Journal of Personnel Psychology Issue 1, 2019

Volume 18 / Number 1 / 2019

Journal of

Personnel Psychology Editor-in-Chief Bernd Marcus Managing Editor Petra Gelléri Associate Editors Tanja Bipp Ian Gellatly Ilke Inceoglu Jonas Lang Laurenz Meier Cornelia Niessen Xin-An Zhang

Test development and construction: Current practices and advances “This book is indispensable for all who want an up-to-date resource about constructing valid tests.”

Prof. Dr. Johnny R. J. Fontaine, President of the European Association of Psychological Assessment, Faculty of Psychology and Educational Sciences, Ghent University, Belgium

Karl Schweizer / Christine DiStefano (Editors)

Principles and Methods of Test Construction Standards and Recent Advances

(Series: Psychological Assessment – Science and Practice – Vol. 3) 2016, vi + 336 pp.

US $69.00 / € 49.95 ISBN 978-0-88937-449-2 Also available as eBook This latest volume in the series Psychological Assessment – Science and Practice describes the current state-of-the-art in test development and construction. The past 10–20 years have seen substantial advances in the methods used to develop and administer tests. In this volume many of the world’s leading authorities collate these advances and provide information about current practices, thus equipping researchers and students to successfully construct new tests using the best modern standards and

www.hogrefe.com

techniques. The first section explains the benefits of considering the underlying theory when designing tests, such as factor analysis and item response theory. The second section looks at item format and test presentation. The third discusses model testing and selection, while the fourth goes into statistical methods that can find group-specific bias.

The final section discusses topics of special relevance, such as multitraitmultimethod analyses and development of screening instruments.

Journal of

Personnel Psychology Volume 18 / Number 1/2019

Editor-in-Chief

Bernd Marcus, Organizational and Personnel Psychology, Institute of Business Administration, University of Rostock, Ulmenstr. 69, 18057 Rostock, Germany. Tel. +49 381 498-4080, Fax +49 381 498-4419, E-mail: bernd.marcus@uni-rostock.de

Managing Editor

Petra Gelléri, Work and Organizational Psychology, Faculty of Psychology, University of Hagen, Universitätsstr. 33, 58084 Hagen, Germany, Tel. +49 2331 987-2745, Fax +49 2331 987-2179, E-mail: jpp.editorial.office@gmail.com

Associate Editors

Tanja Bipp, University of Würzburg, Germany Ian Gellatly, University of Alberta, Canada Ilke Inceoglu, University of Exeter, UK Jonas Lang, Ghent University, Belgium Laurenz Meier, University of Neuchâtel, Switzerland Cornelia Niessen, Erlangen, Germany Xin-An Zhang, Shanghai Jiao Tong University, China

Editorial Board

Mike Ashton, Canada Arnold Bakker, The Netherlands Gerhard Blickle, Germany Diana Boer, Germany Janine Bosak, Ireland John Campbell, USA Oliver Christ, Germany Neil Christiansen, USA Brian Connelly, Canada Jeremy Dawson, UK Nele de Cuyper, Belgium Filip De Fruyt, Belgium Evangelia Demerouti, The Netherlands Deanne den Hartog, The Netherlands Patrick Dunlop, Australia Jörg Felfe, Germany Steffen Giessner, The Netherlands Richard Goffin, Canada Martin Götz, Switzerland Barbara Griffin, Australia Peter Harms, USA Alex Haslam, UK Sarah Hezlett, USA Giles Hirst, Australia Stefan Höft, Germany Thomas Jønsson, Denmark Anita C. Keller, The Netherlands Rudolf Kerschreiter, Germany Ulla Kinnunen, Finland Martin Kleinmann, Switzerland Cornelius König, Germany Franciska Krings, Switzerland Kibeom Lee, Canada Klaus Melchers, Germany

Publisher

Hogrefe Publishing, Merkelstr. 3, 37085 Göttingen, Germany, Tel. +49 551 99950-0, Fax +49 551 99950-425, E-mail publishing@hogrefe.com, Web http://www.hogrefe.com North America: Hogrefe Publishing, 7 Bulfinch Place, Suite 202, Boston, MA 02114, USA, Phone (866) 823-4726, Fax (617) 354-6875, E-mail publishing@hogrefe.com

Production

Christina Sarembe, Hogrefe Publishing, Merkelstr. 3, 37085 Göttingen, Germany, Tel. +49 551 99950-0, Fax +49 551 99950-425, E-mail publishing@hogrefe.com

Subscriptions

Hogrefe Publishing, Herbert-Quandt-Str. 4, 37081 Göttingen, Germany, Tel. +49 551 99950-900, Fax +49 551 99950-998

Advertising/Inserts

Marketing, Hogrefe Publishing, Merkelstr. 3, 37085 Göttingen, Germany, Tel. +49 551 99950-423, Fax +49 551 99950-425, E-mail marketing@hogrefe.com

ISSN

ISSN-L 1866-5888, ISSN-Print 1866-5888, ISSN-Online 2190-5150

Copyright Information

Ó 2018 Hogrefe Publishing. This journal as well as the individual contributions and illustrations contained within it are protected under international copyright law. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without prior written permission from the publisher. All rights, including translation rights, reserved.

Publication

Published in 4 issues per annual volume.

Subscription Prices

Calendar year subscriptions only. Rates for 2019: Institutions – from US $308.00/ 232.00 (print only; pricing for online access can be found in the journals catalog at hgf.io/journals2019); Individuals – US $159.00/ 114.00 (print & online). Postage and handling – US $16.00/ 12.00.

Single copies – US $81.00/ 62.50 + postage and handling.

Payment

Payment may be made by check, international money order, or credit card, to Hogrefe Publishing, Merkelstr. 3, 37085 Göttingen, Germany, or, for North American customers, to Hogrefe Publishing, 7 Bulfinch Place, Suite 202, Boston, MA 02114, USA.

Abstracting Services

Abstracted/indexed in Social Sciences Citation Index (SSCI), Social SciSearch, Current Contents/Social and Behavioral Sciences (CC/S&BS), PsycINFO, PsyJOURNALS, PSYNDEX, Scopus, IBZ, IBR, ERIH.

Electronic Full-Text

The full-text of the Journal of Personnel Psychology is available online at www.econtent.hogrefe.com and also in PsycARTICLES.

Bertolt Meyer, Germany John P. Meyer, Canada Karin S. Moser, UK Klaus Moser, Germany Peter Muck, Germany Laetitia Mulder, The Netherlands Ioannis Nikolaou, Greece Sandra Ohly, Germany Janneke K. Oostrom, The Netherlands Lisa Penney, USA Deborah Powell, Canada Floor Rink, The Netherlands Ann Marie Ryan, USA Paul R. Sackett, USA Jesus F. Salgado, Spain Niclas Schaper, Germany Sebastian Schuh, China Birgit Schyns, France Meir Shemla, The Netherlands Roy B. L. Sijbom, The Netherlands Christiane Spitzmüller, USA Daan Stam, The Netherlands Thomas Staufenbiel, Germany Sebastian Stegmann, Germany H. Canan Sümer, Turkey Klaus J. Templer, Singapore Robert Tett, USA Christian Vandenberghe, Canada Rolf van Dick, Germany Chockalingam Viswesvaran, USA S. Arzu Wasti, Turkey Juergen Wegge, Germany Despoina Xanthopoulou, Greece Ingo Zettler, Denmark

2017 Impact Factor 1.146, 5-year Impact Factor 2.088, Journal Citation Reports (Clarivate Analytics, 2018)

Journal of Personnel Psychology (2019), 18(1)

Ó 2019 Hogrefe Publishing

Contents Original Articles

Is It Possible to Improve Test Takers’ Perceptions of Ability Tests

by Providing an Explanation?

Klaus G. Melchers and Barbara Körner

1

Not All Leaving Is Created Equal: Differentiating the Factors of Organizational and Occupational Turnover Intentions Huisi (Jessica) Li, Kun Yu, Youhuang Huang, and Xiaotong Jin

10

When Do Those High in Trait Self-Control Suffer From Strain?

The Interplay of Trait Self-Control and Multiple Stressors Kai Externbrink, Stefan Diestel, and Martina Krings

23

Perceived Overqualification and Psychological Well-Being Among Immigrants: The Moderating Role of Personal Resources Maria Wassermann and Annekatrin Hoppe

34

Research Note

Measuring High-Quality Work Relationships: A Test of Model and Gender Invariance Michael T. Warren and Meg A. Warren

46

News and Announcements

Awards for Outstanding Achievements as Authors and Reviewers 2018

53

Ó 2019 Hogrefe Publishing

Journal of Personnel Psychology (2019), 18(1)

Original Article

Is It Possible to Improve Test Takers’ Perceptions of Ability Tests

by Providing an Explanation?

Klaus G. Melchers1

and Barbara Körner2

1

Work and Organizational Psychology, Institute of Psychology and Education, Ulm University, Ulm, Germany

2

Work and Organizational Psychology, Department of Psychology, University of Zurich, Switzerland

Abstract: Previous meta-analytic findings have revealed that explanations can improve applicants’ perceptions of selection procedures.

However, they also suggest that these positive effects do not generalize to ability tests. Given some limitations of previous studies and the small empirical basis for the corresponding meta-analytic results, we had another look at whether perceptions of ability tests can be improved by providing an explanation. In two experimental studies, participants had to complete either an attention or a general mental ability test. In the explanation group, a justification was given concerning the content, relevance, and predictiveness of the test. In contrast, no explanation was given in the control group. Providing an explanation significantly improved test takers’ fairness perceptions.

Keywords: applicant reactions, explanations, procedural fairness, ability tests

In addition to the psychometric properties of a selection procedure, it is also important to consider how a procedure is perceived by applicants because these perceptions are related to several relevant outcomes including perceptions of organizational attractiveness, intentions to proceed with a selection process, to accept a job offer, to reapply, or to recommend the organization to others (cf. Hausknecht, Day, & Thomas, 2004, or McCarthy, Bauer, Truxillo, Anderson, et al. 2017, for reviews). Furthermore, recent research also found effects of applicants’ perceptions on actual job offer acceptance (Harold, Holtz, Griepentrog, Brewer, & Marsh, 2016; Konradt, Garbers, Böge, Erdogan, & Bauer, 2017). In light of these findings, it seems unfortunate that valid psychometric tests are often perceived less positively in comparison with other more expensive selection procedures like work samples or interviews (Anderson, Salgado, & Hülsheger, 2010; Hausknecht et al., 2004).

Providing explanations to applicants has been suggested as a cost-effective and straightforward way to positively influence their perceptions of such tests and meta-analytic work by Truxillo, Bodner, Bertolino, Bauer, and Yonce (2009) found that explanations can have beneficial effects on perceptions of a selection procedure. Unfortunately, this meta-analysis did not confirm such a beneficial effect of explanations for ability tests such as tests of general mental ability (GMA) or of specific abilities. This is rather unfortunate given the good validity of ability tests (e.g., Bertua, Ó 2019 Hogrefe Publishing

Anderson, & Salgado, 2005; Schmidt & Hunter, 2004).

However, a limitation of Truxillo et al.’s meta-analytic result is that it is based on a rather small empirical basis.

Furthermore, many of the previous studies on effects of explanations on perceptions of ability tests did not use explanations that were related to the content and job relevance of the test.

The aim of our research was to address these limitations and thereby to extend previous research on the beneficial effects of explanations. Specifically, we hypothesized that providing a justification that focuses on the content of a test and the relevance and predictiveness of it can improve test takers’ perceptions of this test – similar to findings for other selection procedures (Truxillo et al., 2009). Accordingly, we set up two studies to test whether this is indeed the case.

Review of Previous Research During the last decades, several theoretical frameworks have been put forth that describe relevant determinants of applicant reactions (cf. Truxillo & Bauer, 2011). The most influential framework was proposed by Gilliland (1993; also see Gilliland & Steiner, 2012) whose model of applicant reactions is based on organizational justice theory. It defines a set of justice rules that are pertinent to the Journal of Personnel Psychology (2019), 18(1), 1–9 https://doi.org/10.1027/1866-5888/a000212

2

perceived fairness of a selection process and that comprise aspects of procedural, distributive, informational, and interpersonal justice. According to the model, these justice rules determine the perceived fairness of a selection procedure and applicants react negatively in situations when the rules are violated. Most of the applicant reactions research during the last two decades used this model (see, e.g., reviews

by McCarthy, Bauer, Truxillo, Anderson, et al., 2017, or Truxillo & Bauer, 2011), and most of Gilliland’s rules were shown to be relevant for subsequent applicant reactions (Hausknecht et al., 2004; McCarthy, Bauer, Truxillo, Anderson, et al., 2017). Furthermore, in their recent review of applicant reactions research, McCarthy, Bauer, Truxillo, Anderson, et al. noted that the provision of explanations to applicants was one of the aspects that were most consistently related to fairness perceptions.

Conceptually, explanations can contribute to several of the procedural justice rules suggested in Gilliland’s (1993) model. First, explanations are relevant with regard to the provision of information about the administration of selection procedures or the necessity to use certain procedures.

Furthermore, explanations can also be used to provide information on the job-relatedness and predictiveness of a selection procedure. And explanations can also be used to provide information about other applicants (e.g., in case of a rejection that others were more qualified) or about practices by other organizations that use or do not use similar selection procedures.

Possible explanations fall into two major categories: excuses and justifications. Excuses are explanations in which the actor denies full responsibility for a process or an outcome because this process or outcome was (at least to some degree) beyond his or her control. In contrast to this, justifications are explanations in which the actor stresses the appropriateness or ethicality of a process or outcome. According to Truxillo et al.’s (2009) meta-analysis, both justifications and excuses have positive effects on applicants’ perceptions, but especially justifications seem to be beneficial for perceptions of a selection procedure.

Furthermore, explanations also contributed to improved perceptions of the organization and to an enhancement of applicants’ test taking motivation. Finally, according to the meta-analysis, the positive effects of explanations hold regardless of the selection outcome.

Unfortunately, as already noted above, Truxillo et al.

(2009) found no beneficial effect of explanations that were related to ability tests such as GMA tests or tests of specific abilities. And even though previous research found that applicants in general perceive GMA tests as more favorable in comparison with some other selection procedures such as personality or integrity tests (Anderson et al., 2010; Hausknecht et al., 2004), this relatively positive perception is not true for all ability tests. A comparison of several Journal of Personnel Psychology (2019), 18(1), 1–9

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

common German ability tests by Kersting (2008), for example, revealed considerable differences with regard to the overall acceptability of different tests and found that especially more abstract tests were perceived rather negatively by test takers. Accordingly, it would be beneficial if these perceptions could be improved by explanations.

A limitation of Truxillo et al.’s (2009) meta-analytic result for ability tests is that it rests on a rather small empirical basis. Specifically, the meta-analytic estimate for ability tests is based on only six primary studies. And we are not aware of any additional studies since then that dealt with the effects of explanations on test takers’ perceptions of ability tests even though Hausknecht et al. (2004) already called for more research “to examine the potential benefits of interventions that could improve applicant reactions . . . such as providing explanations” (p. 674) quite some time ago.

With regard to the specific primary studies that investigated effects of explanations in the context of ability tests – and in line with Truxillo et al.’s (2009) meta-analytic result – positive effects were only found in a few cases.

Specifically, LaHuis, Perreault, and Ferguson (2003) found higher procedural fairness perceptions and more positive perceptions of content and predictive validity after an explanation, and Rolland and Steiner (2007) found positive effects on perceived predictive validity and propriety of questions. Both studies used selection simulations, and participants were told that the respective tests were good predictors of job performance. In addition, Rolland and Steiner also provided a justification for the alleged selection decision in which participants were given “comparative information of participants’ scores to the results of other ‘applicants’” (p. 2808). However, in both studies, participants had to complete not only a GMA test but also a personality test (LaHuis et al., 2003) or an overt integrity test (Rolland & Steiner, 2007) and the explanation was directed at both tests at the same time. Furthermore, specific effects concerning perceptions of the GMA test were not collected in the former study and not reported in the latter one.

In contrast to these few positive effects of explanations on perceptions of ability tests, results from other studies (Horvath, Ryan, & Stierwalt, 2000; Lievens, De Corte, & Brysse, 2003) or with additional groups or additional variables in the studies by LaHuis et al. (2003) and by Rolland and Steiner (2007) found nonsignificant effects when treatment groups were compared to a no-explanation control group. In these studies, participants were provided with information about test reliability and validity (Lievens et al., 2003), the use of much more rigorous tests by other organizations (Horvath et al., 2000), or the need for “a better way of weeding out poor performers” after a “wave of poor hiring decisions” (Horvath et al., 2000, p. 315). Furthermore, a more extensive explanation in LaHuis et al.’s Ó 2019 Hogrefe Publishing

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

(2003) study that also covered information about the way how job requirements were identified and about test content (i.e., cognitive ability and personality) had no effect.

Finally, an explanation that informed test takers about the hope that the test did not discriminate with regard to race, gender, age, and ethnicity even led to significantly lower face validity perceptions in comparison with the control group (Horvath et al., 2000).

As can be seen in this review of earlier research, many of the explanations that were used in previous studies with ability tests were not specifically related to the content and the relevance of the test for the job in question. Instead, they targeted aspects such as reduced discrimination or practices by other organizations or they contained relatively abstract information on reliability and validity. In contrast to this, research related to other selection procedures such as personality tests repeatedly found that explanations concerning the job relevance and the predictiveness of the test results led to improved perceptions of job relatedness (e.g., Holtz, Ployhart, & Dominguez, 2005).

Given the state of previous research on effects of explanations on test takers’ perceptions of ability tests, we wanted to evaluate the hypothesis that a justification that focuses on test content and relevance as well as on the predictiveness of the test can improve test takers’ perceptions.

To test this hypothesis, we conducted two studies. In Study 1, we used a specific ability test that was perceived rather negatively in previous research, and in Study 2, we used a GMA test to evaluate the generalizability of the results from the first study.

Study 1 Method Sample The sample consisted of 96 university students (Mage = 23.17 years, SD = 3.82; 49% females), most of whom were majoring in the natural sciences, engineering, or medicine.

No psychology majors beyond the first semester were allowed to participate to prevent that study participants had previous training in psychometric testing. Of the participants, 73% had previously already applied for a job and 28% had completed selection tests before. Additionally, 49% of the participants already had work experience.

Procedure We asked participants to describe an attractive job in their field of study for which they would apply after completing their degree. Then, they were told that the following test is often used in selection settings and that successful

Ó 2019 Hogrefe Publishing

3

completion of it influences hiring decisions. To further enhance their motivation for the later administration of the test (a common attention test, see below for more information), we promised participants gift certificates for a local cinema for the 10 individuals with the highest test scores.

Following this instruction, participants were randomly allocated to the control group and the explanation group.

Participants in the explanation group were provided with a justification containing three important pieces of information about the use of the attention test in selection contexts: Specifically, test takers were informed (a) that the test is an established procedure to measure attention and concentration, (b) that attention and concentration are important for challenging jobs that require simultaneous attention to various different aspects, and (c) that performance on this test is predictive of one’s ability for such tasks. Following this explanation, participants were provided with the actual test instruction and the test was administered. In contrast to the explanation group, participants in the control group were only presented with the test instruction but not the justification concerning content, job relatedness, and predictive validity of the test. After the test, all participants had to complete a questionnaire concerning their perceptions of the test.

Ability Test A shortened version of d2-R test of attention (Brickenkamp, Schmidt-Atzert, & Liepmann, 2010) was used. This paper– pencil test measures selective attention and its suitability and validity for tasks that require simultaneous attention has been well-documented (cf. Brickenkamp et al., 2010).

However, Kersting’s (2008) comparison of several common German ability tests had revealed that the d2-R was perceived rather negatively.

In the d2-R, participants are presented with rows of letters and their task is to cross out any letter d that has two dashes printed below or above (i.e., ds with two dashes printed below, two dashes printed above, or one dash printed below and another one printed above are valid targets). Furthermore, each row also contains various distractor letters that are similar to the target (i.e., ps with dashes or ds with one or three dashes). Participants have 20 s to complete each row before they are told to carry on with the next row irrespective of how far they have proceeded in the current row.

For the current study, all participants were administered 9 rows of letters. In line with the test manual, test performance was calculated by subtracting all errors (letters that were incorrectly crossed out as well as missed targets) from the overall number of letters, which was determined by summing the number of letters completed in each row.

Journal of Personnel Psychology (2019), 18(1), 1–9

4

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

Test Takers’ Perceptions We used different scales to assess test takers’ perceptions.

Specifically, we measured perceived face validity (α = .85, e.g., “I did not understand what the test had to do with my targeted job.”) and perceived predictive validity (α = .85, e.g., “My performance on the test is a good indicator of my ability to do the targeted job.”) with two 5-item scales that were adapted from Smither, Reilly, Millsap, Pearlman, and Stoffey (1993) and that were also used in previous research on the effects of explanations on test takers’ perceptions (e.g., Holtz et al., 2005). Overall procedural fairness (α = .70, e.g., “Overall, the test used was fair.”) was measured with a 3-item procedural justice scale from Bauer et al. (2001). Finally, test motivation (α = .79, e.g., “Doing well on this test was important to me.”) was assessed with three items from Arvey, Strickland, Drauden, and Martin (1990). Five-point scales (1 = strongly disagree to 5 = strongly agree) were used for all items.

Table 1 shows descriptive information and correlations for all study variables. In line with previous research (Brickenkamp et al., 2010), there was a significant negative correlation between age and test performance, r = .26, p < .05.

Means and SDs for the dependent variables for the two groups and effect sizes from tests of mean differences between the two groups are presented in Table 2. In line with our hypothesis, means for test takers’ perceptions were more positive in the explanation group. Accordingly, independent-samples t-tests also confirmed that the ratings were significantly higher for face validity, t(94) = 3.72, p < .01, Cohen’s d = .76, predictive validity, t(94) = 3.96, p < .01, Cohen’s d = .81, and procedural fairness t(94) = 2.03, p < .05, Cohen’s d = .41, for the group that had received a justification compared to the control group.

However, the difference for participants’ test motivation was not significant, t < 1. Furthermore, there was also no significant difference between the groups with regard to their test performance, t < 1.

Results First, we determined whether the experimental groups were comparable with regard to demographic and background variables. We found no significant differences with regard to age, gender, number of previous job applications, or previous experience with selection tests, all ts < 1.63, all ps > .10.

Discussion In contrast to previous research (e.g., Truxillo et al., 2009), our results suggest that providing an explanation can have beneficial effects on test takers’ perceptions of an ability test that measures attention: Participants who received a justification before the test had more positive perceptions

Table 1. Descriptive statistics and intercorrelations of study variables in Study 1 Variables 1. Age

M

(SD)

23.17

(3.82)

1.

2.

3.

4.

5.

2. Gendera

0.49

(0.50)

3. Face validity

2.92

(0.86)

.10

.20

4. Predictive validity

2.48

(0.82)

.14

.18

5. Procedural fairness

3.15

(0.72)

.01

.16

.46**

.51**

6. Test motivation

4.24

(0.56)

.02

.20

.21*

.13

.15

143.06

(31.39)

.26*

.19

.16

.07

.09

7. Test performance

6.

.28** .66**

.02

a

Note. N = 96. Gender was coded as 0 = male, 1 = female. *p < .05, **p < .01.

Table 2. Means and standard deviations for the two experimental groups in Study 1 No explanation (n = 46)

Explanation (n = 50)

Variables

M

(SD)

M

(SD)

Face validity

2.60

(0.86)

3.21

(0.75)

Effect size .76**

Predictive validity

2.16

(0.75)

2.78

(0.77)

.81**

Procedural fairness

2.99

(0.76)

3.29

(0.66)

.41*

Test motivation

4.22

(0.60)

4.26

(0.52)

.08

146.39

(31.17)

140.00

(31.59)

.20

Test performance

Note. Effect sizes depict Cohen’s d. Positive d values indicate that the experimental group that had received an explanation had higher means than the noexplanation control group. Asterisks represent significance levels from independent-samples t-tests. *p < .05, **p < .01.

Journal of Personnel Psychology (2019), 18(1), 1–9

Ó 2019 Hogrefe Publishing

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

after the test with regard to face validity, predictive validity, and procedural fairness. However, the justification did not affect participants’ test motivation or test performance significantly.

Even though the present results seem promising in light of Truxillo et al.’s (2009) findings with regard to the possibility to improve applicants’ perceptions of ability tests, some caution is still necessary. Specifically, considering the specific nature of the ability test that we used, our results cannot be taken as evidence that a justification will always suffice to improve test takers’ perceptions of other ability tests and especially of GMA tests as well. However, the question of whether our results generalize to GMA tests seems highly relevant given that there is evidence that GMA tests have better criterion-related validity (e.g., Schmidt & Hunter, 2004) than attention tests or perceptual tests in general (cf. Hunter & Hunter, 1984) and also given their higher usage. Furthermore, as already noted above, previous research also found that applicants’ perceptions of GMA tests are more favorable in comparison with personality or integrity tests (Anderson et al., 2010; Hausknecht et al., 2004). In contrast to this, the d2-R was perceived more negatively than most of the other ability tests covered in Kersting’s (2008) study. Thus, given the rather negative perceptions of the d2-R, it might be easier to improve test takers’ perceptions (at least to a certain degree) of this test than of a GMA test. Therefore, we set up Study 2 to evaluate to which degree the results from Study 1 generalize to a GMA test.

Study 2 Method Sample Altogether, 149 students from a German university took part in Study 2. However, five participants were excluded from the analyses because they were psychology majors beyond the first semester (even though we had informed participants from psychology beforehand that we only searched for freshmen). As in Study 1, the rationale for this was to prevent that study participants had previous training in psychometric testing. Furthermore, another participant was excluded because of insufficient knowledge of the German language. Therefore, the final sample for Study 2 consisted of 143 participants (Mage = 20.03 years, SD = 2.50; 55% females) from various fields of study. The largest groups were majoring in computer science (27.3%), psychology (17.5%), and biochemistry (11.9%). Of the participants, 60% had already applied for a job in the past and 23% had completed selection tests before. Additionally, 18% currently held a job. Ó 2019 Hogrefe Publishing

5

Procedure The procedures were identical to Study 1 with a few exceptions, the first of which was that the justification was adjusted to the GMA test that we used. Thus, test takers in the explanation group were informed (a) that the test is an established procedure to measure cognitive abilities such as comprehension, reasoning, and problem solving, (b) that these abilities are important to quickly acquire new job knowledge and to solve new tasks and problems on the job in a swift and flexible way, and (c) that performance on this test is predictive for one’s ability for such tasks. This justification was inserted before the general test instruction for the GMA test. It was printed in a frame and in boldface to increase its salience.

In contrast to the explanation group, participants in the control group were only presented with the standard test instruction but not the justification concerning content, job relatedness, and predictiveness of the test.

Again, all participants had to complete a questionnaire concerning their test perceptions once they had completed the test. Furthermore, we also included a manipulation check after the items related to test takers’ perceptions.

This manipulation check asked participants to indicate whether they had received an explanation why the test they had just completed can be used for selection purposes and participants had to answer this question with yes or no.

Finally, we also asked for participants’ final high school grade point average (GPA).

Ability Test GMA was assessed with the 10-minute test by Musch et al.

(2011). This test is a short test that assesses GMA within 10 min. It is comprised of 32 items that measure both fluid (e.g., numerical sequences) and crystallized cognitive ability (e.g., “Which of the following fractions is the smallest? 1) 2/4; 2) 2/5; 3) 1/3; 4) 3/8”). Thus, similar to the Wonderlic Personnel Test (Wonderlic Inc., 2002), the items are taken from different ability domains with the aim of yielding an overall score for individuals’ GMA. In line with this, the 10-minute test proved to be highly g-saturated in studies

by Ostapczuk, Musch, and Lieberei (2011) and Ostapczuk, Wagner, and Musch (2014). Its split-half reliability (using an odd–even split) was .76 in the present sample.

Test Takers’ Perceptions We used the same scales as in Study 1 to assess test takers’ perceptions. The internal consistencies were .78 for perceived face validity and for perceived predictive validity, .65 for overall procedural fairness, and .73 for test motivation.

Results First, we again determined whether the experimental groups were comparable with regard to demographic and Journal of Personnel Psychology (2019), 18(1), 1–9

6

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

test performance in the GMA test, and had a better GPA (in Germany, lower grades represent better performance).

Furthermore, to our initial surprise, the correlation between GMA test scores and GPA was not significant, r = .11.

However, after controlling for age and mother tongue, this correlation increased and turned significant, r = .22, p < .05.

To test whether the explanation improved test takers’ perceptions, we conducted separate analyses for participants who answered the manipulation check correctly and those who provided an incorrect answer to it. The rationale for this was that it seems unlikely that the experimental groups differ on the basis of the explanation if participants did not pay attention to this explanation. Means and SDs for the dependent variables and effect sizes from statistical tests of differences between the experimental groups are presented in Table 4.

In line with our hypothesis, means for test takers’ perceptions were more positive in the explanation group when we considered data for participants with a correct answer to the manipulation check. Furthermore, independentsamples t-tests also confirmed that the ratings were significantly higher for face validity, t(87) = 2.01, p < .05, Cohen’s d = .46, and for predictive validity, t(87) = 2.48, p < .05,

background variables and also whether participants had correctly answered the manipulation check. Similar to Study 1, we did not find any significant differences with regard to age, gender, mother tongue, number of previous job applications, or high school GPA, all ts < 1.12, all ps > .27. However, participants in the control group (M = 0.30, SD = 0.46) had more experience with selection tests than in the explanation group (M = 0.11, SD = 0.32), t(124.62) = 2.76, p < .05. Furthermore, it turned out that 50 of the 143 participants answered the manipulation check incorrectly. These participants were not distributed equally among the two groups. Instead, the majority of them (i.e., 33) were from the condition that had received an explanation (initial n = 70) whereas only 17 were from the control condition (initial n = 69; cf. Table 4). Four additional participants (two from each group) did not answer the manipulation check at all and were therefore excluded from further analyses.

Table 3 shows descriptive information and correlations for all study variables and for the manipulation check. As can be seen, there were significant correlations between gender on the one hand and test motivation, r = .19, test performance, r = .20, and high school GPA, r = .33, all ps < .05. Thus, women were less motivated, showed lower

Table 3. Descriptive statistics and intercorrelations of study variables in Study 2 Variables

M

(SD)

20.03

(2.50)

2. Gendera

0.55

(0.50)

3. Face validity

2.98

(0.75)

.05

.09

4. Predictive validity

2.37

(0.74)

.22**

.09

5. Procedural fairness

2.99

(0.75)

.10

.05

.23**

.35**

6. Test motivation

3.97

(0.74)

.05

.19*

.01

.15

.09

7. Test performance

19.79

(4.26)

.12

.20*

.02

.05

.08

.42**

8. High school GPA

2.00

(0.63)

.15

.33**

.03

.10

.07

.09

9. Manipulation checkb

0.64

(0.48)

.02

.07

.03

.11

.03

.01

1. Age

1.

2.

3.

4.

5.

6.

7.

8.

.02 .40**

a

.11 .04

.02

b

Note. N = 143. GPA = grade point average. Smaller GPA values indicate better grades. Gender was coded as 0 = male, 1 = female. Manipulation check was coded as 0 = incorrect, 1 = correct. *p < .05, **p < .01.

Table 4. Means and standard deviations for the two experimental groups in Study 2 Incorrect manipulation check No explanation (n = 17) Variables

M

Correct manipulation check

Explanation (n = 33)

(SD)

M

(SD)

No explanation (n = 52) Effect size

M

Explanation (n = 37)

(SD)

M

(SD)

Effect size

Face validity

2.96

0.66

2.95

0.74

.01

2.86

0.67

3.21

0.89

.46*

Predictive validity

2.27

0.84

2.25

0.79

.02

2.27

0.66

2.64

0.74

.53*

Procedural fairness

2.96

0.83

2.97

0.73

.01

2.96

0.74

3.08

0.71

.17

Test motivation

3.55

0.74

4.19

0.70

.90**

4.01

0.72

3.88

0.74

.18

19.00

3.94

20.61

4.18

.39

21.08

4.06

17.81

4.20

.79**

Test performance

Note. Effect sizes depict Cohen’s d. Positive d values indicate that the experimental group that had received an explanation had higher means than the no-explanation control group. Asterisks represent significance levels from independent-samples t-tests. *p < .05, **p < .01.

Journal of Personnel Psychology (2019), 18(1), 1–9

Ó 2019 Hogrefe Publishing

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

Cohen’s d = .53, for the group that had received a justification. However, for procedural fairness, the means did not differ significantly, t < 1. As in Study 1, the difference for participants’ test motivation was not significant, t < 1.

Finally, however, participants who had received an explanation obtained significantly lower scores in the GMA test, t(87) = 3.69, p < .01, Cohen’s d = .79 In contrast to participants who answered the manipulation check correctly, test takers’ perceptions between the two groups did not differ for participants who failed the manipulation check, all ts < 1. To our surprise, however, the group with the explanation reported higher test motivation, t(48) = 3.00, p < .01, Cohen’s d = .90.

Finally, given that participants in the control group had significantly more experience with selection tests than the explanation group, we also repeated the comparisons between the two groups and used experience with selection tests as a covariate. These additional analyses led to the same pattern of results. Furthermore, given the unexpected difference for performance in the GMA test among those participants who answered the manipulation check correctly, we also repeated the analyses with mother tongue as a covariate. However, this also did not change the general pattern of results and did not reduce the obtained difference.

Discussion The most important result was that explanations were also effective in improving two of the three test taker perception variables related to a GMA test – at least among those participants who answered the manipulation check correctly.

However, there were two surprising results in addition to the high percentage of participants who did not read our instructions carefully. The first were the relatively low values for the test taker perception variables. These values seemed comparable to those from Study 1 even though we had not intended to choose a test that provokes negative reactions in the first place but a test that is comparable to other common tests like the Wonderlic (Wonderlic Inc., 2002). The second and even more surprising effect was that the explanation group obtained significantly lower test scores after the explanation. Thus, given that an explanation is unlikely to alter participants’ true ability level, it seems as if the explanation impaired participants in performing at their true level. It might have been that the explanation increased the personal relevance of the test results and even though our study did not represent a high-stakes situation participants might have experienced higher test taking anxiety after the explanation of the test content and its relevance. Unfortunately, we did not measure anxiety in the present study to test this suggestion.

Ó 2019 Hogrefe Publishing

7

However, in line with this suggestion, Hausknecht et al.

(2004) found a substantial negative relationship between anxiety and performance in selection procedures.

General Discussion In contrast to meta-analytic findings by Truxillo et al.

(2009) and to the corresponding primary studies, the present results clearly suggest that providing an explanation can have beneficial effects on test takers’ perceptions of ability tests: participants who received a justification before the completion of the test had more positive perceptions with regard to face validity, predictive validity, and – at least for the attention test – also with regard to procedural fairness. However, in Study 2, these enhanced perceptions were only found for participants who were able to answer the manipulation check correctly, which confirmed that they had indeed paid attention to the explanation.

We suggest that the main reason why the present study found positive effects of an explanation in contrast to earlier research is that we used a justification that focused on the content of the test, explained the job relevance of the targeted constructs, and stressed the predictiveness of the test. Furthermore, our results suggest that this kind of explanation is relevant not only when tests seem rather abstract (as in Study 1) but also when tests with common GMA items (as in Study 2) are used. In both cases, it seems beneficial to provide test takers with an explanation of what the test is measuring and why it is related to work outcomes.

In contrast to Truxillo et al.’s (2009) meta-analytic results, the explanation in our studies did not increase test motivation, and in Study 2, it even impaired test performance. Concerning test motivation, it might well be that the explanation did not make much of a difference because of the relatively high motivation in the first place. Thus, the null effect for test motivation might be due to a ceiling effect. However, if the negative effect on test performance in the GMA test replicates in future research and if it is really due to increased anxiety as suggested above, then it should be explored how the relevance of the test content can be explained in an alternative way that provokes less anxiety. Alternatively, the explanation could be supplemented with an uncertainty reduction explanation. In a recent study by McCarthy, Bauer, Truxillo, Campion, et al. (2017), such an explanation was combined with an explanation about test content and the testing process as well as with a social fairness explanation that stressed respect for test takers who had to complete a work sample test afterward. It was found that this extended explanation led to significantly improved fairness perceptions.

Journal of Personnel Psychology (2019), 18(1), 1–9

8

Limitations and Lines for Future Research A possible objection concerning the beneficial effects on test takers’ perceptions might be related to the absolute level of the perception variables that was observed in the experimental groups: Even though the explanation improved these perceptions significantly and even though most effect sizes were in the moderate to large range, the means for the different perceptions were still close to the actual midpoint of the scale. However, this is comparable with other primary studies (e.g., Horvath et al., 2000; Rolland & Steiner, 2007).

Another limitation that concerns the generalizability of our results is our use of student samples. Previous research suggests that effect sizes in applicant reactions research differ between field studies with actual applicants on the one hand and simulated selection processes with student participants on the other hand. Future research is therefore needed to evaluate the actual effects of providing explanations to test takers’ in high-stakes selection settings. However, it seems possible that the present studies provide more conservative estimates of the actual effects of explanations in comparison with field studies with actual applicants, because additional analyses by Truxillo et al.

(2009) revealed that explanations have larger effects in field studies. Furthermore, this parallels meta-analytic evidence from Hausknecht et al. (2004) who also found that applicants’ perceptions were more strongly related to many outcome measures in authentic selection contexts than in hypothetical or simulated ones. Thus, test takers in real high-stakes selection contexts for whom the test and the test outcome are more relevant than for study participants in selection simulations might be more receptive to the provision (or to the lack) of explanations.

In addition, it might also be that the high percentage of participants who did not carefully read the instructions and the explanation does not generalize to high-stakes settings where applicants are probably more anxious to not miss relevant pieces of information. However, given that it is not uncommon in laboratory research (Oppenheimer, Meyvis, & Davidenko, 2009) that substantial numbers of participants fail to read and follow the instructions, it might be that this carelessness also contributed to smaller effects of explanations in earlier simulation studies. Again, future research is needed to evaluate this suggestion.

Practical Implications Taken together, our results suggest that test administrators should explain to test takers what a given ability test is measuring and why it is important in the context of a selection process. Our hope is that providing such explanations will help to improve perceptions of ability tests in the long Journal of Personnel Psychology (2019), 18(1), 1–9

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

run. However, to avoid negative side effects on test performance, additional steps might be beneficial. Furthermore, test administrators should also try to ensure that applicants do indeed perceive the explanation.

References Anderson, N., Salgado, J. F., & Hülsheger, U. R. (2010). Applicant reactions in selection: Comprehensive meta-analysis into reaction generalization versus situational specificity. International Journal of Selection and Assessment, 18, 291–304. https://doi.org/10.1111/j.1468-2389.2010.00512.x Arvey, R. D., Strickland, W., Drauden, G., & Martin, C. (1990).

Motivational components of test taking. Personnel Psychology, 43, 695–716. https://doi.org/10.1111/j.1744-6570.1990. tb00679.x Bauer, T. N., Truxillo, D. M., Sanchez, R. J., Craig, J. M., Ferrara, P., & Campion, M. A. (2001). Applicant reactions to selection: Development of the selection procedural justice scale (SPJS).

Personnel Psychology, 54, 388–420. https://doi.org/10.1111/ j.1744-6570.2001.tb00097.x Bertua, C., Anderson, N., & Salgado, J. F. (2005). The predictive validity of cognitive ability tests: A UK meta-analysis. Journal of Occupational and Organizational Psychology, 78, 387–409. https://doi.org/10.1348/096317905X26994 Brickenkamp, R., Schmidt-Atzert, L., & Liepmann, D. (2010). d2-R: Aufmerksamkeits-und Konzentrationstest [d2-R: Attention and concentration test]. Göttingen, Germany: Hogrefe.

Gilliland, S. W. (1993). The perceived fairness of selection systems: An organizational justice perspective. Academy of Management Review, 18, 694–734. https://doi.org/10.2307/258595 Gilliland, S. W., & Steiner, D. D. (2012). Applicant reactions to testing and selection. In N. Schmitt (Ed.), The Oxford handbook of personnel assessment and selection (pp. 629–666). New York, NY: Oxford University Press.

Harold, C. M., Holtz, B. C., Griepentrog, B. K., Brewer, L. M., & Marsh, S. M. (2016). Investigating the effects of applicant justice perceptions on job offer acceptance. Personnel Psychology, 69, 199–227. https://doi.org/10.1111/peps.12101 Hausknecht, J. P., Day, D. V., & Thomas, S. C. (2004). Applicant reactions to selection procedures: An updated model and meta-analysis. Personnel Psychology, 57, 639–683. https://doi. org/10.1111/j.1744-6570.2004.00003.x Holtz, B. C., Ployhart, R. E., & Dominguez, A. (2005). Testing the rules of justice: The effects of frame-of-reference and pre-test validity information on personality test responses and test perceptions. International Journal of Selection and Assessment, 13, 75–86. https://doi.org/10.1111/j.0965-075X.2005.00301.x Horvath, M., Ryan, A. M., & Stierwalt, S. L. (2000). The influence of explanations for selection test use, outcome favorability, and self-efficacy on test-taker perceptions. Organizational Behavior and Human Decision Processes, 83, 310–330. https://doi.org/ 10.1006/obhd.2000.2911 Hunter, J. E., & Hunter, R. F. (1984). Validity and utility of alternative predictors of job performance. Psychological Bulletin, 96, 72–98. https://doi.org/10.1037/0033-2909.96.1.72 Kersting, M. (2008). Zur Akzeptanz von Intelligenz- und Leistungstests [The acceptance of intelligence and achievement tests].

Report Psychologie, 33, 420–433.

Konradt, U., Garbers, Y., Böge, M., Erdogan, B., & Bauer, T. N. (2017). Antecedents and consequences of fairness perceptions in personnel selection. Group & Organization Management, 42, 113–146. https://doi.org/10.1177/1059601115617665 Ó 2019 Hogrefe Publishing

K. G. Melchers & B. Körner, Explanations and Fairness Perceptions

LaHuis, D. M., Perreault, N. E., & Ferguson, M. W. (2003). The effect of legitimizing explanations on applicants’ perceptions of selection assessment fairness. Journal of Applied Social Psychology, 33, 2198–2215. https://doi.org/10.1111/j.15591816.2003.tb01881.x Lievens, F., De Corte, W., & Brysse, K. (2003). Applicant perceptions of selection procedures: The role of selection information, belief in tests, and comparative anxiety. International Journal of Selection and Assessment, 11, 67–77. https://doi.org/10.1111/ 1468-2389.00227 McCarthy, J. M., Bauer, T. N., Truxillo, D. M., Anderson, N. R., Costa, A. C., & Ahmed, S. M. (2017). Applicant perspectives during selection: A review addressing “So what?”, “What’s new?”, and “Where to next?”. Journal of Management, 43, 1693–1725. https://doi.org/10.1177/0149206316681846 McCarthy, J. M., Bauer, T. N., Truxillo, D. M., Campion, M. C., Van Iddekinge, C. H., & Campion, M. A. (2017). Using pre-test explanations to improve test-taker reactions: Testing a set of “wise” interventions. Organizational Behavior and Human Decision Processes, 141, 43–56. https://doi.org/10.1016/j.obhdp.

2017.04.002 Musch, J., Ostapczuk, M., Hilbig, B. E., Auer, T. S., Brandt, M., Cüpper, L., . . . Endfelder, E. (2011). 10-Minuten-Test [10minute test] (Unpublished test). University of Düsseldorf, Germany Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009). Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of Experimental Social Psychology, 45, 867–872. https://doi.org/10.1016/j.jesp.2009.03.009 Ostapczuk, M., Musch, J., & Lieberei, W. (2011). Der “Analytische Test”: Validierung eines neuen eignungsdiagnostischen Instruments zur Erfassung von schlussfolgerndem Denken [The Analytical Test: Validation of a new personnel selection tool assessing reasoning]. Zeitschrift für Arbeits- und Organisationspsychologie, 55, 1–16. https://doi.org/10.1026/0932-4089/ a000031 Ostapczuk, M., Wagner, M., & Musch, J. (2014). Revisiting the Rybakov figures: A classical and probabilistic analysis of the psychometric properties of a test of spatial visualization. Swiss Journal of Psychology, 73, 57–67. https://doi.org/10.1024/ 1421-0185/a000125 Rolland, F., & Steiner, D. D. (2007). Test-taker reactions to the selection process: Effects of outcome favorability, explanations, and voice on fairness perceptions. Journal of Applied Social Psychology, 37, 2800–2826. https://doi.org/10.1111/ j.1559-1816.2007.00282.x

Ó 2019 Hogrefe Publishing

9

Schmidt, F. L., & Hunter, J. (2004). General mental ability in the world of work: Occupational attainment and job performance.

Journal of Personality and Social Psychology, 86, 162–173. https://doi.org/10.1037/0022-3514.86.1.162 Smither, J. W., Reilly, R. R., Millsap, R. E., Pearlman, K., & Stoffey, R. W. (1993). Applicant reactions to selection procedures.

Personnel Psychology, 46, 49–76. https://doi.org/10.1111/ j.1744-6570.1993.tb00867.x Truxillo, D. M., & Bauer, T. N. (2011). Applicant reactions to organizations and selection systems. In S. Zedeck (Ed.), APA handbook of industrial and organizational psychology (Vol. 2, pp. 379–397). Washington, DC: American Psychological Association.

Truxillo, D. M., Bodner, T. E., Bertolino, M., Bauer, T. N., & Yonce, C. A. (2009). Effects of explanations on applicant reactions: A meta-analytic review. International Journal of Selection and Assessment, 17, 346–361. https://doi.org/10.1111/j.14682389.2009.00478.x Wonderlic Inc. (2002). Wonderlic personnel test and scholastic level exam: User’s manual. Libertyville, IL: Author.

History Received March 18, 2017 Revision received February 28, 2018 Accepted March 12, 2018 Published online January 9, 2019 Acknowledgments We thank Kim Bentlage, Lisa Leusch, Julia Mies, Linda Miller, Anja Roth, and Ebru Sezer for their help with data collection.

ORCID Klaus G. Melchers https://orcid.org/0000-0003-4211-6450 Klaus G. Melchers Work and Organizational Psychology Institute of Psychology and Education Ulm University Albert-Einstein-Allee 41 89069 Ulm Germany klaus.melchers@uni-ulm.de

Journal of Personnel Psychology (2019), 18(1), 1–9

Original Article

Not All Leaving Is Created Equal Differentiating the Factors of Organizational and Occupational Turnover Intentions Huisi (Jessica) Li1, Kun Yu2 , Youhuang Huang2, and Xiaotong Jin3 1

Johnson Graduate School of Management, Cornell University, Ithaca, NY, USA

2

School of Labor and Human Resources, Renmin University of China, Beijing, China

3

Business School, Jilin University, Changchun, China

Abstract: Drawing on social cognitive career theory (SCCT) and the focus–congruence approach, this study examines how organizational and occupational turnover is differently influenced by work-related factors. Using a sample of 2,353 teachers in rural China, results first showed that negative relationships between organization-focused predictors (i.e., leader support, colleague support, and physical resources for work) and organizational turnover intentions were significant. Similarly, the negative relationship between occupation-focused predictors (i.e., occupational satisfaction, salary satisfaction, and occupational choice motivations) and occupational turnover intentions was also significant.

Moreover, occupation-focused predictors have a stronger relationship with occupational turnover intentions than organizational turnover intentions, and vice versa. Implications for research and practice are discussed.

Keywords: turnover intentions, social cognitive career theory, focus–congruence approach, organizational turnover, occupational turnover

Employee turnover has captured researchers’ attention since the beginning of the 20th century (Hom, Mitchell, Lee, & Griffeth, 2012), as it costs organizations heavily in areas such as financial operations (Hom et al., 2012).

Employees could make two decisions regarding turnover: to leave the organization (organizational turnover; Hom, 2011) or to leave the occupation (occupational turnover; Fimian, Fastenau, & Thomas, 1988; Lane, Mathews, & Presholdt, 1988). However, scholarship on turnover has largely limited its focus to organizational turnover, leaving unanswered questions about occupational turnover (Cotton & Tuttle, 1986). This is a critical void as occupational turnover is correlated with employee retention regarding both organizational and occupational matters (Yousaf, Sanders, & Abbas, 2015).

Moreover, leaving an organization and switching an occupation are considered related but separated processes (Blau, 1985). In general, occupational turnover is much more difficult for individuals than organizational turnover and occurs much less often (Blau, 2007). For example, drawn from US data, turnover patterns of teachers show that every year 7.4% of teachers move to a different school (between 1987 and 2000) but over a lifetime only 7% of teachers move to another occupation (till 2003; Ingersoll, 2003). Treating organizational and occupational turnover as synonymous ignores the potentially different etiologies of these phenomena, because they may have different Journal of Personnel Psychology (2019), 18(1), 10–22 https://doi.org/10.1027/1866-5888/a000216

individual and/or situational antecedents. As a result, the differentiated relationships of various antecedents with organizational and occupational turnover are unknown.

This represents a critical gap in our knowledge. Studying destinations of leaving (different organizations or new occupations) informs researchers about the relevant pre-turnover work attitudes and decision processes contributing to each type of movement (Hom et al., 2012). Thus, providing empirical evidence to substantiate the distinctiveness of organizational and occupational turnover in their nomological networks is much needed.

In addition, although few studies have found that organizational turnover was negatively associated with organizational commitment and that occupational turnover was negatively associated with occupational commitment (Chang, Chi, & Miao, 2007; Yousaf et al., 2015), they mainly treated organizational turnover and occupational turnover as theoretically separated systems and tested it in separated analyses. However, it is also possible that both organizational and occupational turnover are under the influence of the same set of individual and/or situational antecedents but differ in degree of influence. To our knowledge, research considering the quantity difference of antecedents of organizational and occupational turnover is absent from the literature. Thus, to better uncover the relationship between organizational and occupational turnover, drawing from an alternative lens and substantiating Ó 2019 Hogrefe Publishing

H. Li et al., Organizational and Occupational Turnover

the relative importance of individual and/or situational antecedents are also much needed.

The present study aims to contribute to the understanding of the basis of employees’ occupational turnover intentions as distinguished from organizational turnover intentions, as well as to the understanding of the relationship between organizational and occupational turnover.

Results of the current investigation could provide the evidence of the distinctiveness between organizational and occupational turnover as well as the resemblance between them, especially regarding work-related antecedents. The sample we used in the current study is from a representative survey of 2,353 primary and junior secondary school teachers in the Gansu Province in the northwestern interior of China. Between 2000 and 2013, the supply of full-time primary school teachers in rural China dropped by 44% (Ministry of Education of the People’s Republic of China, 2000, 2014). Interestingly, two related but distinct turnover problems simultaneously contribute to the teacher shortage in rural China: teacher migration and teacher attrition (Ministry of Education of the People’s Republic of China, 2008). Teacher migration occurs when teachers move from rural schools to other places. In contrast, teacher attrition occurs when teachers leave the teaching profession altogether to pursue other lines of work (Johnson, Berg, & Donaldson, 2005), which can be either other professions in the same school (e.g., staff) or jobs outside (e.g., salesperson). More formally, teacher migration fits the traditional definition of organizational turnover (i.e., people leaving organizations, regardless of whether they stay or leave their occupations), whereas teacher attrition is consistent with definitions of occupational turnover (i.e., people leaving their occupations, regardless of whether they stay or leave their organizations). Thus, we believe that this large sample of teachers is particularly relevant and appropriate for the study of the relationship between organizational and occupational turnover.

Theoretical Background and Hypotheses To understand the relationship between various factors and either organizational or occupational turnover intentions, the focus–congruence approach is of particular relevance (Klein, Molloy, & Brinsfield, 2012). The focus–congruence approach suggests that predictors and criteria are more strongly related when they are expressed and measured at the same level of specification and with the same focus (Klein et al., 2012; Smith, 1976). For example, Blau (1985) found a significant negative correlation between organizational commitment and organizational turnover Ó 2019 Hogrefe Publishing

11

intentions – but not between organizational commitment and occupational turnover intentions. Alternatively, he found a significant negative correlation between occupational commitment and occupational turnover intentions but not between occupational commitment and organizational turnover intentions. Thus, relationships between congruent foci should be stronger than when foci differ (Klein et al., 2012).

Identifying Antecedents Social cognitive career theory (SCCT; Lent, Brown, & Hackett, 2002) predicts that individuals’ career choices, including occupational and organizational choices, are influenced by both situational and individual factors. Since leaders and peers are two critical situational factors that influence employees’ salience of self-concept orientations (Lord & Brown, 2004), we first propose support from the leader and colleagues as antecedents of turnover intention.

Considering that employees within an organization have beliefs on both specific working conditions and the general job and career (James, Hater, Gent, & Bruni, 1978; James & Tetrick, 1986), we also propose two factors regarding employees’ perceptions of and attitudes towards working conditions (i.e., salary satisfaction and working environment support), and two factors regarding employees’ general attitudes toward their career (i.e., occupational satisfaction and career motivation) as antecedents of turnover intentions.

Because the purpose of this study is not to identify new predictors of organizational or occupational turnover intentions, but rather to differentiate the most predictive and impactful antecedents of each type of turnover intentions, the predictors in this study have appeared in previous research and meta-analyses (e.g., Allen, Bryant, & Vardaman, 2010; Cotton & Tuttle, 1986; Griffeth, Hom & Gaertner, 2000; Maertz & Griffeth, 2004; Woo & Maertz, 2012). Based on the focus–congruence approach (Klein, et al, 2012), we propose that factors with an organizational focus (e.g., leader support, colleague support, and working environment) would be more strongly related to organizational than occupational turnover intentions, while those with an occupational focus (e.g., occupational satisfaction, salary satisfaction, and intrinsic motivation for choosing an occupation) would be more influential for occupational turnover intentions than for organizational turnover intentions.

Wanting to Leave the Organization Leader Support Employees consider their leaders as organizational representatives. According to social exchange theory, when leaders give their followers much support, the followers feel Journal of Personnel Psychology (2019), 18(1), 10–22

12

H. Li et al., Organizational and Occupational Turnover

indebted to the leaders and the organization, and thus are more likely to be engaged in tasks and committed to the organization (Gerstner & Day, 1997; Liden, Sparrowe, & Wayne, 1997). Leader support should be considered a work perception with an organizational focus rather than an occupational focus, and thus, we propose that it has a stronger relationship with organizational turnover intentions than with occupational turnover intentions.

Colleague Support Colleagues comprise the immediate context in which an employee works. Employees who are embedded in a social web at work develop a disinclination to leave (Lee, 2004).

Similar to leader support, daily shared work experiences with colleagues do not generalize to the entire occupation.

Researchers (Kerr & Slocum, 1987; Kopelman, Brief, & Guzzo, 1990) have argued that variation in employee retention across organizations is related to the organizational culture which is organization-specific and varies among companies in the same industry (Brightman & Sayeed, 1990; Chatman, 1991; Rentsch, 1990; Sheridan, 1992).

Based on this reasoning, colleague support should be considered to have an organizational focus rather than an occupational focus, and thus, we propose that it would have a stronger relationship with organizational – rather than occupational – turnover intentions.

Physical Resources for Work Besides psychological and social support provided by leaders and peers as a form of job resources, organizations are also expected to provide the physical resources needed for an employee to perform work-related tasks (Bakker & Demerouti, 2007); indeed, it is almost taken for granted that organizations provide employees with adequate physical resources such as work materials (Mintzberg, 1979). In order to establish stable patterns of employee behavior, organizations should design the physical environment in order to coordinate and divide labor (Mintzberg, 1979).

For most occupations, the levels of physical resources from the organization vary from one employer to another, which is particularly evident in this study.

Based on the above arguments, we propose the following hypotheses: Hypothesis 1: (a) Leader support (b) colleague support, and (c) physical resources for work are negatively associated with organizational turnover intentions.

Hypothesis 2: Organization-focused factors (i.e., leader support, colleague support, and physical resources for work) have a stronger relationship with organizational turnover intentions than with occupational turnover intentions.

Journal of Personnel Psychology (2019), 18(1), 10–22

Wanting to Leave the Occupation Occupational Satisfaction Job satisfaction has been studied extensively as a predictor of turnover (Mobley, Griffeth, Hand, & Meglino, 1979).

However, the majority of researchers have only studied overall job satisfaction (Chapman & Lowther, 1982; Ingersoll, 2001), leaving out consideration of one’s satisfaction toward one’s occupation (for an exception, see Ng & Feldman, 2007). In this study, occupational satisfaction was examined and proposed to be more closely related to an employee’s occupational turnover intentions than to organizational turnover intentions.

Salary Satisfaction Whether salary satisfaction is an organization-specific or an occupation-wide factor depends on the dispersion of pay for the same type of job across different organizations and occupations. In this study, we expect salary satisfaction of teachers to have an occupational focus and thus be more closely related to an employee’s occupational turnover intentions than to organizational turnover intentions.

Intrinsic Occupational Choice Motivation Intrinsic motivation involves engaging in work for its own sake because the work itself is interesting or satisfying (Deci & Ryan, 1985). Since whether work is interesting or satisfying is mainly determined by the occupation rather than the organization, it should be expected that employees’ occupational turnover intentions should be better predicted by intrinsic occupational motivation than organizational turnover intentions.

Based on the above arguments, we propose the following hypotheses: Hypothesis 3: (a) Occupational satisfaction, (b) salary satisfaction, and (c) intrinsic occupational choice motivation are negatively associated with occupational turnover intentions.

Hypothesis 4: Occupation-focused factors (i.e., occupational satisfaction, salary satisfaction, and intrinsic occupational choice motivation) have a stronger relationship with occupational turnover intentions than with organizational turnover intentions.

Method Sample and Procedures The data in this study were collected in rural areas of northwestern China in 2007 as part of a large and publicly available dataset called the Gansu Survey of Children and Ó 2019 Hogrefe Publishing

H. Li et al., Organizational and Occupational Turnover

Families, a rigorous multilevel and longitudinal study of rural children’s welfare (more information can be found at http://china.pop.upenn.edu.). A total of 2,353 primary or junior secondary school teachers from 198 schools comprised the sample of this study, with 11.88 (SD = 10.97) teachers per school on average. Among the 2,353 teachers, 60.4% were male, 83.1% were married, 24% had only finished their secondary education (with 76% had finished their tertiary-level education), 51.3% were from the same township, and 41.2% were teaching in primary schools versus secondary schools. The average age was 36.72 years.

They had been teaching for 14.41 years and worked in the same school for 8.30 years on average. All surveys were administrated in person by trained research assistants. All items were developed in Chinese by the research team based on past research, as well as careful piloting and discussion with local teachers and principals to ensure that the items were suitable for this setting.

Measures Organizational and Occupational Turnover Intentions Two items measuring turnover intentions were used in the study. They were “I want to move to a different school” (organizational turnover intentions) and “I want to change my occupation” (occupational turnover intentions). This method was similar to that used in previous research on turnover intentions (Krausz, Koslowsky, Shalom, & Elyakim, 1995). These two constructs were measured on a 3-point Likert scale (1 = completely disagree, 2 = not sure, and 3 = completely agree).

Leader Support Four items measured leader support. They were “My principal has high expectations for me,” “My principal respects me very much,” “My principal offers me opportunities for self-development,” and “My principal offers good suggestions on my teaching.” This construct was measured on a 5-point Likert scale (1 = completely disagree, 2 = somewhat disagree, 3 = not sure, 4 = somewhat agree, and 5 = completely agree). Cronbach’s α = .78.

Colleague Support Four items measured colleague support. They were “I have a lot of opportunities to discuss teaching with my colleagues,” “The activities organized by the teaching section/ department are valuable,” “I get along well with my colleagues in the school,” and “The teachers in my school are highly motivated to work.” This construct was measured on a 3-point Likert scale (1 = completely disagree, 2 = not sure, and 3 = completely agree). Cronbach’s α = .63.

Ó 2019 Hogrefe Publishing

13

Physical Resources for Work Two items measured physical resources for work. They were “The school has adequate teaching materials and equipment” and “The school has adequate books and journals that I can refer to for preparation of my classes.” This construct was measured on a 3-point Likert scale (1 = completely disagree, 2 = not sure, and 3 = completely agree).

Cronbach’s α = .82.

Occupational Satisfaction One item, “I am satisfied with my job as a teacher,” measured occupational satisfaction on a 3-point Likert scale (1 = completely disagree, 2 = not sure, and 3 = completely agree).

Salary Satisfaction One item, “I am satisfied with my salary,” measured this construct on a 3-point Likert scale (1 = completely disagree, 2 = not sure, and 3 = completely agree).

Intrinsic Occupational Choice Motivation Intrinsic occupational motivation was measured by three items: “I chose to be a teacher because I love being together with students,” “I chose to be a teacher because I believe that education is very important for the development of the country,” and “I chose to be a teacher because I had always wanted to be a teacher since I was little.”

These constructs were measured on a 3-point Likert scale (1 = completely disagree, 2 = not sure, and 3 = completely agree). Cronbach’s α was .60.

Control Variables Teachers’ age, sex (0 = male and 1 = female), education attainment level (0 = secondary level and 1 = tertiary level), marital status (0 = married, 1 = unmarried), tenure in the teaching profession and in the current school (in years), level of school taught (0 = primary and 1 = secondary), and whether they were working in their original hometowns (0 = no and 1 = yes) were controlled in the present study.

We included these variables because, similar to teachers in other countries (Lachman & Diamant, 1987), younger, unmarried male teachers with greater human capital were found to be less satisfied with positions as teachers in rural China, while teachers who are more socially similar to the local community were found to be more satisfied (Sargent & Hannum, 2005). Moreover, it is also possible that, like younger teachers, the older teachers are also more likely to leave their jobs than middle-aged teachers. Older teachers may want to change positions or even professions because their accumulated expertise let them easier to get better-paid positions whether in or out of teacher occupation in cities. As a result, a nonlinear and parabolic relationship between age and turnover intentions could occur.

Journal of Personnel Psychology (2019), 18(1), 10–22

14

Thus, we also controlled the age groups to rule out the possible nonlinear effect of age on turnover intentions. Based on Caldwell et al.’s (2004) classification of age, we computed two dummy-coded age group variables: “age (1),” with participants younger than 25 coded as 1 and others as 0, and “age (2),” with participants older than 55 coded as 1 and others as 0.

Measurement Issues Although all scales measured in the current study were developed based on past research and had been tested in pilot studies, they had not been validated in previous research. To address this issue, we collected a small validation sample of 74 employees from a trading company in China. In the validation data, each self-developed scale was measured and paired with a well-developed scale measuring the same theoretical construct. Well-established scales include leader support (Greenhaus, Parasuraman, & Wormley, 1990), colleague support (Caplan, Cobb, French, Van Harrison, & Pinneau, 1975), physical resources for work (Schneider, Parkington, & Buxton, 1980), occupational satisfaction (Greenhaus et al., 1990), salary satisfaction (Heneman & Schwab, 1985), intrinsic occupational motivation (Centers & Bugental, 1966), and occupational and organizational turnover intentions (Kelloway, Barling, & Shah, 1993). Among the 74 employees, 28 were female (37.8%), and 46 were male (62.2%). The average age was 26.12 years (SD = 4.11), and 97.3% had a college degree or above. The average tenure of these employees was 38.64 months.

The validation test showed that Cronbach’s α of the criterion scales (i.e., well-developed scales) ranged from .90 to .99, indicating that all established measures had good internal consistency. Moreover, Cronbach’s α of self-developed scales (used in the present study) in the validation data ranged from .84 to .97, which indicated very good reliability and provided supplemental evidence of reliability for these self-developed scales. Furthermore, every self-developed scale had a high correlation with the established measure – from .73 to .95 – indicating that these self-developed scales used in the current research have a good psychometric property.

Analysis Regarding the nested structure of the data (i.e., teachers nested within schools), physical resources for work was treated as a school-level variable and should be aggregated from individual ratings. Other study variables were all 0treated as individual-level variables. Thus, we calculated intra-class correlation coefficients (ICCs; Shrout & Fleiss, Journal of Personnel Psychology (2019), 18(1), 10–22

H. Li et al., Organizational and Occupational Turnover

1979) and rwg and rwg(j) testing for within-team agreement (James, Demaree, & Wolf, 1984) to make sure that physical resources for work has considerable between-group variance, and the other variables, on the contrary, have minimal between-group variance. To provide complete information on the justification for aggregation, ICCs and rwg/rwg(j) were reported for all study variables. For organizational turnover intentions, ICC1 (i.e., amount of variance explained by group membership) = 0.167, ICC2 (i.e., reliability of group means) = 0.706, and rwg = 0.296; for occupational turnover intentions, ICC1 = 0.112, ICC2 = 0.602, and rwg = 0.327; for leader support, ICC1 = 0.132, ICC2 = 0.645, and rwg(j) = 0.838; for colleague support, ICC1 = 0.137, ICC2 = 0.655, and rwg(j) = 0.765; for physical resources for work, ICC1 = 0.186, ICC2 = 0.733, and rwg(j) = 0.379; for occupational satisfaction, ICC1 = 0.073, ICC2 = 0.486, and rwg = 0.556; for salary satisfaction, ICC1 = 0.097, ICC2 = 0.565, and rwg = 0.200; and for intrinsic occupational choice motivation, ICC1 = 0.120, ICC2 = 0.622, and rwg(j) = 0.689. According to LeBreton and Senter (2008), although rwg(j) of physical resources for work is relatively low and considered “weak agreement,” it has an ICC2 greater than .70 and ICC1 greater than .15. Since physical resources for work was treated as a school-level variable aggregated from individual ratings, these levels of agreement are sufficient to justify aggregation. Moreover, both turnover intentions and all antecedents except physical resources for work were treated as individual-level variables in the subsequent analysis. For that all study variables have ICC1s from .07 to .19, showing “medium” effects of group membership; we then used multilevel modeling with Mplus 7 that allows us to test our research hypothesis at both individual level (Level-1) and group level (Level-2) of analysis while controlling for group-level variance of individual-level predictors (Bliese, 2000).

Four sets of analyses were performed using Mplus 7.2 (Muthén and Muthén 1998–2012). We first tested the impact of each predictor on respective types of turnover intentions by conducting separated regressions with one respective predictor and control variables included in each regression. Next, we put all predictors and control variables in two multilevel regression models. We provided the model formula for organizational turnover intentions as follows: Level-1:

OrTIij ¼ β0j þ β1j LSij þ β2j CSij þ β3j Ageij þ β4j Age1ij þ β5j Age2ij þ β6j Sexij þ β7j Marij þ β8j Eduij þ β9j OrTenij þ β10j OcTenij þ β11j Townij þ β12j Gradij þ eij Ó 2019 Hogrefe Publishing

H. Li et al., Organizational and Occupational Turnover

Level-2:

β0j ¼ γ 00 þ γ 01 PR 0j þ u0j And the model formula for occupational turnover intentions is as follows: Level-1:

15

turnover intentions and occupational turnover intentions demonstrates that they should be considered as separate – although related – constructs. Most control variables significantly correlated with the two dependent variables, except for marital status. All independent variables were significantly correlated with both dependent variables in the predicted directions.

OcTIij ¼ β0j þ β1j OSij þ β2j SSij þ β3j IMij þ β4 Ageij þ β5 Age1ij þ β6 Age2ij þ β7 Sexij þ β8 Marij þ β9 Eduij þ β10 OrTenij þ β11 OcTenij þ β12 Townij þ β13 Gradij þ eij where OrTI = organizational turnover intentions; LS = leader support; CS = colleague support; PR = physical resources for work; OS = occupational satisfaction; SS = salary satisfaction; IM = intrinsic occupational choice motivation; Age1 = dummy variable indicating younger than 25; Age2 = dummy variable indicating older than 55; Mar = marital status; Edu = educational level; OrTen = organizational tenure; OcTen = occupational tenure; Town = from the same town; Grad = teaching grade. Then, we used the net regression method (Cohen, Cohen, West, & Aiken, 2003, p. 157 & p. 642) to calculate the difference between betas for different dependent variables to further support the focus–congruence hypotheses. Moreover, we used grand-mean centering for all of our analyses to facilitate the interpretation of the model results.

Results Preliminary Analysis To ensure the discriminant validity of the self-rated variables, a confirmatory factor analysis (CFA) was performed using Mplus 7.2 (Muthén and Muthén 1998–2012). The 6-factor model (i.e., all variables are independent of each other, except that two single-item turnover measures were merged as one factor and two single-item satisfaction measures were merged as another factor) provided a generally good fit to the data, with w2(104) = 544.14, p < .01, comparative fit index (CFI) = .95, Tucker-Lewis index (TLI) = .94, and root-mean-square error of approximation (RMSEA) = .04. According to the chi-square difference tests, the 6-factor model fit the data significantly better than the one-factor model (i.e., combine all variables for active divergence), w2(119) = 4,405.08, p < .01, CFI = .52, TLI = .46, and RMSEA = .12. Thus, the discriminant validity of the study variables was supported.

Descriptive statistics are presented in Table 1. The moderate correlation (r = .35, p < .01) between organizational

Ó 2019 Hogrefe Publishing

Hypotheses Test Hypothesis 1 predicts that leader support, colleague support, and physical resources for work are negatively associated with organizational turnover intentions. Similarly, Hypothesis 3 predicts that occupational satisfaction, salary satisfaction, and intrinsic occupational choice motivation are negatively associated with occupational turnover intentions. We ran a series of analyses to test the above two hypotheses. First, we ran six regression models with each model including one of the two types of turnover intentions and one respective predictor (and controls) incorporated.

The results revealed that, after controlling for sex, age (both linear and nonlinear), marital status, educational level, tenure in the teaching profession and in the current school, level of school taught, and whether teachers were working in their original hometowns, the association between organizational turnover intention and leader support (B = 0.44, p < .001), colleague support (B = 1.15, p < .001), and physical resources for work (B = 0.95, p < .001) was significant. Similarly, the relationship between occupational turnover intention and occupational satisfaction (B = 1.60, p < .001), salary satisfaction (B = 0.50, p < .001), and intrinsic occupational choice motivation (B = 1.60, p < .001) was also significant. Thus, Hypothesis 1 and Hypothesis 3 were both supported.

Next, as shown in Table 2, we simultaneously put all predictors into the models predicting organizational turnover intentions and occupational turnover intentions. In general, occupational turnover intentions were negatively related to occupational satisfaction (B = 1.31, p < .001), salary satisfaction (B = 0.30, p < .001), and intrinsic occupational choice motivation (B = 1.07, p < .001), while organizational turnover intentions were negatively related to leader support (B = 0.40, p < .001), colleague support (B = 0.74, p < .001), and physical resources for work (B = 0.69, p < .001). This approach provided additional support of Hypothesis 1 and Hypothesis 3.

Furthermore, we tested whether organization-focused factors (i.e., leader support, colleague support, and physical resources for work) have a stronger relationship with organizational turnover intentions than with occupational turnover intentions (Hypothesis 2), and whether occupation-focused factors (i.e., occupational satisfaction,

Journal of Personnel Psychology (2019), 18(1), 10–22

Journal of Personnel Psychology (2019), 18(1), 10–22 –

0.59 3.65 2.62 1.76 2.63 1.85 2.51 1.73 1.86

10. Teaching grade

11. Leader support

12. Colleague support

13. Physical resources

14. Occupational satisfaction

15. Salary satisfaction

16. Intrinsic motivation

17. Occupational turnover intentions

18. Organizational turnover intentions

.06**

–

5

–

.04

.08**

.04*

.00

.10**

.06**

0.80 .04*

.12**

.27** .04* .16** .01

.09**

.01

.04

7

– .25**

.60**

.09**

.09**

.02

– .17**

9

–

.12**

.27**

.04*

.04

.09**

.17**

.13**

–

.04

.13**

.02

10

.10** .11** .12** .08**

.14**

.09**

.08**

.18**

.32**

(.78)

11

.33**

.21**

.28**

.34**

(.63)

12

.16**

.20**

.16**

(.82)

13

– .36**

.22**

14

– .19**

15

(.60)

16

–

17

.06** .20** .28** .22** .17** .16** .17** .35**

.17** .14** .28** .12** .49** .25** .40**

.11** .21**

.11** .13**

.10** .12**

.10** .00

.03

.05*

.16** .15** .11** .09**

.20**

.08** .18**

.08** .13** .19**

.08**

.12**

.02

.04*

.02

.05* .01

8

.42** .27** .09** .16**

.11** .24**

.06** .01

.12**

–

.29** .30**

.07** .01

.15**

.09**

.02

.04

.07** .03

.19** .05*

.06**

.08**

.07**

0.78 .04*

0.49

0.84 .05*

0.63

0.75

0.42

0.71 .04*

6

.43** .52**

0.49 .12** .24** .05** .15** .03

.13**

.26**

.35**

.29** .11**

4

Note. N = 2,353. Reliability coefficients (Cronbach’s α) are reported along the diagonal in parentheses. Sex was coded as 0 = male and 1 = female; Age (1) = dummy variable indicating younger than 25; Age (2) = dummy variable indicating older than 55; educational level was coded as 0 = secondary level and 1 = tertiary level; marital status was coded as 0 = married and 1 = unmarried; teaching grade was coded as 0 = primary and 1 = secondary; work in hometown was coded as 0 = no and 1 = yes. *p < .05; **p < .01.

0.51

9. Work in hometown

.26** .08**

8.30

8. Organizational tenure 0.50 .20**

.04

.59** .26**

.07** .52**

.43** .52**

0.43

0.37 .09**

–

.43** .06**

.19** .42**

3

0.19 .14**

0.29

2

7.71 .11**

0.76

6. Educational level

–

.95** .37**

0.83

5. Marital status

1

9.69 .29**

0.49

SD

14.41 10.15 .26**

0.04

4. Age (2)

7. Occupational tenure

0.09

36.72

0.40

M

3. Age (1)

2. Age

1. Sex

Variable

Table 1. Means, standard deviations, reliabilities, and correlations

16 H. Li et al., Organizational and Occupational Turnover

Ó 2019 Hogrefe Publishing

H. Li et al., Organizational and Occupational Turnover

17

Table 2. Results of net regression analyses Variables

Organizational turnover intentions

Occupational turnover intentions

Age

0.01 (0.01)

0.03y (0.02)

Age (1)

0.41* (0.27)

Differencesa 0.02 (0.02)

0.78*** (0.19)

0.37 (0.34) 0.15 (0.45)

Age (2)

0.02 (0.35)

0.13 (0.28)

Sex

0.11 (0.10)

0.07 (0.10)

0.04 (0.14)

0.16 (0.13)

Educational level

0.26y (0.13)

0.02 (0.14)

0.24 (0.19)

Organizational tenure

0.02* (0.01)

0.01 (0.01)

0.01 (0.01)

Occupational tenure

0.00 (0.01)

0.01 (0.01)

0.01 (0.02)

y

0.42** (0.13)

0.26 (0.18)

Marital status

Hometown working

0.18 (0.10)

0.01 (0.10)

0.19 (0.14)

Teaching grade

0.12 (0.12)

0.20y (0.11)

0.32y (0.16)

IV: Organization-focused factors Leader support

0.40*** (0.07)

0.21** (0.07)

0.19* (0.10)

Colleague support

0.74*** (0.12)

0.29* (0.12)

0.45** (0.17)

Physical resources for work

0.69*** (0.20)

0.07 (0.15)

0.62* (0.25)

Occupational satisfaction

0.20** (0.07)

1.31*** (0.08)

1.11*** (0.11)

Salary satisfaction

0.20*** (0.06)

0.30*** (0.06)

0.10 (0.08)

Intrinsic occupational choice motivation

0.16 (0.10)

1.07*** (0.11)

0.91*** (0.15)

IV: Occupation-focused factors

Within-group R2

0.133***

0.379***

Between-group R2

0.143y

0.008

Note. N = 2,353. Standard errors are reported in parentheses. aReported difference between betas for different dependent variables from a single sample. y p < .10; *p < .05; **p < .01; ***p < .001.

salary satisfaction, and intrinsic occupational choice motivation) have a stronger relationship with occupational turnover intentions than with organizational turnover intentions (Hypothesis 4). In other words, we examined if the magnitude of the relationship between the antecedents and either type of turnover intentions was indeed significantly different. To do so, we used a test of the differences between betas for two dependent variables from a single sample (Cohen, Cohen, West & Aiken, 2003, p. 157 & p. 642).

The results showed that occupation-focused predictors have stronger relationships with occupational turnover intentions than organizational turnover intentions, for occupational satisfaction (difference = 1.11, p < .001) and intrinsic occupational choice motivation (B = 0.91, p < .001), except for salary satisfaction (B = 0.10, p > .05). Thus, Hypothesis 2 was partially supported in that two, but not all, of the occupation-focused predictors have a stronger association with occupational turnover intentions than with organizational turnover intentions. In contrast, three organization-focused predictors have stronger relationships with organizational turnover intentions than occupational turnover intentions, for leader support (B = 0.19, p < .05), colleague support (B = 0.45, p < .01), and physical resources for work (B = 0.62, p < .05). This provided full support for Hypothesis 4.

Ó 2019 Hogrefe Publishing

Discussion Drawing on social cognitive career theory (SCCT; Lent et al., 2002) and the focus–congruence approach (Jackofsky & Peters, 1983; Klein et al., 2012), and with a sample of 2,353 teachers in China, the current study firstly showed that three proposed organization-focused factors, namely leader support, colleague support, and physical resources for work, are negatively associated with organizational turnover intentions. Moreover, these three organization-focused factors have a stronger relationship with organizational turnover intentions than with occupational turnover intentions. Similarly, three proposed occupation-focused factors, namely occupational satisfaction, salary satisfaction, and intrinsic occupational choice motivation, are negatively associated with occupational turnover intentions. Furthermore, these three occupation-focused factors have a stronger relationship with occupational turnover intentions than with organizational turnover intentions.

Theoretical Implications The present research contributes to the literature on turnover in the following ways. First, the current study uncovers

Journal of Personnel Psychology (2019), 18(1), 10–22

18

how employees distinguish between intentions to leave their organizations and occupations, which supported the uniqueness of employees’ intentions to change occupations as distinguished from leaving the organization. In line with earlier theorizing (e.g., Jackofsky, 1984; Jackofsky & Peters, 1983; Krausz et al., 1995; Wright & Bonett, 1992), these results suggest that organizational and occupational turnover intentions are constructs that were modestly correlated (r = .35), sharing 13% of their variances.

Second, to address prior calls to assess organizational and occupational factors in turnover research (Woo & Maertz, 2012), we identified six individual and situational factors based on SCCT (Lent, Brown, & Hackett, 2002). Along with the focus–congruence approach (Klein et al., 2012), the findings show that associations were stronger when the foci of turnover intentions and potential antecedents match than when they did not (Jackofsky & Peters, 1983).

While leader support, colleague support, and physical resources for work were negatively associated with organizational turnover intentions, more importantly, we identified occupational satisfaction, salary satisfaction, and intrinsic occupational choice motivation as predictors of occupational turnover intentions. Based on these findings, the present study sheds new light on the turnover research taking the perspective of switching occupations rather than changing organizations.

Third, while most previous research only focused on predicting organizational turnover intentions, this study is one of the few to simultaneously measure different dimensions of turnover intentions and test the quantity difference of those predictors’ impact on them. Using net regression analyses, the present study revealed that occupationfocused predictors contributed more to the prediction of occupational turnover intentions than organization-focused predictors. In contrast, organization-focused predictors had stronger relationships with organizational turnover intentions than occupational turnover intentions. One unexpected result is that the negative relationships between salary satisfaction and two types of turnover intentions were both significantly negative, with no significant difference in the strength of relationships. This is perhaps because the salary level of schools in rural China is not only lower than that of other occupations but also lower than that of schools in urban areas; therefore, pay dissatisfaction triggers both types of turnover intentions.

Finally, this study contributes to the focus–congruence approach by generalizing it to a wider range of organizational phenomena. Because of varying beliefs about one’s different roles and social identities (Hogg & Terry, 2000; Tajfel, 1972; Turner, 1982), an employee is able to form different attitudes toward organizational and occupational turnover. Similarly to turnover intentions, other important work attitudes

Journal of Personnel Psychology (2019), 18(1), 10–22

H. Li et al., Organizational and Occupational Turnover

(e.g., commitment, identification, attachment, and trust) that are often treated as single-target constructs might also have more than one focus, reflect different constructs, and be related to different antecedents. For example, Zettler, Friedrich, and Hilbig (2011) divided career commitment into self-related work commitment and other-related work commitment and found a difference of predictivity of Machiavellianism on these two types of career commitment.

Limitations and Future Directions Due to the cross-sectional data and potential common method variance in this study, definitive conclusions about the differentiated models of turnover intentions require further work. Although it is worth noting that, as we employed multilevel modeling and some level-2 antecedents (e.g., physical resources for work) with different teachers’ ratings aggregated at school level were included to predict individuals’ turnover intentions, the concern over common method bias should be mitigated. In spite of this, the present study did not employ methods that would warrant causal inference, and “predictor” and “antecedent” were used in a general sense. To make better causal inferences, a longitudinal approach is needed in future research.

Future researchers might also use objective data collected from other sources, such as supervisors, colleagues, or customers, to further reduce possible common method bias.

In addition, we used the comparison of beta coefficients from two regression models for two types of turnover intentions. The test of comparative effects may cause bias if the comparison is unfair (Cooper & Richardson, 1986). Unfair comparison occurs when the operationalization or measurement of the competing factors is not equally cared, or when the value of competing factors is not set at comparable levels along those factors’ respective value distributions (Cooper & Richardson, 1986). Although we tried to operationalize and measure all predictors in an equivalent way, we acknowledge that possible unfair comparison may still exist and disturb the results. Future research could address this point and adopt alternative strategies that offer fairer comparisons between regression models. Besides, due to the limitation of the data, some constructs (e.g., dependent variables) in the current study were measured using singleitem scales, which constrains the validity of the research findings. Future research could use well-developed scales instead to get more robust evidence of the relationship between two types of turnover and their antecedents.

Second, future research should use samples from various types of organizations and occupations to test turnover attitudes with different foci. In other types of organizations

Ó 2019 Hogrefe Publishing

H. Li et al., Organizational and Occupational Turnover

(or bigger schools), the dynamics within work groups might have a significant impact on turnover intentions.

Researchers who studied the importance of work-group identification relative to organizational identification found that the former was stronger than the latter, as well as more predictive of organizational attitudes and behavior such as job satisfaction, turnover intentions, job involvement, and job motivation (Van Knippenberg & Van Schie, 2000). In addition, some suggestions for management and human resources in other occupations may not apply to the education system but are useful for other organizations. For instance, Dalton and Todor (1993) suggested that changes in absenteeism and intra-organization transfer policies could reduce turnover. For many organizations, especially those that are large and decentralized, breaking down boundaries between workgroups or having more flexible schedules might boost retention.

Also, future studies could examine the mechanisms or boundary conditions of the relationship between organizational and occupational turnover intentions. Blau (1989) suggested that occupational turnover intentions may have incremental effects on turnover behavior beyond the effects of organizational turnover intentions. It also seems there might be a reciprocal relationship between organizational and occupational turnover intentions (Chang, Chi, & Miao, 2007). As Woo and Maertz (2012) reasoned, occupational turnover intentions may positively relate to organizational turnover intentions because changing occupations implies changing organizations when some employees do not have the opportunity to change occupations within a given organization. Occupational attachment might yet reduce turnover intentions under the condition that this particular occupation is compatible, supported, and given adequate status within the organization, and not readily available in other organizations (Lee, Carswell, & Allen, 2000). The opposite condition may translate into higher organizational turnover intentions for those who identify significantly with their occupation (Woo & Maertz, 2012).

Finally, although in the present study we treated organizational turnover intentions and occupational turnover intentions as two “parallel” choices that individuals have when considering career transition, we admitted that the actual behavior of organizational turnover and occupational turnover might be not so “parallel.” For that occupational turnover is much more difficult for individuals than organizational turnover and occurs much less often (Blau, 2007), the strength and mechanisms may be different for the links between the two types of turnover intentions and actual turnover behavior. Given the rare data on actual occupational turnover (Blau, 2007), this line of investigation is absent. Future research could address this issue and collect behavior data on the two types of turnover. In doing so, it will be possible to examine questions such as whether the Ó 2019 Hogrefe Publishing

19

relationship between organizational turnover intentions and behavior is stronger than the relationship between occupational turnover intentions and behavior.

Practical Implications Differentiating between antecedents for organizational turnover intentions and occupational turnover intentions has practical implications for turnover interventions. For example, in the case of the intervention to curb teacher turnover in rural China, policymakers and education administrators should separate the issue of teacher shortage into two problems, namely teachers’ migration to more economically developed areas and teachers’ attrition from the occupation. At the country or regional level, policymakers should pay attention to both the school-specific factors and the occupation-wide factors that might contribute to teachers’ intentions to withdraw. Little is likely to be achieved, in terms of tackling the inequality of education quality between urban and rural areas, if policies focus only on occupation-wide factors, for example, selecting teachers that choose the occupation for intrinsic reasons. A teacher might have high levels of intrinsic motivation but still might leave a resource-constrained and low-performing school for a more desirable school.

On the other hand, by only focusing on interventions at the school level, such as the social exchange between principals and teachers, colleague support among teachers, or school physical resources, policymakers might still fail to retain teachers in the education system. In order to keep teachers both in their current schools and in the occupation, both school-specific and occupation-wide factors need to be addressed simultaneously. At the school level, administrators and teachers must be well informed to make the best use of school-specific resources, such as the social exchange between principals and teachers, colleague support among teachers, and physical resources for teachers to use. For example, although a school might not have the power to raise its teachers’ salaries, it can cultivate collaboration and cohesion among teachers and between teachers and the principal.

Conclusion In summary, this study provides evidence of the uniqueness of employees’ intentions to change occupations as distinguished from leaving the organization. Among factors that impacted these two types of withdrawal intentions, in general, associations were stronger when the foci of turnover intentions and potential antecedents matched than when they did not. For future research, the focus–congruence Journal of Personnel Psychology (2019), 18(1), 10–22

20

approach offers a useful perspective for examining this phenomenon in organizations, so researchers and practitioners can arrive at a more nuanced understanding of both organizational and occupational commitment and withdrawal.

References Allen, D. G., Bryant, P. C., & Vardaman, J. M. (2010). Retaining talent: Replacing misconceptions with evidence-based strategies. Academy of Management Perspectives, 24, 48–65. https:// doi.org/10.5465/AMP.2010.51827775 Bakker, A. B., & Demerouti, E. (2007). The job demands–resources model: State of the art. Journal of Managerial Psychology, 22, 309–328. https://doi.org/10.1108/02683940710733115 Blau, G. (1985). The measurement and prediction of career commitment. Journal of Occupational Psychology, 58, 277–288. https://doi.org/10.1111/j.2044-8325.1985.tb00201.x Blau, G. (1989). Testing the generalizability of a career commitment measure and its impact on employee turnover. Journal of Vocational Behavior, 35, 88–103. https://doi.org/10.1016/00018791(89)90050-X Blau, G. (2007). Does a corresponding set of variables for explaining voluntary organizational turnover transfer to explaining voluntary occupational turnover? Journal of Vocational Behavior, 70, 135–148. https://doi.org/10.1016/j.jvb.2006.07.007 Bliese, P. D. (2000). Multilevel random coefficient modeling in organizational research: Examples using SAS and S-PLUS. In F. Drasgow & N. Schmitt (Eds.), Measuring and analyzing behavior in organizations (pp. 401–445). San Francisco, CA: Jossey-Bass.

Brightman, H. J., & Sayeed, L. (1990). The pervasiveness of senior management’s view of the cultural gaps within a division. Group & Organization Management, 15, 266–278. https://doi.org/ 10.1177/105960119001500303 Caldwell, S. D., Herold, D. M., & Fedor, D. B. (2004). Toward an understanding of the relationships among organizational change, individual differences, and changes in person– environment fit: A cross-level study. Journal of Applied Psychology, 89, 868–882. https://doi.org/10.1037/0021-9010.89.

5.868 Caplan, R., Cobb, S., French, J., Van Harrison, R., & Pinneau, S. (1975). Demands and worker health: Main effects and organizational differences. Washington, DC: US Government Printing Office.

Centers, R., & Bugental, D. E. (1966). Intrinsic and extrinsic job motivations among different segments of the working population. Journal of Applied Psychology, 50, 193–197. https://doi. org/10.1037/h0023420 Chang, H. T., Chi, N. W., & Miao, M. C. (2007). Testing the relationship between three-component organizational/occupational commitment and organizational/occupational turnover intention using a non-recursive model. Journal of Vocational Behavior, 70, 352–368. https://doi.org/10.1016/j.jvb.2006.10.001 Chapman, D. W., & Lowther, M. A. (1982). Teachers’ satisfaction with teaching. Journal of Educational Research, 75, 241–247. https://doi.org/10.1080/00220671.1982.10885388 Chatman, J. A. (1991). Matching people and organizations: Selection and socialization in public accounting firms. Administrative Science Quarterly, 36, 459–484. https://doi.org/10.5465/AMBPP. 1989.4980837 Cohen, J., Cohen, P. C., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum.

Journal of Personnel Psychology (2019), 18(1), 10–22

H. Li et al., Organizational and Occupational Turnover

Cooper, W. H., & Richardson, A. J. (1986). Unfair comparisons.

Journal of Applied Psychology, 71, 179–184. https://doi.org/ 10.1037/0021-9010.71.2.179 Cotton, J. L., & Tuttle, J. M. (1986). Employee turnover: A metaanalysis and review with implications for research. Academy of Management Review, 11, 55–70. https://doi.org/10.5465/AMR. 1986.4282625 Dalton, D. R., & Todor, W. D. (1993). Turnover, transfer, absenteeism: An interdependent perspective. Journal of Management, 19, 193–219. https://doi.org/10.1016/0149-2063(93) 90052-O Deci, E. L., & Ryan, R. M. (1985). Self-determination. Hoboken, NJ: Wiley.

Fimian, M. J., Fastenau, P. S., & Thomas, J. A. (1988). Stress in nursing and intentions of leaving the profession. Psychological Reports, 62, 499–506. https://doi.org/10.2466/pr0.1988.62.

2.499 Gerstner, C. R., & Day, D. V. (1997). Meta-analytic review of leader-member exchange theory: Correlates and construct issues. Journal of Applied Psychology, 82, 827–844. https:// doi.org/10.1037/0021-9010.82.6.827 Greenhaus, J. H., Parasuraman, S., & Wormley, W. M. (1990).

Effects of race on organizational experience, job performance evaluations, and career outcomes. Academy of Management Journal, 33, 64–86. https://doi.org/10.2307/256352 Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis of antecedents and correlates of employee turnover: Update, moderator tests, and research implications for the next millennium. Journal of Management, 26, 463–488. https://doi. org/10.1177/014920630002600305 Heneman, H. G. III, & Schwab, D. P. (1985). Pay satisfaction: Its multidimensional nature and measurement. International Journal of Psychology, 20, 129–141. https://doi.org/10.1080/ 00207598508246743 Hogg, M. A., & Terry, D. J. (2000). Social identity and selfcategorization processes in organizational contexts. Academy of Management Review, 25, 121–140. https://doi.org/10.5465/ AMR.2000.2791606 Hom, P. W. (2011). Organizational exit. In S. Zedeck (Ed.), Handbook of industrial/organizational psychology (2nd ed., pp. 67–117).

Washington, DC: American Psychological Association.

Hom, P. W., Mitchell, T. R., Lee, T. W., & Griffeth, R. W. (2012).

Reviewing employee turnover: Focusing on proximal withdrawal States and an expanded criterion. Psychological Bulletin, 138, 831–858. https://doi.org/10.1037/a0027983 Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: An organizational analysis. American Educational Research Journal, 38, 499–534. https://doi.org/10.3102/00028312038003499 Ingersoll, R. M. (2003). Is there really a teacher shortage? A research report. Seattle, WA: University of Washington Center for the Study of Teaching and Policy (CTP).

Jackofsky, E. F. (1984). Turnover and job performance: An integrated process model. Academy of Management Review, 9, 74–83. https://doi.org/10.5465/AMR.1984.4277940 Jackofsky, E. F., & Peters, L. H. (1983). Job turnover versus company turnover: Reassessment of the March and Simon participation hypothesis. Journal of Applied Psychology, 68, 490–495. https://doi.org/10.1037/0021-9010.68.3.490 James, L. R., Demaree, R. G., & Wolf, G. (1984). Estimating withingroup interrater reliability with and without response bias.

Journal of Applied Psychology, 69, 85–98. https://doi.org/ 10.1037/0021-9010.69.1.85 James, L. R., Hater, J. J., Gent, M., & Bruni, J. (1978). Psychological climate: Implications from cognitive social learning theory and interactional psychology. Personnel Psychology, 31, 783–813. https://doi.org/10.1111/j.1744-6570.1978.tb02124.x

Ó 2019 Hogrefe Publishing

H. Li et al., Organizational and Occupational Turnover

James, L. R., & Tetrick, L. E. (1986). Confirmatory analytic tests of three causal models relating job perceptions to job satisfaction. Journal of Applied Psychology, 71, 77–82. https://doi.org/10.1037/0021-9010.71.1.77 Johnson, S. M., Berg, J. H., & Donaldson, M. L. (2005). Who stays in teaching and why: A review of the literature on teacher retention. Boston, MA: Harvard Graduate School of Education, Project on the Next Generation of Teachers.

Kelloway, E. K., Barling, J., & Shah, A. (1993). Industrial relations stress and job satisfaction: Concurrent effects and mediation.

Journal of Occupational Behavior, 14, 447–457. https://doi.org/ 10.1002/job.4030140506 Kerr, J., & Slocum, J. W. (1987). Managing corporate culture through reward systems. Academy of Management Executive, 1, 99–107. https://doi.org/10.5465/AME.1987.4275817 Klein, H. J., Molloy, J., & Brinsfield, C. (2012). Reconceptualizing workplace commitment to redress a stretched construct: Revisiting assumptions and removing confounds. Academy of Management Journal, 37, 130–151. https://doi.org/10.5465/ amr.2010.0018 Kopelman, R. E., Brief, A. P., & Guzzo, R. A. (1990). The role of climate and culture productivity. In B. Schneider (Ed.), Organizational climate and culture (pp. 282–318). Oxford, UK: Jossey-Bass.

Krausz, M., Koslowsky, M., Shalom, N., & Elyakim, N. (1995).

Predictors of intentions to leave the ward, the hospital, and the nursing profession: A longitudinal study. Journal of Organizational Behavior, 16, 277–288. https://doi.org/10.1002/job.

4030160308 Lachman, R., & Diamant, E. (1987). Withdrawal and restraining factors in teachers’ turnover intentions. Journal of Organizational Behavior, 8, 219–232. https://doi.org/10.1002/job.

4030080304 Lane, I. M., Mathews, R. C., & Presholdt, P. H. (1988). Determinants of nurses’ intentions to leave their profession. Journal of Organizational Behavior, 9, 367–372. https://doi.org/10.1002/ job.4030090408 LeBreton, J. M., & Senter, J. L. (2008). Answers to 20 questions about interrater reliability and interrater agreement. Organizational Research Methods, 11, 815–852. https://doi.org/ 10.1177/1094428106296642 Lee, K., Carswell, J. J., & Allen, N. J. (2000). A meta-analytic review of occupational commitment: Relations with person- and workrelated variables. Journal of Applied Psychology, 85, 799–811. https://doi.org/10.1037/0021-9010.85.5.799 Lee, P. C. B. (2004). Social support and leaving intention among computer professionals. Information & Management, 41, 323–334. https://doi.org/10.1016/S0378-7206(03)00077-6 Lent, R. W., Brown, S. D., & Hackett, G. (2002). Social cognitive career theory. In D. Brown (Ed.), Career choice and development (pp. 255–311). San Francisco, CA: Jossey-Bass.

Liden, R. C., Sparrowe, R. T., & Wayne, S. J. (1997). Leadermember exchange theory: The past and potential for the future.

Research in Personnel and Human Resources Management, 15, 47–119.

Lord, R. G., & Brown, D. J. (2004). Leadership processes and follower self-identity. Mahwah, NJ: Erlbaum.

Maertz, C. P., & Griffeth, R. W. (2004). Eight motivational forces and voluntary turnover: A theoretical synthesis with implications for research. Journal of Management, 30, 667–683. https://doi.org/10.1016/j.jm.2004.04.001 Ministry of Education of the People’s Republic of China. (2000).

Number of teachers, staff & workers in primary schools.

Retrieved from http://www.moe.gov.cn/publicfiles/business/ htmlfiles/moe/moe_591/200506/7923.html

Ó 2019 Hogrefe Publishing

21

Ministry of Education of the People’s Republic of China. (2014).

Number of Educational Personnel in Primary Schools. Retrieved from http://www.moe.gov.cn/publicfiles/business/htmlfiles/ moe/s8490/201412/181903.html Ministry of Education of the People’s Republic of China. (2008).

National education supervision report 2008. Retrieved from http://www.moe.gov.cn/publicfiles/business/htmlfiles/moe/ moe_914/201001/xxgk_81660.html Mintzberg, H. (1979). The structuring of organizations: A synthesis of the research (SSRN Scholarly Paper No. ID 1496182).

Rochester, NY: Social Science Research Network Mobley, W. H., Griffeth, R. W., Hand, H. H., & Meglino, B. (1979).

Review and conceptual analysis of the employee turnover process. Psychological Bulletin, 86, 493–522. https://doi.org/ 10.1037/0033-2909.86.3.493 Muthén, L. K., & Muthén, B. O. (1998–2012). Mplus user’s guide.

Los Angeles, CA: Muthén & Muthén.

Ng, T. W. H., & Feldman, D. C. (2007). Organizational embeddedness and occupational embeddedness across career stages.

Journal of Vocational Behavior, 65, 336–351. https://doi.org/ 10.1016/j.jvb.2006.10.002 Rentsch, J. R. (1990). Climate and culture: Interaction and qualitative differences in organizational meanings. Journal of Applied Psychology, 75, 668–681. https://doi.org/10.1037/ 0021-9010.75.6.668 Sargent, T., & Hannum, E. (2005). Keeping teachers happy: Job satisfaction among primary school teachers in rural northwest china. Comparative Education Review, 49, 173–204. https://doi. org/10.1086/428100 Schneider, B., Parkington, J. J., & Buxton, V. M. (1980). Employee and customer perceptions of service in banks. Administrative Science Quarterly, 25, 252–267. https://doi.org/10.2307/ 2392454 Sheridan, J. E. (1992). Organizational culture and employee retention. Academy of Management Journal, 35, 1036–1056. https://doi.org/10.2307/256539 Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420–428. https://doi.org/doi:10.1037/0033-2909.86.2.420 Smith, P. C. (1976). Behavior, results, and organizational effectiveness: The problem of criteria. In M. D. Dunnette (Ed.), Handbook of industrial and organizational psychology (pp. 745–775).

Chicago, IL: Rand McNally.

Tajfel, H. (1972). Social categorization (English translation of “La categorisation sociale”). In S. Moscovici (Ed.), Introduction à la Psychologie Sociale (pp. 272–302). Paris, France: Larousse.

Turner, J. C. (1982). Towards a cognitive redefinition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations (pp. 15–40). Cambridge, UK: Cambridge University Press.

Van Knippenberg, D., & Van Schie, E. C. M. (2000). Foci and correlates of organizational identification. Journal of Occupational and Organizational Psychology, 73, 137–147. https://doi. org/10.1348/096317900166949 Woo, S. E., & Maertz, C. P. (2012). Assessment of voluntary turnover in organizations: Answering the questions of why, who, and how much. In N. Schmitt (Ed.), The Oxford handbook of personnel assessment and selection (pp. 570–594). New York, NY: Oxford University Press.

Wright, T. A., & Bonett, D. G. (1992). The effect of turnover on work satisfaction and mental health: Support for a situational perspective. Journal of Organizational Behavior, 13, 603–615. https://doi.org/10.1002/job.4030130606 Yousaf, A., Sanders, K., & Abbas, Q. (2015). Organizational/ occupational commitment and organizational/occupational

Journal of Personnel Psychology (2019), 18(1), 10–22

22

turnover intentions. Personnel Review, 44, 470–491. https:// doi.org/10.1108/PR-12-2012-0203 Zettler, I., Friedrich, N., & Hilbig, B. E. (2011). Dissecting work commitment: The role of Machiavellianism. Career Development International, 16, 20–35. https://doi.org/10.1108/ 13620431111107793

H. Li et al., Organizational and Occupational Turnover

Kun Yu School of Labor and Human Resources Renmin University of China Beijing, 100872 China yuk@ruc.edu.cn

History Received May 22, 2017 Revision received April 11, 2018 Accepted April 19, 2018 Published online January 9, 2019 Acknowledgments The authors thank Christine Min Wotipka, Jennifer Adams, John Hausknecht, and the participants of Stanford ICE/IEAPA and Cornell ILR Human Resource Studies workshops for their valuable feedback on earlier versions of this paper.

Funding This research was supported by the National Natural Science Foundation of China (NSFC Grant 71702184).

ORCID Kun Yu https://orcid.org/0000-0002-9373-8367

Journal of Personnel Psychology (2019), 18(1), 10–22

Ó 2019 Hogrefe Publishing

Original Article

When Do Those High in Trait Self-Control Suffer From Strain?

The Interplay of Trait Self-Control and Multiple Stressors Kai Externbrink1 , Stefan Diestel2, and Martina Krings1 1

Department of Business and Psychology, FOM University of Applied Sciences, Dortmund, Germany

2

International School of Management, Dortmund, and University of Wuppertal, Germany

Abstract: We explore the limits of the protective function of trait self-control in coping with sources of stress. Inspired by integrative selfcontrol theory (ISCT) we predict that trait self-control only buffers the relationship between self-control demands and irritation when individuals have to cope with one source of stress, whereas in cases of two stressors, trait self-control fails to attenuate adverse effects.

Samples consisted of occupational students (N = 163) and partly or fully or not formally employed students (N = 135). Job-related self-control demands (SCDs) did not predict strain when trait self-control was high and the other stressor (academic SCDs or weekly study time) was low, whereas strain was disproportionally higher and predicted by SCDs when trait self-control was low or the other stressor was high.

Keywords: academic and job-related self-control demands, study and working time per week, trait self-control, irritation

Trait self-control reflects interindividual differences in the ability to override spontaneous impulses, motivational blockades, and habitual response patterns (Tangney, Baumeister, & Boone, 2004). Growing evidence demonstrates trait self-control is a promising precursor for selfregulatory functioning in nearly all life domains (de Ridder, Lensvelt-Mulders, Finkenauer, Stok, & Baumeister, 2012).

Not only academic and job performance are predicted by trait self-control (Stumm, Thomas, & Dormann, 2010; Zettler, 2011), but also stable interpersonal relationships, psychological health, and higher self-acceptance (Shoda, Mischel, & Peake, 1990; Vohs & Faber, 2007). Finally, and more relevant for personnel psychology, supervisors’ trait self-control buffers deleterious effects of dysfunctional emotional regulation on their resulting tendency to abusive supervision (Yam, Fehr, Keng-Highberger, Klotz, & Reynolds, 2016). Conversely, in cases of high abusive supervision, followers with high trait self-control show less supervisor-directed aggression as compared to followers with low trait self-control (Lian et al., 2014). Inspired by such and similar findings, Schmidt, Hupke, and Diestel (2012) have identified trait self-control as a protective resource, which prevents high psychological strain, especially when demands are high.

Although trait self-control seems highly important for employees’ psychological health, recent theoretical models suggest boundary conditions under which trait self-control may not stabilize well-being and even those with high trait

Ó 2019 Hogrefe Publishing

self-control experience disproportionately high strain in response to high demands. In particular, integrative self-control theory (ISCT; Kotabe & Hofmann, 2015; Lian, Yam, Ferris, & Brown, 2017) proposes a trait component in the capacity to exert volitional self-control and implies that simultaneously occurring demands that exceed the capacity may result in disproportionately high strain, even when the trait component is well developed (i.e., trait self-control is high). In other words, trait self-control will protect individuals from high strain, if and only if they have to cope with one demand.

We believe the limits of trait self-control become notably manifest for those who are faced with multiple sources of stress in different domains putting high demands on selfcontrol, and thus taxing the control capacity (e.g., Diestel & Schmidt, 2011). A prototypical example is occupational students who have to cope with stressors in two domains, the academic and the job-related setting. While occupational study programs address skills shortages in organizations and enable employees to gain extra qualifications without restricting professional activity, integrating study and work into weekly schedules can be stressful and can result in psychological strain, because lectures are held in the evenings after a working day and at weekends. Therefore, based on two samples with students who were either fully or partially employed, we examine the moderating role of trait self-control under conditions of two sources of stress. We thereby focus on self-control demands (SCDs),

Journal of Personnel Psychology (2019), 18(1), 23–33 https://doi.org/10.1027/1866-5888/a000218

24

especially overcoming inner resistance (the requirement to override motivational deficits to complete unattractive tasks that cannot be postponed or evaded; Schmidt & Diestel, 2015), because past research has repeatedly revealed that such SCDs are strongly positively related with indicators of psychological strain, such as burnout symptoms, impaired work engagement, and absenteeism. In conclusion, we predict a three-way interaction of trait self-control and job-related and academic SCDs on strain: Trait selfcontrol attenuates the positive relation of one stressor to psychological strain, if and only if the other stressor is low, whereas trait self-control does not prevent strain in cases of coping with both stressors simultaneously.

We thereby contribute to the literature in the following ways: firstly, we apply experimental findings on academic SCDs (Oaten & Cheng, 2005) to a sample of occupational students for whom self-control is highly relevant. Secondly, and theoretically more important, the three-way interaction of trait self-control and both stressors may reveal psychological contingencies under which trait self-control does not buffer the deleterious effects of demands on strain and fails to provide protective resilience in coping with (disproportionately) high stress.

Volitional Self-Control in the Academic and Occupational Context In increasingly complex and dynamic working environments (Sonnentag & Frese, 2013), employees have to cope with regulatory demands, such as emotional labor (Hülsheger & Schewe, 2011), time pressure (Prem, Paškvan, Kubicek, & Korunka, 2018), problem-solving (Schmitt, Zacher, & Frese, 2012) and SCDs (Schmidt & Diestel, 2015) which cause them to engage in self-control (Prem, Kubicek, Diestel, & Korunka, 2016). In addition, Prem et al. (2016) demonstrated that exercising self-control mediates the relation of such regulatory demands to strain.

In addition to these and similar findings, basic research provides more nuanced insights into how exercising selfcontrol (e.g., overcoming motivational blockades, attentional control, or impulse regulation tasks) causes impairments in executive functioning, lower self-control performance, and exhaustion (Dang, 2018; Hagger, Wood, Stiff, & Chatzisarantis, 2010). To explain the so-called egodepletion effect (Muraven & Baumeister, 2000), Kotabe and Hofmann (2015) delineated the ISCT, which distinguishes between three phases of the self-control process (activation, exertion, and enactment). A desire–goal conflict, which may result from unattractive, but important tasks and is experienced as an inner motivational blockade, activates the intention to exert self-control to overcome Journal of Personnel Psychology (2019), 18(1), 23–33

K. Externbrink et al., The Limits of Trait Self-Control

inner resistances. The exertion of self-control depends on a limited, depletable, but restorable control capacity (Kotabe & Hofmann, 2015, p. 626), which was originally proposed by Muraven and Baumeister (2000) and involves some kind of energetic resource for “will-power.” If high desire–goal conflicts (SCDs) exceed or deplete the capacity, behavioral enactment of self-control will fail and individuals will experience increasing strain. Although the idea of a limited “resource” capacity inspired a controversial debate about its existence and several meta-analytical replication studies (Carter, Kofler, Forster, & McCullough, 2015; Dang, 2018), the mechanistic framework of ISCT and other models of executive functioning (Miyake et al., 2000) provide a well-developed basis for predictions of self-control failures and psychological strain at work (Lian et al., 2017). Additionally, behavioral enactment of self-control may also be impaired by external constraints, like additional sources of stress, which prevent the control capacity from restoring, and thus disproportionately increase the risk of self-control failures and associated strain symptoms.

Supporting this view, Diestel and Schmidt (2011) found that job-related SCDs and emotional dissonance mutually amplify each other in their deleterious effects on strain.

However, not only the occupational setting, but also the academic context can put high demands on volitional self-control and thus cause psychological strain. In their longitudinal studies at universities, Oaten and Cheng (2005, 2006) repeatedly reported increasing indicators of strain, such as emotional distress and psychosomatic symptoms, and self-control failures during academic examination periods as compared to students not facing examination stress. In line with ISCT, such findings indicate that time pressure, learning- and problem-solving demands, as well as academic examinations, are aversive and thus tax students’ control capacity. Academic tasks may in general require high self-control, because lectures are often complicated, schedules are usually fixed not allowing for flexible planning and decision making, and academic projects are increasingly complex (e.g., Thomas & Mengel, 2008). Consistent with this argument, burnout and other indicators of psychological strain among students have been repeatedly reported (Stoeber, Childs, Hayward, & Feast, 2011). In sum, employees as well as students are faced with SCDs and are likely to experience high strain as a result of coping with such sources of stress. However, given that many employees are enrolled in academic programs and students increasingly have to work parallel to their studies (in order to self-finance or to gain occupational experience), a growing number of individuals may have to cope with multiple stressors in academic and occupational settings with consequences for their psychological health and wellbeing. Ó 2019 Hogrefe Publishing

K. Externbrink et al., The Limits of Trait Self-Control

The Interplay of Trait Self-Control, Job-Related SCDs, and Academic SCDs In both academic and job-related settings, trait self-control has emerged as a strong predictor of performance and well-being (de Ridder et al., 2012; Tangney et al., 2004).

Moreover, Schmidt et al. (2012) found that trait selfcontrol attenuated the positive relationship between jobrelated SCDs and psychological strain (see also, Yam et al., 2016).

Despite this promising protective function of trait selfcontrol in coping with sources of stress, ISCT suggests that strain disproportionately increases in cases of simultaneous coping with at least two SCDs, even for those who exhibit high levels of trait self-control. In particular, the trait component of the control capacity reflects interindividual differences in the general ability to volitionally regulate behavior and assures effective regulatory functioning under conditions of goal–desire conflicts (high SCDs) by conserving and efficiently investing the limited resources in control efforts. In support of this argument, Diestel, Rivkin, and Schmidt (2015) found that both trait self-control and sleep quality attenuated the positive relations of emotional dissonance to strain. According to their explanation, trait selfcontrol seems to stabilize well-being through the effective allocation of resources as provided by sleep quality (see also, Beedie & Lane, 2012). However, trait-driven regulatory functioning should only enable successful coping with SCDs, and thus prevent strain, if the overall control capacity was not already depleted and/or enactment constraints, such as other stressors, did not impede restoration of the capacity. In other words, even when trait self-control is high, simultaneous occurring demands, which draw on the same control capacity, should result in high strain, because successful coping with one stressor is constrained by the other stressor. In conclusion, high trait self-control should attenuate positive relations of one stressor to strain, when the other stressor is low, whereas in cases of two stressors, strain should increase, regardless of the level of trait self-control.

Hypothesis 1: Trait self-control and academic and jobrelated SCDs interact in predicting psychological strain: The relationship of one stressor to strain is weakest when the other stressor is low and trait self-control is high. In comparison, in all other cases (the other stressor is high and/or trait self-control is low), SCDs are stronger and positively related to strain.

Ó 2019 Hogrefe Publishing

25

Study 1 Method We conducted a cross-sectional survey among occupational students of a German university of applied sciences. Participants were recruited via mailings and during lectures. The sample consisted of 163 participants studying for a bachelor degree parallel to a full-time job, of which 65.03% were female. Age was assessed in categories: 60.47% were between 18 and 25 years, 33.13% between 26 and 33 years, and 6.13% were older than 34.

Irritation As an indicator of psychological strain, we assessed irritation using the Irritation Scale which was developed by Mohr, Rigotti, and Mueller (2005). The scale involves several perceived emotional and cognitive strain symptoms, such as rumination, difficulties to detach, feelings of anger and impulsivity (eight items, e.g., “I anger quickly”). We focused on irritation because those symptoms reflect failures in self-control (anger and deficits in impulse regulation) and are conceptually related to a wide range of other strain variables (such as distress, depressive symptoms, and daytime dysfunction), which are predicted by demands on self-control (Oaten & Cheng, 2005; Schmidt et al., 2012). All items were scored on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree).

Trait Self-Control Dispositional self-control was measured with the Self-Control Scale by Bertrams and Dickhäuser (2009; 13 items, e.g., “I am good at resisting temptation”). Items were rated on a 5-point scale (1 = not at all like me, 5 = very much like me).

Job-Related and Academic SCDs We used five items to measure job-related demands on overcoming inner resistances (e.g., “Some of my work tasks are such that I really need to force myself to get them done”; Schmidt & Diestel, 2015). To assess academic SCDs, we adapted the content of the five items to the academic context by changing the word “work” to “academic” (e.g., “Some of my academic tasks are such that I really need to force myself to get them done”). Prior research showed that students are well able to discriminate different forms of SCDs across time and content domains (Oaten & Cheng, 2005). All items were scored on a 5-point rating scale (1 = not at all, 5 = a great deal). Confirmatory factor analyses (CFAs) provide support for the discriminative validity of both sources of SCDs [2-factor model: w2(34) = 60.13, p < .01, RMSEA = .069, CFI = .959, SRMR = .047; 1-factor

Journal of Personnel Psychology (2019), 18(1), 23–33

26

K. Externbrink et al., The Limits of Trait Self-Control

Table 1. Descriptive results, α reliabilities, and intercorrelations (Study 1) M

SD

α

1. Irritation

3.45

0.97

.77

2. Gender

1.65

0.48

–

3. Age

2.50

1.23

–

.14*

.06

4. Job-related self-control demands

2.67

0.80

.84

.44**

.05

.14*

5. Academic self-control demands

3.04

0.84

.89

.41**

.09

.04

.55**

6. Trait self-control

3.10

0.51

.75

.33**

.07

.10

.56**

1.

2.

3.

4.

5.

.09

.60**

Note. Gender: 1 = male, 2 = female; age: 1 = 18–21 years, 2 = 22–25 years, 3 = 26–29 years, 4 = 30–33 years, 5 = 34–36 years, 6 = 36+ years. *p < .05, **p < .01.

model: w2 (35) = 174.62, p < .01, RMSEA = .156, CFI = .782, SRMR = .095].

Table 2. Results of multiple regression analysis for predicting irritation (Study 1) β

t

Gender

.05

0.71

Age

.09

1.26

Job-related self-control demands

.28

3.19**

Academic self-control demands

.23

2.52*

Trait self-control

.02

0.21

Model

Results We tested the proposed interactions using moderated regression analyses, which were performed with the process module by Hayes (2013). To avoid multicollinearity, we meancentered all predictors before forming the product terms (Cohen, Cohen, West, & Aiken, 2003). In the first step, we introduced age and gender as potential covariates as well as the three predictors (job-related and academic demands on overcoming inner resistances, trait self-control). In the second step, all three two-way interactions were included into the equations. Finally, in the third step, the proposed three-way interaction was analyzed to test Hypothesis 1.

Table 1 shows means, standard deviations, intercorrelations, and reliabilities of all study variables. Table 2 provides results of regressions. After controlling for age and gender, job-related and academic SCDs exhibited significant positive relations to irritation, whereas trait self-control did not predict irritation. Finally, and consistent with Hypothesis 1, regression analysis revealed a significant three-way interaction of trait self-control and both SCDs on irritation (ΔR2 = 0.02; p < .05).

We conducted simple slope analysis and depicted the form of interaction (see Figure 1). Specifically, the interaction of academic and job-related SCDs was more pronounced when trait self-control was high. That is, in cases of high trait self-control, the relation of one stressor to irritation was insignificant, when the other stressor was also low, whereas the positive relationship was stronger in the presence of the other stressor. In contrast, in cases of low trait self-control, the positive relations of one stressor to irritation were positive and significant regardless of the level of the other stressor.

Discussion of Study 1 and Hypotheses Development for Study 2 According to our findings in Study 1, trait self-control will moderate the positive relationship between one source of Journal of Personnel Psychology (2019), 18(1), 23–33

Step 1 – Predictors

R2

.24

F

10.09**

Step 2 – Two-way interaction Job-Related Self-Control Demands Academic Self-Control Demands

.10

0.95

Job-related Self-Control Demands Trait Self-Control

.08

0.71

Academic Self-Control Demands Trait Self-Control

.02

0.17

ΔR2

.01

ΔF

0.37

Step 3 – Three-way interaction Job-Related Self-Control Demands Academic Self-Control Demands Trait Self-Control

.23

ΔR2

.02

ΔF

4.67*

2.16*

Note. Gender: 1 = male, 2 = female; age: 1 = 18–21 years, 2 = 22–25 years, 3 = 26–29 years, 4 = 30–33 years, 5 = 34–36 years, 6 = 37+ years. *p < .05, **p < .01.

SCDs and strain, if and only if the other demand is low.

Put differently, in cases of simultaneous coping with two stressors, trait self-control fails to prevent one from feeling strained. We thereby identified boundary conditions of the protective function of trait self-control. From a perspective of ISCT, the present interaction pattern is in line with the mechanistic notion that the trait component of the control capacity is only able to stabilize well-being when coping with one stressor. In other words, the trait component determines whether strain disproportionately increases only in cases of two stressors (high trait self-control) or even in Ó 2019 Hogrefe Publishing

K. Externbrink et al., The Limits of Trait Self-Control

27

Figure 1. Three-way interaction of trait self-control and academic and job-related SCDs on irritation (N = 163).

the presence of one stressor (low trait self-control), whereas, however, two stressors overtax the overall capacity during self-control exertion and inevitably result in disproportionately high strain.

Despite initial evidence for our proposition, our results are subject to several methodological limits. First, complex interactions are often contingent upon specific contextual circumstances and may not materialize in different samples (Shieh, 2007). Thus, we cannot rule out that the present interaction results from specific distributional characteristics of our sample. Second, and related to the first issue, our sample was exclusively restricted to fully employed students whose objective SCDs are probably quite homogeneous, because all of them hold full-time employment and are enrolled in an occupational study program. That is, our findings may not be generalizable to other populations, which involve full-time students with part-time jobs also facing different sources of stress. Third, although well-established measures of SCDs discriminate between occupational roles, work tasks, and professional groups with different levels of job requirements (Schmidt & Neubach, 2010), an indicator, which is largely unaffected by selfreport biases and directly assesses the amount of workload, may improve our design and provide more valid findings about the hypothesized interaction patterns.

To address these limitations, we conducted a second study with a more heterogeneous sample, which consists of students without parallel employment, with part-time employment and those who are fully employed. We introduced average study and working time per week (in hours) as more objective indicators (as compared to perceptual indicators of SCDs) for the degree of demands in both spheres (Valcour, 2007). According to a meta-analysis (Sparks, Cooper, Fried, & Shirom, 1997), working hours Ó 2019 Hogrefe Publishing

are positively related to strain and thus reflect the extent to which employees are potentially faced with stressful events at work and required to regulate themselves according to job demands. In addition, working hours have been found to impair work–life balance and interfere with duties in other life domains (Valcour, 2007). In a similar vein, we argue that study time per week indicates the degree of one’s entanglement in tasks, processes, and duties related to academic studies.

Consequently, we conceptualized study and working time per week as potential sources of SCDs, which tax the control capacity and thus interact with each other, and directly assessed SCDs as well as trait self-control in predicting psychological strain. In particular, regardless of the level of trait self-control, those with high study and working time per week should experience disproportionately high strain, when they are faced with high SCDs in the other domain. In contrast, in cases of low study and working time per week, trait self-control should attenuate the positive relationship of SCDs to strain. Thus, we propose the following additional hypotheses: Hypothesis 2: Trait self-control, job-related SCDs, and study time per week (in hours) interact in predicting psychological strain: The relationship of job-related SCDs to strain is weakest when the study time per week is low and trait self-control is high. In comparison, in all other cases (the study time per week is high and/or trait self-control is low), SCDs are stronger and positively related to strain.

Hypothesis 3: Trait self-control, academic SCDs, and working time per week (in hours) interact in predicting psychological strain: The relationship of academic Journal of Personnel Psychology (2019), 18(1), 23–33

28

K. Externbrink et al., The Limits of Trait Self-Control

SCDs to strain is weakest when the working time per week is low and trait self-control is high. In comparison, in all other cases (the working time per week is high and/or trait self-control is low), SCDs are stronger and positively related to strain.

Results Drawing from the analytical procedure in Study 1, we applied moderated regression analyses to test the threeway interactions between academic and job-related SCDs, trait self-control, and study time per week, as well as working time per week. That is, for each of the three hypotheses, we introduced the main effect variables (gender, age, jobrelated and academic SCDs, as well as trait self-control) in the first step, whereas all two-way interactions and the three-way interaction were integrated in the second and third steps, respectively.

Table 3 displays descriptive statistics, intercorrelations, and reliabilities, while the results of the moderated regression analyses are depicted in Table 4. Whereas job-related SCDs and trait self-control were significantly associated with irritation with signs corresponding to expectations, academic SCDs failed to exhibit positive relations to irritation. In line with prior studies, both study and working time per week significantly predicted irritation. In the final steps, only the interaction between job-related SCDs, study time per week, and trait self-control explained significant proportions in variance of irritation over and beyond that accounted for by the main and two-way interaction effects (Hypothesis 2). The incremental amount of explained variance was 7% (p < .01). The other three-way interactions were insignificant in predicting irritation (Hypotheses 1, 3, and 4).

As in Study 1, on the basis of the simple slope method, we analyzed the form of the significant three-way interaction between job-related SCDs, study time per week, and trait self-control (Figure 2): When trait self-control was high, the positive relations of SCDs to irritation were amplified as a function of study time per week, whereas in cases of low trait self-control, study time per week attenuated the positive relation of job-related SCDs to irritation. In particular, SCDs were significantly and positively associated with irritation, when either trait self-control and study time per week were high or both were low. In conclusion, and consistent with Hypothesis 2, those with high trait self-control did not report an increase in irritation with increasing

Hypothesis 4: Trait self-control, study times per week, and working time per week (in hours) interact in predicting psychological strain: The relationship of study times per week to strain is weakest when the working time per week is low and trait self-control is high. In comparison, in all other cases (the working time per week is high and/or trait self-control is low), study times per week are stronger and positively related to strain.

Study 2 Method Again, we recruited participants from two German universities of applied sciences via mailings and during lectures. In contrast to Study 1, the participants were either fully employed (39.1%) or partially employed (32.3%) or not formally employed, but may have a side job (28.6%). In sum, a final sample of 135 students who were studying for a bachelor or master degree provided data for all relevant study variables. Participants were between 18 and 35 years of age (M = 23.73, SD = 3.68), and 67.4% were women.

We assessed irritation, trait self-control, job-related and academic SCDs on the basis of the same measures as in Study 1. Again, according to CFA, both measures of SCDs seem to reflect different constructs [2-factor model: w2(34) = 95.63, p < .01, RMSEA = .116, CFI = .916, SRMR = .047; 1-factor model: w2(35) = 781.08, p < .01, RMSEA = .397, CFI = .000, SRMR = .187]. In addition, to assess study and working time per week, we asked the participants the following questions: “On average, how many hours do you study in a typical week?” and “On average, how many hours do you work in a typical week?” (see also Valcour, 2007).

Table 3. Descriptive results, α reliabilities, and intercorrelations (Study 2) M

SD

α

1. Irritation

3.30

1.34

.90

2. Gender

1.67

0.47

–

23.73

3.68

–

.04

.07

4. Job-related self-control demands

2.86

1.00

.91

.36**

.01

5. Academic self-control demands

3.29

0.96

.89

.15

.10

.04

.35**

6. Trait self-control

3.11

0.63

.83

.32**

.07

.06

.41**

.41**

7. Working time per week (in hr)

24.51

17.36

–

.04

.13

.40**

.01

.16

.03

8. Study time per week (in hr)

22.33

15.99

–

.20*

.05

.39**

.00

.07

.05

3. Age

1.

2.

3.

4.

5.

6.

7.

.07 .13

.52**

Note. Gender: 1 = male, 2 = female. *p < .05, **p < .01.

Journal of Personnel Psychology (2019), 18(1), 23–33

Ó 2019 Hogrefe Publishing

K. Externbrink et al., The Limits of Trait Self-Control

29

Table 4. Results of multiple regression analysis for predicting irritation (Study 2) Hypothesis 1

Hypothesis 2

Hypothesis 3

Hypothesis 4

β

t

β

t

β

t

β

t

Gender

.10

1.29

.11

1.41

.11

1.36

.13

1.62

Age

.02

0.16

.03

0.31

.02

0.18

.03

0.35

Working time per week (in hr)

.24

2.42**

.30

3.16**

.28

2.52**

.33

3.17**

Study time per week (in hr)

.32

3.35**

.46

4.64**

.33

3.49**

.40

3.70**

Job-related self-control demands

.31

3.19**

.24

2.76**

.25

2.70**

.27

3.02**

Academic self-control demands

.10

1.06

.10

1.09

.07

0.76

.08

0.81

.22

2.31**

.21

2.40**

.25

2.72**

.35

2.86**

Model Step 1 – Predictors

Trait self-control R2

.26

.26

.26

.26

F

6.22**

6.22**

6.22**

6.22**

Step 2 – Two-way interaction Job-Related Self-Control Demands Academic Self-Control Demands

.18

1.68

Job-Related Self-Control Demands Trait SelfControl

.02

0.23

Academic Self-Control Demands Trait SelfControl

.13

1.18

.03

0.36 .06

0.66

Working Time Per Week Academic Self-Control Demands

.09

0.93

Working Time Per Week Trait Self-Control

.04

0.45

Study Time Per Week Job-Related Self-Control Demands

.02

0.24

Study Time Per Week Trait Self-Control

.06

0.72

Working Time Per Week Study Time Per Week

.03

.29

.04

.30

.13

1.30

ΔR2

.02

.02

.00

.01

ΔF

0.87

0.86

0.11

0.80

Step 3 – Three-way interaction Job-Related Self-Control Demands Academic Self-Control Demands Trait Self-Control

.09

0.85

Job-Related Self-Control Demands Study Time Per Week Trait Self-Control

.32

Academic Self-Control Demands Working Time Per Week Trait Self-Control

3.56** .01

Study Times Per Week Working Time Per Week Trait Self-Control

0.10 .16

ΔR2

.004

0.07

.00

.01

ΔF

0.729

13.70**

0.00

1.48

1.22

Note. Gender: 1 = male, 2 = female. *p < .05, **p < .01.

job-related SCDs, when they did not spend much time in academic study. In all other cases, job-related SCDs resulted in strain or strain was constantly high.

General Discussion In the present research, we examined the interplay of different kinds of stressors and trait self-control in predicting psychological strain. In exploring the boundary conditions of the protective function of trait self-control, we tested moderating effects of trait self-control on the positive Ó 2019 Hogrefe Publishing

relationships of two different simultaneously occurring stressors to irritation. Drawing from ISCT (Kotabe & Hofmann, 2015), we proposed that trait self-control enables successful coping with only one stressor, whereas in cases of multiple stressors even those with high trait self-control will experience disproportionately high strain. On the basis of our proposition, we derived hypotheses on interaction effects of trait self-control with job-related and academic SCDs as well as study and working times per week as potential sources of stress.

In the first sample which consisted of occupational students enrolled in an occupational study program, trait Journal of Personnel Psychology (2019), 18(1), 23–33

30

K. Externbrink et al., The Limits of Trait Self-Control

Figure 2. Three-way interaction between job-related SCDs, study time per week, and trait self-control (N = 135).

self-control buffered the positive relationship of job-related SCDs to strain when academic SCDs were low, whereas in cases of low trait self-control or high academic SCDs, jobrelated SCDs were strongly and positively related with strain. In the second sample which was more heterogeneous and involved students who were fully or partially or not formally employed, we sought to replicate the interaction of both SCDs and trait self-control on strain. In addition, we extended the spectrum of potential stressors by including study and working times per week as demands, which are less contaminated by self-report biases. We found an interaction of trait self-control, job-related SCDs, and study hours per week in predicting irritation: For individuals with low trait self-control, job-related SCDs were positively associated with irritation when study times per week were low, whereas in cases of high study times per week, irritation was constantly high. In contrast, under conditions of high trait self-control, both stressors mutually amplified each other in their effects on irritation. However, the other interaction effects – even the interaction of both SCDs and trait self-control of Study 1 – failed to reach significance. In sum, given that two of five analyses were in line with our hypotheses, our data thus provided mixed support for our propositions.

In an attempt to interpret the significant interactions, we believe that in general, in cases of low trait self-control, the control capacity should be more rapidly depleted when one (occupational or academic) domain is demanding. That is, regardless of one or two stressors, strain increases with increasing levels of demands. In comparison, in cases of high trait self-control, the capacity is only depleted by two sources of stress, while those with high trait self-control are able to put high effort into coping with one stressor Journal of Personnel Psychology (2019), 18(1), 23–33

when the other is low. Thus, the relationship of one stressor to strain is amplified by the other stressor when trait selfcontrol is high. In our studies, such an interaction pattern did not emerge for all combinations of SCDs and study or working times per week. To some extent, both significant interactions indicate that even those who are well able to resist distractions, override their response tendencies or overcome inner motivational resistances and thus exhibit high self-discipline experience high psychological strain in the presence of multiple stressors. While most empirical evidence indicates that trait self-control prevents strain and self-control deficits in cases of high stress, we were able to identify the limits of the protective function of trait selfcontrol in coping with sources of stress. In the following, we elaborate on the implications of our findings.

Theoretical Implications We see two main theoretical implications. Firstly, past research repeatedly revealed interactions of two stressors, which are hypothesized to put high demands on self-control and mutually amplified each other in their deleterious effects on psychological strain (e.g., Diestel & Schmidt, 2011; Zapf, Seifert, Schmutte, Mertini, & Holz, 2001).

According to ISCT, given that the control capacity is limited, simultaneous coping with two stressors, which draw on the same capacity, produces higher levels of strain than the sum of their main effects. In extending our knowledge about the effects of multiple stressors, the present interaction patterns show that multiple stressors may not only lead to disproportionately high strain, but also impede the usage of protective resources for coping with stress. In line with Ó 2019 Hogrefe Publishing

K. Externbrink et al., The Limits of Trait Self-Control

the job demandsâ&#x20AC;&#x201C;resources model (Bakker & Demerouti, 2007), which disentangles processes of strain, as well as motivation, and proposes that demands can attenuate the beneficial effects of resources on well-being, the simultaneous occurrence of multiple stressors can constitute a boundary condition under which personal resources fail to reduce strain.

Secondly, as noted above, we could not replicate the interaction effect between trait self-control, job-related SCDs, and academic SCDs on irritation in the second sample. In addition, the two other interactions with different combinations also did not emerge. Therefore, the present findings from both studies are not consistent and should be interpreted tentatively. On the one hand, the presence of two high SCDs in different contexts might have been differentially experienced: that is, whereas occupational students (Sample 1) may have suffered more from SCDs because of, for example, dysfunctional or straining time schedules (e.g., lectures after working time, working and studying in the evening, more complicated coordination with others), regular students probably benefit from more autonomy and thus are better able to cope with overcoming inner resistances. On the basis of additional subgroup analyses in Sample 2, we found a stronger (but insignificant) three-way interaction of both SCDs and trait self-control with signs corresponding to expectations for those who have at least a part-time job (as compared to those who were not formally employed). We interpret this finding as an indication for the argument that the perception of two simultaneously occurring SCDs may somewhat differ between types of employment.

On the other hand, study time per week does not directly reflect the extent to which individuals are required to engage in volitional self-control. However, Prem et al.

(2016) have revealed mediating effects of SCDs in the positive relations of workload, as well as time pressure, to indicators of strain, and thus demonstrated that dysfunctional conditions such as quantitative demands exert their deleterious effects on strain through exerting volitional self-control, which taxes the control capacity. That is, at least to some extent, study or working time per week can put high demands on volitional self-control. In light with the strong relationship of job-related SCDs to irritation in Study 2, study times per week may have caused employees to engage in volitional self-control (even more than academic SCDs or working times per week), because they had to cope with SCDs at work, were already depleted or strained, and thus perceived high study times as a SCD, which interact other SCDs and trait self-control. Since the form of both significant interaction effects is somewhat similar, we argue that the same underlying psychological mechanism, which relies on a dysfunctional interplay between the limited control capacity and additional enactment constraints, materiĂ&#x201C; 2019 Hogrefe Publishing

31

alizes in both interactions. The extent to which potential sources of stress require high volitional self-control might depend upon the specific occupational setting. Whereas the present results are somewhat strongly suggestive, but not highly convincing in terms of an exact replication, however, both significant interactions indicate that trait self-control can fail to prevent irritation when coping with two stressors.

Limitations Our study has limitations. Firstly, previous research has manipulated state self-control in laboratory settings by varying (more or less objectively) SCDs. Field experiments on academic SCDs use examination periods as a dichotomous indicator for high SCDs and therefore may suffer from variance restrictions. In contrast, we measured academic and job-related SCDs as continuous variables, because such measures capture natural variation in perceptions of demands.

Secondly, our results are based upon cross-sectional selfreport data and do not allow causal conclusions in a strict sense. However, because our analyses draw from two samples and we employed different operationalizations of the stress variables, we can provide some confidence in our findings. Notably, in Study 2, we used more objective indicators of sources of stress (compared to the self-report SCDs-measures), which reflect the amount of time spent in the academic and job domains and thus strengthen our design. Notwithstanding, experience event sampling or longitudinal studies may provide additional evidence for our propositions.

Thirdly, we only focused on irritation as an indicator of psychological strain, which is typically more broadly conceptualized compared to emotional and cognitive symptoms (Sonnentag & Frese, 2013). However, we believe that irritation particularly indicates deficits in the enactment phase of exercising self-control (Lian et al., 2017), because its symptoms refer to perceived deficits in detachment and behavioral failures in exerting self-control (e.g., anger). Nevertheless, future research may also consider other indicators of strain, such as exhaustion or even physiological measures (Ilies, Aw, & Lim, 2016).

Fourthly, given the complexity of our results, the present sample sizes do not seem large enough to have strong confidence in them. However, in light of simulation studies (Dawson & Richter, 2006), our sample sizes provide a sufficient basis for taking our results seriously.

Fifthly, our measures and results do not necessarily show whether the level of experienced strain is accounted for by two different kinds of stressors, which require volitional self-control, in both settings or might be due to an additional source of stress in the same setting. For example, Journal of Personnel Psychology (2019), 18(1), 23â&#x20AC;&#x201C;33

32

in Sample 1, some of the participants might have high jobrelated SCDs at a level that would be experienced if they were working a second job in the evening (instead of studying). Such distributional patterns imply quadratic effects of SCDs on strain: In both studies, however, additional analyses failed to provide evidence of nonlinear relations of SCDs to irritation. Nevertheless, the question of perceptual biases in self-report assessments of specific demands still remains somewhat methodologically unanswered (Spector, Fox, & Van Katwyk, 1999).

Practical Implications The present findings may have good applicability in advising individuals about whether or not they can handle two large commitments (work and study) simultaneously. We see self-regulation as a central issue for career counseling of occupational students. Universities, for example, may provide support for students who are working parallel to their academic studies through stress-reduction programs, psychological counseling, or self-management trainings.

Additionally, online-based self-assessments, which provide differential feedback about one’s psychological wellbeing, could include SCDs, trait self-control, and other relevant occupational characteristics (e.g., workload). Based on such feedback, employees can better decide whether they are capable of enrolling in an occupational study program without taking much risk of psychological strain.

Conversely, students may also benefit from such selfassessments when they face multiple demands due to (side) jobs. An example for such self-assessments is provided by Kallus and Kellmann (2015) who developed and validated a questionnaire for recovery and stress.

Finally, to prevent high SCDs in the academic context, structural support may be facilitated by blended learning, flexible scheduling, and autonomy in selecting subjects, as well as examinations. Integrated learning methods (such as blended learning) may facilitate self-regulation and well-being through the usage of various online and offline tools, which reduce time pressure and enlarge autonomy (Rossett, 2002).

In sum, our findings accentuate the importance of focusing on potentially stressful conditions for those who are faced with multiple demands in different life domains. Individual characteristics or personal resources may not always provide sufficient protection, especially when demands exceed one’s capacity for self-control.

References Bakker, A. B., & Demerouti, E. (2007). The job demands–resources model: State of the art. Journal of Managerial Psychology, 22, 309–328. https://doi.org/10.1108/02683940710733115

Journal of Personnel Psychology (2019), 18(1), 23–33

K. Externbrink et al., The Limits of Trait Self-Control

Beedie, C.J., & Lane, A.M. (2012). The role of glucose in selfcontrol: Another look at the evidence and an alternative conceptualization. Review Personality and Social Psychology, 16, 143–153. https://doi.org/10.1177/1088868311419817 Bertrams, A., & Dickhäuser, O. (2009). Messung dispositioneller Selbstkontroll-Kapazität [Measuring dispositional self-control capacity]. Diagnostica, 55, 2–10. https://doi.org/10.1026/00121924.55.1.2 Carter, E. C., Kofler, L. M., Forster, D. E., & McCullough, M. E. (2015). A series of meta-analytic tests of the depletion effect: Self-control does not seem to rely on a limited resource.

Journal of Experimental Psychology: General, 144, 796–815. https://doi.org/10.1037/xge0000083 Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum.

Dang, J. (2018). An updated meta-analysis of the ego depletion effect. Psychological Research, 82, 645–651. https://doi.org/ 10.1007/s00426-017-0862-x Dawson, J., & Richter, A. W. (2006). Probing three-way interactions in moderated multiple regression. Journal of Applied Psychology, 91, 917–926. https://doi.org/10.1037/0021-9010.

91.4.917 de Ridder, D. T., Lensvelt-Mulders, G., Finkenauer, C., Stok, F. M., & Baumeister, R. F. (2012). Taking stock of self-control: A meta-analysis of how trait self-control relates to a wide range of behaviors. Personality and Social Psychology Review, 16, 76–99. https://doi.org/10.1177/1088868311418749 Diestel, S., & Schmidt, K.-H. (2011). Costs of simultaneous coping with emotional dissonance and self-control demands at work.

Journal of Applied Psychology, 96, 643–653. https://doi.org/ 10.1037/a0022134 Diestel, S., Rivkin, W., & Schmidt, K.-H. (2015). Sleep quality and self-control capacity as protective resources in the daily emotional labor process. Journal of Applied Psychology, 100, 809–827. https://doi.org/10.1037/a0038373 Hagger, M. S., Wood, C., Stiff, C., & Chatzisarantis, N. L. D. (2010).

Ego depletion and the strength model of self-control: A metaanalysis. Psychological Bulletin, 136, 495–525. https://doi.org/ 10.1037/a0019486 Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis. New York, NY: Guilford Press.

Hülsheger, U. R., & Schewe, A. F. (2011). On the costs and benefits of emotional labor: A meta-analysis of three decades of research. Journal of Occupational Health Psychology, 16, 361–389. https://doi.org/10.1037/a0022876 Ilies, R., Aw, S. S., & Lim, V. K. (2016). A naturalistic multilevel framework for studying transient and chronic effects of psychosocial work stressors on employee health and wellbeing. Applied Psychology, 65, 223–258. https://doi.org/ 10.1111/apps.12069 Kallus, K. W., & Kellmann, M. (2015). Recovery-stress questionnaires. Frankfurt am Main, Germany: Pearson.

Kotabe, H. P., & Hofmann, W. (2015). On integrating the components of self-control. Perspectives on Psychological Science, 10, 618–638. https://doi.org/10.1177/1745691615593382 Lian, H., Yam, K. C., Ferris, D. L., & Brown, D. (2017). Self-control at work. Academy of Management Annals, 11, 703–732. https:// doi.org/10.5465/annals.2015.0126 Lian, H., Brown, D. J., Ferris, D. L., Liang, L. H., Keeping, L. M., & Morrison, R. (2014). Abusive supervision and retaliation: A selfcontrol framework. Academy of Management Journal, 57, 116–139. https://doi.org/10.5465/amj.2011.0977 Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex “frontal

Ó 2019 Hogrefe Publishing

K. Externbrink et al., The Limits of Trait Self-Control

lobe” tasks: A latent variable analysis. Cognitive Psychology, 41, 49–100. https://doi.org/10.1006/cogp.1999.0734 Mohr, G., Rigotti, T., & Mueller, A. (2005). Irritation – ein Instrument zur Erfassung psychischer Beanspruchung im Arbeitskontext. Skalen- und Itemparameter aus 15 Studien [Irritation – an instrument assessing mental strain in working contexts. Scale and item parameters from 15 studies].

Zeitschrift für Arbeits- und Organisationspsychologie, 49, 44–48. https://doi.org/10.1026/0932-4089.49.1.44 Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of limited resources: Does self-control resemble a muscle? Psychological Bulletin, 126, 247–259. https://doi.org/ 10.1037/0033-2909.126.2.247 Oaten, M., & Cheng, K. (2005). Academic examination stress impairs self-control. Journal of Social and Clinical Psychology, 24, 254–279. https://doi.org/10.1521/jscp.24.2.254.62276 Oaten, M., & Cheng, K. (2006). Improved self-control: The benefits of a regular program of academic study. Basic and Applied Social Psychology, 28, 1–16. https://doi.org/10.1207/ s15324834basp2801_1 Prem, R., Kubicek, B., Diestel, S., & Korunka, C. (2016). Regulatory job stressors and their within-person relationships with ego depletion: The role of state anxiety, self-control effort, and job autonomy. Journal of Vocational Behavior, 92, 22–32. https:// doi.org/10.1016/j.jvb.2015.11.004 Prem, R., Paškvan, M., Kubicek, B., & Korunka, C. (2018). Exploring the ambivalence of time pressure in daily working life. International Journal of Stress Management, 25, 35–43. https://doi. org/10.1037/str0000044 Rossett, A. (2002). The ASTD e-learning handbook. New York, NY: McGraw-Hill.

Schmidt, K.-H., & Diestel, S. (2015). Self-control demands: From basic research to job-related applications. Journal of Personnel Psychology, 14, 49–60. https://doi.org/10.1027/1866-5888/ a000123 Schmidt, K.-H., Hupke, M., & Diestel, S. (2012). Does dispositional capacity for self-control attenuate the relation between selfcontrol demands at work and indicators of job strain? Work & Stress, 26, 21–38. https://doi.org/10.1080/02678373.2012.660367 Schmidt, K.-H., & Neubach, B. (2010). Selbstkontrollanforderungen bei der Arbeit – Fragebogen zur Erfassung eines bislang wenig beachteten Belastungsfaktors [Self-control demands – Questionnaire for measuring a job stressor largely neglected thus far]. Diagnostica, 56, 133–143. https://doi.org/10.1026/ 0012-1924/a000015 Schmitt, A., Zacher, H., & Frese, M. (2012). The buffering effect of selection, optimization, and compensation strategy use on the relationship between problem solving demands and occupational well-being: A daily diary study. Journal of Occupational Health Psychology, 17, 139–149. https://doi.org/10.1037/ a0027054 Shieh, G. (2007). A unified approach to power calculation and sample size determination for random regression models.

Psychometrika, 72, 347–360. https://doi.org/10.1007/s11336007-9012-5 Shoda, Y., Mischel, W., & Peake, P. K. (1990). Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification. Developmental Psychology, 26, 978–986. https://doi.org/10.1037/0012-1649.26.6.978 Sonnentag, S., & Frese, M. (2013). Stress in organizations. In W. C. Borman, D. R. Ilgen, & R. J. Klimoski (Eds.), Comprehensive handbook of psychology (Vol. 12, pp. 453–491). Hoboken, NJ: Wiley.

Sparks, K., Cooper, C., Fried, Y., & Shirom, A. (1997). The effects of hours of work on health: A meta-analytic review. Journal of Occupational & Organizational Psychology, 70, 391–408. https://doi.org/10.1111/j.2044-8325.1997.tb00656.x

Ó 2019 Hogrefe Publishing

33

Spector, P. E., Fox, S., & Van Katwyk, P. T. (1999). The role of negative affectivity in employee reactions to job characteristics: Bias effect or substantive effect? Journal of Occupational and Organizational Psychology, 72, 205–218. https://doi.org/ 10.1348/096317999166608 Stoeber, J., Childs, J. H., Hayward, J. A., & Feast, A. R. (2011).

Passion and motivation for studying: Predicting academic engagement and burnout in university students. Educational Psychology, 31, 513–528. https://doi.org/10.1080/01443410.

2011.570251 Stumm, S., Thomas, E., & Dormann, C. (2010). Selbstregulationsstärke und Leistung–Dualer Prädiktor im dualen Hochschulstudium [Self-regulatory strength and performance: Dual predictor in cooperative university education]. Zeitschrift für Arbeits- und Organisationspsychologie A&O, 54, 171–181. https://doi.org/10.1026/0932-4089/a000029 Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High selfcontrol predicts good adjustment, less pathology, better grades, and interpersonal success. Journal of Personality, 72, 271–322. https://doi.org/10.1111/j.0022-3506.2004.00263.x Thomas, J., & Mengel, T. (2008). Preparing project managers to deal with complexity – advanced project management education. International Journal of Project Management, 26, 304–315. https://doi.org/10.1016/j.ijproman.2008.01.001 Valcour, M. (2007). Work-based resources as moderators of the relationship between work hours and satisfaction with work– family balance. Journal of Applied Psychology, 92, 1512–1523. https://doi.org/10.1037/0021-9010.92.6.1512 Vohs, K. D., & Faber, R. J. (2007). Spent resources: Self-regulatory resource availability affects impulse buying. Journal of Consumer Research, 33, 537–547. https://doi.org/10.1086/510228 Yam, K. C., Fehr, R., Keng-Highberger, F. T., Klotz, A. C., & Reynolds, S. J. (2016). Out of control: A self-control perspective on the link between surface acting and abusive supervision.

Journal of Applied Psychology, 101, 292–301. https://doi.org/ 10.1037/apl0000043 Zapf, D., Seifert, C., Schmutte, B., Mertini, H., & Holz, M. (2001).

Emotion work and job stressors and their effects on burnout.

Psychology and Health, 16, 527–545. https://doi.org/10.1080/ 08870440108405525 Zettler, I. (2011). Self-control and academic performance: Two field studies on university citizenship behavior and counterproductive academic behavior. Learning and Individual Differences, 21, 119–123. https://doi.org/10.1016/j.lindif.2010.11.002 History Received April 30, 2017 Revision received May 21, 2018 Accepted June 6, 2018 Published online January 9, 2019 Authorship Kai Externbrink and Stefan Diestel contributed equally to the manuscript and share first authorship.

ORCID Kai Externbrink https://orcid.org/0000-0001-9600-3393 Kai Externbrink Department of Business and Psychology FOM University of Applied Sciences Lissaboner Allee 7 44269 Dortmund Germany kai.externbrink@fom.de

Journal of Personnel Psychology (2019), 18(1), 23–33

Original Article

Perceived Overqualification and Psychological Well-Being Among Immigrants The Moderating Role of Personal Resources Maria Wassermann

and Annekatrin Hoppe

Department of Psychology, HU Berlin, Germany

Abstract: Migration is often driven by immigrants’ hope of improving their job situation. However, in the host country, they are at risk of holding jobs below their qualifications. This study examines the relationship between perceived overqualification and psychological well-being (depressive symptoms and life satisfaction) among 176 Italian immigrants in Germany along with the buffering role of optimism and meaningmaking. The results show that perceived overqualification is associated with higher levels of depressive symptoms and lower levels of life satisfaction. Optimism moderates the relationship between perceived overqualification and life satisfaction: the relationship is attenuated with increasing optimism. We conclude that interventions that enhance optimism could help immigrants cope with perceived overqualification.

Keywords: perceived overqualification, psychological well-being, optimism, meaning-making, immigrants

Overqualification is a form of lower-quality employment in which people are unable to fully utilize their education, experience, knowledge, or skills (Maynard, Joseph, & Maynard, 2006). Perceived overqualification refers to the awareness of a person regarding the mismatch of their qualifications and job requirements (Fine & Nevo, 2008).

Previous research has demonstrated that perceived overqualification can be negatively related to psychological well-being (for a meta-analysis, see Harari, Manapragada, & Viswesvaran, 2017).

Migration is often driven by immigrants’ hope of improving their job situation and career prospects (Bartolini, Gropas, & Triandafyllidou, 2017). However, in the host country, immigrants have a particularly high risk for overqualification (e.g., Crollard, de Castro, & Tsai, 2012; Frank & Hou, 2017). Whereas skill utilization at work can help immigrants to achieve full professional development, social recognition, and self-actualization in the host country (Adler, 1977; Aycan & Berry, 1996), overqualification makes career progression and integration into the host country more arduous. There is some evidence for a negative association between overqualification and psychological wellbeing in immigrant samples (e.g., Chen, Smith, & Mustard, 2010; Dean & Wilson, 2009; Frank & Hou, 2017). As the migration of skilled workers has substantially increased in the last decades (Seibert & Wapler, 2012), research needs Journal of Personnel Psychology (2019), 18(1), 34–45 https://doi.org/10.1027/1866-5888/a000219

to address this population more profoundly and identify protective factors that buffer the negative relationship between overqualification and psychological well-being.

Accordingly, the present study examines personal resources as protective factors with regard to the relationship between perceived overqualification and indicators of psychological well-being among immigrants. More specifically, we investigate whether optimism and meaning-making buffer the negative relationship between perceived overqualification with depressive symptoms and life satisfaction, respectively.

This study contributes to the literature in two ways: first, building on the person–environment (P-E) fit theory (Edwards, Caplan, & Van Harrison, 1998; Kristof-Brown, Zimmerman, & Johnson, 2005), it broadens our knowledge about the association between perceived overqualification and psychological well-being among a specific sample of immigrant workers. Thus, with this study, we respond to the need for more population-specific research on perceived overqualification (Erdogan, Bauer, Peiró, & Truxillo, 2011; Liu & Wang, 2012). By investigating immigrant workers, we focus on a group that is particularly at risk of perceiving overqualification following a critical life event and, thus, in need of interventions that enhance protective factors.

Second, by drawing on the conservation of resources (COR) theory (Hobfoll, 1989), we investigate optimism Ó 2019 Hogrefe Publishing

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

and meaning-making as potential protective factors for immigrants who perceive overqualification. In doing so, we respond to the call for more research on moderators of perceived overqualification (e.g., Erdogan et al., 2011).

The identification of personal resources as potentially protective factors is relevant for workers and organizations, as personal resources can be enhanced and trained (Luthans, Avey, Avolio, Norman, & Combs, 2006) and thus provide an interesting starting point for interventions in the work setting.

Theoretical Background Perceived Overqualification and Psychological Well-Being Overqualification is a poor fit between a person and his or her job. It constitutes a demands–abilities misfit, because a person’s qualifications exceed his or her job demands. Further, it constitutes a needs–supplies misfit, because a person’s need to do an intrinsically motivating job and utilize his or her skills is poorly satisfied (Liu, Luksyte, Zhou, Shi, & Wang, 2015). To maintain well-being, people strive toward a good fit with their job. A person–job fit provides optimal conditions for workers to pursue their interests and perform their work (Lanivich, Brees, Hochwarter, & Ferris, 2010). It, thereby, can enhance their resources, such as professional success and advancement, adequate income, status at work, and acknowledgment of their professional accomplishments (see Hobfoll, 2001). On the contrary, a person–job misfit such as overqualification can threaten existing resources (e.g., by causing a loss of skills; De Grip, Bosma, Willems, & Van Boxtel, 2008; Desjardins & Rubenson, 2011) and make access to future resources more difficult (e.g., access to better job positions or a higher salary; Luksyte, Spitzmueller, & Maynard, 2011). Thus, according to P-E fit theory, a misfit such as overqualification is a stressful experience that impairs psychological well-being (Edwards et al., 1998; Kristof-Brown et al., 2005).

Apart from an objective person–job misfit, also a subjectively perceived misfit by the person, such as the perception of being overqualified for one’s job, can be related to poor psychological well-being. It is possible that in some cases, overqualification is a conscious choice, especially if it leads to intermediate resource gain or protection (e.g., the benefit of having a job at all; see Erdogan et al., 2011). However, the majority of research indicates that it is a threat for resources, as perceived overqualification is linked to impaired psychological well-being (Harari et al., 2017).

In this study, we investigate the association between a person–job misfit, in terms of perceived overqualification, Ó 2019 Hogrefe Publishing

35

and two distinct indicators of psychological well-being: First, we consider depressive symptoms, including feelings of sadness, loneliness, despair, and dejection (American Psychiatric Association, 2013). Depressive symptoms can be an indication of a serious health impairment, which is of high importance for organizations as it can lead to long-term sickness absence and create substantial costs (Goldberg & Steury, 2001). Depressive symptoms can be especially harmful for non-natives, as their inclination to utilize professional help is lower (Lindert, Schouler-Ocak, Heinz, & Priebe, 2008). Second, we consider life satisfaction, which is a conscious judgment about the quality of one’s life in general based on individual criteria (Pavot & Diener, 1993). Migration is often motivated by the desire of individuals to improve their quality of life and happiness (Nowok, Van Ham, Findlay, & Gayle, 2013). Therefore, life satisfaction is a highly relevant outcome for immigrants.

Previous research has mainly shown an association between perceived overqualification and job-related outcomes, such as job satisfaction (e.g., Maynard et al., 2006; Wu, Luksyte, & Parker, 2015), withdrawal behavior (e.g., Erdogan & Bauer, 2009; Maynard & Parfyonova, 2013), and counterproductive work behavior (Luksyte et al., 2011). Compared with research on job-related outcomes, research on the association between perceived overqualification and indicators of general psychological well-being is scarce. There is some evidence that perceived overqualification is positively associated with depressive symptoms (Johnson & Johnson, 1992, 1996), whereas the opportunity to utilize skills is negatively associated with depressive symptoms (Griffin, Greiner, Stansfeld, & Marmot, 2007).

More specifically, with regard to our study population, Chen and colleagues showed that overqualification was related to persistent feelings of sadness, depression, and loneliness among immigrants (Chen et al., 2010). However, they applied objective measures of overqualification.

Similarly, perceived overqualification has been associated with lower levels of life satisfaction: Roh and colleagues demonstrated that life satisfaction of workers who perceive overqualification was as low as life satisfaction among unemployed people (Roh, Chang, Kim, & Nam, 2014). With regard to our study population, George and colleagues investigated internationally trained engineers and showed that life satisfaction was lower among those who held a job outside of the engineering field and who were therefore likely to perceive overqualification (George, Chaze, Fuller-Thomson, & Brennenstuhl, 2012). Frank and Hou (2017) showed a negative relationship between an objective measure of overqualification and life satisfaction among immigrants. However, to our knowledge, no studies have specifically looked into the relationship of perceived overqualification and life satisfaction among immigrants.

Journal of Personnel Psychology (2019), 18(1), 34–45

36

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

Building on the P-E fit theory (Edwards et al., 1998; Kristof-Brown et al., 2005) and prior research, we assume an adverse relationship between perceived overqualification and depressive symptoms and life satisfaction among immigrants.

Hypothesis 1: Perceived overqualification is positively related to depressive symptoms.

Hypothesis 2: Perceived overqualification is negatively related to life satisfaction.

Personal Resources as Moderators Resources are defined as valued objects, conditions, energies, and personal characteristics (Hobfoll, 2001). COR theory (Hobfoll, 1989) states that people strive to build resources and feel stressed when resources are threatened or lost. As argued above, overqualification can be a threat for resources as unused skills are at risk of becoming obsolete (De Grip et al., 2008; Desjardins & Rubenson, 2011) and access to adequate job positions and salary is limited (Luksyte et al., 2011). Furthermore, COR theory posits that people must invest resources in order to recover from previous resource loss or to interrupt an ongoing negative spiral of resource loss. Previous research on perceived overqualification has shown that the relationship between perceived overqualification and well-being can be moderated by resources, such as empowerment (Erdogan & Bauer, 2009), emotional support (Johnson & Johnson, 1997), and job autonomy (Wu et al., 2015). With this study, we aim to expand this research and identify personal resources as moderators in the interplay between perceived overqualification and psychological well-being.

Personal resources are defined as aspects of the self that enhance resiliency (Hobfoll, Johnson, Ennis, & Jackson, 2003) and “motivate and facilitate goal-attainment even in the face of adversity or challenge” (Van den Heuvel, Demerouti, Bakker, & Schaufeli, 2010, p. 129). Immigration is a critical life event that is accompanied by a loss of resources, such as the loss of social networks and support systems immigrants had in their home country (Schwarzer, Hahn, & Schröder, 1994). Therefore, personal resources may be especially important for immigrants perceiving overqualification.

Hobfoll (2011) argues that people with more resources (e.g., personal resources) can make use of these resources when facing a stressful situation (e.g., perceived overqualification) and thus are less vulnerable to resource loss.

Accordingly, we claim that immigrants who have higher levels of personal resources cope better with perceived overqualification. More specifically, we expect that the per-

Journal of Personnel Psychology (2019), 18(1), 34–45

sonal resources optimism and meaning-making can buffer the adverse relationship between perceived overqualification and immigrants’ psychological well-being. Van den Heuvel and colleagues (2010) argue that both optimism and meaning-making can be important resources for workers to deal with job stressors or challenges at work.

Optimism is the generalized expectancy that positive things will happen in the future and that things will turn out well (Mäkikangas, Feldt, Kinnunen, & Mauno, 2013; Scheier & Carver, 1985). Previous research has provided evidence that optimism is directly related to enhanced psychological well-being, including lower levels of depressive symptoms (e.g., Ruthig, Haynes, Stupnisky, & Perry, 2009; Scheier & Carver, 1992; Wong & Lim, 2009) and higher life satisfaction (e.g., Daukantaite & Zukauskiene, 2012; Hayes & Weathington, 2007; Wong & Lim, 2009).

Whereas these direct relationships are well-established, the potential of optimism as moderator of the relationship between perceived overqualification and depressive symptoms and life satisfaction is less clear. We assume that optimism acts as a buffer based on the following rationale.

Optimists are aware of the bright side of things. When perceiving overqualification, they are likely to use their positive state of mind to focus on benefits (e.g., having excess energy for nonwork demands such as learning the host country language; see Erdogan et al., 2011) and to identify resources that can help them to cope with the adversity of the situation (e.g., social support), resulting in lower levels of stress and higher well-being (Van den Heuvel et al., 2010). Furthermore, optimists actively take up and maintain action to deal with adverse conditions instead of denying them (Aspinwall & Taylor, 1997; Nes & Segerstrom, 2006). We therefore assume that, even when they perceive overqualification, they strive more persistently toward their career goals than pessimists (see Fugate & Kinicki, 2008) and, thus, reduce the resource loss that can result from overqualification. To our knowledge, there is no study that investigates optimism as a moderator of the relationship between perceived overqualification and psychological well-being. We have, however, found studies that reveal optimism as a buffer in relation to other job stressors, including daily hassles (Fry, 1995), role stress (Garrosa, Moreno-Jiménez, Rodríguez-Muñoz, & Rodríguez-Carvajal, 2011), time pressure, and poor organizational climate (Mäkikangas & Kinnunen, 2003).

Building on COR theory (Hobfoll, 1989), we propose that optimism buffers the relationship between perceived overqualification and psychological well-being among immigrants.

Hypothesis 3: Optimism moderates the positive relationship between perceived overqualification and

Ó 2019 Hogrefe Publishing

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

depressive symptoms: the relationship is attenuated with increasing optimism.

Hypothesis 4: Optimism moderates the negative relationship between perceived overqualification and life satisfaction: the relationship is attenuated with increasing optimism.

Meaning-making is defined as “the ability to integrate challenging or ambiguous situations into a framework of personal meaning” (Van den Heuvel, Demerouti, Schreurs, Bakker, & Schaufeli, 2009, p. 509). People with high levels of meaning-making actively engage in the construction of meaning by means of values-based reflection and thereby regulate their psychological well-being. Previous research has shown that successful meaning-making (in terms of finding meaning) is associated with lower levels of depressive symptoms (e.g., Helgeson, Reynolds, & Tomich, 2006; Park, Park, & Peterson, 2010) and higher levels of life satisfaction (for a review, see Park, 2010).

Beyond these well-established direct associations, we expect meaning-making to buffer the relationship between perceived overqualification and depressive symptoms and life satisfaction. As part of the process of meaningmaking, values-based reflection can diminish the uncertainty inherent to ambiguous situations and establish a sense of acceptance and control (Van den Heuvel, Demerouti, Bakker, & Schaufeli, 2013). Furthermore, meaningmaking can reduce the discrepancy between the appraised meaning of a stressful event and higher-level goals or beliefs (see Joseph & Linley, 2005; Park, 2011). When perceiving overqualification, immigrants might reduce this discrepancy by making downward comparisons with less fortunate others (e.g., unemployed immigrants) or by imagining less favorable hypothetical scenarios (e.g., being unemployed themselves). Further meaning-making strategies include selectively focusing on positive aspects of the situation (e.g., on the chance to build up social networks and acquire labor market experience), finding acceptable and modifiable reasons for the situation (e.g., the lack of language skills), or downgrading one’s achievement aspirations (for a summary, see Park, 2010, 2011).

Even though meaning-making has not been considered in the context of perceived overqualification, it has been shown to buffer the negative relationship between other job stressors (i.e., time pressure) and well-being outcomes (i.e., vigor) among nurses (Wassermann, Hoppe, Reis, & von Uthmann, 2014).

Building on COR theory (Hobfoll, 1989), we propose that meaning-making buffers the relationship between perceived overqualification and psychological well-being among immigrants.

Ó 2019 Hogrefe Publishing

37

Hypothesis 5: Meaning-making moderates the positive relationship between perceived overqualification and depressive symptoms: the relationship is attenuated with increasing meaning-making.

Hypothesis 6: Meaning-making moderates the negative relationship between perceived overqualification and life satisfaction: the relationship is attenuated with increasing meaning-making.

Method Procedure and Sample An online survey was conducted among Italian immigrants in Germany in 2014. The participants were recruited through posts in online social networks as well as via mailing lists of cultural associations. A total of 427 people started the questionnaire after reading a short introduction (including information about voluntariness of participation, data confidentiality, and anonymity). Of those, 270 (63.2%) finished the questionnaire and provided a full dataset.

The final sample consisted of 176 Italian nationals who lived and worked in Germany at the time the data were collected. We only included participants who were working full time (127; 72.2%), part time (26; 14.8%), or on a freelance basis (23; 13.1%) because the level of perceived overqualification was only assessed among these groups.

The participants of this study were living in 35 different German cities and communities and had migrated to Germany between 1972 and 2014. The majority (80.7%) migrated after 2005. Among our participants, 94 (53.4%) were female. The study participants were between 19 and 59 years old (M = 35.27, SD = 7.92).

Measures Perceived overqualification was assessed with nine items of the Scale of Perceived Overqualification (Maynard et al., 2006), ranging from 1 (= strongly disagree) to 7 (= strongly agree) (e.g., “My job requires less education than I have”).

Optimism was measured with the Scale Optimism-Pessimism-2 (Kemper, Beierlein, Kovaleva, & Rammstedt, 2013; Italian version: Kemper, Wassermann, Hoppe, Beierlein, & Rammstedt, 2015). This short scale contains two items (e.g., “Optimists are people who look to the future with confidence and who mostly expect good things to happen.

How would you describe yourself? How optimistic are you in general?”). The items were answered on a rating scale

Journal of Personnel Psychology (2019), 18(1), 34–45

38

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

ranging from 1 (= not at all optimistic/pessimistic) to 7 (= very optimistic/pessimistic). The validity and reliability of the Italian adaptation of the scale had been confirmed and reported based on data that included the sample of the present study (Kemper et al., 2015).

Meaning-making was measured with seven items (Van den Heuvel et al., 2009) that were rated on a scale ranging from 1 (= strongly disagree) to 5 (= strongly agree) (e.g., “I actively take the time to reflect on events that happen in my life”).

Depressive symptoms were measured with four items of the short form of the Center for Epidemiological Studies Depression Scale (CES-D; Melchior, Huba, Brown, & Reback, 1993; Radloff, 1977; Italian version: Fava, 1982).

Each item assessed the frequency of a depressive symptom during the last seven days (e.g., “I felt depressed”) and was rated on a scale from 1 (= rarely/none of the time; < 1 day) to 4 (= most or all of the time; 5–7 days).

Life satisfaction was measured with five items of the Satisfaction with Life Scale (Diener, Emmons, Larsen, & Griffin, 1985; Italian version: Prezza & Costantini, 1998) that were answered on a scale ranging from 1 (= strongly disagree) to 5 (= strongly agree) (e.g., “In most ways my life is close to my ideal”).

We computed confirmatory factor analyses (CFA) using the software R and Lavaan package (Rosseel, 2012). First, we tested a model with all five constructs loaded onto a single factor, w2(324, N = 176) = 1,567.8, p = .000, CFI = .49, RMSEA = .148 (CI [.141, .155]), SRMR = .147. Second, we tested a model with three factors with both indicators of psychological well-being and both personal resources loaded onto a common factor, w2(321, N = 176) = 817.0, p = .000, CFI = .80, RMSEA = .094 (CI [.086, .101]), SRMR = .098. Third, we tested a model with five factors and all items loaded onto their respective construct, w2(314, N = 176) = 588.4, p = .000, CFI = .89, RMSEA = .070 (CI [.062, .079]), SRMR = .078. The third model showed better fit indices than the first model (Δw2 = 979.4, Δdf = 10, p < .001) and the second model (Δw2 = 228.6, Δdf = 7, p < .001).

All items were provided in the Italian language. For perceived overqualification and meaning-making, a standard forward–backward translation procedure involving bilingual

translators was employed to create Italian-language versions of the scales. For all other scales, Italian-language versions were available as indicated above.

Data Analysis We ran two hierarchical regression analyses in SPSS 22: one analysis for each of the two dependent variables. We entered the predictors in two steps. In Step 1, we entered perceived overqualification to test the direct association between perceived overqualification and depressive symptoms (Hypothesis 1) and life satisfaction (Hypothesis 2).

Optimism and meaning-making were also added in Step 1.

In Step 2, we entered the product terms of perceived overqualification and both personal resources to test whether optimism and meaning-making moderate the relationship between perceived overqualification and psychological well-being (Hypotheses 3–6).

All predictors were centered around their grand mean in order to facilitate interpretation of the results (Aiken & West, 1991). A bootstrap approach with n = 1,000 bias-corrected resamples was used, because not all study variables were normally distributed (Bollen & Stine, 1990).

Results Table 1 shows the means, standard deviations, internal consistencies, and zero-order correlations of the study variables. Duration of residence and gender of the participants did not correlate with any study variables. Age was negatively associated with depressive symptoms (r = .24, p < .01).

Perceived Overqualification and Psychological Well-Being We had proposed a negative relationship between perceived overqualification and psychological well-being. As shown in Table 2, the results from Step 1 revealed a significant positive association between perceived overqualification and

Table 1. Means, standard deviations, Cronbach’s α (in parentheses on the diagonal), and correlations between study variables Variable

M

SD

1

1. Perceived overqualification

3.41

1.75

(.94)

2

3

4

2. Optimism

4.95

1.43

.21**

(.87)

3. Meaning-making

3.75

0.67

.04

.25**

(.74)

4. Depressive symptoms

1.55

0.57

.33**

.42**

.18*

(.76)

5. Life satisfaction

3.62

0.99

.43**

.41**

.27**

.54**

5

(.91)

Note. N = 176. *p < .05, **p < .01.

Journal of Personnel Psychology (2019), 18(1), 34–45

Ó 2019 Hogrefe Publishing

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

39

Table 2. Hierarchical regression of depressive symptoms and life satisfaction Depressive symptoms Step 1 Ba

Life satisfaction

Step 2 SEa

Ba

Step 1 SEa

Ba

Step 2 SEa

Ba

SEa

Perceived overqualification (POQ)

.17**

.05

.16**

.05

.42**

.07

.39**

.07

Optimism (O)

.19**

.04

.19**

.05

.26**

.06

.29**

.06

Meaning-making (MM)

.05

.04

.06

.05

.17**

.06

.15*

.06

.04

.05

.14*

.06

POQ O POQ MM

.08

.05

.04

.08

R2

.249**

.273**

.358**

.375**

Change in F

19.04**

2.82

31.91**

2.35

Note. N = 176; B = unstandardized regression coefficients; SE = standard error; R2 = variance explained in each step. aResults based on 1,000 bootstrapped samples. *p < .05, **p < .01.

depressive symptoms (B = .17, CI [.08, .26], p = .001). This supports Hypothesis 1.

Furthermore, we found a significant negative association between perceived overqualification and life satisfaction (B = .42, CI [ .57, .27], p = .001). This supports Hypothesis 2.

The Moderating Role of Personal Resources Optimism as a Moderator The results of Step 2 showed that the two-way interaction between optimism and perceived overqualification on depressive symptoms was not significant (B = .04, CI [ .13, .06], p = .430). Therefore, Hypothesis 3 was not supported. However, the two-way interaction between optimism and perceived overqualification on life satisfaction was significant (B = .14, CI [.01, .26], p = .025, see Figure 1).

For a more specific test, we conducted a simple slope analysis as proposed by Aiken and West (1991) using PROCESS (Model 1, Hayes, 2013). The simple slope indicated that perceived overqualification had a stronger negative relationship with life satisfaction for participants with low levels of optimism ( 1 SD, B = .28; SE = 0.04, t = 6.43, p = .000) than for participants with high levels of optimism (+1 SD, B = .15; SE = 0.05, t = 3.00, p = .003). Therefore, our findings confirm Hypothesis 4.

Meaning-Making as a Moderator The results from Step 2 revealed that the two-way interaction between meaning-making and perceived overqualification on depressive symptoms was not significant (B = .08, CI [ .17, .01], p = .116). Likewise, the two-way interaction between meaning-making and perceived overqualification on life satisfaction was not significant (B = .04, CI [ .22, .11], p = .603). Therefore, our findings do not confirm Hypothesis 5 and Hypothesis 6. Ă&#x201C; 2019 Hogrefe Publishing

Figure 1. Two-way interaction of perceived overqualification and optimism on life satisfaction.

We ran all models with duration of residence, age, and gender as control variables. The results were essentially the same and are available from the first author upon request.

Discussion Consistent with P-E fit theory (Edwards et al., 1998; KristofBrown et al., 2005), in this study, higher perceived overqualification was related to lower psychological well-being among immigrant workers. More specifically, we found a positive relationship between perceived overqualification and depressive symptoms and a negative relationship between perceived overqualification and life satisfaction.

Furthermore, in line with COR theory (Hobfoll, 1989), optimism buffered the negative relationship between perceived overqualification and life satisfaction. Against our expectations, optimism did not moderate the positive relationship between perceived overqualification and depressive

Journal of Personnel Psychology (2019), 18(1), 34â&#x20AC;&#x201C;45

40

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

symptoms. Also, meaning-making did not moderate the relationship between perceived overqualification and depressive symptoms and life satisfaction, respectively.

Previous research has linked P-E fit theory and COR theory by describing the person–job fit as a condition that facilitates the accumulation of resources such as a higher salary and better job positions. Therefore, a good person–job fit can be seen as a resource itself (Lanivich et al., 2010).

Whereas P-E fit theory predicts individual-level outcomes such as psychological well-being, COR theory explains interactions between constructs that threaten resources (e.g., person–job misfit) and constructs that have the potential to protect resources (e.g., optimism). Our results support the assumption that perceived overqualification represents a person–job misfit that is related to lower psychological well-being. Whereas the adverse relationship between perceived overqualification and psychological well-being has been demonstrated mainly among native workers (e.g., Johnson & Johnson, 1992, 1996; Roh et al., 2014), in this study, we surveyed a specific sample of immigrant workers, namely Italian immigrants in Germany. At the time that data for this study were collected, the unemployment rate in Germany was relatively low (5.0%). In contrast, Italy faced an unemployment rate more than twice as high (12.7%), particularly among young people (42.7%; OECD, 2018). In the preceding years, there was a significant increase of immigration from Italy to Germany (e.g., +35% from 2012 to 2013; Bundesamt für Migration und Flüchtlinge [Federal Office for Migration and Refugees], 2015). Our results indicate that – also when facing a high risk of unemployment in their home country – immigrants are not satisfied with taking any job, but that they want a job that allows them to fully utilize their education, experience, knowledge, and skills and that doing such a job might be relevant for their psychological well-being. Our findings on the adverse relationship between perceived overqualification and psychological well-being are important for both organizations and immigrant workers. There is sound evidence that depressive symptoms reduce job performance and lead to long-term sickness absence (Glozier, 2002; Lagerveld et al., 2010). Furthermore, poor psychological well-being can threaten the adjustment of immigrants to the host country. For example, depressive symptoms might interfere with foreign language acquisition because they are related to attention and memory deficits (e.g., Hammar & Årdal, 2009). Low satisfaction with life in the host country is related to higher intentions to leave the host country (Mara & Landesmann, 2013).

Our findings on the moderating role of personal resources only partly support the assumptions derived from COR theory (Hobfoll, 1989). As proposed, optimism diminished the negative relationship between perceived overqualification and life satisfaction. Thus, optimism Journal of Personnel Psychology (2019), 18(1), 34–45

appears to be a personal resource that can reduce resource loss and foster immigrants’ resiliency in the face of an unfavorable job situation. When facing barriers in the labor market of the host country, immigrants might be required to take on a job for which they are overqualified, and they have to maintain the motivation to succeed (Zikic, Bonache, & Cerdin, 2010). Optimism can help them to master this challenge. Optimists cope more actively with stressors and pursue their goal more persistently than pessimists (Aspinwall & Taylor, 1997; Nes & Segerstrom, 2006). To this end, optimists are likely to use self-determination strategies (e.g., self-leadership and job crafting) more effectively (e.g., Bakker & van Woerkom, 2017) which, in turn, can help them to overcome perceived overqualification in the long term. The positive expectation of holding a better job in the future might weaken the negative relationship between the currently perceived P-E misfit (i.e., perceived overqualification) and life satisfaction among optimists. Based on the ideas of Caplan (1983), who first addressed the relevance of anticipated P-E fit, Sen (1992) showed that the interaction of present and anticipated fit is more strongly related to job satisfaction than present fit alone. Taking into account anticipated P-E fit and factors that promote positive expectations (such as optimism) as moderators would advance our knowledge on the relationship between perceived overqualification and psychological well-being.

Interestingly, against our expectations, optimism did not moderate the positive relationship between perceived overqualification and depressive symptoms. Whereas life satisfaction is a conscious cognitive judgment of one’s life in general, depressive symptoms may evolve on an unconscious level and constitute a rather affective component of psychological well-being (Pavot & Diener, 1993). Therefore, whereas the assessment of life satisfaction can be affected by an optimistic long-term perspective of the job situation, depressive symptoms are likely to be mainly affected by immediate experiences of overqualification. Overqualification is often related to structural barriers (Zikic et al., 2010) that are, at least in the short term, resistant to the influence of active coping by optimists (Mäkikangas & Kinnunen, 2003).

Compared to optimism, meaning-making does not buffer the relationship between perceived overqualification and depressive symptoms or life satisfaction. Previous research has shown that meaning-making can be related to better adjustment to stressors, but also to rumination (Park, 2010). Rumination is defined as repetitive negative thoughts (Michael & Snyder, 2005) and can maintain or increase negative feelings. Park (2010) pointed out that the benefits of meaning-making are more related to finding meaning than the process of meaning-making itself. However, the meaning-making measure in our study focused more on the effort Ó 2019 Hogrefe Publishing

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

of meaning-making than on its success. On the one hand, meaning-making efforts might be beneficial for immigrants if they result in increased feelings of meaningfulness (e.g., by seeing benefits in overqualification). On the other hand, the effort of finding meaning might also result in higher awareness of the futility of overqualification. In consequence, potentially positive and negative aspects of meaning-making might balance each other out and account for the nonsignificant findings in our study.

Last but not least, we would like to point out that – in line with prior research (e.g., Park, 2010; Scheier & Carver, 1992; Wong & Lim, 2009) – optimism and meaning-making were directly related to higher psychological well-being in our study. Optimism was negatively associated with depressive symptoms and positively associated with life satisfaction. Meaning-making was positively related to life satisfaction. Therefore, optimism and meaning-making should be considered as valuable personal resources for immigrants. However, in our study, the moderating role of these personal resources in the interplay between perceived overqualification and psychological well-being was less pronounced than expected.

Limitations Some limitations of this study need to be considered when interpreting the results. First, its cross-sectional nature made it impossible to investigate cause-and-effect relations.

Therefore, we are unable to rule out that depressive symptoms or dissatisfaction with life – vice versa – increase perceptions of overqualification. Also, based on COR theory (Hobfoll, 1989), it is possible that, beyond the postulated moderating role of personal resources, perceiving overqualification leads to the depletion of personal resources over time which, in turn, impairs psychological well-being. Longitudinal studies with three or more points of measurement should investigate the relationship between perceived overqualification, personal resources, and well-being by applying mediation analyses.

Second, the self-report data might artificially inflate the correlations between the constructs, because of common method variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003; Spector, 1994). Even though self-report is the most suitable method for assessing psychological well-being and personal resources, additional objective information on overqualification would have been desirable (Maltarich, Reilly, & Nyberg, 2011). It is important to acknowledge that perceived overqualification and objective overqualification are related but different constructs (see Liu & Wang, 2012). Perceived overqualification is assumed to be triggered by an objective mismatch between the qualifications of a worker and his or her job demands. Therefore, it is Ó 2019 Hogrefe Publishing

41

likely to mediate the relationship between objective overqualification and psychological well-being. However, perceived overqualification is also affected by other factors, such as personality traits and context variables (Liu & Wang, 2012; Maltarich et al., 2011), and can, therefore, differ from objective overqualification.

Third, Internet-based surveys facilitate the recruitment of people who are normally difficult to recruit (Khazaal et al., 2014), which is the case when surveying immigrants. However, this sampling method also results in a self-selected sample.

Finally, we solely studied Italian immigrants in Germany.

Italian immigrants can easily cross national boarders within the EU and are legally entitled to equal treatment with regard to access to employment, guaranteed by the Treaty on the Functioning of the European Union. It is unclear whether our results could be generalized to the situation of immigrants from other countries, particularly non-EU countries.

Direction for Future Research It is reasonable to focus on populations at risk when studying perceived overqualification. However, future research should systematically compare the relationship of perceived overqualification and psychological well-being between immigrants and natives. More specifically, longitudinal studies with matched samples (i.e., immigrants and natives matched with regard to sociodemographics and work characteristics) would be required to investigate how immigrants differ from natives when perceiving and dealing with overqualification. There is some evidence in the literature that the response of immigrants to job characteristics differs from that of native workers who hold similar jobs (see, e.g., Hoppe, 2011; Hoppe, Toker, Schachler, & Ziegler, 2017), including their response to overqualification (Frank & Hou, 2017).

Furthermore, future research should investigate the effect of perceived overqualification on psychological well-being over and beyond that accounted for by wellestablished job stressors (e.g., high workload and role ambiguity) and adverse job characteristics (e.g., low salary and involuntary part-time work).

Finally, as our knowledge about potential moderators of perceived overqualification is constantly increasing, it is time to develop and evaluate workplace interventions that aim to enhance resources among overqualified workers and to move forward our knowledge on resource-based interventions in the workplace (Michel, O’Shea, & Hoppe, 2015). Following Briner and Walshe (2015), we regard it as worthwhile to develop and evaluate resource-based interventions for workers in need, specifically for the underserved population of immigrant workers.

Journal of Personnel Psychology (2019), 18(1), 34–45

42

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

Practical Implications Based on our findings and the results of previous studies, it is important to reduce barriers that prevent immigrants from obtaining jobs that match their professional skills.

These barriers include a lack of language skills, social networks, and local know-how and failure to acknowledge foreign qualifications (Zikic et al., 2010). Researchers have pointed out the importance of employing international talents under adequate conditions in order to maintain their well-being and health (e.g., Chen et al., 2010; George et al., 2012; Smith & Frank, 2005) but also in order to justify the remarkable effort companies put into recruiting international employees (Buchanan, Scott, Yu, Schutz, & Jakubauskas, 2010).

Nevertheless, overqualification is common among immigrants, and measures are required to help immigrants deal with the adversity of perceived overqualification. Based on our findings, a promising approach could be to strengthen optimism. This can be achieved, for example, in short interventions fostering positive work reflection which can be integrated into the workday (e.g., Clauss et al., 2018). Furthermore, employers should provide immigrants who perceive overqualification with working conditions that help them to sustain positive future expectations, for example, by providing interesting tasks, promotion prospects, and options for personal development (e.g., through language training and mentoring programs). Additionally, employers should ensure best possible rewards (e.g., an appropriate salary) and other job-related resources in order to compensate for the lack of skill utilization (Erdogan et al., 2011; Van den Heuvel et al., 2010).

Conclusion This study demonstrates that a poor person–job fit in terms of perceived overqualification is associated with higher levels of depressive symptoms and lower levels of life satisfaction among immigrants. Optimism can diminish the negative relationship between perceived overqualification and life satisfaction. Besides improving immigrants’ access to jobs that fit their qualifications, another strategy to help immigrants deal with perceived overqualification could be to enhance optimism through person-level interventions.

References Adler, S. (1977). Maslow’s need hierarchy and the adjustment of immigrants. International Migration Review, 11, 444–451. https://doi.org/10.2307/2545398 Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Journal of Personnel Psychology (2019), 18(1), 34–45

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5) (5th ed.). Washington, DC: Author.

Aspinwall, L. G., & Taylor, S. E. (1997). A stitch in time: Selfregulation and proactive coping. Psychological Bulletin, 121, 417–436. https://doi.org/10.1037/0033-2909.121.3.417 Aycan, Z., & Berry, J. W. (1996). Impact of employment-related experiences on immigrants’ psychological well-being and adaptation to Canada. Canadian Journal of Behavioural Science/Revue Canadienne des Sciences du Comportement, 28, 240–251. https://doi.org/10.1037/0008-400X.28.3.240 Bakker, A. B., & van Woerkom, M. (2017). Flow at work: A selfdetermination perspective. Occupational Health Science, 1, 47–65. https://doi.org/10.1007/s41542-017-0003-3 Bartolini, L., Gropas, R., & Triandafyllidou, A. (2017). Drivers of highly skilled mobility from Southern Europe: Escaping the crisis and emancipating oneself. Journal of Ethnic and Migration Studies, 43, 652–673. https://doi.org/10.1080/ 1369183X.2016.1249048 Bollen, K. A., & Stine, R. (1990). Direct and indirect effects: Classical and bootstrap estimates of variability. Sociological Methodology, 20, 15–140. https://doi.org/10.2307/271084 Briner, R. B., & Walshe, N. D. (2015). An evidence-based approach to improving the quality of resource-oriented well-being interventions at work. Journal of Occupational and Organizational Psychology, 88, 563–586. https://doi.org/10.1111/joop.12133 Buchanan, J., Scott, L. D., Yu, S., Schutz, H., & Jakubauskas, M. (2010). Skills demand and utilisation: An international review of approaches to measurement and policy development (OECD Local Economic and Employment Development Working Papers, No. 2010/04). Paris, France: OECD Publishing. https://doi.org/10.1787/5km8zddfr2jk-en Bundesamt für Migration und Flüchtlinge. (2015). Migrationsbericht des Bundesamts für Migration und Flüchtlinge im Auftrag der Bundesregierung: Migrationsbericht 2013 [2013 Migration Report of the Federal Ministry of Migration and Refugees]. Nuremberg, Germany: Bundesministerium des Inneren.

Caplan, R. D. (1983). Person-environment fit: Past, present, and future. In C. L. Cooper (Ed.), Stress research: New directions for the 1980s (pp. 35–78). London, UK: Wiley.

Chen, C., Smith, P., & Mustard, C. (2010). The prevalence of overqualification and its association with health status among occupationally active new immigrants to Canada. Ethnicity & Health, 15, 601–619. https://doi.org/10.1080/ 13557858.2010.502591 Clauss, E., Hoppe, A., O’Shea, D., González Morales, M. G., Steidle, A., & Michel, A. (2018). Promoting personal resources and reducing exhaustion through positive work reflection among caregivers. Journal of Occupational Health Psychology, 23, 127–140. https://doi.org/10.1037/ocp0000063 Crollard, A., de Castro, A., & Tsai, J. H. C. (2012). Occupational trajectories and immigrant worker health. Workplace Health & Safety, 60, 497–503. https://doi.org/10.3928/2165079920121016-68 Daukantaite, D., & Zukauskiene, R. (2012). Optimism and subjective well-being: Affectivity plays a secondary role in the relationship between optimism and global life satisfaction in the middle-aged women. Longitudinal and cross-cultural findings. Journal of Happiness Studies, 13, 1–16. https://doi.org/ 10.1007/s10902-010-9246-2 Dean, J. A., & Wilson, K (2009). “Education? It is irrelevant to my job now. It makes me very depressed. . .”: Exploring the health impacts of under/unemployment among highly skilled recent immigrants in Canada. Ethnicity &Health, 14, 185–204. https:// doi.org/10.1080/13557850802227049

Ó 2019 Hogrefe Publishing

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

De Grip, A., Bosma, H., Willems, D., & Van Boxtel, M. (2008). Job– worker mismatch and cognitive decline. Oxford Economic Papers, 60, 237–253. https://doi.org/10.1093/oep/gpm023 Desjardins, R., & Rubenson, K. (2011). An analysis of skill mismatch using direct measures of skills (OECD Education Working Papers, No. 63). Paris, France: OECD Publishing. https://doi.org/10.1787/5kg3nh9h52g5-en Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71–75. https://doi.org/10.1207/s15327752jpa4901_13 Edwards, J. R., Caplan, R. D., & Van Harrison, R. (1998). Personenvironment fit theory: Conceptual foundations, empirical evidence, and directions for future research. In C. L. Cooper (Ed.), Theories of organizational stress (Vol. 28, pp. 28–67).

Oxford, UK: Oxford University Press.

Erdogan, B., & Bauer, T. N. (2009). Perceived overqualification and its outcomes: The moderating role of empowerment. Journal of Applied Psychology, 94, 557–565. https://doi.org/10.1037/ a0013528 Erdogan, B., Bauer, T. N., Peiró, J., & Truxillo, D. M. (2011).

Overqualified employees: Making the best of a potentially bad situation for individuals and organizations. Industrial and Organizational Psychology, 4, 215–232. https://doi.org/ 10.1111/j.1754-9434.2011.01330.x Fava, G. A. (1982). Versione italiana del CES-D per la valutazione degli stati depressivi [The Italian version of CES-D for the evaluation of depressive states]. In R. Canestrari (Ed.), Nuovi metodi in psicometria (pp. 87–93). Florence, Italy: Organizzazioni Speciali.

Fine, S., & Nevo, B. (2008). Too smart for their own good? A study of perceived cognitive overqualification in the workforce. The International Journal of Human Resource Management, 19, 346–355. https://doi.org/10.1080/09585190701799937 Frank, K., & Hou, F. (2018). Over-education and well-being: How does education-occupation mismatch affect the life satisfaction of university-educated immigrant and non-immigrant workers? Ethnicity & Health, 23, 884–901. https://doi.org/ 10.1080/13557858.2017.1316832 Fry, P. S. (1995). Perfectionism, humor, and optimism as moderators of health outcomes and determinants of coping styles of women executives. Genetics, Social, and General Psychology Monographs, 121, 211–245.

Fugate, M., & Kinicki, A. J. (2008). A dispositional approach to employability: Development of a measure and test of implications for employee reactions to organizational change. Journal of Occupational and Organizational Psychology, 81, 503–527. https://doi.org/10.1348/096317907X241579 Garrosa, E., Moreno-Jiménez, B., Rodríguez-Muñoz, A., & Rodríguez-Carvajal, R. (2011). Role stress and personal resources in nursing: A cross-sectional study of burnout and engagement.

International Journal of Nursing Studies, 48, 479–489. https:// doi.org/10.1016/j.ijnurstu.2010.08.004 George, U., Chaze, F., Fuller-Thomson, E., & Brennenstuhl, S. (2012). Underemployment and life satisfaction: A study of internationally trained engineers in Canada. Journal of Immigrant & Refugee Studies, 10, 407–425. https://doi.org/10.1080/ 15562948.2012.717827 Glozier, N. (2002). Mental ill health and fitness for work. Occupational and Environmental Medicine, 59, 714–720. https://doi. org/10.1136/oem.59.10.714 Goldberg, R. J., & Steury, S. (2001). Depression in the workplace: Costs and barriers to treatment. Psychiatric Services, 52, 1639– 1643. https://doi.org/10.1176/appi.ps.52.12.1639 Griffin, J. M., Greiner, B. A., Stansfeld, S. A., & Marmot, M. (2007).

The effect of self-reported and observed job conditions on depression and anxiety symptoms: A comparison of theoretical

Ó 2019 Hogrefe Publishing

43

models. Journal of Occupational Health Psychology, 12, 334– 349. https://doi.org/10.1037/1076-8998.12.4.334 Hammar, Å., & Årdal, G. (2009). Cognitive functioning in major depression – A summary. Frontiers in Human Neuroscience, 3, 1–7. https://doi.org/10.3389/neuro.09.026.2009 Harari, M. B., Manapragada, A., & Viswesvaran, C. (2017). Who thinks they’re a big fish in a small pond and why does it matter?

A meta-analysis of perceived overqualification. Journal of Vocational Behavior, 102, 28–47. https://doi.org/10.1016/j. jvb.2017.06.002 Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press.

Hayes, C. T., & Weathington, B. L. (2007). Optimism, stress, life satisfaction, and job burnout in restaurant managers. The Journal of Psychology, 141, 565–579. https://doi.org/10.3200/ JRLP.141.6.565-580 Helgeson, V. S., Reynolds, K. A., & Tomich, P. L. (2006). A metaanalytic review of benefit finding and growth. Journal of Consulting and Clinical Psychology, 74, 797–816. https://doi. org/10.1037/0022-006X.74.5.797 Hobfoll, S. E. (1989). Conservation of resources: A new attempt at conceptualizing stress. The American Psychologist, 44, 513– 524. https://doi.org/10.1037/0003-066X.44.3.513 Hobfoll, S. E. (2001). The influence of culture, community, and the nested-self in the stress process: Advancing conservation of resources theory. Applied Psychology, 50, 337–421. https://doi. org/10.1111/1464-0597.00062 Hobfoll, S. E. (2011). Conservation of resource caravans and engaged settings. Journal of Occupational and Organizational Psychology, 84, 116–122. https://doi.org/10.1111/j.20448325.2010.02016.x Hobfoll, S. E., Johnson, R. J., Ennis, N., & Jackson, A. P. (2003).

Resource loss, resource gain, and emotional outcomes among inner city women. Journal of Personality and Social Psychology, 84, 632–643. https://doi.org/10.1037/0022-3514.84.3.632 Hoppe, A. (2011). Psychosocial working conditions and well-being among immigrant and German low wage workers. Journal of Occupational Health Psychology, 16, 187–201. https://doi.org/ 10.1037/a0021728 Hoppe, A., Toker, S., Schachler, V., & Ziegler, M. (2017). The effect of change in supervisor support and job control on change in vigor: Differential relationships for immigrant and native employees in Israel. Journal of Organizational Behavior, 38, 391–414. https://doi.org/10.1002/job.2151 Johnson, G. J., & Johnson, W. R. (1992). Subjective underemployment and psychosocial stress: The role of perceived social and supervisor support. The Journal of Social Psychology, 132, 11– 21. https://doi.org/10.1080/00224545.1992.9924684 Johnson, G. J., & Johnson, W. R. (1996). Perceived overqualification and psychological well-being. The Journal of Social Psychology, 136, 435–445. https://doi.org/10.1080/ 00224545.1996.9714025 Johnson, G. J., & Johnson, W. R. (1997). Perceived overqualification, emotional support, and health. Journal of Applied Social Psychology, 27, 1906–1918. https://doi.org/10.1111/j.15591816.1997.tb01631.x Joseph, S., & Linley, P. A. (2005). Positive adjustment to threatening events: An organismic valuing theory of growth through adversity. Review of General Psychology, 9, 262–280. https:// doi.org/10.1037/1089-2680.9.3.262 Kemper, C. J., Beierlein, C., Kovaleva, A., & Rammstedt, B. (2013).

Entwicklung und Validierung einer ultrakurzen Operationalisierung des Konstrukts Optimismus-Pessimismus – Die Skala Optimismus-Pessimismus-2 (SOP2) [Development and validation of an ultrashort measure for the construct of

Journal of Personnel Psychology (2019), 18(1), 34–45

44

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

optimism-pessimism – The Scale Optimism-Pessimism-2 (SOP2)]. Diagnostica, 59, 119–129. https://doi.org/10.1026/ 0012-1924/a000089 Kemper, C. J., Wassermann, M., Hoppe, A., Beierlein, C., & Rammstedt, B. (2015). Measuring dispositional optimism in large-scale studies – Psychometric evidence for German, Spanish, and Italian versions of the Scale Optimism-Pessimism-2 (SOP2). European Journal of Psychological Assessment, 33, 403–408. https://doi.org/10.1027/1015-5759/a000297 Khazaal, Y., Van Singer, M., Chatton, A., Achab, S., Zullino, D., Rothen, S., & Thorens, G. (2014). Does self-selection affect samples’ representativeness in online surveys? An investigation in online video game research. Journal of Medical Internet Research, 16, e164. https://doi.org/10.2196/jmir.2759 Kristof-Brown, A. L., Zimmerman, R. D., & Johnson, E. C. (2005).

Consequences of individuals’ fit at work: A meta-analysis of person–job, person–organization, person–group, and person– supervisor fit. Personnel Psychology, 58, 281–342. https://doi. org/10.1111/j.1744-6570.2005.00672.x Lagerveld, S. E., Bültmann, U., Franche, R. L., Van Dijk, F. J. H., Vlasveld, M. C., Van der Feltz-Cornelis, C. M., . . . Nieuwenhuijsen, K. (2010). Factors associated with work participation and work functioning in depressed workers: A systematic review.

Journal of Occupational Rehabilitation, 20, 275–292. https:// doi.org/10.1007/s10926-009-9224-x Lanivich, S. E., Brees, J. R., Hochwarter, W. A., & Ferris, G. R. (2010). P-E Fit as moderator of the accountability–employee reactions relationships: Convergent results across two samples. Journal of Vocational Behavior, 77, 425–436. https://doi. org/10.1016/j.jvb.2010.05.004 Lindert, J., Schouler-Ocak, M., Heinz, A., & Priebe, S. (2008).

Mental health, health care utilisation of migrants in Europe.

European Psychiatry, 23, 14–20. https://doi.org/10.1016/ S0924-9338(08)70057-9 Liu, S., Luksyte, A., Zhou, L., Shi, J., & Wang, M. (2015).

Overqualification and counterproductive work behaviors: Examining a moderated mediation model. Journal of Organizational Behavior, 36, 250–271. https://doi.org/10.1002/ job.1979 Liu, S., & Wang, M. (2012). Perceived overqualification: A review and recommendations for research and practice. Research in Occupational Stress and Well-being, 10, 1–42. https://doi.org/ 10.1108/S1479-3555(2012)0000010005 Luksyte, A., Spitzmueller, C., & Maynard, D. C. (2011). Why do overqualified incumbents deviate? Examining multiple mediators. Journal of Occupational Health Psychology, 16, 279–296. https://doi.org/10.1037/a0022709 Luthans, F., Avey, J. B., Avolio, B. J., Norman, S. M., & Combs, G. M. (2006). Psychological capital development: Toward a microintervention. Journal of Organizational Behavior, 27, 387– 393. https://doi.org/10.1002/job.373 Mäkikangas, A., Feldt, T., Kinnunen, U., & Mauno, S. (2013). Does personality matter? A review of individual differences in occupational well-being. Advances in Positive Organizational Psychology, 1, 107–143. https://doi.org/10.1108/S2046-410X (2013)0000001008 Mäkikangas, A., & Kinnunen, U. (2003). Psychosocial work stressors and well-being: Self-esteem and optimism as moderators in a one-year longitudinal sample. Personality and Individual Differences, 35, 537–557. https://doi.org/10.1016/S0191-8869 (02)00217-9 Maltarich, M. A., Reilly, G., & Nyberg, A. J. (2011). Objective and subjective overqualification: Distinctions, relationships, and a place for each in the literature. Industrial and Organizational Psychology, 4, 236–239. https://doi.org/10.1111/j.17549434.2011.01332.x

Journal of Personnel Psychology (2019), 18(1), 34–45

Mara, I., & Landesmann, M. (2013). Do I stay because I am happy or am I happy because I stay? Life satisfaction in migration, and the decision to stay permanently, return and out-migrate.

(NORFACE Migration Discussion Paper No. 2013-08). Retrieved from http://www.norface-migration.org/publ_uploads/ NDP_08_13.pdf Maynard, D. C., Joseph, T. A., & Maynard, A. M. (2006). Underemployment, job attitudes, and turnover intentions. Journal of Organizational Behavior, 27, 509–536. https://doi.org/10.1002/ job.389 Maynard, D. C., & Parfyonova, N. M. (2013). Perceived overqualification and withdrawal behaviours: Examining the roles of job attitudes and work values. Journal of Occupational and Organizational Psychology, 86, 435–455. https://doi.org/10.1111/ joop.12006 Melchior, L. A., Huba, G., Brown, V. B., & Reback, C. J. (1993). A short depression index for women. Educational and Psychological Measurement, 53, 1117–1125. https://doi.org/10.1177/ 0013164493053004024 Michael, S. T., & Snyder, C. R. (2005). Getting unstuck: The roles of hope, finding meaning, and rumination in the adjustment to bereavement among college students. Death Studies, 29, 435– 458. https://doi.org/10.1080/07481180590932544 Michel, A., O’Shea, D., & Hoppe, A. (2015). Designing and evaluating resource-oriented interventions to enhance employee well-being and health. Journal of Occupational and Organizational Psychology, 88, 459–463. https://doi.org/ 10.1111/joop.12131 Nes, L. S., & Segerstrom, S. C. (2006). Dispositional optimism and coping: A meta-analytic review. Personality and Social Psychology Review, 10, 235–251. https://doi.org/10.1207/ s15327957pspr1003_3 Nowok, B., Van Ham, M., Findlay, A. M., & Gayle, V. (2013). Does migration make you happy? A longitudinal study of internal migration and subjective well-being. Environment and Planning A, 45, 986–1002. https://doi.org/10.1068/a45287 OECD. (2018). Unemployment rate (indicator). Retrieved from https://data.oecd.org/unemp/unemployment-rate.htm Park, C. L. (2010). Making sense of the meaning literature: An integrative review of meaning making and its effects on adjustment to stressful life events. Psychological Bulletin, 136, 257–301. https://doi.org/10.1037/a0018301 Park, C. L. (2011). Meaning, coping, and health and well-being. In S. Folkman (Ed.), The Oxford handbook of stress, health, and coping (pp. 227–241). New York, NY: Oxford University Press.

Park, N., Park, M., & Peterson, C. (2010). When is the search for meaning related to life satisfaction? Applied Psychology: Health and Well-Being, 2, 1–13. https://doi.org/10.1111/j.17580854.2009.01024.x Pavot, W., & Diener, E. (1993). Review of the satisfaction with life scale. Psychological Assessment, 5, 164–172. https://doi.org/ 10.1037/1040-3590.5.2.164 Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies.

Journal of Applied Psychology, 88, 879–903. https://doi.org/ 10.1037/0021-9010.88.5.879 Prezza, M., & Costantini, S. (1998). Sense of community and life satisfaction: Investigation in three different territorial contexts.

Journal of Community & Applied Social Psychology, 8, 181–194. https://doi.org/10.1002/(SICI)1099-1298(199805/06)8:3<181:: AID-CASP436>3.0.CO;2-4 Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. https://doi.org/10.1177/ 014662167700100306

Ó 2019 Hogrefe Publishing

M. Wassermann & A. Hoppe, Perceived Overqualification and Psychological Well-Being

Roh, Y. H., Chang, J. Y., Kim, M. U., & Nam, S. K. (2014). The effects of income and skill utilization on the underemployed’s self-esteem, mental health, and life satisfaction. Journal of Employment Counseling, 51, 125–141. https://doi.org/10.1002/ j.2161-1920.2014.00047.x Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48, 1–36. https://doi. org/10.18637/jss.v048.i02 Ruthig, J. C., Haynes, T. L., Stupnisky, R. H., & Perry, R. P. (2009).

Perceived academic control: Mediating the effects of optimism and social support on college students’ psychological health.

Social Psychology of Education, 12, 233–249. https://doi.org/ 10.1007/s11218-008-9079-6 Scheier, M. F., & Carver, C. S. (1985). Optimism, coping and health: Assessment and implications of generalized outcome expectancies. Health Psychology, 4, 219–247. https://doi.org/ 10.1037/0278-6133.4.3.219 Scheier, M. F., & Carver, C. S. (1992). Effects of optimism on psychological and physical well-being: Theoretical overview and empirical update. Cognitive Therapy and Research, 16, 201–228. https://doi.org/10.1007/BF01173489 Schwarzer, R., Hahn, A., & Schröder, H. (1994). Social integration and social support in a life crisis: Effects of macrosocial change in East Germany. American Journal of Community Psychology, 22, 685–706. https://doi.org/10.1007/BF02506899 Seibert, H., & Wapler, R. (2012). Zuwanderung nach Deutschland: Aus dem Ausland kommen immer mehr Akademiker [Migration to Germany: More and more academics are coming from abroad] (IAB-Kurzbericht, 21). Nuremberg, Germany: Institut für Arbeitsmarkt und Berufsforschung.

Sen, M. (1992). Retrospected and anticipated fits: An exploration into their differential effects in a sample of Indian managers.

Work & Stress, 6, 153–162. https://doi.org/10.1080/ 02678379208260349 Smith, P., & Frank, J. (2005). When aspirations and achievements don’t meet. A longitudinal examination of the differential effect of education and occupational attainment on declines in selfrated health among Canadian labour force participants. International Journal of Epidemiology, 34, 827–834. https://doi.org/ 10.1093/ije/dyi047 Spector, P. E. (1994). Using self-report questionnaires in OB research: A comment on the use of a controversial method.

Journal of Organizational Behavior, 15, 385–392. https://doi. org/10.1002/job.4030150503 Van den Heuvel, M., Demerouti, E., Bakker, A. B., & Schaufeli, W. B. (2010). Personal resources and work engagement in the face of change. In J. Houdmont & S. Leka (Eds.), Contemporary occupational health psychology (Vol. 1, pp. 124–150). Chichester, UK: Wiley.

Ó 2019 Hogrefe Publishing

45

Van den Heuvel, M., Demerouti, E., Bakker, A. B., & Schaufeli, W. B. (2013). Adapting to change: The value of change information and meaning making. Journal of Vocational Behavior, 83, 11–21. https://doi.org/10.1016/j.jvb.2013.02.004 Van den Heuvel, M., Demerouti, E., Schreurs, B. H., Bakker, A. B., & Schaufeli, W. B. (2009). Does meaning making help during organizational change? Development and validation of a new scale. Career Development International, 14, 508–533. https:// doi.org/10.1108/13620430910997277 Wassermann, M., Hoppe, A., Reis, D., & von Uthmann, L. (2014).

Sinnstiftung als persönliche Ressource bei Altenpflegekräften.

Zu direkten und moderierenden Effekten von Sinnstiftung auf emotionale Erschöpfung und Vitalität [Meaning-making as a personal resource among geriatric nurses: Direct and moderating effects on emotional exhaustion and vigour]. Zeitschrift für Arbeits- und Organisationspsychologie, 58, 51–63. https:// doi.org/10.1026/0932-4089/a000142 Wong, S. S., & Lim, T. (2009). Hope versus optimism in Singaporean adolescents: Contributions to depression and life satisfaction. Personality and Individual Differences, 46, 648– 652. https://doi.org/10.1016/j.paid.2009.01.009 Wu, C. H., Luksyte, A., & Parker, S. K. (2015). Overqualification and subjective well-being at work: The moderating role of job autonomy and culture. Social Indicators Research, 121, 917– 937. https://doi.org/10.1007/s11205-014-0662-2 Zikic, J., Bonache, J., & Cerdin, J. L. (2010). Crossing national boundaries: A typology of qualified immigrants’ career orientations. Journal of Organizational Behavior, 31, 667–686. https://doi.org/10.1002/job.705 History Received December 14, 2016 Revision received April 19, 2018 Accepted July 5, 2018 Published online January 9, 2019 ORCID Maria Wassermann https://orcid.org/0000-0003-4361-7193 Maria Wassermann Department of Psychology HU Berlin Unter den Linden 6 10099 Berlin Germany maria.wassermann@hu-berlin.de

Journal of Personnel Psychology (2019), 18(1), 34–45

nline free o issue le samp

Journal of Individual Differences Editor-in-Chief Martin Voracek University of Vienna, Austria Associate Editors André Beauducel, Germany Philip J. Corr, UK Sam Gosling, USA Jürgen Hennig, Germany Philipp Y. Herzberg, Germany Aljoscha Neubauer, Austria

ISSN-Print 1614-0001 ISSN-Online 2151-2299 ISSN-L 1614-0001 4 issues per annum (= 1 volume) Subscription rates (2019) Individuals US $159.00 / € 114.00 (print & online) Institutions From US $308.00 / € 232.00 (print only; pricing for online access can be found in the journals catalog at hgf.io/journals2019) Postage / Handling US $16.00 / € 12.00

www.hogrefe.com

About the Journal Researchers, teachers, and students interested in all areas of individual differences (e.g., gender, temperament, personality, intelligence) and their assessment in human and animal research will find the Journal of Individual Differences useful. It publishes manuscripts dealing with individual differences in behavior, emotion, cognition, and their developmental aspects. This includes human as well as animal research. The Journal of Individual Differences is conceptualized to bring together researchers working in different areas ranging from, for example, molecular genetics to theories of complex behavior. Moreover, it places emphasis on papers dealing with special methodological and conceptual issues in basic science as well as in their applied fields (assessment of personality and intelligence).

Karl-Heinz Renner, Germany Willibald Ruch, Switzerland Astrid Schütz, Germany Andrzej Sekowski, Poland Jutta Stahl, Germany

Manuscript Submissions All manuscripts should be submitted online at www.editorialmanager.com/jindivdiff, where full instructions to authors are also available.

Electronic Full Text The full text of the journal – current and past issues (from 1999 onward) – is available online at econtent.hogrefe.com/loi/jid (included in subscription price). A free sample issue is also available here.

Abstracting Services The journal is abstracted / indexed in Current Contents / Social and Behavioral Sciences (CC / S&BS), Social Sciences Citation Index (SSCI), Scopus, PsycINFO, and PSYNDEX. Impact Factor (Journal Citation Reports®, Clarivate Analytics): 2017 = 1.283

Integrative perspectives on motivation and volition “This is an excellent and valuable volume. It is a wonderful collection of pieces on motivation that serves as an apt tribute to an unusually creative and generous scholar.”

Andrew J. Elliot, PhD, Professor of Psychology, Department of Clinical & Social Sciences in Psychology, University of Rochester, NY, USA

Nicola Baumann / Miguel Kazén / Markus R. Quirin / Sander L. Koole (Editors)

Why People Do the Things They Do Building on Julius Kuhl’s Contributions to the Psychology of Motivation and Volition 2018, xii + 434 pp.

US $87.00 / € 69.95 ISBN 978-0-88937-540-6 Also available as eBook How can we motivate students, patients, employees, and athletes? What helps us achieve our goals, improve our well-being, and grow as human beings? These issues, which relate to motivation and volition, are familiar to everyone who faces the challenges of everyday life. This comprehensive book by leading international scholars provides integrative perspectives on motivation and volition that build on the work of German psychologist Julius Kuhl.

The first part of the book examines the historical trail of the European and American research traditions of motivation and volition and their integration in Kuhl’s theory of personality systems interactions (PSI). The sec-

www.hogrefe.com

ond part of the book considers what moves people to action – how needs, goals, and motives lead people to choose a course of action (motivation).

The third part of the book explores how people, once they have committed themselves to a course of action, convert their goals and intentions into action (volition). The fourth part shows what an important role personality plays in our motivation and actions. Finally, the fifth part of the book discusses how integrative theories of motivation and volition may be applied in coaching, training, psychotherapy, and education.

This book is essential reading for everyone who is interested in the science of motivating people.

Research Note

Measuring High-Quality Work Relationships A Test of Model and Gender Invariance Michael T. Warren1

and Meg A. Warren2

1

Department of Psychology, Western Washington University, Bellingham, WA, USA

2

Department of Management, Western Washington University, Bellingham, WA, USA

Abstract: Research on high-quality relationships (HQR) between coworkers has garnered considerable interest, yet the original HQR measure (Carmeli, 2009) has been adapted in disparate ways (e.g., including vs. omitting the vitality subscale). Continued application is further complicated by incomplete reporting on the measure’s factor structure. Relatedly, findings that women often experience relationships differently than men highlight the need to test whether the HQR measure functions similarly across genders. We surveyed 401 employees (50% women) to test four competing models and examine invariance across genders. Findings revealed that HQR is best conceptualized as six distinct but correlated constructs. Further, full scalar invariance was observed across genders, indicating that the measure functioned equivalently and can be used in gender comparisons involving HQR constructs.

Keywords: positive work relationships, relationship quality, positive organizational scholarship, gender, measurement invariance

Positive coworker relationships are intangible organizational assets that predict numerous benefits, such as learning, creativity, and performance (Brueller & Carmeli, 2011; Carmeli, Dutton, & Hardin, 2015; Carmeli & Gittell, 2009).

According to Dutton and Heaphy (2003), high-quality work relationships are characterized by structural features of emotional carrying capacity (authentic and constructively expressed emotions), tensility or resilience, and openness; as well as emotional experiences of positive regard, mutuality, and vitality. This theory has gained traction and inspired empirical research, in part due to the availability of its self-report measure (Carmeli, 2009). However, empirical examinations of high-quality relationships (HQR) have often deviated from the original conceptual framing by only examining certain elements (e.g., Brueller & Carmeli, 2011; Carmeli, 2009), by developing alternative operationalizations that do not involve any of the six elements (Carmeli & Gittell, 2009), and by forming new constructs that borrow elements from HQR (Carmeli et al., 2015).

Although there is merit in separately examining elements of HQR and testing links with theory-informed outcomes, the fragmentation of HQR may also reflect inconclusive psychometric investigation of integrated HQR measures that involve all six elements. An integrated measure conceptually aligns with the view that HQR is best epitomized by relationships that function at a high level across all six domains, and suggests creating total scores by averaging across domains. To date, two studies have tested whether Journal of Personnel Psychology (2019), 18(1), 46–52 https://doi.org/10.1027/1866-5888/a000217

such an integrated approach is warranted, and determined that there are two second-order HQR factors: structural capacities and emotional experiences (Carmeli, 2009; Carmeli, Brueller, & Dutton, 2009). However, neither study included the vitality domain, model fit was either poor (Carmeli, 2009) or incompletely reported (Carmeli et al., 2009), and no tests of alternative models were reported.

Inconsistency in measuring HQR obscures whether studies that claim to examine HQR are in fact measuring the same construct. Further, Carmeli’s (2009) measure carries theoretical assumptions about the structure of HQR (1) that have yet to be thoroughly tested, and (2) pending such testing should be utilized in a theory-consistent manner to improve employees’ lives. For instance, the organizing framework of structural versus emotional aspects, if empirically distinct and critical, might influence whom to target for interventions, for example, dyads/groups for structural changes; individuals for emotions. The inconsistent modeling of HQR breeds confusion and inhibits such advances in intervention research.

We build on past work by examining four competing models of HQR described below and depicted in Figure 1.

Given the conceptual distinctiveness of each HQR element, it is plausible that the six domains are best modeled as separate constructs. Therefore, Model 1 specifies six correlated first-order factors where the items load onto their respective factors. Yet, theory and research indicate that the shared variance among the six HQR domains can be modeled as Ó 2019 Hogrefe Publishing

M. T. Warren & M. A. Warren, Measuring High-Quality Relationships

47

Figure 1. Four competing models examining the structure of high-quality relationships.

structural capacities and emotional experiences (Carmeli, 2009; Carmeli et al., 2009; Dutton & Heaphy, 2003).

Accordingly, Model 2 specifies two second-order factors representing structural capacities and emotional experiences, upon which the six (three each) first-order factors load (cf. Carmeli, 2009). Also according to theory, all six elements are conceptualized as indicators of HQR (Dutton & Heaphy, 2003), and therefore, they may cohere in an overarching HQR construct rather than requiring separate structural and emotional factors. Indeed, a strong test of the tenability of Model 2 (two higher-order factors) would be to compare it against a model in which all six elements load onto a single overarching HQR construct. Therefore, Model 3 specifies one second-order HQR factor upon which all six first-order factors load. Finally, we consider a model that similarly involves an overarching HQR construct, but that does not differentiate between the six theorized elements. Although the six domains are conceptually distinct, we test their empirical tenability by considering Model 4: one overarching HQR construct upon which all items directly load (irrespective of the domain they represent). We examine the empirical fit of these competing models to inform best practices for modeling HQR.

1

HQR and Gender A related psychometric issue concerns measuring HQR across genders.1 Gender differences have been found in relational aspects of work (Lee, Kesebir, & Pillutla, 2016), and in the display, perception, and evaluation of emotion in work relationships (Ragins & Winkel, 2011). When power imbalances across genders are strong and salient, women tend to guess others’ emotions, remain attuned to interpersonal nuances, and appraise the impact of their behaviors on others (Lips, 2010). It follows that women and men may interpret questions about work relationships differently, and the HQR measure may function differently across genders.

Comparisons of HQR scores across genders can be misleading if women and men interpret the items differently, or if salient reference frameworks (e.g., social norms) lead one gender to give higher ratings in the absence of true differences (Chen, 2008). For instance, one HQR item is “If I get upset with my co-workers, I know they will try to understand me.” Compared to men, for whom overt expression of anger is normalized, women risk being perceived as “out of control,” and are therefore forced to express anger

We do not intend to imply a binary (i.e., female or male) framework of gender. However, we did not seek out gender nonconforming participants and only one participant self-identified as “other,” making it impossible to systematically examine those who do not identify as female or male.

Ó 2019 Hogrefe Publishing

Journal of Personnel Psychology (2019), 18(1), 46–52

48

in subtle forms (Brescoll & Uhlmann, 2008). Therefore, what counts as “being upset” might be different for women and men, creating gender disparities in what the scale measures. A useful way to examine the equivalence of a scale across disparate groups is by testing measurement invariance (Chen, 2008). Past HQR research has not examined invariance across genders, although gender has the capacity to be critical to the study of relationship quality. We examine the measurement invariance of the HQR measure to ensure that it is defensible to make gender comparisons in associations and means involving the HQR measure, and to ensure that researchers can interpret HQR scores the same way across genders.

In sum, the current study has two aims: Aim 1: Compare the four competing models to determine which fits best for each gender.

Aim 2: Test the measurement invariance of the HQR measure across genders.

Method Participants Participants were recruited through Amazon’s Mechanical Turk, and were (1) adults living in the USA, (2) employed by their current organization for at least 6 months, (3) had regular face-to-face interaction with at least three coworkers, and (4) read English.

The questionnaire was pilot tested for comprehension.

The survey was completed by 484 participants. Data from participants who failed attention check questions, and therefore failed to demonstrate attentive responding, were excluded. Responses from duplicate IP addresses were screened out. The overwhelming majority of excluded cases (67) were duplicate IP addresses. This resulted in a total of 401 participants (50% female; 81% Caucasian).

The estimated median age was 33 years, and a wide range of occupations was represented (e.g., engineering, food services, law). Positions ranged from directors and executives (7%), to middle-level managers and professionals (53%), to frontline staff such as tellers and janitors (39%; 1% did not identify). The average organizational tenure was 6.26 years (SD = 5.17).

Measures HQR was measured with Carmeli’s (2009) scale. Participants responded to statements about emotional carrying Journal of Personnel Psychology (2019), 18(1), 46–52

M. T. Warren & M. A. Warren, Measuring High-Quality Relationships

capacity (4 items; α = .80); tensility (6 items; α = .92), openness-based connectivity (4 items; α = .80), positive regard (5 items; α = .91), mutuality (4 items; α = .90), and vitality (6 items; α = .92), using a scale from 1 (= Strongly Disagree) to 7 (= Strongly Agree).

Data Analysis We used Mplus version 8 (Muthén & Muthén, 1998–2017) to examine confirmatory factor analysis models. Weighted least squares means and variance-adjusted (WLSMV) estimation with delta parameterization was used. WLSMV is appropriate for ordered categorical observed variables (Flora & Curran, 2004), performs better (e.g., less parameter bias) than robust maximum likelihood estimation (Bandalos, 2014), and is appropriate for small samples (Beauducel & Herzberg, 2006; Sass, Schmitt, & Marsh, 2014). WLSMV tests of scalar invariance (see below) require each item to have an equal number of response categories across groups. When Strongly Disagree was not used by one gender, we met the above condition by collapsing the Strongly Disagree and Disagree categories for both genders. No data were missing, as we had IRB approval to require participants to respond to all items.

Since chi-square is sensitive to sample size, we evaluated model fit using the comparative fit index (CFI), TuckerLewis index (TLI), root-mean-square error of approximation (RMSEA), and the weighted root-mean-square residual (WRMR). CFI and TLI values larger than .90 and RMSEA values less than .08 indicate acceptable fit (Little, 2013).

Standards have yet to be developed for WRMR, but lower values indicate better fit.

Structure of HQR Confirmatory factor analyses tested the four competing HQR models depicted in Figure 1. Because WLSMV w2 values are not directly comparable, model fit comparisons were performed with the Δw2 DIFFTEST option in Mplus (Asparouhov & Muthén, 2006). DIFFTEST performs better than other methods (e.g., ΔCFI) when comparing nested models with WLSMV estimation (Sass et al., 2014). A significance threshold of α = .05 was used to determine if the more parsimonious model (i.e., the model with more df) in a comparison had worse fit.

Measurement Invariance Across Genders Measurement models were separately fit for women and men, and then the data were combined into a multigroup configural invariance model. In this model, the same specifications were used across genders, but no constraints were placed on any of the parameters (e.g., factor loadings, thresholds). Ó 2019 Hogrefe Publishing

M. T. Warren & M. A. Warren, Measuring High-Quality Relationships

49

Table 1. Alternate factor structures for high-quality relationships for women and men Model

w2

df

Models compared

Δw2

Δdf

p

CFI

TLI

RMSEA

WRMR

.968

.964

.086

1.075

Women (n = 200) 1. Six correlated first-order factors: E, T, O, P, M, V items load onto their respective factors

897.05

362

–

–

–

2. Two second-order factors: Three factors (E, T, O) load onto one factor; three factors (P, M, V) load onto the other factor

874.88

370

2 vs. 1

19.80

8

.013

.970

.967

.083

1.103

3. One second-order factor: Six factors (E, T, O, P, M, V) load onto one factor

939.23

371

3 vs. 1

58.66

9

< .001

.966

.963

.088

1.168

1,788.05

377

4 vs. 1

450.34

15

< .001

.916

.910

.137

1.857

1. Six correlated first-order factors: E, T, O, P, M, V items load onto their respective factors

735.53

362

–

–

–

.965

.961

.072

0.945

2. Two second-order factors: Three factors (E, T, O) load onto one factor; three factors (P, M, V) load onto the other factor

750.75

370

2 vs. 1

28.41

8

< .001

.964

.961

.072

0.984

3. One second-order factor: Six factors (E, T, O, P, M, V) load onto one factor

773.83

371

3 vs. 1

47.06

9

< .001

.962

.959

.073

1.039

1,720.51

377

4 vs. 1

469.46

15

.875

.865

.133

1.789

4. One first-order factor: All items load onto one factor Men (n = 201)

4. One first-order factor: All items load onto one factor

< .001

Note. Selected models are in bold. With WLSMV estimation, differences in w2 are not distributed as w2 and therefore nested models were compared with the Mplus DIFFTEST, the results of which are presented in the Δw2 column. It is noteworthy that for women, Model 2 appeared to fit better (e.g., lower w2 and RMSEA; higher CFI and TLI), but the Δw2 DIFFTEST was significant and therefore indicates that the more parsimonious Model 2 fit significantly worse than Model 1 (the Δw2 DIFFTEST is unidirectional and is only significant when the nested, parsimonious model has worse fit). WLSMV = weighted least squares mean and variance estimator; CFI = comparative fit index; TLI = Tucker-Lewis index; RMSEA = root-mean-square error of approximation; WRMR = weighted root-mean-square residual. E, T, O, P, M, and V represent the six facets of high-quality relationships (E = emotional carrying capacity; T = tensility, O = openness-based connectivity; P = positive regard; M = mutuality; V = vitality).

A metric invariance model was specified to examine equivalence of the factor loadings across genders. When metric invariance holds, it is likely that the two groups interpreted the items in a similar way (Chen, 2008). Different interpretations of a given item would result in stronger factor loadings for one group than the other, with larger loadings for the gender that interprets the item in a manner more aligned with the meaning of the underlying HQR domain. To test metric invariance, equality constraints across genders were placed on the factor loadings,2 and the constrained (metric invariance) model was compared to the unconstrained (configural) model. A nonsignificant Δw2 is evidence of metric invariance.

A scalar invariance model was specified to examine equivalence of the item thresholds across genders. When scalar invariance holds, it is likely that the two groups used similar reference points in determining their responses (Chen, 2008), and therefore, factor means can be com2

3

pared across groups. To test scalar invariance, additional equality constraints were placed on the item thresholds, and the constrained (scalar invariance) model was compared to the metric invariance model. A nonsignificant Δw2 is evidence of scalar invariance.3

Results Structure of HQR Results for the competing models in Figure 1 are summarized separately for women and men in Table 1. Model 1 (six correlated first-order factors) generally had adequateto-good fit for both genders [for women: w2(362) = 897.05, p < .001; CFI = .968; TLI = .964; RMSEA = .086; WRMR = 1.075; for men: w2(362) = 735.53, p < .001;

Since our structure of HQR results showed that Model 1 (six correlated first-order factors) fit best for both genders, there were no second-order factor constraints to consider in the measurement invariance tests.

There are additional forms of invariance (e.g., item residuals, factor variances) that could have been tested. We restricted our analyses to factor loading and threshold invariance because these determine whether comparisons of covariances and means across genders are justified.

Ó 2019 Hogrefe Publishing

Journal of Personnel Psychology (2019), 18(1), 46–52

50

M. T. Warren & M. A. Warren, Measuring High-Quality Relationships

Table 2. Tests of measurement invariance across genders for model with six correlated first-order factors (Model 1) w2

Model

df

Models compared

Δw2

Δdf

p diff

CFI

TLI RMSEA WRMR

Baseline women (n = 200)

897.05

362

.968

.964

.086

1.075

Baseline men (n = 201)

735.53

362

.965

.961

.072

0.945

1. Configural invariance (no constraints)

1,624.87

724

.968

.964

.079

1.429

2. Metric invariance (factor loadings constrained)

1,599.65

747

2 vs. 1

24.63

23

.370

.969

.967

.075

1.442

3. Scalar invariance (thresholds constrained)

1,480.83

886

3 vs. 2

163.62

139

.075

.979

.980

.058

1.560

Note. With WLSMV estimation, differences in w are not distributed as w and therefore nested models were compared with the Mplus DIFFTEST, the results of which are presented in the Δw2 column. Six items had one unused response category (1 = Strongly Disagree), so categories 1 and 2 were collapsed for both genders to permit the scalar (threshold) invariance test. Consequently, the number of constrained thresholds in the scalar invariance model was six fewer than expected (i.e., 139 rather than 145). CFI = comparative fit index; TLI = Tucker-Lewis index; RMSEA = root-mean-square error of approximation; WRMR = weighted root-mean-square residual.

2

2

CFI = .965; TLI = .961; RMSEA = .072; WRMR = 0.945], although for women the RMSEA of .086 did not clear the .08 threshold for adequate fit. Model 2 (two second-order factors with three first-order factors loading onto each) also fit the data quite well and had similar approximate fit indexes (e.g., for men: CFI = .964). However, the Δw2 test indicated that Model 2 fit significantly worse than Model 1 for both genders. Thus, empirically, the model that was derived from HQR theory (Model 2) did not fit as well as the model with six distinct but correlated HQR constructs (Model 1). Model 3 (one second-order factor upon which all six first-order factors loaded) also fit the data quite well overall, but the Δw2 test indicated that it fit significantly worse than Model 1. Finally, Model 4 (one first-order factor that all items load onto) displayed poor-to-adequate fit, and the Δw2 test indicated that it fit significantly worse than Model 1. Therefore, Model 1 had the best fit and was selected for subsequent analyses for both genders.

model did not differ significantly from the configural model (see Table 2). These findings indicated that the factor loadings were invariant across genders, suggesting that women and men interpreted the items similarly, and justifying gender comparisons of covariances involving the HQR constructs. Notably, any such comparisons of covariances would further require that the covariances be standardized, for example, by using a phantom variable approach within structural equation modeling (Little, 2013; Rindskopf, 1984), or by comparing correlation coefficients or standardized regression weights.

Finally, scalar invariance was tested. After placing equality constraints on the item thresholds, the scalar invariance model did not differ significantly from the metric invariance model (see Table 2). These findings indicated that the item thresholds were invariant across genders, suggesting that women and men used similar points of reference in determining their responses, and justifying gender comparisons of HQR constructs’ means. In short, the HQR measure demonstrated evidence of full scalar invariance.4

Measurement Invariance Across Genders Tests of invariance examined whether the measure functions the same for women and men. This was accomplished by testing whether the items’ factor loadings and thresholds were equal across genders.

Data from both genders were combined into a configural invariance multigroup model, which had adequate-to-good fit depending on the fit index under consideration (see Table 2). These findings indicated that the same model specifications were reasonable for both genders.

Metric invariance was tested next. After placing equality constraints on the factor loadings, the metric invariance 4

Discussion and Conclusion This study tested competing models of Carmeli’s (2009) HQR scale, and examined whether the scale functions the same way for women and men. Diverging from theory (Dutton & Heaphy, 2003) and prior confirmatory factor analyses (Carmeli, 2009; Carmeli et al., 2009), our analysis of competing models found that HQR was best modeled as six distinct but correlated factors: emotional carrying capacity, tensility, openness-based connectivity, positive regard,

Following Sass et al. (2014), we also performed analyses with robust maximum likelihood (MLR) estimation to test both the structure and measurement invariance. Converging with our WLSMV results, Model 1 was most appropriate for women, but the more parsimonious Model 3 fit just as well for men. These findings somewhat temper our confidence that Model 1 is best supported empirically, although WLSMV is generally superior for ordered categorical data (Bandalos, 2014; Flora & Curran, 2004). Since Model 1 still fit well for men and fit best for both genders in our primary analyses, we used it as the baseline for MLR tests of measurement invariance. Results indicated full scalar invariance with adequate model fit. Finding similar results with MLR increases our confidence in the full scalar invariance of the HQR measure. Results from the MLR analyses are available upon request from the first author.

Journal of Personnel Psychology (2019), 18(1), 46–52

Ó 2019 Hogrefe Publishing

M. T. Warren & M. A. Warren, Measuring High-Quality Relationships

mutuality, and vitality. This finding has two divergent sets of implications: on the one hand, assuming the theory is correct, the scale would need to be refined into a more psychometrically robust measure of the two higher-order HQR constructs – structural capacities and emotional experiences. On the other hand, the theory itself may need to be refined. Since the HQR domains did not have overwhelmingly high correlations with each other, the data suggest that it is more accurate to conceptualize the six domains as separate constructs rather than as manifestations of two underlying HQR constructs. In practice, the scale seems to be well-suited to produce scores for the six domains, but these domains should not be further combined into total scores for structural capacities and emotional experiences (nor overall HQR). That said, the distinction between structural and emotional aspects may hold practical value, for example, by differentiating between useful intervention targets (structural capacities) and their affective consequences (emotional experiences).

The second goal of this study was to examine gender invariance of the measure. Although relational experiences at work differ by gender (e.g., Lee et al., 2016), we found that the HQR measure functions similarly for women and men. We found full scalar invariance, suggesting that women and men interpreted the survey items similarly and used similar reference frames (e.g., norms) to determine their responses (Chen, 2008). These findings indicate that researchers are justified in comparing associations (and means) across genders, and practitioners can interpret HQR domain scores the same way across genders.

Our findings should be viewed in light of certain limitations. First, empirical findings are just one piece of the puzzle in theory refinement. We hope our findings will stimulate discussion about HQR theory while offering useful directions for its modification. Second, the vitality aspect of HQR was not strictly relationship-oriented, as some items measured general vitality at work (e.g., “I am most vital when I am at work.”). By straying from the relationship context, such vitality items might have compromised the cohesiveness (and thus the fit) of the higher-order factor models.5 Future research should develop relationshiporiented vitality items. Third, as with past HQR research, the measure captures global judgments of HQR domains across all work colleagues, whereas Dutton and Heaphy’s (2003) theory discusses temporally brief, high-quality dyadic connections. Nonetheless, global judgments may have the broadest influence on organizational outcomes. Finally, the HQR measure has been examined primarily in a North American individualistic cultural context. Future research should examine differential item functioning for individual5

51

istic versus collectivistic cultures, as well as across ethnicities and industries.

In conclusion, the present study extends empirical work on measurement of HQR in the workplace. Burgeoning research on HQR reveals that the construct has been modeled in disparate ways, with limited empirical investigation of the measure. Our results show that a model with six distinct factors is the best application of the measure. We hope this clarification will guide researchers and practitioners to measure HQR as six separate variables. In addition, our findings demonstrate that the measure functions – and is likely interpreted – similarly across genders. Thus, scholars and practitioners concerned about gendered nuances of relationships may confidently compare associations and means across genders.

References Asparouhov, T., & Muthén, B. (2006). Robust chi square difference testing with mean and variance adjusted test statistics.

Retrieved from www.statmodel.com/download/webnotes/webnote10.pdf Bandalos, D. L. (2014). Relative performance of categorical diagonally weighted least squares and robust maximum likelihood estimation. Structural Equation Modeling, 21, 102–116. https:// doi.org/10.1080/10705511.2014.859510 Beauducel, A., & Herzberg, P. Y. (2006). On the performance of maximum likelihood versus means and variance adjusted weighted least squares estimation in CFA. Structural Equation Modeling, 13, 186–203. https://doi.org/10.1207/ s15328007sem1302_2 Brescoll, V. L., & Uhlmann, E. L. (2008). Can an angry woman get ahead? Status conferral, gender, and expression of emotion in the workplace. Psychological Science, 19, 268–275. https://doi. org/10.1111/j.1467-9280.2008.02079.x Brueller, D., & Carmeli, A. (2011). Linking capacities of high-quality relationships to team learning and performance in service organizations. Human Resource Management, 50, 455–477. https://doi.org/10.1002/hrm.20435 Carmeli, A. (2009). Positive work relationships, vitality, and job performance. In C. E. J. Hartel, N. M. Ashkanasy, & W. J. Zerbe (Eds.), Research on emotions in organizations: The effect of affect in organizational settings (Vol. 5, pp. 45–71). Bingley, UK: Emerald.

Carmeli, A., Brueller, D., & Dutton, J. E. (2009). Learning behaviours in the workplace: The role of high-quality interpersonal relationships and psychological safety. Systems Research and Behavioral Science, 26, 81–98. https://doi.org/ 10.1002/sres.932 Carmeli, A., Dutton, J. E., & Hardin, A. E. (2015). Respect as an engine for new ideas: Linking respectful engagement, relational information processing, and creativity among employees and teams. Human Relations, 68, 1021–1047. https://doi.org/ 10.1177/0018726714550256 Carmeli, A., & Gittell, J. H. (2009). High-quality relationships, psychological safety, and learning from failures in work

In support of this point, in the two models we tested with higher-order factors (Models 2 and 3), vitality had the lowest factor loading for both genders.

Ó 2019 Hogrefe Publishing

Journal of Personnel Psychology (2019), 18(1), 46–52

52

organizations. Journal of Organizational Behavior, 30, 709–729. https://doi.org/10.1002/job.v30:610.1002/job.565 Chen, F. F. (2008). What happens if we compare chopsticks with forks? The impact of making inappropriate comparisons in cross-cultural research. Journal of Personality and Social Psychology, 95, 1005–1018. https://doi.org/10.1037/a0013193 Dutton, J. E., & Heaphy, E. D. (2003). The power of high-quality connections. In K. S. Cameron, J. E. Dutton, & R. E. Quinn (Eds.), Positive organizational scholarship (pp. 263–278). San Francisco, CA: Berret-Koehler.

Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9, 466– 491. https://doi.org/10.1037/1082-989X.9.4.466 Lee, S. Y., Kesebir, S., & Pillutla, M. M. (2016). Gender differences in response to competition with same-gender coworkers: A relational perspective. Journal of Personality and Social Psychology, 110, 869–886. https://doi.org/10.1037/pspi0000051 Lips, H. M. (2010). A new psychology of women: Gender, culture, and ethnicity (3rd ed.). Long Grove, IL: Waveland Press.

Little, T. D. (2013). Longitudinal structural equation modeling. New York, NY: Guilford Press.

Muthén, L. K., & Muthén, B. O. (1998–2017). Mplus user’s guide (8th ed.). Los Angeles, CA: Muthén & Muthén.

Ragins, B. R., & Winkel, D. E. (2011). Gender, emotion and power in work relationships. Human Resource Management Review, 21, 377–393. https://doi.org/10.1016/j.hrmr.2011.05.001 Rindskopf, D. (1984). Using phantom and imaginary latent variables to parameterize constraints in linear structural

Journal of Personnel Psychology (2019), 18(1), 46–52

M. T. Warren & M. A. Warren, Measuring High-Quality Relationships

models. Psychometrika, 49, 37–47. https://doi.org/10.1007/ BF02294204 Sass, D. A., Schmitt, T. A., & Marsh, H. W. (2014). Evaluating model fit with ordered categorical data within a measurement invariance framework: A comparison of estimators. Structural Equation Modeling, 21, 167–180. https://doi.org/10.1080/ 10705511.2014.882658 History Received June 21, 2017 Revision received April 27, 2018 Accepted May 11, 2018 Published online January 9, 2019 ORCID Michael T. Warren https://orcid.org/0000-0002-2958-333X Meg A. Warren Department of Management College of Business and Economics Western Washington University Parks Hall 432 516 High Street Bellingham WA 98225 USA meg.warren@wwu.edu

Ó 2019 Hogrefe Publishing

News and Announcements Awards for Outstanding Achievements as Authors and Reviewers 2018 The Journal of Personnel Psychology is delighted to announce the winners of the Best Paper Award and the Best Reviewers Award for 2018, chosen by a committee composed of members of the editorial team.

Best Paper Award 2018 The 2018 Best Paper Award goes to: Roulin, N., & Powell, D. M. (2018). Identifying applicant faking in job interviews: Examining the role of criterionbased content analysis and storytelling. Journal of Personnel Psychology, 17, 143â&#x20AC;&#x201C;154. https://doi.org/10.1027/18665888/a000207

Ă&#x201C; 2019 Hogrefe Publishing

Best Reviewer Awards The truly unsung heroes in the process of publishing research are the anonymous external reviewers, who often contribute substantially to the quality of the final product without being recognized at all. In an attempt to change this at least in a few of the most deserved cases, the Journal of Personnel Psychology distinguishes outstandingly helpful reviewers with its annual Best Reviewer Award.

For 2018, these awards go to Janneke K. Oostrom (Vrije Universiteit Amsterdam, The Netherlands) and Annika Wilhelmy (University of Zurich, Switzerland).

Congratulations to the award winners!

Journal of Personnel Psychology (2019), 18(1), 53 https://doi.org/10.1027/1866-5888/a000221

Instructions to Authors Aims and Scope of Journal of Personnel Psychology: The journal welcomes excellent empirical and theoretical contributions to basic and applied research in personnel psychology and related methodology. Reviews are also welcome, as are replications of previous research. Articles deal with all fields in personnel psychology, such as personnel selection, performance measurement, motivation, leadership, organizational commitment, personnel development and training, new test developments, and job analysis. As many topics in personnel psychology are closely related to issues in other branches of psychology or, more generally, the social sciences and human resource management, the journal is open to contributions of an interdisciplinary nature.

Journal of Personnel Psychology publishes the following types of articles: Original Articles, Review Articles, Research Notes, Registered Reports, and Hybrid Registered Reports.

Manuscript submisson: All manuscripts should be submitted electronically at http://www.editorialmanager.com/jppsy Detailed instructions to authors are provided at http://www. hogrefe.com/j/jpp Copyright Agreement: By submitting an article, the author confirms and guarantees on behalf of him-/herself and any coauthors that the manuscript has not been submitted or published elsewhere, and that he or she holds all copyright in and titles to the submitted contribution, including any figures, photographs, line drawings, plans, maps, sketches, and tables, and that the article and its contents do not infringe in any way on the rights of third parties. The author indemnifies and holds harmless the publisher from any third-party claims.

The author agrees, upon acceptance of the article for publication, to transfer to the publisher the exclusive right to reproduce and distribute the article and its contents, both physically and in nonphysical, electronic, or other form, in the journal to which it has been submitted and in other independent publications, with no limitations on the number of copies or on the form or the extent of distribution. These rights are transferred for the duration of

copyright as defined by international law. Furthermore, the author transfers to the publisher the following exclusive rights to the article and its contents: 1. The rights to produce advance copies, reprints, or offprints of the article, in full or in part, to undertake or allow translations into other languages, to distribute other forms or modified versions of the article, and to produce and distribute summaries or abstracts.

2. The rights to microfilm and microfiche editions or similar, to the use of the article and its contents in videotext, teletext, and similar systems, to recordings or reproduction using other media, digital or analog, including electronic, magnetic, and optical media, and in multimedia form, as well as for public broadcasting in radio, television, or other forms of broadcast.

3. The rights to store the article and its content in machinereadable or electronic form on all media (such as computer disks, compact disks, magnetic tape), to store the article and its contents in online databases belonging to the publisher or third parties for viewing or downloading by third parties, and to present or reproduce the article or its contents on visual display screens, monitors, and similar devices, either directly or via data transmission.

4. The rights to reproduce and distribute the article and its contents by all other means, including photomechanical and similar processes (such as photocopying or facsimile), and as part of so-called document delivery services.

5. The right to transfer any or all rights mentioned in this agreement, as well as rights retained by the relevant copyright clearing centers, including royalty rights to third parties.

Online Rights for Journal Articles: Guidelines on authors’ rights to archive electronic versions of their manuscripts online are given in the document ‘‘Guidelines on sharing and use of articles in Hogrefe journals’’ on the journal’s web page at www.hogrefe.com/ j/jpp

September 2016

Cultural diversity – challenge and opportunity “It’s a book that we were all waiting for, and will be useful not only to psychologist practitioners and students, but also to stakeholders and policy makers in education.”

Bruna Zani, Professor of Social and Community Psychology, Department of Psychology, Alma Mater Studiorum-University of Bologna, Bologna, Italy; EFPA Executive Council Member

Alexander Thomas (Editor)

Cultural and Ethnic Diversity How European Psychologists Can Meet the Challenges 2018, x + 222 pp.

US $56.00 / € 44.95 ISBN 978-0-88937-490-4 Also available as eBook Culture and diversity are both challenge and opportunity. This volume looks at what psychologists are and can be doing to help society meet the challenges and grasp the opportunities in education, at work, and in clinical practice. The increasingly international and globalized nature of modern societies means that psychologists in particular face new challenges and have new opportunities in all areas of practice and research.

The contributions from leading European experts cover relevant intercultural issues and topics in areas as di-

www.hogrefe.com

verse as personality, education and training, work and organizational psychology, clinical and counselling psychlogy, migration and international youth exchanges.

As well as looking at the new challenges and opportunities that psychologists face in dealing with people from increasingly varied cultural backgrounds, perhaps more importantly they also explain and discuss how psychologists can deepen and acquire the intercultural competencies that are now needed in our professional lives.

The first structured resource for psychologists that combines mindfulness with character strenghts Ryan M. Niemiec

Mindfulness and Character Strengths A Practical Guide to Flourishing

2014, xx + 274 pp. + CD with meditation exercises US $39.80 / € 27.95 ISBN 978-0-88937-376-1 Also available as eBook At the core of this hands-on resource for psychologists and other practitioners, including educators, coaches, and consultants, is MindfulnessBased Strengths Practice (MBSP), the first structured program to combine mindfulness with the character strengths laid out in the VIA Institute’s classification developed by Drs. Martin E. P. Seligman and Christopher Peterson. This 8-session program systematically boosts awareness and application of character strengths – and so helps people flourish and lead more fulfilling lives.

The author’s vast experience working with both mindfulness and character strengths is revealed in his sensitive and clear presentation of the conceptual, practical, and scientific elements of this unique combined approach. It is not only those who are new to mindfulness or to character strengths who will appreciate the detailed primers on these topics in the first

www.hogrefe.com

section of the book. And the deep discussions about the integration of mindfulness and character strengths in the second section will benefit not just intermediate and advanced practitioners. The third section then leads readers step-by-step through each of the 8 MBSP sessions, including details of session structure and content, suggested homework, 30 practical handouts, as well as inspiring quotes and stories and useful practitioner tips. An additional chapter discusses the adaption of MBSP to different settings and populations (e.g., business, education, individuals, couples). The mindfulness and character strengths meditations on the accompanying CD support growth and development.

This highly accessible book, while primarily conceived for psychologists, educators, coaches, and consultants, is suitable for anyone who is interested in living a flourishing life.