranging from 1 (not at all) to 5 (extremely).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

185

Hugs and Their Constituent Components | Koshar and Knowles

Procedure The following procedure is depicted in Figure 1. Participants were greeted by a male experimenter in a white lab coat. All participants were then asked to read and sign an informed consent form. Control condition. After completing the consent process, participants in the control condi tion were then led into the lab room and asked to complete the trait measures on the computer using Qualtrics (www.qualtrics.com). Four participants completed the trait measures online, less than one week before arriving for the lab session. After completing the trait measures, the participants informed the experimenter that they were done, and the experimenter took them back to the wait ing area where they had arrived and instructed the participants to wait for 1 minute. This was the participants’ first 1-minute break. After the break, the experimenter led the participants back to the lab room. The experimenter then informed them that they would be giving a 5-minute speech to the camera in the lab room about why they would be a good candidate for their ideal job. The experi menter also informed participants that the camera was monitored live by three judges, including the experimenter; in actuality it was only monitored by the experimenter. Participants then began the TSST, which consisted of a 3-minute preparation period, a 5-minute speech task, and a 5-minute math task (Kirschbaum, 2015). Participants were first given 3 minutes to prepare for their speech task before performing the 5-minute speech. During the speech task, if the participants stopped speaking for an extended period of time, the experimenter would tell the participants

via intercom to continue, as there was still time remaining. After the speech was completed, the second 1-minute break took place, with participants remaining in the lab room. Then, participants were instructed to serially subtract 17 from 2023 (one of the more difficult math tasks proposed for the TSST) for another 5 minutes and were told to restart if they made a mistake (Kudielka, Hellhammer, Kirschbaum, Harmon-Jones, & Winkielman, 2007). Afterward, the experimenter informed the participants that they now had a third and final 1-minute break; the participants again remained in the lab room. After the break, the participants completed the state measures on the computer using Qualtrics (www.qualtrics.com). Vest condition. The procedure for the vest condition was nearly identical to the above pro cedure for the control condition, except during the first 1-minute break the experimenter helped participants put on the weighted pressure vest. Four weighted pressure vests (Hyper Wear Inc., Austin, TX) weighing 10 pounds each were used, one in each of the four commonly available sizes (small, medium, large, and extra-large). Each par ticipant selected the size that best fit. Participants in the vest condition wore the vest throughout the experiment until the final 1-minute break, when the experimenter helped participants remove the vest. Hug condition. If the participant was in the hug condition, then the procedure was also similar to that of the control condition but with a few major changes. One such change is that, at the beginning of the study, the participant and friend were told that, during the study, they would not be

FIGURE 1 State of Procedure Control

Vest

Hug

Friend

End of Study Wait for 1 min

Wait for 1 min

Put on Vest, Wait for Remainder of 1 min

Continue Wearing Vest, Wait for 1 min

Remove Vest, Wait for Remainder of 1 min

Recieved Hug for 15 s, Wait With Friend for Remainder of 1 min

Wait With Friend for 1 min

Speech Portion of TSST

Recieve Hug for 15 s, Wait Alone for Remainder of 1 min

Math Portion of TSST

Be With Friend for 15 s, Wait Alone for Remainder of 1 min

Recieve Hug for 15 s, Wait Alone for Remainder of 1 min Be With Friend for 15 s, Wait Alone for Remainder of 1 min

Figure 1. A flow chart demonstrating the procedural differences between the four conditions in the present study. TSST = Trier Social Stress Test.

186

Pottest Measures

Koshar and Knowles | Hugs and Their Constituent Components

allowed to talk to or touch each other without per mission from the experimenter. This was to ensure that communication and touch that was not a part of the protocol—both of which could generate social support or reduce stress and anxiety—were relatively constant across participants and condi tions. They were also informed that the participant would be doing tasks in the lab room while the friend primarily remained in the waiting area. The procedure was then identical to the control condition until the first 1-minute break, when the participant and friend were instructed to hug each other, using both hands and arms for 15 seconds. Then, both the participant and the friend waited together for the rest of the break. The procedure was then identical to the control condition until the second break, when the friend was asked to follow the experimenter to the lab room and hug their friend in the same way as before for 15 seconds. After the hug was over, the experimenter asked the friend to return to the waiting area. The participant and friend then waited out the rest of the second break alone. The procedure was then identical to the control condition until the third and final break, when the same 15-second hug was repeated once more before the friend was asked to return to the waiting area. Friend condition. The friend condition was nearly identical to the hug condition. However, instead of receiving a hug during the first break, the participant and friend just waited together for 1 minute in the waiting area. The gender of the friend in both the hug and the friend conditions was not recorded, although the experimenter anecdotally observed that the majority of partici pants brought a same-sex friend. During the two other minute-long breaks, the friend entered the second room and stood near the participant for the same 15-second interval that the hug took place in the hug condition, but the friend did not touch or talk to them before being asked to leave after 15 seconds. All participants finished the protocol by com pleting the state measures.

with vest removal (n = 2), or the participant being an acquaintance or friend of the experimenter (n = 9).1 In preliminary analyses involving gender, the one participant with nonbinary gender was also excluded, but this participant was included for analyses that did not involve gender. First, a two-way Multivariate Analysis of Variance with gender and condition was run on all trait variables (anxiety, social anxiety, depres sion, stress, and social support) in order to verify that random assignment was successful. There were no significant main or interaction effects. To confirm that there were no effects for any individual variables, additional two-way Analyses of Variance (ANOVAs) with gender and condition were run on each of the trait variables. There were no significant effects involving condition. A series of one-way ANOVAs were then run to examine whether significant differences in our outcome variables emerged as a function of condition. Inconsistent with our hypotheses, these analyses revealed no significant differences between conditions in participants’ state anxiety, F(3, 136) = 0.36, p = .79, ηp2 = .01, state social anxiety, F(3, 136) = 0.48, p = .69, η p2 = .01, state stress, F(3, 135) = 0.66, p = .58, ηp2 = .02, state negative affect, F(3, 136) = 0.37, p = .77, ηp2 = .01, or state social support, F(3, 136) = 1.48, p = .22, ηp2 = .03. Descriptive statistics for state variables across conditions (and the whole sample) are viewable in Table 1. No other state or trait measures or analyses had any results of note.2 TABLE 1 Mean State Variable Scores Across Conditions Control (n = 37)

Hug (n = 36)

Vest (n = 34)

Friend (n = 33)

Full Sample (n = 140)

Anxiety

2.68(0.58)

2.69(0.73)

2.72(0.56)

2.83(0.75)

2.73(0.65)

Social Anxiety

2.60(1.20)

2.72(1.32)

2.49(1.03)

2.81(1.24)

2.65(1.20)

Stressa

2.97(1.14)

3.17(1.08)

3.09(1.11)

3.34(1.13)

3.14(1.11)

Negative Affect

2.13(0.77)

2.20(0.86)

2.14(0.73)

2.32(0.98)

2.20(0.83)

Social Support

1.49(0.08)

1.92(1.08)

1.74(0.93)

1.58(0.90)

1.68(0.94)

Note. Standard deviations are provided in parentheses. a One participant in the friend condition did not complete this scale and was not included in analyses involving state stress. This participant was included in all other analyses.

Results All of the scales were reduced by reverse-scoring appropriate items and averaging the scores. Fifteen participants were dropped from analyses for a variety of reasons such as failing to follow protocol (n = 3), knowing revealing information about the study beforehand (n = 1), experimenter error

Variable

Additional analyses using all 155 participants yielded results which were largely the same as those presented here. These additional tests are available upon request to the first author. 2 Exploratory ANOVAs were also run with gender and each of the state and trait measures (anxiety, social anxiety, stress, depression, and social support). No significant effects involving condition were found. These analyses are also available upon request to the first author. 1

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

187

Hugs and Their Constituent Components | Koshar and Knowles

Discussion

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

188

Contrary to predictions, our brief interventions utilizing hugs, weighted vests, and the presence of a friend were all ineffective at reducing state anxiety, state social anxiety, state stress, and state negative affect during a stressful situation, relative to a control condition. All interventions also failed to significantly increase state social support, relative to a control condition. Overall, these results do not appear, at least at first glance, to align with previous research suggesting that all of these conditions can produce fast-acting beneficial effects, such as rapid anxiolytic and supportive responses to massage and warm contact, respectively (Field, 2010; Light et al., 2005). However, further examination suggests that these results may nonetheless make sense when interpreted through an established theoretical framework. There are a number of possible explanations for our null results. The most likely of these explanations is that there are very little to no actual benefits conferred from being hugged, wearing a weighted pressure vest, or being in the presence of a friend. However, this explanation is not aligned with the limited amount of previous research sug gesting that similar brief interventions have at least some measurable benefits (Field, 2010; Light et al., 2005; Novak et al., 2012). The research indicating the importance of context and interpretation is also inconsistent with this explanation (Jakubiak & Feeney, 2017). Another possible explanation for our results is that the beneficial effects of touch, pressure, and the presence of a friend are heavily moderated by the perception or interpretation of the touch, pressure, or presence of a friend (Jakubiak & Feeney, 2017). This idea is robustly supported in a number of studies on touch and pressure, which suggest that many of its benefits are due to the release of oxytocin via vagal stimulation (Field, 2010; Jakubiak & Feeney, 2017). Oxytocin has been clearly linked to perceived social support, as well as dopaminergic and serotoninergic pathways, which could be why massage has antidepressant and anxiolytic properties (Baskerville & Douglas, 2010; Field, 2010; Marazziti et al., 2012). If oxytocin is indeed at the root of all of the benefits related to touch, pressure, and social support, then it would make sense that these benefits would only present themselves in positive contexts because oxytocin is a context-dependent hormone (e.g., Jakubiak & Feeney, 2017). As summarized in Jakubiak and Feeney’s model of affective touch (2017), most

of the benefits of oxytocin are not realized unless they are paired with a positive interpretation of the oxytocin-releasing stimulus—and no such positive interpretation was provided during this study. Negative contexts and interpretations may also be why previous research on weighted pressure vests was so mixed and largely unclear; children and nonnormative populations may be more likely to interpret new situations and sensations negatively (Fertel-Daly et al., 2001; Losinski et al., 2017). It is also possible that these interventions are effective, but only when implemented for long periods of time (e.g., wearing the vest for a longer period of time), or when used repeatedly (e.g., wearing the vest for several short intervals a day for several weeks). Long-term intervention strategies (e.g., massaging a participant daily for several weeks) have been used effectively in previous research (for a systematic review, see Field, 2010). This explanation, however, conflicts with several other studies that demonstrate relatively rapid changes in oxytocin and its accordant benefits (e.g., Jakubiak & Feeney, 2017; Light et al., 2005). The presence of a friend condition might have also failed to reach significance as it only measured the effect of the presence of a friend when behavior between friends was restricted (i.e., they were not allowed to touch or talk to one another). It is pos sible that some benefits could have been gained if social behavior had not been restricted. Another likely explanation for the null effect is that oxytocin was released in the presence of a friend, but that it was moderated heavily by the confusing context and possibly negative interpretation of a psychological experiment (Jakubiak & Feeney, 2017). Of all possible explanations, this study appears to align most closely with Jakubiak and Feeney’s aforementioned model of affectionate touch (2017). Our findings were consistent with that model and suggest that the effects from the basic actions of wearing a weighted vest, being hugged by a friend, or being in the presence of a friend have very little psychological value when they are not interpreted positively. Limitations The study, however, is not without limitations. We were unable to recruit sufficient numbers of participants to achieve the desired power of .80. Assuming a medium effect size, our study had .73 power, according to a G*Power analysis. This study also had a surprisingly young sample (even for college students), with a mean age of 19.55 years

Koshar and Knowles | Hugs and Their Constituent Components

(SD = 1.24). Given the limited age range and that all participants attended the same undergraduate institution, it is possible that this sample was not representative of college students. The gender of the friend in both the hug and friend condition was also not accounted for, and it is likely that the gender of the friend is a significant moderator because the gender of both the hugger and individual receiving the hug has a major role in touch behaviors and touch responses during a hug (Stier & Hall, 1984). This may be due to concerns about the possibility of intimacy, as well as social norms (Stier & Hall, 1984). The study might also have lacked ecological validity; that is, the forced, clinical nature of the experiment might have made a lab-based hug irreconcilable with a real, genuine hug borne from prosocial impulses (Sbordone, 1996). This may well be the case, and, if it is indeed what occurred, serves to further support the model of affectionate touch, in that the context and interpretation of the hug is what makes a hug genuine and beneficial, as opposed to the action itself (Jakubiak & Feeney, 2017). Demand characteristics might also have con tributed to the results (Nichols & Maner, 2008). It is thoroughly possible that participants believed that the hug (or any of the other conditions) was meant to be beneficial (or perhaps harmful) and therefore were biased to report more positive (or negative) results. However, this seems somewhat unlikely given the null results across all conditions, and no indications of negative bias from participants, but it is a potential factor that should be considered. Future Research Future research should focus on how context, meaning, and interpretation impact both social interactions and tactile stimulation. It would be interesting to see what the benefits of a naturally occurring hug in a positive context are, without the artificial and often negative sentiment of a directed hug. And although research on naturally occurring hugs is needed, future studies should also examine directed hugs because their effects are both more easily measurable and more easily administrable as an adjunct to therapy. Further studies examining the context of touch and pressure, particularly in the form of hugs, should be performed, in an effort to truly understand the benefits that may be gained. Additional work is needed to see if these null results were real and replicable, or if they were spurious.

Conclusion In sum, brief interventions with hugs, weighted pressure vests, or the presence of a friend did not reduce anxiety, stress, or negative affect. They also did not significantly increase social support. As such, it appears that these brief interventions are not viable as strategies for reducing anxiety, stress, or negative affect. However, given previous research, it is possible that the effects of hugs, weighted pressure vests, and the presence of a friend were attenuated by the lack of positive con text or interpretation during the stress test. Further research on the medicinal properties of touch and pressure must be done, focusing in particular on how context and interpretation moderate the effects of oxytocin released in response to tactile stimulation.

References Baskerville, T. A., & Douglas, A. J. (2010). Dopamine and oxytocin interactions underlying behaviors: Potential contributions to behavioral disorders. CNS Neuroscience & Therapeutics, 16, e92–e123. https://doi.org/10.1111/j.1755-5949.2010.00154.x Champagne, T., Mullen, B., Dickson, D., & Krishnamurty, S. (2015). Evaluating the safety and effectiveness of the weighted blanket with adults during an inpatient mental health hospitalization. Occupational Therapy in Mental Health, 31, 211–233. https://doi.org/10.1080/0164212X.2015.1066220 Chen, H. Y., Yang, H., Chi, H. J., & Chen, H. M. (2013). Physiological effects of deep touch pressure on anxiety alleviation: The weighted blanket approach. Journal of Medical and Biological Engineering, 33, 463–470. https://doi.org/10.5405/jmbe.1043 Christenfeld, N., Gerin, W., Linden, W., Sanders, M., Mathur, J., Deich, J. D., & Pickering, T. G. (1997). Social support effects on cardiovascular reactivity: Is a stranger as effective as a friend? Psychosomatic Medicine, 59, 388–398. https://doi.org/10.1097/00006842-199707000-00009 Classen, C. (2012). The deepest sense: A cultural history of touch. https://doi.org/10.5406/illinois/9780252034930.001.0001 Cohen, S., Janicki-Deverts, D., Turner, R. B., & Doyle, W. J. (2015). Does hugging provide stress-buffering social support? A study of susceptibility to upper respiratory infection and illness. Psychological Science, 26, 135–147. https://doi.org/10.1177/0956797614559284 Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396. https://doi.org/10.2307/2136404 Cottam, N., Dodman, N. H., & Ha, J. C. (2013). The effectiveness of the Anxiety Wrap in the treatment of canine thunderstorm phobia: An open-label trial. Journal of Veterinary Behavior: Clinical Applications and Research, 8, 154–161. https://doi.org/10.1016/j.jveb.2012.09.001 DuPont, R. L., Rice, D. P., Miller, L. S., Shiraki, S. S., Rowland, C. R., & Harwood, H. J. (1996). Economic costs of anxiety disorders. Anxiety, 2, 167–172. https://doi.org/10.1002/(SICI)1522-7154(1996)2:4<167::AID-ANXI2>3.0.CO;2-L Edelson, S. M., Edelson, M. G., Kerr, D. C. R., & Grandin, T. (1999). Behavioral and physiological effects of deep pressure on children with autism: A pilot study evaluating the efficacy of Grandin’s Hug Machine. American Journal of Occupational Therapy, 53, 145–152. https://doi.org/10.5014/ajot.53.2.145 Eisinga, R., Te Grotenhuis, M., & Pelzer, B. (2013). The reliability of a two-item scale: Pearson, Cronbach, or Spearman-Brown? International Journal of Public Health, 58, 637–642. https://doi.org/10.1007/s00038-012-0416-3 Fertel-Daly, D., Bedell, G., & Hinojosa, J. (2001). Effects of a weighted vest on attention to task and self-stimulatory behaviors in preschoolers with pervasive developmental disorders. American Journal of Occupational Therapy, 55, 629–640. https://doi.org/10.5014/ajot.55.6.629 Field, T. (2010). Touch for socioemotional and physical well-being: A review. Developmental Review, 30, 367–383. https://doi.org/10.1016/j.dr.2011.01.001

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

189

Hugs and Their Constituent Components | Koshar and Knowles

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

190

Field, T., Morrow, C., Valdeon, C., Larson, S., Kuhn, C., & Schanberg, S. (1992). Massage reduces anxiety in child and adolescent psychiatric patients. Journal of the American Academy of Child & Adolescent Psychiatry, 31, 125–131. https://doi.org/10.1097/00004583-199201000-00019 Fish, R. E., Foster, M. L., Gruen, M. E., Sherman, B. L., & Dorman, D. C. (2017). Effect of wearing a telemetry jacket on behavioral and physiologic parameters of dogs in the open-field test. Journal of the American Association for Laboratory Animal Science, 56, 382–389. Floyd, K., Mikkelson, A. C., Tafoya, M. A., Farinelli, L., La Valley, A. G., Judd, J., . . . Wilson, J. (2007). Human affection exchange: XIV. Relational affection predicts resting heart rate and free cortisol secretion during acute stress. Behavioral Medicine, 32, 151–156. https://doi.org/10.3200/bmed.32.4.151-156 Floyd, K., & Riforgiate, S. (2008). Affectionate communication received from spouses predicts stress hormone levels in healthy adults. Communication Monographs, 75, 351–368. https://doi.org/10.1080/03637750802512371 Fromme, D. K., Jaynes, W. E., Taylor, D. K., Hanold, E. G., Daniell, J., Rountree, J. R., & Fromme, M. L. (1989). Nonverbal behavior and attitudes toward touch. Journal of Nonverbal Behavior, 13, 3–14. https://doi.org/10.1007/bf01006469 Gamse, R., Lembeck, F., & Cuello, A. C. (1979). Substance P in the vagus nerve. Naunyn-Schmiedeberg’s Archives of Pharmacology, 306, 37–44. https://doi.org/10.1007/bf00515591 Gee, B. M., Peterson, T. G., Buck, A., & Lloyd, K. (2016). Improving sleep quality using weighted blankets among young children with an autism spectrum disorder. International Journal of Therapy and Rehabilitation, 23, 173–181. https://doi.org/10.12968/ijtr.2016.23.4.173 Grandin, T. (1992). Calming effects of deep touch pressure in patients with autistic disorder, college students, and animals. Journal of Child and Adolescent Psychopharmacology, 2, 63–72. https://doi.org/10.1089/cap.1992.2.63 Green, S. A., & Ben-Sasson, A. (2010). Anxiety disorders and sensory overresponsivity in children with autism spectrum disorders: Is there a causal relationship? Journal of Autism and Developmental Disorders, 40, 1495–1504. https://doi.org/10.1007/s10803-010-1007-x Grewen, K. M., Girdler, S. S., Amico, J., & Light, K. C. (2005). Effects of partner support on resting oxytocin, cortisol, norepinephrine, and blood pressure before and after warm partner contact. Psychosomatic Medicine, 67, 531–538. https://doi.org/10.1097/01.psy.0000170341.88395.47 Gringras, P., Green, D., Wright, B., Rush, C., Sparrowhawk, M., Pratt, K., . . . Wiggs, L. (2014). Weighted blankets and sleep in autistic children—A randomized controlled trial. Pediatrics, 134, 298–306. https://doi.org/10.1542/peds.2013-4285 Holt-Lunstad, J., Birmingham, W. A., & Light, K. C. (2008). Influence of a ‘warm touch’ support enhancement intervention among married couples on ambulatory blood pressure, oxytocin, alpha amylase, and cortisol. Psychosomatic Medicine, 70, 976–985. https://doi.org/10.1097/psy.0b013e318187aef7 Huang, C. J., Webb, H. E., Zourdos, M. C., & Acevedo, E. O. (2013). Cardiovascular reactivity, stress, and physical activity. Frontiers in Physiology, 4, 314. https://doi.org/10.3389/fphys.2013.00314 Jakubiak, B. K., & Feeney, B. C. (2017). Affectionate touch to promote relational, psychological, and physical well-being in adulthood: A theoretical model and review of the research. Personality and Social Psychology Review, 21, 228–252. https://doi.org/10.1177/1088868316650307 Kane, A., Luiselli, J. K., Dearborn, S., & Young, N. (2004). Wearing a weighted vest as intervention for children with autism/pervasive developmental disorder. Scientific Review of Mental Health Practice, 3, 19–24. Kashdan, T. B., & Steger, M. F. (2006). Expanding the topography of social anxiety: An experience-sampling assessment of positive emotions, positive events, and emotion suppression. Psychological Science, 17, 120–128. https://doi.org/10.1111/j.1467-9280.2006.01674.x Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K. R., & Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62, 593–602. https://doi.org/10.1001/archpsyc.62.6.593 King, C., Buffington, L., Smith, T. J., & Grandin, T. (2014). The effect of a pressure wrap (ThunderShirt®) on heart rate and behavior in canines diagnosed with anxiety disorder. Journal of Veterinary Behavior: Clinical Applications and Research, 9, 215–221. https://doi.org/10.1016/j.jveb.2014.06.007 Kirschbaum, C. (2015). Trier Social Stress Test. In I. P. Stolerman & L. H Price (Eds.) Encyclopedia of Psychopharmacology (pp. 1755–1758).

https://doi.org/10.1007/978-3-642-36172-2_53 Kirschbaum, C., Pirke, K. M., & Hellhammer, D. H. (1993). The ‘Trier Social Stress Test’–a tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology, 28, 76–81. https://doi.org/10.1159/000119004 Koole, S. L., Tjew A Sin, M., & Schneider, I. K. (2014). Embodied terror management: Interpersonal touch alleviates existential concerns among individuals with low self-esteem. Psychological Science, 25, 30–37. https://doi.org/10.1177/0956797613483478 Krauss, K. E. (1987). The effects of deep pressure touch on anxiety. American Journal of Occupational Therapy, 41, 366–373. https://doi.org/10.5014/ajot.41.6.366 Kudielka, B. M., Hellhammer, D. H., Kirschbaum, C., Harmon-Jones, E., & Winkielman, P. (2007). Ten years of research with the Trier Social Stress Test—revisited. In E. Harmon-Jones & P. Winkielman (Eds.), Social neuroscience: Integrating biological and psychological explanations of social behavior (pp. 56–83). New York, NY: Guilford. Light, K. C., Grewen, K. M., & Amico, J. A. (2005). More frequent partner hugs and higher oxytocin levels are linked to lower blood pressure and heart rate in premenopausal women. Biological Psychology, 69, 5–21. https://doi.org/10.1016/j.biopsycho.2004.11.002 Lin, H. Y., Lee, P., Chang, W. D., & Hong, F. Y. (2014). Effects of weighted vests on attention, impulse control, and on-task behavior in children with attention deficit hyperactivity disorder. American Journal of Occupational Therapy, 68, 149–158. https://doi.org/10.5014/ajot.2014.009365 Lindgren, L., Westling, G., Brulin, C., Lehtipalo, S., Andersson, M., & Nyberg, L. (2012). Pleasant human touch is represented in pregenual anterior cingulate cortex. Neuroimage, 59, 3427–3432. https://doi.org/10.1016/j.neuroimage.2011.11.013 Losinski, M., Cook, K., Hirsch, S., & Sanders, S. (2017). The effects of deep pressure therapies and antecedent exercise on stereotypical behaviors of students with autism spectrum disorders. Behavioral Disorders, 42, 196–208. https://doi.org/10.1177/0198742917715873 Losinski, M., Sanders, S. A., & Wiseman, N. M. (2016). Examining the use of deep touch pressure to improve the educational performance of students with disabilities: A meta-analysis. Research and Practice for Persons with Severe Disabilities, 41, 3–18. https://doi.org/10.1177/1540796915624889 Löwe, B., Kroenke, K., & Gräfe, K. (2005). Detecting and monitoring depression with a two-item questionnaire (PHQ-2). Journal of Psychosomatic Research, 58, 163–171. https://doi.org/10.1016/j.jpsychores.2004.09.006 Marazziti, D., Baroni, S., Giannaccini, G., Betti, L., Massimetti, G., Carmassi, C., & Catena-Dell’Osso, M. (2012). A link between oxytocin and serotonin in humans: Supporting evidence from peripheral markers. European Neuropsychopharmacology, 22, 578–583. https://doi.org/10.1016/j.euroneuro.2011.12.010 Marteau, T. M., & Bekker, H. (1992). The development of a six-item shortform of the state scale of the Spielberger State—Trait Anxiety Inventory (STAI). British Journal of Clinical Psychology, 31, 301–306. https://doi.org/10.1111/j.2044-8260.1992.tb00997.x Matsunaga, M., Isowa, T., Yamakawa, K., Tsuboi, H., Kawanishi, Y., Kaneko, H., . . . Ohira, H. (2011). Association between perceived happiness levels and peripheral circulating pro-inflammatory cytokine levels in middle-aged adults in Japan. Neuroendocrinology Letters, 32, 458–463. Mazurek, M. O., Vasa, R. A., Kalb, L. G., Kanne, S. M., Rosenberg, D., Keefer, A., . . . Lowery, L. A. (2013). Anxiety, sensory over-responsivity, and gastrointestinal problems in children with autism spectrum disorders. Journal of Abnormal Child Psychology, 41, 165–176. https://doi.org/10.1007/s10802-012-9668-x Mulaik, J. S., Megenity, J. S., Cannon, R. B., Chance, K. S., Cannella, K. S., Garland, L. M., . . . Massey, J. A. (1991). Patients’ perceptions of nurses’ use of touch. Western Journal of Nursing Research, 13, 306–323. https://doi.org/10.1177/019394599101300302 Nichols, A. L., & Maner, J. K. (2008). The good-subject effect: Investigating participant demand characteristics. The Journal of General Psychology, 135, 151–166. https://doi.org/10.3200/genp.135.2.151-166 Novak, T., Scanlan, J., McCaul, D., MacDonald, N., & Clarke, T. (2012). Pilot study of a sensory room in an acute inpatient psychiatric unit. Australasian Psychiatry, 20, 401–406. https://doi.org/10.1177/1039856212459585 Ogden, P., & Minton, K. (2000). Sensorimotor psychotherapy: One method for processing traumatic memory. Traumatology, 6, 149–173. https://doi.org/10.1177/153476560000600302 Park, C. L., Armeli, S., & Tennen, H. (2004). The daily stress and coping process

Koshar and Knowles | Hugs and Their Constituent Components

and alcohol use among college students. Journal of Studies on Alcohol, 65, 126–135. https://doi.org/10.15288/jsa.2004.65.126 Peters, L., Sunderland, M., Andrews, G., Rapee, R. M., & Mattick, R. P. (2012). Development of a short form Social Interaction Anxiety (SIAS) and Social Phobia Scale (SPS) using nonparametric item response theory: The SIAS-6 and the SPS-6. Psychological Assessment, 24, 66–76. https://doi.org/10.1037/a0024544 Porges, S. W. (2001). The polyvagal theory: Phylogenetic substrates of a social nervous system. International Journal of Psychophysiology, 42, 123–146. https://doi.org/10.1016/s0167-8760(01)00162-3 Sbordone, R. J. (1996). Ecological validity: Some critical issues for the neuropsychologist. In R. J. Sbordone & C. J. Long (Eds.), Ecological validity of neuropsychological testing (pp. 15–41). Delray Beach, FL, England: Gr Press/St Lucie Press. Spitzer, R. L., Kroenke, K., Williams, J. B., & Löwe, B. (2006). A brief measure for assessing generalized anxiety disorder: The GAD-7. Archives of Internal Medicine, 166, 1092–1097. https://doi.org/10.1001/archinte.166.10.1092 Stephenson, J., & Carter, M. (2009). The use of weighted vests with children with autism spectrum disorders and other disabilities. Journal of Autism and Developmental Disorders, 39, 105. https://doi.org/10.1007/s10803-008-0605-3 Stier, D. S., & Hall, J. A. (1984). Gender differences in touch: An empirical and theoretical review. Journal of Personality and Social Psychology, 47, 440–459. https://doi.org/10.1037//0022-3514.47.2.440 Stock, S., & Uvnäs-Moberg, K. (1988). Increased plasma levels of oxytocin in response to afferent electrical stimulation of the sciatic and vagal nerves and in response to touch and pinch in anaesthetized rats. Acta Physiologica, 132, 29–34. https://doi.org/10.1111/j.1748-1716.1988.tb08294.x Van der Kolk, B. A. (2015). The body keeps the score: Brain, mind, and body in the healing of trauma. New York, NY: Penguin. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of

Personality and Social Psychology, 54, 1063–1070. https://doi.org/10.1037//0022-3514.54.6.1063 Weiss, S. J. (1990). Effects of differential touch on nervous system arousal of patients recovering from cardiac disease. Heart & Lung: The Journal of Critical Care, 19, 474–480. White, S. W., Oswald, D., Ollendick, T., & Scahill, L. (2009). Anxiety in children and adolescents with autism spectrum disorders. Clinical Psychology Review, 29, 216–229. https://doi.org/10.1016/j.cpr.2009.01.003 Wilhelm, F. H., Kochar, A. S., Roth, W. T., & Gross, J. J. (2001). Social anxiety and response to touch: Incongruence between self-evaluative and physiological reactions. Biological Psychology, 58, 181–202. https://doi.org/10.1016/s0301-0511(01)00113-2 Zimet, G. D., Dahlem, N. W., Zimet, S. G., & Farley, G. K. (1988). The Multidimensional Scale of Perceived Social Support. Journal of Personality Assessment, 52, 30–41. https://doi.org/10.1207/s15327752jpa5201_2 Author Note. Preman Koshar, https://orcid.org/0000-00017475-652X, Department of Psychology, Franklin & Marshall College; Megan L. Knowles, Department of Psychology, Franklin & Marshall College. Preman Koshar is now at the Department of Psychiatry at The University of North Carolina at Chapel Hill. This study was supported by a Leser Grant provided by Franklin & Marshall College, as well as by personal contributions from Kevin Koshar, Megan L. Knowles, Sharon Koshar, and Glenn Koshar. Correspondence concerning this article should be addressed to Preman Koshar, Department of Psychiatry at The University of North Carolina at Chapel Hill. E-mail: pkoshar@fandm.edu

191

https://doi.org/10.24839/2325-7342.JN25.2.192

Who Am I? Identity Development During the First Year of College Madelynn D. Shell* , David Shears, and Zoe Millard The University of Virginia’s College at Wise

ABSTRACT. The first year of college provides emerging adults with time to explore their identity, but changes in exploration and commitment may differ across different aspects of identity. It was hypothesized that (a) exploration and commitment would be stable across the first year, but that fall exploration would prompt greater commitment in spring, indicating that individuals are moving toward identity achievement; (b) there would be differences in these patterns across the domains; and (c) exploration would be negatively associated with satisfaction with life, although commitment would be positively associated with life satisfaction. A total of 98 college students reported on satisfaction with life and identity exploration and commitment within 8 domains in the fall and spring of their first year. Results demonstrated that exploration and commitment were stable over time. In general, greater fall commitment predicted less spring exploration, whereas greater fall exploration predicted less spring commitment, but these patterns differed by domain. In addition, global exploration was negatively related to life satisfaction, and peer relationship identities were particularly important in predicting psychological well-being. These findings suggest that identity development is not complete by the end of the first year of college, and that students would benefit from support as they the transition from exploration to commitment. Keywords: identity development, emerging adulthood, college transition, psychological well-being

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

192

dentity development is the process by which adolescents and emerging adults search for answers to fundamental questions that surround who they are, including their political and religious views, career choices, and future achievements (Ole, 2016). Identity development is often characterized as beginning in adolescence and continuing through emerging adulthood (Arnett, 2000; Kunnen, Sappa, van Geert, & Bonica, 2008). However, because of differing opportunities for exploration or commitment, identity development in some domains (i.e., components of identity, such as religion or political views) may develop earlier or later than others. Thus, it is important to understand not just identity development in general, but domainspecific identity development as well (Kunnen et al., 2008).

Contextual factors may play an important role in the process of identity development. In par ticular, the transition from high school to college may result in significant changes in a number of domains such as romantic relationships, friend ships, and academics (Azmitia, Syed, & Radmacher, 2013; Kunnen et al., 2008). This transition is a time when students explore new potential identities and begin to commit to one of their choosing (Luyckx, Goossens, Soenens, & Beyers, 2006). Consistent with this idea, previous work has found that high school students often had neither explored nor committed to an identity, whereas college students were more likely to be currently exploring identity options (Verschueren, Rassart, Claes, Moons, & Luyckx, 2017). Because the first year of college may be a time of identity instability and development (Azmitia et al., 2013), it is important to understand

*Faculty mentor

Shell, Shears, and Millard | Identity Development College

the process of identity development during this time. These potential shifts in identity could also have important impacts on psychological well-being. Identity Exploration and Commitment Identity development occurs as emerging adults explore and learn about different identity alterna tives and consider which to adopt as their own. It involves two separate processes: exploration and commitment (Kunnen et al., 2008). Exploration occurs when individuals search for and collect information on alternative identities or perspec tives. Commitment occurs when emerging adults have selected an identity that they plan to adopt as their own (Kunnen et al., 2008). Historically, exploration and commitment have been used to develop four identity statuses: achievement, foreclosure, moratorium, and diffusion (Marcia, 1966). Achievement, considered an optimal end result (Marcia, 1966), is when identity exploration has occurred and a commitment has taken place. Foreclosure is when a commitment has been made with no exploration. Moratorium involves current exploration but no commitment. Lastly, diffusion is when individuals have not explored or made a commitment. Although it is generally expected that exploration will lead to commitment, commitment and identity achievement are not always final, as emerging adults may also revisit their identity and explore new views even after making a commitment (Marcia, 1966). Despite the previous categorical treatment, exploration and commitment happen on a con tinuum; thus, it may be more useful to investigate identity development in a continuous fashion (McLean, Syed, & Shucard, 2016), particularly when investigating changes over time. By taking a dimensional approach to identity development, one can detect changes in exploration and commitment that may not be captured when creating categories. Previous evidence has suggested that continuity in exploration and commitment is common over the course of one year (Meeus, van de Schoot, Keijsers, Schwartz, & Branje, 2010). However, there is also gradual change in exploration and commitment over time, as emerging adults move toward identity achievement (Meeus et al., 2010). As students prog ress in identity development, it would be expected that greater exploration would be associated with greater commitment (and therefore identity achievement). In contrast, if individuals are still in the process of identity development, exploration and commitment may be negatively related (i.e., foreclosure or moratorium).

Multiple Domains of Identity Although identity development is often described as a singular process, emerging adults develop multiple domain-specific identities, and this process may occur with different timing for differ ent domains (McLean et al., 2016). For example, individuals might have made a commitment in their friendship identity (with or without explora tion), while simultaneously actively exploring their career identity with no commitment. Eight identity domains have been examined in previous research (McLean et al., 2016): career, family, friendship, romantic relationships, religion, politics, general values, and gender roles. Changes in career. Career decisions are often multistep processes that may involve extensive exploration in college (Gati, Krausz & Osipow, 1996; Rosemond & Owens, 2018). The selection of an undergraduate major may set the trajectory of emerging adults’ future career options, thus declaring a major may prompt career identity exploration. Declaring a major is often required in the first few semesters of college, therefore students are forced to make a commitment relatively early in their college career (Rosemond & Owens, 2018). Although some students may use the first year of college to thoroughly explore potential majors and long-term careers prior to making a commitment, others may select a major without much consider ation of alternatives. Making a career commitment has a positive impact on the college experience, whereas exploration of career goals is associated with negative attitudes about college (Waterman & Waterman, 1970). This may be a reflection of the stress associated with career exploration. Thus, academic major selection may drive career identity exploration during the first year of college. Changes in relationships. Similar to career identity, social relationships may also change sig nificantly during the first year of college. Changes in family dynamics, particularly parent-child relationships, often occur when students move away from home (Guassi Moreira & Telzer, 2015; Mounts, Valentiner, Anderson, & Boswell, 2006). This may lead students to reconsider their roles in the family, leading to exploration. Furthermore, the family itself may affect identity exploration, as emerging adults with more family support engage in more exploration (Jourdan, 2006) and may be more likely to develop their family and other identities. Therefore, more positive support from family members can result in more positive identity exploration experiences.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

193

Identity Development College | Shell, Shears, and Millard

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

194

Friendship dynamics often change as well dur ing the transition to college. Comfort from budding college friendships aids in establishing identity development across domains (Barry, Madsen, Nelson, Carroll, & Badger, 2009). However, the loss of high school friendships and the formation of new friendships in a larger environment (Oswald & Clark, 2003) may cause instability as students explore the types of friends and friendships they prefer and make commitments to their roles as friends. Along with changing friendships, college introduces emerging adults to a different dating scene. As students are establishing more mature romantic relationships, they may do further exploration and commitment as to who they are as a romantic partner. Previous evidence has suggested that romantic relationship quality and dissolu tion are often related to identity development in emerging adulthood (Barry et al., 2009; Norona, Olmstead, & Welsh, 2017). Thus, the dramatic changes in interpersonal environment during the transition to college may prompt emerging adults to engage in exploration and commitment in a number of relationship domains. Changes in beliefs and opinions. The shift from high school to college may also prompt a period of questioning of personal beliefs and opinions such as politics, religion, general values, and gender roles, which could lead to identity instability. For example, religious exploration may occur when emerging adults engage in critical discussions about religious and value-related ideas with their peers, parents, or college faculty (Foster & LaForce, 1999). However, evidence has also suggested that students tend to adopt religious ideals early on from their parents or other significant adults (Copen & Silverstein, 2007). Thus, it is possible that rather than being something to explore, religion could instead provide stability and comfort to students. If this were the case, during the transition to college students may be less inclined to explore or com mit to new religions, resulting in greater stability in religious identity compared to other domains. Likewise, other domains involving more abstract beliefs (e.g., politics, gender roles, and values) may also develop later in adulthood. For many such domains, students initially adopt the beliefs of their parents during high school (Johnson, 2017; Wray-Lake, 2019). These beliefs may remain stable throughout college as emerging adults focus on more pressing and immediate aspects of identity (e.g., career and relationship identities), but may be revised in adulthood.

Identity Development and Psychological Well-Being The process of identity development may also impact social or emotional well-being, including satisfaction with life. Evidence has suggested that subjective well-being and satisfaction with life are influenced by domains that are important and relevant to an individual (Diener, 1984). Thus, a link between identity development and satisfaction with life would demonstrate the importance of identity in psychological well-being. Consistent with this idea, committing to an identity has been associ ated with higher quality of life and mental health (Azmitia et al., 2013; Berzonsky, 2003; Luyckx et al., 2006; Oleś, 2016; Waterman & Waterman, 1970). In contrast, identity exploration in college has been associated with poorer mental health (Luyckx et al., 2006; Oleś, 2016) and lower satisfaction with college (Waterman & Waterman, 1970). However, there is also some evidence that exploration may have positive impacts on well-being (Berzonsky, 2003; Kunnen et al., 2008; Luyckx et al., 2006). The process of exploration may be stressful in the short-term, although it may produce more positive identity outcomes in the long term. If some areas of identity are more important than others, there may be domain-specific differences in the impacts of identity development on satisfaction with life. The Present Study The current study investigated changes and stability in identity exploration and commitment across multiple domains during the first year of college, and investigated the link between identity devel opment and psychological well-being. Students reported on exploration and commitment in eight domains of identity development, and also on satisfaction with life, during the fall and spring of their first year of college. Hypothesis 1 focused on general expectations of stability versus change in both global and domain-specific identity develop ment. It was hypothesized that, although there would be stability in exploration and commitment (Hypothesis 1a), there would also be progress in identity development. It was expected that fall exploration would predict greater commitment in spring (Hypothesis 1b), a pattern indicative of identity achievement. In contrast, fall commit ment was expected to be negatively associated with spring exploration (Hypothesis 1c), because individuals who have committed may be less likely to continue to explore other options. Hypothesis 2 predicted that that there would be differences in

Shell, Shears, and Millard | Identity Development College

these patterns across domains. In particular, it was expected that identities that were more directly influenced by the college environment (such as career or relationships) would be more likely to experience change toward identity achievement (i.e., positive relationship between fall explora tion and spring commitment, greater support for Hypothesis 1b). In contrast, in domains related to more abstract beliefs and opinions (e.g., politics, religion), individuals may be less likely to engage in exploration if they initially made a commitment (i.e., greater support for Hypothesis 1c). Finally, Hypothesis 3 focused on the relation between identity development and satisfaction with life. It was expected that, across domains, exploration would be negatively associated with satisfaction with life (Hypothesis 3a), whereas commitment would be positively associated with life satisfaction (Hypothesis 3b).

Method Participants The current study was conducted at a four-year pub lic liberal arts institution. Following Institutional Review Board approval at the University of Virginia’s College at Wise, college students were invited to complete a series of questionnaires during fall and spring of their first year. In fall, 144 students completed the survey, and in spring, 108 participants completed it. Only the 98 participants (55 women, 56%; 43 men, 44%) who completed questionnaires in both fall and spring were used in the present analyses. Of those students, 91% identified as European American, 10% as African American, 2% as Latino/a, and 2% as American Indian/Native American. Measures The Ego Identity Process Questionnaire (Balistreri, Busch-Rossnagel, & Geisinger, 1995) was used to assess identity development in eight domains (career, family, friendship, romantic relationships, religion, politics, values, and gender roles) in both fall and spring. Thirty-two questions assessed explo ration and commitment in each domain (see Table 1 for Cronbach’s α for this study at each time point). Exploration questions included items such as, “I have tried to learn about different occupational fields to find the best one for me.” Commitment questions included items such as, “I have definitely decided on the occupation I want to pursue.” Participants responded on a 6-point Likert-type

scale, from 1 (strongly disagree) to 6 (strongly agree). Items were reverse-scored as needed so that higher scores indicated more exploration or more com mitment. The global exploration and commitment scores included 16 items for each construct, and the mean was taken across all domains. For the domain scores there were two items each assessing explora tion and commitment, and the domain score was the mean of the two items. See Table 1 for means and standard deviations for each domain. The Satisfaction With Life Questionnaire (Diener, Emmons, Larsen, & Griffin, 1985) was used to assess psychological well-being in both fall and spring. The questionnaire had five statements, which participants responded to on a 7-point Likert-type scale of 1 (strongly disagree) to 7 (strongly agree). Example items included, “If I could live my life over, I would change almost nothing,” and “In most ways my life is close to my ideal” (see Table 1 for descriptive statistics). TABLE 1 Descriptive Statistics for All Variables Fall

Spring

Global exploration

3.73

0.63

.69

3.63

0.62

.68

Global commitment

4.24

0.73

.78

4.18

0.69

.76

Career exploration

4.44

0.86

4.32

1.05

Career commitment

4.08

1.21

4.21

1.19

Family exploration

3.90

1.13

3.77

0.98

Family commitment

4.51

1.20

4.43

1.06

Friendship exploration

3.82

1.05

3.77

1.01

Friendship commitment

4.28

1.01

4.31

0.99

Romantic relationships exploration

3.94

1.15

3.71

1.19

Romantic relationships commitment

4.22

1.25

4.37

1.18

Religion exploration

3.16

1.14

3.11

1.06

Religion commitment

4.58

1.04

4.54

1.10

Politics exploration

3.95

1.24

3.86

1.32

Politics commitment

3.83

1.44

3.70

1.29

Values exploration

3.47

1.02

3.35

1.14

Values commitment

4.62

1.09

4.40

1.05

Gender roles exploration

3.17

1.31

3.12

1.14

Gender roles commitment

3.81

1.38

3.54

1.21

Satisfaction with life

5.24

1.35

5.20

1.22

.91

SUMMER 2020 .89

Note. N = 98.

PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

195

Identity Development College | Shell, Shears, and Millard

Procedure Participants were assessed during the first eight weeks of the fall and the last 12 weeks of the spring semester in their first year of college. In fall, the majority of participants (77%) completed surveys within the first four weeks (the median date for fall surveys was 2.5 weeks after the first day of classes). There was a mean of approximately 6.5 months (194 days, SD = 23.77) between the administration of the fall and spring surveys. At each time point, participants were contacted via phone or email, and then came to the lab to complete the question naires under the supervision of a trained research assistant.

Results First, bivariate correlations were run among all vari ables for descriptive purposes, to establish general relations among variables (see Supplemental Table 1 at https://osf.io/6tjyw). Of particular interest was the relationship between exploration and commitment within domain, which could indicate the stage of identity development. In fall, there was a significant negative relationship between exploration and commitment in global identity (r = -.32, p < .01) as well as for politics (r = -.41, p < .01), values (r = -.25, p < .01), and gender roles (r = -.65, p < .01). Thus, for these domains, higher exploration was associated with less commitment, and less exploration was associated with more com mitment. Likewise, in spring, there were significant negative correlations between exploration and commitment in all domains except for family and religion (rs = -.21 to -.51, all ps < .05). This evidence suggests that within each time point individuals may be in either moratorium (exploring without commitment) or foreclosure (made a commitment without exploration).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

196

Stability and Change Over Time in Exploration and Commitment First, analyses explored the relationship between exploration and commitment over time during the first year of college to address Hypothesis 1. Linear regressions were used to identify whether fall explo ration or commitment predicted spring exploration and commitment (see Table 2). The interaction between fall exploration and fall commitment was included in preliminary models, but was not significant for any analyses, thus was excluded from the final models. The results for global identity supported Hypothesis 1a by demonstrating that exploration and commitment were quite stable;

greater fall global exploration predicted greater spring exploration, and likewise greater fall global commitment predicted more spring commitment. Contrary to Hypothesis 1b, greater fall global exploration predicted less spring commitment. However, consistent with the Hypothesis 1c, greater fall commitment predicted less spring exploration. Thus, during the first year of college, students did not appear to complete identity achievement, but rather were more likely to continue the identity task they were engaged in during the fall. Next, fall exploration and commitment within each domain were used to further test Hypothesis 1 and predict spring exploration or commitment in the same domain (see Table 2). These analyses indicated several domain-specific patterns of identity development that differed from the global patterns, consistent with Hypothesis 2. The most common pattern, shared by friendship, politics, and values domains, demonstrated stability in both exploration and commitment, consistent with Hypothesis 1a. In addition, consistent with Hypothesis 1c, for each of these domains, more commitment in the fall predicted less exploration in spring. This pattern is also consistent with the Hypothesis 2 expectation that domains related to abstract beliefs (i.e., politics and values) would demonstrate the patterns predicted by Hypothesis 1c, although the same pattern was not expected for friendship, which is more directly impacted by the college transition. Another common pattern was found for religion and gender role identities. Again, there was significant stability in both exploration and commitment within these domains (consistent with Hypothesis 1a), but students who engaged in more exploration in fall were lower in commitment in spring, contrary to Hypothesis 1b. There was also no significant relationship between fall commitment and spring exploration, contrary to Hypothesis 1c and Hypothesis 2. Next, career identity demonstrated a unique pattern. Although career commitment was stable, as expected, fall exploration was not a significant predictor of spring exploration, contrary to expec tations in Hypothesis 1a. Thus, exploration in the career domain was the only indicator that was not stable across the first year of college. Contrary to Hypothesis 1b, greater fall career exploration predicted less spring commitment. The lack of this relationship in the career domain contradicts Hypothesis 2, that domains directly related to the college transition would experience more growth

Shell, Shears, and Millard | Identity Development College

toward identity achievement. In addition, greater fall commitment did predict less spring exploration, consistent with Hypothesis 1c, although according to Hypothesis 2 this pattern was not expected to be as prominent in a domain directly influenced by the college transition. It is important to note, however, that although the overall model for spring career exploration was significant, the predictors only explained a small portion of the variance (R2 = .07). There were several patterns displayed in only one domain. For example, romantic relationship identity mirrored the pattern for global explora tion and commitment, with evidence for stability in both exploration and commitment, consistent with Hypothesis 1a. There was also a negative relationship between exploration and commitment, consistent with Hypothesis 1c and inconsistent with Hypothesis 1b (according to Hypothesis 2, this domain was expected to demonstrate significant support for Hypothesis 1b). Finally, the pattern for family identity indicated only stability in explora tion and commitment, consistent with Hypothesis 1a and inconsistent with Hypotheses 1b and 1c, although the model predicting spring family exploration did not meet the standard threshold of significance (p = .06).

exploration, commitment, and satisfaction with life could predict change in satisfaction with life across the first year of college (see Table 3). The initial model included a Fall Exploration x Fall Commitment interaction, but it was not significant in any of the analyses and therefore was excluded from final models. As expected according to Hypothesis 3a, higher global exploration in fall pre dicted significant decreases in satisfaction with life but, contrary to Hypothesis 3b, global commitment did not predict change in satisfaction. However, this pattern differed across domains. Consistent with global results and Hypothesis 3a, exploration in friendship and political identity domains had a negative impact on life satisfaction. The pattern for commitment was less clear. Consistent with Hypothesis 3b, fall commitment with regard to romantic relationship identity improved satisfaction with life over the first year of college. However, fall commitment to a political identity decreased life satisfaction, contrary to Hypothesis 3b. Finally, nei ther exploration nor commitment in career, family, religion, values, or gender roles predicted changes in life satisfaction across the first year of college.

Discussion This study explored identity development in multiple domains during the first year of college. Overall, findings demonstrated that, consistent with Hypothesis 1a, there was significant stability in exploration and commitment from fall to spring.

Satisfaction With Life To address the final hypothesis, that identity development may impact psychological well-being, linear regressions were used to identify whether fall

TABLE 2 Fall Exploration and Commitment Within Domain Predict Spring Exploration and Commitment in Some Domain Global Sp Explore

Career Sp Commit

Sp Explore

Family Sp Commit

Sp Explore

Romantic Relationships

Sp Commit

Sp Explore

Sp Commit

.53

<.01

-.26

<.01

.17

.10

-.22

.02

.22

.03

-.03

.78

.22

.03

.07

.47

.36

<.01

-.25

.01

Fall commitment within domain

-.26

<.01

.50

<.01

-.22

.03

.42

<.01

.07

.52

.31

<.01

-.32

<.01

.40

<.01

-.25

.01

.41

<.01

.42

<.01

.39

<.01

.07

.01

.24

<.01

.06

.09

<.01

.15

<.01

.17

<.01

.20

<.01

.25

<.01

Religion

Sp Explore

Fall exploration within domain

Friendship Sp Commit

Politics

Values

Gender Roles

Sp Explore

Sp Commit

Sp Explore

Sp Commit

Sp Explore

Sp Commit

Sp Explore

Sp Commit

Fall exploration within domain

.31

<.01

-.20

.03

.50

<.01

-.10

.34

.27

.01

-.11

.22

.32

.01

-.32

.01

Fall commitment within domain

-.18

.06

.43

<.01

-.20

.03

.40

<.01

-.23

.02

.52

<.01

-.22

.07

.24

.04

.14

<.01

0.25

<.01

.37

<.01

.20

<.01

.16

<.01

.31

<.01

.24

<.01

.27

<.01

Note. Sp Explore is spring exploration. Sp Commit is spring commitment. R and corresponding p value relate to whole model. N = 98 2

197

Identity Development College | Shell, Shears, and Millard

As expected by Hypothesis 1c, global commitment predicted less exploration over time, but contrary to expectations of Hypothesis 1b, global exploration predicted less (rather than more) commitment over time. Thus, global findings suggested that students may not yet be progressing toward identity achieve ment (contrary to Hypothesis 1b), but rather engaging in either exploration or commitment throughout the first year of college. Furthermore, these patterns differed by domains, consistent with Hypothesis 2, although domain-specific findings generally underscored the pattern of stability in either exploration or commitment, rather than progress toward identity achievement in domains directly affected by the college transition. Finally, as expected by Hypothesis 3a, global exploration pre dicted decreases in satisfaction with life, particularly in the friendship and political domains of identity. In contrast, global commitment did not predict life satisfaction, although these findings differed some what by domain. Thus, this study demonstrated that the processes of exploration and commitment may differ for different domains of identity, and these processes may also have differential impacts on psychological well-being.

in fall were lower in commitment in spring, and those higher in commitment in fall were lower in exploration in spring. Mapping this on to the traditional categorical discussion of identity status (e.g., Meeus et al., 2010), this suggests that students who start out higher on exploration are likely remaining in moratorium (continuing exploration without commitment), rather than transitioning to achievement (as predicted in Hypothesis 1b). These students may enter college actively considering and exploring potential identities. During the first year of college, they may be continuing to explore identity options. Although they may eventually increase their commitment over time (Meeus et al., 2010; Verschueren et al., 2017), the lack of positive relationship between fall exploration and spring commitment, indicated that few students made direct progress toward identity achievement by the end of their first year. In contrast, students who were high on commitment in fall were less likely to engage in exploration (consistent with Hypothesis 1c), thus likely remaining in achievement (if they previously explored) or foreclosure (if they did not explore prior to entering college). These students may be unwilling to explore even in contexts that encourage it, such as starting college or moving away from home, and instead remain committed to their initial identity. The college transition appears to encourage continued exploration for those who explore early on, but leads to doubling down on commitment for those who enter college commit ted to an identity. This is consistent with Meeus and colleagues (2010), who found movement toward identity achievement over four years, but not one year. Thus, the transition to identity achievement may not have been captured in the time frame of the present study.

Stability and Change in Exploration and Commitment Consistent with Hypothesis 1a, there was significant stability in exploration and commitment (Meeus et al., 2010). Students who began their first year as higher in exploration were likely to continue to explore identity options in spring, and likewise students high in commitment were likely to remain committed in spring. Global exploration and commitment were also inversely related over time, indicating that students who engaged in exploration

TABLE 3 Fall Satisfaction With Life, Exploration, and Commitment Predict Spring Satisfaction with Life Global

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

198

Career

Family

Friendship

Romantic Relationships

Religion

Politics

Values

Gender Roles

Fall Satisfaction With Life

.48

<.01

.52

<.01

.53

<.01

.50

<.01

.43

<.01

.50

<.01

.56

<.01

.47

<.01

.46

<.01

Fall Exploration Within Domain

-.22

.02

.05

.58

-.11

.22

-.19

.04

-.08

.39

-.12

.20

-.22

.02

-.15

.10

-.19

.11

Fall Commitment Within Domain

.04

.72

.00

.97

.06

.51

.02

.86

.21

.03

.08

.36

-.24

.02

.12

.20

.02

.84

.31

<.01

.26

<.01

.29

<.01

.31

<.01

.32

<.01

.30

<.01

.31

<.01

.32

<.01

.32

<.01

Note. Dependent variable for all analyses in spring satisfaction with life. R and corresponding value relate to the whole model. N = 94 2

Shell, Shears, and Millard | Identity Development College

Differences in Identity Development Across Domain As expected in Hypothesis 2, results indicated that there were differences in the process of develop ment in different identity domains. However, in contrast to predictions, even in domains directly impacted by the transition to college, students did not yet appear to be making progress toward identity achievement. First, over time, fall career commitment predicted less exploration and more commitment in spring, but early exploration did not predict later exploration, and was associated with less commitment. Thus, the process of taking general education courses and choosing a college major may promote exploration in the career domain, but this exploration does not appear to be completed by the end of the first year. The fact that these patterns were unique to career identity underscores that colleges provide an important opportunity for career exploration, and this may drive occupational identity achievement (Fouad & Ghosh, 2016). However, these patterns did not indicate progress toward identity development, at least within the first year, thus were inconsistent with Hypothesis 2. For friendship, political, and value identities, in addition to stability in both exploration and commitment, entering college having greater com mitment predicted less exploration in spring. Thus, making an early commitment in friendship, politi cal, or value domains appeared to deter students from engaging in exploration in these domains during their first year of college. Many of these students may be in foreclosure status, in which they have made a commitment without exploration, and continue to stay with that decision despite opportunities to explore. On the other hand, some students might have explored prior to entering college, and thus might enter college in identity achievement. These students might be happy with their identity and therefore not considering other options. Although Hypothesis 2 predicted this pattern for the abstract domains of political and value identities, it was initially hypothesized that students would experience growth toward identity achievement in friendship identity as a result of the transition to college. The common pattern in these domains may be rooted in the fact that students might have had extensive exposure to potential friendship, political, and value identities prior to college entry, and therefore might be hesitant to consider alternatives. Consistent with this idea, evidence has suggested that civic and political

identity development often begins in adolescence (Johnson, 2017; Wray-Lake, 2019), so many of these students might have already explored and made a commitment before they enter college. Similar to the previous pattern, religion and gender roles demonstrated stability in both exploration and commitment. However, higher exploration in these domains in fall was associated with lower in commitment in spring. This is in contrast to expectations that individuals would move from exploration into commitment during their first year, although according to Hypothesis 2 these patterns were expected to be less likely in abstract domains such as these. The negative relation between exploration and commitment indicates that exploration in these domains may be particularly likely to lead individuals to continue to question their identity. Consistent with this, some evidence has suggested that religion serves to promote identity development (King, 2003), so during the process of exploration in fall, students might realize how little they know about their religious or gender role options and therefore be less likely to make a commitment in spring. Romantic relationship identity mirrored the global pattern; those who were currently exploring their romantic identity were less likely to make a commitment, and those who had made a commit ment were less likely to explore. This is in contrast to the Hypothesis 2 expectation that relationship domains would be impacted by the college transi tion and therefore experience growth toward achievement. It could be that students have either entered college in a romantic relationship or with an idea of their romantic identity and refuse to consider alternatives (foreclosure) or have entered with plans to explore their romantic options with no interest in making a commitment (morato rium). Emerging adults just entering college may be transitioning between the more casual affection phase of a relationship and more complex bonded love (Seiffge-Krenke, 2003), and these patterns of romantic relationship identity development may reflect these different relationship goals. Finally, the pattern for family identity indicated only stability in exploration and commitment, with no relationship between exploration and commit ment over time. Family relationships may be set in place earlier in adolescence, and therefore, may be unlikely to change at college entry, even in the face of substantial changes such as moving out and establishing long-distance relationships.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

199

Identity Development College | Shell, Shears, and Millard

Overall, there was the most substantial support for Hypothesis 1a, and stability in exploration and commitment. It also appeared that rather than the expected progress toward identity development (Hypothesis 1b), students were instead engaged in either exploration or commitment, but were not transitioning between the two. Development in the different domains of identity reflected pieces of this global pattern, suggesting that the general process may be the same but the timing of changes within domains might be different. In contrast to Hypothesis 2 expectations, findings for domains directly impacted by the college transition as well as those related to more abstract beliefs demon strated that students were stably engaging in either exploration or commitment within the first year of college, rather than transitioning toward identity achievement.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

200

Identity Development and Satisfaction With Life Consistent with Hypothesis 3a and previous research (Kunnen et al., 2008; Luyckx et al., 2006; Oleś, 2016), greater global exploration in fall predicted significant decreases in satisfaction with life over the course of the year. The process of exploration and questioning one’s beliefs or values may create stress that decreases satisfaction with life. Conversely, global commitment did not predict sat isfaction with life. This is in contrast to Hypothesis 3b and previous research that demonstrated links between commitment and psychological well-being (Azmitia et al. 2013; Berzonsky, 2003; Oleś, 2016). However, these effects differed across domains. First, peer relationship identities appeared to have a significant impact on psychological well-being. Friendship exploration was associated with poorer satisfaction with life, consistent with Hypothesis 3a. Because students generally rely on precollege friendships for social support during the first year of college (Shell & Absher, 2019), questioning friendship identity could create strain in these relationships and lead to less social support. Alternatively, students may engage in friendship exploration in fall because of a significant loss of social support during the college transitions (Oswald & Clark, 2003), leading to a decrease in life satisfaction. Conversely, consistent with Hypothesis 3b, commitment in romantic relationship identity had a positive impact on satisfaction with life, con sistent with previous research linking commitment to better psychological well-being (Berzonsky, 2003; Olés, 2016). Students high on commitment to a romantic identity may be entering college already in

a romantic relationship, and the emotional support from that relationship could improve satisfaction with life. Even if students are not in a committed relationship, making a commitment about prefer ence and expectations for romantic partners could relieve the stress of actively exploring potential romantic options. Combined, this evidence demon strates that identity with regard to peer relationships has a significant impact on psychological well-being during the first year of college. Political identity also appeared to be par ticularly linked to well-being, although both exploration and commitment in the political domain were associated with a decrease in wellbeing over the first year. Political identity develops in adolescence as a result of parental influences and high school community engagement experiences (Wray-Lake, 2019), and therefore may be firmly rooted in students’ family values, especially in more conservative and rural areas such as the one in which this study was conducted (Feinberg, Tullett, Mensch, Hart, & Gottlieb, 2017). This could make challenges to identity in this domain particularly impactful. Students who explore political views may experience stress as a result of reconsidering previously accepted assumptions, particularly if these views contrast those of their families. However, commitment in political identity was also negatively associated with satisfaction with life. Exposure to alternative ideas may cause stress or dissatisfaction for students who enter college having already com mitted to a political perspective. Overall, it appears that the effects of exploration and commitment may differ depending on domain. Implications This study demonstrated significant stability in exploration and commitment during the first year of college, although the two constructs were inversely related over time. Thus, colleges might support identity development by providing structured opportunities for exploration, as well as encouraging commitment following explora tion to help students progress toward identity achievement. However, it also demonstrated that different domains of identity develop at different rates, highlighting the importance of investigating domains of identity separately. Exploration of career identity was particularly common, but did not predict later commitment. Greater support and guidance for career exploration may ensure that first year college students have thoroughly investigated potential careers and are making

Shell, Shears, and Millard | Identity Development College

commitments. For friendship, political, and value identities, students who have committed appear to be less likely to explore over time, thus these identities may not change much in the first year of college. In contrast, for religion and gender roles, students who begin exploring are less likely to make a commitment. This suggests that students may ben efit from support for exploration in these domains in particular; classes or discussion groups may help students explore and may motivate commitment. Furthermore, associations between identity status and psychological well-being suggested that peer relationship identities may be particularly linked to satisfaction with life, thus colleges might focus on supporting such relationships. Limitations Despite these contributions, there were several limitations to this study. First, the current study only investigated identity development in the first year of college; continuing farther into college may lead to better understanding of individual trajec tories of identity development, as well as a better picture of the timing of identity achievement. The relationship between identity and psychological well-being may also differ over a longer period of time. Second, findings demonstrated links between identity development and a very general measure of psychological well-being (satisfaction with life). Identity development within each domain may be associated with different measures of psychological well-being. Finally, there are some limitations to generalizability. Many students were local to the rural Appalachian community in which the study was conducted, which tends to be predominantly Christian and culturally conservative. This may impact the amount of reliance on family (Russ, 2010) as well as early expectations of responsibility, leading to earlier commitment. In addition, identity development processes may differ for emerging adults who do not attend college, as they may have fewer opportunities for exploration. Conclusions Altogether, this study provided important informa tion about the identity development process during the first year of college. Consistent with hypotheses, exploration and commitment were generally stable over time. Students who had more commitment in fall did less exploration in spring, and those who did more exploration in fall were less likely to make a commitment in spring. This demonstrates that identity development is certainly not complete

by the end of the first year of college. This study also highlighted the importance of investigating domains of identity separately, as patterns differed across domains. It also underscored the importance of continuing to investigate identity development throughout students’ entire college career, as exploration and commitment to various domains may change later in college. Finally, it suggested that colleges may increase psychological well-being by providing support for peer relationship identities and exploration in general.

References Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469–480. https://doi.org/10.1037//0003-066x.55.5.469 Azmitia, M., Syed, M., & Radmacher, K. (2013). Finding your niche: Identity and emotional support in emerging adults’ adjustment to the transition to college. Journal of Research on Adolescence, 23, 744–776. https://doi.org/10.1111/jora.12037 Balistreri, E., Busch-Rossnagel, N., & Geisinger, K. (1995). Development and preliminary validation of the Ego Identity Process Questionnaire. Journal of Adolescence, 18, 179–192. https://doi.org/10.1006/jado.1995.1012 Barry, C. M., Madsen S. D., Nelson, L. J., Carroll, J. S., & Badger, S. (2009). Friendship and romantic relationship qualities in emerging adulthood: Differential associations with identity development and achieved adulthood criteria. Journal of Adult Development, 16, 209–222. https://doi.org/10.1007/s10804-009-9067-x Berzonsky, M. D. (2003). Identity style and well-being: Does commitment matter? Identity, 3, 131–142. https://doi.org/10.1207/s1532706xid030203 Copen, C., & Silverstein, M. (2007). Transmission of religious beliefs across generations: Do grandparents matter? Journal of Comparative Family Studies, 38, 497–510. https://doi.org/10.3138/jcfs.38.4.497 Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95, 542–575. https://doi.org/10.1037//0033-2909.95.3.542 Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). Satisfaction With Life Scale. Journal of Personality Assessment, 49, 71–75. https://doi.org/10.1207/s15327752jpa4901_13 Feinberg, M., Tullett, A. M., Mensch, Z., Hart, W., Gottlieb, S. (2017). The political reference point: How geography shapes political identity. PLOS One, 12, 1–13. https://doi.org/10.1371/journal.pone.0171497 Foster, J. D., & LaForce, B. (1999). A longitudinal study of moral, religious, and identity development in a Christian liberal arts environment. Journal of Psychology and Theology, 27, 52–68. https://doi.org/10.1177/009164719902700105 Fouad, N., & Ghosh, A. (2016). Career exploration among college students. Journal of College Student Development, 57, 460–464. https://doi.org/10.1353/csd.2016.0047 Gati, I., Krausz, M., & Osipow, S. H. (1996). A taxonomy of difficulties in career decision making. Journal of Counseling Psychology, 43, 510–526. https://doi.org/10.1037/0022-0167.43.4.510 Guassi Moreira, J. F., & Telzer, E. H. (2015). Changes in family cohesion and links to depression during the college transition. Journal of Adolescence, 43, 72–82. https://doi.org/10.1016/j.adolescence.2015.05.012 Johnson, M. R. (2017). Understanding college students’ civic identity development: A grounded theory. Journal of Higher Education Outreach and Engagement, 21, 31–60. Jourdan, A. (2006). The impact of the family environment on the ethnic identity development of multiethnic college students. Journal of Counseling and Development, 84, 328–340. https://doi.org/10.1002/j.1556-6678.2006.tb00412.x King, P. E. (2003). Religion and identity: The role of ideological, social, and spiritual contexts. Applied Developmental Science, 7, 197–204. https://doi.org/10.1207/s1532480xads0703_11 Kunnen, E. S., Sappa, V., van Geert, P. L. C., & Bonica, L. (2008). The shapes of commitment development in emerging adulthood. Journal of Adult Development, 3, 113–131. https://doi.org/10.1007/s10804-008-9042-y Luyckx, K., Goossens, L., Soenens, B., & Beyers, W. (2006). Unpacking commitment and exploration: Preliminary validation of an integrative

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

201

Identity Development College | Shell, Shears, and Millard

model of late adolescent identity formation. Journal of Adolescence, 29, 361–378. https://doi.org/10.1016/j.adolescence.2005.03.008 Marcia, J. E. (1966). Development and validation of ego-identity status. Journal of Personality and Social Psychology, 3, 551–558. https://doi.org/10.1037/h0023281 McLean, K. C., Syed, M., & Shucard, H. (2016). Bringing identity content to the fore: Links to identity development processes. Emerging Adulthood, 4, 356–364. https://doi.org/10.1177/2167696815626820 Meeus, W., van de Schoot, R., Keijsers, L., Schwartz, S. J., & Branje, S. (2010). On the progression and stability of adolescent identity formation: A five-wave longitudinal study in early-to-middle and middle-to-late adolescence. Child Development, 81, 1565–1581. https://doi.org/10.1111/j.1467-8624.2010.01492.x Mounts, N. S., Valentiner, D. P., Anderson, K. L., & Boswell, M. K. (2006). Shyness, sociability, and parental support for the college transition: Relation to adolescents’ adjustment. Journal of Youth and Adolescence, 35, 68–77. https://doi.org/10.1007/s10964-005-9002-9 Norona, J. C., Olmstead, S. B., & Welsh, D. P. (2017). Breaking up in emerging adulthood: A developmental perspective of relationship dissolution. Emerging Adulthood, 5, 116–127. https://doi.org/10.1177/2167696816658585 Oleś, M. (2016). Dimensions of identity and subjective quality of life in adolescents. Social Indicators Research, 126, 1401–1419. https://doi.org/10.1007/s11205-015-0942-5 Oswald, D. L., & Clark, E. M. (2003). Best friends forever?: High school best friendships and the transition to college. Personal Relationships, 10, 187–196. https://doi.org/10.1111/1475-6811.00045 Rosemond, M. M., & Owens, D. (2018). Exploring career development in emerging adult collegians. Education, 138, 337–351. Russ, K. A. (2010). Working with clients of Appalachian culture. Retrieved from https://www.counseling.org/resources/library/vistas/2010-v-online/ Article_69.pdf Seiffge-Krenke, I. (2003). Testing theories of romantic development from adolescence to young adulthood: Evidence of a developmental sequence. International Journal of Behavioral Development, 27, 519–531.

https://doi.org/10.1080/01650250344000145 Shell, M. D., & Absher, T. N. (2019). Effects of shyness and friendship on socioemotional adjustment during the college transition. Personal Relationships, 26, 386–405. https://doi.org/10.1111/pere.12285 Verschueren, M., Rassart, J., Claes, L., Moons, P., & Luyckx, K. (2017). Identity statuses throughout adolescence and emerging adulthood: A large scale study into gender, age, and contextual differences. Psychologica Belgica, 57, 32–42. https://doi.org/10.5334/pb.348 Waterman, A. S., & Waterman, C. K. (1970). The relationship between ego identity status and satisfaction with college. The Journal of Educational Research, 64, 165–168. https://doi.org/10.1080/00220671.1970.10884127 Wray-Lake, L. (2019). How do young people become politically engaged? Child Development Perspectives, 13, 127–132. https://doi.org/10.1111/cdep.12324 Author Note. Madelynn D. Shell, https://orcid.org/0000-0003-3316-2025, The University of Virginia’s College at Wise, David Shears, Department of Psychology, The University of Virginia’s College at Wise; Zoe Millard, https://orcid.org/0000-0001-7049-5625, Department of Psychology, The University of Virginia’s College at Wise. Zoe Millard is now at the Department of Occupational Therapy at Radford University, Radford, VA. David Shears is now at the Department of Counselor Education at Longwood University, Farmville, VA. This study was supported by The University of Virginia’s College at Wise Office of Admissions, Scholarship, and Financial Aid and Office of Academic Affairs. Supplementary materials are available at https://osf.io/6tjyw. We would like to thank Colton Chase Collins for his significant contributions to editing and revision of the initial draft of this manuscript.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

202

https://doi.org/10.24839/2325-7342.JN25.2.203

Planning to Practice: Action and Coping Plans Increase Days of Meditation Practiced Jonathan N. Cloughesy , Alissa J. Mrazek* University of California Santa Barbara

, Michael D. Mrazek*, and Jonathan W. Schooler*

ABSTRACT. Despite the growing prevalence of mindfulness, few studies have investigated self-regulatory strategies to develop and sustain a meditation practice over the long-term. In the current research, participants (N = 109) naĂŻve to mindfulness and meditation were randomly assigned to an active control or self-regulatory intervention designed to increase the frequency of meditation practice and habit strength over a 4-week practice period. The intervention led to more days of meditation practiced (z = 2.44, p = .02), but did not produce stronger habits of practice (z = 0.63, p = .53). Common perceived barriers to daily practice included busyness, forgetfulness, and lack of motivation. Common enabling factors to daily practice included setting practice reminders and identifying a suitable practice location. Implications, limitations, and future directions are discussed.

Open Materials badge earned for transparent research practices. Materials are available at https://osf.io/aj6te

Keywords: mindfulness, meditation, action plans, coping plans, habit strength

mpirical investigation into the effects of mindfulness meditation over the past three decades has demonstrated a range of psychological and physiological benefits, spurring unprecedented global interest in the practice. Growing interest in meditation has been accompanied by a multitude of books, online courses, and mobile applications that have made high-quality meditation instruction more accessible than ever before (Mrazek et al., 2018). Accordingly, national statistics document a more than threefold increase in meditation practice from 2012 to 2017, with approximately 35 million U.S. adults estimated to have practiced meditation in 2017 alone (Clarke, Barnes, Black, Stussman, & Nahin, 2018). Despite the increasing prevalence of meditation practice among the general public, little is known about how to help novice meditation practitioners maintain a longterm practice. Evidence has suggested that greater total time spent practicing meditation is associated with improved mindfulness and psychological wellbeing among practitioners (Carmody & Baer, 2008;

*Faculty mentor

Huppert & Johnson, 2010; Vettese, Toneatto, Stea, Nguyen, & Wang, 2009). At the same time, greater single-session practice time is associated with reductions in practice adherence over time (Adams et al., 2018), and mindfulness-based intervention participants often struggle to achieve compliance with practice recommendations (Quach, Gibler, & Jastrowski Mano, 2017; Rosenzweig et al., 2010). Consequently, evaluating behavioral strategies that may help to sustain a meditation practice over the long-term is essential if meditation practitioners are to fully benefit from their practice. A prominent behavioral strategy employed to sustain behavior over the long-term is habit formation (Duhigg, 2012; Galla & Duckworth, 2015; Gardner, 2015). Indeed, research has shown habits help individuals maintain greater behavioral consistency across a number of domains, including exercise, diet, sleep, schoolwork, and meditation (Galla & Duckworth, 2015; Lally, Chipperfield, & Wardle, 2008). Habits strengthen over time when a behavior is repeated within a specific context, leading to the automatic initiation of that behavior when the

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

203

Planning to Practice | Cloughesy, Mrazek, Mrazek, and Schooler

associated context is encountered (Gardner, 2015). By shifting the responsibility for the initiation of the behavior to an automatic process, habits reduce reli ance on momentary self-control (Gardner, Lally, & Wardle, 2012). Thus, forming a habit of meditation may help practitioners avoid self-regulatory failures, accomplish their practice goals, and continue to meditate over the long term. Crucial to successful habit formation are the characteristics of the context in which the behavior is enacted. If contextual factors make the behavior’s initiation more difficult, habit formation will falter. Conversely, contexts that remove fric tion and facilitate action can enhance behavioral automaticity and habit acquisition. Given the importance of context, planning when and where to carry out a behavior is critical to the intentional development of habitual behavior. Such plans are referred to in the psychological literature as action plans and have been shown to translate behavioral intentions into actual behavior (Gollwitzer, 1999; Gollwitzer & Sheeran, 2006; Hagger & Luszczynska, 2014). Action plans create a mental link between the relevant context and an individual’s intent to act, facilitating behavior initiation when that context is encountered (Parks-Stamm, Gollwitzer, & Oettingen, 2007). However, well-thought-out action plans may still fail to elicit the desired behavior. Internal barriers to action, such as fatigue, doubt, or low mood, along with external barriers, such as social pressure, time constraints, or an unsuitable environment, can prevent intention from turning into action. Coping plans, which involve planning a response to antici pated barriers, can be used to circumvent these obstacles to action (Kwasnicka, Presseau, White, & Sniehotta, 2013). Like action plans, coping plans mentally link anticipated barriers to a planned response. Encountering a barrier activates the coping response and allows the individual to persist with the intended behavior (Sniehotta, Scholz, & Schwarzer, 2006). Taken together, action plans and coping plans may provide a complementary approach to bolstering long-term adherence to meditation practice.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

204

Present Study The paucity of research into behavioral strategies to facilitate long-term meditation practice underscores a critical gap in the literature on meditation instruc tion. To fill this gap, the present research evaluated the impact of two complementary evidencebased self-regulation strategies on adherence to

mindfulness meditation practice. Specifically, the current investigation assessed the impact of the action and coping plans on number of days of meditation practiced and habit strength over four weeks. We predicted that subjects assigned to create action and coping plans would practice meditation more often over a four-week daily practice period and score higher on a measure of habit strength compared to subjects assigned to an active control.

Method Participants and Design The sample consisted of 109 undergraduate students at a large public university in the south western United States. Participants were recruited on a rolling basis over 5 weeks from an introductory level psychology class that required participation in research to receive course credit. Participants were eligible if they reported no prior experience with mindfulness or meditation practice. Participants were randomly assigned to either an action and coping plan intervention or active control condi tion. Informed consent was obtained from all participants in the study. Procedure Prior to initiating the study, approval was granted from the University of California Santa Barbara institutional review board to conduct this research. In-lab Time 1. At Time 1 (T1), participants were brought into an experimenter room where they watched a 30-minute digital mindfulness crash course designed by our lab (see online supplemen tary materials at https://osf.io/aj6te). At the end of the crash course, participants practiced a brief mindful breathing meditation and received instruc tions to practice mindful breathing meditation for 5 minutes each day over the next 4 weeks. Participants then received the first part of the self-regulatory intervention or active control (described below). All participants were provided a daily tracking sheet on which they were instructed to track their meditation practice starting the following day. Practice period. Over the 4 weeks following T1, participants recorded whether or not they had prac ticed meditation each day on the provided tracking sheet. One week into the 4-week practice period, all participants additionally received an online survey sent via email. The survey contained the second part of the intervention or active control (described below). Two days before each participant’s posttest session, participants were sent an email reminder to return to posttesting.

Cloughesy, Mrazek, Mrazek, and Schooler | Planning to Practice

In-lab Session 2. Four weeks after T1, partici pants returned to the lab for posttesting (T2). An online questionnaire was administered to assess habit strength. Additionally, participants were provided with two open-ended prompts that asked them to describe which factors had helped and which factors had hindered them in maintaining a daily practice. Treatment condition. Part 1. Participants assigned to the treatment condition completed an action plan at T1. First, the instruction to practice mindful breathing meditation daily over the next 4 weeks was reiter ated. Second, participants were guided through the creation of their action plan. Participants were first instructed to consult their class schedule and personal calendars in order to determine the best time and place for them to practice meditation on each day of the week. For each day of the week, participants were asked to stick to the time and place they had selected across all 4 weeks. When choosing a time and place for each day of the week, participants were presented with a list of four criteria designed to help them select a practice time and location. Specifically, participants were told: “Your mindful breathing exercise should occur (a) during a natural transition in your day, (b) when you have enough time, (c) near a place you feel comfortable practicing, and (d) where there won’t be a lot of distractions.” Participants were then provided with a worksheet containing a daily calendar that ranged from 6 a.m. to midnight for each day of the week (see online Supplementary Materials). At the appropriate time on the calendar, participants wrote down the location they intended to practice for each day of the week. Participants also wrote down the activity that they expected to directly precede their planned meditation time for each day of the week. Next, participants created an “enjoyment strategy” that they would use to reward themselves after practicing meditation. Two suggestions were provided (“Before doing mindful breathing, remind yourself this is a chance to set down your burdens and find peace of mind” and “When you finish mindful breathing, take a moment to give yourself a little credit for doing something good for yourself”), but participants were also encouraged to develop their own enjoyment strategies if they so desired. Participants wrote down their chosen enjoyment strategy on the provided worksheet. The reverse side of the worksheet contained the 4-week daily tracking sheet, on which participants were asked to indicate if they had or had not practiced meditation (see online Supplementary Materials).

Part 2. One week into the 4-week daily practice period, participants received a survey that helped them revise their action plan and enjoyment strategy and create a coping plan (referred to as an “obstacle strategy”). First, participants were asked to revise their action plan if they had not practiced mindful breathing on each day so far. Second, participants were instructed to think about the most common barrier to daily practice that they had encountered and to form a coping plan to overcome this obstacle. Third, participants were encouraged to revise their enjoyment strategy if they felt they could make it more effective. Active control. Part 1. Participants assigned to receive the active control were asked to respond to four questions requesting feedback on the digital mind fulness crash course. Participants were asked how enjoyable, relevant, and valuable they had found the course on a 1–7 scale. Participants were then asked to provide recommendations to improve the course. As in the treatment condition, participants were instructed to practice 5 minutes of mindful breathing meditation each day over the next 4 weeks and to track their practice on a provided worksheet (see online supplementary materials). Participants were then told that getting comfortable with practicing an activity is one of the best ways to bring it into your daily routine, and subsequently practiced an additional 5-minute mindful breathing meditation. Part 2. The control condition also received a survey 1 week into the 4-week practice period. The control survey consisted of three steps. First, par ticipants were asked to review their daily tracking sheet. Second, participants were asked to list five of their existing habits. Third, participants were asked to pick one of the five habits they had listed that had been most beneficial to their life and explain how it has benefited their life. Measurement. Habit strength. The 12-item Self-Report Habit Index (Verplanken & Orbell, 2003) was used to measure habit strength (e.g., “Practicing mindful breathing is something I do frequently”; “Practicing mindful breathing is something I do without having to consciously remember”; “Practicing mindful breathing is something that’s typically ’me’”). The index is widely used to measure habit strength of an identified behavior. Items were rated from 1 (disagree) to 6 (agree). Internal reliability was high (α = .95).

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

205

Planning to Practice | Cloughesy, Mrazek, Mrazek, and Schooler

Days of meditation practice. Days of meditation practiced was recorded by participants on a daily practice tracking sheet. Number of unique days meditated from 1 day after T1 until the day before T2 was aggregated to create a total score from 0 to 27. Enablers and barriers to practice. Enablers and barriers to daily practice were assessed by openended responses to two prompts (“Over the last 4 weeks, what factors did you find helpful when trying to maintain your daily mindful breathing practice? Try to think of at least two.”; “Over the last 4 weeks, what types of challenges did you face when trying to maintain your daily mindful breathing practice? Try to think of at least two.”). Analytic plan. Only participants who returned to posttesting were included in the analyses. In the treatment condition, 39 returned to posttest ing, while 40 returned in the control condition. All participants who returned to posttesting were analyzed to assess the effect of condition on habit strength. Due to nonnormal distribution of data, a Mann-Whitney U test was run to test the effect of condition on habit strength. Of the 79 participants who returned to nontesting, 69 participants (treat ment n = 34; control n = 35) also returned their daily tracking sheet containing data on days of meditation practice. A chi-square test of homoge neity revealed no differences between conditions in proportions of nonreturners or returners who did not bring tracking sheet (p = .97). Participants who did not provide tracking sheet data were not included in the analysis assessing effect of condition on meditation practice. Again, due to nonnormal distribution of data, a Mann-Whitney U test was run to test the effect of condition on days of meditation practice.

Results

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

206

Descriptive Statistics Days of meditation practice. On average, par ticipants practiced meditation less than half the total number of days in the practice period (M = 12.06, SD = 7.52). Participants who received the self-regulatory intervention practiced more (M = 14.38, SD = 8.10) than participants who received the active control (M = 9.80, SD = 6.23). Habit strength. On average, participants reported low levels of habit strength (M = 2.22, SD = 1.07). Participants who received the selfregulatory intervention reported similar habit strength (M = 2.28, SD = 1.07) to participants who received the active control (M = 2.15, SD = 1.07).

Correlation between days of practice and habit strength. A moderate-to-large correlation was observed between days of meditation practice and habit strength (r = .49, p < .01). This correla tion held in both the treatment condition (r = .57, p < .01) and control condition (r = .43, p = .01). Main Analyses The intervention increased days of meditation practice. A Mann-Whitney U test was performed to determine the effect of condi tion on days of meditation practice. Days of meditation practice was significantly greater in the treatment condition (mean rank = 40.97) than control condition (mean rank = 29.20), U = 798.00, z = 2.44, p = .02. The intervention did not change habit strength. A Mann-Whitney U test was performed to determine the effect of condition on habit strength. Habit strength did not significantly differ between the treatment condition (mean rank = 41.65) and control condition (mean rank = 38.39), U = 844.50, z = 0.63, p = .53. Enablers and barriers to practice. Of the 79 participants who returned to posttesting, 72 pro vided a written response to items assessing enablers and barriers to daily practice. After responses were coded by the authors, five major themes that enabled participants to practice daily were identified: 31.9% (n = 23) set practice reminders, 26.4% (n = 19) practiced in a quiet and comfortable location, 18.1% (n = 13) practiced in the morning or evening, 16.7% (n = 12) recalled the benefits of the practice, and 8.3% (n = 6) practiced at the same time and same place each day. Five major themes related to common barriers to daily practice were also identified: 72.2% (n = 52) reported being too busy to practice, 62.5% (n = 45) reported forgetting to practice, 41.7% (n = 30) reported motivational barriers to practicing, 16.7% (n = 12) reported difficulties identifying a comfortable and quiet practice location, and 12.5% (n = 9) reported dif ficulties creating a practice schedule.

Discussion The present study found that a self-regulatory intervention increased the days of meditation practiced over a 4-week daily practice period among individuals who had no prior experience with mindfulness or meditation. However, the effect of the intervention did not extend to increased habit strength. There are a number of reasons why the lack of effect on habit strength might have been

Cloughesy, Mrazek, Mrazek, and Schooler | Planning to Practice

observed. For one, lack of context consistency across the practice period might have limited the development of habit strength. Although partici pants were encouraged to practice meditation at the same time and place on a specific day across the practice period (e.g., same time and place every Monday), the intervention did not require participants to practice meditation in the same location or at the same time across each day of the week. Instead, participants were encouraged to consult their weekly schedule when planning their meditations in order to account for the variability in the typical undergraduate student’s schedule. Although this allowed for increased customization and personalization of practice time and location on different days of the week, this approach likely created considerable variability in practice context across days. The lack of context consistency across days might have hindered the acquisition of context-behavior associations that underlie habit strength. Second, although research has suggested that missing a single day of a target behavior does not have a detrimental impact on habit formation (Lally, van Jaarsveld, Potts, & Wardle, 2009), missing many days in a row may hinder habit development (Armitage, 2005). Given that separate action plans were created for each day of the week, some action plans might have worked better than others, potentially contributing to lapses in practice over time. Although the practice period lasted 27 days, participants receiving the intervention meditated an average of just over 14 days, indicating that missed days of practice were common. Third, fac tors intrinsic to the practice of meditation may have restricted habit strength from developing. Research has suggested that complex behaviors may not become as automatic, and therefore may have lower habit strength maximums compared with simple tasks (Verplanken, 2006; Wood, Quinn, & Kashy, 2002). Although ostensibly simple, the initiation of a meditation practice session can involve a series of cognitive, affective, and behavioral processes that cumulatively serve to increase behavioral complex ity. For example, beginning a meditation practice session might involve evaluating whether one has sufficient time, overcoming motivational barriers, and locating a quiet and private place to practice. For this reason, meditation can be seen as a more complex behavior and, as a result, may have a lower maximum of habit strength, further confining differences between conditions. Indeed, research has shown that habit formation can take between 18 and 254 days depending on the complexity of

the behavior and the consistency with which it is performed (Lally, van Jaarsveld, Potts, & Wardle, 2009). The action plan, coping plan, and enjoyment strategies crafted by participants in the intervention were designed to bolster behavioral frequency and increase habit strength by encouraging daily practice while accounting for the variability of a student’s changing daily schedule. However, lack of context consistency, variability in daily practice, and the complexity of initiating a meditation ses sion may all have hampered the development of habit strength. Although differences were not observed between conditions on measures of habit strength, the intervention did lead to more days of medita tion practice for participants in the treatment condition. It is especially promising that these results were observed among a population that lacked prior experience with mindfulness or medi tation. Many mindfulness-based training programs include participants who are motivated and autono mously driven to practice meditation, indicating that this intervention may work just as well, if not better, in these contexts. Indeed, research has suggested that action plans work best when sup ported by high levels of commitment (Gollwitzer, 1999; Sheeran, Webb, & Gollwitzer, 2005). Among the present sample, days of practice accounted for more than 20% of the variance in habit strength, suggesting that interventions resulting in increased practice could lead to stronger habits. This finding points to the potential for action and coping plan interventions to lay the framework for long-term meditation practice. However, additional support may be necessary to bolster daily practice and context consistency. For example, encouraging context consistency across days of the week, or administering weekly check-ins that facilitate plan revision and barrier identification may help practitioners achieve greater context consistency, optimize action plans, and overcome new barriers as they arise. Although the treatment condition received a check-in one week into the practice period, they did not continue to receive check-ins throughout the rest of the practice period. This study also assessed the factors that partici pants reported to have helped or hindered daily practice. A substantial percentage of the sample reported being too busy to practice 5 minutes of meditation per day. However, 5 minutes represents an inconsequential amount of time in the day. Although a majority of the participants reported being too busy, it is more likely that most simply felt

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

207

Planning to Practice | Cloughesy, Mrazek, Mrazek, and Schooler

too busy. This distinction is important. If partici pants were truly too busy, then future interventions may need to focus on helping practitioners reduce commitments in order to incorporate meditation practice into their day. However, given the more likely scenario that the participants felt too busy, busyness can be conceptualized as a motivational barrier. Future research might aim to develop strategies that increase practitioners’ motivation over time, for example, by delivering targeted motivational messages directly before a scheduled practice time. A substantial percentage of partici pants also reported forgetting to practice, as well as other motivational barriers such as fatigue, not seeing value in the practice, or simply not wanting to practice. On the other hand, the most common enabling factor participants reported was setting reminders. Reminders are powerful, not only because they can make one’s plan to meditate more salient, but also because they can simultane ously address common motivational barriers. For example, a reminder set 5 minutes before a planned meditation time could both serve as a reminder to practice and a motivational boost by highlighting the benefits of the practice. Precise and personal ized motivational reminders delivered with digital tools are a promising future direction for address ing common barriers to daily meditation practice.

SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

208

Limitations and Future Directions The study was subject to a number of limitations. First, the multifaceted nature of the intervention makes it difficult to isolate the elements that con tributed most to increased days of practice. Future work could manipulate the presence of action plans, coping plans, and enjoyment strategies inde pendently to precisely identify the contribution of each. Second, the analysis did not look at whether practice trends changed over time. Evaluating whether specific points in time are particularly challenging for most people to stay adherent could allow for the targeted administration of coping plan interventions. Third, the study duration was restricted to a 4-week period, limiting the study’s ability to assess whether habit strength would have continued to develop over a longer period of time. The 4-week duration was necessary in order to incorporate sufficient time for recruiting an adequate number of participants given the con straints of the university’s 11-week quarter system. Given that recruitment was estimated to take 5 weeks and began on the second week of the quarter, a practice period of 4 weeks was necessary. No longterm follow up was conducted, rendering us further

unable to draw conclusions about whether habit formation would have been achieved. These limita tions suggest caution in extrapolating the findings of this study to habit formation. Future studies may benefit from expanding the practice period duration and conducting long-term follow-ups, as well as assessing the duration and consistency necessary to accomplish habit formation of mind fulness meditation. Fourth, the study was unable to examine practice data for participants who did not return to posttesting. Participants who were more engaged with their daily practice might have been more likely to return to posttesting, leading to potentially inflated effects. Future research may consider using digital methods of behavior tracking as to eliminate the need for participants to return to the lab to collect measures of behavior frequency. Fifth, the study only examined participants who had no prior experience with mindfulness or medita tion, limiting the generalizability of this sample to those who are just beginning their mindfulness practice. Future research could look at the effects of action and coping plans on participants who have prior experience with meditation practice. Sixth, the study relied upon data from participant self-report. Non-self-report measures are needed to most accurately measure the effect of self-regulation interventions on meditation practice. Seventh, participants were asked to record their daily practice on a piece of paper we referred to as the daily tracking sheet. It is unlikely that participants would have carried their tracking sheet with them throughout the 4-week practice period, presenting barriers to immediate and accurate self-reporting. Instead, participants may have relied on memory recall to report their meditation practice, increasing the likelihood of error. Future studies may benefit from using digital tracking methods that can utilize reminders to increase the accuracy of self-report behavioral measures. Eighth, participant attrition at posttesting reduced the achieved power of the study, suggesting caution in interpreting the results. Replications with a larger sample size are neces sary before results can be considered conclusive. Last, demographic information on participants was not analyzed, limiting interpretations of the generalizability of the results, as well as examina tions of variance in results among age, sex, and race/ethnicity, and other demographic variables. Although demographic data was collected by the department managing the participant pool, errors in communication resulted in the deletion of the demographic data before it could be shared with the authors of this study.

Cloughesy, Mrazek, Mrazek, and Schooler | Planning to Practice

Concluding Comments Mindfulness meditation holds enormous potential to transform lives. Still, no amount of transforma tion can occur without facing the reality of practice. As a whole, this study highlights the promises of using action plans and coping plans to help naïve mindfulness practitioners develop a long-term meditation practice. The self-regulatory interven tion assessed here increased the overall frequency of days of meditation practiced. However, the intervention was unable to increase habit strength over a 4-week period, demonstrating the persistent challenge of facilitating a long-term practice. As mindfulness continues to expand into educational, occupational, and therapeutic contexts, developing self-regulatory strategies that facilitate a sustainable practice remains more important than ever.

References Adams, Z. W., Sieverdes, J. C., Brunner-Jackson, B., Mueller, M., Chandler, J., Diaz, V., . . . Treiber, F. A. (2018). Meditation smartphone application effects on prehypertensive adults’ blood pressure: Dose-response feasibility trial. Health Psychology, 37, 850–860. https://dx.doi.org/10.1037/hea0000584 Armitage, C. J. (2005). Can the theory of planned behavior predict the maintenance of physical activity? Health Psychology, 24, 235–245. https://doi.org/10.1037/0278-6133.24.3.235 Carmody, J., & Baer, R. A. (2008). Relationships between mindfulness practice and levels of mindfulness, medical and psychological symptoms and well-being in a mindfulness-based stress reduction program. Journal of Behavioral Medicine, 31, 23–33. https://doi.org/10.1007/s10865-007-9130-7 Clarke, T. C., Barnes, P. M., Black, L. I., Stussman, B. J., & Nahin, R. L. (2018). Use of yoga, meditation, and chiropractors among U.S. adults aged 18 and over. (NCHS data brief, No. 325). Hyattsville, MD: National Center for Health Statistics. Duhigg, C. (2012). The power of habits: Why we do what do in life and business. New York, NY: Random House. Galla, B. M., & Duckworth, A. L. (2015). More than resisting temptation: Beneficial habits mediate the relationship between self-control and positive life outcomes. Journal of Personality and Social Psychology, 109, 508–525. https://dx.doi.org/10.1037/pspp0000026 Gardner, B. (2015). A review and analysis of the use of ‘habit’ in understanding, predicting, and influencing, health-related behaviour. Health Psychology Review, 9, 277–295. https://doi.org/10.1080/17437199.2013.876238 Gardner, B., Lally, P., & Wardle, J. (2012). Making health habitual: The psychology of ‘habit formation’ and general practice. British Journal of General Practice, 62, 664–666. https://doi.org/10.3399/bjgp12X659466 Gollwitzer, P. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist, 54, 493–503. https://doi.org/10.1037/0003-066X.54.7.493 Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta‐analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119. https://doi.org/10.1016/S0065-2601(06)38002-1 Hagger, M. S., & Luszczynska, A. (2014). Implementation intention and action planning interventions in health contexts: State of the research and proposals for the way forward. Applied Psychology: Health and Well-Being, 6, 1–47. https://doi.org/10.1111/aphw.12017 Huppert, F. A., & Johnson, D. M. (2010). A controlled trial of mindfulness training in schools: The importance of practice for an impact on well-being. The Journal of Positive Psychology, 5, 264–274. https://doi.org/10.1080/17439761003794148 Kwasnicka, D., Presseau, J., White, M., & Sniehotta, F. F. (2013). Does planning

how to cope with anticipated barriers facilitate health-related behaviour change? A systematic review. Health Psychology Review, 7, 129–145. https://doi.org/10.1080/17437199.2013.766832 Lally, P., Chipperfield, A., & Wardle, J. (2008). Healthy habits: Efficacy of simple advice on weight control based on a habit-formation model. International Journal of Obesity, 32, 700–707. https://doi.org/10.1038/sj.ijo.0803771 Lally, P., van Jaarsveld, C. H. M., Potts, H. M. M., & Wardle, J. (2009). How are habits formed: Modeling habit formation in the real world. European Journal of Social Psychology, 40, 998–1009. https://doi.org/10.1002/ejsp.674 Mrazek, A. J., Mrazek, M. D, Cherolini, C. M, Cloughesy, J. N., Cynman, D. J., Gougis, L. J., . . . Schooler, J. W. (2018). The future of mindfulness training is digital, and the future is now. Current Opinion in Psychology, 28, 81–86. https://doi.org/10.1016/j.copsyc.2018.11.012 Parks–Stamm, E. J., Gollwitzer, P. M., & Oettingen, G. (2007). Action control by implementation intentions: Effective cue detection and efficient response initiation. Social Cognition, 25, 248–266. https://doi.org/10.1521/soco.2007.25.2.248 Quach, D., Gibler, R. C., & Jastrowski Mano, K. E. (2017). Does home practice compliance make a difference in the effectiveness of mindfulness interventions for adolescents? Mindfulness, 8, 495–504. https://doi.org/10.1007/s12671-016-0624-7 Rosenzweig, S., Greeson, J. M., Reibel, D. K., Green, J. S., Jasser, S. A., & Beasley, D. (2010). Mindfulness-based stress reduction for chronic pain conditions: Variation in treatment outcomes and role of home meditation practice. Journal of Psychosomatic Research, 68, 29–36. https://doi.org/10.1016/j.jpsychores.2009.03.010 Sheeran, P., Webb, T. L., & Gollwitzer, P. M. (2005). The interplay between goal intentions and implementation intentions. Personality and Social Psychology Bulletin, 31, 87–98. https://doi.org/10.1177/0146167204271308 Sniehotta, F. F., Scholz, U., & Schwarzer, R. (2006). Action plans and coping plans for physical exercise: A longitudinal intervention study in cardiac rehabilitation. British Journal of Health Psychology, 11, 23–37. https://doi.org/10.1348/135910705X43804 Verplanken, B. (2006). Beyond frequency: Habit as a mental construct. British Journal of Social Psychology, 45, 639–656. https://doi.org/10.1348/014466605X49122 Verplanken, B., & Orbell, S. (2006). Reflection on past behavior: A self-report index of habit strength. Journal of Applied Social Psychology, 33, 1313–1330. https://doi.org/10.1111/j.1559-1816.2003.tb01951.x Vettese, L. C., Toneatto, T., Stea, J. N. Nguyen, L., & Wang, J. J. (2009). Do mindfulness meditation participants do their homework? And does it make a difference? A review of the empirical evidence. Journal of Cognitive Psychotherapy, 23, 198–225. http://dx.doi.org/10.1891/0889-8391.23.3.198 Wood, W., Quinn, J. M., & Kashy, D. A. (2002). Habits in everyday life: Thought, emotion, and action. Journal of Personality and Social Psychology, 83, 1281–1297. https://doi.org/10.1037/0022-3514.83.6.1281 Author Note. Jonathan N. Cloughesy, https://orcid.org/0000-0002-6531-038X, Department of Psychological and Brain Sciences, University of California Santa Barbara; Alissa J. Mrazek, https://orcid.org/0000-0003-3178-1789, Center for Mindfulness and Human Potential, University of California Santa Barbara; Michael D. Mrazek, Center for Mindfulness and Human Potential, University of California Santa Barbara; Jonathan W. Schooler, Department of Psychological and Brain Sciences, University of California Santa Barbara. Jonathan N. Cloughesy is now at the Center for Advanced Hindsight at Duke University, Durham NC. This study was supported by the UCSB Undergraduate Research and Creative Activities Grant. Correspondence concerning this article should be addressed to Jonathan N. Cloughesy, Center for Advanced Hindsight, Duke University, Durham, NC, 27701. E-mail: jcloughesy@gmail.com SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

209

Earn Your Master of Science in Experimental Psychology where comprehensive skills in scientific inquiry and research methodology will give you that NSU edge.

nova.edu/dra SUMMER 2020 PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH

210

Nova Southeastern University is accredited by the Southern Association of Colleges and Schools Commission on Colleges to award associate’s, baccalaureate, master’s, educational specialist, doctorate, and professional degrees. Contact the Commission on Colleges at 1866 Southern Lane, Decatur, Georgia 30033-4097 or call 404-679-4500 for questions about the accreditation of Nova Southeastern University. n Nova Southeastern University admits students of any race, color, sexual orientation, gender, gender identity, military service, veteran status, and national or ethnic origin. 05-041-20RNK

211

Publish Your Research in Psi Chi Journal Undergraduate, graduate, and faculty submissions are welcome year round. Only one author (either first author or coauthor) is required to be a Psi Chi member. All submissions are free. Reasons to submit include • • • •

a unique, doctoral-level, peer-review process indexing in PsycINFO, EBSCO, and Crossref databases free access of all articles at psichi.org our efficient online submissions portal

View Submission Guidelines and submit your research at www.psichi.org/?page=JN_Submissions

Become a Journal Reviewer Doctoral-level faculty in psychology and related fields who are passionate about educating others on conducting and reporting quality empirical research are invited become reviewers for Psi Chi Journal. Our editorial team is uniquely dedicated to mentorship and promoting professional development of our authors—Please join us! To become a reviewer, visit www.psichi.org/page/JN_BecomeAReviewer

Resources for Student Research Looking for solid examples of student manuscripts and educational editorials about conducting psychological research? Download as many free articles to share in your classrooms as you would like. Search past issues, or articles by subject area or author at www.psichi.org/journal_past

Add Our Journal to Your Library Ask your librarian to store Psi Chi Journal issues in a database at your local institution. Librarians may also e-mail to request notifications when new issues are released. Contact PsiChiJournal@psichi.org for more information.