Replication: Influence of Context on Emotion Perception in Facial Expressions
Danica A. Barron and Ralph G. Hale* Department of Psychological Science, University of North Georgia
ABSTRACT
Accurate facial emotion perception during social interactions facilitates personal wellbeing and improves social skills such as empathy and social appraisal (Manstead et al., 1999). In the present study, we attempted to replicate this effect. Participants were presented with an image depicting a face that has been identified by Matsumoto and Ekman (1988) as a universal facial expression of “fear.” The expression is characterized by distinct features which collectively signal a state of apprehension or distress across diverse cultural contexts. Then participants were randomly assigned to 1 of 3 conditions: congruent, incongruent, or no context scenarios. The congruent context scenario suggested the face expressed fear whereas the incongruent scenario suggested the face expressed anger. Participants selected the emotion they felt was expressed by the face in the picture. We expected the congruent and nocontext groups to choose fear as the expression more often than the incongruent group. A 1way ANOVA between the 3 groups found a significant effect, F(2, 42) = 38.50, p < .001, ηp2 = .64. Congruent and no context groups chose fear as the facial expression significantly more often, thereby confirming our hypothesis and replicating previous findings. The effect of context on emotion perception is strong and replicable, suggesting that context plays an important role in emotion perception despite the universality of facial expressions.
Keywords: context, facial expression, emotion perception
The study of emotion perception, particularly in the context of facial expressions, has long intrigued researchers in the field of cognitive psychology. Building upon the seminal work in the domain, such as Carroll and Russell (1996), we sought to contribute to the understanding of how contextual cues influence the interpretation of facial expressions. Although previous literature has explored other theoretical perspectives on facial expressions and cues to emotional perception, there remains a gap in understanding the underlying processes. Although the context effect in language, visual
cognition, and processing ambiguous stimuli has provided insights into the complexities of human perception, the application of these principles in the domain of emotion perception warrants further investigation.
The process of emotion perception plays a critical role in how humans interact and communicate. Martinez (2012) suggested facial configurations are primary points of reference that aid in the interpretation of the internal status of others. The ability to accurately interpret facial expressions facilitates the understanding of the intentions and feelings. This information, however, is not perceived
Replication badge earned for transparent research practices.
in isolation and is embedded within a contextual environment that profoundly impacts its interpretation (Barrett, 2020). Evidence revealing that contextual information modulates the recognition of facial emotion provides valuable insight into social cognition, empathy, and the broad landscape of human emotional understanding.
The replication conducted in this study holds significant value in the field of emotion perception and cognitive psychology. By replication of Carroll and Russell’s (1996) work, we aimed to reinforce the reliability and generalizability of their findings in a contemporary context. Replication studies are essential for previous findings and validating theoretical frameworks. Building on the foundation laid by Carrol and Russell, we sought to extend the understanding of emotion perception through an examination of the mechanisms involved in facial expressions and their interpretation.
Facial Expressions
Facial expressions and their interpretation by others are complex. Expression involves intricately connected facial muscle movements which are unconsciously generated (Manstead et al., 1999). Darwinian perspective hypothesizes that these movements express a variety of human emotions. The behavioral ecology perspective offers a similar view, suggesting these movements are a signal instead of a direct expression of the individual’s intentions. Both theories are grounded in evolution and convey the importance of facial expression as a humanistic tool of displaying social motive and intention. The complex nature of facial expressions sets the stage for understanding the nuances involved in emotion perception, having a broader spectrum of cues and contextual influences.
Emotion Perception
To assess someone’s internal state, many cues are used including situational surroundings, body posture cues, and personal biases of the interpreter (Rajhans et al., 2016). Barrett and Bar (2009) found individual affect to be an unconscious influence on the interpretation of the emotional state of others. Barrett described this unconscious influence as an affective context effect. Previous research has demonstrated that contextual information can modulate facial emotion perception (e.g., Lee et al., 2012). Emotional information triggering advanced neural connections between amygdaloid areas and facial stimuli processing centers, such as the fusiform gyrus and superior temporal sulcus, could serve as a possible mechanism to explain this phenomenon. This enhanced connection is indicative of an evolutionary threat detection system (Mobbs et al., 2007).
The stimulus used as the facial expression for our study provided by Matsumoto and Ekman (1988) is
featured in their seminal work in facial expression research. Their study investigated the identification and interpretation of facial expressions across cultures and lays the groundwork for subsequent research in the field of emotion perception. In their examination of the six primary emotions (i.e., happiness, sadness, surprise, fear, disgust, and anger), they aimed to validate the universality of facial expressions, challenging the previous assumptions that emotional expressions were culturally determined. The Matsumoto and Ekman stimulus was comprised of carefully selected facial expressions depicting these universal emotions and has since become a cornerstone in emotion research, serving as a standard tool for investigating emotion perception, expression, and recognition across populations and contexts. Understanding emotion perception may inform how contextual factors contribute to the broad phenomenon of the context effect.
The Context Effect
The context effect is a heavily studied cognitive effect, and is a perspective that belongs to constructivist theory (Greening, 1998). This theory posits that individuals perceive reality through conceptualization of sensory stimuli, which are then used to make predictions. This conceptualization process is facilitated by top down processing and, as explained by Altman (2002), is akin to the concepts formed around phonemes that create each word (e.g., including the features and organization of letters in written speech). These topdown conceptualizations allow for the momenttomoment prediction humans use to navigate and make sense of the environment. They are learned experiences and statistical regularities, formed out of sensory information received by bottomup sensory organs that received stimulation from the environment (Panichello et al., 2013). For example, consider the human brain trying to make sense of an apple. Red light and particular line orientations reflect off the surface and reach the retina. Organic compounds reach the nose and mouth. A rounded, solid shape with specific texture and weight touches the hand that holds it. The brain combines this multimodal information into a unified interpretation that is perceived and experienced as an apple. Ngo (2015) highlighted that the unification of this sensory information may be a relatively quick process for familiar stimuli, or a slow process for more ambiguous stimuli.
In exploring the context effect, it becomes evident that semantic context plays a crucial role in shaping perceptions and interpretations of emotional stimuli. This is especially true because the semantic context effect encompasses the influence of linguistic and conceptual information on the emotion being perceived. This influence emerges as a central focus of our investigation. As individuals navigate social
interactions, the semantic context surrounding these expressions serves as a powerful cue, guiding the understanding of the other’s emotional state. We hypothesized that semantic context would significantly modulate the interpretation of facial expressions, ultimately shaping an emotional understanding and response. The significance of the context effect is influenced by adaptation to a stimulus, priming effects, and cues. Adaptation to a stimulus refers to the process of sensory systems adjusting to sensory information over time. The more a person encounters a stimulus, the less sensitive and responsive they become to it. This plays a crucial role in how context can elicit differences in perception by shifting focus from repetitive stimuli to detecting changes in the environment (Werner et al., 2005). Priming cues are the exposure to a stimulus that influences the processing of other, later information in the environment. This phenomenon may impact aspects of perception, memory, and decisionmaking. Cues trigger mental associations or concepts that allow memory to interact with perception (Schacter et al., 2004). For example, exposure to an advertisement about a nearby beach may impact the processing of later beachrelated stimuli. The interaction between environmental cues and cognitive processing becomes particularly evident in the perception of ambiguous stimuli.
Ambiguous Stimuli
Ambiguous stimuli are subject to perceptual cycling and mutual exclusivity. Perceptual cycling describes the patterned saccadic eye movements that alternate between two perceptual sets or representations of the brain’s prediction for the sensory information being received from the environment (VanRullen, 2016). Mutual exclusivity is a phenomenon in which sensory information offers conflicting evidence of the representation, thus affecting the stability of the interpretation (Klink et al., 2012). These two distinct features are characteristic of ambiguous stimuli. For example, the Necker cube is a twodimensional representation of a cube that appears as a three dimensional cube in one of two configurations (i.e., pointing up and to the left or pointing down and to the right). Context (e.g., past experience with depth perception, typical view angles for cubes, practice with Necker cubes) can help disambiguate the orientation; however, observers are typically still able to switch orientations exogenously or endogenously by intentionally switching the orientation or allowing for incidental orientation shifts, respectively. The complexities inherent in interpreting ambiguous stimuli parallel the contextual influences that shape the perception of ambiguous emotional stimuli.
Ambiguous Emotion Perception
Emotion perception allows the same ambiguity criteria. As mentioned before, every emotional expression is not completely diagnostic of the internal state of another. Facial expressions or emotional cues are unclear and open to multiple interpretations. Mixed or dimorphous expressions may further complicate perception. Dimorphous expressions result from a combination of elements from two conflicting emotions (Aragón, 2017). For example, when someone experiences intense joy and starts to cry, they are exhibiting a dimorphous expression. There are additional reasons why emotion perception may be ambiguous, including subtle expressions, cultural differences, individual differences, and contextual influences (Biehl et al., 1997; Martin et al.,1996). These contextual ambiguities may elicit negative social outcomes (e.g., miscommunications, misunderstandings, and difficulties empathizing). Additionally, this effect has been shown to be heightened for those diagnosed with autism spectrum disorder (ASD). Stagg (2021) suggested that ASD may be reflective of the difficulty of retrieving and utilizing previously learned concepts and applying them to present perceptions. This may indicate a conceptual network problem for individuals with ASD. Building on the challenges posed by ambiguous emotion perception, the we sought to dive deeper into the interaction between semantic context and facial expression perception.
The Present Study
This replication study delved into the fascinating interplay between semantic context and facial emotion perception. The present study was identical to the original study by Carroll and Russell (1996), including stimuli, measures, manipulation, and analysis procedures. We hypothesized there would be a significant influence of semantics, including the meaning derived from the facial expression and the context provided via written scenarios, on facial expression perception.
Method
Participants
A total of 45 participants completed our study. Participants were recruited from the university’s Electronic Sona/NERD research participant pool. A power analysis conducted using G*Power version 3.1.9.4 (Faul et al., 2007) to determine the required sample size for our one way ANOVA between subjects design. We aimed for a significance level (α) of .05, a desired power of .80, and assumed a large effect size based on previous literature (ηp2 ≥ .14). A large effect was assumed based on previous research into this phenomenon (e.g., ηp2 = .30 in Barrett & Kensinger, 2010, and ηp2 = .33 in
WINTER 2024
Ngo & Isaacowitz, 2015). The power analysis indicated that a minimum sample size of 42 to 66 participants was needed to detect the anticipated effects. All participants were at least 18 years of age or older at the time of participation and provided informed consent before participating. Demographic data—including gender, race, and ethnicity—was not collected. We recognize this is a limitation of this study, and future research will include demographic measurements to promote equity and representation in research (see Roberts et al., 2020). Participants were granted partial course credit as compensation for their participation. This study adhered to the ethical standards and practices outlined by the Declaration of Helsinki and was approved by the university’s Institutional Review Board.
Measures
The study was created using Qualtrics and was administered to participants online through this platform. The Qualtrics study included the consent form, instructions, experiment, and debrief. For the experiment, a single stimulus was used (see Figure 1). The facial expression stimulus was identical to the stimulus used by Matsumoto and Ekman (1988) and Carroll and Russell (1996). The image is a headshot of a woman exhibiting the universal facial expression for fear, as detailed by Tomkins (1992). There were three between subjects conditions in this study: no scenario (n = 15), congruent scenario (n = 15), and incongruent scenario (n = 15). The congruent and incongruent scenarios are identical to those used in Carroll and Russell (1996). See Table 1 to read these scenarios. The congruent scenario suggests the correct interpretation of the facial expression should be fear, whereas the incongruent scenario suggests anger is the correct interpretation of the stimulus.
The emotion selection measure was a fouralternative forced choice response. For the noscenario and congruent scenario conditions, the response items included fear, anger, excitement, and disgust. In the incongruent scenario condition, excitement was replaced with sadness; all others were the same.1 Response items were simplified from the item sets used by Carroll and Russell (1996). The original study also tested for situational dominance unlike the present study; therefore, fewer response items were required here. The simplified set of response items used in the present study was used in the original study. These response items are also consistent with Imbir (2017)
1As a replication study, we used response items from the original set of items in Carroll & Russell (1996). However, future research should use a list of common response items between all groups to avoid design confounds. It is worth noting that no participants selected “excitement” or “sadness” in this experiment, the only two items that differ between conditions.
eight sectors of primary emotions. This model categorizes each emotion, providing a detailed representation of distinct facets of human emotional experience and an overview of the emotional spectrum. This model also provides researchers with a structured framework for understanding and classifying emotional responses. For congruent and incongruent conditions, response items were randomized between participants to avoid order effects. Barron and
Facial Expression Stimulus
Note. This facial expression is commonly associated with fear (Tompkins, 1992; Biehl et al., 1997). However, it can become ambiguous with the addition of scenario-based context. This image is reproduced with permission from Matsumoto and Ekman (1988).
TABLE 1
Congruent Versus Incongruent Scenarios
From Carroll & Russell, 1996
Congruent context scenario
“This is a story about a woman who had never done anything really exciting in her life. One day she decided she had to do something exciting, so she enrolled in a class for parachuting. Today is the day that she will make her first jump. She and her class are seated in the plane as it reaches the right altitude for parachute jumping. The instructor calls her name. It is her turn to jump. She refuses to leave her seat.”
Incongruent context scenario
“This is a story of a woman who wanted to treat her sister to the most expensive, exclusive restaurant in their city. Months ahead, she made a reservation. When she and her sister arrived at the restaurant, they were told by the host that their table would be ready in 45 minutes. Still, an hour passed, and no table. Other groups arrived and were seated after a short wait. The woman went to the host and reminded him of her reservation. He said he would do his best. Ten minutes later, a local celebrity and his date arrived and were immediately shown a table. Another couple arrived and were seated immediately. The woman again went to the host, who said that all the tables were now full, and that it might be another hour before anything was available.”
FIGURE 1
Procedure
To begin this experiment, participants provided informed consent via Qualtrics. Then age data was collected to confirm they were at least 18 years old. Participants were then randomly assigned to one of three conditions: no scenario, congruent scenario, and incongruent scenario. Following randomization, all participants received instructions for the experiment. The instructions stated they would be shown an image and be asked to make a judgement based on this image. The two scenario groups (congruent and incongruent) also received instructions stating they would read a scenario
to help them make their judgement. Participants confirmed they had read the instructions. Then all groups were shown the face stimulus (see Figure 1). The scenario groups were also shown their scenario. The scenario and face were present simultaneously for these two groups. Participants had an unlimited amount of time to make their judgement. The judgement item asked, “What is the woman feeling?” To respond, participants selected one of four emotions from a dropdown menu. Only a single trial was conducted. Then, participants read a debriefing script which explained the purpose of the study and provided them with researcher contact information in case they had any questions.
Results
A one way ANOVA found a significant effect of Condition, F (2, 42) = 38.50, p < .001, η p 2 = .64. All participants in the no scenario ( M = 1.0, SD = 0.0) and congruent scenario (M = 1.0, SD = 0.0) conditions selected fear as the emotion expressed. However, only 23.5% of the incongruent group (M = 0.24, SD = 0.29) selected fear (see Figure 2). Pairwise comparisons found both noscenario and congruent scenario to be significantly different from incongruent scenario, ps < .001. These findings suggest that contextual cues significantly influenced participants interpretation of facial expressions, supporting the hypothesis of contextual biasing on emotion perception. The effect size, which falls into the range of large significance, indicates a substantial impact of the experimental manipulation on participant responses.
Discussion
Distribution of Accuracy
Note.
The purpose of this study was to establish the replicability of contextual biasing on facial expression perception. We conducted a simplified replication of Carroll and Russell’s (1996) study, in which a face showing a universal expression of fear was presented to participants accompanied with no context, congruent context, or incongruent context depending on condition. The context effectively primed participants to adjust their expectations—and therefore their perceptions—for the emotion portrayed in the facial expression. This suggests that the mechanisms responsible for correctly interpreting emotions from facial expressions relies not only on characteristics and patterns of facial features, but also on situational factors leading up to the generation of a particular facial expression. These situational factors were provided in the form of scenarios.
The congruent scenario primed participants to believe the person pictured was afraid. This matched the universal facial expression portrayed in the photograph. However, the incongruent scenario primed
FIGURE 3
FIGURE 2
Facial Expression Stimulus
Note. For the three conditions (no-context, congruent, and incongruent) of this experiment, the percentage of fear responses are shown. For the incongruent condition, most participants did not select fear. See Figure 3 for breakdown of incongruent participant responses.
participants to believe the person pictured was angry. As seen in Figures 2 and 3, this context and priming was effective. Despite the face exhibiting the universal expression for fear, only 4 out of 15 participants (26.67%) in the incongruent condition rated the expression as fear. This is compared to 10 out of 15 participants (66.67%) rating the expression as anger. Alternative explanations to these results may include attention and demand characteristics. Participants’ individual differences in attentional capacity, cognitive biases, or momentary distractions may have influences responses. For instance, participants who were more attentive or had a heightened sensitivity to facial cues might have been more inclined to accurately perceive the emotions. Furthermore, demand characteristics, which refer to the participants’ tendency to alter their behavior or responses based on the perceptions of others, should also be considered. Even though the participants were unaware of the specific hypotheses, they might have unconsciously adjusted their responses to align with perceived expectations or to comply with what they believed researchers wanted to observe. Although their lack of awareness reduces the likelihood of these variations, subtle influences on attention and behavior cannot entirely be ruled out. These findings replicate findings from the previous work of Carroll and Russell (1996), suggesting that facial expression emotion perception is more complex than simply interpreting universal facial expressions based on facial features alone.
Applications
As context is naturally elucidating, it makes sense that it would significantly influence how individuals interpret facial emotion cues. This holds implications beyond the scope of the present study. Bottomup input is rarely processed in isolation without influence from topdown information, and the perceptual processing system develops to establish which aspects of stimuli are most important for focusing selective attention (Barrett, 2020).
The significance of the context effect can be applied to many social interactions (Manstead et al.,1999). Heightened awareness of the interplay between context and socializing could enhance effective communication. For instance, teaching individuals to recognize and consider context when interpreting facial expressions might help them navigate interpersonal relationships more effectively, thereby demonstrating a potential benefit from the implementation of socialemotional learning in childhood education (West & Turner, 2011). The connection between context and emotion perception could also guide the development of artificial intelligence systems, leading them to better mimic human emotional
understanding (Shin, 2021). Additionally, these connections may be useful in clinical settings to treat conditions such as social anxiety and autism spectrum disorder (Stagg et al., 2021). Further, accurate perception of emotions can benefit law enforcement and security personnel in assessing individuals’ emotional states accurately in interviews, interrogations, and airport security checks (Barrett et al., 2011). Together, it is clear the applications of context and facial expression perception interactions are ubiquitous and impactful.
Future Directions
The present study was limited in its scope and design due to a few factors. Importantly, the data was collected online with limited control of participant experience. For instance, we could control what is displayed on Qualtrics, but we could not ensure proper eye gaze or attention during instructions, stimulus presentation, or scenarios. We did replicate previous findings; however, future research should be conducted in person to increase internal validity of the study.
Additionally, this study could be expanded in terms of sample size and sample diversity. A larger, more diverse sample size could improve generalizability beyond a college student population. As mentioned previously, collecting demographic data—including race, ethnicity, and gender—would be useful in determining whether this effect is variable between demographic factors. If not, it would speak to the universality of the impact of context on facial expression perception. This would be significant to establish as the universality of facial expression perception without context has already been established, as described previously.
The design could also be expanded to include a larger variety of faces and expressions. The present study was a simplified replication. However, the original study could be expanded further to include a range of faces from various races, ethnicities, genders, and ages; a wider range of universal facial expressions could be used; and more variety could be included in the scenarios leading to the congruent and incongruent conditions. This increased design variety would provide a more nuanced understanding of the interplay between context and expression perception.
The present study only utilized vision to examine the connection between context and emotion perception. However, future research could investigate the use of context on emotion perception using other sensory systems such as audition. For instance, a scream may be a universal cue for fear, but context could indicate elation. Perhaps other sensory systems could be explored as well. Looking further into the crossmodal effect of context may better inform how this effect impacts
emotion perception abstractly as a global mechanism. Interestingly, this could be explored for typically developing populations and those with neurological developmental delays and disorders as well.
References
Altmann, G. M. (2002). An interactive activation Model of context effects letter perception. In Psycholinguistics: Critical concepts in psychology (pp. 422–450). Routledge.
Aragón, O. R. (2017). “Tears of joy” and “Tears and joy?” personal accounts of dimorphous and mixed expressions of emotion. Motivation and Emotion, 41(3), 370–392. https://doi.org/10.1007/s11031-017-9606-x
Barret, L. F. (2020). How emotions are made: The secret life of the brain. Picador. Barrett, L. F., & Bar, M. (2009). See it with feeling: Affective predictions during Object Perception. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1521), 1325–1334. https://doi.org/10.1098/rstb.2008.0312
Barrett, L. F., & Kensinger, E. A. (2010). Context is routinely encoded during emotion perception. Psychological Science, 21(4), 595–599. https://doi.org/10.1177/0956797610363547
Barrett, L. F., Mesquita, B., & Gendron, M. (2011). Context in emotion perception. Current Directions in Psychological Science, 20(5), 286–290. https://doi.org/10.1177/0963721411422522
Biehl, M., Matsumoto, D., Ekman, P., Hearn, V., Heider, K., Kudoh, T., & Ton, V. (1997). Matsumoto and Ekman’s Japanese and Caucasian facial expressions of emotion (JACFEE): Reliability data and cross-national differences. Journal of Nonverbal Behavior, 21(1), 3–21. https://doi.org/10.1023/a:1024902500935
Carroll, J., & Russell, J. (1996). Do facial expressions signal specific emotions? Judging emotion from the face in context. Journal of Personality and Social Psychology, 70(2), 205–218. https://doi.org/10.1037/0022-3514.70.2.205
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
Greening, T. (1998). Building the constructivist toolbox: An exploration of cognitive technologies. Educational Technology, 38(2), 22–35. https://www.jstor.org/stable/44428434
Imbir, K. K. (2017). Psychoevolutionary theory of emotion (Plutchik). Encyclopedia of personality and individual differences, 1–9. https://doi.org/10.1007/978-3-319-28099-8_547-1
Klink, P. C., Van Wezel, R. J., & van Ee, R. (2012). United we sense, divided we fail: Context-driven perception of ambiguous visual stimuli. Philosophical Transactions of the Royal Society B: Biological Sciences, 367(1591), 932–941. https://doi.org/10.1098/rstb.2011.0358
Lee, T.-H., Choi, J.-S., & Cho, Y. S. (2012). Context modulation of facial emotion perception differed by individual difference. PLoS ONE, 7(3). https://doi.org/10.1371/journal.pone.0032987
sensitivity. Journal of Research in Personality, 30(2), 290–305. https://doi.org/10.1006/jrpe.1996.0019
Martínez, A. M., & Du, S. (2012). A model of the perception of facial expressions of emotion by humans: Research overview and perspectives. Journal of Machine Learning Research, 13, 1589–1608.
Matsumoto, D., & Ekman, P. (1988). Japanese and Caucasian facial expressions of emotion (JACFEE). Unpublished slide set and brochure, Department of Psychology, San Francisco State University.
Mobbs, D., Petrovic, P., Marchant, J. L., Hassabis, D., Weiskopf, N., Seymour, B., Dolan, R. J., & Frith, C. D. (2007). When fear is near: Threat imminence elicits prefrontal-periaqueductal gray shifts in humans. Science, 317(5841), 1079–1083. https://doi.org/10.1126/science.1144298
Ngo, N., & Isaacowitz, D. M. (2015). Use of context in emotion perception: The role of top-down control, cue type, and perceiver’s age. Emotion, 15(3), 292–302. https://doi.org/10.1037/emo0000062
Panichello, M. F., Cheung, O. S., & Bar, M. (2013). Predictive feedback and conscious visual experience. Frontiers in Psychology, 3, 1–8. https://doi.org/10.3389/fpsyg.2012.00620
Rajhans, P., Jessen, S., Missana, M., & Grossmann, T. (2016). Putting the face in context: Body expressions impact facial emotion processing in human infants. Developmental Cognitive Neuroscience, 19, 115–121. https://doi.org/10.1016/j.dcn.2016.01.004
Roberts, S. O., Bareket-Shavit, C., Dollins, F. A., Goldie, P. D., & Mortenson, E. (2020). Racial inequality in psychological research: Trends of the past and recommendations for the future. Perspectives on Psychological Science, 15(6), 1295–1309. https://doi.org/10.1177/1745691620927709
Schacter, D., Dobbins, I. & Schnyer, D. Specificity of priming: A cognitive neuroscience perspective. Nature Review Neuroscience 5, 853–862 (2004). https://doi.org/10.1038/nrn1534
Shin, D. (2021). The perception of humanness in Conversational journalism: An algorithmic information-processing perspective. New Media & Society, 24(12), 2680–2704. https://doi.org/10.1177/1461444821993801
Stagg, S., Tan, L.-H., & Kodakkadan, F. (2021). Emotion recognition and context in adolescents with autism spectrum disorder. Journal of Autism and Developmental Disorders, 52(9), 4129–4137. https://doi.org/10.1007/s10803-021-05292-2
Tomkins, S. S. (1992). Affect, imagery, consciousness (Vol. 1–2). Springer. Webster, M. A. VanRullen, R. (2016, August 23). Perceptual cycles. Trends in Cognitive Sciences https://www.sciencedirect.com/science/article/pii/S1364661316301048
Werner, J. S., & Field, D. J. (2005). Adaptation and the phenomenology of perception. Fitting the Mind to the World Adaptation and After-Effects in High-Level Vision, 241–278. https://doi.org/10.1093/acprof:oso/9780198529699.003.0010
West, R. L., & Turner, L. H. (2011). Understanding interpersonal communication: Making choices in changing times. Wadsworth Cengage Learning.
Author Note
Correspondence concerning this article should be addressed to Ralph G. Hale, University of North Georgia, 3820 Mundy Mill Rd, Oakwood, GA 30566. Email: ralph.hale@ung.edu Influence
Manstead, A. S. R., Fischer, A. H., & Jakobs, E. B. (1999). The social and emotional functions of facial displays. In P. Philippot, R. S. Feldman, & E. J. Coats (Eds.), The social context of nonverbal behavior (pp. 287–313). Cambridge University Press; Editions de la Maison des Sciences de l’Homme. Martin, R. A., Berry, G. E., Dobranski, T., Horne, M., & Dodgson, P. G. (1996). Emotion perception threshold: Individual differences in emotional
Ralph G. Hale https://orcid.org/0000000150268417
We have no known conflict of interest to disclose. Many thanks to Anna Adamo, Pilar AmbrizCalleros, and Kimberly Carpenter for their creative input and contributions to this project.
The Interplay of Depression, Rumination, and Negative Autobiographical Memory
Patricia M. Roberts and Jason R. Finley* Department of Psychology, Southern Illinois University Edwardsville
ABSTRACT. This experiment investigated rumination as a possible mechanism for the phenomenon of depression increasing negative autobiographical memories. Participants recalled a negative autobiographical memory involving school before rating the negative mood intensity of that memory, then half of the participants ruminated on that memory and half of the participants were distracted from it. Participants then rated the memory again, and either ruminated or were distracted for a second time before rating the memory for a third time and completing the Beck Depression InventoryII, the Rumination on Sadness Scale, and demographic questions. The negative mood intensity of the autobiographical memory decreased over time, but did so to a lesser extent when participants ruminated on it versus being distracted (interaction p < .001, d = 1.01). Furthermore, for the rumination condition, participants with greater depression scores reported negativity ratings that decreased at a slower rate over time; for the distraction condition, participants with greater depression scores reported negativity ratings that decreased at a faster rate over time (interaction p = .040, ∆R2 = .047). Depression leads to rumination, and may also amplify the effect of rumination on the negativity of autobiographical memories. The effects of rumination may be due to memory effects such as retrieval practice and mood congruency. Individuals experiencing higher levels of depressive symptom severity may be more likely to experience increasingly negative memories due to rumination.
Keywords: depression, rumination, autobiographical memory, mood congruency, negative mood intensity
Depression is a general negative affective state in which individuals can experience symptoms of unhappiness, lack of motivation, and changes in habits regarding social connections, eating, and sleeping (American Psychological Association [APA], n.d.a) Autobiographical memory includes an individual’s memory for events that they have experienced (APA, n.d.b). Depression influences autobiographical memory by biasing recall to negative events (Hitchcock et al.,
2020; Lyubomirsky & NolenHoeksema, 1995; Peeters et al., 2003; Vrijsen et al., 2001). One potential mechanism by which depression may have this effect is through rumination, which is when an individual repeatedly has similar thoughts to the point of interference with other mental activities (APA, n.d.c). Thus, for the current study, we investigated the interplay between depression, rumination, and autobiographical memory.
PSI
Depression and Autobiographical Memory
Research has shown a negativity bias in memory for simple materials in people with depression. For example, results from a study by Bianchi et al. (2020) found that a group of participants who scored low on a selfreport depression scale recalled more positive words and fewer negative words compared to a group of participants who scored high on the scale, with the latter group recalling fewer positive words and more negative words. The results were based on responses from 1,015 participants who completed a free recall task involving 10 positive and 10 negative words. It is possible that a negativity bias also extends to memories from everyday life.
There are a number of ways to investigate depression’s effects on autobiographical memory. Simply having participants attempt to recall good and bad days is one approach. For example, when participants were asked to remember both positive and negative events from their lives, Hitchcock et al. (2020) found that healthy control participants (who had never been depressed) recalled significantly more positive memories than negative memories. This was in contrast to participants with depression, for whom there was no significant difference in the number of positive memories and negative memories recalled. However, results also showed that participants with depression rated their memories less positively compared to healthy control participants.
Another approach is to use experience sampling to form records of participants’ experiences. For example, Peeters et al. (2003) conducted a study in which participants filled out self report forms reporting their current mood, negative and positive events, and appraisals of the events 10 times a day for six days. Results showed that participants with depression reported fewer positive than negative events and reported negative events as more unpleasant compared to participants without depression. Results also showed that overall, participants with depression had higher base levels of negative mood for memories compared to participants without depression, and this was particularly true for participants with more severe depression and/or who had a longer duration of depression. Studies using experience sampling have also shown that people with depression are biased to recall more negative experiences than they actually had. Urban et al. (2018) conducted a study in which 1,657 participants made daily records of emotional experiences for eight days in a row, and recalled those experiences at the end of the final day. Results showed that participants with a history of depression particularly overestimated how often they had experienced negative emotions, consistent with a negative memory bias due to depression.
Depression and Rumination
One hallmark of depression is the tendency toward repetitive thoughts, known as rumination. A variety of studies have shown such a relationship (Mitchell, 2016). For example, Harrington and Blankenship (2002) found a medium correlation of r = .33 between rumination and depression with 199 participants. NolenHoeksema (2000) interviewed 1,132 participants on two occasions, one year apart, and found that participants with depression at Time 1 experienced greater levels of rumination at Time 1 and Time 2 compared to participants without depression. Additionally, results showed that rumination at Time 1 predicted diagnostic status for depression at Time 2. The relationship between rumination and depression is further evidenced by studies that experimentally manipulate rumination. For example, Lyubomirsky and NolenHoeksema (1995) found that inducing rumination in participants with dysphoria led them to have more negative interpretations of hypothetical situations as compared to participants with dysphoria who were not induced to ruminate.
Rumination as a Mechanism Linking Depression and Negative Memories
Some studies have suggested rumination as a possible reason for depression increasing a bias towards negative memories. Research by Lyubomirsky et al. (1998) experimentally investigated the effects of distraction compared to rumination on mood and autobiographical memory in participants with and without depression. In Experiment 1, there were 72 participants who either ruminated or were distracted before recalling personal memories and rating the emotional valence of those memories. Results showed that depressed participants who ruminated became more depressed, but depressed participants who were distracted became less depressed; there was no significant difference in nondepressed participants who ruminated or were distracted. Additionally, results showed that depressed participants who ruminated rated their autobiographical memories as more negative and less positive than any other group. Park et al. (2004) found similar results with adolescents, showing that there was a greater increase in depressed mood with rumination than distraction for adolescents with depression compared to adolescents without depression.
Current Study
In summary, previous research has suggested that there is a relationship between depression and rumination such that rumination causes an increase in negative mood (Lyubomirsky et al., 1998; Lyubomirsky & NolenHoeksema, 1995; Park et al., 2004). The current
WINTER 2024
experiment sought to further explore such a relationship by measuring the effects of rumination and distraction at multiple timepoints, before and after more than one round of distraction or rumination, thus investigating how rumination changes the emotional context of an autobiographical memory, with depressive symptom severity as a moderator and general rumination tendency as a covariate.
Participants described a negative past event involving school and rated the emotional intensity of the memory three times (Time 1, Time 2, and Time 3). Half of the participants were induced to ruminate about the memory for 3 min between ratings, and the other half were distracted for 3 min between ratings. Finally, participants completed the Beck Depression InventoryII (Beck et al., 1996), the Rumination on Sadness Scale (Conway et al., 2000), and demographic questions.
We investigated three hypotheses. The first hypothesis was that there would be a twoway interaction between time and task such that by distracting from a negative autobiographical memory, the negative mood intensity of the memory would stay the same across Times 1, 2, and 3. In contrast, by ruminating on a negative autobiographical memory, the negative emotional intensity of the memory would increase across Times 1, 2, and 3. This prediction can be seen in Figure 1.
The second hypothesis was that there would be a three way interaction between time, task, and depression such that participants with greater levels of depressive symptom severity in the rumination condition would have a greater increase in negative mood intensity across time compared to participants with lower levels of depressive symptom severity in the rumination condition. That is, depression would amplify the negative effect of rumination.
The third hypothesis was that there would be a main effect of depression so that, even at Time 1, participants with greater depressive symptom severity would have higher ratings of negative mood intensity for their memory.
Method
Participants
Participants were 73 students in the introductory psychology course at Southern Illinois University Edwardsville who participated for partial fulfilment of a course requirement in February 2022. Participants completed the study online via Qualtrics, and gave consent to participate by checking a box. Data were additionally collected from 10 other participants but were excluded from analysis due to incomplete data. The mean age of the participants was 19.49 (SD = 2.06, range = 18–31), with one participant not reporting their
age. There were 61 women and 12 men. With sample size N = 73 the study obtained 80% power to detect effect sizes of size d = 0.66 or greater for betweensubjects t tests, and r = .32 or greater for correlations. This study received ethical approval from the Institutional Review Board of Southern Illinois University Edwardsville (Protocol #1488).
Design
The independent variables were time (Time 1, Time 2, and Time 3; withinsubjects) and task (distraction vs. rumination; betweensubjects). Participants were randomly assigned to either the distraction group ( n = 36) or rumination group (n = 37). The main dependent variable was the negative mood intensity for the autobiographical memory and was measured using a sliding scale with values ranging from 0–100 (0 = extremely positive, 100 = extremely negative), where higher numbers corresponded with a higher negative mood intensity. In addition, depressive symptom severity was measured as a quasiindependent variable using the Beck Depression InventoryII, and rumination frequency was measured as a potential covariate using the Rumination on Sadness Scale.
Materials
The Beck Depression InventoryII (Beck et al., 1996) is a 21item, selfreport questionnaire that was used to measure the level of participants’ depressive symptoms from within the past 2 weeks. Questions use a 4point scale, with higher values indicating greater symptom severity. Across many studies, the BDI II has been found to have high internal consistency, with a mean Cronbach’s alpha of .90 (Wang & Gorenstein, 2013). In the current study, Cronbach’s alpha was .93.
The Rumination on Sadness Scale (Conway et al.,
Note. Time periods were 3 minutes apart.
FIGURE 1
Prediction of Hypothesis 1
2000) is a 13item, selfreport questionnaire that was used to measure the extent to which participants ruminate on sad memories in general. Questions use a 5point scale (1 = not at all, 5 = very much), with higher values indicating greater rumination. Conway et al. (2000) reported a Cronbach’s alpha of .91. In the current study, Cronbach’s alpha was .93.
For both the Beck Depression InventoryII and the Rumination on Sadness Scale, we calculated each participant’s score as the mean of their responses on the questions for each scale. This was done instead of summation in order to allow for missing values.
Procedure
The experiment was completed online using the Qualtrics website after participants signed up on the Sona participant pool website. Participants were first told to describe a single clear and specific negative experience involving school, which they typed in a freeresponse textbox. They then reported how long ago the experience occurred, and rated the negative mood intensity of the experience using a sliding scale from 0 (extremely positive) to 100 (extremely negative). Participants could view the value for their rating when moving the slider. After initially describing the experience, participants were randomly assigned to either think about that negative experience for 3 minutes (rumination condition), or to think about the external stimuli of clouds forming in the sky for 3 minutes (distraction condition).1 Next, participants again rated the negative mood intensity of their past experience on a 0–100 sliding scale for a second time. Then, participants in the rumination group again ruminated on the experience for 3 minutes and participants in the distraction group again focused on the external stimuli of clouds forming in the sky for 3 minutes. The participants then rated the negative mood intensity of their past experience on a 0–100 sliding scale for a third time. Finally, all participants completed the Beck Depression InventoryII, the Rumination on Sadness Scale, and demographic questions (age and
1
distraction condition was derived from Lyubomirsky et al. (1998).
gender). As part of debriefing, participants were given a link to free counseling services at Southern Illinois University Edwardsville. See Figure 2 for a diagram of the procedure. Complete instructions are included at https://osf.io/h657r
Results
Both the Beck Depression Inventory II and the Rumination on Sadness Scale showed high reliability as measured by Cronbach’s alpha (.93 for both). Participant scores on both scales were calculated as the mean of their responses to all items. The obtained range for the Beck Depression InventoryII score was 1–3.24 (M = 1.78, SD = 0.55) with the total possible range being 1–4. The obtained range for Rumination on Sadness Scale score was 1–4.85 (M = 2.71, SD = 0.93) with the total possible range being 1–5. Data are available at https://osf.io/h657r
Hypothesis 1: Rumination Will Increase the Negative Mood
Intensity
Table 1 shows the means and standard deviations of the negativity ratings across time periods, separately for the distraction and rumination conditions. The skewness of the distributions for these ratings were 0.36, 0.10, and 0.65 for the distraction condition, and 1.29, 0.69, and 0.85 for the rumination condition, for Time 1, Time 2, and Time 3 respectively. As illustrated in Figure 3, negativity ratings appeared to decrease over time for both conditions. In order to analyze these decreases, we calculated a simple linear regression slope across Times 1–3 for each participant. 2 These slopes were used for all subsequent analyses, because the change in negativity rating over time was of key interest. Twotailed singlesample t tests compared the mean
2For example, one participant’s three negativity ratings were 76, 51, and 37. A simple linear regression for that participant yields a slope of -19.5. This slope gives a single number representing the overall change in rating across the three time periods for this participant. Having such a single score for each participant simplifies analyses, for example allowing a comparison of mean slopes across conditions using a t test, rather than requiring a 2-way ANOVA.
WINTER 2024
The task for the
FIGURE 2
Diagram Outlining the Basic Procedure
slopes of the distraction and rumination conditions to zero. For the distraction condition the mean slope was 17.07 (SD = 11.67), t(35) = 8.65, p < .001, d = 1.46. For the rumination condition the mean slope was 5.08 (SD = 11.77), t(36) = 2.59, p = .014, d = 0.43. Thus, in both conditions, negativity ratings statistically significantly decreased over time. The singlesample t tests were followed by a twotailed betweensubjects t test comparing the mean slopes between the distraction and rumination conditions (i.e., looking for a 2way interaction between time and task). The mean slope was indeed significantly different in the two conditions such that the negative mood intensity decreased more in the distraction condition than the rumination condition, t(71) = 4.31, p < .001, d = 1.01.
Hypothesis 2: Depression
Will Amplify the Effect of Rumination
First, we analyzed the overall relationship between depressive symptom severity and the negative mood intensity over time (i.e., the slope) and no correlation was found, r(71) = .03, p = .807, suggesting no overall relationship between depressive symptom severity and change in negativity rating. However, when we analyzed this correlation separately for the two conditions, an interesting pattern emerged. For the distraction condition, there was a negative correlation between depressive symptom severity and slope, r(34) = .20, p = .255; for the rumination condition, there was a positive correlation between depressive symptom severity and slope, r(35) = .29, p = .080. Although neither was statistically significant with our sample size, these smalltomedium sized correlations were in the opposite direction, suggesting that the relationship between depressive symptom severity and negativity rating over time depended on the condition (distraction vs. rumination). Thus, we next analyzed depressive symptom severity as a potential moderator using a multiple linear regression to determine if there was an interaction between the effect of depressive symptom severity and condition on negative mood intensity over time. Results of this regression are shown in Table 2, and the interaction was indeed significant. To better understand this interaction, we created a scatterplot, shown in Figure 4. The yaxis shows the direction and extent to which negativity ratings changed over time for a given participant (i.e., the participant’s slope). For the distraction condition, the higher a participant’s depressive symptom severity, the more negative their slope was (i.e., their negativity ratings decreased over time to a greater extent). For the rumination condition, the higher a participant’s depressive symptom severity, the less negative their slope was (i.e., their negativity ratings decreased over time to
a lesser extent). Simple linear regression equations for both conditions are shown in Figure 4. Although we had originally predicted that negativity ratings would increase over time in the rumination condition, in fact the ratings decreased over time for the majority of participants in both conditions. However, as seen in Figure 4, the negativity rating did increase over time for 12 participants as seen by the dots above zero
TABLE 1
Mean (SD) of Negativity Ratings by Condition and Time
Note. Ratings were made on a 0–100 sliding scale, where 0 was extremely positive and 100 was extremely negative. Time periods were 3 minutes apart.
TABLE 2
Regression of Condition on Negativity Rating Over Time With Beck Depression Inventory-II (BDI) as Moderator
Note. Outcome variable is slope of negativity rating across Times 1–3; Condition = distraction (0) vs. rumination (1); BDI = Beck Depression Inventory-II. Overall R2 = .256.
Mean Negativity Ratings by Condition and Time
Note. Error bars represent standard errors. Time periods were 3 minutes apart.
FIGURE 3
on the yaxis. Interestingly, 11 of these 12 cases were in the rumination condition. A sign test verified that this is a significant difference, z = 2.73, p = .006. These results of the sign test allude to the original hypothesis, but do not fully support it.
The Rumination on Sadness Scale was measured to see how much a participant tends to ruminate on their own. This could contribute to how the rumination manipulation affects participants, and thus be
TABLE 3
Regression of Condition on Negativity Rating Over Time With Beck Depression Inventory-II (BDI) as Moderator and Rumination on Sadness Scale (RSS) as Covariate
an additional source of variance. So, we included Rumination on Sadness Scale in an additional regression in order to see if the relationship between task, depressive symptom severity, and negativity rating over time could be clarified when variance due to rumination was explicitly modeled. The results of this regression are shown in Table 3. The regression did not provide any additional insights. A possible reason for this is that the Beck Depression InventoryII and Rumination on Sadness Scale are similar in what they measure; Conway et al. (2009) reported r(211) = .56, p < .001. This shows a large positive correlation between rumination and depressive symptom severity, which is replicated in the current study, r(71) = .68, p < .001.
Hypothesis 3: High Depression Levels Will Correlate With High Negativity Ratings
at Time 1
Finally, to test the third hypothesis, we calculated a correlation between depressive symptom severity and negativity rating at Time 1 and found that it was small and positive but not significant r(71) = .18, p = .136. Thus, there was no statistically significant difference in the initial negativity of the autobiographical memory as a function of depressive symptom severity.
Note. Outcome variable is slope of negativity rating across Times 1–3; Condition = distraction (0) vs. rumination (1); BDI = Beck Depression Inventory-II; Interaction is between Condition and Beck Depression Inventory-II Score; RSS = Rumination on Sadness Scale. Overall R2 = .267.
FIGURE 4
Scatterplot of Negativity Rating Over Time and Beck Depression Inventory-II Score by Condition
Note. Slope of zero (horizontal line) indicates no change in negativity rating over time.
Discussion
Results showed that negative mood intensity ratings of an autobiographical memory decreased across time when participants repeatedly ruminated about the memory and when they were repeatedly distracted. This pattern (see Figure 3) differed from our first hypothesis that negative ratings would increase with rumination and stay the same with distraction (see Figure 1). However, the difference in the effects of the distraction condition compared to the rumination condition was statistically significant, so that the negative memory was viewed less negatively over time for the distraction condition than the rumination condition. That is, the relative difference in the slopes was as we predicted, but the absolute direction of the slopes was not.
The results were somewhat consistent with our second hypothesis that there would be an interaction between depression, time, and task. We did find a threeway interaction (see Figure 4). However, rather than our prediction that participants with a greater level of depression would show a greater increase in negative ratings of their memory across repeated ruminations, they instead showed a diminished decrease relative to less depressed participants. Furthermore, participants with a greater level of depression in the distraction condition showed a greater decrease in negative ratings of their memory across time relative to less depressed participants. Overall, 11 of the 12 participants who showed an increase
in negative mood intensity were in the rumination condition (in line with our first hypothesis); however, placement into the rumination condition alone did not guarantee an increase in negative mood intensity.
Finally, we also hypothesized that participants with a greater level of depression would begin with a more negatively rated autobiographical memory at Time 1. A small correlation between depressive symptom severity and negativity rating at Time 1 showed that the results were in the predicted direction but not statistically significant.
The overall pattern seen in the current study was similar to that found in two previous studies. Lyubomirsky et al. (1998) found that people with dysphoria who ruminated once became more depressed, but people with dysphoria who were distracted once became less depressed. They found no significant changes in depressed mood in people without dysphoria who ruminated versus were distracted. Dysphoric rumination leads to retrieving more negative autobiographical memories and rating autobiographical memories as more negative and unhappy. Park et al. (2004) similarly found a greater increase in depressed mood with rumination versus distraction for participants with depression compared to participants without. In the current study, in the rumination condition, the higher a participant’s depressive symptom severity, the less their negativity rating decreased over time (i.e., the decline was shallower); however, in the distraction condition, the higher a participant’s depressive symptom severity, the more their negativity rating decreased over time (i.e., the decline was deeper). The results of the current study are similar to those of the previous two studies, but instead of finding an increase in negativity over time for the rumination condition, we found a diminished decrease in negativity rating over time.
Overall, evidence implicates rumination as a mechanism for the effect of depression on autobiographical memory. We offer a few possible explanations for this observed effect. First of all, studies of mood dependence and congruency in memory have shown that it is easier for someone to recall a previous episode when their current mood is congruent with their previous mood (Eich & Metcalfe, 1989). Thus, when someone is feeling sad, previous sad experiences may come more readily to mind when they try to think back. This can in turn feed into the availability heuristic, whereby things that come more readily to mind are judged as more frequent or probable (Tversky & Kahneman, 1973), leading depressed people to overestimate how often they have felt sad in the past or might again feel sad in the future. MacLeod and Campbell (1992) found that participants estimated higher probability of future negative events when they had been induced into a sad mood
and asked to retrieve unpleasant memories. Finally, rumination—that is, thinking about a sad memory over and over again—can serve to further strengthen that memory because of the wellestablished retrieval practice effect (i.e., testing effect; Roediger & Butler, 2011), such that the very act of retrieving a memory strengthens that memory, making it even more likely to be retrieved again in the future. Thus, rumination amplifies the effect of depression on autobiographical memory, and also further perpetuates depression itself. There were several strengths and limitations of the current study. One strength was that our results support findings from previous studies and were obtained using different methodology. A second strength was the use of the slider with a 0–100 scale which allowed for a greater range of results for the DV (negativity rating) than would have been obtained using a traditional 5point scale. A third strength was that the memory prompt successfully elicited specific negative memories from college students with an overall high mean negativity rating at Time 1 without a ceiling or floor effect.
One limitation of the current study was that it was completed using only a sample of college students. As such, it may be beneficial to complete a future study using a sample with a larger age range than what is generally provided by college students. Additionally, future research could investigate the extent to which antidepressants might influence the effect of rumination on negative autobiographical memories in participants with depression. Finally, a major source of variance in the current study was likely due to the variety of negative experiences that participants recalled. In order to reduce such variance, a future study could look at a negative experience designed by the researcher, or elicit negative memories with a more specific prompt.
Additionally, it may be beneficial to complete a qualitative analysis on the content of the negative autobiographical memories to investigate if the content impacts the effectiveness of distraction and rumination. Future research could also further investigate the relationship between depression and rumination in regard to how they interact with each other, for example to see if verbally describing a negative autobiographical memory or experience to an acquaintance in conversation (rather than simply thinking about it) leads to a decrease or increase in the negative mood intensity rating for that memory. Another topic of future investigation could be overgeneral autobiographical memories in the context of rumination in participants with depression, similar to studies such as one by Mitchell (2015).
The broader implication of this study is that depressed individuals may be susceptible to persistence or amplification of negative memories because of Roberts and Finley
ruminating on them. But they may be able to avoid this by distracting themselves from rumination. For example, college students who are distressed over a course experience may ruminate on their distress and thus become more distressed. However, by distracting themselves from the distress, students may be able to become less distressed and thus be able to focus more on other memories or cognitive processes. The results could also have beneficial implications for therapy and counseling as teaching patients who are in counseling or therapy to distract themselves from ruminative thoughts could be beneficial in terms of decreasing the intensity of their overall negative affect (Watkins, 2015).
References
American Psychological Association. (n.d.-a). Depression. In APA dictionary of psychology. Retrieved June 19, 2024 from https://dictionary.apa.org/depression
American Psychological Association. (n.d.-b). Autobiographical memory. In APA dictionary of psychology Retrieved June 19, 2024 from https://dictionary.apa.org/autobiographical-memory American Psychological Association. (n.d.-c). Rumination. In APA dictionary of psychology. Retrieved June 19, 2024 from https://dictionary.apa.org/rumination
Beck, A. T., Steer, R. A., & Brown, G. (1996). Beck Depression Inventory–II [Database record]. APA PsycTests. https://doi.org/10.1037/t00742-000
Bianchi, R., Laurent, E., Schonfeld, I. S., Bietti, L. M., & Mayor, E. (2020). Memory bias toward emotional information in burnout and depression. Journal of Health Psychology, 25(10–11), 1567–1575. https://doi.org/10.1177/1359105318765621
Conway, M., Csank, P. A. R., Holm, S. L., & Blake, C. K. (2000). On individual differences in rumination on sadness. Journal of Personality Assessment, 75(3), 404–425. https://doi.org/10.1207/S15327752JPA7503_04
Eich, E., & Metcalfe, J. (1989). Mood dependent memory for internal versus external events. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15(3), 443–445. https://doi.org/10.1037/0278-7393.15.3.443
Harrington, J. A., & Blankenship, V. (2002). Ruminative thoughts and their relation to depression and anxiety. Journal of Applied Social Psychology, 32(3), 465–485. https://doi.org/10.1111/j.1559-1816.2002.tb00225.x
Hitchcock, C., Newby, J., Timm, E., Howard, R. M., Golden, A.-M., Kuyken, W., & Dalgleish, T. (2020). Memory category fluency, memory specificity, and the fading affect bias for positive and negative autobiographical events: Performance on a good day-bad day task in healthy and depressed individuals. Journal of Experimental Psychology: General, 149(1), 198–206. https://doi.org/10.1037/xge0000617
Lyubomirsky, S., Caldwell, N. D., & Nolen-Hoeksema, S. (1998). Effects of ruminative and distracting responses to depressed mood on retrieval of autobiographical memories. Journal of Personality and Social Psychology, 75(1), 166–177. https://doi.org/10.1037//0022-3514.75.1.166
Lyubomirsky, S., & Nolen-Hoeksema, S. (1995). Effects of self-focused
rumination on negative thinking and interpersonal problem solving. Journal of Personality and Social Psychology, 69(1), 176–190. https://doi.org/10.1037/0022-3514.69.1.176
Nolen-Hoeksema, S. (2000). The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. Journal of Abnormal Psychology, 109(3), 504–511. https://doi.org/10.1037/0021-843X.109.3.504
MacLeod, C., & Campbell, L. (1992). Memory accessibility and probability judgments: an experimental evaluation of the availability heuristic. Journal of Personality and Social Psychology, 63(6), 890. https://doi.org/10.1037/0022-3514.63.6.890
Mitchell, A. E. P. (2015). Autobiographical memory response to a negative mood in those with/without a history of depression. Studia Psychologica, 57(3), 229–241. https://doi.org/10.21909/sp.2015.03.696
Mitchell, A. E. P. (2016). Phenomenological characteristics of autobiographical memories: Responsiveness to an induced negative mood state in those with and without a previous history of depression. Advances in Cognitive Psychology, 12(2), 105–114. https://doi.org/10.5709/acp-0190-8
Park, R. J., Goodyer, I. M., & Teasdale, J. D. (2004). Effects of induced rumination and distraction on mood and overgeneral autobiographical memory in adolescent major depressive disorder and controls. Journal of Child Psychology and Psychiatry, 45(5), 996–1006. https://doi.org/10.1111/j.1469-7610.2004.t01-1-00291.x
Peeters, F., Nicolson, N. A., Berkhof, J., Delespaul, P., & deVries, M. (2003). Effects of daily events on mood states in major depressive disorder. Journal of Abnormal Psychology, 112(2), 203–211. https://doi.org/10.1037/0021-843X.112.2.203
Roediger, H. L., & Butler, A. C. (2011). The critical role of retrieval practice in longterm retention. Trends in Cognitive Sciences, 15(1), 20–27. https://doi.org/10.1016/j.tics.2010.09.003
Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5(2), 207–232. https://doi.org/10.1016/0010-0285(73)90033-9
Urban, E. J., Charles, S. T., Levine, L. J., & Almeida, D. M. (2018). Depression history and memory bias for specific daily emotions. PLoS ONE, 13(9). https://doi.org/10.1371/journal.pone.0203574
Vrijsen, J. N., Ikani, N., Souren, P., Rinck, M., Tendolkar, I., & Schene, A. H. (2001). How context, mood, and emotional memory interact in depression: A study in everyday life. [Supplemental Material] Emotion, 23(1), 41–51) https://doi.org/10.1037/emo0000783.supp
Wang, Y. P., & Gorenstein, C. (2013). Psychometric properties of the Beck Depression Inventory-II: A comprehensive review. Brazilian Journal of Psychiatry, 35, 416–431. https://doi.org/10.1590/1516-4446-2012-1048
Watkins, E. (2015). Psychological treatment of depressive rumination. Current Opinion in Psychology, 4, 32–36. https://doi.org/10.1016/j.copsyc.2015.01.020
Author Note
Jason R. Finley https://orcid.org/0000000159218336
We have no conflicts of interest related to this project. Data are available at https://osf.io/h657r
Correspondence concerning this article should be addressed to Jason R. Finley, Psychology Department, Box 1121, Southern Illinois University Edwardsville, Edwardsville, IL 62026.
WINTER 2024
Is My Anger Justified? The Influence of Gendered Racial Identity Centrality on the Relationship Between Internalization of the Sapphire Stereotype and Disengagement Coping Among Black Women
Nailah Johnson, Makyra Farmer, and Danielle D. Dickens* Department of Psychology, Spelman College
ABSTRACT. Historically, Black women have been susceptible to the effects of race and gender related stress, such as the sexually promiscuous Jezebel or the motherly Mammy stereotypes. However, few studies have centered on the influence of the Sapphire stereotype—defined as being overbearing, emasculating, and aggressive—on Black women’s health. Due to these stereotypes, Black women may utilize different coping strategies such as disengagement coping—limiting one’s interpersonal interactions— to manage these stressors. Yet, little research has specifically examined how Black women’s gendered racial identity development relates to internalization of the Sapphire stereotype. The present study aimed to explore whether gendered racial identity centrality moderates the relationship between the Sapphire stereotype and disengagement coping among Black women. Two hundred ninetyeight Black/African American women between the ages of 18–52 completed an online survey via Qualtrics. Based on the Phenomenological Variant of Ecological Systems Theory, we hypothesized that gendered racial identity centrality will moderate the relationship between the Sapphire stereotype and disengagement coping. Consistent with the hypothesis, the results indicated that identity centrality significantly moderated the relationship between the Sapphire stereotype and the use of disengagement coping such that higher identity centrality decreased the association between the Sapphire stereotype and disengagement use (b = .51, SE = 0.09, t = 5.67, p < .001, 95% CI [.33, .69]). This research can be used to develop educational programs to promote healthy identity development among Black women.
Keywords: gendered racial identity centrality, Sapphire stereotype, gendered racialized stereotypes, disengagement coping, coping strategies
Diversity badge earned for conducting research focusing on aspects of diversity.
Historically, Black women have been—and continue to be—susceptible to the effects of race and gender related stress, considering their socially constructed identities of being Black and a woman (Greer, 2011). Lewis et al. (2016) concluded that other people often have gendered racial misconceptions about Black women’s behavior and treat them based on those beliefs. Oftentimes, Black women are labeled as the Angry Black woman, or the promiscuous Jezebel. Another lesserknown stereotype, the Sapphire, describes Black women as loud, emasculating, and aggressive (Bond et al., 2021). These stereotypes are connected to experiences of gendered racial microaggressions, defined as seemingly innocent yet harmful remarks based on one’s gender and race. Experiencing such gendered racial microaggressions has been associated with poorer mental health and coping strategies (Lewis et al., 2017).
In fact, Black women may manage stressful events regarding racism, colorism, sexism, and other forms of oppression by utilizing different coping strategies, such as religion/spirituality, social support, and disengagement coping (Hall, 2018; Holder et al., 2015; Kilgore et al., 2020). Research has shown that the use of more disengagement coping strategies, the process of limiting one’s participation in certain activities, groups, or situations, was associated with poorer mental and physical health among Black women (Lewis et al., 2017). Further research has shown that the importance of one’s race and gender identity, referred to as gendered racial identity centrality (GRIC), may positively or negatively buffer the relationship between gendered racism and coping strategies among Black women (Jones et al., 2021). Yet, minimal research has examined the significance of GRIC in minimizing the effect of internalization of specific stereotypes (e.g., the Sapphire stereotype) and the use of disengagement as a coping strategy. The purpose of the current study was to fill this research gap by understanding how GRIC moderates the relationship between internalization of the Sapphire stereotype and disengagement as a coping strategy.
In the present study, the Phenomenological Variant of Ecological Systems Theory (PVEST) was the main theoretical framework used to explore and contextualize the relationship between gendered racism and the coping strategies Black women use. This theory explains how various risk factors influence potentially stressful experiences, how individuals use various coping mechanisms, and how these mechanisms impact an individual’s behavior and health outcomes (Spencer et al., 1997). Previous research exploring racial discrimination in Black adolescents used PVEST to explain unique protective and risk factors associated with being Black (Seaton et al., 2009; Sellers et al., 2006). In one study, Sellers et al. (2006) used PVEST’s risk and resilience framework to examine psychological wellbeing with racial
discrimination as a risk factor and a stressor and racial identity as a tool for resilience and as a coping strategy. The researchers found more racial discrimination to be associated with more stress, more depressive symptoms, and lower psychological wellbeing, indicating that race is one unique factor related to stress among African American adolescents. Specifically exploring the experiences of Black women, Dickens and Chavez (2018) used PVEST to posit how race, socioeconomic status, and gender serve as risk factors that result in Black women developing reactive coping mechanisms, such as changing their behavior to fit their environment. Through conducting interviews, researchers found that Black women described changing their behavior, appearance, and perspective, also known as identity shifting, to navigate experiences of stereotyping and discrimination in the workplace (Dickens & Chavez, 2018). These results show that race, socioeconomic status, and gender act as unique risk factors that are stressors for Black women, which can lead to maladaptive coping strategies. In the current study, PVEST was used to show that experiences of gendered racialized stereotypes are risk factors that can cause stress for Black women and these stressors can influence the coping strategies that Black women decide to use, such as the use of disengagement coping strategies when faced with the stressors associated with being both Black and a woman.
Black Women’s Identity Development and Related Experiences
One factor that may buffer Black women against the harmful effects of experiences of stereotyping and discrimination is gendered racial identity centrality, which refers to how an individual’s gender and racial identity combined are directly important to that person’s overall sense of self (Jones et al., 2021). For example, some Black women may view being Black and a woman as important to who they are (i.e., their identity) and others may not. Many researchers have examined the extent to which Black women place importance on their race and gender identity (Jones et al., 2021; Lewis et al., 2017; Nelson et al., 2022; Szymanski & Lewis, 2016). Scholars have found mixed results on the influence of GRIC among Black women, such that some research has found that greater GRIC served as a protective factor in the face of gendered racism (racism and sexism; Lewis et al., 2017), and other studies found that greater GRIC was related to poorer outcomes when experiencing gendered racism (Jones et al., 2021; Nelson et al., 2022; Szymanski & Lewis, 2016). Specifically, research has found that more GRIC increased the positive relationship between identity shifting–changing one’s speech and actions to minimize experiences of discrimination–and depressive symptoms (Jones et al., 2021), and increased the positive
Johnson, Farmer, and Dickens | Gendered Racism and Coping Among Black Women
relationship between gendered racism and the use of disengagement coping among Black women (Szymanski & Lewis, 2016). Conversely, researchers also found that less GRIC increased the positive relationship between experiences of gendered racism and disengagement coping (Lewis et al., 2017), indicating mixed results on the influence of GRIC. Based on previous research, it is suspected that experiencing more gendered racism may be associated with Black women using more avoidant coping strategies (e.g., disengagement coping) regardless of their level of GRIC. Overall, more research is needed on the influence of GRIC on Black women’s experiences of stereotyping or discrimination. Particularly, GRIC needs to be examined in relation to the internalization of gendered racialized stereotypes, which stem from gendered racism.
Gendered Racialized Stereotypes – The Sapphire
Unfortunately, Black women are often subjected to negative gendered racialized stereotypes. Most prominently, there is the Jezebel, sexually promiscuous and seductive; the Angry Black woman, domineering and loud; the Mammy, selfsacrificing and nurturing; the Superwoman, emotionally strong and driven; and the welfare queen, lazy and materialistic (Collins, 2000; Holder et al., 2015; Lewis et al., 2016; Nelson et al., 2022). In particular, the Sapphire stereotype describes a Black woman who is emasculating, loud, and aggressive (Bond et al., 2021). The imagery associated with the Sapphire may seem familiar because it is the root of the Angry Black woman stereotype (West, 2018). These stereotypes can be extremely harmful to Black women, especially if they internalize the stereotypes. According to Bailey et al. (2011), internalizing and accepting negative stereotypes is a component of internalized racism, or the acceptance of White dominant culture’s attitudes towards Black people, that goes beyond simply identifying with these negative stereotypes. As Black people accept White dominant attitudes and identify with negative stereotypes, they start to reject Black culture and values. Similarly, metastereotype awareness refers to when people are aware of and have personal beliefs about stereotypes for their own social group (Jerald et al., 2017). For Black women, awareness of negative stereotypes and beliefs about one’s own group has been shown to have negative effects on one’s health (Jerald et al., 2017). Specifically, metastereotype awareness of the Sapphire stereotype was indirectly related to less selfcare and more substance use in Black women (Jerald et al., 2017).
Additionally, Black women who internalize the Sapphire image may struggle to express needs or show anger (Thomas et al., 2004). If gendered racialized
stereotypes—like the Sapphire—are internalized, Black women may feel responsible for the inconvenience of others and change to avoid conflict with other ethnic groups (Lewis et al., 2016; West, 1995). They might feel the need to suppress their anger and other emotions in order to prevent conflict with other people. This reaction may be one way that Black women who internalize the Sapphire stereotype cope, but more research is needed to identify specific coping strategies used by Black women. Given the limited research on Black women’s experiences with the Sapphire, we intentionally focused on the Sapphire stereotype in the present study. Our decision was influenced by the negative outcomes associated with this stereotype that Black women experience, such as less selfcare, more substance use, and difficulty expressing emotions (Jerald et al., 2017; Thomas et al., 2004).
Disengagement as a Coping
Strategy
To help manage stressors, such as experiences of gendered racism, Black women have engaged in behaviors to lessen the impact of life stressors, referred to as coping strategies (Hall, 2018; Holder et al., 2015; Jones, 2022; Kilgore et al., 2020). Research has found that access to social resources and networks in Black communities acted as an important coping strategy for managing emotional well being (Hall, 2018). Black women may also engage in coping strategies such as writing, religion/spirituality, armoring, shifting, and obtaining sponsorships/mentorships to cope with gendered racism (Dickens & Chavez, 2018; Holder et al., 2015; Kilgore et al., 2020). Jones (2022) examined how Black women develop responses to racial microaggressions on a predominantly White campus. The most common theme to emerge in participant responses was the use of disengagement coping, such as pulling out of participating in certain activities, groups, or situations (Lewis et al., 2017). In other words, participants refrained from responding to racial microaggressions in general due to not wanting to be perceived as the “Angry Black woman.”
In addition to choosing to refrain from responding, participants also felt emotionally drained and fearful of the consequences that may follow (Jones, 2022).
Disengagement coping is a seemingly maladaptive strategy because it has been associated with poorer mental and physical health and limitations in behavior (Jones, 2022; Lewis et al., 2017; Szymanski & Lewis, 2016; Thomas et al., 1995; West, 1995). Research has found increased use of disengagement coping to be associated with more experiences of gendered racial microaggressions (GRMs) and more psychological distress, such as poorer mental and physical health, among Black women (Lewis et al., 2017; Szymanski & Lewis, 2016). For example, Black women who used
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
more coping methods that involved disengagement strategies had more depressive symptoms when they experienced racism compared to Black women who used these coping strategies less often (West et al., 2010). Additionally, Williams and Lewis (2019) found more experiences of GRMs to be directly related to using more disengagement coping, which was then related to more depressive symptoms among Black women. As previously mentioned, internalizing gendered racial microaggressions, like stereotypes, can lead to Black women feeling responsible for inconveniencing others, which results in them changing their behaviors to avoid conflict (Lewis et al., 2016; West, 1995). Specifically, Thomas et al. (2004) found that internalization of the Sapphire stereotype was associated with an inability to express emotions in Black women. These findings align with the description of disengagement coping, suggesting that Black women who encounter the Sapphire stereotype might use more disengagement as a coping strategy.
Present Study
Previously, disengagement coping has been studied in relation to gendered racism as a whole and GRIC. However, little research has focused on disengagement coping in relation to a specific gendered racialized stereotype. The current study aimed to fill this gap in the research by exploring whether GRIC moderates the relationship between internalization of the Sapphire stereotype and disengagement as a coping strategy among Black women. This purpose led to the research question: How does GRIC moderate the relationship between internalization of the Sapphire stereotype and disengagement as a coping strategy among Black women? We hypothesized that internalization of the Sapphire stereotype would be positively correlated and GRIC would be negatively correlated with the use of disengagement as a coping strategy (H 1). In other words, increased internalization of the Sapphire stereotype would be associated with increased use of disengagement coping, but higher rates of GRIC would be associated with decreased use of disengagement coping. Additionally, we hypothesized that GRIC would moderate the relationship between internalization of the Sapphire stereotype and disengagement coping strategy among Black women (H2). Specifically, having a greater GRIC would significantly decrease the positive relationship between the Sapphire stereotype and disengagement use.
Method
Participants
The sample included 298 individuals, all of whom identified as Black/African American. Ninetyfour percent of
participants identified as women (n = 280), and 0.3% identified as transgender women (n = 1). Participants ranged from 18 to 52 years old (M = 27.29, SD = 5.78). Participants reported their marital status, with 55.8% being single and never married ( n = 154), 21.4% were single, 21.4% were in a committed relationship (n = 59), 21% being married (n = 58), 0.4% were separated ( n = 1), 1.1% were divorced ( n = 3), and 0.4% were widowed (n = 1). Participants selfidentified as 88.3% straight (n = 242), 8.8% bisexual (n = 24), 1.8% lesbian (n = 5), and 1.1% reported they were unsure of their sexual orientation (n = 3). Participants income level varied greatly; 29.3% had an income below $15,000 ( n = 81), 21% had an income between $15,000 and $30,000 (n = 58), 19.9% had an income between $30,001 and $50,000 (n = 55), 14.5% had an income between $50,001 and $75,000 (n = 40), and 15.2% had an income of $75,001 or more (n = 42; see Table 1).
Procedure
Prior to data collection, approval from Institutional Review Boards (IRBs) at Spelman College and Chicago State University were obtained. After IRB approval, Black women 18 years of age or older were recruited via email and several social media sites in order to gain a diverse sample. Eligibility criteria required participants to selfidentify as a Black/African American woman, 18 years of age or older, and living in the United States. The original study was part of a larger national study investigating stereotypes, discrimination, and health behaviors among Black women ages 18–35.
Participants completed a web based survey via Qualtrics. Invitations to participate in the study were distributed nationally, using the online community bulletin boards targeting African American/Black women, social media sites, professional listservs, campus flyers, and personal and professional contacts. The Qualtrics research panel was also utilized to recruit a larger sample of Black women. Before starting the survey, participants were presented with a consent form, in which they declared their willingness to participate in the study. The survey consisted of demographic information first, and then the study’s measures. The measures were organized in the following order: depression symptoms subscale, GRIC subscale, anxiety symptoms subscale, identity shifting scale, and gendered racism scale. The questionnaire took participants about 30–45 min to complete. Participants who were not recruited via Qualtrics had the chance to enter a raffle to win one of four $25 gift cards, and participants recruited through Qualtrics each received $5 (the standard amount determined by Qualtrics based on survey length and participant demographics).
WINTER 2024
Measures
Demographic Information
Participants were asked to complete a demographic questionnaire. The questionnaire included race, ethnicity, age, gender, sexual orientation, marital status, and household income.
Internalization of the Sapphire Stereotype
Internalization of the Sapphire stereotype was assessed using the Sapphire subscale from the Stereotypical Roles of Black Women Scale (Thomas et al., 2004), a 10item
scale measuring the presence of attitudes reflecting the Sapphire stereotype. Sample items include: “Black women are usually angry with others.” or “If given the chance, Black women will put down Black men.”
Scale items were rated using a 5point Likert scale that ranged from 1 (strongly disagree) to 5 (strongly agree). Responses across items were averaged, in which higher scores indicated higher internalization of the Sapphire stereotype (α = .85).
Gendered Racial Identity Centrality (GRIC)
GRIC was assessed by using an adapted version of the
TABLE 1
Demographic Characteristics
Note. % excludes missing value
8item Centrality subscale from the Multidimensional Model of Racial Identity (MMRI; Sellers et al., 1998).
The scale was modified to specifically measure GRIC. A sample item includes: “I have a strong attachment to other Black women.” Scale items were rated using a 7 point Likert scale that ranged from 1 ( strongly disagree) to 7 (strongly agree). Response options were averaged, with higher scores indicating higher levels of GRIC (α = .77).
Disengagement Coping
The use of disengagement as a coping mechanism was assessed using the behavioral disengagement subscale of the Brief COPE Inventory (Carver, 1997), which consists of two items measuring how often individuals disengage to cope. The items are: “I’ve been giving up trying to deal with it.” and “I’ve been giving up the attempt to cope.” Scale items were rated using a 4point scale, with responses ranging from 1 (I haven’t been doing this at all) to 4 (I’ve been doing this a lot). Response options were then averaged, with higher scores indicating higher use of disengagement coping (α = .76).
Results
Preliminary Analyses
In this study, researchers analyzed secondary data, obtained by the original researchers of a previous study (Jones et al., 2021), to address the specific research question. All the analyses for the current study were conducted using IBM SPSS Statistics (Version 27). Preliminary analyses included calculating the mean, standard deviation, and standard error for GRIC, the Sapphire stereotype, and disengagement coping. By exploring patterns, missing data for the variables ranged from 14–18%. There were no transformations performed to normally distribute the data. In our sample, participants on average had high levels of GRIC (M = 5.05 SD = 1.20) out of 7, meaning that on average, their gender and race combined were central aspects to
Moderation Between the Sapphire Stereotype, Identity Centrality, and Disengagement Coping
Note. Dependent Variable = Disengagement Coping
their identity. Participants in the sample also had low levels of internalizing the Sapphire stereotype (M = 2.48, SD = 0.73) out of 5, and low usage of disengagement coping (M = 1.75, SD = 0.85) out of 4. These low levels could suggest that participants did not highly internalize the Sapphire stereotype nor disengage to cope. G*Power (Faul et al., 2007) was utilized to conduct a priori power analysis, which determined that a sample size of n = 89 is needed for a medium effect size (f2 = .15; Cohen, 1988) at p < .05 level of significance with power at .95. Thus, the obtained sample size of n = 298 exceeds the adequate amount to test the current study hypothesis. Bivariate correlations between GRIC, internalization of the Sapphire stereotype, and disengagement coping among Black women were analyzed using Pearson’s r in SPSS. The results revealed a small negative association between GRIC and the internalization of the Sapphire stereotype, r (242) = .14, p = .03 , a small negative association between GRIC and the use of disengagement, r (251) = .16, p = .01, and a medium positive association between the use of disengagement and the internalization of the Sapphire stereotype, r (242) = .31, p < .001. These results indicated that more GRIC is associated with less internalization of the Sapphire stereotype and less use of disengagement as a coping strategy. These results also suggested that individuals who use disengagement as a coping strategy are more likely to internalize the Sapphire stereotype. Each correlational result also substantiated the study’s first hypothesis, which predicted that high internalization of the Sapphire stereotype would be related with greater use of disengagement coping, and greater levels of GRIC would be related to lower use of disengagement coping.
Moderation Analysis
To test the main hypothesis, whether GRIC moderated the relationship between internalization of the Sapphire stereotype (the predictor) and disengagement as a coping mechanism (the outcome), the researchers used Model 1 from the PROCESS Macro in SPSS version 4.3 (Hayes, 2022). The overall model was significant, R 2 = .14, F (3, 240) = 12.95, p < .001. Both GRIC (β = .10, SE = .04, p = .03) and the Sapphire stereotype (β = .28, SE = .07, p < .001) significantly contributed to the variance in explaining disengagement (see Table 2). Also, the interaction between GRIC and the Sapphire stereotype was significant, ΔR2 = .03, F(1, 240) = 8.54, p = .004, β = −.19. To understand the significant moderation, a simple effects slope analyses was conducted. For Black women who reported lower GRIC, internalizing the Sapphire stereotype more was associated with more disengagement coping ( b = 51, SE = 0.09, t = 5.67, p < .001, 95% CI [.33, .69]). For Black women who
WINTER 2024
TABLE 2
Johnson, Farmer, and Dickens | Gendered Racism and Coping Among Black Women
reported higher GRIC, the relationship between internalizing the Sapphire stereotype and the use of disengagement coping was no longer significant (b = .05, SE = .12, t = 0.40, p = .69, 95% CI [−.20, .29]; see Figure 1). Overall, Black women with higher levels of GRIC internalized the Sapphire stereotype much less and in turn were less likely to engage in disengagement as a coping strategy, compared to Black women with lower levels of gendered racial identity, in which they internalized the Sapphire stereotype more and were more likely to use disengagement to cope.
A multiple linear regression was conducted to test if Black women’s GRIC and their internalization of the Sapphire stereotype significantly predicted Black women’s use of disengagement as a coping strategy. The fitted regression model was Disengagement = 1.29 –0.08 (identity centrality) + 0.35 (Sapphire stereotype). The overall regression was statistically significant, R2 = .11, F(2, 241) = 14.71, p < .001. It was found that GRIC did not significantly predict Black women’s use of disengagement as a coping strategy, (β = −.11, p = .09). It was also found that internalization of the Sapphire stereotype significantly predicted Black women’s use of disengagement as a coping strategy, (β = .30, p < .001).
Discussion
The purpose of this research was to understand how the relationship between internalization of the Sapphire stereotype and the use of disengagement was moderated by GRIC. Acknowledging the role of GRIC in this relationship could confront the maladaptive coping mechanisms concerning gendered racial stress and aid in creating messages promoting healthy gendered racial identity, specifically among Black women. Based on the previous literature, researchers predicted that internalization of the Sapphire stereotype would have a positive association and GRIC would have a negative association with the use of disengagement to cope. Thus, it was also hypothesized that GRIC would decrease the positive link between internalization of the Sapphire stereotype and disengagement coping.
The results confirmed each of the hypotheses, revealing a small negative relationship between participants’ sense of GRIC and the internalization of the Sapphire stereotype, indicating that individuals whose gendered racial identity was central to their overall identity was associated with less internalization of the Sapphire stereotype. Similarly, a negative association between GRIC and the use of disengagement suggested that participants high in identity centrality were less likely to utilize disengagement as a coping strategy. This finding relates to Williams and Lewis’ (2019) study, which indicates that Black women with negative
perceptions of their identity were more likely to use disengagement to cope with gendered racial stressors. The correlation results also suggest that a positive, medium association between the use of disengagement and the internalization of the Sapphire stereotype exists, in which participants who use disengagement as a coping strategy were more likely to internalize the Sapphire stereotype. This finding complements previous work by Thomas et al. (2004), which established an existing relationship between internalizing the Sapphire stereotype and the inability to express emotions. The multiple linear regression preliminary results revealed that although there is a relationship between GRIC and Black women’s use of disengagement as a coping strategy, GRIC does not predict the level of Black women’s use of disengagement as a coping strategy. We can conclude that regardless of how important Black women’s social identities are to them, this does not mean that they will refrain from using disengagement as a coping mechanism. However, results also reveal that Black women who personalize the Sapphire stereotype are more likely to use disengagement as a coping strategy.
Overall, GRIC did moderate the relationship between internalization of the Sapphire stereotype and the use of disengagement, confirming the research hypothesis. For Black women with low identity centrality, they used more disengagement coping when they highly internalized the Sapphire stereotype. For Black women with high identity centrality, there was no longer a significant relationship between disengagement coping and internalization of the Sapphire stereotype. Women with low identity centrality might be internalizing the
Interaction Effect of Identity Centrality on the Sapphire and Disengagement Coping
FIGURE 1
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
stereotype at higher rates and increase their use of disengagement coping, or they might not have good coping strategies when encountering gendered racism so they just choose to limit their interactions in general. Sellers et al. (1998) suggests that individuals with a greater sense of identity centrality are able to recognize encounters of oppression more readily instead of internalizing the previous event. Based on this research, women with low GRIC may not be able to identify when they are experiencing gendered racism, leaving them to be more susceptible to internalizing gendered racial microaggressions.
Previous work also suggests that Black women with lower gender racial private regard use disengagement coping to protect themselves from psychological distress caused by gendered racial stressors (Williams & Lewis, 2019). This finding shows that having a high GRIC does decrease the relationship between the Sapphire stereotype and disengagement coping, which is consistent with the finding that less GRIC increases the relationship between gendered racism and disengagement coping (Lewis et al., 2017). However, it is not consistent with previous work that has found more GRIC to increase the positive relationship between identity shifting and depression symptoms (Jones et al., 2021) and between gendered racism and disengagement coping (Syzmanski & Lewis, 2016). Considering the current study’s findings, having a high GRIC might mean that Black women recognize the value in their identities, so they are not as susceptible to internalizing messages associated with stereotypes, such as the Sapphire stereotype. Additionally, this finding could potentially explain the negative relationship between GRIC and the use of disengagement coping among the current study’s participants. Because GRIC was found to buffer the positive relationship between internalizing the Sapphire and the use of disengagement coping, it is possible that Black women with higher GRIC are using less disengagement coping and internalizing the Sapphire stereotype less; if they internalize the Sapphire stereotype less, then they might use less disengagement coping.
Limitations and Future Research
Despite these important results, a couple limitations were present in the current study. Although this study found significant results for the moderating role of GRIC, there might be other variables that influence the relationship between the Sapphire stereotype and disengagement coping. Other aspects of gendered racial identity, like regard (i.e., how positively one feels about their own race/ethnicity) and ideology (e.g., one’s beliefs about how African Americans should behave in society; Sellers et al., 1998), might moderate this relationship.
Specifically, high regard and ideologies that promote African American culture might decrease the positive relationship between the Sapphire and disengagement coping. It would also be useful for future research to consider how experiencing stressful life events or having a general positive attitude may change this relationship as well (Spencer et al., 1997). In addition to these variables, future researchers should expand the research on the Sapphire stereotype by studying it in relation to various health outcomes in Black women. For example, researchers could examine how depression, anxiety, and other coping strategies (e.g., identity shifting) may relate to Black women’s internalization of the Sapphire stereotype. Additionally, researchers should explore Black women’s uses of adaptive, in addition to maladaptive, coping strategies in response to experiences of gendered racism. As an example, a recent study found that distraction (i.e., stepping away from the stressor to do something enjoyable) to be an adaptive form of disengagement coping that increased positive emotions (Waugh et al., 2020; Waugh et al., 2021).
Another limitation was the average age of the study participants. On average, participants were approximately 27 years old, meaning they were relatively young. Younger participants might have had fewer experiences with gendered racism, relative to older Black women, given their age. If young Black women have had few experiences with the Sapphire stereotype, then they would not have as many opportunities to internalize it. In the future, researchers should consider how the results of the current study may differ with an older sample of Black women. It is possible that an older sample of Black women might have higher rates of internalization because they have had more opportunities to encounter experiences of discrimination and being stereotyped, such as with the Sapphire stereotype. A younger sample should be examined as well because adolescence is a crucial time for identity development (Spencer et al., 1997). Additionally, adolescents may experience gendered racism differently and choose different coping strategies (Sellers et al., 2006; Spencer et al., 1997; Velez & Spencer, 2018). It would also be critical for future researchers to conduct a longitudinal study to see how internalization of the Sapphire stereotype and the use of disengagement coping changes over time for Black girls and women.
Implications
These limitations do not take away from the implications from the findings of the current study. The findings still demonstrate the importance of healthy gender and racial identity development for Black girls and women (Jacobs, 2016; Lewis et al., 2017). GRIC can be a useful tool for
WINTER 2024
Johnson, Farmer, and Dickens | Gendered Racism and Coping Among Black Women
combatting gendered racism (Lewis et al., 2017), so it is important to cultivate a strong sense of gendered racial identity in young Black girls. Past research has suggested for schools to take an active role in combatting negative messages about Black girls and women to help show Black girls that those messages are not true (Burnett et al., 2022; Jacobs, 2016).
The development of educational programs may also be useful to show schools and parents how they can help young Black girls proudly build these parts of their identity. For example, schools could include positive examples of Black women in their curriculum to expose Black girls to positive representation, and parents could incorporate daily words of affirmation to foster selflove in their Black daughters. Schools and parents should also work together to promote positive messages about Black girls and women (Burnett et al., 2022). With this cooperation, the messages at home are being corroborated at school, and Black girls are receiving positive messages from multiple sources. Furthermore, there is a need for programs, like Afrocentric communitybased programs, that teach Black women how to use healthy coping mechanisms to combat gendered racism (Gibbs & Fuery, 1994; Lewis et al., 2013; Linnaberry et al., 2014; ShorterGooden, 2004). Unhealthy coping mechanisms, like disengagement coping, may lead to negative health outcomes, like anxiety and depression. Healthy coping mechanisms, such as religion and social support, give Black women an outlet to deal with gendered racism in a more beneficial manner (Linnaberry et al., 2014; ShorterGooden, 2004).
Conclusion
Using PVEST, the current research explores Black women’s experiences with gendered racialized stereotypes and their choice of coping strategies. Having a strong gendered racial identity can help Black women deal with internalizing the Sapphire stereotype. The present study has implications for schools and parents to support their Black children, indicating the importance of cultivating a strong, positive gendered racial identity in Black girls and women. Developing programs that aim to increase GRIC will give Black girls and women the tools to use their identity as a shield against gendered racism. By studying the Sapphire stereotype and disengagement coping, this research shows the negative toll that gendered racialized stereotypes can have on Black women.
References
Bailey, T.-K. M., Chung, Y. B., Williams, W. S., Singh, A. A., & Terrell, H. K. (2011). Development and validation of the Internalized Racial Oppression Scale for Black individuals. Journal of Counseling Psychology, 58(4), 481–493. https://psycnet.apa.org/doi/10.1037/a0023585
Bond, K. T., Leblanc, N., Williams, P., & Amutah-Onukagha, N. N. (2021). Racebased sexual stereotypes, gendered racism, and sexual decision making among young Black cisgender women. Health Education & Behavior, 48(3), 295–305. https://doi.org/10.1177/10901981211010086
Burnett, M., McBride, M., Green, M. N., & Cooper, S. M. (2022). “When I think of Black girls, I think of opportunities”: Black girls’ identity development and the protective role of parental socialization in educational settings. Frontiers in Psychology, 13, 1–16. https://doi.org/10.3389/fpsyg.2022.933476
Carver, C. S. (1997). You want to measure coping but your protocol’s too long: Consider the brief COPE. International Journal of Behavioral Medicine, 4(1), 92–100. https://doi.org/10.1207/s15327558ijbm0401_6
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.
Collins, P. H. (2000). Black feminist thought: Knowledge, consciousness, and the politics of empowerment (2nd ed.). Routledge. https://doi.org/10.4324/9780203900055
Dickens, D. D., & Chavez, E. L. (2018). Navigating the workplace: The costs and benefits of shifting identities at work among early career U.S. Black women. Sex Roles, 78(1), 760–774. https://doi.org/10.1007/s11199-017-0844-x
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(2), 175–191. https://doi.org/10.3758/bf03193146
Gibbs, J. T., & Fuery, D. (1994). Mental health and well-being of Black women: Toward strategies of empowerment. American Journal of Community Psychology, 22(4), 559–582. https://doi.org/10.1007/BF02506893
Greer, T. M. (2011). Coping strategies as moderators of the relation between individual race-related stress and mental health symptoms for African American women. Psychology of Women Quarterly, 35(2), 215–226. https://doi.org/10.1177/0361684311399388
Hall, J. C. (2018). It is tough being a Black woman: Intergeneration stress and coping. Journal of Black Studies, 49(5), 481–501. https://doi.org/10.1177/0021934718766817
Hayes, A. F. (2022). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (3rd ed.). Guilford Press.
Holder, A. M. B., Jackson, M. A., & Ponterotto, J. G. (2015). Racial microaggression experiences and coping strategies of Black women in corporate leadership. Qualitative Psychology, 2(2), 164–180. https://doi.org/10.1037/qup0000024
Jacobs, C. E. (2016). Developing the “oppositional gaze”: Using critical media pedagogy and Black feminist thought to promote Black girls’ identity development. Journal of Negro Education, 85(3), 225–238. https://doi.org/10.7709/jnegroeducation.85.3.0225
Jerald, M. C., Cole, E. R., Ward, L. M., & Avery, L. A. (2017). Controlling images: How awareness of group stereotypes affects Black women’s well-being. Journal of Counseling Psychology, 64(5), 487–499. http://dx.doi.org/10.1037/cou0000233
Jones, A. M. (2022). Conflicted: How Black women negotiate their responses to racial microaggressions at a historically White institution. Race Ethnicity and Education, 25(5), 738–753. https://doi.org/10.1080/13613324.2021.1924136
Jones, M. S., Womack, V., Jeremie-Brink, G., & Dickens, D. (2021). Gendered racism and mental health among young adult U.S. Black women: The moderating roles of gendered racial identity centrality and identity shifting. Sex Roles, 85, 221–231. https://doi.org/10.1007/s11199-020-01214-1
Kilgore, A. M., Kraus, R., & Littleford, L. N. (2020). “But I’m not allowed to be mad”: How Black women cope with gendered racial microaggressions through writing. Translational Issues in Psychological Science, 6(4), 372–382. http://dx.doi.org/10.1037/tps0000259
Lewis, J. A., Mendenhall, R., Harwood, S. A., & Huntt, M. B. (2013). Coping with gendered racial microaggressions among Black women college students. Journal of African American Studies 17(1), 51–73. https://doi.org/10.1007/s12111-012-9219-0
Lewis, J. A., Mendenhall, R., Harwood, S. A., & Huntt, M. B. (2016). “Ain’t I a woman?”: Perceived gendered racial microaggressions experience by Black women. The Counseling Psychologist, 44(5), 758–780. https://doi.org/10.1177/0011000016641193
Lewis, J. A., Williams, M. G., Peppers, E. J., & Gadson, C. A. (2017). Applying intersectionality to explore the relations between gendered racism and health among Black women. Journal of Counseling Psychology, 64(5), 475–486. https://doi.org/10.1037/cou0000231
Linnaberry, E., Stuhlmacher, A. F., & Towler, A. (2014). From whence cometh their strength: Social support, coping, and well-being of Black women professionals. Cultural Diversity and Ethnic Minority Psychology, 20(4), 541–549. http://dx.doi.org/10.1037/a0037873
Nelson, T., Cardemil, E. V., Overstreet, N. M., Hunter, C. D., & Woods-Giscombé, C. L. (2022). Association between Superwoman Schema, depression, and resilience: The mediating role of social isolation and gendered racial centrality. Cultural Diversity & Ethnic Minority Psychology, 30(1), 95–106. https://doi.org/10.1037/cdp0000533
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
Seaton, E. K., Yip, T., & Sellers, R. M. (2009). A longitudinal examination of racial identity and racial discrimination among African American adolescents. Child Development, 80(2), 406–417. https://doi.org/10.1111/j.1467-8624.2009.01268.x
Sellers, R. M., Copeland-Linder, N., Martin, P. P., & Lewis, R. L. (2006). Racial identity matters: The relationship between racial discrimination and psychological functioning in African American adolescents. Journal of Research on Adolescence, 16(2), 187–216. https://doi.org/10.1111/j.1532-7795.2006.00128.x
Sellers, R. M., Smith, M. A., Shelton, J. N., Rowley, S. A. J., & Chavous, T. M. (1998). Multidimensional Model of Racial Identity: A reconceptualization of African American racial identity. Personality and Social Psychology Review, 2(1), 18–39. https://doi.org/10.1207/s15327957pspr0201_2
Shorter-Gooden, K. (2004). Multiple resistance strategies: How African American women cope with racism and sexism. Journal of Black Psychology, 30(3), 406–425. https://doi.org/10.1177/0095798404266050
Spencer, M. B., Dupree, D., & Hartmann, T. (1997). A Phenomenological Variant of Ecological Systems Theory (PVEST): A self-organization perspective in context. Development and Psychopathology, 9(4), 817–833. https://doi.org/10.1017/s0954579497001454
Szymanski, D. M. & Lewis, J. A. (2016). Gendered racism, coping, identity centrality, and African American college women’s psychological distress. Psychology of Women Quarterly, 40(2), 229–243. https://doi.org/10.1177/0361684315616113
Thomas, A., Witherspoon, K. M., & Speight, S. L. (2004). Toward the development of the stereotypic roles for Black women scale. Journal of Black Psychology, 30(3), 426–442. https://doi.org/10.1177/0095798404266061
Velez, G., & Spencer, M. B. (2018). Phenomenology and intersectionality: Using PVEST as a frame for adolescent identity formation amid intersecting ecological systems of inequality. New Directions for Child and Adolescent Development, 2018(161), 75–90. https://doi.org/10.1002/cad.20247
Waugh, C. E., Leslie-Miller, C. J., Shing, E. Z., Furr, R. M., Nightingale, C. L., & McLean, T. W. (2021). Adaptive and maladaptive forms of disengagement coping in caregivers of children with chronic illnesses. Stress and Health: Journal of the International Society for the Investigation of Stress, 37(2), 213–222. https://doi.org/10.1002/smi.2985
Waugh, C. E., Shing, E. Z., & Furr, R. M. (2020). Not all disengagement coping
strategies are created equal: Positive distraction, but not avoidance, can be an adaptive coping strategy for chronic life stressors. Anxiety, Stress, and Coping, 33(5), 511–529. https://doi.org/10.1080/10615806.2020.1755820
West, C. M. (1995). Mammy, Sapphire, and Jezebel: Historical images of Black women and their implications for psychotherapy. Psychotherapy, 32, 458–466. https://doi.org/10.1037/0033-3204.32.3.458
West, C. M. (2018). Mammy, Sapphire, Jezebel, and the bad girls of reality television: Media representations of Black women. Lectures on the Psychology of Women, 139–158.
West, L. M., Donovan, R. A., & Roemer, L. (2010). Coping with racism: What works and doesn’t work for Black women? Journal of Black Psychology, 36(3), 331–349. https://doi.org/10.1177/0095798409353755
Williams, M. G., & Lewis, J. A. (2019). Gendered racial microaggressions and depressive symptoms among Black women: A moderated mediation model. Psychology of Women Quarterly, 43(3), 368–380. https://doi.org/10.1177/0361684319832511
Author Note
Nailah Johnson https://orcid.org/0009000717884681
Nailah Johnson is now at the Department of Psychology at Cornell University, Ithaca, NY.
Makyra Farmer is now in the Clinical Mental Health Counseling program at Agnes Scott College, Decatur, GA. This research was supported by the National Science Foundation Award # 1832141 and Award # 2010676.
Positionality Statement: All the authors selfidentify as Black American women, which influences their desire to study Black women. These authors acknowledge that their perspectives are influenced by their positions within these dimensions of identity. Correspondence concerning this article should be addressed to Nailah Johnson, Department of Psychology, Cornell University, 116 Reservoir Ave, Ithaca, NY, 14850, United States.
Email: naj45@cornell.edu
Implicit Juror Bias: Does Racial Priming Impact the Race-Crime Congruency Effect?
Evangelina T. Marquez, Kathryn Sperry*, Emilia Griffiths, and Jack Stuart Department of Psychological Science, Weber State University
ABSTRACT. Defendants who commit racially stereotypical crimes are judged more harshly compared to defendants who commit counter stereotypic crimes. This is termed the race crime congruency effect (e.g., Gordon et al., 2001; Jones & Kaplan, 2003). The present study sought to replicate the racecrime congruency hypothesis, and test whether racial priming would attenuate this effect. Participants (N = 361 White participants; 193 MTurk workers and 168 undergraduate participants) viewed a police report in which the crime was either embezzlement or gang crime, and the defendant was either White or Black. Half the participants were primed to think about their own race prior to viewing the police report. We employed a 2 (racecrime congruence) x 2 (defendant race) x 2 (racial prime) x 2 (sample) betweensubjects design. The Black defendant was judged more harshly on several dependent measures, and this effect was largely driven by the undergraduate participants. For most dependent measures, we replicated the race crime congruency effect only for the White defendant. Embezzlement committed by the White defendant led to more confidence in a guilty verdict, more responsibility for the crime, and an internal locus of causality compared to the White defendant who committed gang crime. Priming participants to think about their own race may have increased racial biases. Among the undergraduate sample, the racial prime led participants to give harsher punishment to the Black defendant compared to the White defendant. Implications and future directions are discussed.
Keywords: racecrime congruence, racial bias, racial priming, race salience, jury decisionmaking
RESUMEN. Los acusados que cometen delitos racialmente estereotipados son juzgados con más dureza en comparación con los acusados que cometen delitos contra los estereotipos. Esto se denomina efecto de congruencia entre delitos raciales (p. ej., Gordon et al., 2001; Jones y Kaplan, 2003). El presente estudio buscó replicar la hipótesis de la congruencia entre delitos raciales y probar si la preparación racial atenuaría este efecto. Los participantes (N = 361 participantes Blancos; 193 trabajadores de MTurk y 168 participantes universitarios) vieron un informe
Diversity badge earned for conducting research focusing on aspects of diversity. Open Data and Open Materials badges earned for transparent research practices. Data and materials are available at https://osf.io/ykptf
Racial Prime and Race-Crime Congruency | Marquez, Sperry, Griffiths, and Stuart
policial en el que el delito era malversación de fondos o delito de pandillas, y el acusado era Blanco o Negro. A la mitad de los participantes se les pidió que pensaran en su propia raza antes de ver el informe policial. Empleamos un diseño entre sujetos 2 (congruencia razacrimen) x 2 (raza del acusado) x 2 (primo racial) x 2 (muestra). El acusado Negro fue juzgado con mayor dureza según varias medidas dependientes, y este efecto fue impulsado en gran medida por los participantes universitarios. Para la mayoría de las medidas dependientes, replicamos el efecto de congruencia entre delitos raciales sólo para el acusado Blanco. La malversación de fondos cometida por el acusado Blanco generó más confianza en un veredicto de culpabilidad, más responsabilidad por el delito y un locus de causalidad interno en comparación con el acusado blanco que cometió delitos de pandillas. Preparar a los participantes para que piensen en su propia raza puede haber aumentado los prejuicios raciales. Entre la muestra de estudiantes universitarios, la preferencia racial llevó a los participantes a aplicar un castigo más severo al acusado negro en comparación con el acusado blanco. Se discuten las implicaciones y direcciones futuras.
Implicit racism in our criminal justice system occurs at all stages—from arrest to sentencing. The Black Lives Matter (BLM) movement has highlighted the disproportionate number of unarmed Black males who are shot and often killed by police. According to the Uniform Crime Report, Black Americans make up 12% of the U.S. population yet comprise about 27% of all arrests (Federal Bureau of Investigations, 2019). In 2020, Black Americans were imprisoned at a rate 3.5 times higher than White Americans (Minton & Zeng, 2021). Additionally, Black suspects are 3.23 times more likely than White suspects to die at the hands of a police officer (Schwartz & Jahn, 2020). Research has also demonstrated clear racial biases against Black defendants in the courtroom. In a comprehensive review of racial disparities in the U.S. criminal justice system, Kovera (2019) summarized racial biases in pretrial processing, sentences, pleas, and wrongful convictions. Black defendants were more likely to receive pleas that involved incarceration compared to White defendants (whose pleas are more likely to involve reduced charges). Even after controlling for case variables, Black and Hispanic defendants received harsher sentences than White defendants (Kovera, 2019).
Addressing another form of bias in the criminal justice system, the present study re examined the racecrime congruency effect, whereby jurors judge defendants more harshly for committing crimes that are stereotypic of their race. Additionally, the present study examined whether racial priming can reduce the racecrime congruency effect.
Racial Biases in Criminal Justice
Explicit racism has seemingly been on the decline in recent decades. However, a large body of research provides evidence of more subtle forms of racial bias in our criminal justice system. One example of these subtle biases is provided in literature examining decisions to shoot White vs. Black suspects. In a firstperson shooter task, participants engaged in a video game where they were instructed to shoot armed targets and not shoot unarmed targets (Correll et al., 2007a). This study found that participants set a more lenient threshold for shooting Black targets, and this effect was even larger if they first read an article about a Black criminal prior to engaging in the shooter task. Both community and police officer samples made faster decisions with stereotype congruent targets (e.g., Black armed targets and White unarmed targets; Correll et al., 2007b). A metaanalysis confirmed that across 42 studies, participants are quicker to shoot armed Black targets than armed White targets and have a more lenient threshold for shooting Black targets (Mekawi & Bresin, 2015).
These racial biases have been observed in juror verdicts and sentencing. Experimental and archival studies confirm a clear bias against Black defendants in the courtroom. According to an article published by the National Registry of Exonerations, innocent Black Americans are seven times more likely than White Americans to be falsely convicted of serious crimes (Gross et al., 2022). In a study of over 600 capital cases in Philadelphia, researchers found that when the
Marquez, Sperry, Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
victim was White, Black defendants with stereotypical features were more likely to receive the death penalty (Eberhardt et al., 2006). Furthermore, a metaanalysis found that participants were more likely to render guilty verdicts and recommend harsher sentences for otherrace defendants (Mitchell et al., 2005). Jurors are disproportionately White (Offit, 2021), which often translates into harsher sentences for Black defendants.
Race-Crime Congruency Effect
Certain racial groups seem to be associated with certain types of crimes. Black Americans tend to be associated with crimes such as soliciting, assault, grand theft auto, and gang crime. White Americans tend to be associated with crimes such as fraud, embezzlement, counterfeiting, and rape (Skorinko & Spellman, 2013; Sunnafrank & Fontes, 1983). Previous research indicates that mock jurors are more likely to deem defendants guilty when accused of racestereotypic crimes, termed the racecrime congruency effect (e.g., Gordon et al., 2001; Jones & Kaplan, 2003). For example, Gordon et al. (2001) found that both Black and White defendants were given longer jail sentences and assumed more likely to reoffend if the crime was consistent with racial stereotypes (e.g., a Black defendant charged with burglary and a White defendant charged with embezzlement). Additionally, Jones and Kaplan (2003) found when the crime was consistent with racial stereotypes, participants were more likely to make internal attributions and believe the defendant was more responsible for the crime.
Recent explorations of race crime congruence found stronger effects for Black defendants compared to White defendants. For example, Maeder et al. (2016) found that Black defendants received more guilty verdicts in an autotheft trials compared to the dangerous operation of a motor vehicle while intoxicated (evidence of the racecrime congruency effect), but did not find that White defendants received more guilty verdicts in a fraud trial compared to the other two trials (auto theft and dangerous operation of a motor vehicle). Similarly, Petsko and Bodenhausen (2019; study 1) also replicated the racecrime congruency effect, but only for Black stereotypical crimes.
Racial Priming
Priming is defined as “the incidental activation of knowledge structures, such as trait concepts and stereotypes, by the current situational context” (Bargh et al., 1996, p. 230). In other words, priming occurs when certain thoughts or ideas are brought to the forefront of someone’s mind through recent experiences. To date, no studies have examined whether priming participants to think about their own racial identity impacts perceptions
of defendants, and particularly how priming impacts the racecrime congruency effect. However, a closely related construct is called race salience.
According to Sommers and Ellsworth (2000, p. 1371), “During the course of a trial, racial issues may become salient in any number of ways, including pretrial publicity, voir dire questioning of potential jurors, opening and closing arguments, the nature of police testimony, attorneys’ demeanors, and sometimes the nature of the crime itself.” Much of the race salience literature is anchored in aversive racism theory (Dovidio & Gaertner, 2004). Accordingly, liberal and welleducated White Americans are highly motivated to appear nonracist, despite possessing implicit racial biases. When racial issues are highlighted in a case it can serve as a reminder for jurors to not appear racist, resulting in lower reported racial biases (Sommers & Ellsworth, 2000).
In one of the earliest tests of race salience, Sommers and Ellsworth (2000) found that when racial issues were not salient, White jurors showed the predicted bias against Black defendants. However, when the trial included race salient testimony (in which the defendant yelled: “You know not to talk that way about a White (or Black) man”) the defendant’s race had no impact on White jurors. Sommers and Ellsworth (2001) replicated these findings using a different trial and manipulation of race salience (i.e., a witness testified that the defendant was one of only two Whites (or Blacks) on the team and “had been the subject of racial remarks and criticism”). Some studies have found that race salience even creates an out-group favoritism effect, resulting in more lenient judgments for Black defendants compared to White defendants (e.g., Bucolo & Cohn, 2010; Gamblin & Kehn, 2021).
Interestingly, two recent studies which utilized racially charged media found contradictory results. Livingston and Gurung (2019) found that priming race through media coverage reduced racial biases, consistent with aversive racism theory. However, McManus et al. (2018) did not find evidence supporting the race salience effect when participants were exposed to racially charged media coverage.
The present study sought to examine the impact of priming participants to think about their own racial identity. This is a novel way to conceptualize racial priming in this context. It is unclear whether racial priming would have similar effects as those described in the race salience literature. On the one hand, it’s possible that racial priming might encourage participants to think about the importance of their own race. Through this lens, racial priming would not reduce racial biases, but might actually increase them. On the other hand, racial
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
priming might be seen as analogous to race salience by bringing to light biases and motivating participants to appear unbiased, consistent with aversive racism theory. One study most similar to the racial prime in the present study examined race salience in the context of a pretrial questionnaire. Schuller et al. (2009) gave participants a pretrial questionnaire which asked them to reflect on their own racial biases; those participants were subsequently less likely to find the defendant guilty than those given a racially neutral questionnaire. The authors noted that including a pretrial questionnaire that requires a thoughtful reflection of one’s biases may have “enabled the potential jurors to reflect on the powerful influence of race and this deliberative mindset may have simply made them more cautious in their responses’’ (Schuller et al., 2009, p. 326).
The Present Study
The present study examined three research questions: (1) Will there be evidence of racial biases in judgments of the Black vs. the White defendant? We expect the Black defendant to be judged more harshly on several dependent variables compared to the White defendant; (2) Will we replicate the racecrime congruency effect? Based on the most recent explorations of the racecrime congruency effect (Maeder et al., 2016; Petsko & Bodenhausen, 2019), we expect to replicate the racecrime congruency effect and expect this effect will be stronger for the Black defendant compared to the White defendant; and (3) What impact will racial priming have on racial biases, and particularly the racecrime congruency effect? The novelty of the racial prime makes this hypothesis tenuous. But, based on the race salience literature (Sommers & Ellsworth, 2000; Sommers & Ellsworth, 2001), we expected the racial prime to reduce participants’ racial biases (bringing judgments of the White and Black defendant closer together). If the racecrime congruency effect is replicated, we expect the racial prime will dampen the effect.
Method
Participants and Design
Participants were recruited in one of two ways. First, participants were recruited using Amazon’s Mechanical Turk (MTurk) and were paid $2.00 for their participation. Second, we collected data from a public university and awarded participants course credit upon completion of the survey. Undergraduate participants were recruited using Sona, the online participant management system.
Data Filtering
We began with a sample of 402 MTurk participants. We filtered out individuals who did not pass the attention
check ( n = 74; 18%) and individuals who took the survey twice (n = 8; 2%). Only the participants’ second survey submission was deleted. This left our sample with 320 MTurk participants. We began with a sample of 449 undergraduate participants. We filtered out the individuals who did not pass the attention check (n = 248; 55%). This left our sample with 201 undergraduate participants.
We combined the datasets from the two samples and filtered out any participants who did not identify as White (n = 87; 17%). Participants could select multiple ethnic identities, if “White” was one of the selected identities these participants were included in the final data. This resulted in a final sample size of N = 434.
MTurk Sample Demographics (n = 261)
The MTurk sample was mostly men (n = 180; 69%) with a mean age of 37 (SD = 10.07). Most MTurk participants had a Bachelor’s degree (n = 147; 56%) or a Master’s degree ( n = 48; 18%). The remaining participants had some college (n = 23; 9%), a high school diploma (n = 24; 9%), an Associate’s degree (n = 14; 5%), or a professional degree (n = 1; 0.4%). Nine (3%) MTurk participants indicated more than one racial/ethnic identity (White and another identity). On a scale from 1 (extremely liberal) to 10 (extremely conservative), the average political leaning was 5.86 (SD = 3.17). There were 61 (23%) MTurk workers identified as “Master Workers,” with an average of 50,000 Human Intelligence Tasks (HITs).
Undergraduate Sample Demographics (n = 173)
The undergraduate sample was mostly women (n = 103; 60%) with a mean age of 24 (SD = 8.49). A majority of the sample had some college (n = 129; 75%) or an Associate’s degree (n = 26; 15%). The remaining participants had completed high school (n = 10; 6%) or had a Bachelor’s degree (n = 8, 5%). There were 14 (8%) multiracial participants (participants who identified as White and at least one other racial identity). The average political leaning (on a scale from 1 – 10) was 5.27 (SD = 2.05).
We combined the MTurk and the undergraduate samples and included the sample as a factor in the design. This was a 2 (race crime congruence) x 2 (defendant race) x 2 (racial prime) x 2 (sample) betweensubjects factorial design.
Procedure
The institutional Review Board at Weber State University provided approval prior to data collection (IRB #AY2122165). Participants were shown a crime description which included a pixelated picture of the defendant Racial
WINTER 2024
Marquez, Sperry, Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
(obtained from Petsko & Bodenhausen, 2019; see Appendix). The defendant was either Black or White, and the crime described was either embezzlement (White stereotypical) or gangrelated (Black stereotypical).
Prior to viewing the description of the crime, half of the participants were primed to think about their own racial or ethnic identity. Racial priming was accomplished by instructing participants to fill out the Collective SelfEsteem Scale – Race (CSESR; Luhtanen & Crocker, 1992) either before (racial prime condition) or after the dependent variables. To increase the salience of the racial prime manipulation participants typed their racial identity prior to filling out the measure.
Participants then viewed the description of the crime and followed by questions concerning their verdict and perceptions of the defendant. Pursuant to the instructions used by Petsko and Bodenhausen (2019; Experiment 1), participants were asked to report on “the judgments that the average American juror might make based on the information provided” (italics added). This was to reduce the extent to which social desirability would impact participants’ responses to the survey. Indirect questioning has been shown to be effective in reducing social desirability biases (e.g., Fisher, 1993).
Measures
Collective Self-Esteem Scale–Race
All participants filled out the Collective SelfEsteem Scale – Race (CSES R; Luhtanen & Crocker, 1992).
The CSESR assesses the extent to which one’s racial identity is a source of selfesteem. Participants were asked to think about their racial identity and how they feel being a part of that racial identity. Sample items include: “Overall, my racial/ethnic group is considered good by others,” and “The racial/ethnic group I belong to is an important reflection of who I am.” This scale had high internal consistency in the present study (16 items; α = .82). Participants who were in the racial prime condition completed this measure before viewing the crime description. Participants who were not in the racial prime condition completed this measure after the main dependent measures.
Verdict and Confidence
Participants indicated whether the defendant was guilty or not guilty, followed by their confidence in the verdict (four items on a scale from 1 10; Petsko & Bodenhausen, 2019; α = .79 in the present study). Verdict was coded as 1 (not guilty) and +1 (guilty). A verdictconfidence scale was created by multiplying the verdict by their confidence. A value of 10 indicated highly confident in a not guilty verdict and value of +10 indicated highly confident in a guilty verdict.
Discipline and Punishment
Questions pertaining to disciplinary action and punishment were based on materials by Petsko and Bodenhausen (2019). Participants were asked the likelihood that the average American juror would recommend disciplinary action against the defendant, using a scale from 1 (definitely not) to 5 (definitely would). They were then asked what level of punishment would likely be recommended, from 1 (minimum provided by law) to 9 (maximum provided by law).
Attributions of Responsibility, Stability, and Locus of Causality
There were nine attributional items addressing the three domains of causal attributions: responsibility, stability, and locus of causality (Jones & Kaplan, 2003).
Using a scale from 1 (not at all) to 9 (very much), three questions assessed the extent to which the defendant was responsible for the crime: “Did the defendant intend to commit the crime,” “Was the defendant able to foresee the outcome of his behavior,” and “How responsible was the defendant for his behavior?” (undergraduate sample α = .52; MTurk sample α = .78).
Stability assessed the perceived likelihood of recidivism. Participants were asked about the extent to which the defendant would commit the same crime in the future, any crime in the future, and reform and not commit any crimes in the future (reversecoded). Participants’ responses were made using a scale from 1 (extremely unlikely) to 9 (extremely likely). This scale exhibited relatively low (although acceptable) internal consistency among the MTurk participants (α = .70). The third item did not correlate with the first two items, despite reversecoding the item. Removing the third item resulted in acceptable internal consistency for both the undergraduate sample (α = .79) and the MTurk sample (α = .84). The overall alpha (collapsing across the samples) was .82. We used the composite of the first two items on all analyses.
Three questions assessed the locus of causality: “To what extent was the crime due to personality or the environment,” “To what extent does the crime reflect him or society,” and “To what extent is the crime due to personal or societal reasons?” Responses were made on a scale from 1 (internal) to 9 (external). The reliability for these items was low among the undergraduate sample (α = .65) but high among MTurk participants (α = .89).
Social Desirability Scale–Short Form
The Social Desirability Scale (Crowne & Marlowe, 1960) is a 33item instrument used to assess social desirability response sets. Participants in the present study were presented with the shortened 13item version of this
scale. Participants read each item and decided whether the statement was “true” or “false.” Sample items included: “I sometimes feel resentful when I don’t get my own way,” and “I am always courteous, even to people who are disagreeable.” This scale was not correlated with any key variables in this study and including this scale as a covariate did not alter any results. This scale is not discussed further.
Attention Check and Manipulation Checks
The present study used an attention check adapted from Oppenheimer et al. (2009). Participants were provided with a survey question in which they were instructed to ignore all the possible options and instead click the “other” option and type “I read the instructions.” Any participant who did not type “I read the Instructions” failed the attention check.
For both the race and crime manipulation checks, participants were first asked whether they read a description of a crime and whether they saw the race of the defendant. If they answered “yes,” they were then provided a multiplechoice question in which they were asked to select the correct race and the crime described.
Demographics and MTurk Experience
Participants were presented questions pertaining to their gender identity, age, ethnicity, educational level, and political leanings (1 = extremely liberal, 10 = extremely conservative). Demographic questions were modeled according to the recommendations of Hughes et al. (2016). The updated recommendations (Hughes et al., 2022) were published after data had been collected. MTurk participants were then asked whether they were qualified as “Master Workers” (i.e., those with the highest performance ratings on tasks), and to indicate how many Human Intelligence Tasks (HITs) they had completed on MTurk.
Descriptive Statistics and Correlations for
1
2
3
4
5
6 Locus of
Note. Correlations below the diagonal are the MTurk sample; correlations above the diagonal are the undergraduate sample. Scales: Verdict-Confidence Scale (-10 to +10); Punishment (1 to 9); Discipline (1 to 5); Responsibility (1 to 9); Stability (1 to 9); Locus (1 – internal; 9 – external) * p < .05. ** p < .01. *** p < .001.
Analytic Strategy
We combined the MTurk and SONA samples and treated the combined sample as an independent variable. We created a congruence variable to code White defendant/ embezzlement and Black defendant/gang crime as congruent, and White defendant/gang crime and Black defendant/embezzlement were coded as incongruent. We conducted a 2 (sample) x 2 (racecrime congruence) x 2 (defendant race) x 2 (racial prime) Multivariate Analysis of Covariance (MANCOVA).
The dependent measures included verdictconfidence scale, punishment, discipline, responsibility, stability, and locus of causality. Because the two samples differed regarding the average age, the gender breakdown, and educational level, we included these as covariates. Gender was coded as man/trans man or woman/trans woman. Education level was dichotomized as 4year college degree vs. less than 4 year college degree. Education was a significant covariate for punishment and locus of causality. Age was a significant covariate for verdictconfidence scale, discipline, stability, and locus of causality. Gender was not a significant covariate for any of the dependent variables.
Results
See Table 1 for the means and standard deviations of all dependent variables and interitem correlations. Across both samples, all of the variables were significantly correlated with one another, with the exception of locus of causality. Locus of causality was correlated with most of the dependent variables among the undergraduate sample (all but verdictconfidence, r = .11), but was correlated with only two dependent measures among the MTurk sample (verdict confidence, r = .16; responsibility, r = .17). Compared to the undergraduate sample, the correlations appear to be higher among the MTurk sample which could be a reflection of the lower reliability of some of the measures among the undergraduate sample.
Manipulation Checks
Race Manipulation Check
Participants were asked whether they saw the race of the defendant. If they answered “Yes,” they were directed to a multiplechoice question in which the options included: American Indian, Asian or Asian American, Black or African American, Latino, Native Hawaiian or Pacific Islander, or White. Participants failed the manipulation check if they indicated the wrong race of the defendant (participants in the Black defendant condition who selected a race other than Black and participants in the White defendant condition who selected a race other than White; n = 33).
TABLE 1
Marquez,
Sperry,
Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
We elected to include participants who indicated that they did not see the race of the defendant (n = 79). This decision was made due to the findings in our original sample with all racial/ethnic groups (N = 521), White participants were more likely to say they did not see the race of the defendant compared to nonWhite participants (17% of White participants vs. 8% of nonWhite participants said they did not see the defendant’s race; X2 (1, N = 521) = 5.28, p = .022). This finding suggests that White participants may have attempted to appear “colorblind” on that question, despite encoding the defendant’s race.
Crime Manipulation Check
Similar to the race manipulation check, participants were first asked whether they saw the crime, and followed by answering a multiplechoice question in which the following options were included: insider trading, gangrelated crime, embezzlement, and internet hacking. Participants passed this manipulation check if they indicated the correct crime (embezzlement or gangrelated depending on their condition). We included participants who were in the embezzlement condition who indicated insider trading (n = 20), as that is also considered a Whitecollar crime and would be a test of our key hypotheses.
Of the 434 White participants, 73 participants (68 MTurk and 5 undergraduate) failed at least one of the manipulation checks. The following analyses are conducted on the 361 White participants who passed the both attention check and manipulation checks.
Verdict
Consistent with the hypotheses, there was a main effect of the defendant’s race on the verdictconfidence scale, F(1, 330) = 17.83, p < .001, ηp2 = .05. Participants were more confident in a guilty verdict for the Black defendant (M = 5.52, SE = 0.46) compared to the White defendant (M = 2.87, SE = 0.43). However, the defendant’s race interacted with the sample, F(1, 330) = 5.85, p = .016, ηp 2 = .02. This racial bias on the verdictconfidence scale was driven by the undergraduate sample. In the MTurk sample, there was not a main effect of race on the verdictconfidence scale. However, among the undergraduate sample, participants were more confident in a guilty verdict for the Black defendant compared to the White defendant, F(1, 330) = 19.57, p < .001, ηp2 = .06 (see Figure 1).
There was also an interaction between the racecrime congruence and the defendant’s race on the verdictconfidence scale, F(1, 330) = 8.97, p = .003, η p 2 = .03. However, contrary to our hypothesis, the racecrime congruency effect occurred for the White
defendant but not for the Black defendant (opposite of what was expected). For the White defendant, a racially congruent crime (embezzlement) resulted in more confidence in a guilty verdict than a racially incongruent crime (gang), F(1, 330) = 6.40, p = .012, ηp2= .02. When the defendant was Black, there was not a significant difference between the racecongruent and raceincongruent crimes (see Figure 2).
There was a significant interaction between the sample and the racial prime on the verdictconfidence scale, F(1, 330) = 4.29, p = .039, ηp2 = .01. In the undergraduate sample, the prime slightly increased confidence in a guilty verdict (4.65 vs. 3.58, n.s.). In the MTurk sample, the prime led to slightly lower confidence in a
FIGURE 1
Race X Sample Interaction on Verdict-Confidence Scale (N = 361)
FIGURE 2
Congruency X Race Interaction on Verdict-Confidence Scale (N = 361)
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
Racial Prime and Race-Crime Congruency | Marquez, Sperry, Griffiths, and Stuart
guilty verdict (5.04 vs. 3.51, n.s.). However, these simple main effects were not statistically significant. There was also a significant interaction between the sample and the racecrime congruency, F(1, 330) = 5.22, p = .023, ηp2 = .02. In the undergraduate sample, racecrime congruence led to slightly more confidence in a guilty verdict (4.98 vs. 3.26, n.s.), but the opposite occurred among the MTurk participants (3.69 vs. 4.86, n.s.). Again, the simple main effects were not statistically significant.
Punishment
There was a main effect of the sample on levels of punishment, with MTurk participants giving harsher
Sample x Race Interaction on Punishment (N =
Race x Prime x Sample Interaction on Punish-
n = 193)
punishments overall ( M = 5.40, SE = 0.19) compared to the undergraduate participants ( M = 4.61, SE = 0.22), F(1, 330) = 5.36, p = .021, ηp2 = .02. Consistent with our hypotheses and with the findings on the verdict confidence scale, there was a main effect of the defendant’s race on punishment, F(1, 330) = 13.04, p < .001, ηp2= .04. The Black defendant was given harsher punishment compared to the White defendant. However, this racial bias was driven largely by the undergraduate participants. There was a significant interaction between the sample and the defendant’s race, F(1, 330) = 14.06, p < .001, ηp2= .04. Specifically, the simple main effects revealed the difference between the White and Black defendant was only significant for the undergraduate sample, F(1, 330) = 24.08, p < .001, ηp2= .07 (see Figure 3). There was a significant interaction between the racial prime and the defendant’s race on judgments of punishment, F(1, 330) = 4.89, p = .028, ηp2= .02. Simple main effects revealed when participants were primed, the Black defendant was given harsher punishment compared to the White defendant, F(1, 330) = 18.94, p < .001, ηp2 = .05. When participants were not primed there was no difference between the punishment judgments for the Black and White defendants. However, this interaction should be understood in light of the significant threeway interaction between the racial prime, the defendant’s race, and the sample, F(1, 330) = 4.05, p = .045, ηp2 = .01. To better understand this interaction, we split the file by sample, and examined the interaction between the racial prime and the defendant’s race on punishment. In the undergraduate sample, the interaction between the racial prime and the defendant’s race was significant, F(1, 153) = 6.59, p = .011, ηp 2 = .04. Participants who
Race x Prime x Sample Interaction on Punishment Among Undergraduate Participants (n = 168)
FIGURE 3
361)
FIGURE 4a
ment Among MTurk Participants (
FIGURE 4b
Marquez, Sperry, Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
were racially primed gave harsher punishments for the Black defendant compared to the White defendant. The interaction was not statistically significant in the MTurk Sample (see Figures 4a and 4b).
There was an interaction between the racial prime and the sample, F(1, 330) = 5.06, p = .025, ηp2 = .02. In the MTurk sample, the prime led to significantly lower punishment ratings (M = 5.09, SE = 0.24) compared to the no prime condition (M = 5.70, SE = 0.23), F(1, 330) = 4.32, p = .038, ηp2= .01. The opposite pattern occurred in the undergraduate sample, but the difference was not statistically significant.
Discipline
There was a main effect of the defendant’s race on discipline, F(1, 330) = 11.47, p < .001, ηp2 = .03. Consistent with the hypotheses and the findings reported above on verdict and punishment, the Black defendant was given a harsher level of discipline (M = 3.92, SE = 0.07) compared to the White defendant (M = 3.58, SE = 0.07). However, the main effect should be qualified by a significant interaction between the defendant’s race and the sample, F(1, 330) = 4.60, p = .033, ηp2 = .01. Simple main effects revealed the racial bias against the Black defendant was significant among the undergraduate sample, F(1, 330) = 13.58, p < .001, ηp2 = .04. In the MTurk sample, the difference between the Black and White defendant was not statistically significant (see Figure 5).
Responsibility
The internal consistency for Responsibility items was low in the undergraduate sample (α = .52). Removing items did not seem to improve reliability. Therefore, the following results should be viewed with caution.
There was an interaction between race crime congruency and the defendant’s race on attributions of responsibility, F(1, 330) = 11.35, p < .001, ηp2 = .03. For the White defendant, the racestereotypic crime led to higher ratings of responsibility, F(1, 330) = 5.24, p = .023, ηp2 = .02. Contrary to the hypotheses, the Black defendant who committed a crime that was congruent with racial stereotypes was perceived as less responsible for the crime compared to a Black defendant who committed an incongruent crime, F(1, 330) = 6.22, p = .013, ηp2 = .02 (see Figure 6).
Stability
There was an interaction between racecrime congruence and the defendant’s race on judgments of stability, F (1, 330) = 15.77, p < .001, η p 2 = .05. Consistent with the race crome congurnecy effect hypotheses, when the Black defendant committed a stereotypecongruent crime (gang) he was judged more likely to
reoffend compared to when he committed a stereotypeincongruent crime (embezzlement), F(1, 330) = 6.71, p = .010, ηp2 = .02. However, when the White defendant committed a racially congruent crime, he was perceived as being less likely to reoffend compared to the racially incongruent crime, F(1, 330) = 9.46, p = .002, ηp2= .03 (see Figure 7).
Locus of Causality
Similar to the responsibility items, there was low reliability on the locus of causality items among the undergraduate participants (α = .65), but not the MTurk participants (α = .89). The following results should be viewed with caution due to the lower reliability in the undergraduate sample.
FIGURE 5
Race x Sample Interaction on Discipline (N = 361)
FIGURE 6
Congruency x Race Interaction on Responsibility (N = 361)
Racial Prime and Race-Crime Congruency | Marquez, Sperry, Griffiths, and Stuart
There was a main effect of the race crime congruence on locus of causality, F(1, 330) = 12.61, p < .001, ηp2= .04. Consistent with the hypotheses, when a crime was congruent with racial stereotypes participants perceived the crime as more internally caused (M = 4.96, SE = 0.13) than when the crime was incongruent with racial stereotypes (M = 5.63, SE = 0.14). This main effect should be qualified by the interaction between racecrime congruence and the defendant’s race, F(1, 330) = 18.52, p <.001, ηp2= .05.
Contrary to our hypotheses, the racecrime congruency effect occurred only for the White defendant, F(1, 330) = 33.35, p <.001, ηp2= .09, but not for the Black defendant (see Figure 8). For the White defendant,
if a racially congruent crime (embezzlement) was committed it was perceived as more internally caused compared to a racially incongruent crime (gang). The racial congruence did not impact the locus of causality for the Black defendant.
Discussion
The present study had three hypotheses: (a) Black defendants would be judged more harshly compared to White defendants; (b) Defendants who commit racially congruent crimes would be judged more harshly than defendants who commit racially incongruent crimes (racecrime congruency effect); and (c) Priming participants to think about their own racial identity (a novel implementation of racial priming) would attenuate the racecrime congruency effect.
Hypothesis 1:
Racial Bias in Judgments of White vs. Black Defendants
Our findings provided evidence of racial biases consistent with decades of research (e.g., Kovera, 2019; Minton & Zeng, 2021; Mitchell et al., 2005). When the defendant was Black participants had more confidence in a guilty verdict, gave harsher punishments, and harsher disciplinary action. However, these racial biases were all driven by the undergraduate sample and were not found among the MTurk sample. The undergraduate sample was gathered from a largely White institution (U.S. Department of Education, n.d.). This lack of diversity in an individual’s immediate environment may explain the strong racial biases among the undergraduate sample. However, it is unknown whether the MTurk sample also included individuals from a majority White population or immediate environment. In addition, the undergraduate sample also was drawn from a state with a large population of religious adults (Pew Research Center, 2024). MTurk samples have been found to be less religious than the general population (Burnham et al., 2018; Lewis et al., 2015). Our study controlled for age, gender, and educational status, but religion or religiosity was not measured, and which may have accounted for the racial biases observed in the present study. Previous studies have found positive correlations between measures of religiosity and racial prejudice (Batson et al., 1986; Hall et al., 2010; Johnson et al., 2010).
Another notable difference between the two samples was the bivariate correlations between the dependent measures (Table 1). Specifically, locus of causality was correlated with many dependent variables among the undergraduate sample but was not highly correlated among the MTurk sample. The role that locus of causality has in these racial biases might be worth exploring. Some research has suggested that internal
WINTER 2024
FIGURE 8
Race x Congruency Interaction on Locus of Causality (N = 361).
FIGURE 7
Congruency x Race Interaction on Stability (N = 361)
Marquez, Sperry, Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
attributions (such as genetics) can lead people to believe that the defendant did not have control over committing the crime, but at the same time, increased the perceived likelihood that the defendant would reoffend (Cheung & Heine, 2015). Further, Gordon (1990) found that racestereotypic crimes were perceived as internal, but the connection to case outcome was more complicated. The mediating role of causal attributions in judgments of defendants who commit race stereotypic crimes may partially explain the differences observed in our samples. Continuing to examine racial biases in our criminal justice system is imperative. On the surface, it may seem intuitive, perhaps obvious, that racial biases in mock jury decisionmaking would be found in the present study. Interestingly, we found these racial biases only among the undergraduate participants, and not among the MTurk participants. This finding is novel and warrants further investigation.
Hypothesis 2:
Replicating the Race-Crime Congruency Effect
The racecrime congruency effect was partially replicated on some of the dependent variables but not all. Recent studies found the racecrime congruency effect for Black defendants but not White defendants (e.g., Maeder et al., 2016; Petsko & Bodenhausen, 2019). We replicated this finding only for the attributions of stability. When a Black defendant committed a racially stereotypical crime (gang crime), participants believed he was more likely to reoffend compared to when he committed a counterstereotypic crime (embezzlement). Contrary to our hypotheses, for the verdictconfident scale and locus of causality, the race crime congruence effect emerged only for the White defendant. That is, the White defendant who committed embezzlement led to higher confidence in a guilty verdict and more internal attributions compared to the White defendant who committed gang crime. Bitter and colleagues recently reviewed the literature on the racecrime congruency effect and found that although most studies find the effect is stronger among Black defendants, some studies did find the racecrime congruency effect only for White defendants (Bitter et al., 2023). Several moderators of this effect were identified, and we return to this issue below.
Hypothesis 3: Racial Priming
We expected the racial prime to reduce racial biases, and in particular, to reduce the racecrime congruency effect. This is not the impact that the racial prime had. In fact, among the undergraduate participants, the racial prime actually led to harsher judgments for the Black defendant compared to the White defendant, thus increasing racial bias.
A possible explanation for the prime not reducing racial biases may be attributed to the racial prime manipulation used in the present study. Participants were primed by filling out the Collective SelfEsteem Scale Race. While this scale likely increased the salience of participants’ race it may have also inadvertently increased racial pride . Although racially priming participants to think about their own race could be analogous to race salience (thus highlighting one’s own potential biases), it is possible that racial priming with the Collective SelfEsteem Scale Race actually made participants think about the importance of their racial identity. This may have led to an ingroup favoritism which could potentially explain the lower punishment given to the White defendant when White participants were primed. It is interesting to note that the racial prime increased the racial bias only among the undergraduate sample. This may suggest that the racial prime had an additive effect (among participants with high levels of racial bias, the racial prime exacerbated those preexisting racial biases).
Our study also found that, among the MTurk participants, the racial prime led to lower punishment ratings overall. This is more consistent with the findings of Schuller et al. (2009), who found that asking participants to reflect on their own biases led to fewer guilty verdicts. Perhaps among participants with lower racial biases (as with our MTurk participants) the racial prime led to this more deliberative mindset.
Limitations
An important limitation of the present study was the large number of participants who failed the attention and manipulation checks. The manipulation checks in the present study were placed after the dependent measures and after the attention check. Participants may have failed the manipulation check due to memory decay. The manipulation may have impacted responses on the main dependent variables, but by the time participants arrived at the manipulation checks they did not accurately remember the defendant’s race or crime described. Another likely possibility is participants did not encode the race of the defendant or the crime description. Since it cannot be known which of these occurred (memory decay or lack of encoding), we decided to remove participants who failed either manipulation check. The manipulations were rather subtle and future studies might consider increasing the salience of these manipulations. Recent studies have noted concerns with manipulation checks (e.g., Fayant et al., 2017; Varaine, 2022). Studies have suggested manipulation checks should be included in pilot testing materials instead of in the experimental procedures (Ejelov & Luke, 2020).
Participants in the present study were asked to make judgments that would reflect those of the “average American juror.” The decision to include this instruction was based on Petsko and Bodenhausen (2019), who found racial biases only when the instructions were framed accordingly. Similarly, in the present study, the purpose was to reduce social desirability by allowing participants’ racial biases to emerge under the guise of the “average American jurors’” beliefs rather than their own. However, we cannot be certain to what extent this instruction reduced social desirability. It is possible that participants’ responses did actually reflect their views about racial biases in mainstream America. Petsko and Bodenhausen acknowledged this possibility as well. Although it is possible that the undergraduate sample had higher racial biases compared to the MTurk sample (implications of this discussed further below), perhaps the differences found between the two samples reflect differences in their perspectives of mainstream America rather than their own beliefs.
Data were collected from two samples: Amazon’s Mechanical Turk (MTurk) and an undergraduate sample collected through the research participation pool at the researchers’ institution. These samples might limit the generalizability of our results. MTurk workers tend to be similar to U.S. demographics in terms of gender and race, but they do differ in terms of religious affiliation, with larger numbers of MTurk workers identifying as agnostic or atheist compared to the general population (Burnham et al., 2018; Lewis et al., 2015). The undergraduate sample suffers from generalizability in terms of age and gender (this study had mostly young female participants). We combined across these two samples and included only White participants. We recognize that each of these samples has limited generalizability, and combining the samples does not necessarily negate this limitation. By combining the two samples, we had a wider age demographic and roughly even gender split (we statistically controlled for demographic variables that differed across the samples). It would be important to replicate these findings using additional samples, and perhaps including religiosity as a variable, as the undergraduate sample was drawn from a state with a large number of religious adults (Pew Research Center, 2024). Roughly 97% of students enrolled at the institution where data was collected are instate students (Weber State University, n.d.). This could possibly explain the higher racial biases exhibited by the undergraduate participants compared to the MTurk participants. The crime descriptions employed in the present study were embezzlement (White stereotypic) and gang crime (Black stereotypic). These crime descriptions have frequently been used to examine the racecrime
congruency effect. White collar crimes tend to be associated with White defendants and more violent crimes tend to be associated with Black defendants. However, it is worth noting that these crimes also differ in terms of the likely age of the offender, the reasons for committing the crime, and the target of the crime. Additionally, as a function of different types of crimes, the descriptions of each crime were not identical. This introduces a possible threat to internal validity, as the crimes differ on attributes other than just the racial stereotypicality of the crimes.
Future Directions and Implications
We found stark differences between the MTurk sample and the undergraduate sample in terms of racial biases as well as the impact of racial priming. After controlling for age, gender, and education level, racial biases emerged among the undergraduate sample but not in the MTurk sample. In their systematic review of the racecrime congruency literature, Bitter et al. (2023) noted several moderators of the racecrime congruency effect, including victim race, presence of jury instructions, and evidence strength. The only sample characteristic identified by Bitter and colleagues was the participant’s race. Our findings suggest that there may be additional sample characteristics that act as moderators on the racecrime congruency effect. To fully understand the workings of the racecrime congruency effect, sample characteristics beyond participant race should be explored.
The present study used measures obtained from previous research that examined the race crime congruency effect (e.g., Jones & Kaplan, 2003; Petsko & Bodenhausen, 2019). However, future studies could utilize the Perceptions of Criminal Defendants Scale (PCDS; Crawley et al., 2017). Sample items include: “If guilty, what type of sentence would you recommend?,” “I think this individual would likely commit a crime in the future,” and “I believe this individual probably has a prior criminal record” (Crawley et al., 2017). This scale is very concise, while still including key constructs of interest. Given the large number of participants who failed the attention check in the present study, using a shorter scale for the key dependent variables might mitigate participant fatigue and attention check failure problems in future studies.
Although the racial prime used in the present study did not reduce racial biases, racial priming should be investigated further to understand its potential impact on racial biases and decisionmaking in the courtroom. There are several fruitful areas of research: (a) Our findings suggest that the impact of the racial prime may have a different effect depending on participants’ level of racial bias. The fact that we found a different Racial
WINTER 2024
Marquez, Sperry, Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
effect among the undergraduate sample compared to the MTurk sample suggests that among participants with high levels of racial bias (as we found with our undergraduate sample), the racial prime might increase those preexisting racial biases. (b) It is also important to understand the impact that a racial prime might have on participants with various racial identities. In the present study, we included only White participants due to not having enough nonWhite participants to make meaningful comparisons across racial groups. But certainly racially priming White participants is qualitatively different than racially priming nonWhite participants.
(c) Finding additional ways to unconsciously prime participants to think about their own race would also be important. Schuller et al. (2009) asked participants to reflect on their own racial biases. The prime we utilized in the present study was the Collective SelfEsteem Scale (Luhtanen & Crocker, 1992). Replicating and extending the findings of the present study with additional primes would add to this body of research.
Jurors may be inadvertently primed to think of their own race during voir dire questioning or through pretrial questionnaires. Jurors could also be potentially primed during a racesalient trial or if a juror happens to be the only person of their ethnicity/race on the jury panel. Learning more about racial priming and its effect on case outcomes and juror bias will prove beneficial when working to mitigate discrepancies during a jury trial.
Conclusion
Racial biases continue to exist in every step of the criminal justice system, from arrest to sentencing. Black Americans make up 12% of the U.S. population but make up about 27% of all arrests (Federal Bureau of Investigations, 2019). Additionally, Black suspects are 3.23 times more likely than White suspects to die at the hands of a police officer (Schwartz & Jahn, 2020). Racial biases in jury decision making and sentencing decisions have been welldocumented as well. Compared to White Americans, Black Americans are more likely to be falsely convicted of serious crimes. Black and Hispanic defendants receive harsher sentences compared to White defendants (Kovera, 2019).
It is imperative to continue investigating racial biases in jury decisionmaking. Our findings suggest that racial biases are not a foregone conclusion. Racial attitudes are impacted by the political landscape and evolving norms (thus revisiting these findings continues to add to this literature). In fact, in a systematic review of the racecrime congruency literature, Bitter et al. (2023) found that the older studies consistently reported a racecrime interaction, yet some newer studies have failed to replicate this finding. A metaanalysis confirmed that the
effect size of racial bias decreased over time (from the 1980s to the 2000s; Mitchell et al., 2005). More recently, Sawyer and Gampa (2018) found that both implicit and explicit racial attitudes were less proWhite during the BLM movement than preBLM.
Revisiting racial biases, perhaps with an eye for moderators such as sample characteristics, continues to be a crucial endeavor. This is especially true given the juxtaposition of the BLM movement against the recent mainstream antiBlack rhetoric (e.g., Paul, 2019; Pew Research Center, 2023).
References
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 71(2), 230–244. https://doi.org/10.1037/0022-3514.71.2.230
Batson, C. D., Flink, C. H., Schoenrade, P. A., Fultz, J., & Pych, V. (1986). Religious orientation and overt versus covert racial prejudice. Journal of Personality and Social Psychology, 50(1), 175–181. https://doi.org/10.1037/0022-3514.50.1.175
Bitter, A. N., Smith, O. K. H., Michalski, N. D., Freng, S. (2023). Examining the crossover interaction of the race-crime congruency effect: A systematic review. Race and Social Problems, 15, 408–427. https://doi.org/10.1007/s12552-022-09376-1
Bucolo, D. O., & Cohn, E. S. (2010). Playing the race card: Making race salient in defence opening and closing statements. Legal and Criminological Psychology, 15(2), 293–303. https://doi.org/10.1348/135532508X400824
Burnham, M. J., Le, Y. K., & Piedmont, R. L. (2018). Who is MTurk? Personal characteristics and sample consistency of these online workers. Mental Health, Religion, & Culture, 21(9–10), 934–944. https://doi.org/10.1080/13674676.2018.1486394
Cheung, B. Y., & Heine, S. J. (2015). The double-edged sword of genetic accounts of criminality: Causal attributions from genetic ascriptions affect legal decision-making. Personality and Social Psychology Bulletin, 41(12), 1723–1738. https://doi.org/10.1177/0146167215610520
Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2007a). The influence of stereotypes on decisions to shoot. European Journal of Social Psychology, 37(6), 1102–1117. https://doi.org/10.1002/ejsp.450
Correll, J., Park, B., Judd, C. M., Wittenbrink, B., Sadler, M. S., & Keesee, T. (2007b). Across the thin blue line: Police officers and racial bias in the decision to shoot. Journal of Personality and Social Psychology, 92(6), 1006–1023. https://doi.org/10.1037/0022-3514.92.6.1006
Crawley, D., Ramos, C., & Leyva, J. (2017). Perceptions of criminal defendants scale: Development and validation. Psi Chi Journal of Psychological Research, 22(4), 307–319. https://doi.org/10.24839/2325-7342.jn22.4.307
Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24(4), 349–354. https://doi.org/10.1037/h0047358
Dovidio, J. F., & Gaertner, S. L. (2004). Aversive racism. In M. P. Zanna (Ed.), Advances in experimental social psychology, Vol. 36, (pp 1–52). Elsevier Academic Press. https://doi.org/10.1016/S0065-2601(04)36001-6
Eberhardt, J. L., Davies, P. G., Purdie-Vaughns, V. J., & Johnson, S. L. (2006). Looking deathworthy: Perceived stereotypicality of Black defendants predicts capital-sentencing outcomes. Psychological Science, 17(5), 383–386. https://doi.org/10.1111/j.1467-9280.2006.01716.x
Ejelov, E. & Luke, T. J. (2020). “Rarely safe to assume”: Evaluating the use and interpretation of manipulation checks in experimental social psychology. Journal of Experimental Social Psychology, 87, Article 103937. https://doi.org/10.1016/j.jesp.2019.103937
Fayant, M., Sigall, H., Lemonnier, A., Retsin, E., & Alexopoulos, T. (2017). On the limitations of manipulation checks: An obstacle toward cumulative science. International Review of Social Psychology, 30(1), 125–130. https://doi.org/10.5334/irsp.102
Federal Bureau of Investigations (2019). Crime in the United States 2019 https://ucr.fbi.gov/crime-in-the-u.s/2019/crime-in-the-u.s.-2019/home
Fisher, R. J (1993). Social desirability bias and the validity of indirect
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
questioning. Journal of Consumer Research, 20(2), 303–315 https://doi.org/10.1086/209351
Gamblin, B. W., & Kehn, A. (2021). Race salience and attorney statements: The unique role of defense opening statements and closing arguments. Current Psychology, 40, 2621–2633. https://doi.org/10.1007/s12144-020-01147-8
Gordon, R. A. (1990). Attributions for blue-collar and white-collar crime: The effects of subject and defendant race on simulated juror decisions. Journal of Applied Social Psychology, 20(12), 971–983. https://doi.org/10.1111/j.1559-1816.1990.tb00385.x
Gordon, R. A., Bindrim, T. A., McNicholas, M. L., & Walden, T. L. (2001). Perceptions of blue-collar and white-collar crime: The effect of defendant race on simulated juror decisions. The Journal of Social Psychology, 128(2), 191–197. https://doi.org/10.1080/00224545.1988.9711362
Gross, S.R., et al. (2022). Race and wrongful convictions in the United States 2022. National Registry of Exonerations, Sept. 2022. https://www.law.umich. edu/special/exoneration/Documents/Race%20Report%20Preview.pdf
Hall, D. L., Matz, D. C., & Wood, W. (2010). Why don’t we practice what we preach? A meta-analytic review of religious racism. Personality and Social Psychology Review, 14(1), 126–139. https://doi.org/10.1177/1088868309352179
Hughes, J. L., Camden, A. A., & Yangchen, T. (2016). Rethinking and updating demographic questions: Guidance to improve descriptions of research samples. Psi Chi Journal of Psychological Research, 21(3), 138–151. https://doi.org/10.24839/2164-8204.JN21.3.138
Hughes, J. L., Camden, A. A., Yangchen, T., Smith, G. P. A., Rodriquez, M. M. D., Rouse, S. V., McDonald, C. P., & Lopez, S. (2022). Guidance for researchers when using inclusive demographic questions for surveys: Improved and updated questions. Psi Chi Journal of Psychological Research, 27(4), 232–255. https://doi.org/10.24839/2325-7342.JN27.4.232
Johnson, M. K., Rowatt, W. C., & LaBouff, J. (2010). Priming Christian religious concepts increases racial prejudice. Social Psychological and Personality Science, 1(2), 119–126. https://doi.org/10.1177/1948550609357246
Jones, C. S., & Kaplan, M. F. (2003). The effects of racially stereotypical crimes on juror decision-making and information-processing strategies. Basic and Applied Social Psychology, 25(1), 1–13. https://doi.org/10.1207/s15324834basp2501_1
Kovera, M. B. (2019). Racial disparities in the criminal justice system: Prevalence, causes, and a search for solutions. Journal of Social Issues, 75(4), 1139–1164. https://doi.org/10.1111/josi.12355
Lewis, A. R., Djupe, P. A., Mockabee, S. T., & Wu, J. S. (2015). The (non) religion of Mechanical Turk workers. Journal for the Scientific Study of Religion, 54(2), 419–428. https://doi.org/10.1111/jssr.12184
Livingston, N. J., & Gurung, R. A. (2019). Trumping racism: The interactions of stereotype incongruent clothing, political racial rhetoric, and prejudice toward African Americans. Psi Chi Journal of Psychological Research, 24(1), 52–60. https://doi.org/10.24839/2325-7342.jn24.1.52
Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of one’s social identity. Personality and Social Psychology Bulletin, 18(3), 302–318. https://doi.org/10.1177/0146167292183006
Maeder, E. M., Yamamoto, S., McManus, L. A., & Capaldi, C. A. (2016). Race-crime congruency in the Canadian context. Canadian Journal of Behavioural Science, 48(2), 162–170. https://doi.org/10.1037/cbs0000045
McManus, L., Maeder, E., & Yamamoto, S. (2018). The role of defendant race and racially charged media in Canadian mock juror decision making. Canadian Journal of Criminology & Criminal Justice, 60(2), 266–295. https://doi.org/10.3138/cjccj.2017.0035.r1
Mekawi, Y., & Bresin, K. (2015). Is the evidence from racial bias shooting task studies a smoking gun? Results from a meta-analysis. Journal of Experimental Social Psychology, 61, 120–130. http://dx.doi.org/10.1016/j.jesp.2015.08.002
Minton, T. D., & Zeng, Z. (2021). Jail Inmates in 2020 – Statistical Tables. Bureau of Judicial Statistics. https://bjs.ojp.gov/library/publications/jail-inmates-2020-statistical-tables
Mitchell, T. L., Haw, R. M., Pfeifer, J. E., & Meissner, C. A. (2005). Racial bias in mock juror decision-making: A meta-analytic review of defendant treatment. Law and Human Behavior, 29(6), 621–637. https://doi.org/10.1007/s10979-005-8122-9
Offit, A. (2021). Race-conscious jury selection. Ohio State Law Journal, 82(2), 201–247.
Journal of Experimental Social Psychology, 45(4), 867–872. http://doi.org/10.1016/j.jesp.2009.03.009
Paul, J. (2019). ‘Not black and white, but black and red’: Anti-identity identity politics and #AllLivesMatter. Ethnicities, 19(1), 3–19. https://doi.org/10.1177/1468796818791661
Petsko, C. D., & Bodenhausen, G. V. (2019). Race–crime congruency effects revisited: Do we take defendants’ sexual orientation into account? Social Psychological and Personality Science, 10(1), 73–81. https://doi.org/10.1177/1948550617736111
Pew Research Center (2023, September 26). Black Americans’ experience with news. https://www.pewresearch.org/journalism/2023/09/26/blackamericans-experiences-with-news/
Pew Research Center. (2024, June 17). Religous landscape study adults in Utah . Pew Research Center. https://www.pewresearch.org/topic/religion/ Sawyer, J., & Gampa, A. (2018). Implicit and explicit racial attitudes changed during Black Lives Matter. Personality and Social Psychology Bulletin, 44(7), 1039–1059. https://doi.org/10.1177/0146167218757454
Schuller, R.A., Kazoleas, V., & Kawakami, K. (2009). The impact of prejudice screening procedures on racial bias in the courtroom. Law and Human Behavior, 33(4), 320–328. https://doi.org/10.1007/s10979-008-9153-9
Schwartz, G. L., & Jahn, J. L. (2020). Mapping fatal police violence across U.S. metropolitan areas: Overall rates and racial/ethnic inequities, 2013–2017. PLoS ONE, 15(6), 1–16. https://doi.org/10.1371/journal.pone.0229686
Skorinko, J. L., & Spellman, B. A. (2013). Stereotypic crimes: How group-crime associations affect memory and (sometimes) verdicts and sentencing. Victims & Offenders, 8(3), 278–307. https://doi.org/10.1080/15564886.2012.755140
Sommers, S. R., & Ellsworth, P. C. (2000). Race in the courtroom: Perceptions of guilt and dispositional attributions. University of Michigan Law School Scholarship Repository, 26 1366–1379.
Sommers, S. R., & Ellsworth, P. C. (2001). White juror bias: An investigation of prejudice against Black defendants in the American courtroom. Psychology, Public Policy, and Law, 7(1), 201–229. https://doi.org/10.1037/1076-8971.7.1.201
Sunnafrank, M., & Fontes, N. E. (1983). General and crime related racial stereotypes and influence on juridic decisions. Cornell Journal of Social Relations, 17(1), 1–15.
U.S. Department of Education (n.d.). Weber State University. US Department of Education College Scorecard. https://collegescorecard.ed.gov/school/?230782-Weber-State-University Varaine, S. (2022). How dropping subjects who failed manipulation checks can bias your results: An illustrative case. Journal of Experimental Political Science, 10(2), 299–305. https://doi.org/10.1017/XP
Weber State University (n.d.). About WSU. Weber State University. https://www.weber.edu/AboutWSU/
Author Note
Evangelina T. Marquez https://orcid.org/0009000730591867
Kathryn Sperry https://orcid.org/0000000264924304
Evangelina Marquez graduated with her BS and recently graduated from the Master of Legal Studies program at the University of Utah.
This research was supported by an Undergraduate Research Grant awarded to the first author by the Office of Undergraduate Research at Weber State University. The Institutional Review Board at Weber State University approved this project (IRBAY2122165), and all ethical guidelines for research with human subjects were followed. We have no conflicts of interest to disclose.
Positionality statement: Evangelina Marquez identifies as a nondisabled straight cisgender American Latina woman. Evangelina is a firstgeneration college student. Kathryn Sperry identifies as an educated straight cisgender White woman. Emilia Griffiths identifies as a nondisabled straight cisgender White woman. Jack Stuart identifies as a straight cisgender White man. The authors recognize the influence of these intersecting identities on their worldviews and their interpretation of the findings and the implications of this study.
Correspondence concerning this article should be addressed to the faculty mentor: Kathryn Sperry, Weber State University, 1299 Edvalson St., Ogden, UT, 84408, United States. Email: Kathrynsperry@weber.edu Racial Prime and Race-Crime Congruency | Marquez, Sperry, Griffiths, and Stuart
Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009). Instructional manipulation checks: Detecting satisficing to increase statistical power.
Marquez, Sperry, Griffiths, and Stuart | Racial Prime and Race-Crime Congruency
APPENDIX
Race x Crime Manipulation
Case summary:
The defendant was charged with assaulting two men in a parking lot on the basis of gang membership. In his defense, the defendant stated that his own gang membership had nothing to do with the assault, but that he was instead protecting his girlfriend from being harassed by these men.
Case summary:
The defendant was charged with approving loans for his girlfriend and taking bank funds for his own use (i.e., buying real estate and paying off debts). In his defense, the defendant stated that the company president approved these loans and funds for bank purchases.
Cybermonitoring: What Is It, Who Does It, and Why?
Darcey N. Powell* and Abbie Joseph Department of Psychology, Roanoke College
ABSTRACT.
This study examined emerging adults’ cybermonitoring behaviors before, during, and after a romantic interaction. Given the abundance of information available online, behaviors representative of the less intrusive end of the continuum of cyberstalking (e.g., clicking on profile, viewing tagged photos) have become a common aspect of romantic interactions. Thus, this project focused predominately on those more mundane aspects of information seeking (i.e., cybermonitoring). Emerging adults between the ages of 18 and 29 (N = 240) and who had at least one previous romantic interaction completed an online survey regarding their experiences with cybermonitoring, engagement in cybermonitoring, and their motives for cybermonitoring. Analyses revealed that 94.4% of participants had engaged in cybermonitoring; often using Facebook (74.6%) and Instagram (75.4%) to engage in cybermonitoring on their perspective, current, or former romantic partners. Furthermore, they reported engaging in behaviors differentially, Wilks’ λ = .19, F(17, 203) = 49.85, p < .001, ηp 2 = .81, with less intrusive behaviors (e.g., clicking on profile, viewing tagged photos) commonly used than more intrusive cyberstalking behaviors (e.g., creating fake profile, using location). Additionally, participants reported engaging in more cybermonitoring during their romantic relationship than before, t(191) = 5.78, p < .001, g = 0.35, or after, t(194) = 8.41, p < .001, g = 0.60, and more frequently before the interaction than after, t (192) = 3.70, p < .001, g = 0.21. Lastly, participants predominately reported engaging in cybermonitoring after a breakup for reasons related to curiosity about their expartner. How this information compares to prior research on cyberstalking is discussed. Ultimately, we posit that cybermonitoring may be a more appropriate term for representing emerging adults’ online behaviors revolving around romantic interactions.
Keywords: cyberstalking, cybermonitoring, romantic relationship, emerging adults, social media
Amajority of emerging adults have at least one social media account and use one or more of the platforms at least daily (Jones, 2023). Social media has given individuals access to many people’s
Preregistration and Open Materials badges earned for transparent research practices. The preregistration can be viewed at https://osf.io/rkm9u/. Materials are available at https://osf.io/rkm9u/
photos, whereabouts, life events, dating history, and so forth with just a click of a button (Frampton & Fox, 2021; Kwok & Wescott, 2020; McEwan, 2013). Sometimes, the frequency with which the click of a button occurs
and how that information is used can be concerning and dangerous for recipients (e.g., Burke et al., 2011; Drebing et al., 2014; Marshall, 2012). However, given the vast amount of information available on social media (McEwan, 2013), the click of a button also has the potential to be informative for engagers when seeking information about romantic partners (Frampton & Fox, 2021; Kwok & Wescott, 2020).
Cyberstalking is often conceptualized as a severe form of obsessive relational intrusion and unwanted pursuit behaviors (Cupach & Spitzberg, 1998; De Smet et al., 2011; LanghinrichsenRohling et al., 2000; Marcum & Higgins, 2021). However, the broad category of cyberstalking consists of a variety of behaviors. The behaviors most commonly reported by individuals are typically the least intrusive behaviors, such as checking status updates, wall posts, photos, and friend lists (Lyndon et al., 2011; Marshall, 2012). Significantly less common are the more intrusive cyberstalking behaviors, such as harassing their expartners (Lyndon et al., 2011; Marshall, 2012). Yet, cyberstalking research has predominately focused on the more intrusive behaviors (e.g., Kwok & Wescott, 2020; Marcum & Higgins, 2021).
The current study examined the more commonly reported and less intrusive online informationcollecting behaviors (i.e., cybermonitoring) before, during, and after a romantic interaction. The decision to use the term cybermonitoring, rather than cyberstalking or other terms which have been used in the literature, is supported by Frampton and Fox’s (2021) decision tree. Furthermore, individual differences in cybermonitoring were examined, as much of the cyberstalking research on individual differences was conducted when Facebook was the main form of social media (e.g., Fox et al., 2014: Lyndon et al., 2011; Marshall et al., 2013). Thus, we wondered if the findings may extend to the less intrusive cybermonitoring behaviors and to other social media platforms.
Relevant Past Research
The online informationseeking topic that has received the most attention is cyberstalking. However, most of the past research has focused on its occurrence via Facebook and after a romantic interaction’s dissolution. Platformwise, it is intuitively likely that these findings extend to individuals’ engagement in cyberstalking (and less intrusive cybermonitoring behaviors) on Instagram, Snapchat, and other social media sites as well, but little research has explored this possibility (e.g., Huie et al., 2023). Moreover, individuals can engage in cyberstalking or cybermonitoring behaviors before and during a romantic interaction as well (Kwok & Wescott, 2020). For example, beyond the proportion who use social
media to track their expartners (Marshall, 2012), many also use social media to track the activities of their current partners and evaluate prospective partners (Fox et al., 2014; Kwok & Wescott, 2020).
Tong (2013) reported three types of information that individuals seek about their ex partner when cyberstalking: their general social activities, their new romantic connections, and modes of direct communication (Tong, 2013). This aligns well with individuals’ own reports of why they have cyberstalked; specifically reporting that their motivation stemmed from wanting to know where the individual was, what they have been doing, or who they have been with recently (Marcum et al., 2016). On the other hand, victims of cyberstalking report believing their expartner’s motivation stemmed from being jealous, wanting to seek revenge, or wanting to reinitiate the romantic interaction (Drebing et al., 2014). Given that the term cyberstalking often insinuates the more intrusive behaviors (e.g., checking and using location information), these may not be the reasons why individuals engage in the less intrusive behaviors (e.g., clicking profile, looking at tagged photos). Relatedly, Kwok and Wescott (2020), identified several reasons why individuals may engage with technology for their prospective, current, and prior relationships (e.g., connection with a potential partner, displays of affection, viewing similarities and differences, jealousy, post breakup actions).
Theoretical Foundation
A key theory to understanding individuals’ desire to gain information—whether that be through cybermonitoring behaviors, more invasive cyberstalking behaviors, or even taking a more direct approach of asking a pointed question—is uncertainty reduction theory (Berger, 1979; Berger & Bradac, 1982; Frampton & Fox, 2021). The goal of uncertainty reduction theory is to gain information to reduce the experience of uncertainty. Initiating a relationship with a prospective partner is fraught with uncertainty and the potential of being emotionally hurt. An individual may feel hurt because the prospective partner is not who they thought they were (Couch et al., 2012; Toma et al., 2018), because the prospective partner rejects the individual’s attempt to form a relationship (MacDonald & Leary, 2005), and so forth. As such, many prospective couples “talk” before pursuing a romantic relationship, often communicating via texting and social media (Powell et al., 2021). Connecting with an individual on social media significantly increases the amount of information available to the other person, beyond what is shared during conversations. As such, individuals may engage in cybermonitoring prior to initiating a romantic interaction
Cybermonitoring | Powell and Joseph
to learn more about their prospective partner and evaluate the likelihood of being hurt should a relationship be initiated. Once in a relationship, individuals may seek evidence that their partner is truly committed to them and not seeking alternative partners or considering terminating the relationship. Individuals may also engage in cybermonitoring after a romantic interaction to reflect on past experiences with the other individual or see how they are behaving after the interaction’s dissolution. To date, research has predominately focused on the more intrusive behaviors of cyberstalking, and especially after a relationship’s dissolution, rather than the less intrusive behaviors (i.e., cybermonitoring) that may occur before, during, or after the relationship.
Present Study
The present study was guided by prior work on cyberstalking, and aimed to understand the broader concept of
TABLE 1
Participant Demographics
cybermonitoring, as well as explore individual differences in cybermonitoring before, during, and after a romantic interaction. Prior to data collection, research questions and hypotheses were preregistered on Open Science Framework1 (OSF; https://osf.io/rkm9u/). Guided by prior research, it was hypothesized that (a) the majority of participants with social media would have engaged in some kind of cybermonitoring, (b) less concerning cybermonitoring behaviors would occur more often than more concerning behaviors, (c) participants would engage in more cybermonitoring after the breakup than during the romantic interaction, (d) participants who perceive cybermonitoring to occur more often would engage in more cybermonitoring themselves, (e) women would engage in more cybermonitoring than men, (f) initiators of the recent breakup would engage in less cybermonitoring than noninitiators, (g) participants would engage in more cybermonitoring after the breakup the longer the relationship lasted, and (h) participants who cybermonitor will have done so out of curiosity regarding their expartner at least once. The social media apps and app features used for cybermonitoring, and how friendship with the expartner and the expartner’s privacy settings contribute to cybermonitoring were also explored.
Method
Participants
Participants were recruited through Prolific. To participate, individuals had to be between the ages of 18 and 29, have had at least one romantic partner in the past, be fluent in English, reside in the U.S. or Canada, and have at least a 75% approval rate on their Prolific profile. After removing 28 individuals for failing both attention checks and 10 who indicated not to use their data, we had a final analytic sample size of 240 emerging adults (M = 23.74 years, SD = 3.38; see Table 1 for full sample descriptives). Participants tended to report that their most recent romantic interaction lasted 6 months or more (49.6%), and that dissolution occurred more than 6 months ago (74.0%). However, more than half of participants reported still following their expartner on social media (52.1%) and most reported being able to contact their expartner (75.2%), though less than half (47.5%) reported actually contacting their expartner since the dissolution.
1 The study was initially titled and, therefore, preregistered as “Cyberstalking after ghosting” with a hypothesis to examine differences in cyberstalking behaviors based on how participants’ most recent romantic interaction was terminated, in addition to the hypotheses listed in this paper. However, due to a low number of relationship dissolutions due to ghosting, those hypotheses were not tested and are not discussed in this paper.
WINTER 2024
Measures
Relationship Dissolution
Participants were asked if they had ever had a romantic interaction end (Yes or No). Participants then indicated how long ago their most recent dissolution of a romantic interaction occurred (8point scale: Not applicable to Over 6 months), the length of that romantic interaction prior to its dissolution (8point scale: Not applicable to Over 6 months), and who initiated the dissolution (You, Your ex-partner, Mutual, or other).
Cybermonitoring Behaviors
To reduce the potential for biased responses, we used the phrase “gathering information through social media” rather than using the term cybermonitoring. As such, participants were asked to indicate whether they thought any romantic partner had ever gathered information through social media about them before, during, and after their interaction (5point scale: Definitely not to Definitely yes). They were also asked to indicate whether they had ever gathered information through social media about their most recent expartner before, during, and after their interaction (Yes or No), and how often they gathered information (5point scale: Never to All the time). Furthermore, they selected the social media apps and the features of the apps (see Table 2) that they used to gather information about a prospective, current, or former romantic partner and then more specifically about their most recent expartner.
Next, participants were given a list of 18 behaviors and asked to report how often they had done each toward any romantic partner ever (5point scale: Never to All the time; see Table 3). The behaviors were modeled after and created in response to results from Lyndon et al. (2011) and Marshall (2012). Using the same 5point scale, participants then reported how often they had engaged in each of the 18 behaviors toward their most recent expartner (a) before, (b) during, and (c) after the interaction (see Table 3). Then, participants were asked to indicate their motivations behind their engagement in cybermonitoring behaviors during and after the relationship (Yes or No; see Table 4). The list of motivations was modeled after and created in response to results from Marcum et al. (2016).
Next, participants reported how much they considered themselves friends with their most recent expartner (5point scale: Definitely not to Definitely so), if they had been in contact with their expartner since the breakup (Yes or No) and how often they had been in contact with their expartner since the breakup (6point scale: Only one time to Every day). They also reported if they follow or are still friends with their most recent expartner on any form of social media (Yes or No), if they are still able to contact their expartner (Yes or No), and if they can
still view their expartner’s posts/pictures on any form of social media (Yes or No). Lastly, participants were asked if they had ever used a friend’s social media to seek out information about their expartner (Yes or No) and about their expartner’s new partner (Yes or No).
Demographics
Participants were asked demographic questions (e.g., age, gender, ethnicity, sexual orientation, educational
TABLE 2
Social Media Apps Used for Cybermonitoring
Snapchat features n = 100 (41.7%) n = 63 (26.3%)
Snapchat Maps
Twitter followers/following
Twitter bio
Online ‘active status’ 29.1% 20.1%
Facebook posts/stories 85.5% 79.9%
Facebook tagged posts/photos
Facebook likes/comments
Facebook friends list
Facebook ‘About’ information
VSCO features n = 14 (5.8%) n = 6 (2.5%)
VSCO photos/posts 100.0% 100.0%
TikTok features n = 16 (6.7%) n = 4 (1.7%)
TikTok videos/posts 56.3% 75.0%
TikTok liked videos
TikTok comments
TikTok followers/following 31.3% 50.0%
Dating app features n = 59 (24.6%) n = 23 (10.0%)
Tinder/Bumble/Other dating app profiles/bios 89.8% 91.3%
Note. The subsample n and percentage are listed for each social media app, followed by the percentage for each feature. Participants were excluded from these analyses if they indicated they had not ever used the specific app to gather information.
Cybermonitoring | Powell and Joseph
TABLE 3
Cybermonitoring Within Romantic Interactions
Checking if they liked another individual’s picture/post
Viewing their story/ stories more than once within 24 hours
Viewing their tagged photos
Consistently checking their posts/interactions
Creating a fake profile
Checking their followers/
or friends lists
Checking their location
Using
location to go where they
Checking their activity on an app
Checking their friends’ profiles/ posts/pictures
Sending multiple messages in a row to them
Viewing their ‘liked’ posts
Contacting friends of theirs
Following their friends on social media
Checking their online active status
Note. Participants were excluded from these analyses if they indicated they had not ever gathered information before/ during/after the romantic interaction with their most recent ex-partner.
status, and current relationship status). To assess their perceived norm, they were asked how often they thought others engage in cyberstalking on social media (5point scale: Never to All the time). Lastly, they answered a free response question regarding their beliefs about the purpose of the study and the researchers’ potential hypotheses, and a question regarding whether researchers should use their data for the study (Yes or No).
Procedure
This project was approved by the Institutional Review Board at the first author’s institution. Prior to conducting the main study on Prolific, a pilot study was conducted at a small liberal arts college to ensure the cybermonitoring questions encompassed a wide array of possible behaviors. A total of 106 students from the college’s Psychology Department subject pool were recruited; 38 were removed for failing attention checks, 2 indicated not to use their data, 2 were removed for reporting no prior dissolution of a romantic interaction, and 1 was removed for not completing the majority of the survey. The pilot’s final analytic sample size was 63 (M = 19.29 years, SD = 1.23; 71.4% women, 25.4% men, 1.6% nonbinary, 1.6% did not disclose; 85.7% heterosexual, 9.5% bisexual, 1.6% gay, 3.2% did not disclose). Additional cybermonitoring behaviors and social media app features were added for the full survey after reviewing those mentioned in the pilot study. The full survey was advertised on Prolific, an online platform for conducting research in which potential participants have answered prescreening questions to allow for tailored sample recruitment. Based on funding for the project, 250 participants were sought for the study. Participants who met the criteria and were interested in participating could then select to participate in the Qualtrics survey. After reviewing the study information and indicating their interest to proceed, participants were provided the definitions of romantic interaction (any form of a relationship with someone of a romantic interest; e.g., talking, dating, friends with benefits) and ex-partner (any person with whom you previously had some form of a romantic interaction). Then, they answered questions regarding their experiences with cybermonitoring and their engagement in such behaviors, and demographic questions. There were two questions within the survey asking them to report a specific answer for attention checks. The full survey is posted on OSF (https://osf.io/rkm9u/). On average, the survey took approximately 18 minutes to complete. Participants who passed at least one of the attention checks were compensated $2.25 for participating in the study.
Results
Overall Engagement in Cyberstalking
Supporting Hypothesis 1, among the participants with social media (n = 233), a majority (94.4%) reported engaging in some type of social mediabased information gathering (i.e., cybermonitoring) behaviors. Exploratory descriptive statistics were then conducted to examine the social media platforms and the features of these platforms that participants used to gather information (see Table 2). Most participants reported using Facebook and Instagram. Specifically, they reported using stories or posts and tagged photos on Instagram and Facebook, as well as Snapchat stories and Twitter interactions, to engage in cybermonitoring.
A repeated measures ANOVA revealed a significant difference in the frequency with which participants ever reported engaging in specific cybermonitoring behaviors, F (17, 203) = 49.85, p < .001, ηp2 = .81 (see Table 5). Supporting Hypothesis 2, engaging in less intrusive behaviors like clicking on the expartner’s profile and viewing the expartner’s tagged photos were reported most frequently. Relatedly, engaging in more intrusive behaviors like creating a fake profile, checking the expartner’s location, and using the expartner’s location to go to them were reported least frequently. Participants’ engagement in cybermonitoring before, during, and after their most recent romantic interaction was then examined. Contrary to Hypothesis 3, participants reported greater engagement in cybermonitoring during the romantic interaction (M = 2.04, SD = 0.72) than after the romantic interaction (M = 1.62, SD = 0.69), t(194) = 8.41, p < .001, g = 0.60. Exploratory analyses revealed that participants reported greater engagement in cybermonitoring during the romantic interaction (M = 2.07, SD = 0.74) than before the romantic interaction (M = 1.82, SD = 0.68), t(191) = 5.78, p < .001, g = 0.35. They also reported greater engagement in cybermonitoring before the romantic interaction ( M = 1.81, SD = 0.68) than after the dissolution (M = 1.66, SD = 0.72), t(192) = 3.70, p < .001, g = 0.21. Thus, participants reported most frequently engaging in cybermonitoring during the romantic interaction, followed by before the interaction, and reported the least amount of cybermonitoring after the romantic interaction.
A follow up exploratory repeated measures
ANOVA examined this hypothesis at a more microlevel (i.e., looking at each specific behavior rather than averaging across behaviors). There were main effects of time (i.e., During versus After), F(1, 205) = 67.54, p < .001, ηp2 = .25, and of behavior, F(17, 189) = 31.85, p < .001, ηp2 = .74, which were qualified by a Time X Behavior interaction, F(17, 189) = 9.73, p < .001, η p 2 = .47. For example, sending multiple
messages in a row to the partner was more likely during the relationship than after the breakup, and creating a fake profile was more likely after the relationship than before or during (see Table 3).
Individual Differences in Cybermonitoring
Several individual differences were examined; specifically, whether there were differences in participants’ engagement in cybermonitoring based on their perceived norm of it, their gender, if they initiated the breakup or not, the length of the romantic interaction prior to dissolution, and if they remained friends with the expartner.
Supporting Hypothesis 4, participants who reported engaging in cybermonitoring more often also perceived
TABLE 4
During Interaction (n = 175; 72.9%) To know exactly where the partner was
know who the partner was with
decrease uncertainty about the relationship
see
After Interaction (n = 164; 68.3%)
know who the ex-partner was with
decrease uncertainty about your relationship with the ex-partner
see if your ex-partner has another partner
To see the ex-partner’s new partner
To see if your ex-partner is doing better or worse without you
To avoid going to the same place as your ex-partner
To go to the same place as your ex-partner 9.2
To try to create a friendship with your ex-partner 28.2
Note. Participants were excluded from these analyses if they indicated they had not ever gathered information about their most recent ex-partner before/during/after the romantic interaction
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
it to occur more often by others, r(232) = .28, p < .001. Supporting Hypothesis 5, women’s overall reported engagement in cybermonitoring (M = 2.18, SD = 0.71) was significantly higher than men’s (M = 1.98, SD = 0.61), t(228) = 2.26, p = .025, g = 0.30. An additional series of independent samples t tests were conducted based on engagement in each specific cybermonitoring behavior; a reduced p value of .01 was used to determine significance (see Table 6). Women were more likely than men
to check who liked a romantic partner’s picture or post, to check if romantic partner liked another individual’s picture or post, to consistently check a romantic partner’s posts or interactions, and to check a romantic partner’s followers and following or friends list. Men were more likely than women to contact a romantic partner’s friends.
Supporting Hypothesis 6, initiators of the breakup reported significantly less engagement in cybermonitoring after the dissolution (M = 1.56, SD = 0.60) compared
TABLE 5
Mean Differences in Ever Engaging in Cybermonitoring Behaviors
1 Clicking on their profile -
2 Viewing their profile more than once a day 0.95***
3 Checking who liked their picture/post 0.84***
4 Checking if they liked another individual’s picture/post 1.10***
5 Viewing their story/ stories more than once within 24 hours
6 Viewing their tagged photos
7 Consistently checking their posts/interactions
8 Creating a fake profile
9 Checking their followers/following or friends lists
10 Checking their location
13 Checking
15
16 Contacting friends
17 Following
18 Checking
Note. Bonferroni-corrected estimated marginal mean comparisons. For interpreting the table, the number represents the mean for the row item subtracted from the mean for the column item. For example, participants were more likely to click on their profile than view their profile more than once a day. * p < .05. **p < .01. *** p < .001.
to those who were rejected ( M = 1.85, SD = 0.79), t(152) = 2.54, p = .012, g = 0.41. Supporting Hypothesis 7, participants whose romantic interaction had lasted longer engaged in more cybermonitoring after the breakup than those whose romantic interaction had been shorter, r(237) = .14, p = .015.
An exploratory analysis revealed that participants who more strongly considered themselves and their most recent expartner friends, were more likely to engage in cybermonitoring after the breakup, r(237) = .23, p < .001. Lastly, a series of exploratory independent sample t tests were conducted based on their expartner’s privacy settings; a reduced p value of .01 was used to determine significance. Participants who still followed and/or were still friends with their ex partner on any form of social media engaged in more cybermonitoring after the dissolution (M = 1.74, SD = 0.79) than participants who were not connected to their expartner (M = 1.37, SD = 0.52), t(234) = 4.16, p < .001, g = 0.55. Participants who were still able to contact their ex partner engaged in more cybermonitoring after the breakup (M = 1.65, SD = 0.74) than participants who were not able to contact them (M = 1.31, SD = 0.47), t(234) = 3.15, p = .002, g = 0.49. Participants who were still able to see their expartner’s posts/pictures on any form of social media engaged in more cybermonitoring after the breakup (M = 1.68, SD = 0.75) than participants who were not able to see them (M = 1.21, SD = 0.27; t(234) = 4.67, p < .001, g = 0.70). Participants who reported using a friend’s social media account to seek out information about their expartner engaged in more cybermonitoring after the dissolution (M = 2.41, SD = 0.91) than participants who had never done so (M = 1.45, SD = 0.57), t(234) = 7.94, p < .001, g = 1.54.
Motives for Postdissolutional Cybermonitoring
Descriptives were conducted on the participants who reported engaging in cybermonitoring to examine the motives of these behaviors after the dissolution (n = 164). Supporting Hypothesis 8, participants who engaged in cybermonitoring predominately did so out of reasons related to curiosity regarding their most recent expartner (see Table 4).
Discussion
This study sought to provide an update to the literature by focusing on less intrusive information seeking (i.e., cybermonitoring) behaviors, acknowledging the multiple platforms that can be used for cybermonitoring, and exploring how and why today’s emerging adults engage in cybermonitoring. Seven of our eight hypotheses were supported. A majority of emerging adults had engaged in cybermonitoring and they predominately
reported engaging in the less intrusive behaviors. Moreover, those who perceived such behaviors as the norm were more likely to engage in cybermonitoring, women tended to report engaging in more of the behaviors than men, relationship factors (i.e., who ended it and how long it had lasted) were associated with postdissolution cybermonitoring behaviors, and emerging adults tended to engage in cybermonitoring for reasons related to curiosities. Contrary to expectations, emerging adults reported most frequently engaging in cybermonitoring during their romantic interaction, followed by before it, and that they were least likely to do so after it.
Cybermonitoring as a Broad Concept
Cyberstalking behaviors range from individuals monitoring their ex partner’s social media page to regularly attempting to communicate with or harass the expartner (Burke et al., 2011; Lyndon et al., 2011; Marshall, 2012; Tong, 2013). Many of the cybermonitoring behaviors used in the current study were modeled after studies by Lyndon et al. (2011) and Marshall (2012); other cybermonitoring behaviors were added based
TABLE 6
Participant Gender Differences for Having Ever Engaged in Cybermonitoring
Note. * p < .01
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
on results from the pilot study and features of current social media platforms. The behaviors reported in this study as having been engaged in ranged from clicking on the partner’s profile—typically conceptualized as a less intrusive behavior—to using someone’s location to go where they are—typically conceptualized as a more intrusive (i.e., cyberstalking) behavior. However, aligning with past research (Lyndon et al., 2011; Marshall, 2012) and supporting the authors’ supposition that cybermonitoring is a common aspect of romantic interactions, participants reported significantly greater engagement in the less intrusive (i.e., cybermonitoring) behaviors compared to the more intrusive (i.e., cyberstalking) behaviors.
Cybermonitoring and Social Media
Past research on cyberstalking has predominately focused on individuals’ use of Facebook (Darvell et al., 2011; Fox et al., 2014; Lyndon et al., 2011; Marshall et al., 2012; Marshall et al., 2013; Tong, 2013) but different forms of social media have since become more popular. Participants in the current study used Facebook, as well as many other social media apps (e.g., Instagram, Snapchat, Twitter) to engage in cybermonitoring and gather information about romantic partners and expartners. Although emerging adults’ usage of specific social media sites tended to vary, they predominately used partners’ photos and videos, or the bio the partner had shared on the site, to gather information. Given that today’s emerging adults tend to be more regular users of Instagram and Snapchat (Auxier & Anderson, 2021), it is important for future research on cybermonitoring to capture users’ behaviors on those sites, as well as account for how the features of those sites may be different from the features of Facebook.
Comparing Past Cyberstalking and Present Cybermonitoring Findings
Although individuals may engage in information seeking after a romantic interaction/relationship to see what their expartner is doing (Marshall, 2012), as well as during their romantic interactions to monitor their current partners (Drebing et al., 2014; Fox et al., 2014; Marshall et al., 2013), they may also engage in information seeking before their romantic interaction (Kwok & Wescott, 2020). The current study found that information seeking (i.e., cybermonitoring) occurred most frequently during the romantic interaction compared to before and after, and more frequently before the romantic interaction than afterwards. The latter of these findings demonstrates the importance of conceptualizing cybermonitoring before the romantic interaction as a way to seek information about a prospective partner (Vandenbosch & Eggermont,
2016) due to the abundance of information available on social media about prospective partners (Frampton & Fox, 2021; Kwok & Wescott, 2020; McEwan, 2013). Furthermore, these findings align well with uncertainty reduction theory (Berger, 1979; Berger & Bradac, 1982). Individuals are most motivated to seek ways to reduce uncertainties before and during romantic interactions.
Focused on individual differences, our results reiterated that emerging adults’ perceived norms for information seeking are associated with their own engagement in cybermonitoring (Darvell et al., 2011; Tong 2013). The current study also found that women reported engaging in more cybermonitoring than men. This finding adds to a growing body of research which suggests that the specific behaviors being examined may impact whether men (Drebing et al., 2014; Marcum & Higgins, 2021) or women (Muise et al., 2013) are found to engage in more online information seeking (e.g., cyberstalking, surveillance). Additionally, aligning with a review by Marcum and Higgins (2021), it was men in this current study who were more likely to engage in more intrusive cybermonitoring behaviors, such as contacting a partner’s friends.
Focusing on cybermonitoring after the relationship’s dissolution, the current study demonstrated an association between relationship length prior to the dissolution and individuals’ engagement in cybermonitoring. Furthermore, and aligning with prior research (Langhinrichsen Rohling et al., 2000; Tong, 2013), the current study found that initiators of the breakup engaged in less cybermonitoring after the dissolution than those who were broken up with. However, contrary to prior research that found support for remaining friends after a breakup (Villella, 2010), the current study revealed that individuals who remained friends with their ex partner were more likely to engage in cybermonitoring of their expartner, compared to those who did not report a friendship with their expartner. Moreover, although individuals differ in whether or not they remain connected to their ex partners via social media after the relationship dissolution (Fox et al., 2014; Marshall, 2012), the current study revealed that individuals who were still connected with their expartner on social media were more likely to engage in cybermonitoring. They were also more likely to engage in cybermonitoring if they were able to see their expartner’s social media accounts (e.g., if they had a public rather than private Instagram).
Prior research has noted various motives for why individuals may cyberstalk their expartners, such as seeking to know their social activities and new romantic connections, as well as wanting to communicate with them directly or see where they are (Marcum et al., 2016; Tong,
WINTER 2024
2013). In the current study, participants reported engaging in cybermonitoring because they were curious about their expartner. They also reported doing so because they wanted to decrease uncertainty about the relationship, sought to create a friendship, and wanted to recoup the romantic interaction. For those experiencing uncertainty after the romantic interaction or relationship’s dissolution, these findings align well with uncertainty reduction theory (Berger, 1979; Berger & Bradac, 1982).
An Argument for Using the Term Cybermonitoring Reyns et al. (2012) defined cyberstalking as the use of the internet to repeatedly monitor or harass an individual, and any information found is often used to threaten or intimidate that individual. However, participants in this study reported nonthreatening motives for their information seeking behaviors such as getting to know their current partner better or to see how their expartner was doing. This finding aligns with past work which has used other terms to refer to the same or similar cybermonitoring behaviors and reasons that the emerging adults in this study endorsed (e.g., “partner monitoring” in Darvell et al., 2011; “cyberintimacy” in Kwok & Wescott, 2020; “online surveillance” in Tokunaga, 2016; “Monitoring of attractive peers on social networking sites” in Vandenbosch & Eggermont, 2016). Although unwanted pursuit behaviors mostly occur postbreakup, participants in this study more regularly engaged in cybermonitoring, and especially the less intrusive behaviors, during their romantic interaction. Although prior research has acknowledged that cyberstalking can include less intrusive behaviors (Lyndon et al., 2011; Marshall, 2012), it is usually the more intrusive behaviors that typically are referenced when the term cyberstalking is used (e.g., Kwok & Wescott, 2020; Marcum & Higgins, 2021).
Given the abundance of information available on social media, it seems engaging in behaviors representative of the less intrusive end of the continuum of cyberstalking has become a common aspect of romantic interactions (Frampton & Fox, 2021; Kwok & Wescott, 2020). Thus, we posit that cybermonitoring , rather than cyberstalking, may be a more appropriate term to reflect the behaviors emerging adults report engaging in prior to, during, and after their romantic interactions. Moreover, although not directly studied in this project, the term cybermonitoring likely has a less negative connotation than the term cyberstalking. However, future research is necessary to test that assertion.
Limitations and Future Directions
This study sampled a relatively small number of emerging adults during the COVID19 pandemic, which may
have resulted in heightened cybermonitoring due to stayathome and social distancing protocols. Thus, it is recommended that this study be replicated in a larger, more diverse sample when participants are not living through a global pandemic. Second, although a core set of hypotheses were preregistered, several exploratory analyses were also conducted. As such, there is concern about increased possibility of type I error. Third, this study was predominately descriptive of the full sample’s reports. Future research on cybermonitoring may consider individual differences (e.g., personality or relationship variables) in behaviors. In relation to the items about participants’ most recent romantic interaction, it may be that this particular interaction was qualitatively different from their typical interactions. As such, future research should consider participants’ other prior romantic interactions and dissolutions, as well as consider participants’ current romantic relationship status. A longitudinal study would also be beneficial to examine cybermonitoring across a romantic interaction.
Conclusion
The present study provides a nuanced view into individuals’ cybermonitoring behaviors before, during, and after a romantic interaction. Emerging adults reported using many different social media apps to engage in cybermonitoring. Moreover, they reported engaging in cybermonitoring least frequently after a romantic interaction; rather, they were more likely to engage in cybermonitoring before the romantic interaction and most likely to do so during it. As hypothesized, there were also individual differences in their engagement in cybermonitoring. Per our analyses, engagement in cybermonitoring behaviors seems to predominately consist of less intrusive behaviors and have less harmful motivations, compared to the typical conceptualization of cyberstalking and other unwanted pursuit behaviors.
References
Auxier, B., & Anderson, M. (2021, April). Social media use in 2021. Pew Research Center. https://www.pewresearch.org/internet/2021/04/07/social-media-use-in-2021/ Berger, C. R. (1979). Beyond initial interaction: Uncertainty, understanding, and the development of interpersonal relationships. In H. Giles & R. N. St. Clair (Eds.), Language and social psychology (pp. 122–144). Basil Blackwell. Berger, C. R., & Bradac, J. J. (1982). Language and social knowledge: Uncertainty in interpersonal relations. Edward Arnold.
Burke, S. C., Wallen, M., Vail-Smith, K., & Knox, D. (2011). Using technology to control intimate partners: An exploratory study of college undergraduates. Computers in Human Behavior, 27(3), 1162–1167. https://doi.org/10.1016/j.chb.2010.12.010
Couch, D., Liamputtong, P., & Pitts, M. (2012). What are the real and perceived risks and dangers of online dating? Perspectives from online daters: Health risks in the media. Health, Risk & Society, 14(7), 697–714. https://doi.org/10.1080/13698575.2012.720964
Cupach, W., & Spitzberg, B. (1998). Obsessive relationship intrusion and stalking. In B. H. Spitzberg & W. R. Cupach (Eds.), The dark side of close relationships, pp 233–263. Lawrence Erlbaum Associates.
Darvell, M. J., Walsh, S. P., & White, K. M. (2011). Facebook tells me so: Applying
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
WINTER 2024
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
the theory of planned behavior to understand partner-monitoring behavior on Facebook. Cyberpsychology, Behavior, and Social Networking, 14, 717–722. https://doi.org/10.1089/cyber.2011.0035
De Smet, O., Buysse, A., & Brondeel, R. (2011). Effect of the breakup context on unwanted pursuit behavior perpetration between former partners. Journal of Forensic Sciences, 56(4), 934–941. https://doi.org/10.1111/j.1556-4029.2011.01745.x
Drebing, H., Bailer, J., Anders, A., Wagner, H., & Gallas, C. (2014). Cyberstalking in a large sample of social network users: Prevalence, characteristics, and impact upon victims. Cyberpsychology, Behavior, and Social Networking, 17(2), 61–67. https://doi.org/10.1089/cyber.2012.0231
Fox, J., Osborn, J. L., & Warber, K. M. (2014). Relational dialectics and social networking sites: The role of Facebook in romantic relationship escalation, maintenance, conflict, and dissolution. Computers in Human Behavior, 35, 527–534. https://doi.org/10.1016/j.chb.2014.02.031
Frampton, J. R., & Fox, J. (2021). Monitoring, creeping, or surveillance? A synthesis of online social information seeking concepts. Review of Communication Research, 9, 1–42. https://doi.org/10.12840/ISSN.2255-4165.025
Huie, K., Butler, M., & Percy, A. (2023). Identifying trends and patterns in offending and victimization on Snapchat: A rapid review. Security Journal, 1–18. https://doi.org/10.1057/s41284-023-00400-6
Jones, J. M. (2023, January 9). Social media users more inclined to browse than post content. Gallup https://news.gallup.com/poll/467792/social-mediausers-inclined-browse-post-content.aspx
Kwok, I., & Wescott, A. B. (2020). Cyberintimacy: A scoping review of technology-mediated romance in the digital age. Cyberpsychology, Behavior, and Social Networking, 23(10), 657–666. https://doi.org/10.1089/cyber.2019.0764
Langhinrichsen-Rohling, J., Palarea, R. E., Cohen, J., & Rohling, M. L. (2000). Breaking up is hard to do: Unwanted pursuit behaviors following the dissolution of a romantic relationship. Violence and Victims, 15(1), 73–90. https://doi.org/10.1891/0886-6708.15.1.73
Lyndon, A., Bonds-Raacke, J., & Cratty, A. D. (2011). College students’ Facebook stalking of ex-partners. Cyberpsychology, Behavior, and Social Networking, 14(12), 711–716. https://doi.org/10.1089/cyber.2010.0588
MacDonald, G., & Leary, M. R. (2005). Why does social exclusion hurt? The relationship between social and physical pain. Psychological Bulletin, 131(2), 202–223. https://doi.org/10.1037/0033-2909.131.2.202
Marcum, C. D., & Higgins, G. E. (2021). A systematic review of cyberstalking victimization and offending behaviors. American Journal of Criminal Justice, 46, 882–910. https://doi.org/10.1007/s12103-021-09653-6
Marcum, C. D., Higgins, G. E., & Poff, B. (2016). Exploratory investigation on theoretical predictors of the electronic leash. Computers in Human Behavior, 61, 213–218. https://doi.org/10.1016/j.chb.2016.03.010
Marshall, T. C. (2012). Facebook surveillance of former romantic partners: Associations with post-breakup recovery and personal growth. Cyberpsychology, Behavior, and Social Networking, 15(10), 521–526. https://doi.org/10.1089/cyber.2012.0125
Marshall, T. C., Bejanyan, K., Di Castro, G., & Lee, R. A. (2013). Attachment styles as predictors of Facebook-related jealousy and surveillance in romantic relationships. Personal Relationships, 20(1), 1–22. https://doi.org/10.1111/j.1475-6811.2011.01393.x
McEwan, B. (2013). Sharing, caring, and surveilling: An actor-partner interdependence model examination of Facebook relational maintenance strategies. Cyberpsychology, Behavior and Social Networking, 16(12), 863–869. https://doi.org/10.1089/cyber.2012.0717
Muise, A., Christofides, E., & Desmarais, S. (2013). “Creeping” or just information seeking? Gender differences in partner monitoring in response to jealousy on Facebook. Personal Relationships, 21(1), 35–50. https://doi.org/10.1111/pere.12014
Powell, D. N., Freedman, G., Jensen, K., & Preston, V. (2021). “Talking” as a romantic interaction: Is there consensus? Journal of Couple & Relationship Therapy, 20(4), 384–404. https://doi.org/10.1080/15332691.2020.1867684
Reyns, B. W., Henson, B., & Fisher, B. S. (2012). Stalking in the twilight zone: Extent of cyberstalking victimization and offending among college students. Deviant Behaviors, 33(1), 1–25. https://doi.org/10.1080/01639625.2010.538364
Tokunaga, R. S. (2016). Interpersonal surveillance over social network sites: Applying a theory of negative relational maintenance and the investment model. Journal of Social and Personal Relationships, 33, 171–190. https://doi.org/10.1177/0265407514568749
Toma, C. L., Jiang, C., & Hancock, J. T. (2018). Lies in the eye of the beholder: Asymmetric beliefs about one’s own and others’ deceptiveness in meditated and face-to-face communication. Communication Research, 48(8), 1167–1192. https://doi.org/10.1177/0093650216631094
Tong, S. T. (2013). Facebook use during relationship termination: Uncertainty reduction and surveillance. Cyberpsychology, Behavior, and Social Networking, 16(11), 788–793. https://doi.org/10.1089/cyber.2012.0549
Vandenbosch, L., & Eggermont, S. (2016). The interrelated roles of mass media and social media in adolescents’ development of an objectified selfconcept: A longitudinal study. Communication Research, 43, 1116–1140. http://doi.org/10.1177/0093650215600488
Villella, S. (2010). Broken up but not broken: Satisfaction, adjustment, and communication in post-dissolutional relationships. Communication and Theater Association of Minnesota Journal, 37(3), 27–46. https://cornerstone.lib.mnsu.edu/ctamj/vol37/iss1/3/
Author Note
Darcey N. Powell https://orcid.org/0000000160769741
Darcey N. Powell is now an Associate Professor in the Department of Psychology and Sociology at Texas A&M University – Corpus Christi. Abbie Joseph graduated with her bachelor’s degree from Roanoke College in May of 2021 and is now a clinical trial specialist at the Professional Education & Research Institute. The study was preregistered on OSF (https://osf.io/rkm9u/). The materials are available on OSF (https://osf.io/rkm9u/). The data and analyses are available upon request. This article is based on Abbie Joseph’s honors in the major project at Roanoke College, which was supported by an internal grant for student research. Aspects of the project were presented at the 2021 Society for Personality and Social Psychology annual conference. We have no known conflict of interest to disclose.
Correspondence concerning this article should be addressed to Darcey N. Powell, Department of Psychology and Sociology, Texas A&M University – Corpus Christi, 6300 Ocean Drive, Unit 5827, Corpus Christi, TX 78412. Email: Darcey.Powell@gmail.com Cybermonitoring | Powell and Joseph
Differences in Condom Use Patterns Between Couples and Singles in College-Aged Individuals
Gabrielle O. Stokes, Kylie R. Gulotta, Rachel E. Blansfield, Mackenzie S. Flanders, and Alicia Drais-Parrillo*
Department of Psychology, The Pennsylvania State University
ABSTRACT. Condoms are safe and effective at preventing the risk of sexually transmitted infections (STIs) and pregnancy (Hatcher, 2007; Holmes et al., 2004). However, when an individual enters a committed relationship, their condom use decreases significantly (Copen, 2017; Lehmiller et al., 2014; Matser et al., 2014; Weitzman et al., 2019). We surveyed 219 heterosexual individuals in committed relationships and 87 in noncommitted sexual relationships to determine differences in their condom use patterns. We collected data on individual psychometric characteristics (condom use attitudes, sexual narcissism, perceived invulnerability, concern for sexual pleasure, condom use selfefficacy, and concern for pregnancy and STIs), individual sexual behaviors (birth control usage, type of sexual activity, number of partners, number of sexual encounters), and couple characteristics (commitment and exclusivity). We modeled the predictors using hierarchical regressions. Our model for coupled participants was significant, F(3, 150) = 10.12, p = .002, R2 = .36, and included birth control, pregnancy, and condom use attitudes as significant predictors. Our model for single individuals was also significant, F(3, 82) = 18.02, p < .001, R2 = .38, and included STI concern, pregnancy concern, and condom use attitudes as significant predictors. Understanding predictors of when collegeaged individuals in committed and casual relationships chose to use condoms may allow researchers to be able to create better, targeted tactics to promote condom use in the future.
Keywords: sex, condoms, couples, singles, behavior
Condoms are safe and effective at preventing the risk of sexually transmitted infections (STIs) and pregnancy in sexual encounters (Hatcher, 2007; Holmes et al., 2004). Condom use for contraception and STI prevention increased into the 1980s and 1990s with better access to materials, decrease in cost, and as a result of the HIV/AIDs epidemic (Amy & Thiery, 2015; Youssef, 1993). Yet, consistent use has been on
Open Materials badge earned for transparent research practices. Materials are available at https://osf.io/bgc4m/
the decline since the turn of the century, especially among sexually active young adults (Amy & Thiery, 2015; Copen, 2017). There are many possible causes for this decline such as increased use of other forms of contraceptives including hormonal and longacting reversible contraceptives (LARCs; WalshBuhi & Helmy, 2018). Another cause may be decreased risks associated with the HIV and AIDS virus with more effective
antiretroviral therapy and prophylactic medication like preexposure prophylaxis (PrEP; Spinner et al., 2016; Volberding & Deeks, 2010). Together, these factors appear to have affected both attitudes and behaviors towards condom use.
Most research in the United States on condom use patterns has been focused on two populations. The first of these populations is high school students whose reported condom use behaviors are monitored via the Youth Risk Surveillance System (Centers for Disease Control and Prevention [CDC] et al., 2013), a crosssectional survey of 9–12th graders conducted biannually. The second population where much condom research has focused is communities of men who have sex with men. The majority of past research neglects an important population where condom use (or lack of) is prevalent: collegeaged individuals. It is important to explore why condom use is declining in heterosexual collegeaged populations, as risks associated with not using condoms during vaginal intercourse such as the climbing rates of STIs (Workowski et al., 2015) and the decreased access to abortions across the United States (Wilkinson & Bernard, 2024) are becoming increasingly prevalent. Specifically, the current work seeks to understand predictive factors that influence heterosexual young adults’ condomuse choices.
A prominent predictive factor for condom use is relationship status, with prior studies showing differences in condom use among committed and casual relationships (Copen, 2017; Lehmiller et al., 2014; Matser et al., 2014; Weitzman et al., 2019). Therefore, it follows that single individuals (or those in casual relationships) would have different predictive markers than coupled individuals (those in committed relationships) as the two groups’ decisionmaking strategies about condom use are likely to be different. A review of literature for both of these populations is presented below.
Condom Use and Decision-Making Within Coupled Individuals
After starting a relationship, condom use decreases for heterosexual couples (Copen, 2017; Lehmiller et al., 2014; Matser et al., 2014; Weitzman et al., 2019), especially as the duration of the relationship increases (Amy & Thiery, 2015). Heterosexual men are less likely to use a condom when they feel a stronger commitment to their partner and see them more regularly (CraigKuhn et al., 2019). Another study, focusing on heterosexual women, showed increases in both commitment and intimacy levels were negatively correlated with condom use (Kusunoki & Barber, 2020). One reason for this trend may be that, as a relationship becomes more intimate and committed, there may be less risk associated with
not using a condom. The efficacy of a condom is not changed if someone is in a committed relationship or not. However, if both partners enter into a relationship free of STIs and remain monogamous, then there is no concern for future infection. This means the main, if not only, risk of not using a condom in a committed relationship may be an unintended pregnancy.
Indeed, the reduction of the risk of STI contraction has been found to influence birth control selection in coupled individuals. Couples are likely to start out their relationships using condoms and transition to hormonal birth control measures as they become more intimate (Manlove et al., 2011; Manning et al., 2009). This can be explained by the decreased risk for STI contraction and a larger concern for unintended pregnancies leading to couples choosing more effective contraception (Trussell, 2011); another explanation is an increased level of trust between partners where the male partner trusts his female counterpart to take contraceptive measures even when he is not physically present.
Interestingly in a study analyzing data from the National Longitudinal Study of Adolescent Health from 2001–2002 comprising 322 dating and 406 cohabitating young adult relationships (all over the age of 18), researchers found that not only were cohabitating couples less likely to use condoms, they were also less likely to use hormonal birth control (Wildsmith et al., 2015). A lower likelihood of using condoms was also associated with increased frequency of sexual encounters as well as cohabitation status. These data suggest that as a relationship becomes more serious, couples may be more willing to assume the risk of an unintended pregnancy.
Condom Use and Decision-Making Among Singles
Those who are more likely to have sex with multiple casual partners are more likely to contract and spread STIs (Grabovac et al., 2020). However, collegeaged individuals who have multiple partners may not perceive themselves to be at risk of contracting an STI. The perception that an individual is immune to a bad outcome or “perceived invulnerability” has been found to affect both men” and women’s condom use behaviors (Thompson et al., 1996). This can be explained by considering riskreward modeling. If one feels that an adverse outcome such as STI or unwanted pregnancy is not likely to happen to them, they are less likely to use a condom to decrease their risk at the cost of their sexual satisfaction or against the urging of a partner.
Interestingly, there seems to be a cognitive dissonance involving the temporal relation of one partner to another in condom usage. Weitzman et al. (2019) found that in a review of 2.5 years’ worth of weekly data on 1,003 eighteen to nineteenyearold women,
WINTER 2024
Stokes, Gulotta, Blansfield, Flanders, and Drais-Parrillo |
the women were more likely to use contraception when they had concurrent multiple partners and less likely to use contraception when they had multiple partners consecutively. This suggests that the young women perceive their risk of contracting an STI as higher when they have concurrent partners and lower when they are temporally monogamous. These findings support a cognitive basis for condom use decisions based on perception of risk in each sexual encounter.
Another potential reason for reduced condom use among single individuals is a lack of communication in casual sexual encounters. Single, heterosexual men reported that most of the time that they do not know their partners’ desired condom use nor do they ask, so the decision typically falls on them (Raine et al., 2010). Single young adults also reported their reason for not using condoms is for fear of “ruining the moment” or decreasing pleasure (Fennell, 2014). Indeed, the concern of decreased pleasure when using condoms has been widely cited as a deterrent, especially in young adult populations (Brown et al., 2008; Higgins & Wang, 2015; Randolph et al., 2007). Concern for one’s own pleasure coupled with a lack of communication may be compounded by an individual’s level of sexual narcissism, or the expression of exploitative, selfcentered, or lack of empathy in the sexual domain (Basting et al., 2023; Widman & McNulty, 2010). Those with sexual narcissism may have more power to prioritize themselves in casual relationships where there is less communication between the dyad.
Hypotheses
Our goal was to explore the difference in decisionmaking priorities between committed and casual relationships to compare predictors of condom use decisions within couples and among singles by conducting hierarchical regression analysis of literaturedriven predictors of condom use in each population described below. Ultimately, we hoped to inform public health strategies to increase condom use by attuning educational campaigns toward the key factors affecting young adults’ decisions to use or not use condoms.
For coupled individuals, we predicted that exclusivity and commitment ratings, concern for pregnancy, number of sexual encounters, and a positive attitude towards condoms would predict a higher likelihood of condom use. We also predicted that more effective birth control methods and an increased sense of perceived invulnerability would predict a lower likelihood of condom use. For single individuals, we predicted that increased concern for STI contraction and pregnancy, positive attitudes towards condoms, and condom use selfefficacy would predict a higher likelihood of
condom use. We also predicted that increased concern for self pleasure, an increased sense of perceived invulnerability, and more sexual narcissism traits would predict a lower likelihood of condom use.
Methods
Procedure
We surveyed collegeaged participants to assess their condom use attitudes and behaviors via Qualtrics. Participants were students enrolled in a lower level psychology class at a large, public university in the midAtlantic. The study was approved by the Pennsylvania State University IRB. The students were required to complete a certain number of research hours as part of their coursework and were able to selfselect to be in the current study. The study was posted on Qualtrics, and students took the survey on their personal laptops. Students received survey credit for their participation. Prior to starting the survey, participants were required to read a description of the purpose of the survey and asked to confirm their informed consent to their depersonalized information being used. The survey was estimated to last 10 minutes.
Participants
The sample consisted of 308 collegeaged individuals (ages 18–24). Initially, 597 students consented to the study, but 45% were ineligible due to not having had sex in the previous month. Only 2% of the participants who had engaged in sexual activities in the past month had done so with a samesex partner, analysis was limited to participants who had engaged in penetrative vaginal sex in the past month. This inherently limited the sample to heterosexual couples. Most of the sample were women (69%), with 30% of our sample being men and less than 1% of the sample identifying as nonbinary. Seventyone percent of the sample selfreported being in a committed partnership.
Materials
A pilot study with a previous college aged sample was completed in advance of primary data collection to review applicability of established measures and determine which scale items were essential to our central research questions. In addition, we developed and tested a Condom Use Attitude Scale for this age group in the pilot study. Based on our results, we created a streamlined version of the latter and established scales for inclusion in our survey. Applicable Cronbach’s alpha coefficients for reliability are presented below. Visit https://osf.io/bgc4m/ for specific questions for each scale.
Condom Use Attitudes
Condom use attitudes were assessed using a novel scale. The first edition of the Condom Use Attitudes Scale given in the pilot study consisted of 17 items on a 7point scale. Two questions were excluded from validity and reliability assessments and the resultant 15item scale was considered reliable (α = .86). The scale was further shortened to eliminate redundant items for the purposes of this study. The remaining six items included statements such as “one should use condoms during penetrative sex,” “condom use is necessary when you have multiple sexual partners,” and “condoms interfere with spontaneity.” The shortened version of the scale also had good reliability (α = .71).
Perceived Invulnerability
Perceived invulnerability is a belief that one is protected from harm or bad outcomes and is often correlated with risky behaviors. This invulnerability is widely studied in adolescents and can be reasonably extended to collegeaged students who, as emerging adults, benefit from an extended youth and still may reflect characteristics of adolescence. To test the level of perceived invulnerability in our sample, we used a portion of the Adolescent Invulnerability Scale (Duggan, 2014; Hill et al., 2012) which is scored on a 5point scale with higher scores corresponding with higher levels of invulnerability. The Adolescent Invulnerability Scale has three factors: general invulnerability, danger (or physical/external) invulnerability, and interpersonal (or psychological/ internal) invulnerability. We only included items that loaded onto the danger invulnerability subcategory (α = .81; Hill et al., 2012) and omitted five of the original 12 items while still maintaining a comparable alpha (α = .80). The distribution of the sample was normal.
Condom Use Self-Efficacy
Condom use selfefficacy, or the belief that one is able to decide to use a condom and execute that decision, was measured by the Condom Use SelfEfficacy Scale (Baele et al., 2001). Items were rated on a 7point scale with higher scores corresponding to stronger belief in one’s ability to use and suggest condoms. The Condom Use SelfEfficacy Scale contains six subcategories (Technical Skills, Image Confidence, Emotion Control, Purchase, Assertiveness, and Sexual Control). Our survey included four questions out of the six in the “Technical Skills” category (original α = .84) and three questions out of the six in the “Image Confidence” category (original α = .82). In addition, items were updated in accordance with modern verbiage surrounding sex and sexual activities. The resulting alpha for the Technical Skills in the sample was lower than the original published alpha
(α = .67), but the alpha for the Image Confidence was comparable (α = .81) with the scale being reliable overall (α = .60). The distribution of the samples was normal.
Sexual Narcissism
Sexual narcissism, or the expression of exploitative behavior, selfcentered beliefs, or lack of empathy in the sexual domain, was assessed in our sample using a shortened version of the Sexual Narcissism Scale (Widman & McNulty, 2010). The Sexual Narcissism Scale items are rated on a 5point scale with higher scores corresponding to higher levels of sexual narcissism. The original scale consists of four 5item subscales (Sexual Exploitation α = .76; Sexual Entitlement α = .80; Low Sexual Empathy α = .79; Sexual Skill α = .86, overall α = .85). We excluded questions from the sexual skills subcategory and included three questions from each of the remaining three subscales. Our alpha values for the revised, shortened scale were slightly lower than the original values for two of the subscales (Sexual Exploitation α = .69; Sexual Entitlement α = .80; Low Sexual Empathy α = .67). Overall, the shortened scale remained highly reliable (α = .78). The distribution of our sample was normal.
Analysis
The data were analyzed via hierarchical regression modeling using SPSS software. Composite scores for each of the above scales were computed and used in the final analysis. A description of how categorical predictors were derived can be found below.
Condom Use Patterns
We created a composite variable from five distinct questions to rate each participant’s history of condom use. The questions included as indicators of condom use are shown below with corresponding answers that were awarded a point in brackets. Higher scores corresponded to a more consistent history of condom use.
1. Think back to the start of your sexual relationship with your most recent sexual partner. Did you begin your relationship using condoms? [yes]
2. Which birth control(s) did you and your partner use in the last month? (check all that apply) [barrier methods (condoms, female condoms, diaphragm, cervical cap)]
3. Did you and your sexual partner use condoms the last time you had penetrative sex? [yes]
4. Which of the following best describes why you choose to use condoms? [any answer choice excluding “I never use condoms.”]
5. Which of the following best describes why you choose not to use condoms? [I always use condoms].
WINTER 2024
Stokes, Gulotta, Blansfield, Flanders, and Drais-Parrillo |
Birth Control
Participants were provided a list of seven categories of birth control (i.e., oral contraceptives, barrier methods, natural family planning, Long Acting Reversible Contraceptives [LARCs], sterilization, no forms of birth control, and an “I don’t know” option) and asked to indicate which birth controls they and their partners used. They were allowed to select multiple responses. For analysis purposes, responses were collapsed into three categories corresponding with points based on birth control effectiveness. Responses of “I don’t know” and “my partner and I do not use any forms of birth control” were treated as equivalent. We reasoned that not knowing which, if any, forms of birth control being used is essentially akin to entering a sexual relationship without using birth control. Those two responses received a score of zero. Participants who selected oral contraceptives received a score of one, as oral contraceptives are effective, but their effectiveness is often decreased due to user error and, even with perfect use, are less effective than LARCs (Trussell, 2011). Finally, participants who selected LARCs were given a score of 2 as LARCs are the most effective forms of birth control. Participants who only reported using natural family planning and were given a score of zero as natural family planning is a difficult form of birth control to master and may be equivalent to not using any birth control at all. Participants who selected barrier methods were excluded as condom use is the variable we are attempting to predict and, thus, cannot be included in the analysis. No participants selected sterilization as their form of birth control.
Concern for Sexual Pleasure
The degree of concern about condoms affecting sexual pleasure is a composite of two items. If the participant chose “condoms reduce sexual pleasure” in response to “Which of the following best describes why you choose not to use condoms?” they received a point. The second item asked participants to rate their agreement with the statement “Condoms reduce sexual pleasure,” and points were added based on their selection, such that “strongly disagree,” “disagree,” “somewhat disagree,” or “neither agree nor disagree” resulted in zero points, the selection of somewhat agree received one point, and the selection of “agree” and “strongly agree” yielded two points.
Couple’s Commitment and Exclusivity
Participants rated their most recent sexual relationship on commitment (“the degree of emotional attachment you feel towards your partner”) and exclusivity (“the degree that you are only with your current partner
sexually”) using continuous slider scales from 0 to 100%. For analysis purposes, the responses were grouped into four categories for commitment and exclusivity separately: responses from 0–10% are “no commitment/ exclusivity,” 11%–49% are “low commitment/exclusivity,” 50%–89% are “high commitment/exclusivity,” and 90%–100% are “fully committed/exclusive.” The grouping of responses into categories stems from the extreme bimodal distribution of responses. The overall mean for the sample was 73.98% (SD = 36.27) for commitment and 71.86% (SD = 39.82) for exclusivity. The means for commitment and exclusivity of coupled individuals ( M = 91.78%, SD = 19.78 and M = 88.93%, SD = 26.56, respectively) differed from singles ( M = 29.41%, SD = 30.01 and M = 29.22%, SD = 35.55, respectively) and created the bimodal distributions that were not captured by the mean alone. The commitment and exclusivity ratings were highly correlated for the combined sample of coupled individuals and singles (r = .81, p < .001). For the purposes of analysis, we only used exclusivity as in prior research that exclusivity was a predictor for men and women whereas commitment may only predict condom use for women (Constant et al., 2016; CraigKuhn et al., 2019; Kusunoki & Barber, 2020).
TABLE 1
Characteristics of
Individuals and Singles
encounters in the past month (1–60)a
Note. a range determined by responses. * p < .05
TABLE 2
Coupled and Single Individuals Differences in Condom Use Behaviors and Concerns
Note.
TABLE 3
Hierarchical Regression Results for Condom Use Patterns for Coupled Participants
Results
We predicted that being in a couple would change participants’ perception of condoms based on prior research and our theoretical framework for our model. This difference is confirmed by differences within our coupled and single participants was confirmed by independentsamples t tests and described in Table 1. Unsurprisingly, coupled individuals and singles differed in the degree of commitment and exclusivity that they feel describes their relationship. Individuals in committed relationships reported higher feelings of commitment (M = 91.78%, SD = 19.78) compared to singles (M = 29.41%, SD = 30.01), t(115.83) = 17.95, p < .001, d = 2.45. Similarly, coupled participants also reported higher feelings of exclusivity (M = 88.93%, SD = 26.56) with their current partner compared to singles (M = 29.22%, SD = 35.55), t(126.01) = 14.17, p < .001, d = 1.90. Adjusted degrees of freedom were used when equal variance could not be assumed.
Singles had more sexual partners in the previous month ( M = 1.48, SD = 0.76) compared to coupled individuals (M = 1.13, SD = 0.94), t(194.09) = 3.41, p < .001, d = 0.40. However, coupled individuals had significantly more sexual activity (M = 11.43, SD = 9.93) in the previous month compared to singles (M = 3.67, SD = 6.46), t(228.01) = 7.89, p < .001, d = 0.93. There was no difference between coupled participants and singles in how many forms of birth control they use.
Concern for sexual pleasure, condom use selfefficacy, invulnerability, and condom use attitude scores were not different. This lends credence to the belief that the factors do not determine who is part of a committed relationship yet may still affect how individuals in committed relationships make decisions about condom usage compared to those in casual partnerships. Sexual narcissism scores were different between coupled participants and singles with those who are coupled (M = 1.70, SD = 0.53) having lower sexual narcissism scores compared to singles ( M = 2.03, SD = 0.62), t(139.44) = 4.42, p < .001, d = −0.57.
Coupled Participants
A hierarchical regression approach was used to predict condom use patterns for coupled individuals. The first iteration of the model involved only coupled participants’ rankings of exclusivity and did not significantly predict condom use patterns ( p = .38). Therefore, exclusivity was removed from subsequent steps. The first model presented in Table 3 included birth control methods used by the participants and was significant, F(1, 152) = 19.57, p < .001, R2 = .11. According to this model, as an individuals opted for more efficacious birth control methods (e.g., from no birth control to using
Stokes, Gulotta, Blansfield, Flanders, and Drais-Parrillo | Differences in
an oral contraceptive or from an oral contraceptive to using LARC), condom use behaviors decreased slightly (β = .34, t = 4.42, p < .001). However, birth control methods were only able to explain 11% of the variance in condom use patterns among those in a couple.
Pregnancy concern was added in the second step. This model was significant, F(2, 151) = 19.57, p < .001, R 2 = .32, as were both predictors. Birth control use remained negatively related to condom use (β = .36, t = 5.30, p < .001) and pregnancy concern positively predicted condom use for coupled individuals (β = .46, t = 6.77, p < .001). When perceived invulnerability was added in the third step, the model, although significant, was not statistically different from the previous model. In addition, invulnerability was not a significant predictor (t = 1.52, p = .13) and, thus, was dropped from subsequent steps. Similarly, sexual encounters, added in the next step, was not a strong predictor (t = 0.84, p = .40) and the model did not greatly improve, so the variable was removed.
Condom use attitudes were added in the final step. The model was significant, F(3, 150) = 10.12, p = .002, R 2 = .36, and accounted for more variance than the previous models. Birth control and pregnancy concern remained strong predictors, and as condom use attitudes increase, condom use behaviors increase (β = .21, t = 3.17, p = .002).
Singles
A hierarchical regression approach was used to predict condom use patterns for single individuals. The first model in Table 4 included participants’ STI concern and pregnancy concern and was significant, F(2, 83) = 18.02, p < .001, R2 = .30. According to this model, as the concern for contracting an STI (β = .48, t = 4.19, p < .001) or becoming pregnant (β = .68, t = 5.95, p < .001) increases, so does the likelihood that they have a pattern of consistent condom use. Concern for sexual pleasure was not a significant predictor (t = .94, p = .35) to model history of condom use. Similarly, condom use selfefficacy did not significantly predict a history of condom use (t = 0.43, p = .67). Adding perceived invulnerability in the fourth step also did not significantly predict condom usage (t = 0.64, p = .52). Similarly, when sexual narcissism was added, it was not a strong predictor (t = 1.62, p = .11) so the variable was removed.
Condom use attitudes were added in the final step. The model was significant, F(3, 82) = 18.02, p < .001, R2 = .38, and accounted for more variance than the previous models. STI and pregnancy concern remained strong predictors, and as condom use attitudes increased, the likelihood that a participant had a more consistent history of condom use also increased (β = .30, t = 3.17, p = .002).
Discussion
This study was conducted to investigate predictive factors for the use of condoms in coupled and single collegeaged individuals. We hypothesized that increased concern for STI contraction and pregnancy, positive attitudes towards condoms, and condom use selfefficacy would be
TABLE 4
Hierarchical Regression Results for Condom Use Patterns for Singles
Condom
Perceived Invulnerability
Concern 1.59*** 0.76 2.42 0.42 .44 Pregnancy Concern 1.78*** 1.12 2.44 0.33 .63 Sexual Narcissism -0.34 -0.77 0.08 0.21 -.15 Step 6 .38 .06**
Pregnancy Concern 1.48*** 0.81 2.12 0.34 .53
Condom Use Attitudes 0.57** 0.21 0.92 0.18 .30
Note. CI = Confidence Interval; LL = Lower Limit; UL = Upper Limit * p <.05 ** p <.01 *** p <.001
predictive factors for condom use in single participants. We also predicted that increased concern for selfpleasure, an increased sense of perceived invulnerability, and more sexual narcissism traits would predict a lower likelihood of consistent condom use among singles. Our analyses showed that the significant predictive factors for a single person’s use of condoms were pregnancy and STI concerns and condom use attitudes as predicted. Concern for sexual pleasure, perceived invulnerability, sexual narcissism and condom use selfefficacy were not significant for singles’ pattern of condom use, which did not support our original hypotheses.
For participants in a couple, we originally predicted that exclusivity and commitment ratings, concern for pregnancy, number of sexual encounters, and a more positive attitude towards condoms would predict a higher likelihood of condom use whereas more effective birth control methods and an increased sense of perceived invulnerability would decrease the likelihood of condom use. The results supported our original hypotheses that positive predictive factors for coupled condom use patterns over time were use of other forms of birth control, pregnancy concerns, and condom use attitudes. The number of sexual encounters in the past month, perceived invulnerability, and ratings of exclusivity were not significant predictors for those in couples which did not support our original hypotheses. Overall, our model accounted for 38% of the variance in condom use patterns for single participants and 35% of the variance in condom use patterns for coupled participants.
Differences Between Coupled and Single Participants
Relationship status has consistently emerged as a primary predictor of condom use (Copen, 2017; Lehmiller et al., 2014; Matser et al., 2014; Weitzman et al., 2019). Based on this consistent finding, we used separate models for couples and singles. Analyses comparing couples and singles did show notable differences between the groups, which affirmed our decision to analyze couples and singles separately. We feel the differences between the groups reflect different mental models surrounding the risks and benefits of condom use and are important when considering potential predictors for condom use decisionmaking.
Unsurprisingly, coupled participants had significantly higher ratings of exclusivity and commitment towards their partners. Coupled respondents also reported having more sexual encounters in the past month but with fewer sexual partners than single participants. Those in couples were less concerned with STI risk than their single counterparts, but both coupled and single respondents were concerned about pregnancy
risk. Participants who were in committed relationships were less likely than their single counterparts to have used a condom the last time they had sex and were more likely to be using a LARC, which is consistent with previous research showing changes in birth control usage as couples become more serious (Manlove et al., 2011; Manning et al., 2009). This change could be due to a number of reasons, such as diminished sexual satisfaction, less concern about contracting an STI, or a greater concern about pregnancy risk spurring a move to a more effective form of contraception. Understanding why couples may be more willing to use other forms of contraception instead of condoms could inform targeted public health campaigns to promote condom use.
Predictors of Condom Use in Coupled Individuals
We found that significant predictors for increased condom use in coupled individuals included pregnancy concern and more positive condom use attitudes. The use of more effective forms of birth control was a significant predictor of decreased condom use. One reason this may be the case is that when one is in a committed couple, they are no longer concerned about STI risk. This may lead them to choose a birth control that is more effective, even if that birth control does not provide physical protection from an STI. This choice may be influenced further by a negative view of condoms (e.g., they ruin the mood, are uncomfortable, are expensive) or counteracted by a high concern for pregnancy where an individual may opt for “double protection” with a condom and another form of birth control. These results emphasize that the transition from condoms to hormonal forms of birth control as documented in the literature (Manlove et al., 2011; Manning et al., 2009; Wildsmith et al., 2015) can be modulated by a concern for pregnancy. This pregnancy concern may be an area of targeted public health campaigning for increased condom use and may be further exacerbated by increasing abortion restrictions in the United States (Wilkinson & Bernard, 2024).
Initially, we predicted that commitment and exclusivity among couples would have a notable impact on their pattern of condom use over time. D ue to the highly correlated nature of these variables we chose to only include exclusivity in our analysis due to prior research suggesting that only exclusivity, not commitment, is a predictor of decreased condom use in men (CraigKuhn et al., 2019). We predicted that couples who rated their relationship as more exclusive would be less likely to use condoms; however, we found that exclusivity was not a predictive variable. This may be due to limited variation in responses for exclusivity within our coupled respondents. Future research may benefit from having a
WINTER 2024
Stokes, Gulotta, Blansfield, Flanders, and Drais-Parrillo |
more detailed assessment for exclusivity (and commitment) in order to evaluate how these perceptions affect condom use patterns.
We originally predicted that an increase in the number of sexual encounters in the previous month would predict less frequent condom usage. Prior research pointed to the frequency of sex as a factor in condom usage as couples who had more sex tended to use condoms less (Wildsmith et al., 2015). This might imply a level of trust and perceived commitment as sex becomes more frequent with a monogamous partner or a concern about appearing mistrustful if one asked to use a condom in a committed relationship. Number of sexual encounters was not significantly correlated with predicting condom use in couples. This could be due to the relatively high levels of commitment and exclusivity within our coupled respondents representing a high level of trust. Perhaps once a couple decides not to use condoms, they will not revisit this decision if there are changes in their frequency of sexual encounters due to the trust they feel towards their partner.
Finally, we originally predicted that an increased sense of perceived invulnerability would decrease condom use as the individual would not feel that they were likely to have adverse outcomes (STI or pregnancy) if they did not use a condom. Because pregnancy concern was a significant predictor and perceived invulnerability was not, it may be that the scale used to determine perceived invulnerability did not measure what we thought it would. This may be due to our use of the Adolescent Invulnerability Scale (Duggan, 2014) or that perceived invulnerability generally may not affect perception of risk in sexual situations.
Predictors of Condom Use in Single Participants
We found that significant predictors for increased condom use in single individuals included both STI and pregnancy concern along with more positive condom use attitudes. These findings are important as they are targetable for public health campaigning and sexual health education. Overall, it is important to acknowledge the predictiveness of STI and pregnancy concerns for increasing condom usage. These results reiterate the importance of education on the efficacy of condoms at STI and pregnancy prevention to promote more consistent condom use.
We originally predicted that increased condom use selfefficacy would predict a more constant pattern of condom use; however this was not supported by our data. We introduced the selfefficacy measure to quantify what an individual thinks they know about condoms and their confidence in using them, which previously has shown to influence an individual’s decision to use
condoms (Baele et al., 2001; Wulfert & Wan, 1993). It is possible that selfefficacy may be related to, or even derive from, sexual education. For example, in our sample, both coupled and single participants had a mean composite score of just over four on a sevenpoint scale, corresponding to an average selection of “neutral” when asked questions about how confident they felt in using condoms. Future research might consider improving sexual education to see if there is an effect on condom use selfefficacy and condom use patterns.
We did not find any predictors of decreased condom use in single individuals as concern for selfpleasure, increased ratings of perceived invulnerability, and traits of sexual narcissism were not significant predictors of condom disuse. We originally predicted that concern for selfpleasure and increased sexually narcissistic traits would reduce condom use as single individuals have less communication with their casual partners. This could allow for more selfish behaviors as single individuals may not have an invested interest in their partner. Concerns about condoms decreasing pleasure are well documented in literature (Brown et al., 2008; Fennell, 2014; Higgins & Wang, 2015; Randolph et al., 2007), which made it surprising that it was not a significant predictor in our analysis. Their insignificance may be due to the way they were assessed.
Concern for sexual pleasure was gleaned from only two questions in our survey, not an entire scale, so a more extensive measure of this concern could increase its predictive power. It may also be that concerns about pleasure are overshadowed and trumped by the risks of not using a condom (contracting an STI or becoming pregnant) and that individuals are willing to “bite the bullet” to avoid those adverse outcomes. Potential concerns with how we measured perceived invulnerability were briefly addressed above with further discussion in the limitations section.
Limitations
Although we are confident in our results and the significance of our predictors, we must admit that there are limitations. An inherent limitation of a survey is that the data is selfreported and may be biased by social desirability. This limitation applies to the current study and should be considered when analyzing the conclusions we made based on the results of our analysis. Another inherent limitation is the presence of confounding third variables. One such example is drug and alcohol use during sexual encounters, which is assumed as commonplace among collegeaged individuals.
One significant limitation is the demographic variables that were collected. We did not collect demographic data on race or ethnicity in our sample. We also
PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH
Differences in Condom Use Patterns | Stokes, Gulotta, Blansfield, Flanders, and Drais-Parrillo
did not collect socioeconomic data on our participants. We felt that asking participants about socioeconomic status may have been confusing and resulted in unreliable results as students could report their own personal yearly earnings or their parents yearly earnings, neither of which may accurately depict their current socioeconomic status as a college student. However, regardless of the justification, the lack of demographic data may affect the generalizability of our results.
The use of the Adolescent Vulnerability Scale is another important limitation in our methods as this scale was validated in a different population than ours. There is debate on the true age range of an adolescent, particularly concerning socalled “delayed adolescence” in collegeaged individuals. However, this limitation is important to disclose and may have impacted our results. Another limitation is the categorization of commitment and exclusivity from continuous data. We chose to analyze these data after assigning cutoffs to the percentage reported, which introduces bias and may not have fully captured the participants’ true feelings about the degree of commitment and exclusivity they felt towards their partner.
Future Directions
Understanding predictors of when collegeaged individuals chose to use condoms may allow researchers to be able to create better, targeted tactics to promote condom use in the future. In particular, understanding differences in coupled and single individuals when they exist can inform public health strategies to improve contraceptive education and promote condom use. As a team of reproductiveaged women, we are invested in issues involving women’s rights and sexual education in our country, especially following the politicization of healthcare and access due to the Dobbs v. Jackson Women’s Health Organization ruling. Future research in the coming years is essential to understanding the effects of decreased access to reproductive healthcare across the United States.
A different factor that may affect condom use in young adults is a lack of education about the efficacy of condoms. Numerous studies have shown that sexual educators in the United States education system lack the training to properly educate students on safe sexual practices, which is why many adolescents and adults may be unaware of the efficacy of condoms (RamírezVillalobos et al., 2021; Seaborne et al., 2015). Perhaps if individuals received more robust education on the efficacy of condom use and practice on using a condom, we may see an increase in condom usage and a decrease in STIs and unintended pregnancies among young adults.
References
Amy, J.-J., & Thiery, M. (2015). The condom: A turbulent history. The European Journal of Contraception & Reproductive Health Care, 20(5), 387–402. https://doi.org/10.3109/13625187.2015.1050716
Baele, J., Dusseldorp, E., & Maes, S. (2001). Condom use self-efficacy: Effect on intended and actual condom use in adolescents. Journal of Adolescent Health, 28(5), 421–431. https://doi.org/10.1016/S1054-139X(00)00215-9
Basting, E. J., Barrett, M. E., Garner, A. R., Florimbio, A. R., Sullivan, J. A., Medenblik, A. M., & Stuart, G. L. (2023). Sexual narcissism and hypersexuality relate to sexual coercion in hookups among U.S. university students. Archives of Sexual Behavior, 52(6), 2577–2588. https://doi.org/10.1007/s10508-023-02580-z
Brown, L. K., DiClemente, R., Crosby, R., Fernandez, M. I., Pugatch, D., Cohn, S., Lescano, C., Royal, S., Murphy, J. R., Silver, B., Schlenger, W. E., & For the Project Shield Study Group. (2008). Condom use among high-risk adolescents: Anticipation of partner disapproval and less pleasure associated with not using condoms. Public Health Reports, 123(5), 601–607. https://doi.org/10.1177/003335490812300510
Centers for Disease Control and Prevention (CDC), Brener, N. D., Kann, L., Shanklin, S., Kinchen, S., Eaton, D. K., Hawkins, J., Flint, K. H., & Centers for Disease Control and Prevention (CDC). (2013). Methodology of the Youth Risk Behavior Surveillance System—2013. MMWR. Recommendations and Reports: Morbidity and Mortality Weekly Report. Recommendations and Reports, 62(RR-1), 1–20. https://www.cdc.gov/mmwr/preview/mmwrhtml/rr6201a1.htm
Constant, E., Vallet, F., Nandrino, J.-L., & Christophe, V. (2016). Personal assessment of intimacy in relationships: Validity and measurement invariance across gender. European Review of Applied Psychology, 66(3), 109–116. https://doi.org/10.1016/j.erap.2016.04.008
Copen, C. E. (2017). Condom use during sexual intercourse among women and men aged 15–44 in the United States: 2011–2015 National Survey of Family Growth. National Health Statistics Reports, 105, 1–18.
Craig-Kuhn, M. C., Schmidt, N., Gomes, G., Scott, G., Watson, S., Lederer, A., & Kissinger, P. (2019). O12.2 Partnership context and consistent condom use among young African American men. Oral Presentations, A65.2-A66. https://doi.org/10.1136/sextrans-2019-sti.171
Duggan, P. M. (2014). The Adolescent Invulnerability Scale [dataset]. https://doi.org/10.1037/t31051-000
Fennell, J. (2014). “And isn’t that the point?”: Pleasure and contraceptive decisions. Contraception, 89(4), 264–270. https://doi.org/10.1016/j.contraception.2013.11.012
Grabovac, I., Smith, L., Yang, L., Soysal, P., Veronese, N., Turan Isik, A., Forwood, S., & Jackson, S. (2020). The relationship between chronic diseases and number of sexual partners: An exploratory analysis. BMJ Sexual & Reproductive Health, 46(2), 100–107. https://doi.org/10.1136/bmjsrh-2019-200352
Hatcher, R. A. (2007). Contraceptive Technology. Ardent Media Inc.
Higgins, J. A., & Wang, Y. (2015). The role of young adults’ pleasure attitudes in shaping condom use. American Journal of Public Health, 105(7), 1329–1332. https://doi.org/10.2105/AJPH.2015.302567
Hill, P. L., Duggan, P. M., & Lapsley, D. K. (2012). Subjective invulnerability, risk behavior, and adjustment in early adolescence. The Journal of Early Adolescence, 32(4), 489–501. https://doi.org/10.1177/0272431611400304
Holmes, K. K., Levine, R., & Weaver, M. (2004). Effectiveness of condoms in preventing sexually transmitted infections. Bulletin of the World Health Organization, 82(6), 454–461. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2622864/ Kusunoki, Y., & Barber, J. S. (2020). The Dynamics of intimate relationships and contraceptive use during early emerging adulthood. Demography, 57(6), 2003–2034. https://doi.org/10.1007/s13524-020-00916-1
Lehmiller, J. J., VanderDrift, L. E., & Kelly, J. R. (2014). Sexual communication, satisfaction, and condom use behavior in friends with benefits and romantic partners. The Journal of Sex Research, 51(1), 74–85. https://doi.org/10.1080/00224499.2012.719167
Manlove, J., Welti, K., Barry, M., Peterson, K., Schelar, E., & Wildsmith, E. (2011). Relationship characteristics and contraceptive use among young adults. Perspectives on Sexual and Reproductive Health, 43(2), 119–128. https://doi.org/10.1363/4311911
Manning, W. D., Flanigan, C. M., Giordano, P. C., & Longmore, M. A. (2009). Relationship dynamics and consistency of condom use among adolescents. Perspectives on Sexual and Reproductive Health, 41(3), 181–190. https://doi.org/10.1363/4118109
Matser, A., Heiligenberg, M., Geskus, R., Heijman, T., Low, N., Kretzschmar, M., & Schim Van Der Loeff, M. (2014). The importance of partnership factors and
WINTER 2024
Stokes, Gulotta, Blansfield, Flanders, and Drais-Parrillo | Differences in Condom Use Patterns
individual factors associated with absent or inconsistent condom use in heterosexuals: A cross-sectional study. Sexually Transmitted Infections, 90(4), 325–331. https://doi.org/10.1136/sextrans-2013-051087
Raine, T. R., Gard, J. C., Boyer, C. B., Haider, S., Brown, B. A., Ramirez Hernandez, F. A., & Harper, C. C. (2010). Contraceptive decision-making in sexual relationships: Young men’s experiences, attitudes and values. Culture, Health & Sexuality, 12(4), 373–386. https://doi.org/10.1080/13691050903524769
Ramírez-Villalobos, D., Monterubio-Flores, E. A., Gonzalez-Vazquez, T. T., MolinaRodríguez, J. F., Ruelas-González, Ma. G., & Alcalde-Rabanal, J. E. (2021). Delaying sexual onset: Outcome of a comprehensive sexuality education initiative for adolescents in public schools. BMC Public Health, 21(1), 1439. https://doi.org/10.1186/s12889-021-11388-2
Randolph, M. E., Pinkerton, S. D., Bogart, L. M., Cecil, H., & Abramson, P. R. (2007). Sexual pleasure and condom use. Archives of Sexual Behavior, 36(6), 844–848. https://doi.org/10.1007/s10508-007-9213-0
Seaborne, L. A., Prince, R. J., & Kushner, D. M. (2015). Sexual health education in U.S. physician assistant programs. The Journal of Sexual Medicine, 12(5), 1158–1164. https://doi.org/10.1111/jsm.12879
Spinner, C. D., Boesecke, C., Zink, A., Jessen, H., Stellbrink, H.-J., Rockstroh, J. K., & Esser, S. (2016). HIV pre-exposure prophylaxis (PrEP): A review of current knowledge of oral systemic HIV PrEP in humans. Infection, 44(2), 151–158. https://doi.org/10.1007/s15010-015-0850-2
Thompson, S. C., Anderson, K., Freedman, D., & Swan, J. (1996). Illusions of safety in a risky world: A study of college students’ condom use. Journal of Applied Social Psychology, 26(3), 189–210. https://doi.org/10.1111/j.1559-1816.1996.tb01845.x
Trussell, J. (2011). Contraceptive failure in the United States. Contraception, 83(5), 397–404. https://doi.org/10.1016/j.contraception.2011.01.021
Volberding, P. A., & Deeks, S. G. (2010). Antiretroviral therapy and management of HIV infection. The Lancet, 376(9734), 49–62. https://doi.org/10.1016/S0140-6736(10)60676-9
Walsh-Buhi, E. R., & Helmy, H. L. (2018). Trends in long-acting reversible contraceptive (LARC) use, LARC use predictors, and dual-method use among a national sample of college women. Journal of American College Health, 66(4), 225–236. https://doi.org/10.1080/07448481.2017.1399397
Weitzman, A., Barber, J., & Kusunoki, Y. (2019). Sexual concurrency and contraceptive use among young adult women. Demography, 56(2), 549–572. https://doi.org/10.1007/s13524-019-00762-w
Widman, L., & McNulty, J. K. (2010). Sexual narcissism and the perpetration of
sexual aggression. Archives of Sexual Behavior, 39(4), 926–939. https://doi.org/10.1007/s10508-008-9461-7
Wildsmith, E., Manlove, J., & Steward‐Streng, N. (2015). Relationship characteristics and contraceptive use among dating and cohabiting young adult couples. Perspectives on Sexual and Reproductive Health, 47(1), 27–36. https://doi.org/10.1363/47e2515
Wilkinson, T. A., & Bernard, C. (2024). Abortion restrictions and the impact on families. JAMA Pediatrics, 178(1), 15. https://doi.org/10.1001/jamapediatrics.2023.4735
Workowski, K. A., Bolan, G. A., & Centers for Disease Control and Prevention. (2015). Sexually transmitted diseases treatment guidelines, 2015. MMWR. Recommendations and Reports: Morbidity and Mortality Weekly Report. Recommendations and Reports, 64(RR-03), 1–137. https://www.cdc.gov/mmwr/preview/mmwrhtml/rr6403a1.htm
Wulfert, E., & Wan, C. K. (1993). Condom use: A self-efficacy model. Health Psychology, 12(5), 346–353. https://doi.org/10.1037/0278-6133.12.5.346
Youssef, H. (1993). The history of the condom. Journal of the Royal Society of Medicine, 86(4), 226–228. https://doi.org/10.1177/014107689308600415
Author Note
Alicia DraisParrillo (aad129@psu.edu) https://orcid.org/0009000847144374
Kylie R. Gulotta (krg5504@psu.edu) has now graduated from the Department of Psychology
Rachel E. Blansfield (Rachel.blansfield1@gmail.com) has graduated from the Department of Psychology and is now at the Department of Medicine, Pennsylvania State University, Hershey campus.
Mackenzie S. Flanders (mackenziesflanders@gmail.com) has now graduated from the Department of Psychology. Gabrielle O. Stokes (stokes.g.o4711@gmail.com) has now graduated from the Departments of Psychology and Science in May 2024.
The authors report there are no competing interests to declare. This research received no external funding. Participants were anonymous and gave electronic consent; authors complied with IRB regulations and data was applied anonymously. Scales used for this study are available at https://osf.io/bgc4m/ Correspondence concerning this article should be addressed to Alicia DraisParrillo at aad129@psu.edu
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, doctorallevel, peerreview 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
Doctorallevel 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
Learn About Psi Chi
Psi Chi is the International Honor Society in Psychology. Membership is primarily open to undergraduates, graduate students, transfer students, fulltime and parttime faculty members, and alumni.
See membership benefits and a link to apply at www.psichi.org/page/member_benefits Register an account: http://pcj.msubmit.net/cgi-bin/main.plex
