LEADER - MEMBER EXCHANGE SOCIAL COMPARISONS AND LMX - RELATED EMOTIONS : EXAMINING ATTRIBUTIONS AND INTERPERSONAL JUSTICE AS MODERATORS By Catherine Ott - Holland A DISSERTATION Submitted to Michigan State University in partial fulfillmen t of the requirements for the degree of Psychology Doctor of Philosophy 2015 ABSTRACT LEADER - MEMBER EXCHANGE SOCIAL COMPARISONS AND LMX - RELATED EMOTIONS : EXAMINING ATTRIBUTIONS AND INTERPERSONAL JUSTICE AS MODERATORS By Catherine Ott - Holland Research indicates subordinates view their relationship with their supervisor in comparison to the relationships their supervisor has with other subordinates ( Vidyarthi, Liden, Anand, Erdogan, & Ghosh, 2010 ) . Ho wever, the extent to which these social comparison s create different discrete LMX - related emotions is likely due to the casual attributions subordinates make regarding how their LMX quality came to be built and perceptions that the supervisor shows considerate interpersonal treatment . In this study , attribution theory was used to argue locus of causality for relationship building with supervisor w ould moderate the association between LMX social comparisons on subordinate LM X - related emotions (i.e ., pride, anger, guilt, shame, and gratitude) . I nterpersonal justice was expected to moderate the interaction between LMXSC and supervisor - attributed locus of causality for relationship building with the supervisor in predicting LMX - related anger . I n addition to drawing upon attribution theory, LMX - related emotions were framed using core affect theory (Russell, 2003) , which aids in describing how these emotions are similar and different from emotions discussed elsewhere in the affect literature. Results show that while many expectations based on attribution theory were supported, self - attributed locus of causality for RBS and interpersonal just ice had a surprisingly strong direct association with LMX - related emotions. Copyright by CATHERINE OTT - HOLLAND 2015 iv ACKNOWLEDGEMENTS First and foremost, I would like to express my immense gratitude to Dr. Ann Marie Ryan fo r her truly inspiring mentorship. In addition to rigorous research, consult effectively with organizations, and write papers in a clear, persuasive, and publishable manner, Ann Marie taught me to hold my work to high standards and constantly seek out ways of making my work richer, more polished, and more thoughtful . Ann Marie, thank you so much for all th at you put towards my development ! Thanks are also due to Dr. J. Kevin For d. In addition to sharing his expertise as a consultant and instructor , Kevin has always been both a listening ear and sounding board for students in the program as they experience the challenges of graduate student life. He has often helped me engage in scholarly and professional , those conversations . A big thanks to my committee as a whole (Drs. Ann Marie Ryan, J. Kevin Ford, Fred Morgeson, and John Schaubroeck) for your constructive (albeit challenging ! ) feedback. Your comments and suggestions have helped to make this dissertation more compelling and robust. Thank you to all my friends in People Analy tics at Google for their encouragement as I finished my dissertation. Thank you to Anuj Chandarana, Tina Malm, Abeer Dubey, and Prasad Setty for giving me the time and resources I needed to finish my PhD while beginning my dream job. Your support made a world of a difference in my quality of life during this process. There are several friends who provided me with support in my dissertation writing when I needed support the most . Thank you to Samantha Baard, Sarena Bhatia, Mengyang Cao, Rory Dixon , Christine Kermond, Danielle King , Mariah Kornbluh , and Melissa McDonald for b eing my dissertation writing buddies and some of my best people I know. v Thank yo u to Carol Tweten for taking on the arduous task of proofreading this massive document. Thank you to Christine Kermond, Brent Lyons, Melissa McDonald , and Jem Dillon for reading sections of my dissertation and providing me with feedback and suggestions. T hank you to Katie Corker and Melissa McDonald for initiating our writing group and keeping it alive . Many thanks to Espresso Royale, Strange Matter, Grand River Coffee (RIP), Chapelure, The Coffee House, and Red Rock Café for the steadfast supply of espresso drinks and for allowing me to camp out on a regular basis while I worked on my dissertation. The biggest note of appreciation goes to my parents. Your infinite wisdom , humor, and kindness has helped me every day of graduate school eged to be your daughter. vi TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ................. ix LIST OF FIGURES ................................ ................................ ................................ ............... x i i INTRODUCTION ................................ ................................ ................................ ................. 1 Introduction ................................ ................................ ................................ ...................... 1 Contribution to scholarly research ................................ ................................ ............. 3 Review of LMX Variability Constructs and Research Findings and Integration of Social Comparison Theory ................................ ................................ ....... 4 Leader - member exchange theory ................................ ................................ ............... 5 Overview of constructs used to study LMX variability ................................ ............. 6 Definitions ................................ ................................ ................................ ................. 7 Leader - member exchange differentiation ................................ ...................... 7 Perceptions of leader - member exchange differentiation ............................... 1 1 Relative leader - member exchange ................................ ................................ . 1 1 Leader - member exchange social comparison ................................ ................ 1 2 Operationalizations ................................ ................................ ................................ ... 1 2 Leader - member exchange differentiation ................................ ...................... 1 2 Perceptions of leader - member exchange differentiation ............................... 1 3 Relative leader - member exchange ................................ ................................ . 1 3 Leader - member exchange social comparison ................................ ................ 1 4 Framework for differentiating LMX variability constructs ................................ ....... 1 4 Why LMXSC? A social comparison perspective ................................ ...................... 1 5 Focus on LMXSC over LMX ................................ ................................ ........ 1 5 Focus on LMXSC over RLMX ................................ ................................ ...... 1 7 Focus on LMXSC over relative deprivation ................................ .................. 1 7 Hypotheses and Related Justifications ................................ ................................ ............ 1 9 Review of attributi on theory and locus of causality ................................ .................. 20 ........................... 22 Bridging different conceptualizations of emotions ................................ .................... 2 4 - related emotions ................................ ..................... 29 LMXSC and Locus of causality for RBS: LMX - related emotion as an outcome ..... 30 Interpersonal justice and group values model ................................ ............................ 40 METHOD ................................ ................................ ................................ .............................. 5 2 Pilot study to validate locus of causality for RBS measure ................................ ............. 5 2 Pilot study sample ................................ ................................ ................................ ...... 5 2 Measures ................................ ................................ ................................ .................... 5 3 ................. 5 3 Main study ................................ ................................ ................................ ....................... 70 Sample and procedures ................................ ................................ .............................. 70 Sample size ................................ ................................ ................................ ................ 71 vii Careless response identification ................................ ................................ ................. 7 2 Sample characteristics ................................ ................................ ................................ 7 3 Survey 1 measures ................................ ................................ ................................ ..... 7 4 LMXSC ................................ ................................ ................................ ................ 7 4 ................. 7 5 Leader - Member Exchange ................................ ................................ ................... 7 5 Demographics ................................ ................................ ................................ ...... 7 6 Survey 2 measures ................................ ................................ ................................ ..... 7 6 Interpersonal justice ................................ ................................ ............................. 7 6 LMX - related emotions ................................ ................................ ......................... 7 7 Positive and negative affect ................................ ................................ ................. 81 RESULTS ................................ ................................ ................................ .............................. 8 2 Hypothesis tests ................................ ................................ ................................ ............... 9 3 Ancillary a nalyses: Comparing E motions and PANA models ................................ ........ 10 6 Ancillary analyses: LMXSC and its relationship with LMX ................................ ........... 1 10 Ancillary analyses: Comparing I nformation al J ustice and I nterpersonal J ustice ............ 1 21 Ancillary analyses: Race/Ethnicity ................................ ................................ .................. 12 5 Ancillary analyses: Workgroup size ................................ ................................ ................ 12 6 ................................ ... 12 8 Ancillary analyses: LMXSC x interpersonal justice interaction predicting supervisor - attributed locus of causality for RBS ................................ ............................. 135 Ancillary analyses: Hypothesized interactions with alternative emotions and locus of causality entered as controls ................................ ................................ ........ 13 7 DISCUSSION ................................ ................................ ................................ ........................ 13 9 Findings from hypothesis tests ................................ ................................ ......................... 1 40 Ancillary analyses: Comparing emotions and PANA models ................................ ......... 14 6 Ancillary analyses: LMXSC and its relationship with LMX ................................ ........... 14 7 Ancillary analyses: Comparing I nformation al J ustice and I nterpersonal J ustice ............ 14 8 Ancillary analyses: Workgroup size ................................ ................................ ................ 14 8 ................................ ... 14 9 Theoretical Implications and Future Directions ................................ .............................. 1 51 Practical Implications ................................ ................................ ................................ ....... 15 5 Limitations ................................ ................................ ................................ ....................... 15 6 Conclusion ................................ ................................ ................................ ....................... 15 7 APPENDICES ................................ ................................ ................................ ....................... 15 8 Appendix A: Pilot Survey Consent Form ................................ ................................ ........ 15 9 Appendix B: Example Causal Attribution Measures ................................ ....................... 1 60 ........ 1 6 3 Appendix D: Pilot Survey Measures with Revisions Marked ................................ ......... 16 4 Appendix E: First Consent Form for Online Panel Participants ................................ ...... 16 6 Appendix F: Second Consent Form for Online Panel Participants ................................ .. 16 8 Appendix G: End of Survey Message for Online Panel Participants .............................. 16 9 Appendix H: Leader - Member Exchange Social Comparison ................................ ......... 1 70 viii Appendix I: Leader - Member Exchange ................................ ................................ ........... 1 71 Appendix J: Demographic Questions ................................ ................................ .............. 1 7 2 Appendix K: Interactional Justice ................................ ................................ .................... 1 7 3 Appendix L: LMX - related Emotions ................................ ................................ ............... 17 4 Appendix M: Positive and Negative Affectivity ................................ ............................. 17 5 Appendix N: Example Emotion Measures ................................ ................................ ...... 17 6 Appendix O: Item Statistics of All Main Study Items ................................ ..................... 1 8 2 Appendix P: Factor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study Items ........................ 1 9 2 Appendix Q: Factor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study Predictor Items ......... 194 Appendix R: Descriptive Statistics of All Main Study Items by Race/Ethnicity ............ 19 5 Appendix S : Descriptive Statistics of All Main Study Items by Workgroup Size .......... 19 6 Appendix T : Descriptive Statistics of All Main Study Items by Supervisor Relationship Tenure ................................ ................................ ................................ ......... 19 8 REFEREN CES ................................ ................................ ................................ ...................... 19 9 ix LIST OF TABLES Table 1 Definitions and operationalization of LMX variability constructs .......................... 8 Table 2 Categorization of LMX variability constructs by theory level and subjectiv ity ................................ ................................ ................................ ............................. 15 Table 3 Summary of Pilot Expectations and Findings ................................ .......................... 5 8 T able 4 Promax Rotated Maximum Likelihood Extraction Pattern Matrix for Pilot Items ................................ ................................ ................................ ................................ ....... 5 9 Table 5 Factor Correlation Matrix Based on Promax Rotate d Maximum Likelihood Extraction Factor Analysis of Pilot Items ................................ ................................ .............. 61 Table 6 Pilot measures descriptive statis tics and intercorrelations ................................ ....... 6 4 Table 7 Example responses to ASQ main cause open text ................................ ................... 6 7 T able 8 Descriptive Statistics and Mean Differences of Pilot Scale Measures by ASQ Main Cause Comment Coding ................................ ................................ ................................ ........ 6 8 T able 9 Promax Rotated Maximum Likelihood Extraction Pa ttern Matrix for All Items in the Main Study ................................ ................................ ................................ ............................. 8 4 Table 10 Promax Rotated Maximum Likelihood Extraction Pattern Matrix for Main Study Predictors ................................ ................................ ................................ ............................... 8 6 Table 11 Descriptive Statistics, Reliabilities, and Correlations for Main Study Measures .. 8 9 T able 12 Summary of Hypothesis Findings and Ancillary Analyses ................................ ... 9 2 Table 13 Tests of Hypothesis 2 and 3 with Emotions Regressed on LMXSC, Locus of Causality for RBS, and an LMXSC x Locus of Causality Interaction ................................ .. 9 5 Table 14 Tests of Hypothesis 5, with Gratitude and Anger Regressed on Interpersonal Justice, LMXSC, and an Interactional Justice x LMXSC Interaction ................................ ................ 9 6 Table 15 Tests of Hypothesis 6, with Anger Regressed on a Interpersonal Justice x LMXSC x Supervisor Attributed Locus of Causality for RBS Interaction ................................ .......... 10 4 T able 16 Hypothesis 6 Follow - up Regression, with Anger Regressed on an Interpersonal Justice x Locus of Causality Interaction ................................ ................................ ................ 10 5 x Table 17 Ancillary tests of Hypotheses 2 and 3 with Positiv e Affect and Negative Affect as D Vs ................................ ................................ ................................ ................................ .... 10 8 Table 18 Ancillary tests of Hypothesis 5 with Positive A ffect and Negative Affect as DVs ................................ ................................ ................................ ................................ .... 10 8 Table 19 Ancillary test of Hypothesis 6 with Negative Affect as DV ................................ .. 10 9 T able 20 Promax Rotation Principal Components Analysis Structure for Main Study LMXSC , LMX, and Interactional Justice Items ................................ ................................ .................... 1 1 2 Table 21 Component Correlation Matrix Based on Promax Rotated Principal Components Analysis for Main Study LMXSC, LMX, and Interactional Justice Items ............................ 1 1 3 Table 22 Ancillary tests of Hypothesis 2 and 3 with LMX in place of LMXSC ................. 11 5 Table 23 Ancillary tests of Hypothesis 5 with LMX in place of LMXSC ........................... 11 9 T able 24 Ancillary test of Hypothesis 6 with LMX in place of LMXSC ............................. 1 21 Table 25 Ancillary Tests of Hypothesis 5 using Informational Ju stice in place of Interpersonal Justice ................................ ................................ ................................ ................................ ..... 1 2 2 Table 26 Ancillary Test of Hypothesis 6 with Informational Justice in place of Interpersonal Justice ................................ ................................ ................................ ................................ ..... 12 4 Table 27 Informational Justice x Supervisor - attributed Locus of Causality for RBS predicting Anger ................................ ................................ ................................ ................................ ...... 12 4 T able 28 Regression with LMXSC x Number of Peers Interaction Predicting Shame ........ 12 7 Table 29 Regression with LMXSC x Supervisor Relationship Tenure Interaction Pre dicting Prid e ................................ ................................ ................................ ................................ . 12 9 Table 30 Regression with LMXSC x Self - attributed Locus of Causality for RBS x Supervisor Relationship Tenure Interaction Predicting Shame ................................ ............................... 1 31 Table 31 Regression with LMXSC x Interpersonal Justice x Supervisor Relationship Tenure Interaction Predicting Anger ................................ ................................ ................................ .. 1 3 3 T able 32 Regression with LMXSC x Interpersonal Justice Predicting S upervisor - attributed Locus of Causality for RBS ................................ ................................ ................................ ... 13 6 Table 33 Item statistics of all main study items ................................ ................................ .... 18 2 xi T able 34 Factor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study ................................ ................................ .............................. 19 2 Table 35 Factor Correlation Matrix Based on Promax Rotated Maximum Lik elihood Extraction Factor Analysis of Main Study Predictor Items ................................ ................................ .... 19 4 Table 36 Descriptive Statistics of All Main Study Items by Race/Ethnicity ........................ 19 5 Table 37 Descriptive Statistics of All Main Study Items by Workgroup Size ..................... 19 6 T able 38 Descriptive Statistics of All Main Study Items by Supervisor Relationship Tenure ................................ ................................ ................................ ............... 1 9 8 xii LIST OF FIGURES Figure 1 Study Model ................................ ................................ ................................ ........... 2 0 Figure 2 Hypothesized interaction between LMXSC and Self - attributed Locus of Causality predicting LMX - related Pride ................................ ................................ ................................ 34 Figure 3 Hypothesized interaction between LMXSC and Self - attributed Locus of Causality predicting LMX - related Guilt/ Shame ................................ ................................ .................... 35 Figure 4 Hypothesized interaction between LMXSC and Supervisor - attributed Locus of Causality predicting LMX - related Gratitude ................................ ................................ ......... 3 9 Figure 5 Hypothesized interaction between LMXSC and Supervisor - attributed Locus of Causality predicting LMX - related Anger ................................ ................................ .............. 40 Figure 6 Hypothesized interaction between LMX - related Gratitude and Interpersonal Justice predicting LMX - related Gratitude ................................ ................................ ......................... 4 5 Figure 7 Hypothesized interaction between LMXSC and Interpersonal Justice predicting LMX - related Anger ................................ ................................ ................................ .......................... 4 6 Figure 8 Hypothesized interaction between LMXSC, Supervisor - attributed Locus of Causality, and Interp ersonal Justice predicting LMX - related Anger: High interpersonal justice .......... 50 Figure 9 Hypothesized interaction between LMXSC, Supervisor - attributed Locus of Causality, and Interpersonal Justice predicting LMX - related Anger: Low interpersonal justice ........... 50 F igure 10 LMXSC x Locus of Causality Interaction Predicting Pride ................................ . 9 7 Figure 11 LMXSC x Locus of Causality Interaction Predicting Shame .............................. 9 8 Figure 12 LMXSC x Locus of Causality for RBS interaction predicting Gratitude ............ 100 Figure 13 LMXSC x Locus of Causality for RBS Interaction Predicting Anger ................. 101 Figure 14 LMXSC x Interpersonal Justice Interaction Predicting Anger ............................. 10 3 Figure 15 Supervisor - attributed Locus of Causality for RBS x Interpersonal Justice Interaction Predicting Anger ................................ ................................ ................................ .. 10 5 Figure 16 LMX x Locus of Causality for RBS Interaction Predicting Shame ..................... 11 6 Figure 17 LMX x Locus of Causality for RBS Interaction Predicting Gratitude ................. 11 7 xiii Figure 1 8 LMX x Locus of Causality for RBS Interaction Predicting Anger ...................... 11 8 Figure 1 9 LMX x Interperson al Justice Interaction Predicting Anger ................................ . 1 20 Figure 20 LMXSC x Informational Justice Interaction Predicting Anger ............................ 1 2 5 Figure 21 LMXSC x Informational Justice Interaction Predicting Anger ............................ 12 5 Figure 22 LMXSC X Number of Peers Interaction Predicting Shame ................................ . 12 7 Figure 23 LMXSC X Supervisor Relationship Tenure Predicting Pride .............................. 1 30 Figure 24 LMXSC x Self - attributed Locus of Causality for RBS x Supervisor Relationsh ip Tenure Interaction Predicting Shame ................................ ................................ ..................... 132 Figure 25 LMXSC x Interpersonal Justice x Supervisor Relationship Tenure Interaction Predicting Anger ................................ ................................ ................................ .................... 13 4 Figure 26 LMXSC x Interpersonal Justice Interaction predicting Supervisor - attributed Locus of Causality for RBS ................................ ................................ ................................ .................. 13 7 1 INTRODUCTION Introduction Employees experience many advantages when they have h igh - quality exchange relationships with their supervisor . Effective leader - member exchange (LMX) relationship s come with a variety of benefits, including trust, obligation, affect ive loyalty, professional respect, information, influence, support, voice, positive performance appraisals and career progress ( Liden, Erdogan, Wayne, & Sparrowe, 2006 ) . 1 However, effective LMX is a limited resource (Liden, Sparrow, & W ayne, 1997) , and s upervisors often do not have the time or emotional resources to develop high quality relationships with all subordinates. Given effective LMX is a limited but valued asset , there is reason to assume subordinates compare their relationship with the supervisor to the relationships other subordinates have with the supervisor (Vidyarthi, Liden, Anand, Erdogan, & Ghosh, 2010) . Research on social comparison theory suggests that individuals regularly look at the surroun ding social context, and the resulting social comparisons influence emotions and attitudes (Wood, 1989). Extending this to the workplace context, when subordinates compare their LMX quality to the LMX quality held by other peers working for the same superv isor , these social comparisons are likely to influence how subordinates feel about their LMX. Social comparisons have a powerful capacity to influence how subordinates feel at work (Greenberg, Ashton - James, & Ashkanasy, 2007 ) . Research has shown that whe n subordinates feel they have a better than relationship with the supervisor than their peers, they report higher pay satisfaction, work satisfaction, and leader satisfaction (study 2; Erdogan, 2002). When LMX 1 Within the LMX literature, superiors, leaders, managers, and supervisors are often referred to interchangeably, as are followers, members, and sub ordinates. Here, I rely on the terms supervisor and subordinate to maintain consistency for the reader and to emphasize the relationship is direct. I use the terms leader and member only in situations that directly reference the LMX construct. 2 social comparison perceptions influence LMX - re lated emotions , these emotions are likely to influence subordinate attitudes and behaviors that play a role in the LMX relationship, the workgroup, and the organization. Thus, an important question is, when and how do LMX social comparisons result in diffe ren t LMX - related emotions ? The relationship between these social comparisons and the way subordinates feel about their LMX likely depends on the attributions subordinates make regarding how their LMX came to be established . If subordinates perceive the ir LMX quality is highly influenced by their behavior , perceptions of a relatively poo r quality LMX relationship could translate in to LMX - related guilt or shame , whereas perceptions of a relatively high quality LMX relationship could result in feelings of LMX - related pride . Similar attributions may be made regarding the role of the supervisor in LMX relationship building. If s ubordinates think their LMX quality is highly influenced by sup ervisor behavior , perceptions of a relatively high quality LMX relationsh ip could result in LMX - related gratitude and perceptions of a relatively low quality LMX relationship could result in LMX - related anger . Further, the extent to which supervisor interactions are characterized by dignity and respect ma y buffer the negative effects of anger - inducing contexts. The purpose of this study is to understand how and when leader - member exchange social comparisons generate discrete LMX - related emotions . The focal question is : what links a perception of having high (or l ow) standing with the supervisor relative to other subordinates to different LMX - related emotions ? To expand our understanding of this area, I use d attribution theory as a theoretical lens to suggest locus of causality for relationship building supervisor moderates the influence of LMX social comparisons on LMX - related emotions . I also propose interpersonal justice moderates this interaction, such that the anger felt 3 by individuals who view their supervisor as driving their LMX quality and feel they have a relatively poor quality LMX will be buffered when the supervisor shows high interpersonal justice. Contribution to scholarly research This study makes three core contributions to the scholarly literature. First, r ecent research has sought to expand our understanding of how and when variance in the quality of within - group LMX relationships elicit s positive (or negative) outcomes ( e.g. , Haynie, Cullen , Lester, Winter, & Svyantek, 2014; Li & Liao, 2014 ) . Insight into how and when LMX social comparisons influence LMX - related emotions at work will contribute to our understanding of the psychological mechanisms that link differentiated LMX to LMX - related emotions. Second, this study integrate s attribution theory (e.g. Weiner, 1985) and LMX theory ( Graen & Scandura, 1987 ) to examine how and when LMX - related attributions moderate the associations between LMX social comparisons and LMX - related emotions. Attribution theory is supported by extensive past research (cf. Kelle y & Michela, 1980) and it suggest s meaningful mechanisms that could explain how and when LMX social comparisons lead to different levels of LMX - related emotions. LMX theory is extended by attribution theory to make predictions about how and when LMX - related social comparisons elicit different emotions. Third, the current study advances existing LMX resear ch by using core affect theory to more narrowly define the discrete emotions of interest within a specific conte xt . Th is treatment of affect advances past research in several ways . First, existing LMX research has tended to focus on positive and negative affect. E motions differ from negative and positive affect in that emotions are formed in response to focal - target s , and can be focused internally (i.e. self - conscious emotions) or externally . In this sense, emotions carry more information about 4 perceptions of the social context than do positive or negative affect. Second, approach to emotion diff ers in many ways from other areas of research on affect. C ore affect theory as a framework allows for a more detailed description of the affective components that likely underlie LMX - related emotions, and how these components might differ from affect as it is referred to in other affect research. The introduction is outlined as follows. I first review how prior research has studied within - group variability in LMX relationships. I then describe how leader - member exchange social comparison ( LMXSC ; Vidyart hi et al., 2010 ) , the focal LMX variability construct in this study, relates to social comparison theory. Research on attribution theory is reviewed , and a new construct, locus of causality for (locus of causali ty for RBS) , is proposed . I use core affect theory to define LMX - related emotions and then explicate how locus of causality for RBS interacts with LMXSC to elicit different levels of emotions. Research on interpersonal justice and the fair process effect is drawn upon to hypothesize a three - way interaction hypothesis between LMXSC, locus of causality for RBS, and interpersonal justice on LMX - related anger. Review of LMX Variability Constructs and Research Findings and Integration of Social Comparison Theor y To more fully describe the construct of LMXSC and the reasoning behind its inclusion in the present study, this section proceeds as follows. First, LMX theory is briefly described. I then review the d efinitions and operationalizations of existing within - group LMX variability constructs ( e.g. , LMX differentiation, perceived LMX differentiation, relative LMX, and LMX SC ). A framework is put forth than distinguishes these constructs by level of theory (i.e. , individual - in - context versus group) and perceptual status ( i.e. , subjective vs. objective) . Last, I 5 further discuss LMXSC , how it relates to social comparison theory, and how it differs from related but separate constructs (i.e. , relative deprivation). Leader - member exchange theory Leader - member e xchange describes the dyadic process through which supervisor s establish and routinize roles over time ( Graen & Scandura, 1987; Graen & Uhl - Bein, 1995). dinate, dyads become organized so that unstructured tasks can be effectively managed and completed. The exchange relationship transitions through three different phases: role taking, role making, and role routinization. In the initial stage , role taking , the supervisor sends a role to the member, who receives this role. The subordinate then interprets the role and sends it back to the supervisor. This stage in the exchange is purely economic , wherein the supervisor l evels. Role making occurs after the initial roles have been accepted (Graen & Cashman, 1975). During role making, the supervisor and subordinate further define roles as valued resources are exchanged. As behaviors ju dged effective are strengthened and thos e deemed ineffective are weakened, the dyadic relationship stabilizes into more routine behaviors. This more stable , final stage is labeled role routinization (Graen & Scandura, 1987). Meta - analytic research indicates high quality LMX relates to a variet y of positive subordinate outcomes. Subordinates with high quality LMX tend to enjo y their work experiences more, reporting higher levels of job satisfaction ( =.49, k =88, N =22,520, 95% CI lower=.45, upper=.53; Dulebohn, Bommer, Liden, Brouer, & Ferris, 2012), supervisor satisfaction ( =.68, k =32, N =11,195, 95% CI lower=.57, upper=.79; Dulebohn et al., 2012), and organizational commitment ( =.47, k =58, N =14,208, 95% CI lower=.43, upper=.51; Dulebohn et al., 2012). 6 Further, high quality LMX subordinates a re able to more effectively engage with the work environment, showing higher job performance ( =.34, k =108, N =25,322, 95% CI lower=.30, upper=.37; Dulebohn et al., 2012 ), and increased levels of citizenship behavior ( =.39, k =27, N =25,322, 95% CI lower=.30 , upper=.37; Dulebohn et al., 2012). Because of the benefits associated with high quality supervisor - subordinate exchanges, subordinates are likely to view h igh quality LMX as desirable. Overview of constructs used to study LMX variability One of the major contributions LMX theory has made to the leadership literature is the idea that supervisors have different types of relationships with specific subordinate s (Liden et al., 2006). In other words, when supervisors focus on the needs of individual subordinates , supervisor - subordinate relationships are likely to differ in quality and nature. LMX research has looked at the array of LMX dyads that exist within teams in several different ways. To review the construct definitions and operationalizations from this subdomain of LMX research , a ProQuest database search was conduct ed to find scholarly articles and dissertations containing currently focu ses on four k ey constructs: LMX differentiation, perceived LMX differentiation, LMX social comparison (LMXSC), and relative LMX (RLMX). T able 1 summarizes exemplar definitions of these four constructs. Operationalizations are listed with s pecific articles and disserta tions using each operationalization cited . 2 Both definitions and operationalizations are discussed below. 2 Redundan t definitions particularly those found in unpublished dissertations were excluded from the list of definitions. The list of operationalizations reflects all articles and dissertations within the search results that measured these constructs. 7 Definitions Leader - member exchange differentiation . Definitions of LMX differentiation found in the current search tended to emphasize that the construct occurs at the group - level and reflects the variance of within - summarizes these aspects of LMX differentiation and was cited regularly by articles and dissertations focusing on LMX differentiation. 8 Table 1 Definitions and operationalization of LMX variability constructs LMX Differentiation Example definitions Operationalizations Citations using operationalization Harris, Li, & Kirkman (2014) - level construct that refers to the degree of within - group; operates as a contextual construct that provides members with information that helps form justice perceptions and develop identities as group Haynie, Cullen, Lester, Winter, & Svyantek (2014) "A prev alent organizational phenomena in which supervisors form relationships of differing quality with their subordinates. Supervisors may treat their subordinates differently due to time and resource constraints, leader - member similarities, or equitable treatme Henderson, Liden, Glibkowski, & Chaudhry (2009) engaging in differing types of exchange patterns with subordinates, forms different quality exchange relationships (ranging from low to high) with them." interactive exchanges that occur between leaders and members, the nature of which (transactional versus social exchange) may differ across dyads within a work group." Liden, Erdogan, W ayne, & Sparrowe (2006) Within - group variance/ SD Anand (2012)* Chen, Yu, & Son (2014) Choi (2013)* Ford & Seers (2006) (item SDs averaged together) Erdogan & Bauer (2010) George (1999)* Han & Bai (2014) Haynie, Cullen, Lester, Winter, & Svyantek (2014) Henderson, Wayne, Shore, Bommer & Tetrick (2008) Han (2014)* Kwak (2011)* Lau (2008)* Le Blanc & González - Romá (2012) Liao, Liu, & Loi (2010) Liden, Erdogan, Wayne, & Sparrowe (2006) Ma & Qu (2010) Mayer (200 5 )* Naidoo, Scherbaum, Goldstein, & Graen (2011) Nishii & Mayer (2009) Stewart & Johnson (2009) Willis (2014)* Sum of absolute differences between individual score and group mean McClane (1991) R wg Boi es & Howell (2006) Average Deviation Tordera & González - Romá (2013) 9 Table Perceived LMX Differentiation Example Definitions Operationalizations Citations using operationalization Hooper & Martin (2008) relationships are Rate the LMX relationship quality of each of their team members (including themselves), compute mean and SD of these ratings Henderson, Wayne, Shore, Bommer, & Tetrick (2008) Relative Leader - Member Excha nge (RLMX) Example Definitions Operationalizations Citations using operationalization Henderson, Wayne, Shore, Bommer, & Tetrick (2008) Harris, Li, & Kirkman (2014, p. 317) and LMXSC, on the other hand, are disparity measures of group heterogeneity because they contain meaningful information about differences in status (i.e. , high RLMX or LMXSC indicates a higher quality relationship Subtract the mean i ndividual - level LMX score within a group from each group member's composite LMX score Henderson, Wayne, Shore, Bommer, & Tetrick (2008) Leader - Member Exchange Social Comparison (LMXSC) Example Definitions Operationalizations Citations using operationalization Vidyarthi, Liden, Anand, Erdogan, & Ghosh (2010, p. 850) - group social comparison with work group members as the reference point, LMX has no social reference obtained directly from focal employees, and [RLMX] is actual degree to which the focal individual's LMX differs from the average leader - 6 - item social compariso n measure Vidyarthi, Liden, Anand, Erdogan, & Ghosh (2010) Note. *indicates a dissertation. 10 Seve ral researchers expanded on the potential causes of LMX differentiation . These definitions tended to focus on the supervisor as causing LMX differentiation, but further description of LMX differentiation emergence alludes to other contributing factors. For example, Henderson, Liden , Glibkow ski, and Chaudhry (2009) posit LMX differentiation is caused by the supervisor different quality exchange relations hips (ranging from low to high) (p. 519) , but then expand this causation to suggest LMX differentiation results from (p. 519) between the supervisor and subordinate . Further, Haynie et al. (2014) suggested that LMX differentiation arise s as supervisors form relationships of differing quality with their subordinates (p. 914). However, Haynie et al. (2014) go on to suggest that differentiated relationships could also resu lt from co p. 914 ), a - p. 914 ) , or variance in p. 914 ). Thus, definitions of LMX differentiati on suggest there are different possible causes that could contribute to differentiated LMX relationships. Harris, Li, and Kirkman also posits potential outcomes of LMX differentiation. The researchers suggest LMX differentiation prov id es contextual information which them form justice perceptions and deve lop identities as group members (p. 316). This definition implies that within - group variance of LMX relationships should in part contribute to perspective s on their group membership . To summarize, LMX differentiation is typically defined as a group level construct, focusing on within - group variance in the quality of LMX rela tionships. Although supervisor s are assumed to determine the extent to which within - group LMX relationships vary in the group , definitions allude to other important determinants, such as subordinate behavior. LMX 11 differentiation is also described as influencing how group members view their membership identities . Perceptions of leader - mem ber exchange differentiation. Perceived LMX differentiation was an in frequently exa mined LMX variability construct. The only definition found, from Hooper and Martin (2008), describes the construct primarily as a perception of LMX Thus, perceived LMX differentiation can be viewed as individual perceptions of a group descriptor (that is, differentiated LMX relationships as a quality of the group). Relative leade r - member exchange. Several studies within the search results examined relative leader - member exchange (RLMX). Henderson, Wayne , Shore, Bommer, and Tetrick (2008) describe a work Henderson et al. (2008, p. 1210) later refer red to RLMX as occurring at the individual - within - group level of theory (Kelley, 1968), often referred to as a frog - pond approach (Firebaugh, 1980). In comparison to group - level constructs such as LMX dif ferentiation, individual - within - normative within the group (Henderson et al., 2008). Social comparison theory , discussed later, suggests that individuals are motivated to understand t heir relative standing and this information influences attitudes and behaviors. Harris et al. framework for different types of diversity indices, RLMX (and LMXSC, discussed next) would be considered disparity measures. Harrison and Klein (2007) describe disparity measures as reflecting differences in the concentration of valued social assets. In the LMX context, possessing a high ing for the average level of LMX, may be 12 viewed as having a higher concentration of a valued resource (e.g. , receiving more attention, feedback, and opportunities from the supervisor ) th an other members of the group. Leader - member exchange social comparis on. Last, LMXSC was mentioned by several of the articles and dissertations in the search. Vidyarthi et al. (2010) defined it as a social s own LMX and that of coworkers (p . 850). Similar to the relationship between perceived L MX differentiation and LMX differentiation, Vidyarthi et al. (2010 , p. 850 ) employees X differs from the ave rage leader - on subordinate - in - group level RLMX. A comparison between LMXSC and LMX was also made by Vidyarthi and colleagues (2010). The researchers p osit ed within - group social comparison with work group members as the ref erence point (p. 850). This definition is distinguished from LMX, which reflects the quality of the exchange relationship betwee n supervisor and subordinate ive judgment or reference point (p. 850). Next, I discuss the operationalizations of these four constructs and propose an organizing framework for explaining their similarities and differences. Operational izations Leader - member exchange differentiation . LMX d ifferentiation was the most commonly studied construct across the search results, and perhaps for that reason, there were several different operationalizations used to measure it. The majority of the studies utilized the standard deviation or variance of LMX scores with in a given work group (21 total). That is, 13 subordinate s within a work group were assessed using an LMX measure, such as the LMX - 7, and their scores were aggregated in order to obtain the variance. Several alternative approaches were used to measur e LMX differentiation. McClane (1991) summed the absolute differences between individual scores and the group mean. Boi es and Howell (2006) used R wg scores, arguing that these scores serve as an indicator of within - group agreement. To r dera and González - Romá (2013) similarly used a measure of agreement , Average Deviation. Despite these alternative approaches to assessing LMX differentiation, the use of standard deviation and variance was overwhelmingly normative. Perceptions of leader - member exchange differ entiation . Looking to research on perceived LMX differentiation, t hree . In the measure, subordinate s rate the LMX quality of all subordinates of their direct supervisor on a one to five scale of LMX quali ty , including the quality of their own LMX relationship with their supervisor. The standa rd deviation of these scores are then computed for each individual and taken as a n indication of perceived LMX differentiation . For example, if a person worked with five other subordinates of the same supervisor, that person might think his or her relationship is good (a rating of 4), whereas two coworkers have satisfactory ratings (ratings of 3), and three coworkers have very go would be SD =.98, which would then be used as an index of perceived LMX differentiation within the group. Relative leader - member exchange. elative to others in the group. To assess this, Henderson et al. (2008) sub tracted the group mean level of LMX from each subordinate - reported LMX score. Thus, even if LMX scores within a 14 group are generally low or high on average, RLMX scores still reflect whether self - reported LMX scores are higher or lower than others within the group. Leader - member exchange social comparison. Because LMXSC is a social comparison et al. (2010) developed a six - i tem measure of LMXSC that asks about these perceptions more directly. Items in this measure include , working relationship I have with my manager is more effective than the relationships most re loyal to me compared to Framework for differentiating LMX variability constructs To organize these four constructs into a unified framework, I categorized each by level of theory and perceptual status . The resulting f ramework can be found in Table 2 . Here I categorize LMX differentiation and perceived LMX differentiation as group - focused constructs, whereas RLMX and LMXSC are considered as frog - pond or individual - within - group constructs. The constructs are also organized by subjective (perceptions) and objective (using actual aggregated data to unders tand the influence of the group ) . According to this framework, RLMX and LMXSC are related in that they both focus on the individual - within - group . Empirical r esearch supports their linkage : Vidyarthi et al. (2010) found RLMX and LMXSC to show a strong positive relationship ( 30 - 30 = .79, p <.01). Although both LMX diffe rentiation and perceived LMX differentiation focus on group - level variability and thus a similar relationship would be expected, no tests of this relationship were found. 15 Table 2 Categorization of LMX variability constructs by theory level and subjecti v ity Frog - Pond/Individual - in - Context Group - Level Objective Relative Leader - Member Exchange (RLMX ) LMX Differentiation Subjective Leader - Member Exchange Social Exchange (LMXSC ) Perceived LMX Distribution Why LMXSC? A social comparison perspective the comparison between one's own LMX and that of coworker s. (p. 850). Focus on LMXSC over LMX. I focus on LMX social comparisons here instead of examining LMX for two main re asons. First, within - group social comparisons are assumed to stimulate strong er emotional and attitudinal r eactions from subordinate s than LMX evaluations that do not explicitly incorporate a social comparison aspect . Supervisor s do not have the time, socioemotional, or tangible resources to have high quality LMX relationships with all subordinate s ( Dansereau , Graen, & Haga, 1975; Vidyarthi et al., 2010), making high quality LMX a valued but limited resource amongst subordinates . H aving access to that resource takes on a greater importance within the social context than it would in isolation. Whereas high LMX is linked to positive outcomes ( Dulebohn et al., 2012) , high LMXSC makes that relationship significant within the social co ntext and interpersonally contentious. Similarly, low LMXSC is like ly to incite especially strong negative affective and attitudinal outcomes. Social comparison 16 , and thus, should be associat ed with stronger emotions and attitudes. S ocial comparison theory further supports the notion that social comparisons are made frequently and have a strong influence on attitudes and behaviors. Social comparison theory was originally put forth by Festinger (1954), and has since been the focus of extensive research in social psychology ( e.g., Wood, 1989, ( p p. 520 - 521). Social comparison theory posits people are motivated to evaluate their opinions and abilities, to make accurate assessments of oneself (self - evaluation), and to improve (self - improvement ) self - image (self - enhancement; Wood , 1989). When objective, non - social means of comparison are un available, people make social evaluations (Festinger, 1954). Empirical studies show work - related social comparisons are linked to employee attitudes and behaviors. For example, Brown, Ferris, Heller, and Keeping (2007) observed affective commitment and job satisfaction are negatively associated with upward social comparisons (e.g. , comparing oneself to colleagues who are better off) and positively relate d to downward comparisons (e.g. , comparin g oneself to colleagues who are worse off). Through the mediation of these job attitudes, employees who made upward comparisons were more likely to search out labor market alternatives to their current job, whereas employees making downward compari sons we re less likely to result in job search behaviors. Although LMX ratings may be influenced by within - group LMX social comparisons (and vice versa), LMXSC explicitly integrates and LMX does not. E mpirical research has also sup ported the notion that LMXSC perceptions are related but distinct 17 from LMX ratings (Vidyarthi et al., 2010). Accordingly, I expect that LMXSC would influence LMX - related emotions for reasons conceptually distinct from LMX. A second reason for examining LMXSC above and beyond LMX is that subordinate s are more likely to search for causal attributions when self - evaluations are made in a social context . S ocial contexts have been shown to both prompt social comparisons and influence attributions . For example, i n order to appear favorably within social contexts, successes are typically attributed to internal factors and failures to external factors; however, external attributions may be made towards successes in order to appear modest (Kelley & Michela , 1980). When subordinate s perceive their LMX differs substantially from the LMX held by other coworkers , they will be more motivated to seek out an explanation for their LMX quality . These attributions then also play a role in the linkages between LMXSC and LMX - re lated emotions. Focus on LMXSC over RLMX. RLMX captures information about the individual - in - context, similar to LMX SC . Yet a s Vidyarthi et al . (2010) note, because perceptions have a stronger influence on attitudes and behavio rs than actual differentiation, LMXSC is likely to have p articularly meaningful association with LMX - related emotions . Thus, I chose to focus on LMXSC perceptions instead of RLMX score s . Focus on LMXSC over relative deprivation. Another construct similar to LMXSC is relative deprivation. Research integrating relative deprivation with LMX has attempt ed to explain a problem space similar to th model (i.e., Bolino & Turnley, 2009) . R elative deprivation is defined as n someone who perceives a discrepancy Crosby, 1976, p. 56). Bolino and Turnley (2009) proposed that deprivation, which then leads to attitudinal and behavioral outcomes. Both research on relative 18 suggest that social comparisons could However, the present model diffe rs from extant attempts to integrate LMX with relative deprivation (i.e., Bolino & Turnley, 2009) in several ways . First, t he present model has greater construct clarity. does not make clear whet her relat ive deprivation is an affect, cognition , or both . Although some research has broken down the cognitive and affective parts of the process of perceiving relative deprivation ( e.g. , Smith , Pettigrew, Pippin, & Bialosiewicz, 2012), Bol ino and Turnley (2009) treat relative deprivation as a single construct. In contrast, th is process such that LMXSC (a cognitive comparison ) interact s with the locus of causality for relationship b uilding with supervisor (a cognitive attribution ) to influence specific LMX - related emotions (i.e., anger, guilt, shame, pride, and gratitude) . Thus, the present model breaks how Another distinction i s that models of relative deprivation tend to focus on upward social comparisons, but nowhere are downward comparisons addressed ( Bolino & Turnley, 2009; Smith et al., 2012 ). Relative deprivation assesses the extent to which individuals perceive they have be en disadvantaged, meaning the opposite end of the spectrum is a lack of being disadvantaged. Nowhere do these models explore the implications of feeling advantaged in comparison to others. In the present st LMX - related emotions as outcomes of having a relative disadvantage within the group (i.e., the upward comparison of low LMXSC) as well as the outcomes of relative advantage (i.e., the downward comparison of high LMXSC) . 19 In the next chapter, I use attribution theory (Weiner, 1985) and research on interpersonal justice to propose moderators of LMXSC - emotion linkages. Hypotheses and Related Justification s There is limited research on the outcomes of LMX variability constructs, and existing research is restricted to behaviors and attitudes as outcomes. Emotions have not been previously examined as outcome s of LMXSC, yet there is reason to suggest different e motions arise when social comparisons are made (Greenberg et al., 2007 ) . I extend attribution theory to suggest LMX - related emotions is in part determined by the perceptions and attributions made eader . The section proceeds as follows. First, I briefly review existing research on causal or (locus of causality for RBS). C ore affect theory is then used to describ e how emotions are conceptualized in this study . I then use attribution theory and research on fairness to describe the LMX - related emotions that would result from different combinations of LMXSC , interpersonal justice, and locus of causality for RBS. Rela ted hypotheses are put forth. For reference, these hypotheses are modeled in Figure 1. 20 Figure 1 Study Model Review of attribution theory and locus of causality Attribution theory is often used to refer broadly to research on the causal inferences people make and how these inferences influence cognitive, affective, and behavioral outcomes. However, attribution research encompasses a variety of different theories a nd a range of constructs (Kelley & Michela, 1980). Underlying this stream of research is the notion that causal attributions help to develop frameworks for guiding effective adaptation to new situations and environments (Kelley & Michela, 1980; Weiner, 198 5). Attribution theory as we know it today has been elaborated in the past 50 years, starting gesting that individuals vary in the extent to which they believe they can control the events occurring around them. Weiner et al. (1971) e xpanded this idea by suggest ing level of trait ability) or fluctuating (e.g., transient emotions). 21 Weiner (1985) summarized causal attribution research as focusing on three core properties: locus of causality , stability, and controllability. Locus of causality attributions involve judging the cause of an outcome to be either self - induced (internal locus of causality) or caused by another person or object in the surrounding context (external locus of causality). Stability refers to whether or not the attributed cause is expected to be permanent or transient. Con trollability refers to whether the attributed cause was entirely under the authority of the causal agent, or if on attributions of locus of causality, which can be viewed as th e first and most influential attribution made in the attribution process (Weiner, 1985). external attributions. If an employee receives a poor performance rating and blames h im or herself, that employee makes an internal attribution; if the employee blames other coworkers for his or her poor performance rating, that employee makes an external attribution. Here I make assumptions based on the extent to which subordinates view t heir behavior as determining the behavior as playing a causal role in LMX relationship building. It is important to distinguish between locus of causality and locus of control. Locus of control is a stable individual difference, whereas locus of causality refers to a more proximal Further, locus of control emphasizes whether an i ndividual perceives he or she has internal control over his or her environment, or if instead his or her environment is controlled by external factors. As Weiner (1985) notes, locus of control conflates perceptions of locus of causality and controllability . As an example, consider a student who fails an algebra test. There is no question 22 the student did. Thus, the student would be perceived to be the nder the past month, we would likely perceive this failure to be controllable. If the student was in 3 rd grade and was still trying to learn multiplication and division, we would likely view the failure as relationships are built with the supervisor. Expectations regarding controllability are only indi rectly made to the extent that locus attributions are made prior to attributions of controllability. That is, if a person views the supervisor as a greater causal agent in supervisor - subordinate relationship building than subordinates, the question of con trollability, while not answered, would focus more narrowly on the supervisor. interacts with LMXSC to elicit various subordinate LMX - related emotions. I define locus of a causal attribution concerning the extent to which a person or thing determines the quali ty of LMX developed within the dyad . I focus here on self - and supervisor - attributed locus of causality for RBS. How do these loc i of causality for RBS attributions develop? A number of factors likely likely the most important determinant of these attributions. If a supervisor appears t o selectively pay attention to a subordinate, that subordinate may begin to perceive that supervisor behavior 23 tends to drive (or halt) relationship building. If a subordinate has a poor quality LMX but is able to improve this LMX quality over time, that su bordinate may feel his or her own behavior is important for relationship building. toward the supervisor may too indirectly shape locus of causality for RBS attributions. For example, ind ividual differences such as locus of control may influence the extent to which subordinates display proactive behaviors within their supervisory relationships and in turn may affect self - attributed locus of causality for RBS. Similarly, cultural values su ch as power distance ( House et al., 2004) . Workgroup culture may also influence locus of causality for RBS attributions. Certain relationship building behaviors within a group are encouraged or rewarded, whereas others may be discouraged or even punished. In some groups, subordinate proactivity in seeking out more frequent and successful exchanges with the supervisor may be viewed as effective; in other groups, these behaviors may be seen as presumptuous and disrespectful. By focusing on supervisor - and self - attributed locus of causality for RBS, the current model applies attribution theory as a framework that expands our understanding of how LMXSC influences LMX - related emotions (i.e., pride, gratitude, guilt, shame, and anger). That is, self - attributed locus of causality for RBS is expected to elicit pride when LMXSC is high and guilt and shame when LMXSC is low. Supervisor - attributed locus of causality for RBS is expected to res ult in gratitude when LMXSC is high and anger when LMXSC is low. However, before discussing the rationale for these expected relationships, an expanded framework for conceptualizing emotions is needed. 24 Bridging differing conceptualizations of emotions At tribution theory is part of a research stream focusing on emotional appraisals and how sense - attribution framework conceptualizes emotions as a consequence of attributio nal thinking, but this treatment of emotion does not perfectly parallel the way emotions are defined in most affect to the broader affect literature, I frame affect theory. Core affect theory explicates the fundamental elements which underlie emotions, thereby providing an integrative framework that link s different ways of thinking about affect. In th is section, I compare and contrast how emotions are described in attribution theory with how they are described in other areas of affect research. Then, I describe core affect theory, present it as a unifying framework that bridges disparate conceptualiza tions of emotion, and describe how Current research on affect tends to focus on four levels of analyses: emotional traits, moods, emotions, and sensory experiences (Keltner & Lerner, 2010). Emotional traits reflect typical patterns in intraindividual affective experiences that display consistency across contexts. Moods represent diffuse positive ly or negatively valenced affective experiences that are without an immediate target and last for extended periods of time. Emotions, in contrast, are targeted towards an object in a given context (e.g. , feeling angry at someone) and are assumed to be briefer than moods. Sensory experiences are the shortest, most immediate affective exper iences, focusing on transitory pleasure and pain. In general, emotional traits are considered distal and stable, whereas moods, emotions, and sensory experiences are considered to be states that fluctuate with regularity ( Keltner & Lerner, 2010 ). 25 Of the se four levels of analysis, attribution theory focuses on emotions, because emotions represent responses to targets and therefore are therefore inherently linked to appraisals. At first glance, research on emotion appraisals is not inherently at odds with the majority of affect research; both literatures treat emotions as non - permanent and target - focused. Yet some confusion arises regarding the role of cognition. For example, anger has been manipulated in lab research by having study participants write an worst es & Baumeister, 1998; Bushman & Whitaker, 2010). In this scenario, anger towards the confederate or the situation in general is expected to have an abrupt onset with a n egative neurophysiological component. By comparison, attribution theory ascription of a negative, self - (Weiner, 1985, p. 562). That is, according to attribution theory, emotions directly result from cognitive sense - making, which could involve both automatic thought processes and more deliberate cognitions. Another area of confusion involves the propertie s of targets of emotion. In attribution theory, emotions can be targeted towards the quality of long - standing exchanges, processes, or developments. From an attribution - based perspective, an individual may feel proud of the hard work that went into complet regarding his or her high quality relationship with the supervisor. These emotions are different from feelings of pride after seeing an 'A' on a test, anger after a request is rejecte organization, or gratitude immediately after receiving a special privilege from the supervisor. 26 2003) core affect theory attempts to bridge these two conceptions of emotion. Core affect theory posits that emotions are not fundamental entities, but instead, are volved in emotions are core affect and affective quality. Core affect is a non - targeted - reflective feeling that is an integral blend of hedonic (pleasure - displeasure) and arousal (sl eepy - (Russell, 2003, p. 147). Core affe and remains in constant fluctuation. Second, the perception of affective quality - unpleasant and activating - deactivating qualitie core affect and affective quality are the most basic components of affect that, combined with information proce ssing and behavioral planning, comprise the majority of what people refer to as emotions (p. 148). A third entity that contributes to emotions is attributed affect . Attributed affect is the core loss, or liking a funny person at a party. Both evaluations and liking fit within the domain of attributed affect. Russell (2003) notes that these attributions are typically quick and automatic, but can be deliberate (p. 154). Attributed affect involves labeling what emotion is felt and what target it is felt towards. According to core affect theory, these three elements (a nd other less primary elements) be perceived in terms of affective quality, which may lead to a change in core affect and 27 subsequently a change in attributed affect. However, affective quality, core affect, and attribute d affect can occur in any order and are assumed to be constantly in flux. Furthermore, the ive element, but can be formed from any combination of these elements. Core affect theory suggests that when individuals categorize and report their emotions, they report emot ional meta - perceptions core affect, perceived affective quality, and attributed affect, but they are also subject to the fallibilities of information processing and are interpreted in the context of pr e - existing mental scripts. As Russell (2003) - experience is mediated by an interpretive Returning to the discrepancies in how emotions are conceptualized in attribution theory as compared to existing research on emotions, core affect theory would suggest both conceptualizations result in measures that reflect emotional meta - perceptions compri sed in part by affect attributions. That is, emotions in both attribution theory and research on emotions refer to a psychologically constructed experience that links core affect to a focal target. However, attribution theory and research on emotions tre at cognition and time different ly , because they emphasize different forms of information processing and different potential time frames of - perceptions that, to a greater extent than in attributions and involve information processing that works to summarize affect ive experiences across time (e.g. , feeling proud of the hard work that went into a project). Other areas o f research often focus on emotional meta - 28 underlying framework, in the present study I define LMX - related emoti ons as emotional meta - perceptions that reflect a summary evaluation of affective elements (e.g., core affect, perceived heavily influenced by affect attributions To understand the further review specific characteristics of each emotion within the model. [emotion] generated by appraisal that o ne is responsible for a socially valued outcome or for being a socially - context - specific pride (i.e. and not pride as a general positive view towards the self , 2000, p. 56) self - conscious emotions which require a self - appraisal whereas other emotions (e.g. , anger) do not necessarily require the same level of self - evaluation (Tracy & Robins, 2006). Guilt is thought to occur when individuals blame themselves for a negative asp ect of their behavior, whereas shame occurs when individuals blame themselves for a stable, negative aspect of themselves (Tracy & Robins, responsibility for wrongdoing b y another person or entity and often includes the goal of these emotions become important in understanding the predictions set forth by the model, discussed furth er in the following sections. It is important to note that model predictors are not emotions refer to little or no affective experiences typically associated with that emotion. 29 LMX - related emotions When subordinates make an assessment of whether their LMX is of a better or worse quality than the LMX quality of coworkers with the same supervisor, they are assessing the extent to which they have special privileges, opportunities, and resources that oth ers do not have. should elicit an affective reaction. A social comparison - affect linkage is supported by social comparison research. Specifically, the contrast effect refers to the observation that upwards comparisons tend to elicit increased negative affect, whereas downwards comparisons tend to elicit incre ased positive affect (Greenberg et al., 2007). Research has shown regardless of level of self - este em, moods are improved following a downward comparison ( Gerrard, Gibbons , & Boney McCloy 199 3 ). Furthermore, researchers have argued self - conscious emotions like shame, guilt, and pride are almost always the consequence of social comparisons (Baldwin & Bac cus, 2004). What is the mechanism underlying this effect? Buunk , Collins, Taylor, VanYperen, & Dakof ( 1990) suggest that when individuals make an upward comparison, they realize both that they are not as well off as others and that they could be better o ff than they presently are. Accordingly, those with low LMXSC should be more likely to feel shame, guilt, and anger towards their LMX, because they are likely to feel their lower levels of resources set them at a relative disadvantage. Conversely, when in dividuals make a downward comparison, they notice their relative advantage and observe they could be worse off than they presently are. This downward counterfactual contributes to positive emotions like gratitude and pride. Empirical research supports thi s notion: downward comparisons have shown associations with feelings of gratitude (Em mons & McCullough, 2003) and pride (Webster, Duvall, Gaines, & Smith, 2003). 30 For these reasons, LMXSC was expected to be positively associated with positive emotions (i. e., pride, gratitude) and negatively associated with negative emotions (i.e., guilt, shame, anger). H1: LMXSC will be associated with subsequent emotions, such that H1 a : LMXSC will be positively associated with positive emotions (pride, gratitude) H1 b : LMXSC will be negatively associated with negative emotions (guilt, shame, anger). LMXSC and Locus of causality for RBS: LMX - related emotion as an outcome Attribution theory suggests attributions are made when outcomes are perceived to be a success or f ailure. For subordinates, high LMXSC is arguably a perceived success. Attribution theory also posits that when individuals experience a successful outcome and attribute the success to their own behavior, they are especially likely to experience increased self - esteem (Brown & Marshall, 2001) and positive, self - related emotions (Weiner, 1985). Regarding the interactive effect of self - focused locus of causality and successful outcomes on pride, Weiner - esteem Pride is conceptualized as a positive, self - related emotion (Baldwin & Baccus, 2004). Further, as noted earlier, pride is defined as [emotion] generated by appraisal that one is responsible for a socially valued outcome or for being a socially - Fischer, 1995, p. 66). This definition emphasizes both a) the presence of a socially valued outcome or status, and b) the present study, high LMXSC subordinates perceive they possess a socially valued status; high self - attributed locus of causality for RBS subordinates are likely to feel their LMX quality is a 31 consequence of their actions and merits. Accordingly, subordinates with high LMXSC and high self - attributed locus of causality for RBS would be especially likely to experience feelings of LMX - related pride. Attribution theory contends sham e and guilt will occur when individuals experience a failure they could have been avoided (Weiner, 1985). When subordinates think they have a poor quality relationship with their supervisor as compared to the LMX relationships, they are likely to feel deprived of the better social standing peers possess through their relationships with the supervisor. According to Lewis (1971), guilt and shame both deal with a self - inflicted actual self. If low LMXSC subordinates believe their own own behavior and consequently experience guilt. If they view their contribution to their LMX quality as stemming from their per sonality or ability, they are expected to experience shame. Thus, as per attribution theory, low LMXSC subordinates who believe their own behavior has determined their LMX quality are especially likely to experience LMX - related g uilt and shame . Research on the interactional approach to justice (also referred to as the two - stage model of fairness or fa ir - process model) has put forth and tested predic tions similar to those suggested by attribution theory. As a result of the similar ities, tests of the interactional approach to justice indirectly Krehbiel & Cropanzano, 2000 ; Weiss et al. , 1 999 ). In the interactional approach to justice , emotions are assumed to emerge through the in teraction of outcom e favorability appraisals ( is this a good or bad outcome for me personally? ) and appraisals of processes and procedures that lead to that outcome ( why and how did the outcome come to be? ). In the current study, LMXSC could be viewed as a type of outcome f avorability , in that high LMXSC is a desirable outcome likely to elicit positive emotions and low LMXSC is 32 an unattractive outcome likely to elicit negative emotions. Locus of causality for RBS could arguably contribute to subordinates understanding of wh y and how they have their level of LMXSC. Consequently, this interaction focuses on outcome favorability (i.e., being why and how that outcome came to be (i.e., locus of causality for RBS). This is particularly true for self - attributions: individuals with high self - attributed locus of causality for RBS are likely to feel they have some control over relationship - building with their supervisor . It is important t o note that these two frameworks do not provide an exact match; for example, locus of is not an assessment of procedural or interactional justice. However, highlighting the similarities between these two models adds further empirical support Just as the current study anticipates LMX - related pride w ill be highe st among high LMXSC and high self - attributed locus of causality for RBS subordinates, Weiss et al. (1999) found pride was highest when positive outcomes re sulted from a fair procedure. Just as LMX - related guilt and shame are expected to be experienced by low LMXSC and high self - attributed locus of causality for RBS subordinates , Barclay, Skarlicki, and Pugh (2005) observed that negative inward - focused emotions (i.e. , shame and guilt) were highest when individuals reported fair procedures (high procedural o r high interactional justice) and unfavorable outcomes. Expectations for the LMXSC - self - attributed locus of causality for RBS interaction are as follows. The main effect of LMXSC on LMX - related emotions , as put forth in Hypothesis 1, is expected to be o bserved. Attribution theory would suggest that for high LMXSC subordinates, should be positively related to LMX - related pride . Further, the theory suggests subordinates with high LMX SC and 33 high self - attributed locus of causality for RBS should exhibit the highest level of LMX - related pride. O utcome favorability has shown a main effect on pride in previous research (Weiss et al., 1999) ; thus, those with high LMXSC and low self - attribut ed locus of causality for RBS should still exhibit reasonably high levels of LMX - related pride. For low LMXSC subordinates, Tracy and Robins (200 4 , p. 113 ) noted people are motivated to maint ain their pride and self - esteem and will modify their thinking ( i.e. , tak e credit for successes while externalizing failures) to preserve positive self - evaluation s . Thus, even subordinates with low LMXSC and low self - attributed locus of causality for RBS should have moderate levels of LMX - related pride. However when a negative outcome is s elf - attributed (i.e., low LMXSC and high self - attributed locus of causality for RBS), subordinates will view their low LMXSC as a self - inflicted failure and will not be able to salvage their self - esteem. Consequently, these subordinate s should experience the lowest levels of LMX - related pride. Expectations for LMX - related guilt and shame are as follows. H igh LMXSC subordinates hold a valued social status . R egardless of the extent to which they view their LMX as driven by their own beha vior, there is little reason to expect such subordinates would experience anything more than low levels of LMX - related guilt and shame. For individuals with low LMXSC, increasing levels of self - attributed locus of causality for RBS should be positively as sociated with LMX - related shame and guilt. Low LMXSC subordinates who do not see their LMX as driven by their own behaviors should feel some LMX - related guilt and shame, purely based on their undesirable social status (i.e. , low LMXSC ) . Yet as attribution theory suggests, shame and guilt should be highest when subordinates feel worse off (low LMXSC) and attribute this to their own behaviors (high self - attributed locus of causality for RBS). This hypothesis is displayed graphically in Figures 2 and 3. 34 H2: LMXSC will interact with self - attributed locus of causality for RBS such that: H2 a : Self - attributed locus of causality for RBS will accentuate the positive main effect of LMXSC on LMX - related pride, such that subordinates with high LMXSC and high self - attributed locus of causality for RBS will have the highest level of pride, and subordinates with low LMXSC and high self - attributed locus of causality for RBS will have the lowest level of pride. Among those with low self - attributed locus of causality for RBS, those with high LMXSC will report higher pride than those with low LMXSC. Figure 2 Hypothesized interaction between LMXSC and Self - attributed Locus of Causality predicting LMX - related Pride H2b: Self - attributed locus of causality for RBS will moderate the effect of LMXSC on LMX - related guilt . S ubordinates with low LMXSC and high self - attributed locus of causality for RBS will have the highest level of guilt . Subordinates with low LMXSC and 0 1 2 3 4 5 Low LMXSC High LMXSC Pride H2a Pride by Self Locus and LMXSC High Self Locus Low Self Locus 35 low self - attributed locus of causality for RBS will have the second highest level of guilt. Those with high LMXSC will show the lowest levels of guilt, regardless their level of self - attributed locus of causality for RBS. H2c: Self - attributed locus of causality for RBS will moderate the effect of LMXSC on LMX - related shame. S ubordinates with low LMXSC and high self - attributed locus of causality for RBS will have the highest level of shame. Subordinates with low LMXSC and low self - attributed locus of causality for RBS will have the second highest level of shame. Those with high LMXSC will show low levels of shame, regardless their level of self - attributed locus of causality for RBS. Figure 3 Hypothesized interaction between LMXSC and Self - attributed Locus of Causality predicting LMX - related Guilt/Shame Attribution theory holds different expectations when success is attributed to external forces. When an outside person or situation is viewed as causin g a positive, self - relevant 0 1 2 3 4 5 Low LMXSC High LMXSC Shame/Guilt H2b and H2c Shame/Guilt by Self Locus and LMXSC High Self Locus Low Self Locus 36 outcome, individuals are expected to feel gratitude towards that person or situation (Weiner, 1985). As discussed earlier, gratitude is defined as a positive emotion in response to a gift. In the context of LMXSC, subordinates who perceive they have high LMXSC and believe the them a higher quality relationship relative to other subordinates. Therefore, high LMXSC subordinates who be lieve their supervisor has determined their LMX quality should be particularly likely to experience LMX - related gratitude. As noted earlier, a responsibility for wrongdoing by another person or entity and often includes the goal of another individual or entity is appraised as being accountable for a wrongdoing. When a subordinate perceives superviso r behavior has strongly influenced the LMX quality and thinks his or her LMX quality , that subordinate is likely to feel displeased by this social comparison and blame the supervisor for these feelings . Attribution theory posits t hat when outcomes have a negative impact on the self and are viewed as resulting from arbitrary and unjustified behavior of others, anger is likely to result (Weiner, 1985). As such, employees has driven the quality of their LMX are particularly likely to feel LMX - related anger. Some empirical research on the interactional approach to justice support these expectations for LMX - related anger. B oth Weiss et al. (1999) and Krehbiel and Cropanzano (2000) observed anger was highest when negative outcomes resulted from an unfavorably biased (Weiss et. al., 1999 ) or unfair procedure (Krehbiel & Cropanzano, 2000). Barclay, Skarlicki, and Pugh (2005) observed negative outward - focused emotions such as anger to be highest when 37 outcomes were unfavorable and procedurally unfair. It is, however, important to note that inferences derived from the interactional approach to justice are only weakly applicable to supervisor - attributed locus of causality for RBS. Supervisor - attributed locus of causality for RBS may be viewed as unfair to the extent that it prohibits subordinates from influencing the quality of their LMX. For example, if a supervisor is rarely in communication with a subordinate, the ir LMX quality as unfair. However, supervisor behavior may influence LMX quality in ways that do not leave subordinates powerless to change their LMX. Thus, inferences from studies testing the interacti onal approach to justice are considered here with caution. The specific expectations for the LMXSC - supervisor - attributed locus of causality for RBS interaction are as follows. For LMX - related gratitude, LMXSC - emotions main effect is anticipated as per Hy pothesis 1. Attribution theory would indicate that LMX - related gratitude is highest when one experiences high LMXSC (a positive outcome) and perceive s or (caused by an external force). In considering subordinates with high LMXSC who do not view the supervisor as strongly influencing their LMX, o utcome favorability has shown a positive main effect on hap piness (Weiss et al., 1999). Thus, there is reason to expect LMX - related gratitude will still be moderately high for hi gh LMXSC, low supervisor - attributed locus of causality for RBS subordinates. Looking next to low LMXSC subordinates, those who think supervisor behavior has little to do with their LMX quality likely purview (e.g. , task demands). For this reason, subordinates with low LMXSC and low supervisor - attributed locus of causality for RBS are expected to have low, but not the lowest , levels of gratitude. Attribution theory suggests those who perceive superviso r behavior to be a 38 strong determinant of LMX quality and low LMXSC will likely feel wronged by their supervisor, and thus, should exhibit the lowest levels of gratitude. Turning next to LMX - related anger, again an LMXSC - emotion main effect is anticipated. Because subordinates with high LMXSC have a privileged status and are less likely to feel deprived, they are anticipated to show low levels of LMX - related anger irrespective of their level of supervisor - attributed locus of caus ality for RBS. Supervisor - at tributed locus of causality for RBS is expected to show a positive relationship with anger for low LMXSC. Subordinates with low LMXSC and low supervisor - attributed locus of causality for RBS are less likely to blame their supervisor for their LMX quality, but may still perceive a deficit in the LMX quality that moderate levels of LMX - related supervisor and LMXSC is lo w, a ttribution theory suggests LMX - related anger should be highest. This hypothesis is displayed graphically in Figure 4 and 5. H3: LMXSC will interact with supervisor - attributed locus of causality for RBS such that: H3a: Supervisor - attributed locus of ca usality for RBS will accentuate the positive main effect of LMXSC on LMX - related gratitude, such that subordinates with high LMXSC and high supervisor - attributed locus of causality for RBS will have the highest level of gratitude, and subordinates with low LMXSC and high supervisor - attributed locus of causality for RBS will have the lowest level of gratitude. Among those with low supervisor - attributed locus of causality for RBS, those with high LMXSC will report higher grati tude than those with low LMXSC. 39 Figure 4 Hypothesized interaction between LMXSC and Supervisor - attributed Locus of Causality predicting LMX - related Gratitude H3b: Supervisor - attributed locus of causality for RBS will accentuate the negative main effect of LMXSC on LMX - related anger for individuals with low LMXSC, such that subordinates with low LMXSC and high supervisor - attributed locus of causality for RBS will have the highest level of anger. Low LMXSC subordinates with low supervisor - attributed locus of causality for RBS wi ll report less anger. Subordinates with high LMXSC will report low anger, regardless of their level of supervisor - attributed locus of causality for RBS. 0 1 2 3 4 5 Low LMXSC High LMXSC Gratitude H3a Gratitude by Supervisor Locus and LMXSC High Supervisor Locus Low Supervisor Locus 40 Figure 5 Hypothesized interaction between LMXSC and Supervisor - attributed Locus of Causality predicting LMX - related Anger Interpersonal justice and group values model Anger is frequently the focal emotion in justice theories when injustices are at their highest. Thus far, anger has been discussed as most likely when subordinates experienc e low LMXSC and perceive high supervisor - attributed locus of causality for RBS. Yet LMX - related anger could be attenuated by respectful and polite interpersonal treatment from the supervisor (Lind & Tyler, 1988) . Here, I describe why interpersonal justice is expected to show a three - way interaction with LMXSC and locus of causality for RBS in predicting levels of LMX - related anger. Interpersonal justice is a dimension of interactional justice, which was first referred to in a chapter by Bies and Moag (1986). In their study, Bies and Moag asked job applicants how company recruiters should act, and several themes emerged: open and can did communication about decision - making (truthfulness), sufficient explanations of decision - making outcomes 0 1 2 3 4 5 Low LMXSC High LMXSC Anger H3b Anger by Supervisor Locus and LMXSC High Supervisor Locus Low Supervisor Locus 41 (justification), polite and considerate interpersonal behavior (respect), and a lack of biased or prejudicial statements (propriety). Greenberg (199 3) later expanded this research by suggesting interactional justice is comprised of two factors, informational justice and interpersonal justice. Informational justice was intended to describe justification and truthfulness, whereas interpersonal justice w as intended to describe respect and propriety rules. Note that interpersonal justice is the focus here because it fits best with past research on the buffering effect of fairness (Brockner & Wiesenfeld , 1996) ; informational justice will also be collected f or supplementary analyses. The group values model provides some support for the importance of interpersonal justice in how subordinates think and feel. Although is explained in terms of pro cedural justice, the m odel highlights the importance of both procedural and interpersonal justice. The g roup values model posits individuals value group membership because it fulfills the psychological need for identity and self - esteem. Procedures and interactions are perceived fair to the extent that they uphold shared values, such as treating members with dignity and respect. Individuals are in fact expected to care about receiving polite interpersonal above and beyond having immediate, instrumental control over th eir environment (Lind & Tyler, 1988, p. 238) According to the model, dis respectful interpersonal behavior is noteworthy to individuals because it violates a shared value for treating members respectfully , and denies target individuals the esteem otherwise associated with full group membership. Brockner and Wiesenfeld (1996) suggest that within the group values model, fair treatment reassure s individuals their identity and esteem needs will be met in the future. From this perspective , fair treatment may the psychological consequences of negative 42 outcomes , in that respectful treatment provides an indication that the underlying relationship has a relational (versus transactional) quality and esteem and identity needs w ill be fulfilled in the future (Brockner & Wiesenfeld , 1996). In model, the extent to which the supervisor displays polite, considerate treatment may buffer levels of LMX - related anger , particularly amongst individuals with both low LMXSC and high supervisor - attributed locus of causality for RBS. That is, individuals with low LMXSC and a high supervisor attributed locus of causality for RBS are likely to feel disappointed in their LMX quality; however, respectful treatment from r weaken the amount of LMX - related anger felt . Given interpersonal justice is implied here as playing a role in determining the level of LMX - related anger felt, a pertinent question is: what role does interpersonal justice play in the LMX - related emotions , if any ? Interpersonal justice is not expected to determine levels of self - conscious LMX - related emotions (pride, guilt, and shame), because the self - conscious emotion s are expected to emerge insofar as successes and failures are self - attributed . , LMXSC) and the extent to which their behavior is attributed to determining LMX quality (i.e. , locus of causality for RBS) focus on perc eptions of the defini ng characteristics of self - conscious emotions. Interpersonal justice, by comparison, is a treatment, which is externally - focused and likely not relevant for LMX - related pride, guilt, or shame. For these reasons, interpersonal justice is also - conscious emotions, nor is it expected to mod erate the interaction of self - attributed locus of causality for RBS and LMXSC on 43 self - conscious emotions. Simply put, respectful tre relevant information to expected relationships with self - conscious emotions. P olite and considerate treatment should however should play a role in the extent to which externally - focused emotions are experienced. Gratitude and anger are emotions that are felt in relation to a person or thing assumed to cause a self - relevant outcome; in the current study, the focuses on the perception of respe ctful and considerate treatment from the supervisor, there is reason to assume interpersonal justice would be important factor in determining the level of LMX - related anger and gratitude . Interpersonal justice has been linked to anger in past research (Ru pp, McCance, Spencer, & Sonnentag, 2008). Thus, when inconsiderate and rude treatment is displayed by the supervisor , subordinates are anticipated to feel greater LMX - related anger. Although there is limited research on gratitude and interpersonal justice, reason would suggest that when subordinates feel their supervisor shows considerate and polite treatment, they will feel more LMX - related gratitude . Accordingly, interpersonal justice is expected to show a main effect on LMX - related gratitude and ange r. H 4: Interpersonal justice will be associated with emotions, such that a) interpersonal justice will be positive ly associated with LMX - related gratitude b) interpersonal justice will be negatively associated with LMX - related anger Interpersonal justice is also expected to moderate the relationship between LMXSC and externally - focused emotions. It is important to note that interpersonal justice and LMXSC are expected to show a slight positive association, given they both revolve around positive treatment ex conceptually 44 different in several ways. within the context of the LMX relationship held by their peers . LMXSC is explicitly a perceived social the supervisor, but supervisor within a dyad. It is also imp ortant to note that LMX focuses broadly on the quality of social exchanges between the supervisor and subordinate (i.e. , the extent to which expectations are met), but interpersonal justice focus es on one specific aspect of that social exchange : the extent to which the subordinate experiences respect and propriety from the supervisor. The g roup values model would argue polite and considerate treatment represents an especially potent characteristic of social exchange quality (Lind & Tyler, 1988) , and should be especially important in determining outcomes. This hypothesis is displayed graphically in Figures 6 and 7. H5: The effect of LMXSC on externally - focused LMX - related emotions will be moderated by interpersonal justice, such that H5a: LMXSC will exhibit a main effect on gratitude, with high LMXSC subordinates experiencing high er levels of gratitude than low LMXSC subordinates. Interpersonal justice will be positively associated with gratitude , and this positive relationship will be higher for high LMXSC. 45 Figure 6 Hypothesized interaction between LMX - related Gratitude and Interpersonal Justice predicting LMX - related Gratitude H5b: LMXSC will exhibit a main effect on anger, with low LMXSC subordinates experiencing higher levels of anger than high LMXSC subordinates. Interpersonal justice will exhibit a negative main effect on anger , and this negative relationship will be strongest for low LMXSC subordinates . 0 1 2 3 4 5 Low LMXSC High LMXSC Gratitude H5a Gratitude by Interpersonal Justice and LMXSC High Interpersonal Justice Low Interpersonal Justice 46 Figure 7 Hypothesized interaction between LMXSC and Interpersonal Justice predicting LMX - related Anger Attribution theory posits that when outcomes have a negative impact on the self and are viewed as resulting from arbitrary and unjustified behavior of others, anger is likely to result (Weiner, 1985) . I expect that when interpersonal justice is high, there will be less of a main effect of LMXSC on anger than when interpersonal justice is low. I also expect that when interpersonal justice is low, the effect of a supervisor locus of control will have a steeper positive slop associated with anger when LMXSC is low and interpersonal justice is low. When interpersonal justice is high, the effect of a supervisor - locus on anger for low LMX SC indi viduals will be weaker. Expectations for this three - way interaction are as follows. First, h igh LMXSC , high interpersonal justice perceive they are treated with respect and dignity by the supervisor. That is, this condition represents a combination of the most positive perceptions that could be held by subordinates. 0 1 2 3 4 5 Low LMXSC High LMXSC Anger H5b Anger by Interpersonal Justice and LMXSC High Interpersonal Justice Low Interpersonal Justice 47 Because these two perceptions provide a strong positive reinforcement, the extent to which supervisor behavior determines LMX quality is not expected to exert influence on the low levels of anger experienced by these subordinates. Thus, high LMXSC, high interpersonal justice subordinates should experience the lowest levels of LMX - related anger, irrespective of their level of supervisor - attributed locus of causality for RBS. High LMXSC , low interpersonal justice subordinates perceive the supervisor does not show them respect and consideration, but also perceive they have a relatively higher quality LMX than their peers possess . T hese subordinates are expected to exhibit a low level of LMX - related anger , because even though these subordinates experience poor interpersonal treatment, their LMXSC would indicate that they are still able to maintain a more effective relation ship with the supervisor than other peers. However, these subordinates will experience higher levels of LMX - related anger than subordinates with high LMXSC and high interpersonal justice perceptions , given high LMXSC - high interpersonal justice subordinate s exhibit the most positive combination of perceptions. Also, f or high LMXSC, low interpersonal justice subordinates , their interpersonal treatment and relative social standing in terms of LMX will likely have a stronger influence on their levels of LMX - re lated anger than their level of supervisor - attributed locus of causality for RBS. Low LMXSC, low interpersonal justice subordinates will experience the highest levels of LMX - related anger, because these subordinates have a comparatively worse relationship with the supervisor relative their peers and perceive the supervisor is impolite and disrespectful towards them. The LMX - related anger that is experienced in this scenario is exacerbated by the extent to which the supervisor is viewed to be culpable in d etermining the LMX quality that has been developed over time. If a rude supervisor who favors other peers is not viewed as 48 determining the LMX quality that emerges, the subordinate may instead be making other causal locus attributions. In such a scenario, the subordinate may blame a lack of resour ces in the task environment or an unsupportive environment (Shoss, Eisenberger, Restubog, & Zagenczyk, 2013) as Or, the subordinate may b lame him or herself for the quality of the relationship that has developed over time. Whatever the alternative causal locus, the effect of low LMXSC and low interpersonal justice on LMX - related anger should be less strong when supervisor - attributed locus o f causality for RBS is low. When supervisor - attributed locus of causality for RBS is high amongst low LMXSC, low interpersonal justice subordinates, attribution theory would suggest LMX - related anger should be especially high. As was discussed previously , attribution theory would suggest that when a negative outcome (low LMXSC) is attributed to an external force (high supervisor - attributed locus of causality for RBS), LMX - related anger should be highest (Weiner, 1985) . The g roup values model would suggest breaches in respectful interpersonal treatment deny individuals the esteem associated with full group membership. Accordingly, LMX - related a nger will be higher for subordinates with low LMXSC, high supervisor - attributed locus of causality for RBS, and low interpersonal justice, in comparison to low LMXSC, high supervisor - attributed locus of causality for RBS subordinates who experience high in terpersonal justice. When subordinates experience low LMXSC and high interpersonal justice , their relatively lower standing in LMX quality suggests they should experience higher levels of LMX - related anger than high LMXSC subordinates. However among low LMXSC - high interpersonal justice subordinates, the extent to which supervisor - attributed locus of causality for RBS is 49 positively associated with LMX - related anger will be lessened given the presence of respectful, considerate supervisory treatment. This hypothesis is displayed graphically in Figures 8 and 9. H 6 : Interpersonal justice will display a three - way interaction with LMXSC and locus of causality for RBS on LMX - related anger LMXSC will exhibit a main effect on LMX - related anger, with lo w LMXSC subordinates reporting more anger than high LMXSC subordinates. Interpersonal justice will have a main effect on LMX - related anger, such that all combinations of the LMXSC - supervisor locus of control will exhibit higher anger scores when interpers onal justice is high as compared to when interpersonal justice is low. When interpersonal justice is high, the slope of anger is less steep for low LMXSC individuals as supervisor causal locus attributions increase; when interpersonal justice is low, the slope of anger for low LMXSC is steeper as supervisor causal locus attributions increase. The level of anger experienced by high LMXSC subordinates will be unaffected by supervisor - attributed locus of causality for RBS. 50 Figure 8 Hypothesized interact ion between L MXSC, Supervisor - attributed Locus of C ausality, and Interpersonal Justice predicting LMX - related Anger: High interpersonal justice Figure 9 Hypothesized interaction between LMXSC, Supervisor - attributed Locus of Causality, and Interpersonal Justice predicting LMX - related Anger: Low interpersonal justice interpersonal justice to explore when and how LMXSC links to different LMX - related emotions. Further, co re affect 0 1 2 3 4 5 Low LMXSC High LMXSC Anger H6 Anger by Supervisor Locus and LMXSC (High Interpersonal Justice) High Supervisor Locus Low Supervisor Locus 0 1 2 3 4 5 Low LMXSC High LMXSC Anger H6 Anger by Supervisor Locus and LMXSC (Low Interpersonal Justice) High Supervisor Locus Low Supervisor Locus 51 theory provides a theoretical lens for conceptualizing LMX - related emotions. In this model, LMXSC was expected to show a main effect on each emotion, but based on attribution theory, this main effect was expected to be moderated by levels of self - and supervisor - attributed locus of causality for RBS. Interpersonal justice was also expected to play a role in LMX - related emotions. The current study anticipate d interpersonal justice moderates the LMXSC - supervisor - attributed locus of causality inter action on LMX - related anger. For externally - focused emotions (gratitude and anger) interpersonal justice is also expected to exhibit a main effect and moderat e the LMXSC - emotion linkage. 52 METHOD Pilot study to validate locus of c ausality for RBS measure L ocus of causality for RBS is a new construct put forth in this dissertation, and for this reason, I reviewed measures to determine how similar attributional constructs are measured and created a measure based on the review (see the measur e section below for further description . ) The pilot study assessed the validity of the new locus of causality for RBS measure by comparing it to other measures of locus of causality ( i.e. , evidence of convergent validity ) and measures of related but distin ct constructs like cause globality (i.e. , evidence of discriminant validity). Pilot study sample The study was posted on Mechanical Turk with a recruitment description looking for - time $1 sur vey. Only Mechanical Turk workers with a 99% task approval rate and 5000 previously approved tasks were allowed to see the recruitment posting and participate in the study. Data was initially collected from 12 participants to ensure the survey and Mecha nical Turk were both set up correctly. An error was found in the CDSII responses options , and consequently, CDSII data from these 12 participants are not included in the overall pilot study findings. All other data from these participants was included in t he overall pilot results. The full pilot con sisted of 374 MTurk participants . Several Two attention checks were included in the survey ; individuals who did not re spond correctly to either of these attention checks were flagged. Individuals who showed no variance in their either responses to all CDSII items or all locus of causality for RBS items were 53 also flagged. 54 respondents were flagged one or more times. Bec ause respondents could have been flagged once but otherwise responded attentively (e.g. showed no variance on the CDSII items, but varied in responses to the locus of causality for RBS items and passed both attention checks), only the 20 respondents who we re flagged two or more times were removed from the overall data. 77 responses with duplicate IP addresses were also removed from the data. Although there are potentially legitimate reasons for duplicate IP addresses (e.g. a husband and wife who complete surveys on the same computer), a conservative approach was taken to ensure high quality data. By removing these 77 responses and the 20 careless responses, a total of 95 responses were removed from the data leaving data from 279 participants. The survey consent form can be found in Appendix A. Measures Below I critique available attribution measures (found in Appendi x B ) and discuss how I went about creating a new measure of locus of causality for RBS. First, Questionnaire (ASQ) measures respondent s explanatory style. Respondents are given a set of think would likely be the main cause of that event. Respondents are then asked about the w likely is it that the main cause you gave will 54 cause that you gave something that affects just this situation, or does it affect other areas of your The ASQ does not fit th is assess locus attributions, but instead focuses on the stability and globality of attributions. Even if this measure were to be modified to capture locus attributio ns, the layout would be most continuous measurement o f the self and supervisor as cause s . Second, the ASQ takes an event - based approach to measuring attributional style, u sing specific hypothetic al incidents. The construct I aim to measure here, locus of causality for RBS, is not framed as an event - based construct, but rather is assumed to be attributions regarding the behaviors that contributed to relationship building wit hin the LMX dyad . Specific interactions with the supervisor may influence these attributions, but these attributions are made towards relationship building in general . Third, the Dykema et al. (1996) assess es attributional style as an individual differen ce . In th is study, I focus on attributions regarding . Although these perceptions could be linked to individual differences, I frame locus of causality for RBS as a more proximal context - specific attribution, and not an i ndividual difference. Second, McAuley, Duncan, and Russell (1992) Revised Causal Dimension Scale (CDSII) requires individuals to provide an open - ended response describing their causal attributions to a situation. Respondents then rate several properties of the attributed cause, including the locus of causality, stability, and controllability (internal and external) of the cause. This measure has been used to study locus of causality elsewhere in the organizational sciences (Tay, Ang, & Van Dyne, 2006). 55 locus of causality for RBS as existing on unipolar dimensions (i.e. , the self and the supervisor as existing on independent dimensions). However, the locus of causal ity items provided useful content for scale construction. provide relevant item content. and idiomatic, and for these reasons, a poor quality item. assesses attributions made by subordinates regarding abusive supe rvision displayed by their supervisor. This measure divides attributions into three subdimensions: internal attributions, external attributions, and relational attributions. oes two reasons. First, the items focus on abusive supervision as the target , whereas the current study looks at relationship building as the target. This becomes especially problematic given the relational attribution items refer to the r elationship as a potenti al cause of abusive supervis ion instead of using the relationship as the target ( My behavior toward me is due, in part, to the relationship we have ) This target discrepancy is also true for internal an the scope of locus of causality. Despite the focus on abusive supervision as a target and the items addressing controllability, useful content was retained from several of the items for new measure construction ( e.g. , synonyms like 56 Fourth, Pretzer, Epstein, and Fleming 1991) m easure comes from marital research. The . Of specific interest here were the subscales focused on locus of causality towards behaviors : Attribution of Causality t o Own Behavior, Attribution of Causality to Own Personality , Personality. 3 Because locus of causality for RBS is based on how LMX quality is developed and thus is focused on social exchanges , behavioral attributions towards relationship building were deemed more appropriate than personality attributions . So, I focus more specifically on Attribution of Causality to Own Behavior and . Although only two items were available, they provide content useful for the current context be After reviewing these measures, I created the measure found in Appendix C . Similar to the ASQ and CDSII, I instruct participants to consider what causes determine relationship building within their workgroup: Often in workgroups, some subordinat es have better relationships with their supervisor in comparison to the relationships other coworkers have with the supervisor. Other subordinates may have worse relationships with their supervisor in comparison to the relationships other coworkers have wi th the supervisor. Think about the reason or reasons for why this might be. The items below concern your impressions or opinions of what causes effective relationship building with your supervisor. Rate the extent to which you agree with these To create item content, I modified items from the se established measures . For example, 3 I was unable to reach the researchers to gain access to the full measure, so I chose to review the published example items. 57 rovoked my supervisor to act the way he or In the pilot study, I include d this newly constructed measure and the two most commonly used attribution measures : the ASQ and CDSII. To assure all three measures focus on the same target (relatio were made to the ASQ and CDSII instructions and items . The revised mea sure can be found in Appendix D . R emoved text is marked in strikeout, and added text is indicated with boldface. Table 3 shows a summary of pilot study expectations and findings , which are discussed further in the presented order. Because pilot items represented distinct but related constructs (e.g. self - attributed locus of causality for RBS and personal controllabil ity), the first expectation was that the different causal attributions measured would load onto separate but related factors when entered into an exploratory factor analysis (EFA). A maximum likelihood estimation EFA with an oblique promax rotation was use d to test this expectation. The patter n matrix can be found in Table 4 . Although six factors met the Kaiser criterion cut - off (i.e., eigenvalues > 1), the sixth factor contained only one item loading over .3 ( CDSII Item 4). Because this sixth factor explai ned the least amount of variance and did not contribute substantially to interpretation of the items, a 5 - factor solution was instead adopted. 58 Table 3 Summary of Pilot Expectations and Findings Expectation Convergent or Discriminant? Supported? 1. Different scale constructs will map onto separate, related EFA factors. Discriminant Yes 2. Self - and supervisor - attributed locus of causality for RBS will show a weak to moderate negative relationship with one another. Discriminant Yes ( r = - .37) 3. Self - a) weak to moderate size positive relationship with CDSII locus b) null to weak size relationship with CDSII external control c) null to weak size relationship with CDSII stability d) weak to moderate size positive relationship with CDSII personal control e) null to weak size relationship with ASQ stability f) null to weak size relationship with ASQ globality a: Convergent b: Discriminant c: Discriminant d : Convergent e: Discriminant f: Discriminant a: Yes ( r =.39*) b: Yes ( r = - .07) c: Yes ( r =.13 *) d: Yes ( r =.38 *) e: Yes ( r =14*) f: No ( r =33*) 4. Supervisor - a) null to weak size negative relationship with CDSII locus b) weak to moderate size positive relationship with CDSII external control c) null to weak size relationship with CDSII stability d) null to weak size relationship with CDSII personal control e) null to weak size relationship with ASQ stability f) null to weak size relationship with ASQ globality a: Discriminant b: Convergent c: Disc riminant d: Discriminant e: Discriminant f: Discriminant a: Yes ( r = - 12) b: Yes ( r =.13* ) c: Yes ( r = - .06 ) d: Yes ( r = - .08 ) e: Yes ( r =10 ) f: Yes ( r = - .16*) 5. ASQ open - text comments coded as self attributions will show higher levels of self - attributed locus of causality for RBS than all other comment categories Convergent Yes 6. ASQ open - text comments coded as supervisor attributions will show higher levels of supervisor - attributed locus of causality for RBS than all other comment categories. Convergent Yes Note. - 0.3, moderate=0.4 - 0.6, strong=0.7 - 0.9, perfect=1. 59 Table 4 Promax Rotated Maximum Likelihood Extraction Pattern Matrix for Pilot Items Factor 1 2 3 4 5 Locus of causality RBS Self 1 .92 Locus of causality RBS Self 2 .68 Locus of causality RBS Self 3 .78 Locus of causality RBS Self 4 .84 Locus of causality RBS Self 5 .83 Locus of causality RBS Self 6 .79 Locus of causality RBS Supervisor 1 .69 Locus of causality RBS Supervisor 2 .66 Locus of causality RBS Supervisor 3 .82 Locus of causality RBS Supervisor 4 .79 Locus of causality RBS Supervisor 5 .66 Locus of causality RBS Supervisor 6 .83 CDSII Locus 1 .49 CDSII Locus 6 .71 CDSII Locus 9 .74 CDSII External Control 5 .72 CDSII External Control 8 .71 CDSII External Control 12 .70 CDSII Stability 3 .66 CDSII Stability 7 .35 .49 CDSII Stability 11 .58 CDSII Personal Control 2 .68 CDSII Personal Control 4 .70 CDSII Personal Control 10 .90 ASQ Stability ASQ Globality Eigenvalue 6.94 3 .56 2.21 1.98 1.45 Variance explained 26.67 % 13.70 % 8.48% 7.61 % 5.58 % Note. Loadings under .30 are suppressed. Pairwise deletion was used, item N s ranged from 265 to 278. 60 The five - fac tor solution accounted for 62.07 % of the overall variance in items. Factor 1 was interpreted as the self - attributed locus of causality for RBS factor. This factor explained 26.99% of the overall variance, was driven by self - attributed locus of causality for RBS items with loadings ranging from .68 to .92 . No other items loaded on this factor above .3 . T he second factor explained 13.70 % of the overall item variance, and was driven by items from the CDSII locus of causality and personal control items. The CDSII personal control er which you [high] have power/ [low] have no power ). [high] inside of you ) and 9 ( ) also showed high item loadings on the second factor. Interestingly, the CDSII stability item 7 ( [hig h] stable over time / [low] variable also loaded .35 on this factor. The item loadings suggest this second factor can be in terpreted as the locus of control factor, emphasizing a self - attributed locus of causality and a perception of stable, per sonal control. The third factor was interpreted as the supervisor - attr ibuted locus of causality for RBS . This factor explained 8.48% of the total variance in the items, and was driven by the six supervisor - attributed locus of causality for RBS items , wit h item loadings ranging from .63 to .83 . The fou rth factor was interpreted as the external control factor, which accounted for 7.61 % of the total item variance. All three CDSII external control items loaded on this factor , with loadings ranging from .69 t o .73. Last, the f ifth factor was interpreted as the perceived causal stability factor . This factor explained 5.58 % of the total item variance, and was driven by CDSII stability items (e.g. [high] permanent ). Interestingly, ASQ stabilit y and globality item s did not exhibit loadings over .3 on any of the five factors. 61 The rotated factor corr elations can be found in Table 5 . The self - attributed locus of causality for RBS and locus of control (i.e. first and second) factors showed the highest positive correlation ( r =.4 7 ); this was not unexpected given attribution theory suggests controllability attributions are temporally de pendent on locus attributions. Table 5 Factor Correlation Matrix Based on Promax Rotated Maxi mum Likelihood Extraction Factor Analysis of Pilot Items Factor Factor 1 2 3 4 5 1 - 2 .47 - 3 - .40 - .16 - 4 - .12 - .19 .17 - 5 .06 .28 - .03 .07 - Note. Pairwise deletion was used, N ranged from 265 to 278. The self - attributed locus of causality for RBS factor exhibited a moderate negative relationship with the supervisor - attributed locus of causality for RBS factor ( i.e., factors 1 and 3, r = - .40 ), suggesting some individuals were high (or low) on both types of attributions. The locus o f control and causal stability factors also showed a weak positive relationship ( i.e., factors 2 and 5, r =.28), which is somewhat unsurprising given items loading highly on the locus of control factor suggest personal control is somewhat permanent and trai t - like. F or example, the CDSII item 10 states [high] have power / [low] have no power . The supervisor - attributed locus of causality for RBS and external control factors showed a weak positive relationship ( i.e. , factors 3 and 4, r =.17 ). This positive relationship was expected, given these factors were driven by specific (supervisor - attributed) and global (external), non - self causal attributions. The factor structure was as expected and factors show ed appropriate relationships with other measures of related but distinct factors . The fact that factor structure of the CDSII and 62 ASQ items did not perfectly reflect their hypothesized structure (e.g. CDSII personal control, stability, and locus items all lo aded highly on factor 2; ASQ stability did not load on a factor with the CDSII stability items) indicates that these measures are less than ideal for measuring causal attributions. Despite this fact, scale scores, reliabilities, and intercorrelations for t he CDSII and ASQ were generated using the original scale items in order to better compare findings to existing research using these measures. Table 6 shows measure descriptive statistics, intercorrelations, and scale reliabilities. Self - and supervisor - attributed locus of causality for RBS were further shown to have good internal consistency (respectively, =.92; =88). CDSII subdimensions exhibited weaker levels of internal consistency (locus, =.67; external control, =.73; stability, =.61; persona l control, =.85). However, it is important to note that the CDSII subdimensions have a lower number of items than the locus of causality for RBS scales (respectively, 3 and 6 items), and scale reliabilities can be artificially inflated by the number of it ems . Also, these reliabilities are similar to those found in the original measurement paper (locus, =.67; external control, =.82; stability, =.67; personal control, =.79; McAuley, Duncan, & Russell, 1992). As can be seen in the correlations, the pilo t data supported almost all expectations. Self - and supervisor - attributed locus of causality for RBS exhibited a moderate ly size negative relationship with one another ( r = - .37 , p <.05 ) . Self - attributed locus of causality for RBS showed moderate sized relati onships with the CDSII locus and personal control scales (respectively, r =.39, p <.05; r =.38, p <.05), and had weak relationships with CDSII external control and stability (respectively, r =.07 , n.s. ; r =.13 , p <.05 ) and ASQ stability ( r =.14, p <.05). Supervisor - attributed locus of causality for RBS exhibited a weak positive relationship with CDSII external control ( r =.13, p <.05), and weak to null relationships with CDSII locus, stability, and personal control 63 (respectively, r =.12, n.s.; r = - .06 , n .s.; r = - .08 , p <.05) and ASQ stability and globality (respectively r =.10 , n.s.; r =.16, p <.05). Only one expectation was not met: the ASQ globality subscale showed an unexpectedly high relationship with self - attributed locus of causality for RBS ( r =.33, p <.05). 64 Table 6 Pilot measures descriptive statistics and intercorrelations Items N M SD 1 2 3 4 5 6 7 8 Locus of Causality for RBS 1. Self - Attributed 6 278 3.52 .91 (.92) 2. Supervisor - Attributed 6 278 3.48 .76 - .37* (.88) ASQ 3. Stability 1 278 5.86 1.29 .14* - .10 ( - ) 4. Globality 1 277 4.51 2.20 .33* - .16* .31* ( - ) CDSII 5. Locus 3 268 4.89 1.55 .39* - .12 .08 .34* (.67) 6. External Control 3 268 4.84 1.57 - .07 .13* .01 - .13* - .11 (.73) 7. Stability 3 268 5.26 1.53 .13* - .06 .09 .10 .30* .03 (.61) 8. Personal Control 3 268 5.83 1.71 .38* - .08 .04 .27* .71* > - .01 .33* (.85) Note. * p <.05. Locus of causality for RBS items were rated on a 5 - point Likert scale, ASQ items were rated on a 7 - point Likert scale, and CDSII were r ated on a 9 - point Likert scale. 65 As part of the ASQ, participants wrote in an open text the main cause of the relationship quality they have with their supervisor. I coded responses into different categories based on content, and examples of coded re sponses are displayed in Table 7 . 58 responses were coded as self - focused attributions. Responses coded as self - focused attributions emphasized the its and/or behavior 4 - focused attributions. Responses coded as supervisor - personality traits and/or behavior. For example, one respondent said the main cause of his or her ut 26 responses were coded as focusing on an exter nal cause. External causes included characteristics of the relationship that are beyond the control of the supervisor or subordinate (e.g. the supervisor is a relative), characteristics of the work environment (e.g. working at a desk next to the boss), and the hierarchical nature of the supervisor - subordinate relationship (e.g. the supervisor has formal authority in ways the subordinate does not). 128 responses were coded as an attribution towards the interaction of the supervisor and respondent. Many of t hese responses reflected characteristics of the relationship exchanges with the supervisor. For example, 14 hold with their supervisor. Also coded as an inte raction between the supervisor and subordinate were responses referring to traits and behaviors displayed by both the supervisor and the subordinate ( "We both try to be understanding and open minded of each other's situations and 4 Although the definition of locus of causality for RBS focuses on behavior, it was often not possible to differentiate between behavior and tra it attributions in the short, open - text responses. For this reason, both behaviors and traits were coded into the self and supervisor attributions. 66 10 responses were coded as not applicable (NA). These responses did 67 Table 7 Example responses to ASQ main cause open text Self One's own personality and/or behavior "My ability to do my work well and keep clients happy is the clear basis for my relationship with my supervisor. He is clearly happy and treats me well because of it." "I'm really funny and make people laugh. My boss looks for me just for a laugh, it helps him make the day pass more smoothly I think it will continue to affect me because my boss needs that. He is always stressful and when he laughs, he forgets for a miute the overflow of work he has and it puts him in a better mood. It doesn't just affect my supervisor , it affect anyone that works with me because seriously, who doesn't like a good laugh every now and then." "My hard work is what determines the relationship with my supervisor." Supervisor Personality traits and/or behaviors exhibited by the supervisor "To be honest at the start we didn't have a good work relationship, Things were a bit sour at the start as he tried to maintain his ego over his position to others. After quite some time he realized the importance he could obtain through me an d made steadyprogress healing the relationship by recommending me for a promotion." "The main cause for the quality of our relationship is my bosses personality." "One of the main causes for the quality of our relationship is that she genuinely cares as a person about employees well being. She likes to include them in fun and having a relationship more than just supervisor and employee." "The main cause of the relationship is his particular aversion to feeling disrespected, and the fact that one has to walk on eggshells to communicate effectively without setting him off." External Causes Characteristics of the relationship that are beyond the control of the supervisor or subordinate "He is my father. I work at his law firm to help with filing and deliveries. I will be attending law school soon." Characteristics of the work environment (e.g. task procedures, spatial proximity) "The overall environment, customs and rules that are found in the industry as a whole (in my local area)." Hierarchical nature of the relationship "He is the BOSS and I am not. That is the only cause of any relationship we have." Interaction Characteristics of the relationship exchanges "Mutual respect for each other" Behaviors displayed by both the su pervisor and the subordinate "He is reasonable and kind and I am reasonable and kind." "We both try to be understanding and open minded of each other's situations and deal accordingly." 68 Table 8 shows descriptive statistics for other pilot scales by the five ASQ open - text categories. As is marked in bold, the average level of self - attributed locus of causality for RBS items was highest amongst respondents who wrote a self - focused attribution in the ASQ open - text ( M =4.12) , in comparison to other open - text attributions categories . A one - way ANOVA showed significant mean differences in self - attributed locus of causality for RBS across ASQ open - text categories with a large effect size, F (4,273)=45.24, p <.01, 2 =.39. A post - hoc Bonferroni revealed participants reporting ASQ self - focused attributions had on average significantly higher levels of self - attributed locus of causality for RBS in comparison to all other open - text categories, with the exception of the NA responses. Table 8 Descriptive S tatist ics and Mean D ifferences of Pilot S cale M easures by ASQ Main Cause C omment C oding Self Supervisor External Interaction NA Total Locus of Causality for RBS Self N 60 54 27 128 9 278 M 4.12 2.52 2.98 3.77 3.57 3.52 SD .52 1.08 .88 .51 .97 .91 F (4,273)=45.24, p <.01, 2 =.39 Supervisor N 60 54 27 128 9 278 M 2.93 4.12 3.35 3.48 3.72 3.48 SD .80 .63 .76 .57 .86 .76 F (4,273)=23.28, p <.0, 2 =.25 ASQ Stability N 60 54 27 128 9 278 M 6.22 5.74 5.78 5.87 4.44 5.86 SD 1.20 1.09 1.05 1.29 2.35 1.29 F (4,273)=4.20, p <.01, 2 =.06 Globality N 60 54 27 127 9 277 M 5.05 3.31 3.41 4.98 4.67 4.51 SD 2.10 2.26 2.34 1.91 2.29 2.20 F (4,272)=9.06, p <.01, 2 =.12 Note. Locus of causality for RBS items were rated on a 5 - point Likert scale, ASQ items were rated on a 7 - point Likert scale, and CDSII were rated on a 9 - point Likert scale. 69 Table 8 Self Supervisor External Interaction NA Total CDSII Locus N 58 51 26 123 10 268 M 4.66 4.21 4.42 5.37 5.07 4.89 SD 1.41 1.61 1.47 1.47 1.36 1.55 F (4,263)=6.91, p <.01, 2 =.10 External Control N 58 51 26 123 10 268 M 4.96 4.94 4.59 4.77 5.00 4.84 SD 1.44 1.77 1.65 1.55 1.41 1.57 F (4,263)=.38, p= .82, 2 =.01 Stability N 58 51 26 123 10 268 M 5.14 5.02 5.27 5.49 4.50 5.26 SD 1.42 1.80 1.25 1.49 1.51 1.53 F (4,263)=1.72, p= .15, 2 =.03 Personal Control N 58 51 26 123 10 268 M 5.63 5.05 5.45 6.37 5.20 5.83 SD 1.59 2.03 1.84 1.43 1.61 1.71 F (4,263)=7.20, p <.01, 2 =.10 Also of interest, the average level of supervisor - attributed locus of causality for RBS items was expected to be highest amongst respondents who wrote a supervisor - focused attribution in the ASQ open - text. This expectation was supported by the data, with ASQ supervisor attributions having the highest levels of supervisor - attributed locus of causality for RBS ( M =4.12). A one - way ANOVA showed ASQ categories differed significantly in their levels of supervisor - attributed locus of causality with a large effect size, F (4,273)=23.28, p <.01, 2 =.25. Similar to the self - attribution analyses, a post - hoc Bonferroni revealed participants reporting ASQ supervisor - focused attributions had on average significantly higher levels of supervisor - attributed locus of causality for RBS in comparison to all other open - text categories, with the exception of the NA responses. The lack of a significant difference between focal ASQ categories and NA responses on levels of self - and supervisor - attributed locus of causality for RBS was surprising; however, this 70 finding may be influenced by the low number of NA responses (i.e., n =9). In general, comparing mean levels of both self - and supervisor - attr ibuted locus of causality for RBS by coded ASQ responses provided support for the construct validity of the locus of causality for RBS measure. Overall, as can be seen in Table 3 , the locus of causality for RBS measures were shown to be psychometrically sound in comparison to other similar measures. Consequently, the developed scale was used in the main study. Main Study Sampl e and procedures Data were collected using two surveys with a one week separation between the surv eys. Research indicates data collected through Mechanic Turk to meet psychometric standards (e.g., high internal consistencies within measures and high test - retest reliabilities when tested with a 3 - week gap), even when low compensation amounts (e.g., $.1 0) are used (Buhrmester, Kwang, & Gosling, 2011). Recent reviews suggest MTurk provides more diverse samples than the WEIRD (Western, Educated, Industrialized, Rich, and Democratic) participant data typically collected in organizational and student samples used in organizational psychology (Land ers & Behrend, 2014). C onvenience samples of this type are only problematic when the sample creates range restriction or when key variables that differentiate the sample are not measured or controlled (Landers & Beh rend, 2014). In this study, MTurk workers are not assumed to differ in the key variables from a student or organizational example, there is no reason to anticipate L MXSC levels would differ across Mechanical Turk, working student, or employee samples. Several filters recommended by Amazon were used to ensur e high data quality , which are described further below . 71 When cross - sectional survey designs are used, common method variance (CMV) is a potential concern. Podsakoff, MacKenzie, Lee, & Podsakoff (2003) recommend creating temporal, proximal, psychological, or methodological separation of measurement. Here, tempo ral separation between three of the independent variables (LMXSC, locus of causality for RBS) and the dependent variables (emotions) was created by collecting that data on occasions separated by one week. 5 I methodologically separated the constructs measur ed by varying response formats and scale endpoints across the measure. Filters were set so that MTurk workers who have a 99% or higher approval rate on their tasks, and who have been approved on 5,000 tasks previously could see the study and participate . Participants w ere paid $1 for the first survey and $3 for the second survey. The first survey consent form is found in Appendix E , the second consent form is found in Appendix F , and the end of survey message for this sample can be found in Appendix G . S ample size - way interactions, and one three - way interaction would be required as predictors. A multiple regression power analysis with an anticipated effect size of .0 5 , a desire statistical power level of .80, 9 predictors , and a probability level of .05 indicate d a minimum sample of 321 participants was required to detect an effect (Soper, 2015). Initial survey data was collected from 500 participants, and all 500 part icipants were invited to participate in the second survey. The first and second survey data were linked by MTurk worker ID numbers that participants provided. After removing 50 duplicate responses (i.e. individuals who filled out the first and/or second survey multiple times using the same ID/IP address), 396 responses were successfully linked across the two surveys, 6 6 responses consisted 5 For survey length purposes, interpersonal justice was measured at time 2. 7 2 of only the first survey, and 12 respons es consisted of only the second survey. Unlinked responses to the first survey were assumed to be largely due to study attrition; unlinked responses to the second survey were individuals who reported the study ID instead of their worker ID, or did not repo rt an ID number. Careless r esponse i dentification lack of variance in study measures were used to flag careless responses. Two explicit attention checks and one mea sure variance check were used in the first survey; one explicit attention check and two measure variance checks were used in the second survey. Nine participants who failed two or more checks were flagged as careless respondents, and excluded from analyses . After the nine careless responses and 50 duplicate responses were excluded from the data, 4 65 cases remained with 389 linked responses, 6 4 unlinked responses to the first survey, and 12 unlinked responses to the second survey. To see if survey response time had an influence on the way participants responded, three graduate students were asked to take both surveys at a normal pace and their response times were recorded. On average, the first survey took these students ~6 minutes, and the second survey to ok ~5 minutes . Respondents who took more than 60% less than these times on either survey were flagged (i.e., <2.4 minutes for the first survey, <2 minutes on the second survey). 26 participants were flagged for their times on the first survey; 25 participa nts were flagged for their times on the second survey. The average s cale scores for flagged respondents were compared to the averages for other respondents using independent samples t - tests; no significant differences were found between these groups. Conse quently, no respondents were excluded based on the amount of time spent on the survey. 73 Sample characteristics Survey recruitment was restricted to the U.S. MTurk workers only (based on IP addresses) . Among the final sample, the average respondent age was 33.2 years, with a 9.92 year standard deviation. Of those who completed the first survey, 340 respondents identified as White/Caucasian (76%) , 30 identified as Black (7%) , 38 identified as Asian/Pacific Islander (8%) , 16 identified as Multiracial (4%) , and 25 identified as Hispanic (6%) . Because employment status was not formally verified in MTurk (i.e., MTurk workers were invited to opt - in to the survey if they worked 20 hours or m ore per week), work experience was assessed both to further understand characteristics of the sample and to verify work experience variables displayed patterns that real workers would likely self - report. Supervisor eptually related to , and were included with the intention of running ancillary exploratory analyses. As might be expected in a working sample, reported work experience was longest on average ( M =12.6 years, SD =8.63) followed by o rganization tenure ( M =4.93 years, SD =4.36), job tenure ( M =3.98 years, SD =3.53), and supervisor tenure ( M =3.37 years, SD =3.02). Also as per expectations, experience variables were highly correlated, with the lowest correlation emerging between work experien ce and supervisor tenure ( r =.45, p <.05), and the largest correlation emerging between job tenure and organization tenure ( r =.78, p <.05). Respondents reported having 14.5 peers working for the sample supervisor on average; however, the standard deviation i ndicated this number varied widely ( SD =14.17). 74 Although no data was collected on the types of jobs these respondents held, additional evidence supporting their employment status was found in open text responses to the prompt, Please feel free to provide any other thoughts on your relationship with your characteristics they (dis)like about their supervisors, and incidents that have transpired within the LMX relationship. Many of these comments possess a level of detail that would indicate respondents were reflecting on their actual day - to - day experiences hospital, language barriers, and the way decision making styles influence team dynamics. Generally speaking, the comments provided indicated respondents had jobs that played a role in their emotions and stress levels on a regular basis. Survey 1 measures The first survey measured LMXSC (Appendix H ) , locus of causality for RBS (Appendix C ) , LMX (Appendix I ) , and demographics (Appendix J ) . LMXSC. The 6 - item LMXSC c omparison measure used here was developed by Erdogan (2002). The items can be found in Appen dix H . Past measure validation efforts described here are paraphrased from Vidyarthi et al. (2010). Erdogan (2002) originally created an 8 - item measure , attempting to parallel the items of the LMX - MDM measure. The researcher tested the psychometric properties of the measure using data from 261 undergraduate students. A principal components analysis with an oblique rotation was conducted on the 8 LMXSC items, LMX items from Liden and Maslyn (1998), and interactional justice ite ms f rom Niehoff and Moorman (1993). The results showed two LMXSC items loaded on the same components as LMX, and for this reason, those two items were excluded from analyses. The LMXSC factor had no cross - 75 loadings on the LMX or interactional justice compon ents over .20 , and these three factors explained 71.93% of the variance. As Vidyarthi et al. (2010) describe , the measure was further validated using a field sample of 205 employees at a manufacturing company. LMXSC and LMX were positively but moderatel y correlated ( r =.39, p <.01). LMXSC and LMX quality fit well on two separate factors in a confirmatory factor analysis ( 2 (131)= 313.12, CFI=.93, RMSEA=.08) , and an alternative model specifying a correlation between LMXSC and LMX created a significantly w orse fitting model ( 2 (1 ) = 48.53, p <.01). LMXSC was significantly related to interactional justice when LMX was controlled for in a random coefficient regression model ( = .12, SE =.06, p <.05). In all studies measuring LMXSC, reliability was high (Erdogan, 2002, undergraduate sample, = .86, field sample, = .84; Vidyarthi et al., 2010 , = .86). Respondents rated items using a 7 - point Likert scale of agreement . Locus of Causality for Relationship B Based on its favorable psychometric properties observed in the pilot study, the measure in Appendix C was used to assess s elf - and supervisor - attributed locus of causality for RBS. Leader - Member Exchange. LMX was measured us ing a seven - item measure (revised LMX - 7) created by Graen and Uhl - Bien (1995). This me asure can be found in Appendix I . The LMX - 7 is the most commonly used LMX measure (Dulebohn et al., 2012). In their meta - analysis, Gerstner and Day (1997) recommend using the LMX - 7 measure for overall exchange quality due to its higher than average reliability and stronger correlations with outcomes. In a more recent meta - analysis, Dulebohn et al. (2012) found the scale used to measure LMX did not ationships with other constructs. As per Graen and Uhl - use d a different rating scale. 76 Demographics. Demographic information were collected using the measure in Appendix J . Work experience variables were collected to better understand characteristics of the sample (e.g. different forms of tenure should be correlated) to further verify respons es were from working adults. Workgroup size and tenure were included for supplemental analyses. Survey 2 measures The second survey measured interactional justice (Appendix K ) , LMX - related emotions (Appendix L ) , and positive and negative affect (Appendix M ) . Interpersonal justice. To measure interpersonal justice, I used supervisor - focused measure of interactional justice. Only interpersonal justice items were used in hypothesis tests. Informational justice items were collected for anci llary analyses. Past research has shown this measure to have sound psychometric properties. In a sample of 301 university students, Colquitt (2001) found that when distributive, procedural, informational, and interpersonal justice subscale items were inclu ded in the same model, a four factor model exhibited superior fit ( 2 = 769.50, df =406, CFI=.92, RMSEA=.055) in comparison to a three factor model that collapsed interpersonal justice and informational justice into a more general interactional justice factor ( 2 = 965.40, df =413, CFI=.88, RMSEA=.067) and a one factor model summarizing all the justice items ( 2 = 2,057.28, df =424, CFI=.65, RMSEA=.113). This four factor model also exhibited superior fit in a field sample of 337 employees at an automobile m anufacturing company ( 2 = 845.52, df =406, CFI=.94, RMSEA=.057) in comparison to the three factor model ( 2 = 1,776.75, df =413, CFI=.81, RMSEA=.099) and the one factor model ( 2 = 3,235.90, df =424, CFI=.61, RMSEA=.140). These scales show good internal consist ency (university sample, interpersonal justice, = .79 ; field sample, interpersonal justice, = .92 ). 77 These interactional justice items combined have shown to be highly correlated with perceived organizational support ( r =.41, p <.01) and supervisory trust ( r = .66, p <.01; Ambrose & Schminke, 2003). This me asure can be found in Appendix K . Items were rated using a 1 - 5 frequency scale (1= Never to 5= A great deal). LMX - related Emotions . Because emotions are measured in a variety of ways, I chose to review exta nt research measuring pride, gratitude, guilt, and /or shame. The measures referred to here can be found in Appendix L . I first review single item measures, qualitative measures , and last , multiple item measures. Most articles I reviewed from the field of organizational psychology measure emotions using single item measures rated on Likert scales . Barclay, Skarlicki, and Pugh (2005) used single item s to ask laid - off employees the extent to which they felt guilt, shame, and anger regarding the ir co layoff. Weiss et al. (1999) , Rupp and Spencer (2006) , and Coulter and Pinto (1995) also used Likert ratings of emotion adjectives for their measures. Otterbacher and asks respondents to select an adjective from 1 1 guilt - relevant adjectives ; however many of the PGI adjectives are not at a basic reading level (e.g. , chagrined). I decided against the use of single items to measure pride, gratitude, guilt, and shame , because multiple - item measures tend to exhibit bett er psychometric properties than single item measures and emotions are a focal dependent variable of the model. Qualitative approaches have also been used to measure emotions. Friedman et al. (2004) Word Count to assess the level of anger in communications sent to an online dispute negotiation service . In the current study, participants are not focusing on a specific , emotionally laden situation, but instead are reflecting 78 on their relationship with t heir supervisor . For this reason, a qualitative assessment of emotions was not deemed a viable option for the current study. I next reviewed multi ple item measures of pride, gratitude, guilt, shame, and anger ; I describe and comment on them in that order. As noted previously, p ride is [emotion] generated by appraisal that one is responsible for a socially valued outcome or for being a socially - Measures of pride in organizational psychology of ten look at pride as it relates to organizational membership. For example, Tyler and Blader (2003) assessed pride using items that focus on positive impressions r how membership T hese items focus on feelings towards the organization and not towards indi cating these items would need to be modified to be applicable to the current study. The State Shame and Guilt Scale (SSGI ; Marschall, Sanfter, & Tagney, 1994 ) includes a generalized pride subscale . As per Mascolo and the SSGI pride items focus on feeling valued This subscale has shown high internal consistency in past research ( = .93; Stoebe r, Hutchfield, & Wood, 2008). For the current study, I used the pride subscale from the SSGI, with instructions that ask the respondent t o rate how they feel about the relationship they have with their supervisor. This measure can be found in Appendix N . Several - reply simultaneously to two descriptors (e.g. , 79 into two items (e.g. , valuable as one item, worthwhile as another) to clarify the meaning of each item. Several multiple item gratitude measures are frequently used in research today; I considered four in my review . Both McCullough, Emmons, and Tsang (2002) and Adler and Eagly (2005) measure gratitude as a disposition, making their measure a poor fit for the current study. Watkins, W and appreciation of others. The items probe respondent beliefs (e.g. , I believe that I am a very fortunate person not the emotion s of interest in this study. Gordon, Im pett, Kogan, Oveis, (2012) measure assesses the extent to which respondents act appreciative ly and the extent to which they feel appreciated by their relationship partner . Although the relationship often tell my partner how much I appreciate her/him the measure was originally crafted for research on romantic relationships, and several items have content that r uck with a sense of awe and . used item content (e.g. framed similar ly to the SSGI items. Respondents responded to these items in relation to how they feel about their relationship with their supervisor. These items can be found in Appendix N . Looking next to guilt and shame , recall that guilt is assumed to occur when i ndividuals blame negative aspects of their behavior on themselves, and shame is assumed to occur when individuals blame negative aspects of their ability or character on themselves (Tracy & Robins, 80 2006, p. 1340) . A review of existing measures showed m any focus on chronic guilt (e.g. , Buss & Durkee, 1957; Klass, 1987; Kugler & Jones, 1992; Mosher, 1966) or chronic shame (e.g. , Cook, 1989), As with pride, Mars c subscales provide relatively straightforward items that focus on generalized feelings of guilt (e.g. , shame (e.g. , shown favorable internal consistency in past research (guilt, =.89; shame, =.88, Stoeber et al. , 2008). For the present study, I reframe so that respondents answer regarding how they feel about the ir relationship with their supervisor. Again, double - barreled items were broken into separate items for added item content clarity. The resulting items can be found in Appendix N . Lastly, only one multiple item measure of state anger was found . (19 88 ) State Trait Anger Expression Inventory (STAXI) has a state - anger subscale which consists of a number of adjectives respondents use to report th eir current emotional state. In a principal axis factor analysis, these items loaded on a factor (i.e. , all loadings were >.43) separate from other subscales reflecting anger co ping techniques and trait anger (Fuqua et al., 1991). For this study, I chose to use several of these adjectives in a similar format as was used for pride, guilt, shame, and gratitude (see Appendix L ). I dropped adjectives that were strongly worded and did not seem appropriate for a workplace survey (i.e. , feel like hitting). These items were rated on a 5 - point Likert scale ranging from 1=not at all to 5= extremely. 81 Positive and n egative a ffect. Much of the existing LMX research uses positive and negative affect (PANA) framework of affect ( e.g. , Aquino, Lewis, & Bradfield, 19 99). As a emotion measure , I measured positive and negative affect . To do this, I used 20 - item Positive and Negative Affect Schedule ( PANAS ) measure. Wats on et al. (1993) describe the measure as assessing positive and negative moods, with on of subjective distress and unpleasurable engagement that subsumes a variety of aversive mood states, including anger, contempt, disgust, The scale consists of 10 adjective items measuring positiv e affect (e.g. , enthusiastic) and 10 adjective items negative affect (e.g. , nervous) , rat ed on a 5 - point Likert scale (i.e., very slightly or not at all , a little , moderately , quite a bit , and extremely ) . Watson et al. (1993) showed PA and NA scales that assess how participants felt in the past several weeks showed reasonabl e internal consistency ( n =586, =.87 and =.87, respectively) and were correlated in the expected direction with the Beck Depression I nventory (NA, r =.58; PA, r = - .36) and STAI State Anx iety Scale ( NA, r =.51; PA, r = - .35). feelings regarding their relationships with their superviso r. The instructions read: Read each item and then, using the scale belo w, mark the appropriate answer in the space next to that word. Indicate to what extent you feel this way about your relationship with your manager/immediate This measure can be found in Appendix M . 82 RESULTS The plan of analysis was as follows. First, the psychometric rigor of all study measures was assessed. Based on the observation that measures met psychometric standards , I next conducted study hypotheses tests using bivariate correlations (i.e., H1 and H4) and hierarchical regression s (i.e., H2, H3, H5, and H6.) Then, I conducted ancillary analyses wherein I analyzed data from measures of constructs that were conceptually similar to study constructs (i.e., LMXSC and LMX, interpersonal and informal justice, and emotions and PANA) . Alth ough these supplemental measures afforded the possibility of conducting a range of ancillary analyses, the most intuitive and interes ting research questions focus ed on a) the extent to which these supplemental measures were relate d to/distinct from focal s tudy measures, and b) if and how the results of th e hypothesis tests would have changed had been used in place of a given study measure. Thus, these two questions largely guided the ancillary analyses using LMX, informational justice, and positive/negative affect. Next, I looked at whether study measures showed significant differences based on race/ethnicity. These analyses were considered ancillary, as there was no theoretical reason to expect differences based on race/ethnicity. Last, I ran ancillary analyses on workgroup size and w ere made regarding when and how they would influence key study constructs and their relationships with one another. Thus, I explored the main effe ct of workgroup size and LMX tenure on study variables, as well as if/how they moderated study hypotheses. I also ran several alternative models to see if they presented plausible alternative frameworks for understanding the data. These models included testing whether a LMXSC - interpersonal justice interaction 83 predicted supervisor - attributed locus of causality fo r RBS, as well as testing hypothesis models with alternative categories of emotions (self - conscious vs. externally - focused) and alternative locus of causality for RBS attributions (self - vs. supervisor - ) entered as controls. Findings based on this analysis plan are reported in the sections that follow. T o assess the psychometric rigor of the study measures, EFAs were conducted to examine the emergent factor structure among item s. 6 First, I ran a maximum likelihood extraction EFA on all study items . ( In - d epth comparisons of study items with LM X and PANA items are discussed in later sections . ) Because study measures were assumed to be related, factors were allowed to correlate using an oblique Promax rotation. The resulting factor structure is show n in Tabl e 9 . An eight - factor solution was extracted using the Kaiser criterion of eigenvalues over 1. 6 Item descriptive statistics a nd intercorrelations for all study measures can be found in Appendix O. 84 Table 9 Promax Rotated Maximum Likelihood Extraction Pattern Matrix for All Main Study Items Factor 1 2 3 4 5 6 7 8 LMXSC1 .72 LMXSC2 .60 LMXSC3 .82 LMXSC4 .83 LMXSC5 .93 LMXSC6 .92 LoCRBS Self 1 .82 LoCRBS Self 2 .64 LoCRBS Self 3 .73 LoCRBS Self 4 .73 LoCRBS Self 5 .71 LoCRBS Self 6 .65 LoCRBS Supervisor 1 .76 LoCRBS Supervisor 2 .58 LoCRBS Supervisor 3 .71 LoCRBS Supervisor 4 .70 LoCRBS Supervisor 5 .71 LoCRBS Supervisor 6 .75 Interpersonal Justice 1 .63 Interpersonal Justice 2 .72 Interpersonal Justice 3 .31 .65 Interpersonal Justice 4 .45 Informational Justice 1 .31 Informational Justice 2 .70 Informational Justice 3 .67 Informational Justice 4 .67 Informational Justice 5 .44 Pride 1 .91 Pride 2 .91 Pride 3 .89 Pride 4 .84 Pride 5 .79 Pride 6 .88 Pride 7 .83 Guilt 1 .69 Guilt 2 .55 Eigenvalue 20.00 5.53 3.48 2.61 1.77 1.35 1.28 1.15 Variance explained 37.04% 10.24% 6.44% 4.83% 3.27% 2.49% 2.38% 2.13% 85 Guilt 3 .62 Guilt 4 .76 Guilt 5 .76 Shame 1 .84 Shame 2 .71 Shame 3 .92 Shame 4 .81 Shame 5 .79 Shame 6 .72 Shame 7 .55 Gratitude 1 .48 Gratitude 2 .51 Gratitude 3 .83 Gratitude 4 .68 Anger 1 .66 Anger 2 .78 Anger 3 .79 Anger 4 .80 Eigenvalue 20 .00 5.53 3.48 2.61 1.77 1.35 1.28 1.15 Variance explained 37.04% 10.24% 6.44% 4.83% 3.27% 2.49% 2.38% 2.13% Note. Loadings under .30 are suppressed. Pairwise deletion was used, item N s = 395 - 453 . Overall, items tended to load on factors with other items from the same measure. The first factor, which accounted for 37. 04 % of the variance, was driven by self - conscious negative emotion items (i.e., guilt and shame ) . The second factor account ed for 10.24 % of the total variance and was driven by positive emotions ( i.e., gratitude and pride) and one interpersonal justice ite . The third factor explained 6.4 4 % of the total variance and was driven by LMXSC items. The fourth factor was largely driven by self - attributed locus of causality for RBS, and the fifth was largely driv en by supervisor - attributed locus of causality for R BS (respectively explaining 4.83 % and 3.2 7 % of the total variance). Informational justice items loaded highly on the si xth factor, which explained 2. 49 % of the total variance. The seventh factor explained 2. 38 % of the total variance, and was driven by the anger items . Finally, the eight h factor explained 2. 13 % of the total variance, and 86 was driven by interpersonal justice, with the third interpersonal justice item cross - loading with the second factor. The factor correlation matrix can be found in Appendix P. Hypothesis 5 l justice and positive emotions may have contributed to the interpersonal justice cross - loading . To avoid cross - loadings based on the hypot hesized relationships between IVs and DVs and to assess whether predictor items in isolation emerged on their respective factors , I ran a second factor analysis using only predictor items . As can be seen in Table 1 0 , items loaded cleanly on their respectiv e factors with no cross - loadings. The factor correlation matrix is featured in Appendix Q. Table 1 0 Promax Rotated Maximum Likelihood Extraction Pattern Matrix for Main Study Predictors Factor 1 2 3 4 LMXSC 1 .75 LMXSC 2 .62 LMXSC 3 .83 LMXSC 4 .85 LMXSC 5 .95 LMXSC 6 .90 Locus of causality RBS Self 1 .86 Locus of causality RBS Self 2 .68 Locus of causality RBS Self 3 .75 Locus of causality RBS Self 4 .71 Locus of causality RBS Self 5 .73 Locus of causality RBS Self 6 .69 Locus of causality RBS Supervisor 1 .75 Locus of causality RBS Supervisor 2 .57 Locus of causality RBS Supervisor 3 .70 Locus of causality RBS Supervisor 4 .72 Locus of causality RBS Supervisor 5 .70 Locus of causality RBS Supervisor 6 .77 Interpersonal Justice 1 .85 Interpersonal Justice 2 .94 Interpersonal Justice 3 .91 Interpersonal Justice 4 .62 Eigenvalue 7.39 3.37 2.28 1.65 Variance explained 33.59% 15.32% 10.36% 7.51% 87 Note. Loadings under .30 are suppressed. Pair wise deletion was applied, item N s = 399 - 453. Psychometric properties of the measures used in the study were also examined using confirmatory factor analyses (CFA). T he current sample was of ~400 cases did not meet the 10 cases per parameter rule of thumb (Kline, 2004). However, item parceling is not desirable for multidimensional CFAs (Bandalos, 2002). Thus, item parceling was not used, and the smaller sampled size must be taken into account when interpreting results prese nted here. First, a nine factor model was run with LMXSC, self - attributed locus of causality for RBS, supervisor - attributed locus of causality for RBS, interpersonal justice, pride, guilt, shame, gratitude, and anger scale items loading on t heir own respective factor s and all factors correlating to one another. The model yielded acceptable fit, 2 (1091) = 3257 . 80 , CFI = .87, RMSEA = .0 7 , SRMR = .08. Second, b ased on the close relationship between shame and guilt both conceptually and in the EFA resu lts shame and guilt were collapsed into a single factor. The resulting eight factor model showed significantly more misfit than the nine factor model , 2 (8) = 210. 55 , p <.01, CFI = .8 5 , RMSEA = .07, SRMR = .08. Third, positive and negative emotions were collapsed on their own factor s , creating a six - factor model. This six factor model also showed significantly more mis fit than the nine factor model, 2 (21) = 610.139 , p <.01; CFI = .8 3 , RMSEA = .07, SRMR = .08. Last, a two factor model was tested, with all predictor items loading on one factor and all emotions loading on a second factor. This model exhibited by far 2 ( 35 ) = 4663.05 , p <.01; CFI = . 58 , RMSEA = . 11 , SRMR = . 15 . Thus, the nine factor model proved to be the best fitting model, and measure items were used to generate each respective scale score. Next, I ran scale d escriptive statistics, correlations , and internal consistencies for all measures , which are displayed in Table 1 1 . consistency, and correlations with other variables are discussed in depth in a supplementary 88 section. In general, a ll multiple item scales showed accepta ble reliabilities, ranging from = .85 (supervisor - attributed locus of causality for RBS) to = .95 (pride). Because locus of causality for RBS was viewed as a moderating variable within the particularly strong relationships with other study variables. Surprisingly, self - attributed locus of control for RBS exhibited significant relationships with most study variables and showed stronger associations with emotions than hypothesized LMXSC - emotio n relationships. This difference was particularly large for negative emotions, where self - attributed locus of causality for RBS showed negative relationships that were close to twice as strong as those with LMXSC. In general, supervisor - attributions were n egatively linked to positive perceptions, like interpersonal justice ( r = - .20, p <.05) and pride ( r = - .22, p <.05), and positively related to negative perceptions like shame ( r =.22, p <.05) and anger ( r =.26, p <.05). The weak negative relationship between grati tude and supervisor - attributed locus of causality for RBS ( r = - .15, p <.05 ) was counter to expectations, given gratitude is assumed to involve external causal attributions. 89 Table 1 1 Descriptive Statistics, Reliabilities, and Correlations for Main Study M easures Items M SD N 1 2 3 4 5 6 7 8 9 1. LMXSC 6 4.12 1.35 452 (.92) 2. LoCRBS Self 6 3.49 .76 453 .56 (.88) 3. LoCRBS Sup. 6 3.41 .69 453 - .10 - .26 (.85) 4. LMX 7 3.57 .83 453 .58 .51 - .21 (.92) 5. Inter. J. 4 4.33 .74 401 .31 .36 - .20 .58 (.88) 6. Inform. J. 4 3.88 .78 401 .29 .38 - .14 .63 .69 (.87) 7. Pride 7 4.01 1 .00 401 .46 .47 - .22 .68 .72 .68 (.95) 8. Guilt 5 1.5 .8 0 400 - .11 - .28 .19 - .42 - .53 - .50 - .55 (.88) 9. Shame 6 1.38 .75 401 - .16 - .33 .22 - .43 - .60 - .52 - .63 .79 (.92) 10. Gratitude 4 3.41 1.17 401 .42 .46 - .15 .61 .62 .63 .74 - .41 - .47 11. Anger 4 1.48 .94 401 - .22 - .40 .26 - .48 - .59 - .52 - .59 .67 .73 12. P. Affect 10 3.02 .93 401 .53 .48 - .02 .59 .43 .54 .63 - .25 - .28 13. N. Affect 10 1.41 .63 401 - .15 - .28 .21 - .46 - .66 - .55 - .62 .73 .77 14. Work Exp. 1 12.6 8.63 402 .01 .03 - .03 .10 .03 .02 .02 .03 .04 15. Job Tenure 1 3.98 3.53 422 <.01 <.01 .04 .11 .03 - .01 .05 > - .01 <.01 16. Org. Tenure 1 4.93 4.36 420 <.01 > - .01 .01 .13 .05 > - 0.01 .04 - .02 > - .01 17. Sup. Tenure 1 3.37 3.02 421 .14 .05 .01 .19 .04 .04 .06 - .01 .02 18. N Peers 1 14.5 14.17 453 - .06 - .06 .02 - .09 - .02 - .06 - .06 .12* .02 19. Age 1 33.2 9.92 452 - .05 - .01 > - .01 .08 .07 .04 .06 - .01 - .03 Note . All correlations in bold were significant, p <.05. Reliability coefficients are in parentheses on the diagonal. LoC for RBS - Self= self - - Supervisor= supervisor - attributed l ocus MXSC was rated on a 7 - point Likert scale; all other scales were rated on a 5 - point Likert scale entered as 61. 90 Table 1 1 10 11 12 13 14 15 16 17 18 19 10. Gratitude (.87) 11. Anger - .50 (.94) 12. Positive Affect .62 - .33 (.93) 13. Negative Affect - .47 .74 - .28 (.92) 14. Years of work experience .05 .04 .05 .05 ( - ) 15. Years in current position .12 .02 .05 - .01 .50 ( - ) 16. Years at organization .12 - .01 .06 .01 .54 .78 ( - ) 17. Years working for supervisor .10 .01 .09 - .01 .45 .72 .71 ( - ) 18. N Peers - .09 .07 - .04 <.01 .01 .07 <.01 - .01 ( - ) 19. Age .03 .03 .01 - .02 .78 .52 .55 .45 .03 ( - ) LMX showed a higher internal consistency alpha ( =.92) than others have observed using the LMX - 7 (average internal consistency across meta - analysis, =.89, Gerstner & Day, 1997). The mean and standard deviation were relatively similar to the descriptive statistics found in other samples (e.g. current study, M =3.57, SD =.83; Maslyn & Uhl - Bien, 2001, M =3.77, SD =78). LMX showed moderate ly size relationships with all study IVs and DVs, wit h the exception of supervisor - attributed locus of causality for RBS ( r = - .21). LMX showed particularly strong relationships with pride ( r =.68, p <.05), informational justice ( r =.63, p <.05), and gratitude ( r =.61, p <.05). LMX exhibited weak but significant pos itive relationships with years of work Interpersonal and informational justice showed acceptable internal consistency (respectively, =.88, and =.87), despite being slightly lower than alphas previously observed in field samples ( interpersonal , =.92 ; informational, =. 90; Colquitt, 2001) . Though standard deviations were of a similar size for both dimensions of justice, interpersonal justice was higher on average ( M =4.33) than informational justice ( M =3.88). As will be discussed further in the sections that follow, both subscales showed moderate to strong relationships with dif ferent 91 emotions. These relationships were stronger than LMXSC - emotions linkages; the strongest linkage was r = - .50, p <.05 (i.e., guilt and informational justice), whereas the weakest LMXSC - emotion linkage was r = - 11, p <.05 (i.e., guilt and LMXSC). The relati onships with interactional justice dimensions and emotions are likely inflated due to common method variance. Negative affect from the PANA scale showed the lowest mean (1.41) and the lowest standard deviation (.64) of all scales. The highest correlatio n to emerge was that between guilt and shame ( r =. 79 , p <.05 ). Workplace experience variables largely did not correlate with study measures. Positive r =.11, p <.05). A s des cribed in the sample description, a ll workplace experience variables showed moderate ly sized positive correlations with one another . For reference, the findings from all hypothesis tests and additional analyses that paralleled hypothesis tests are in Table 1 2 . Note that columns 3, 4, and 5 are ancillary analyses Because these analyses were examined in an exploratory manner and are not a test of a pr iori was conducted. 92 Table 1 2 Summary of Hypothesis Fin dings and Ancillary Analyses LMXSC PANA LMX Informational Justice Hypotheses Supported? Supported? Supported? Supported? Hypothesis 1 H1a: LMXSC+ Positive Emotions Supported Supported Supported - H1b: LMXSC - Negative Emotions S upported Supported Supported - Hypothesis 2 H2a: LocRBS x LMXSC Pride Partially supported Not supported Not supported - H2b: LoCRBS x LMXSC Guilt Not supported Not supported Not supported - H2c: LoCRBS x LMXSC Shame Not supported Not supported Not supported - Hypothesis 3 H3a: LoCRBS x LMXSC Gratitude Supported Not supported Supported - H3b: LoCRBS x LMXSC Anger Partially supported Not supported S upported - Hypothesis 4 H4a: IJ + Gratitude Supported Supported - Supported H4b: IJ - Anger Supported Supported - Supported Hypothesis 5 H5a: IJ x LMXSC Gratitude Not supported Not supported Not supported Not supported H5b: IJ x LMXSC Anger Partially supported Not supported Supported Partially supported Hypothesis 6 H6: IJ x LMXSC x LoCRBS Anger Not supported Not supported Not supported Not supported 93 Hypothesis tests Hypothesis 1 a predicted LMXSC would be positively associated with positive emotion s. P ride and gratitude showed moderate positive correlations with LMXSC ( Table 1 1 ; respectively, r =.46, p <.05; r =.4 2 , p <.05). Table 1 3 shows that in second step of the hierarchical regression used to test hypothes e s 2 and 3, LMXSC exhibited a significant main effec t on pride ( =. 30 , p <.05), and gratitude ( =.35, p <.05). LMXSC also showed a significant positive main effect on gratitude in the second step of the hypothesis 5 regression found in Table 1 1 ( =. 25 , p <.05). Thus, H1a was supported. Hypothesis 1b predicted LMXSC would be negatively associated with negative emotions. G uilt, shame , and anger showed weak negative correlations with LMXSC ( Table 11 ; respectively, r = - .1 1 , p <.05; r = - .1 6 , p <.05 ; r = - .2 2 , p <.05). The hierarchical regressions in Table 1 3 indicate LMXSC had a significant main effect on anger ( step 2; = - .1 3 , p <.05); but no significant main effect on guilt ( step 2; =.0 7 , n.s. ) or shame ( step 2; = - .0 2 , n.s. ). Interestingly, when interpers onal justice was included in a regression ( see Table 1 4 ), LM XSC showed no main effect on anger ( = - .0 3 , n.s.). Although regression s did not show consistent significant main effects of LMXSC on negative emotions , the hypothes e s focuse d on the bivariate relationships (i.e. correlations) and did not explicitly indicate other variables as controls (as occurs in the regressions). Thus, based on the significance of the bivariate correlations, H 1 b was supported. Hypothesis 2a predicted LMXSC would interact with self - attributed locus of causality for RBS in predicting pride. As part of this interaction, LMXSC was expected to show a positive main effect. Those high on both predictors were expected to report the highest level of pride. Tho se with low LMXSC and high self - attributed locus of causality for RBS were expected to 94 have the lowest level of pride. Individuals with low self - attributed locus of causality for RBS and high LMXSC were expected to report higher pride than those low on bot h predictors. 95 Table 1 3 Tests of Hypo thesis 2 and 3 with Emotions Regressed on LMXSC, Locus of C au sality for RBS, and an LMXSC x Locus of Causality I nteraction Self - attributed Locus of Causality for RBS Pride Guilt Shame Predictors variables SE B SE B SE B Step 1 .27* .09* .1 2 * Locus of causality for RBS .07 .4 1 .3 1 * .06 - .34 - .33* .0 6 - .36 - .37* LMXSC .04 .21 .2 8 * .03 .0 4 .0 7 .03 .0 2 .0 4 Step 2 .01* <.01 .02* Locus of causality for RBS .08 . 31 . 24 * .07 - .3 3 - .31* .0 7 - .26 - .27* LMXSC .04 . 22 . 30 * .03 .0 4 .0 7 .03 .01 .0 2 LoC for RBS x LMXSC .04 - . 09 - .1 2 * . 0 4 .0 1 .0 2 .03 .09 .16* Supervisor - attributed Locus of Causality for RBS Gratitude Anger SE B SE B Step 1 .18* .1 0 * Locus of causality for RBS .08 - .1 7 - .1 0 * .06 .3 1 . 24 * LMXSC .04 .34 .40* .03 - .1 3 - . 19 * Step 2 .02* .0 3 * Locus of causality for RBS .08 - .13 - .08 .06 .2 8 .2 1 * LMXSC .04 .30 .35* .03 - . 09 - .1 3* LoC for RBS x LMXSC .05 .1 7 .16* .04 - .1 4 - .1 7 * Note. * p <.05. Ns ranged 387 - 388 . 96 Table 1 4 Tes ts of Hypothesis 5, with Gratitude and Anger Regressed on Interpersonal J ustice, LMXSC, and an Interactional Justice x LMXSC I nteraction Self - attributed Locus of Causality for RBS Gratitude Anger Predictors variables SE B SE B Step 1 .44* .3 7 * Interpersonal Justice .06 .85 .54* .05 - .7 4 - .5 9 * LMXSC .03 .21 .2 5 * .03 - .0 3 - .0 4 Step 2 < .01 .02* Interpersonal Justice .07 .83 .53* .05 - .65 - .5 3 * LMXSC .03 .21 .2 5 * .03 - .0 2 - .0 3 IJ x LMXSC .04 - .0 4 - .04 .03 .1 2 .1 5 * Note. * p <.05. N = 388 for regressions shown here. IJ= Interpersonal Justice. Table 1 3 shows a significant interaction between LMXSC and self - attributed locus of causality for RBS ( = - .1 2 , p <.05) which explained significantly more variance in pride than only the main effects of LMXSC and self - attributed locus of causality for RBS ( =.01, p <.05). This interaction is displayed in Figure 10 . Three aspects of the interaction were as expected. First, a main effect of LMXSC emerged, with high LMXSC respondents showing higher pride than low LMXSC respondents. Second, those with high LMXSC and high locus of causality for RBS showed the highest levels of pride. Third, those with low locus of causality for RBS and high LMXSC had higher pride than those with low self - attribu ted locus of causality for RBS and low LMXSC. However, low LMXSC , high self - attributed locus o f causality for RBS individuals did not meet expectations. That is, instead of showing the lowest levels of pride, respondents with low LMXSC and high self - attri buted locus of causality for RBS showed the third highest levels of pride. T hose with low LMXSC and low self - attributed locus of causality for RBS showed lower levels of pride. Th e interaction shows the self - attributed locus of causality for RBS had a posi tive main effect on levels of pride for both high and low LMXSC respondents ( = .2 4 , p <.05). Thus, H2a was partially supported. 97 Figure 10 LMXSC x Locus of Causality Interaction Predicting Pride N ote. N = 388 . S elf - attributed locus of causality for RBS was also predicted to interact with LMXSC in predicting shame ( H 2b) and guilt ( H 2c) . Subordinates with high LMXSC were expected to show the low est level of guilt and shame regar dless their level of self - attributed locus of causality for RBS. For low LMXSC individuals, self - attributed locus of causality was expected to be positively related to levels of guilt and shame , with low LMXSC - high self - attributed locus of causality individuals showing the highest levels of guilt and shame. As can be seen in Table 1 3 , this interaction did not have a significant effect on guilt ( =. 02 , n.s.) , indicating H2b was not supported . S elf - attributed locus of causality for RBS showed a significant negative main effect on guilt that was not hypothesized ( = - .31, p <.05) . The i nteraction was significant when shame was used as the dependent variable ( =.16, p <.05) , adding significant variance explained above and beyond the main effects ( =.02, 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 Low LMXSC High LMXSC Pride Low self-attributed locus of causality for RBS High self-attributed locus of causality for RBS 98 p <.05). This interaction, shown graphically in Figure 11 , took on a different form than was expected. Among high LMXSC individuals, self - attributed locus of causality for RBS showed a weak negative association with shame. Among low LMXSC individuals, self - attributed locus of causality for RBS was expected to be positively related to sha me, but instead, it was negatively related to shame. Self - attributed locus of causality for RBS indeed showed a significant main effect on shame (step 2, =. - .27 , p<.05), whereas LMXSC exhibited no main effect in the model (step 2, = .0 4, n.s.). Low LMXSC, high self - attributed locus of causality individuals were expected to show the highest level of shame, and instead, this group reported the lowest level of shame. Based on these findings, H2c was not supported. Figure 11 LMXSC x Locus of Causality Interaction Predicting Shame Note. N = 388 . Hy pothesis 3a predicted LMXSC would interact with supervisor - attributed locus of causality for RBS in predicting gratitude. Those with high supervisor - attributed locus of 0.6 0.8 1 1.2 1.4 1.6 1.8 Low LMXSC High LMXSC Shame Low self-attributed locus of causality for RBS High self-attributed locus of causality for RBS 99 causality for RBS a nd high LMXSC were expected to show the highest gratitude; those with high supervisor - attributed locus of causality for RBS and low LMXSC were expected to show the lowest gratitude. LMXSC was expected to show a main effect, such that for individuals report ing low supervisor - attributed locus of causality for RBS , high LMXSC respondents would exhibit higher levels of gratitude tha n low LMXSC respondents . As shown in Table 1 3 , the interaction between LMXSC and supervisor - attributed locus of causality for RBS was significant ( = .16 , p <.05) and explained significantly more variance in gratitude above and beyond the main effects ( =.02 , p <.05). As Figure 12 shows , LMXSC had a main effect ( =.35, p <.05) . High LMXSC respondents showed high levels of gratitude, with high supervisor - attributed locus of causality for RBS respondents showing marginally higher levels of gratitude than low supervisor - attributed locus of causality for RBS respondents . For low LMXSC respondents, s upervisor - attributed locus of causality for RBS had a strong negative main effect, with low LMXSC - high supervisor - attributed locus of causality respondents reporting the lowest levels of gratitude. Overall, the results supported H3a. 100 Figure 12 LMXSC x Locus of Causality for RBS interaction pr edicting G ratitude Note. N = 388 . Hypothesis 3 b predicted LMXSC would interact with supervisor - attributed locus of causality for RBS in predicting anger. LMXSC was expected to show a main effect, with high LMXSC respondents displaying low levels of anger regardless of the level of supervisor - attributions and low LMXSC respo ndents displaying higher levels of anger as supervisor - attributed locus of causality for RBS increased. Table 1 3 shows the LMXSC - supervisor - attributed locus of causality for RBS interaction was significant ( = - .1 7 , p < .05) and explained significantly more variance in anger than only main effects ( =.0 3 , p <.05). The interaction graph ca n be found in Figure 13 . As expected, low LMXSC respondents display ed higher levels of anger as supervisor - attributed locus of causalit y for RBS increased. High LMXSC respondents reported low er anger, with high 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 Low LMXSC High LMXSC Gratitude Low supervisor- attributed locus of causality for RBS High supervisor- attributed locus of causality for RBS 101 supervisor - attributed locus of causality for RBS respondents showing slightly higher anger than low supervisor - attributed locus of causality for RBS. Counter to the expected main e ffect of LMXSC, individuals reporting low LMXSC and low supervisor - attributed locus of causality for RBS had the lowest levels of anger, lower than individuals with high LMXSC. Despite this discrepancy, many expectations were met, and H3b was partially sup ported. Figure 13 LMXSC x Locus of C ausality for RBS I nteraction P redicting Anger Note. N = 388. Hypothesis 4 predicted that interpersonal justice would be positively related to gratitude (H4a) and negatively related to anger (H4b). Table 1 1 show s interpersonal justice exhibited a positive relationship with gratitude ( r =.6 2 , p <.05) and a negative relationship with anger ( r = - . 5 9, p <.05). The second step of the regressions in Table 1 4 also showed interpersonal justice to have a positive main effect o n gratitude ( =.53, p <.05) and a negative main effect on anger ( = - .5 3 , p <.05). Thus, H4a and H4b were supported. 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 Low LMXSC High LMXSC Anger Low supervisor- attributed locus of causality for RBS High supervisor- attributed locus of causality for RBS 102 Hypothesis 5a predicted interpersonal justice and LMXSC would interact in predicting gratitude. LMXSC was predicted to show a main effect, wi th high LMXSC respondents showing higher gratitude than low LMXSC respondents. For high LMXSC individuals, interpersonal justice was expected to have a strong positive main effect on gratitude. For low LMXSC, interpersonal justice was expected to show a we aker main effect on gratitude. Table 1 4 shows no significant interpersonal justice - LMXSC interaction when gratitude was entered as the dependent variable. Hypothesis 5b predicted LMXSC and interpersonal justice would interact in predicting anger. LMXSC w as expected to show a main effect, with low LMXSC respondents showing higher anger than high LMXSC respondents. For both high and low LMXSC respondents, interpersonal justice was expected to be negatively associated with anger; however, this relationship w as expected to b e stronger when LMXSC was low. Table 1 4 shows the LMXSC - interpersonal justice interaction was significant ( =.1 5 , p <.05) and explained significantly more variance in anger than only the main effects of LMXSC and interpersonal justice ( = .02, p <.05). The graph of this inter action can be found in Figure 14 . As per expectations, higher levels of interpersonal justice were related to lower level s of reported anger. Also in line with expectations, the highest level of anger was displayed by lo w LMXSC, low interpersonal justice respondents. Whereas the lowest level of anger was expected to come from participants reporting high interpersonal justice and high LMXSC, those with high interpersonal justice and low LMXSC instead reported the lowest levels of anger. Thus, H5b was partially supported. 103 Figure 14 LMXSC x Interpersonal Justice I nteraction P redicting Anger Note. N =388. Hypothesis 6 predicted that interpersonal justice would display a three - way inte raction with LMXSC and supervisor - attributed locus of causality for RBS in predicting anger. LMXSC and interpersonal justice w ere expected to show negative main effects on anger. Low LMXSC were expected to report increasing levels of anger as supervisor at tributions increased; this slope was expected to be steeper when interpersonal justice is low and less steep when interpersonal justice is high. Further, anger levels experienced by high LMXSC respondents were expected to be unaffected by supervisor attrib utions. As can be seen in step 3 of Table 1 5 , the three - way interaction was no t significant, indicating H6 was not supported. An unexpected two - way interaction was observed between interpersonal justice and supervisor - attributed locus of causality for RB S ( =. - 1 4 , p <.05) . As can be seen in step 2 of Table 1 6 , the sam e interaction was significant when LMXSC was not 0 0.5 1 1.5 2 2.5 Low LMXSC High LMXSC Anger Low Interpersonal Justice High Interpersonal Justice 104 included in the model ( = - .1 8 , p <.05) and explained significant variance beyond the model main effects ( =.03, p <.05) . Figure 15 shows a graphical representation of this interaction (based on the Table 1 6 model). Those who reported high levels of interpersonal justice reported low levels of anger, regardless of their level of supervisor - attributed locus of causality for RBS. Respon dents who perceived low levels of interpersonal justice exhibited higher levels of ange r, with high supervisor attributions showing the highest level of anger and those with low supervisor attributions showing slightly lower levels of anger Table 1 5 Tests of Hypothesis 6, with A nger R egressed on a I nterpersonal Justice x LMXSC x S upervisor A ttributed L ocus of C ausality for RBS Interaction Anger Predictors variables SE B Step 1 .3 9 * Interpersonal Justice .05 - .7 1 - .5 7 * LMXSC .03 - .0 2 - .0 3 LoCRBS Supervisor .05 .19 .14* Step 2 .03* Interpersonal Justice .0 6 - .6 4 - .5 1 * LMXSC .03 < .01 < .0 1 LoCRBS Supervisor .05 .1 5 .1 1 * LMXSC x IJ .04 .0 3 .0 4 LMXSC x LoCRBS Supervisor .04 - .0 2 - .0 2 IJ x LoCRBS Supervisor .07 - .2 2 - .15* Step 3 <.01 Interpersonal Justice .0 6 - .6 3 - .5 1 * LMXSC .03 < .01 < .0 1 LoCRBS Supervisor .06 .1 4 .1 1 * LMXSC x IJ .0 2 .05 .0 3 LMXSC x LoCRBS Supervisor .04 - .0 2 - .0 2 IJ x LoCRBS Supervisor .08 - . 20 - .1 4 * IJ x LMXSC x LoCRBS Supervisor .05 .02 .03 Note. * p <.05. N = 388 . IJ=Interpersonal Justice; LoCRBS=locus of causality for RBS. 105 Table 1 6 Hypothesis 6 Follow - up Regression, with Anger R egressed on an Interpersonal Justice x Locus of Causality I nteraction Anger Predictors variables SE B Step 1 .38* Interpersonal Justice .05 - .7 2 - .57* LoCRBS Supervisor .05 .20 .15* Step 2 .03* Interpersonal Justice .05 - .65 - .52* LoCRBS Supervisor .05 .1 6 .1 2 * IJ x LoCRBS Supervisor .06 - .2 6 - .1 8 * Note. * p <.05. N = 389 . IJ=Interpersonal Justice; LoCRBS=locus of ca usality for RBS. Figure 15 Supervisor - attributed Locus of Causality for RBS x Interpersonal Justice I nteraction P redicting Anger Note. N = 389. Figure 15 shows regardless the level of supervisor - attributed locus of causality for RBS , interpersonal justice exhibited a negative relationship with anger. However, 0 0.5 1 1.5 2 2.5 Low Interpersonal Justice High Interpersonal Justice Anger Low supervisor- attributed locus of causality for RBS High supervisor- attributed locus of causality for RBS 106 when supervisor - attributed locus of causality for RBS was high, the negative relationship between interpersonal justice and anger was much s teeper than when supervisor - attributed locus of causality for RBS was low. Consequently, the angriest respondents were those with high supervisor - attributions and low perceived interpersonal justice, and the least angry respondents were those with high supervisor - attributions and high perceived interpersonal justice. Ancillary a nalyses: Comparing E motions and PANA models The current study focuses on attribution s assumed to distinguish different forms of positive emotions (i.e. pride vs. gratitude) and negative emotions (i.e., anger vs. guilt/shame). However, the PANA model of affect would group these emotions into two general categories based on their valence (positive and negative affect), ignoring any related attributions. To this point , strong positive relationships were observed between PA and positive emotions (i.e. pride , r =.6 3 , p <.05 ; gratitude . r =. 62, p <.05 ) , and NA and negative emotions (i.e., anger, r =.74 , p <.05 ; guilt , r =.7 3, p <.05 ; shame , r =.7 7, p <.05 ) . PA and NA showed a small negative relationship ( r = - .28 ). Weak negative relationships between PA and NA have been reported in prior studies w hen participants are asked to think about affect ac ross the past few weeks or the past year ( e.g. Watson, Clark, & Tellgen, 1988). A negative relationship was also observed between PA and each negative emotion (i.e. guilt, r = - .2 5 ; shame, r = - . 28 ; anger , r = - . 33 ), and NA and each positive emotion (i.e., pride, r = - .6 2 ; gratitude, r = - .47). Three CFA models were run to test the factor structure of the PA NAS and emotion items. First, positively - valenced ( i.e., pride, gratitude, and PA) and negatively - valenced items ( i.e., guilt, sham e, anger, and NA ) were loaded on separate factors which were allowed to correlate. This model was poor fitting, 2 (1033) = 6162. 30 , CFI = .70, RMSEA = .11, SRMR = .10. A seven - factor model was then run with all emotions, PA, and NA loaded on their own respective 107 but correlated factors. This produced a better fitting model, 2 (20) = 2454.248, p <.05, CFI = .84, RMSEA = .08, SRMR = .07. A six factor model which loaded guilt and shame items on the sam e factor showed significantly more misfit than the seven factor model, 2 (6) = 204.93, p <.05, CFI = .83, RMSEA = .08, SRMR = .07 Thus, the items were best explained by the seven factor model that reflected their construct dimensionality. In addition t measures of emotions, I also explored whether study hypotheses would yield similar results if positive and negative affect were used instead as outcomes. Table 11 shows that similar to hypothesis 1, LMXSC was significantly correlated with positive affect ( r =.5 3 , p <.05) and negative affect ( r = - .1 5 , p <.05). Tests of hypothesis 2 and 3 using PANA outcomes are shown in Table 1 7 . As can be seen in the sec ond step of regressions in both t ables, the interaction terms used for hypotheses 2, 3, and 5 were nonsignificant when positive and negative affect were used as outcomes. 108 Table 1 7 Ancillary tests of Hypothes e s 2 and 3 with Positive Affect and Negative Affect as DVs Self - attributed Locus of Causality for RBS Positive Affect Negative Affect Predictors variables SE B SE B Step 1 .3 3 * .0 8 * Locus of causality for RBS .06 .3 3 . 2 7 * .05 - .25 - .3 0 * LMXSC .03 .2 6 .3 8 * .03 .01 .0 2 Step 2 <.01 .01 Locus of causality for RBS .07 .3 3 .2 7 * .06 - .20 - .24* LMXSC .03 .2 6 .3 8 * .03 <.01 .01 LoC RBS x LMXSC .04 .01 .01 .03 .05 .11 Supervisor - attributed Locus of Causality for RBS Positive Affect Negative Affect SE B SE B Step 1 .2 8 * .06* Locus of causality for RBS .06 .0 7 .0 5 .0 5 .17 .1 9 * LMXSC .03 .3 6 .5 4 * .02 - .0 6 - .1 2 * Step 2 <.01 <.01 Locus of causality for RBS .06 .0 8 .0 6 .0 5 .16 .18* LMXSC .03 .3 5 . 52 * .02 - .0 5 - .1 1 * LoC RBS x LMXSC .04 .0 6 .0 7 .03 - .02 - .04 Note. * p <.05. N = 388 for all regressions shown here. LoCRBS=locus of causality for RBS. Table 18 Ancillary tests of Hypothesis 5 with Positive Affect and Negative Affect as DVs Self - attributed Locus of Causality for RBS Positive Affect Negative Affect Predictors variables SE B SE B Step 1 .3 6 * .4 4 * Interpersonal Justice .05 .37 .30* .03 - .5 8 - .6 8 * LMXSC .03 . 30 .4 4 * .02 .0 3 .0 6 Step 2 <.01 <.01 Interpersonal Justice .06 .40 .32* .04 - .5 8 - .6 8 * LMXSC .03 . 30 .4 4 * .02 .0 3 .0 6 IJ x LMXSC .03 .04 .05 .02 .01 <.01 Note. * p <.05. N = 388 for regressions shown here. IJ=Interpersonal Justice; LoCRBS=locus of causality for RBS. 109 The expectations for hypothesis 4 were also supported in Table 1 1 . Interpersonal justice showed a moderate ly size d positive relationship with positive affect ( r =.43, p <.05), and a negative relationship with negative affect ( r = - .66, p < .05). As can be seen in Table 18 , a significant interpersonal justice - LMXSC interaction did not emerge when positive and negative affect were entered as the dependent variables. In Table 19 , hypothesis 6 was conducted using negative affect as the depen dent variable. The three - way interaction was non - significant (as were all two - way interactions), indicating hypothesis 6 was not supported when NA was used in place of anger. Table 19 Ancillary test of Hypothesis 6 with Negative Affect as DV Negative A ffect Predictors variables SE B Step 1 .45* Interpersonal Justice .03 - .56 - .67* LMXSC .02 .03 .07 LoCRBS Supervisor .04 .07 .08* Step 2 <.01 Interpersonal Justice .04 - .57 - .67* LMXSC .02 .02 .05 LoCRBS Supervisor .04 .08 .08* LMXSC x IJ .03 > - .01 - .01 LMXSC x LoCRBS Supervisor .03 .04 .06 IJ x LoCRBS Supervisor .05 - .03 - .03 Step 3 <.01 Interpersonal Justice .04 - .58 - .68* LMXSC .02 .02 .04 LoCRBS Supervisor .04 .10 .11* LMXSC x IJ .03 .02 .05 LMXSC x LoCRBS Supervisor .03 .04 .07 IJ x LoCRBS Supervisor .05 - .08 - .08 IJ x LMXSC x LoCRBS Supervisor .03 - .06 - .13 Note. * p <.05. N = 388 . IJ=Interpersonal Justice; LoCRBS=locus of causality for RBS. 110 Overall, similar to Hypothes e s 1 and 4, LMXSC and interpersonal justice were significantly associated with PA or NA . H owever, none of the hypothesized interactions explained incremental variance when PA or NA were entered as outcomes. A ncillary a nalyses: LMXSC and its relationship with LMX In this section, the p sychometric properties of LMXSC are examined and compared to findings from the two existing published studies on LMXSC (i.e., Erdogan, 2002; Vidyarthi et al. , 2010 ) . Then, the relationship between LMXSC and LMX is examined through a n umber of analyses. Looking first at the descriptive statistics , internal consistencies and intercorrelations , in the current study the average level of LMXSC was close to the mid - point of the 7 - point Likert scale ( M =4.12) and showed a wide standard deviat ion ( SD =1.35). Vidyarthi et al (2010) and Erdogan (2002) similarly found average responses were close to the midpoint and exhibited wide standard deviation s on a 5 - point Likert scale (Erdogan, 2002; undergraduate sample, M =3.95, SD =1.19; field sample, M =3. 02, SD =1.10; Vidyarthi et al., 2010; M =3.46, SD =.92). The LMXSC scale showed higher internal consistency in the current study ( =.92) than has been observed previously (Erdogan, 2002; undergraduate sample, =.86; field sample, =.84; Vidyarthi et al., 2010 ; =.86). As described previously, LMXSC showed the expected associations with emotions, PA, and NA. LMXSC was unrelated to work experience, current position tenure, organizational tenure, or age. O nly the length visor exhibited a significant (albeit weak) relationship with LMXSC ( r =.14, p <.05). In contrast, Vidyarthi et al. (2010) observed significant positive relationships of LMXSC with organizational tenure ( r =.21, p <.05) and with age ( r =.19, p <.05). 111 Looking next to the relationship between LMXSC and LMX , a positive correlation was expected as the two constructs overlap conceptually. The LMXSC - LMX relationship observed here was positive and moderate in size ( r =.58, p <.05 ). Vidyarthi et al. (2010) used the same LMX measure as the current study (i.e., LMX - 7, Graen & Uhl - Bein, 1995) and observed a similarly - sized relationship ( r =.55, p <.05). Erodgan (2002) used a different measure (LMX - MDM, Liden & Maslyn, 1998) and observed a slightly weaker relationship ( Erdogan, 2002; field sample, r =.38, p <.05 ). In an effort to parallel validation efforts described in Vidyarthi et al (2010), a principal components analysis with an oblique rotation was conducted with 1) LMXSC, 2) LMX, and 3) interactional justice ite ms . It is important to note the PCA described in Vidyarthi et al. (2010) used different measures than the current study to assess LMX (Liden & Maslyn, 1998, as compared to Graen & Uhl - Bien, 1995 here ), and interactional justice (Niehoff & Moorman, 1993, as compared to Colquitt, 2001 here ). 7 The results of th is PCA using data from the current study can be found in Table 2 0 . Listwise deletion was used, as the matrix was not positive - definite when pairwise deletion was applied. Items loaded onto factors with t heir respective scales, yielding a four - factor solution. The components exhibited high loadings (in order) from LMXSC items (. 62 or higher ; 45.66% variance explained ) , LMX items (. 68 or higher ; 14.84% variance explained ) , informational justice items (.38 or higher ; 6.20% variance explained ) and interpersonal justice items (.78 or higher ; 4.83% variance explained ). There were no cross - loadings over .30. C ollectively, these components explained 71.52 % of the variance. These results were similar to the valida tion efforts Vidyarthi et al (2010) described, where a three - 7 Vidyarthi et al. (2010) describe a PCA and CFA using data from Erodgan (2002). These analyses were not included in the original Erdogan (2002) dissertation document, and presumably w ere conducted separately using the findings to the statistics reported in Vidyarthi et al. (2010). 112 factor solution with LMXSC, LMX, and interactional justice items (the scale did not differentiate between interpersonal and informational) loaded on separate factors , explaining 71.23% of the var iance and show ing no cross - loadings over .20. Table 2 0 Promax Rotation Principal Components Analysis Structure for Main Study LMXSC, LMX, and Interactional Justice Items Component 1 2 3 4 LMXSC 1 .80 LMXSC 2 .62 LMXSC 3 .91 LMXSC 4 .87 LMXSC 5 .95 LMXSC 6 .91 LMX 1 .68 LMX 2 .76 LMX 3 .76 LMX 4 .89 LMX 5 .83 LMX 6 .80 LMX 7 .98 Interpersonal Justice 1 .84 Interpersonal Justice 2 .82 Interpersonal Justice 3 .78 Interpersonal Justice 4 .92 Informational Justice 1 .38 Informational Justice 2 .98 Informational Justice 3 .78 Informational Justice 4 .90 Informational Justice 5 .66 Eigenvalue 10.05 3.26 1.36 1.06 Variance explained 45.66% 14.84% 6.20% 4.83% Note. Loadings under .30 are suppressed. Listwise deletion was applied, N = 379. 113 Table 2 1 Component Correlation Matrix Based on Promax Rotated Principal Components Analysis for Main Study LMXSC, LMX, and Interactional Justice Items Components 1 2 3 4 1 - 2 .56 - 3 .26 .58 - 4 .31 .57 .63 - Note. Listwise deletion was applied, N =379. As can be seen in the component correlation matrix displayed in Table 21 , the LMXSC and LMX factors were moderately related (components 1 and 2, r =.56). LMXSC also showed weak relationships with the informational and interpersonal justice components (component 3, r =.26; component 4, r =.31). Referring back to Table 1 1 , LMXSC showed weak positive relationships with interpersonal justice and informati onal justice ( r =.31, p <.05; r =.29, p <.05). Erdogan (2002) observed a similar sized LMXSC - interactional justice relationship (undergraduate sample, r =.31, p <.05; field sample, r =.32, p <.05) using a single interactional s (1993) scale. As in parallel to the validation efforts described in Vidyarthi et al. (2010) , a CFA was conducted to further test the LMXSC and LMX factor structure . In the analys is described in Vidyarthi et al. (2010) , LMXSC and LMX fit well on two separate factors in a confirmatory factor analysis , 2 (131)= 313.12, CFI=.93, RMSEA=.08, and an alternative model specifying a 2 (1) =48.53, p <. 01. , 2 (66)= 393.95, CFI=.93, RMSEA=.11. It should be noted that LMX item 7 displayed a negative residual variance in this model (i.e. a Heywood case); to ensure the resid ual covariance matrix 114 was positive - definite, the LMX item 7 residual was constrained to 0. When LMXSC and LMX were allowed to correlate, the model exhibited improved fit , 2 (2)= 160.82, p <.01; CFI=.96, RMSEA=.08. As a final comparison between LMXSC an d LMX, I ran hypothesis tests with LMX ente red in place of LMXSC . Hypothesis 1 suggested that LMXSC would be positively related to positive emotions (pride and gratitude) and negatively related to negative emotions (anger, guilt, and sh ame). As can be seen in Table 1 1 , LMX also showed positive relationships with positive emotions (i.e. pride, r =.68, p <.05; gratitude, r =.61, p <.05), and negative relationships with negative emotions (i.e. guilt, r = - .4 2 , p <.05; shame, r = - . 43 , p <.05; anger, r = - .4 8 , p <.05). Tests of hypothesis 2 and 3 using LMX in place of LMXSC can be found in Table 2 2 . As can be seen in the second step of Table 2 2 nonsignificant when LMX was used. An LMX - self - attributed lo cus of causality for RBS interaction did, however, significantly predict shame ( step 2, =.1 2 , p <.05), and explained significant variance beyond the model main effects ( =.0 1 , p <.05). This interaction is d isplayed graphically in Figure 16 . 115 Table 2 2 Ancillary tests of Hypothesis 2 and 3 with LMX in place of LMXSC Self - attributed Locus of Causality for RBS Pride Guilt Shame Predictors variables SE B SE B SE B Step 1 .48* .18* .20* Locus of causality for RBS .06 .21 .16* .06 - .08 - .08 .05 - .15 - .15* LMX .05 .71 .60* .05 - .36 - .38* .05 - .32 - .35* Step 2 .01 <.01 .01* Locus of causality for RBS .06 .17 .13* .06 - .05 - .05 .06 - .10 - .10 LMX .05 .70 .59* .05 - .35 - .37* .05 - .30 - .33* LoC for RBS x LMX .05 - .08 - .07 .06 .06 .06 .05 .10 .12* Supervisor - attributed Locus of Causality for RBS Gratitude Anger SE B SE B Step 1 .37* .26* Locus of causality for RBS .07 - .04 - .03 .06 .23 .17* LMX .06 .84 .60* .05 - .49 - .45* Step 2 .01* .04* Locus of causality for RBS .07 - .02 - .01 .06 .17 - .14* LMX .06 .79 .57* .05 - .41 - .37* LoC for RBS x LMX .08 .17 .10* .06 - .28 - .21* Note. * p <.05. N s ranged 388 - 389 . LoCRBS=locus of causality for RBS. 116 As can be seen in Figure 16 , a main effect of LMX was observed with low LMX respondents reporting higher levels of shame that those with high levels of LMX. For high LMX respondents, levels of shame decreased slightly as levels of self - attributed locus of causality for RBS increased. Similar to the H2c pattern observed with LMXSC, self - attributed locus of causality for RBS had a stron ger negative main effect on shame on those with high levels of LMX. As such, interaction pattern was similar to that which emerged when LMXSC was entered; however, it did not reflect the interaction as put forth in the hypothesis. Figure 16 LMX x Locus of C ausality for RBS Interaction Predicting Shame Note . N = 389. An LMX - supervisor - attributed locus of causality for RBS interaction significantly predicted gratitude (Table 2 2 , step 2, =.10, p <.05) and explained significant variance beyond the model main effects ( =.01, p <.05). A graph of this inter action is displayed in Figure 17 . LMX exhibited a strong main effect, with high LMX respondents reporting higher levels of 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 Low LMX High LMX Shame Low self-attributed locus of causality for RBS High self-attributed locus of causality for RBS 117 gratitude than those with low LMX. Supervisor - attributed locus of causality showed a slight positive relationship with gratitude for those with high LMX and a slight negative relationship with gratitude for those with low LMX. In general, the LMX - supervisor - attributed locus of ns and the pattern that emerged for (see Figure 12 ) . Figure 17 LMX x Locus of C ausality for RBS Interaction P redicting Gratitude Note . N =389. LMX also interacted with supervisor - attributed locus of causality in predicting anger above and beyond main effects (Table 2 2 , step 2, = - .21, p <.05; =.04, p <.05). This interaction is di splayed graphically in Figure 1 8 . As can be seen in Figure 10 , LMX had a main effect with low LMX respondents reporting higher levels of anger than those with high levels of LMX. For high LMX respondents, supervisor - attributed locus of causality for RBS had a very weak negative association with anger. For those with low levels of LMX, supervisor - attributed 2 2.5 3 3.5 4 4.5 Low LMX High LMX Gratitude Low supervisor- attributed locus of causality for RBS High supervisor- attributed locus of causality for RBS 118 locus of control for RBS showed a strong positive association with anger. This pattern reflected from one difference: at low levels of supervisor - attributed locus of causality for RBS, LMX displayed a main effect and LMXSC did not ( see Figure 13 ). Figure 1 8 LMX x Locus of Causality for RBS Interaction Predicting Anger Note . N =389. When LMX r eplaced LMXSC in tests of H5, an LMX - interpersonal justice interaction term did not significantly predict gratitude above and be yond the main effects. It did, however, significantly predict anger ( Table 23, step 2, =.2 0 , p <.05), and explained significant variance beyond the mo del main effects ( =.0 3 , p <.05). This interaction is di splayed graphically in Figure 1 9 . 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 Low LMX High LMX Anger Low supervisor- attributed locus of causality for RBS High supervisor- attributed locus of causality for RBS 119 Table 2 3 Ancillary tests of Hypothesis 5 with LMX in place of LMXSC Self - attributed Locus of Causality for RBS Gratitude Anger Predictors variables SE B SE B Step 1 .48* .39* Interpersonal Justice .07 .63 .40* .06 - .61 - .49* LMX .06 .53 .38* .05 - .22 - .20* Step 2 <.01 .03* Interpersonal Justice .07 .58 .37* .06 - .49 - .40* LMX .06 .52 .37* .05 - .19 - .18* IJ x LMX .07 - .10 - .06 .06 .24 .20* Note. * p <.05. N = 389 for regressions shown here. IJ=Interpersonal Justice. In the LMX - interpersonal justice interaction, LMX had a main effect on anger when interpersonal justice was low. That is, when interpersonal justice was low, those with low LMX reported higher anger than respondents with high levels of LMX. When interpersonal justice was high, anger was lowest, regardl ess the level of reported LMX. As a result, interpersonal justice displayed a w eaker negative association with anger for high LMX respondents and a stronger negative association with anger for low LMX respondents. This pattern was similar to the interaction expected in H5b, but H5b anticipated a main effect of LMXSC when interpersona l justice was low, and here no main effect was found. Still, the pattern more closely represented H5b expectations than the LMXSC findings, which showed low LMXSC - high interpersonal having the lowest levels of anger (see Figure 14 ). 120 Figure 1 9 LMX x Interpersonal Justice Interaction P redicting Anger Note. N = 389. As can be seen in Table 2 4 , when H6 was tested with LMX instead of LMXSC, a three - way interaction between LMX , interpersonal justice , and supervisor - attributed locus of causality for RBS did not significantly predict levels of anger. 0 0.5 1 1.5 2 2.5 Low LMX High LMX Anger Low Interpersonal Justice High Interpersonal Justice 121 Table 2 4 Ancillary test of Hypothesis 6 with LMX in place of LMXSC Anger Predictors variables SE B Step 1 .41* Interpersonal Justice .06 - .59 - .47* LMX .05 - .20 - .18 LoCRBS Supervisor .05 .17 .13 Step 2 .04* Interpersonal Justice .06 - .50 - .40* LMX .05 - .14 - .13* LoCRBS Supervisor .05 .12 .09* LMX x IJ .06 .14 .12* LMX x LoCRBS Supervisor .07 - .16 - .12* IJ x LoCRBS Supervisor .08 - .06 - .05 Step 3 <.01 Interpersonal Justice .07 - .48 - .37* LMX .05 - .14 - .13* LoCRBS Supervisor .06 .07 .06 LMX x IJ .07 .11 .09 LMX x LoCRBS Supervisor .07 - .16 - .12* IJ x LoCRBS Supervisor .09 .02 .01 IJ x LMX x LoCRBS Supervisor .07 .12 .12 Note. * p <.05. N = 389 . IJ=Interpersonal Justice; LoCRBS=locus of causality for RBS. In these supplemental analyses , LMXSC showed psychometric properties that were relatively similar to those observed in Vidyarthi et al. (2010) and Erodogan (2002). The curre nt hypothesis models in place of LMXSC, results were similar but not identical to the findings observed with LMXSC in the model (e.g. no significant LMX - locus of causa lity interaction when pride was the DV). Ancillary a nalyses: Comparing Informational Justice and Interpersonal Justice Because informational and interpersonal are highly related, additional analyses were run to see whether hypotheses held when informatio nal justice was entered in H4, H5, and H6. 122 I nformational justice exhibited significant positive relationships with positive emotions ( Table 1 1 ; pride, r =.6 8 , p <.05; gratitude, r =.6 3 , p <.05) and negative relationships with negative emotions ( Table 1 1 ; guilt, r = - . 50 , p <.05; shame, r = - .5 2 , p <.05; anger, r = - .5 2, p <.05) , indicating H4 was supported when informational justice was used in place of interpersonal justice. Table 2 5 shows the results of H5 when informational justice was used. There was no signifi cant LMXSC - informational justice interaction when gratitude was used as the dependent variable. A significant interaction did appear when anger was entered as the model outcome ( = .1 6 , p <.05 ; ² =.0 2 , p <.05) . This interaction i s shown graphically in Figure 20 . Table 2 5 Ancillary T est s of Hypothesis 5 using Informational Justice in place of Interpersonal J ustice Self - attributed Locus of Causality for RBS Gratitude Anger Predictors variables SE B SE B Step 1 .45* .30* Informational Justice .06 .82 .55* .05 - .62 - .52* LMXSC .03 .22 .26* .03 - .05 - .07 Step 2 <.01 .02* Informational Justice .06 .80 .53* .05 - .56 - .48* LMXSC .03 .22 .26* .03 - .04 - .06 Inf J x LMXSC .04 - .06 - .07 .03 .12 .16* Note. * p <.05. N =388. Inf J= Informational Justice; LoCRBS=locus of causality for RBS. Th e interaction shown in Figure 20 shows a similar pattern t for an LMXSC - for interpersonal justice, informational justice had a negative main effect on anger, regardless the level of LMXSC. Yet the resu lts also informational justice was so dramatic on low LMXSC respondents that low LMXSC - high informational justice individuals showed the lowest level of anger. This steep main effect of 123 informati onal justice on anger for low LMXSC individuals was also observed when interpersonal justice was used (see Figure 14 ). Figure 20 LMXSC x Informational Justice Interaction Predicting Anger Note. N =389. Table 26 shows the results of H6 when informational justice was used in the model. As can be seen in step 3, no three - way interaction emerged between LMXSC, supervisor - attributed locus of causality, and informational justice. 0 0.5 1 1.5 2 2.5 Low LMXSC High LMXSC Anger Low Informational Justice High Informational Justice 124 Tabl e 2 6 Ancillary T est of H ypothesis 6 with I nformational J ustice in place of I nterpersonal J ustice Anger Predictors variables SE B Step 1 .33* Informational Justice .05 - .59 - .50* LMXSC .03 - .03 - .05 LoCRBS Supervisor .06 .24 .18* Step 2 .03* Informational Justice .05 - .52 - .44* LMXSC .03 - .01 - .02 LoCRBS Supervisor .06 .18 .14* LMXSC x InfJ .04 .05 .07 LMXSC x LoCRBS Supervisor .04 - .03 - .04 InfJ x LoCRBS Supervisor .07 - .20 - .14* Step 3 <.01 Informational Justice .05 - .51 - .43* LMXSC .03 - .01 - .02 LoCRBS Supervisor .06 .16 .12* LMXSC x InfJ .04 .03 .04 LMXSC x LoCRBS Supervisor .03 - .03 - .04 InfJ x LoCRBS Supervisor - .17 - .17 - .12* InfJ x LMXSC x LoCRBS Supervisor .05 .05 .07 Note. * p <.05. N = 388 . Inf J= Informational Justice; LoCRBS=locus of causality for RBS. Table 2 7 Informational Justice x Supervisor - attributed Locus of Causality for RBS predicting Anger Anger Predictors variables SE B Step 1 .32* Interpersonal Justice .05 - .60 - .51* LoCRBS Supervisor .06 .25 .19* Step 2 .03* Interpersonal Justice .05 - .54 - .45* LoCRBS Supervisor .06 .21 .16* IJ x LoCRBS Supervisor .06 - .25 - .18* Note. * p <.05. N =389. Inf J= Informational Justice; LoCRBS=locus of causality for RBS. 125 Similar to the results for interpersonal justice, informational justice showed a significant two - way interaction with supervisor - attributed locus of causality for RBS as part of Hypothesis 6 . The results of this interaction without LMXSC included in the mo del can be found in Table 2 7 and are di spl a yed graphically in Figure 21 . Figure 21 LMXSC x Informational Justice Interaction Predicting Anger Note. * p <.05. N =389. To summarize, informational justice shows patterns similar to that of interpersonal justice: informational justice had significant relationships with focal emotions, and showed the same general patterns for H5 and H6. Ancillary a nalyses: Race/Ethnicity Study measures did not differ significantly by race/ethnicity group . The descriptive statistics and tests of mean differences can be found in Appendix R . Due to an error in the survey, data on gender was not collected. 0 0.5 1 1.5 2 2.5 Low Informational Justice High Informational Justice Anger Low supervisor- attributed locus of causality for RBS High supervisor- attributed locus of causality for RBS 126 Ancillary a nalyses: Workgroup size relative standing among peers working for the same supervisor, analyses were conducted to explore whether workgroup size (i.e. the number of peers report ing to the same sup ervisor) had an influence on study findings. Respondents were asked to provide their workgroup size on a continuous interval scale; mean differences across sc ale scores (see Appendix S ), responses were placed into five groupings : those with 1 - 3 peers, 4 - 10 peers, 10 - 20 peers, 20 - 60 peers, and 60 or more peers. Only two variables showed significant mean differences across these groups: guilt and shame (respectiv ely, F (4,383)=3.30, p =.01; F (4,384)=2.49, p =.04). Both guilt and shame tended to increase as workgroup size increased , with the exception of the 20 - 60 peer group, which showed slightly lower means). C ontinuous data was not available for the 17 respondents reporting 60 or more peers . Data from these extremely large workgroups were not assumed to provide additional s ubstantive insights (i.e. knowledge of and social comparisons with specific peers would become increasingly diffuse at numbers highe r than 60). Further, extremely large workgroup numbers could be considered outliers that introduce noise to analyses. Based on this reasoning , the 17 respondents with 60 or more peers were excluded from the analyses which follow. Workgroup size did not significantly s with pri de, guilt, gratitude, or anger, nor did it significantly moderate interp 127 gratitude or anger. A significant interaction did emerge between LMXSC and shame. These find ings are reported in Table 28 and shown graphically in Figure 22 . Table 28 Regression with LMXSC x Number of Peers Interaction Predicting Shame Shame Predictors variables SE B Step 1 .03* Number of Peers < .0 1 > - .01 . 05 * LMXSC .03 - . 10 - . 19 * Step 2 .02* Number of Peers <.01 > - .01 - . 05 * LMXSC .0 3 - .1 0 - . 1 9 * Peers x LMXSC <.01 .01 .1 1 * Note. * p <.05. N =3 74. Figure 22 LMXSC X Number of Peers Interaction Predicting Shame Note. N =3 74. 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 Low LMXSC High LMXSC Shame Low Number of Peers High Number of Peers 128 As Figure 22 indicates, low LMXSC respondents reported high levels of shame which decreased as the workgroup size increased. For high LMXSC, shame was lower when the workgroup size was smaller; in larger groups, high LMXSC respondents reported slightly higher levels of shame. W orkgroup size was also tested as a moderator of the H ypothesis 2, 3, and 5 interactions ; results ind icated workgroup size did not significantly moderate any of these interactions. Ancillary a nalyses: Length of Relationship with O Of the tenure - related variables measured in this study, the length of the relationship held supervisor (i.e. supervisor relationship tenure) was most closely related to study variables and most conceptually related to the theoretical framework (i.e. , LMX relationships are assumed to change over time). For these reasons, analyses were conducted t o see how findings might differ based on supervisor relationship tenure. Participants provided the number of years and months they had worked for their current supervisor. Responses were grouped into five categories: 0 - 6 months, 6 - 12 months, 1 - 2 years, 2 - 5 years, and 5 or more years. Descriptive statistics of study measure by these groupings can be found in Appendix T . There were significant mean differences across groupi ngs for the following variables: LMXSC, LMX, interpersonal justice, and LMX - related gratitude. For both LMXSC and LMX, means tended to steadily increase as the relationship length increased . Interpersonal justice tended to be highest among those who worked with their supervisor for 6 months or less ( M =4.55) . Those in the 6 - 12 month group showed the lowest levels of interpersonal justice and a wider standard deviation ( M =4.08, SD =1.02). Perceived interpersonal justice steadily increased at higher levels of s upervisor relationship tenure. Gratitude was generally stable across groupings, but those who had worked with their supervisor for 1 - 2 years showed notably lower 129 levels of gratitude ( M =3.04). Those who report ed working with their supervisor for 5 or more years showed on average the highest level of gratitude ( M =3.68). Next, supervisor relationship tenure was examined as a moderator of the associations moderate LM The length of relationship ide. These results are shown in Table 29 and displayed graphically in Figure 23 . Table 29 Regression with LMXSC x Supervisor Relationship Tenure Interaction Predicting Pride Pride Predictors variables SE B Step 1 .22* Supervisor Relationship Tenure .02 > - .01 - .01 LMXSC .04 .34 .47* Step 2 .01* Supervisor Relationship Tenure .02 .01 .03 LMXSC .04 .35 .47* Supervisor Relationship Tenure x LMXSC .01 - .03 - .12* Note. * p <.05. N =3 60 . 130 Figure 23 LMXSC X Supervisor Relationship Tenure Predicting Pride Note. N =360. Figure 23 shows that although LMXSC had the expected main effect on pride, when respondents had worked with their supervisor for longer periods of time, the effect of LMXSC was weakened (i.e. scores increased for low LMXSC respondents and decreased for high LMXSC respondents). Analyses were also conducted to determine whether s upervisor relationship tenure was a potential moderator of H ypothes e s 2, 3, and 4. There were no significant three - way interactions between locus of causality, LMXSC, and supervisor relationship tenure when pride, guilt, gratitude, or anger were entered as the dependent variable. A three - way interaction did emerge for shame. The results are listed in Table 3 0 , and displayed graphically in Figure 24 . 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 Low LMXSC High LMXSC Pride Low supervisor relationship tenure High supervisor relationship tenure 131 Table 3 0 Regression with LMXSC x Self - attributed Locus of Causality for RBS x Supervisor Relationship Tenure Interaction Predicting Shame Shame Predictor variables SE B Step 1 .12* LMXSC .04 .01 .02 LoCRBS Self .06 - .34 - .35* Supervisor Relationship Tenure .01 .01 .04 Step 2 .02* LMXSC .03 > - .01 <.01 LoCRBS Self .07 - .24 - .24* Supervisor Relationship Tenure .01 .01 .04 Sup Ten x LoCRBS Supervisor .02 .01 .02 LMXSC x LoCRBS Supervisor .03 .10 .18* LMXSC x Sup Ten .01 > - .01 > - .01 Step 3 .04* LMXSC .03 <.01 .01 LoCRBS Self .07 - .26 - .26* Supervisor Relationship Tenure .02 - .01 - .03 Sup Ten x LoCRBS Supervisor .02 .01 .04 LMXSC x LoCRBS Supervisor .03 .10 .18* LMXSC x Sup Ten .01 > - .01 - .01 LMXSC x Sup Ten x LoCRBS Supervisor .01 .02 .12* Note. * p <.05. N =361. Sup Ten= Supervisor Relationship Tenure . 132 Figure 24 LMXSC x Self - attributed Locus of Causality for RBS x Supervisor Relationship Tenure Interaction Predicting Shame Note. N =361. In this interaction, those with low self - attributed locus of causality for RBS tended to show higher levels of shame than those with high self - attributed locus of causality for RBS. LMXSC showed unexpected interactions with relationship tenure within high and low levels of self - attributed locus of causality for RBS. When self - attributed locus of causality for RBS was high, shame decreased among low LMXSC respondents as the relationship length increased. High LMXSC respondents, however, showed increased sham e as the relationship length increased. When self - attributed locus of causality for RBS was low, low LMXSC individuals displayed increasing levels of shame as the relationship length increased. High LMXSC respondents showed decreasing levels of shame as t he relationship length increased. 133 A significant interaction also emerged between supervisor relationship tenure, interpersonal justice, and LMXSC in predicting anger. No significant interaction emerged when gratitude was the DV. This finding is reported in Table 31 and the interaction is displayed graphically in Figure 25 . Table 3 1 Regression with LMXSC x Interpersonal Justice x Supervisor Relationship Tenure Interaction Predicting Anger Anger Predictors variables SE B Step 1 .36* LMXSC .03 - .03 - .04 Interpersonal Justice .06 - .72 - .58* Supervisor Relationship Tenure .01 .01 .04 Step 2 .03* LMXSC .03 - .03 - .04 Interpersonal Justice .06 - .64 - .52* Supervisor Relationship Tenure .01 .02 .06 LMXSC x Int. J .04 .13 .18* Sup Ten x Int. J .02 - .03 - .07 LMXSC x Sup Ten .01 > - .01 - .01 Step 3 .01* LMXSC .03 - .03 - .04 Interpersonal Justice .06 - .61 - .49* Supervisor Relationship Tenure .01 .02 .07 LMXSC x Int. J .04 .14 .18* Sup Ten x Int. J .02 - .02 - .05 LMXSC x Sup Ten .01 > - .01 - .01 LMXSC x Sup Ten x Int J. .01 - .03 - .10* Note. * p <.05. N =361. Sup Ten= Supervisor Relationship Tenure . 134 Figure 25 LMXSC x Interpersonal Justice x Supervisor Relationship Tenure Interaction Predicting Anger Note. N =361. As shown in Figure 25 , interpersonal justice had main effect l evel on anger, wit h individuals reporting high interpersonal justice reporting notably lower levels of anger than those reporting low interpersonal justice. The moderation of supervisor relationship tenure and LMXSC, though, differed by levels of interpersonal justice. When interpersonal justice was perceived to be high, LMXSC had a strong effect on the leve l of reported anger among respondents who worked for their supervisor for a relat ively shorter length of time. As the length of the relationship with the supervisor increased, the effect of LMXSC on anger weakened with high LMXSC respondents expressing slightly more anger and low LMXSC respondents expressing slightly less anger. Whe n interpersonal justice i s perceived to be low, the moderation showed a different pattern. When respondents who had worked for their supervisor for a relatively shorter period of 135 time, LMXSC again showed a main effect on levels anger. Given the main effect of interpersonal justice, these levels of anger were higher that when interpersonal justice was high. As the length of the relationship with the supervisor increased, the main effect of LMXSC appeared to weaken among high LMXSC respondents. Low LMXSC resp ondents did not follow this pattern: for these individuals, a longer relationship with the supervisor resulted in the highest level of anger of all groups. This indicates the effect of LMXSC tends to weaken the longer supervisors and subordinates work tog ether, expect when subordinates perceived they had relatively low quality LMXSC within the workgroup and felt the supervisor has not been respectful. Ancillary a nalyses: LMXSC x interpersonal justice interaction predicting supervisor - attributed locus of causality for RBS - attributed locus of causality for RBS interact in predicting emotions, it could also be that LMXSC and interpersonal justice interact in predicting super visor - attributed locus of causality. That is, when individuals perceive their supervisor displays rude and inconsiderate treatment, the supervisor may be perceived as taking control over the LMX quality, leading to high levels of supervisor - attributed locu s of causality for RBS. This negative association between interpersonal justice and supervisor - attributed locus of causality for RBS may then be exacerbated by perceptions of low LMXSC. In support of this notion, Table 11 shows interpersonal justice and supervisor - attributed locus of causality demonstrated a significant negative correlation ( r =.20, p <.05). A hierarchical regression tested the interaction, and can be found in Table 3 2 . Interpersonal justice showed a negative main effect on supervisor attr ibutions in the first step ( = - .18, p <.05) which became non - significant when an LMXSC - interpersonal j ustice interaction was entered on the second step. The interaction was significant ( =.19, p <.05) and is displayed graphically in Figure 26 . As 136 per expecta tions, the interaction pattern showed supervisor - attributed locus of causality for RBS was especially high among low LMXSC respondents who perceived low interpersonal justice. These levels decreased as perceptions of interpersonal justice increased. Those with high LMXSC reported low supervisor - attributed locus of causality for RBS when interpersonal justice was low, and surprisingly, these levels of supervisor attributions increased as interpersonal justice increased. Although the exact psychological mec hanism driving this interaction pattern is unknown, the results indicate supervisor - attributed locus of causality may play a more Table 32 Regression with LMXS C x Interpersonal Justice Predicting Supervisor - attributed Locus of Causality for RBS Supervisor - Attributed Locus of Causality for RBS Predictors variables SE B Step 1 .05* Interpersonal Justice .05 - .17 - .18* LMXSC .03 - .04 - .08 Step 2 .03* Interpersonal Justice .05 - .09 - .10 LMXSC .03 - .04 - .07 IJ X LMXSC .03 .11 .19* Note. * p <.05. N =388. IJ= Interpersonal Justice. 137 Figure 26 LMXSC x I nterpersonal J ustice I nteraction predicting S upervisor - attributed L ocus of C ausality for RBS Note. N =388. Ancillary a nalyses: Hypothesized interactions with alternative emotions and locus of causality entered as controls I also explored whether the same findings emerged when alternative categories of emotions (self - conscious vs. externally - focused emotions) or alternative locus of causality for RBS attributions (self - vs. supervisor - attributed) were entered into regression models as controls. To test this, I first r an H 2 regression models with LMX - related exter nally focused emotions (i.e., gratitude and anger) entered as control variables , and ran H 3, H 5, and H 6 regression models with LMX - related self - conscious emotions (i.e., pride, guilt, and shame) e ntered as control variables . When self - conscious emotions we re entered into m odels predicting externally - focused emotions (i.e., H3, H5, and H6 models), results paralleled those that emerged from models without control variables . When externally - focused emotions were entered into the H2 models, the partially 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 Low LMXSC High LMXSC Supervisor Locus of Causality for RBS Low Interpersonal Justice High Interpersonal Justice 138 suppor ted H2a (pride) finding that emerged without controls did not appear . Similar to the findings without controls, H2b (guilt) and H2c (shame) were not supported when externally - focused emotions were entered. The three - way interaction predicting anger in H6 remained unsupported. Thus, the results of hypothesis tests generally remained similar with or without the alternative category of em otions entered as controls , but were not identical. I also ran regression models with alternative forms of locus of causa lity entere d as a control. When supervisor - attributed locus of causality was entered as a control in H2 models, the partial support for H2a (pride) was not observed in the results, whereas H2b (guilt) continued to be unsupported. H2c (shame) became partial ly supported when supervisor - attributed locus of causality was entered as a control, with the main effect of self - attributed locus of causality for RBS being stronger when LMXSC was low and weaken ed when LMXSC was high. When self - attributed locus of caus ality for RBS was entered as a control into H3, H3a bserved in the data. H6 continued to be unsupported. Overall, minor differences emerged in the results when hypothesi s models were conducted with and without alternative forms of locus of causality for RBS entered as controls. 139 DISCUSSION Supervisors often have numerous dyadic LMX relationships which vary in quality ( Dansereau et al. , 1975). As these differentiated relationships emerge , subordinates are likely to form social comparisons regarding how their LMX quality compares to the LMX quality held by peers (Vidyarthi et al., 2010; Erodogan, 2002). Although these social comparisons are assum ed existing research has not fully explored when and why L MXSC might be linked to LMX - related emotions . In this study, both attribution theory and research on the fair process effect were used to suggest at tributions regarding relationship building and interpersonal justice would play a role in the association between These expectations were tested using survey data from large sample of working adults. Based on th e findings that emerged , it appears that relationship building attributions and interpersonal justice do play a role in association with LMX - related emotions ; however, their hypotheses . Specifically, a ttributio ns moderated the effect of LMXSC on LMX - related pride, shame, gratitude, and anger , but n o such moderation occurred for guilt. Despite the emergence of these interactions, only with LMX - related gratitude did attribution s moderate the effect of LMXSC i n a manner that aligned with attribution theory ; with LMX - related pride and anger , moderation results only partially matched expectations. W ith LMX - related shame , the LMXSC - attribution interaction resulted in an entirely different pattern of results than was expected. Interpersonal justice e xhibited particularly strong association s with LMX - related emotions, but the interpersonal justice - LMXSC interaction hypothesized to predict levels of externally - focused LMX - related emotions did not appear when LMX - relate d gratitude was the outcome . Further, an interpersonal justice - LMXSC interaction only partially matched 140 expectations when LMX - related anger was the outcome . Interpersonal justice was also expected to act as a buffer in the LMXSC - locus of causality interact ion , with interpersonal justice buffering the effect of supervisor locus of causality attributions among low LMXSC individuals. No such moderation appeared . F indings for each hypothesis are discussed more fully in the sections t hat follow. Findings from hypothesis tests Hypothesis 1 tested the assumption that LMXSC is related to the emotions subordinates feel regarding their LMX . In the current study, LMXSC was positively associated with positive LMX - related emotions , and negatively associated with negative LMX - related emotions. However, the linkage between LMXSC and positive emotions was far stronger than the linkage between LMXSC and negative emotions, indicating this type of social comparison may play a more impor tant role in the emergence of LMX - related positive emotions than negative emotions. Indeed, when self - attributed locus of causality for RBS was entered in the same regression, LMXSC was not a significant predictor of negative self - conscious emotions. In th e context of LMX, it may be that downward social comparisons elicit stronger affective outcomes than upward social comparisons. In hypothesis 2, self - attributed l ocus of causality was expected to play a moderating role in the association between LMXSC an d self - conscious LMX - related emotions ( i.e., pride, guilt, and shame.) T he moderation aside, it is important to note that self - attributed locus of causality had stronger than expected relationships with many variables, including LMXSC . In fact, s elf - attrib uted locus of causality for RBS was more strongly associated with LMX - related emotions than the hypothesized LMXSC - emotion associations. Consequently, s elf - attributed locus of causality for RBS appeared to have an overwhelming strong association with LMX - related self - 141 conscious emotions that altered the moderations in ways that were not expected. As an example, with LMX - related pride , self - attributed locus of causality for RBS was expected to have a positive effect on those experiencing a positive outc ome (i.e., high LMXSC) , and was expected to worsen the effect of a negative outcome (i.e., lower LMX - related pride for those reporting low LMXSC ). I nstead, those with lower LMXSC had increas ed levels of pride as self - attributed locus of causality for RBS increased (although lower than those with high LMXSC). Similar findings were observed with shame. Whereas t hose with high LMXSC were expected to show low shame, those with low LMXSC were expected to feel increasing levels of shame as self - attributed lo cus of causality increased. The opposite result was observed amongst low LMXSC participants . I nstead of showing the highest expectations, those with low LMXSC and high self - attributed locus of causality for RBS showed the lowest level of shame. Although these results were not anticipated, the interaction pattern was consistent across two of the three self - conscious emotions ( guilt showed no significant interaction). S e veral possible reasons may explain why low LMXSC individuals may report more positive LMX - related emotions. First, attribution theory posits individuals look at their outcome and reflect on probable causes of th at outcome. This assumes causal attributions are a secondary thought process that accentuate s the positive effect of success and the negative effect of failure. Alternatively, self - attributed locus of causality for RBS may have a positive unique effect on experienced emotions above and beyond the effect of LMXSC. For example, individuals with low LXMSC and high self - attributed locus of causality for RBS may think, Despite my low status within the group, I do my best to take action within my LMX relationship and for that /not asha Similarly, those with low 142 LMXSC and low self - a relatively worse LMX quality , but I tried and nothing worked From this standpoint, self - attributed locus of causality for RBS could be negatively linked to shame and positively linked to pride in ways that are distinct The above example assume s self - attributed locus of causality for RBS is interpreted as the amount of effort an individual has put into their supervisory relationship. Yet these attributions could also have a distinct effect if they were interpreted as the extent to which one has agency within the LMX relationship . For example, those with low LMXSC and high self - attributed locus of causality may feel proud/unashamed of their LMX because despite their low status, they feel they have agency in terms of how their relationship is built. Those with low LMXSC and low self - attributed locus of causality for RBS may feel powerless within the relationship. The notion that perceived personal agency leads to positive outcomes for individuals is supported by many other streams of research in psychology , including the constructs that comprise core self - evaluations (Judge & Bono , 2001). show the least amount of pride and the highest amount of shame expectations (i.e., low LMXSC and high self - attributed locus of causality for RBS) are most likely to be defensive about their LMX quality. Ta n gney and Dearing (2003) noted that measures that ask respondents directly about their experiences of shame are likely to invite defensive denial response patterns . It may be th but low LMXSC - low self - attributed locus of causality for RBS individuals exhibit defensive denial and thus do not report negative self - conscious emotions. 143 Although shame and guilt where highly correlated, an LMXSC - locus of causality for RBS interaction did not appear for LMX - related guilt. This suggests the interaction between LMXSC and self - attributed locus of causality for RBS may predict LMX - related emotions that involve feelings regarding a stable aspect o f oneself . That is, p attributions made sense of self one is responsible for a socially valued outcome or for being a socially - ); shame focuses more exclusively on views of stable aspects of the self. Guilt, alternatively, focuses on (negative) Another consideration in the lack of findings for guilt is that LMX - related guilt items refer to non - specific behaviors which were va guer and required higher cognitive recall from respondent s . Specifically, g In the context of a standing exchange relationship, past actions that elicit guilt would likely be specific failed exchanges with the supervisor . In order to report feelings of guilt, the respondent must think back to specific unsuccessful interactions wherein they perceived they were at fault , and then report feeling bad or tension regarding that interaction . Whereas other emotion items ask ed respondents to report emotions that may be more easily accessed (e.g. feeling powerless, appreciative, or valuable), the guilt items used here required more cognitive recall . Hypothesis 3 suggested that supervisor - attribu ted locus of causality for RBS would - focused emotions (i.e. anger and gratitude). Whereas the strength of relationship s between study variables and self - attribut ed locus of causality for RBS was surprising , th e direction of relationships between super visor attributions and other variables was unexpected. As noted previously, supervisor - attributed locus of 144 causality for RBS in concept was not assumed to be perceived as negatively valenced or unfair . Supervisor - attributed locus of causality was thought to be both positive and negative externally - focused emotions (i.e. anger and gratitude). Yet supervisor - attributed locus of causality showed unexpected (albeit weak) positive associations with negative outcomes (e.g. shame), and negative associations with positive outcomes (e.g. interactional justice, LMXSC, LMX). In spite of these unexpected characteristics of supervisor - attributed locus of causality for RBS, the interactions between LMXSC and supervisor - attributed locus of causality for RBS showed patterns that were largely as expected. The only major difference between the observed results and the hypothesized interaction s was that anger was low when supervisor - attributed locus of causality for RBS was low. These findings indicate that LMXSC is more strongly linked with LMX - is assumed to be driven by the supervisor. ectations, those who perceived their supervisor to be treating them with respect and consideration were more likely to express increased levels of LMX - related gratitude and decreased levels of LMX - related anger. This finding fits with existing research li nking interpersonal justice to anger ( e.g. Rupp, McCance, Spencer, & Sonnentag, 2008), and adds to existing research by identifying a linkage between interpersonal justice and LMX - related gratitude. However, in interpreting the notably strong justice - emoti on bivariate relationships, it is important to acknowledge these relationships are likely inflated by common method variance. As part of hypothesis 5, i nterpersonal justice was expected to buffer the negative effects of low LMXSC and heighten the positive effects of high LMXSC on externally - focused LMX - related emotions . A moderation was observed with LMX - related anger , but not with LMX - 145 related gratitude. Findings for LMX - related anger partially matched expectations. As per expectations, interperson al justice was associated with lowered level s of anger experienced by high LMXSC respondents. Unexpectedly, those with low LMXSC who observed high interpersonal justice reported the lowest level of anger, even lower than high LMXSC - high interpersonal justi ce respondents. It may be that when low LMXSC individuals experience high than when high LMXSC individuals experience the same behavior. That is, when a supervis or shows consideration to even the members with whom he or she has the weakest relationship, that supervisor may be seen as transformational or at the very least, egalitarian . This fits with 7 , p.25 ) assertion that individuals react less negatively to supervisors who supervisors who show inconsistent interpersonal justice ( ). Last, counter the expectations made in hypothesis 6, a significant three - way interaction between supervisor - attributed locus of causality for RBS, LMXSC, and interpersonal justice predicting LMX - related anger was not observed. There are several potential reason s why this interaction did not emerge. From a practical standpoint, the contributing two - way interactions (LMXSC - locus of causality, LMXSC - interpersonal justice) did not match study expectations. At a more conceptual level, the fair process effect suggests procedural and interactional justice can buffer the negative effects of low distributive justice. Here, LMXSC may not be an important enough outcome that the interpersonal justices would necessarily be required to buffer the anger induced by low LMXSC. In deed, LMXSC showed weaker association s with anger than self - and supervisor - attributed locus of causality for RBS, LMX, interpersonal and informational justice. 146 Ancillary a nalyses: Comparing Emotions and PANA models Turning next to the analyses on emotio ns and the PANA S items, attributions did not moderate effects of LMXSC on PA or NA , despite observed LMXSC - PA and LMXSC - NA associations of similar size and direction to LMXSC - emotion associations. Th ese results support the notion that LMX - related emotions are distinguished by attributional underpinnings, whereas PA and NA do not specify an attributional component. Further, LMX - related emotions were conceptualized here as more cool and cognitive emotional metaperceptions . Although PA and NA items were also i LMX, responses to PA and NA items may still reflect . Tests of the PANA and emotion measurement model showed a model allowing all emotion and PANAS items to load on separate but related factors was be tter fitting than a two - factor model comprised of positive and negative valenced items. This measurement model finding indicates the emotions of interest here cannot be fully explained by an overarching PA and NA model. Given the strong bivariate relatio nship between shame and guilt items, there is reason to question their discriminant validity. Existing literature has observed high correlations between guilt and shame, but in general, the constructs are considered conceptually distinct in their focus. ( T angney, Miller, Flicker, & Barlow, 1996 ) In support of their distinctiveness, the seven factor model with guilt and shame on separate factors showed significantly less misfit than the six factor model that loaded shame and guilt items on a single factor . The fact that hypothesis 2 was supported for shame, but was not supported for guilt provides additional evidence of discriminant validity. Given these findings, I chose to analyze guilt and shame as separate 147 constructs. However, future research should focu s on how the subtle differences in these constructs can be more effectively measured. Ancillary a nalyses: LMXSC and its relationship with LMX Next, the relationship between LMXSC and LMX was further explored. LMXSC showed psychometric characteristics simi lar to those observed in existing studies (i.e., Vidyarthi et al., 2010; Erdogan, 2002). Overall, the data supported the notion that LMXSC and LMX are related but distinct constructs that share substantive variance but have meaningful differences in their focus. R unning the hypotheses with LMX entered in place of LMXSC allowed me to explore the question, do attributions and interpersonal justice also moderate LMX quality - emotion relationships ? Although these analyses had a fundamentally different focus t han the original hypotheses, results tended to show a similar pattern as was found with LMXSC , outside of stronger LMX - emotion linkages than LMXSC - emotion linkages . associations with LMX - related shame, gratitude, and anger. With LMX - related shame, self - attributed locus of causality greatly reduced LMX - related shame among low LMX respondents, but made little difference for high LMX individuals. Thus, attributing as influential in relationship building may be an important factor in lowering LMX - related shame With externally - focused emotions, the effect of LMX on emotions w as accentuated when the supervisor was viewed as driving the relationship quality , indicating s upervisor - attributed locus of causality may exacerbate both the positive and negative effects of LMX on emotions. Analyses were also run to see whether interpersonal justice buffered the negative effects of low LMX quality on externally focused LMX - relat ed emotions . As with LMXSC, an LMX - 148 interpersonal justice interaction was observed with anger, but not with gratitude . The results for LMX - related anger matched expectations put forth by H5b: LMX was not important in determining levels of anger when interp ersonal justice was high, but was especially important when interpersonal justice was low. Thus, although LMX is a different target than LMXSC, many expectations put forth by attribution theory and the fair process effect can also help to ociation with LMX - related emotions. Ancillary a nalyses: Comparing Informational Justice and Interpersonal Justice Next, given interpersonal and informational justice are both facets of interactional justice and tend to be highly correlated , analyses were moderating effects also held for informational justice . As was the case with interpersonal justice, informational justice did moderate the LMXSC - anger linkage, but not the LMXSC - gratitude linkage . With LMX - rela ted anger , the results matched the LMXSC - interpersonal justice moderation findings. Informational justice was negatively associated with levels for anger for both high and low LMXSC individuals, but the effect was stronger for low LMXSC individuals. Low LMXSC - high informational justice respondents in fact reported the lowest levels of LMX - related anger. This suggests the honesty and straightforwardness of communication with a supervisor may improve (i.e. lower) the level of LMX - related anger subordinates feel, especially when individuals feel they have relatively poor quality LMX. supervisor is respectful and honest appear to have similar effects on the extent to which LMX SC elicits LMX - related anger. Ancillary a nalyses: Work group size Given social comparisons are likely to change as the size of the comparison group incr eases, it was also of interest to look at whether workgroup size was associated with the 149 and hypothesized relationships. LMX - related guilt and shame increased as d ; LMX - related pride, gratitude, and anger did not differ based on workgroup size. Further, the effect of LMXSC on shame decreased as the workgroup size increased. These findings suggest negative LMX - related self - conscious emotions tend to be lower in smaller workgroups, but are more likely to stem from LMXSC than with large groups. Future research should be directed towards understanding how and when specific small group processes influen ce LMXSC and moderate its relationship with outcomes, both at group and individual levels. Ancillary a I also examined how LMX relationship tenure influenced study variables and hypotheses. LMX theory w ould suggest the quality of the exchange between a supervisor and subordinate change s over time as the dyad In the changed the longer the dyad worked together. The longer individuals worked with their current supervisor, the m ore likely they were to rate their LMX as high quality and of a relatively higher quality than the LMX levels observed among peers. LMX relationship tenure also had an effect on LMX - related gra titude , with those working for their supervisor for 5 or more years reporting the highest level of gratitude. Time, however, likely cannot fully account for the positive effects of a longer : j ust as the attraction - selection - attrition (ASA ; Schneider, 1987 ) theory suggests employees are more likely to seek out, enter , and remain members of s likely seek out, enter , and remain in LMX dyads that are effective. Thus, in a cross - sectional design , we might expect LMX tenure to be associated with positive LMX perceptions simply because subordinates with longer relationship tenure have elected to remain working for the same supervisor. 150 riables did not change significantly based on the length of the LMX relationship. Relationship length did not appear to play a significant role in most LMX - related emotions (i.e. pride, guilt, shame, anger, Appendix T ), nor was it an important factor in th e extent to which an individual made self or supervisor locus of causality for RBS attributions. These findings suggest that while LMX and LMXSC may stabilize over time, LMX - related emotions and locus of causality for RBS attributions may vary even among d yads who have extensive LMX tenure. Last, interpersonal justice significantly differed based on relationship length, with the high mean levels of interpersonal justice reported among those who recently started working for their supervisor, and a notable drop appeared among those who have worked for their supervisor for 6 - 12 months. Beyond the 6 - 12 month relationship length , interpersonal justice steadil y in creased with relationship length . S everal other variables (i.e. LMX guilt, shame, and anger) showed similar (albeit nonsignificant) blips in means, where those in the first six months of their relationship with the supervisor reported particularly positive experiences, and those who had worked for their supervisor for 6 - 12 months reporting more negative experiences. Research on organizational social ization suggests that in the first several months of a new job, newcomers experience a boost in job satisfaction beyond past job satisfaction levels Shipp, Payne, & Culbertson, 2009). Here it appears LMX relationships may also show a honeymoon and hangover effect, particularly in terms of interpersonal justice. It may be that while subordinates are getting socialized into a new role, supervisors are particularly attentive to the needs and concerns of the newcomer. Further, supervisors (and subordinates) may engage in greater levels of impression management in new relationships. As subordinates become more autonomous in 151 their role, supervisors may focus less on showing polite and conside rate treatment to the subordinate and more on task accomplishment. S ubordinates may then perceive the interpersonal treatment their supervisor displays to be less considerate than the treatment they experienced initially. LMX relationship tenure moderated the relationship between LMXSC and LMX - related pride, with the association between LMXSC and pride weakening the longer subordinates worked for their supervisor . For relatively new LMX relationships, LMXSC was associated with feeling capable, valuable, ; this was less the case as LMX tenure increased. Still, relationship tenure did not moderate the relationship between LMXSC and LMX - related guilt, shame, anger, and gratitude , indicating the role of social compari sons in LMX - tenure . LMX tenure also showed two significant three - way interactions: an interaction with LMXSC and self - attributed locus of causality for RBS predicting LMX - related shame , and an interaction with LMXSC and interpersonal justice predicting LMX - related anger. The findings from these interactions were not entirely intuitive, but indicate that LMX tenure may sometimes exacerbate the effects of LMXSC on emotions and at other ti mes buffer the effects. Future studies should focus on the mechanisms for when and why negative LMX - related emotions occur at later stages of LMX relationships. Theoretical I mplications and Future Directions Several findings observed in this study have important implications for LMX theory ; these findings are listed as follows. 152 Although LMXSC plays a role in subordinates LMX - related emotions, its association s were small relative to interpersonal justice and thinking LMXSC is a relatively new area of research within the broader stream of LMX resear ch. Although the present study showed LMXSC as having some effect on LMX - relation emotions, associations were s mall and weaker than the associations with other predictors. As researchers continue to examine how LMXSC relates to workplace and career attitudes and behaviors, it is important to interpret findings relative to other findings of other relevant variables. When subordinates experience a sense of personal relationship (e.g. high self - attributed locus of causality for RBS) , they experience more LMX - related positive emotions. dings indicate viewing Future research should seek to explore the role self - attributed locus of causality for RBS plays in other workplace and ca reer behaviors. Although LMX theory would suggest supervisor - subordinate dyads change in nature over time, LMX relationship tenure did not moderate most LMXSC - LMX - related associations, nor did it moderate hypotheses. As noted previously, it appears that LMXSC and interactions involving LMXSC did not differentially predict LMX - related emotions based on LMX relationship tenure. Whether subordinates are at the beginning of an LMX relationship and after the relationship has been established, it appears LMXSC exhibits the same 153 relationship with emotions and attributions have the same effect on the LMXSC - emotion relationships. As LMX relationship tenure increa ses, LMXSC increases. Perhaps intuitive ly , this study shows that as subordinates who have worked for their supervisor for a long period of time, they are more likely to perceive themselves as having a relatively higher quality relationship with the supervi sor than their peers. Below are several other future research questions stemming from the current study. What role do discrete LMX - related emotions play in LMX - , job - , and organization - related attitudes and behaviors? The current study focused on potential antecedents of LMX - related emotions; future studies should explore how LMX - related emotions relate other attitudes and behaviors at work. Is RLMX associated with LMX - related emotions? If so, do locus of causality for RBS attributions and interpersonal justice similarly moderate the effects of RLMX on LMX - related emotions? Existing research assumes LMXSC is a perception that mediates the linkage between RLMX and subordinate outcomes (Vidyarthi et al., 2010) . Future res earch should examine whethe r attributions and interpersonal justice moderate the direct linkages between RLMX and LMX - related emotions. Does task type or leader decision making style moderate the linkages between LMX variability and different subordinate outcomes? More democratic task structures (e.g. compensatory tasks) and decision making styles may make LMX variability more surprising and uncomfortable than in hierarchical contexts. Additional research on this topic may be merited. 154 More broadly, researchers should consider whether LMXSC adds significant value to our und erstanding of employee cognition, affect, and behavior . Although social comparisons have consistently shown associations with important employee outcomes ( Greenberg et al., 2007 ), the current study showed LMXSC was more weakly associated with focal outcomes of this study findings, LMXSC does not appear to be a n immensely important factor in how subordinates feel regarding their LMX. A lso, a lthough this study focused on self - and supervisor - attributed locus of causality for RBS, open text responses in the pilot study showed that most participants thought both they and framework might not have yielded valuable insights. Specifically, a ttribution theory does not make predictions about the effects of multiple causal loci, and thus it would not suggest a LMXSC - - related emotions. Further, rceived causes of abusive supervision suggest joint indirect aggression, or OCBs. Thus , it seems unlikely th of causa lity attributions would have provided meaningful insight in the current study. This study also has several implications for research on attribution theory and affect, listed here. Current measures of causal attributions show weak psychometric properties. The measure developed to measure self - and supervisor - attributed locus of causality for RBS was more psychometrically sound than many existing measures. Focusing research on 155 more context - specific attributions (as in the current study) may prove more effective than the existing, more generalized approach to attribution measurement. Core affect theory provides a useful framework for understanding affect as part of attribution theory and other areas of affect . Core affect theory allows affect researchers to make subtle distinctions in affect constructs, in a sense differentiating Future studies should utilize core affect theory as a way of explicating their affective constructs and comparing and contrasting those construc ts to existing research. Direct measures of negative, self - conscious emotions may elicit defensive response patterns. In the current study, respondents who were expected to exhibit the highest levels of negative, self - conscious emotions instead reported p articularly positive feelings. As Tagney and Dearing (2003) suggest, using situational measures may mitigate this concern in future studies. Practical Implications and employees. Results from this study indicated that supervisor - attributed locus of causality for RBS was generally linked to negative outcomes for subordinates. To ensure subordinates do not solely blame their supervisor when misunderstandings occur wit hin the dyad , supervisors should . In these conversations, both parties should discuss how they contributed to an issue and what both could do differently in the future. For employees , results indicate those who see themselves as driving the effectiveness of their relationship with the supervisor feel more positively about their relationship quality. For this reason, and 156 take owners hip over not only their work, but also their LMX quality. E mployees should seek to strengths, weaknesses, goals, motivations, and constraints, and use this information to strategize and engage in work content and processes th at meet mutual needs. Limitations There were several limitations to this study that have been noted throughout the discussion, and are summarized here . First, c ommon method variance (CMV) may have inflated the correlations between justice and LMX - related emotions. CMV is less of a concern when focusing on interactions; still, future research should test whether the findings for interpersonal justice hold when justice and emotion measures are temporally separated. Second, guilt and shame were highly correl ated, indicating low discriminant validity. Additional research on LMX - related emotions should consider using different measures of shame and guilt to see if differences in these constructs can be assessed more effectively. Third, PA and NA were similarly highly correlated with positive and negative emotions, calling into question whether emotions are specifically the requisite affective outcome of the model. The results of the affect measurement model provides some evidence of discriminant validity. Furth er, the observation that hypothesis tests yielded different results when PANAS items were the dependent variable supports the notion that emotions are a more appropriate model outcome. Still, additional research should focus on the extent to which LMX - rela ted emotions and LMX - related PA and NA are distinctive from one another, both theoretically and empirically. Fourth , for both the pilot and the main study, online panel studies were used as a sample. Although highly rigorous data cleaning was applied to sc reen out duplicate and careless respondents, it was not possible to formally verify participants have a full time job. This leaves open the possibility that participants could theoretically lie about these characteristics to be eligible to participate in t he study. Still, it 157 is worth noting that the same issue could occur in an undergraduate sample, wherein students could falsely state they have work experience to increase subject pool study eligibility . Conclusion This study proposed LMXSC is linked to di fferent LMX - related emotions, and these supervisor as well as perceptions of interpersonal justice. In a two - part survey of employed adults, interpersonal justice and causal locus attributions regarding LMX relationship building were associated with LMX - related emotions, and in many cases moderated the extent to which LMXSC was associated with LMX - related emotions. This study reveals th e linkage between LMXSC and how su bordinates feel regarding their LMX is not necessarily straightforward; attributions regarding why and how their LMX came to be are also important. Research in organizational psychology can integrate th ese findings to help better explain the mechanisms thr ough which LMX variability leads to positive and negative outcomes. 158 APPENDICES 159 Appendix A: Pilot Survey Consent Form Pilot Survey: Causal Attribution Measurement Properties Please read the information below completely and carefully: In this questionnaire, you will be answering several questions about your relationship with your manager/immediate supervisor and your beliefs about how that relationship developed. Because this study focuses on the psychometric properties of si milar measures, you will be asked to answer very similar questions about the same topic. You will receive $1 for your participation in this study. Please be aware that your confidentiality will be protected to the maximum extent allowable by law. This means that all disseminated data from this and future surveys will not contain any identifying information that links individuals to their responses , a nd under no circumstances will anyone other than the Michigan State University research er have access to identifying information . All data will be stored o n a password - protected computer, on a secure system, at Michigan State University. The data will be kept for the regulatory minimum of three years after the study is completed. The results of the survey may be disseminated or published in an aggregated, group form, but the identities of all employees in these publications will remain completely anonymou s . Please note that in addition to the researcher at Michigan State University, the MSU the records and research results. Please note that your participati on is voluntary. This means that refusal to participate will involve NO penalty. We will not be keeping record of who does not participate. You are free to withdraw your consent and discontinue participation at any point during the survey. You may choose n ot to answer specific questions or to stop participating at any time. We very much appreciate your time and participation! If you have any questions about this survey, our larger research effort , or to report an injury, please contact Catherine Ott - Hollan d by phone: (262) 497 - 2486, email: ottholla@msu.edu, or regular mail: 348 Psychology Building, East Lansing, MI 48824 or Dr. Ann Marie Ryan, 333 P sychology Building , East Lansing, MI 48824 or ryanan@msu.edu . If you have any questions or concerns about you r role and rights as a research participant, would like to obtain information or offer input, or would like to register a complaint about this research study, you may contact, anonymously if you wish, the Michigan State University Human Research Protection Program at 517 - 355 - 2180, FAX 517 - 432 - 4503, or e - mail irb@msu.edu, or regular mail at: 408 W. Circle Dr., 207 Olds Hall, MSU, East Lansing, MI 48824. If you consent to participation in this study, please press the button below to continue. If you do not agree to participate in this study, you may close out of this webpage. 160 Appendix B: Example Causal Attribution Measures Attributional Style Questionnaire Dykema, Bergbower, Doctora, and Peterson (1996) Please try to imagine yourself in th e following situations. If such a situation happened to you, what do you think might have caused it? While situations like these may have many causes, we want you to choose only one - THE MAIN CAUSE, THAT IS, WHAT MADE THIS SITUATION HAPPEN TO YOU. Please w rite the main cause in the box after each situation. Next, answer two questions about the cause you provided. First, how likely is it that the main cause you gave will continue to affect you? Second, is the main cause that you gave something that affects j ust this situation, or does it affect other areas of your life? To summarize, please: Read each situation and vividly imagine it happening to you. Decide what you feel would be the one main cause for the situation if it happened to you. Write down the on e main cause in the box provided. Answer the two questions about the main cause. Try to imagine yourself in the following situation. of breaking the law How likely is it that the main cause you gave will continue to affect you? W ill never affect you - 3 - 2 - 1 0 1 2 3 Will always affect you Is the main cause that you gave somet hing that affects just this situation , or does it affect other areas of your life? Just affects this sort of event - 3 - 2 - 1 0 1 2 3 Affects all other areas 161 Revised Causal Dimension Scale (CDSII) McAuley, Duncan, and Russell (1992) Instructions: Think about the reason or reasons you have written above. The items below concern your impressions or opinions of this cause or causes of your performance. Circle one number for each of the following questions. Is the cause(s) of something: That reflects an aspect of yourself 9 8 7 6 5 4 3 2 1 Reflects an aspect of the situation Manageable by you 9 8 7 6 5 4 3 2 1 Not manageable Permanent 9 8 7 6 5 4 3 2 1 Temporary You can regulate 9 8 7 6 5 4 3 2 1 You cannot regulate Over which others have control 9 8 7 6 5 4 3 2 1 Over which others have no control Inside of you 9 8 7 6 5 4 3 2 1 Outside of you Stable over time 9 8 7 6 5 4 3 2 1 Variable over time Under the power of other people 9 8 7 6 5 4 3 2 1 Not under the power of other people Something about you 9 8 7 6 5 4 3 2 1 Something about others Over which you have power 9 8 7 6 5 4 3 2 1 Over which you have no power Unchangeable 9 8 7 6 5 4 3 2 1 Changeable Other people can regulate 9 8 7 6 5 4 3 2 1 Other people cannot regulate Note. The total scores for each dimensions are obtained by summing the items, as follows: 1,6,9= locus of causality; 5, 8, 12= external control; 3,7,11=stability; 2,4,10= personal control. Attributions to abusive supervision Burton, Taylor, and Barber (2014) Inte rnal attributions 2. I need to look in the mirror to examine why my supervisor treats me the way he or she does. 3. I probably provoked my supervisor to act the way he or she does. 4. I am at fault for the way my supervisor treats me at work. External attributions 5. My supervisor chooses to act the way he or she does. she is.) 7. The cause of Relational attributions 9. The relationship I have with my supervisor is one of the reason he or she a cts the way he or she does toward me. 162 Marital Attitude Scale Subscales Pretzer, Epstein, and Fleming (1991) Perceived Ability of Couple to Change Relationship Example item: Expectancy of Improvement in Relationship Example item: I think our relationship might improve. Attribution of Causality to Own Behavior Example item: My problems with my partner are caused by the things I say and do. Attribution of Causality to Own Personality Example item: My personality would have to change for my partner and me to get along better. Example item: If my partner did something differentl he/she is. Attribution of Malicious Intent to Spouse Example item: I doubt if my partner del iberately does things to irritate me. Attribution of Lack of Love to Spouse 163 Instructions: Often in workgroups, some subordinates have better relationships with their supervisor in comparison to the relationships other coworkers have with the supervisor. Other subordinates may have worse relationships with their supervisor in comparison to the r elationships other coworkers have with the supervisor. Think about the reason or reasons for why this might be. The items below concern your impressions or opinions of what causes effective relationship building with your supervisor. Rate the extent to w hich you agree with the following statements. The quality of the relationship that has developed between you and your supervisor Is mostly due to the actions I take Reflects something about me Is incited mostly by me Is mostly the consequence of how I c hose to do my work Is generally the result of how I spend my time Tends to be determined by the things I say or do Is mostly due to the actions my supervisor takes Reflects something about my supervisor Is incited mostly by my Is m ost ly the consequence of how my supervisor choses to do his or her work Is generally the result of how my supervisor spends his/her time Tends to be determined by the things my supervisor says or does 1=Strongly disagree 2=Disagree 3=Neither agree nor disagree 4=Agree 5=Strongly agree 164 Appendix D: Pilot Survey Measures with Revisions Marked Revised Causal Dimension Scale (CDSII) McAuley, Duncan, and Russell (1992) Instructions: Think about the reason or reasons you have written above the quality of your relationship with your direct supervisor. The items below concern your impressions or opinions of this cause or causes of your performance the quality of that relationship. Circle one number for each of the following questions. Is the cause(s) of you r relationship with your direct supervisor something: That reflects an aspect of yourself 9 8 7 6 5 4 3 2 1 Reflects an aspect of the situation Manageable by you 9 8 7 6 5 4 3 2 1 Not manageable Permanent 9 8 7 6 5 4 3 2 1 Temporary You can regulate 9 8 7 6 5 4 3 2 1 You cannot regulate Over which others have control 9 8 7 6 5 4 3 2 1 Over which others have no control Inside of you 9 8 7 6 5 4 3 2 1 Outside of you Stable over time 9 8 7 6 5 4 3 2 1 Variable over time Under the power of other people 9 8 7 6 5 4 3 2 1 Not under the power of other people Something about you 9 8 7 6 5 4 3 2 1 Something about others Over which you have power 9 8 7 6 5 4 3 2 1 Over which you have no power Unchangeable 9 8 7 6 5 4 3 2 1 Changeable Other people can regulate 9 8 7 6 5 4 3 2 1 Other people cannot regulate Note. The total scores for each dimensions are obtained by summing the items, as follows: 1,6,9= locus of causality; 5, 8, 12= external control; 3,7,11=stability; 2,4,10= personal control. Attributional Style Questionnaire Dykema, Bergbower, Doctora, and Peterson (1996) Please try to imagine yourself in the following situations. the quality of the relationship that has been built between you and your direct supervisor. If such a situation happened to you, what do you think might have caused it your relationship to be of this quality ? While situations like these may have many causes, we want you to choose only one - THE MAIN CAUSE, THAT IS, WHAT MADE THIS SITUATION HAPPEN TO YOU WHAT DETERMIN ES THE RELATIONSHIP QUALITY YOU HAVE WITH YOUR SUPERVISOR . Please write the main cause in the box after each situation. Next, answer two questions about the cause you provided. First, how likely is it that the main cause you gave will continue to affect yo u? Second, is the main cause that you gave something that affects just this situation your relationship with your supervisor , or does it affect other areas of your life? To summarize, please: 165 Read each situation and vividly imagine it happening to you. De cide what you feel is would be the one main cause for the situation if it happened to you. the quality of the relationship you have with your supervisor? Write down the one main cause in the box provided. Answer the two questions about the main cause. Try to imagine yourself in the following situation. H ow likely is it that the main cause you gave will continue to affect you? Will never affect you - 3 - 2 - 1 0 1 2 3 Will always affect you Is the main cause that you gave something that affects just this situation your relationship with your supervisor , or does it affect other areas of your life? Just affects this sort of event - 3 - 2 - 1 0 1 2 3 Affects all other areas 166 Appendix E : First Consent Form for Online Panel Participants Relationships in the Workplace Survey Part 1 Please read the information below completely and carefully: This is the first questionnaire of a two - part survey study on your perceptions of your workplace, with a special focus on your relationship with your supervisor. We expect that it will take about 1 0 minutes for you to compl ete this survey. T h e survey focuses on how employees view their relationships with their colleagues and supervisors, and what factors employees think contribute to effective relationships at work. You will receive $1 for your participation in this study. You will receive an invitation from MTurk to participate in a second online questionnaire approximately 1 week from today for $3. It is very important to our research that you respond to both questionnaires. Please be aware that your confidentiality wil l be protected to the maximum extent allowable by law. This means that all disseminated data from this and future surveys will not contain any identifying information that links individuals to their responses , and under no circumstances will anyone other t han the Michigan State University research er have access to identifying information . All data will be stored o n a password - protected computer, on a secure system, at Michigan State University. The data will be kept for the regulatory minimum of three years after the study is completed. The results of the survey may be disseminated or published in an aggregated, group form, but the identities of all employees in these publications will remain completely anonymous . Please note that in addition to the researcher at Michigan State University, the MSU HRPP records and research results. Please note that your participation is voluntary. This means that refusal to participate will involve NO penalty. We will not be keeping record of who does not participate. You are free to withdraw your consent and discontinu e participation at any point during the survey. You may choose not to answer specific questions or to stop participating at any time. We very much appreciate your time and participation! If you have any questions about this survey, our larger research eff ort , or to report an injury, please contact Catherine Ott - Holland by phone: (262) 497 - 2486, email: ottholla@msu.edu, or regular mail: 348 Psychology Building, East Lansing, MI 48824 , or Dr. Ann Marie Ryan, 333 Psychology Building, East Lansing, MI 48824 or ryanan@msu.edu. If you have any questions or concerns about your role and rights as a research participant, would like to obtain information or offer input, or would like to register a complaint about this research study, you may contact, anonymously if you wish, the Michigan State University Human Research Protection Program at 517 - 355 - 2180, FAX 517 - 432 - 4503, or e - mail irb@msu.edu, or regular mail at: 408 W. Circle Dr., 207 Olds Hall, MSU, East Lansing, MI 48824. 167 If you consent to participation in this study, please enter your MTurk ID number. We will use this as an anonymous identifier to link your two surveys. Then, click the continue arrow. If you do not agree to participate in this study, you may close out of this webpage. 168 Appendix F : Second Con sent Form for Online Panel Participants Relationships in the Workplace Survey Part 2 Please read the information bel ow completely and carefully: This is the second questionnaire of a two - part survey study on your perceptions of your workplace, with a spec ial focus on your relationship with your supervisor . If you received an invitation to participate in this study, you should have par ticipated in an earlier survey 1 week ago. We expect that it will take about 1 0 minutes for you to complete this survey. T h e survey focuses on how employees view their relationships with their supervisors, and the physical symptoms of stress. You will receive $3 for participating in this survey. Please be aware that your confidentiality will be protected to the maximum extent allowable by law. This means that all disseminated data from this and future surveys will not contain any identifying information that links individuals to their responses , and under no circumstances will anyone other than the Michigan State University res earch er have access to identifying information . All data will be stored o n a password - protected computer, on a secure system, at Michigan State University. The data will be kept for the regulatory minimum of three years after the study is completed. The results of the survey may be disseminated or published in an aggregated, group form, but the identities of all employees in these publications will remain completely anonymous . Please note that in addition to the researcher at Michigan State University, the MSU HRPP records and research results. Please note that your participation is voluntary. This means that refusal to participate will involve NO penalty. We will not be keeping record of who does not participate. You are free to withdraw your consent and discontinu e participation at any point during the survey. You may choose not to answer specific questions or to stop participating at any time. We very much appreciate your time and participation! If you have any questions about this survey, our larger research eff ort , or to report an injury, please contact Catherine Ott - Holland by phone: (262) 497 - 2486, email: ottholla@msu.edu, or regular mail: 348 Psychology Building, East Lansing, MI 48824 , or Dr. Ann Marie Ryan, 333 Psychology Building, East Lansing, MI 48824 or ryanan@msu.edu. If you have any questions or concerns about your role and rights as a research participant, would like to obtain information or offer input, or would like to register a complaint about this research study, you may contact, anonymously if you wish, the Michigan State University Human Research Protection Program at 517 - 355 - 2180, FAX 517 - 432 - 4503, or e - mail irb@msu.edu, or regular mail at: 408 W. Circle Dr., 207 Olds Hall, MSU, East Lansing, MI 48824. If you consent to participation in this study, please enter your MTurk ID number. We will use this as an anonymous identifier to link your two surveys. Then, click the continue arrow. If you do not agree to participate in this study, you may close out of t his webpage. 169 Appendix G : End of Survey Message for Online Panel Participants The researchers at Michigan State University would like to thank you for your participation in their two - part survey looking at e built with their supervisors. This study examined how thoughts and feelings employees experience regarding their relationship with their supervisor are related to perceptions of a fair environment and physiological side effects of stress. A brief repor can be sent to you by request at ottholla@msu.edu. For additional information on how to develop an effective relationship with your supervisor, check out the following articles Forbes 14 Tips for Improving your Relationship with your Boss Fast Company 12 Ways to Build the Best Relationship with your Boss 170 Appendix H : Leader - Member Exchange Social Comparison Please rate the extent to which you agree with the following statements regarding your relationship with your manager/immediate supervisor. I have a better relationship with my manager than most others in my work group. When my manager cannot make it to an important meeting, it is likely that s/he will ask me to fill in. Relative to the others in my work group, I receive more support from my manager. The working relationship I have with my manager is more effective than the relationships most members of my group have with my manager. My manager is more loyal to me compared to my coworkers. My manager enjoys my company more than he/she enjoys the company of other group members. 1= Strongly disagree 2= Disagree 3= Somewhat disagree 4= Neither agree or disagree 5= Somewhat agree 6= Agree 7= Strongly agree 171 Appendix I : Leader - Member Exchange Answer the following questions with regard to your relationship with your manager/ immediate supervisor. 1. Do you know where you stand with your do you usual ly know how satisfied your leader is with what you do? Rarely/Occasionally /Some times/Fairly Often /Very Often 2. How well does your supervisor understand your job problems and needs? Not a bit/A little/ A fair am ount/Quite a bit/ A great deal 3. How well does your supervisor recognize your potential? Not at all/ A little /Moderately/Mostly/ Fully 4. Regardless of how much formal authority he/she has built into his/her position what are the chances that your supervisor would use his/her power to help you solve problems in your work? None/ S mall/Moderate/High/ Very High 5. Again, regardless of the amount of formal authority your supervisor has, what are the chances N one/Small/Moderate/High/ Very High 6. I have enough confidence in my supervisor that I would defend and justify his/her decision if he/she were not present to do so. Strongly Disagree/Disagree/Neutral/Agree/Strongly Agree 7. How would you characterize you r working relationship with your supervisor ? Extremely Ineffective/Worse than Average/Average/Better than Average/Extremely Effective 172 Appendix J : Demographic Questions Reminder: All survey responses are anonymous. Only the researchers will access to s urvey data and results of this study will only be presented in aggregate. Please answer the following questions about yourself: Work Experience: _____ years, _____ months How long have you been in your current position? ____years, ____ months How long have you worked for your organization? ____years, ____ months How long have you worked with your current supervisor? ______ years _____ months How many other employees work for current supervisor? ____________ My age is: ___________ Are yo u of Hispanic, Latino, or Spanish origin? No, not of Hispanic, Latino, or Spanish origin Yes, Mexican, Mexican American, or Chicano Yes, Puerto Rican Yes, Cuban Yes, another Hispanic, Latino, Spanish origin _________________ My race is (mark one or more) White Black, African American, or Negro American Indian or Alaska Native Asian Indian Chinese Filipino Other Asian___________________ Japanese Korean Vietnamese Native Hawaiian Guamanian or Chamorro Samoan Other Pacific Islander_______________ Some other race_______________ 173 Appendix K : Interactional Justice Supervisor - focused interactional justice Colquitt (2001) The following items refer to your manager/immediate supervisor. To what extent: (Interpersonal justice) Has he/she treated you in a polite manner? Has he/she treated you with dignity? Has he/she treated you with respect? Has he/she refrained from improper remarks or comments? (Informational justice) Has he/she been candid in his/her communications with you? Has he/she explained the procedure s thoroughly? Were his/her explanations regarding the procedures reasonable? Has he/she communicated details in a timely manner? 1 Never 2 Rarely 3 Occasionally 4 A moderate amount 5 A great deal 174 Appendix L : LMX - related Emotions (pride subscale) pleased about what I have done (guilt subscale) (shame subscale) (gratitude subscale) (anger subscale) Not at all Very Little Somewhat To a great extent (Optional) Please feel free to provide any other thoughts on your relationship with your supervisor. [open text box] 175 Appendix M : Positive and Negative Affectivity Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measu res of positive and negativ e affect: The PANAS scale. Journal of Personality and Social Psychology, 54 , 1063 - 1070. Read each item and then, using the scale below, mark the appropriate answer in the space next to that word. Indicate to what extent you feel this way about your relationship with your manager/immediate supervisor. 1 VERY SLIGHTLY OR NOT AT ALL 2 A LITTLE 3 MODERATELY 4 QUITE A BIT 5 EXTREMELY _____i nterested _____ irritable _____ distressed _____ alert _ ____ excited _____ ashamed _____ upset _____ inspired _____ strong _____ nervous _____ guilty _____ determined _____ scared _____ attentive _____ hostile _____ jittery _____ enthusiastic _____ active _____ proud _____ afraid 176 Appendix N : Example Emotion Measures Examples of single item emotion measures Guilt, Shame, Anger Barclay, L. J., Skarlicki, D. P., & Pugh, S. D. (2005). Exploring the Role of Emotions in Injustice Perceptions and Retaliation. Journal of Applied Psychology , 90(4), 6 29 - 643. Participants were asked to think about how they felt when they were being laid off and to indicate the extent to which they felt specific emotions. Pride, Anger, Guilt Weiss, H. M., Suckow, K., & Cropanzano, R. (1999). Effects of justice conditions on dis crete emotions. Journal of Applied Psychology, 84 (5), 786. works and gro up them into emotion categories. how you feel about what just - point scale which ranges from not at all to somewhat to very much . Although we were only interested in the four emotions of happiness, pride, anger and guilt, we also asked about such primary emotions as anxiety, compassion, disgust, envy, embarrassment, fear, and love. These were included in order to reduce the demand characteristics for the specific emotions tested. No a priori predictions were made about thes Anger Rupp, D. E., & Spencer, S. (2006). When customers lash out: the effects of customer interactional injustice on emotional labor and the mediating role of discrete emotions. Journal of Applied Psychology , 91(4), 971. happiness were measured using the anger and happiness subscales from the discrete emotions inventory originally developed by Weiss (1999) and modified by Mattern, Bedwell, and Rupp (2004). The two scales consisted for 7 and 10 items, respectively, and cont ained single - word emotional adjectives (e.g. anger, fury, joy, and happiness) to which participants indicated the extent which they were currently feeling each emotion using a scale ranging from 1 177 Anger, Guilt Coulter, R. H., & Pinto, M. B. (1995). Guilt appeals in advertising: what are their effects?. Journal of Applied Psychology, 80 , 697 - 705. - irritation, happy - amused, and guilt (see Table 2) based on previous resea rch (Batra & Ray, 1986; Edell & Burke, 1987) and freely elicited emotional responses obtained in Phase 1 of this study. Participants used a 7 - point Likert - type scale, ranging from 1 (not at all) to 7 (very strongly), to respond to the following question: We would like you to tell us how the advertisement made you feel. For each of the [emotions] Guilt Otterbacher, J. R., & Munz, D. C. (1973). State - trait measure of experiential guil t. Journal of Consulting and Clinical Psychology, 40, 115. Perceived Guilt Index (PGI) The G - Trait scale of the PGI is essentially a single - item measure. People are asked to select one adjective from a list of 11 adjectives (varying in level of guilt) th at best describes how they Unforgivable Disgraceful Degraded Marred Reproachable Chagrined Fretful Pent - up Restrained Undisturbed Innocent Qualitative approaches Anger Friedman, R., Anderson, C., Brett, J., Olekalns, M., Goates, N., & Lisco, C. C. (2004). The positive and negative effects of anger on dispute resolution: evidence from electronically mediated disputes. Journal of Applied Psychology , 89, 369. ed responses. Multi - item measures Pride 178 Tyler, T. R., & Blader, S. L. (2003). The group engagement model: Procedural justice, social identity, and coopera tive behavior. Personality and Social Psychology Review, 7, 349 - 361. My company is one of the best companies in its field; People are impressed when I tell them where I work My company is well respected in its field; I think that where I work reflects well on me; I am proud to tell others where I work. Pride, Guilt, Sha me Marschall, D. E., Sanftner, J., & Tangney, J. P. (1994). The State Shame and Guilt Scale. George Mason University, Fairfax, VA. They showed pride alpha =.93, shame .88, guilt .89 Please answer the following questions thinking about how you feeling as you think about the transgression you described. Be as honest and as accurate as possible. Use the following scale: Strongly Disagree 1 2 3 4 5 Strongly Agree Pride I feel go od about myself. I fe el worthwhile, valuable. I feel capable, useful. I feel proud. I feel pleased about what I have done. Guilt I feel remorse, regret. I feel tension of what I have done. I cannot stop thinking about the bad thing that I hav e done. I feel like apologizing, confessing. I feel bad about what I have done. Shame I want to sink into the floor and disappear. I feel small. I feel that I am a bad person. I feel humiliated, disgraced. I feel worthless, powerless. Gratitude McCullough, M. E., Emmons, R. A., & Tsang, J. A. (2002). The Grateful Disposition: A Conceptual and Empirical Topography. Journal of Personality and Social Psychology , 82, 112 - 127. I have so much in life to be thankful for. 179 If I had to list eve rything that I felt grateful for, it would be a very long list. I am grateful to a wide variety of people. As I get older I find myself more able to appreciate the people, events, and situation s that have been part of my life history. Long amounts of time can go by before I feel grateful for something or someone else. Gratitude Watkins, P. C., Woodward, K., Stone, T., & Kolts, R. (2003). Gratitude and happiness: Development of a measure of gra titude and relationships with subjective well - being . Social Behavior and Personality , 31 , 431 - 452. Gratitude, Appreciation, and Resentment Test (GRAT) For some reason I never seem to get the breaks that others get. More bad things have happened to me in my life than I deserve. I never seem to get the breaks that other people do. gone through in my life, I really feel like the world owes me something. I think that life has handed me a short stick I basically feel like life has ripped me off It sure seems like ot hers get a lot more benefits in life than I do. Life has been good to me It seems like people have frequently tried to impede my progress. I believe that I am a very fortunate person. At Christmas I never seemed to get as many presents or presents that were as good as others received. I believe that things in life that are really enjoyable are just as available to me as they are to Ross Perot or Donald Trump. Every Fall I really enjoy watching the leaves change colors. I really enjoy the changing seasons Oftentimes I have been overwhelmed at the beauty of nature. I love the green of spring . The simple pleasures of life are the best pleasures of life. The simple pleasu res of life are the best pleasures of life. privilege supported me and helped me along the way. 180 I feel deeply appreciative for the things others have done for me in my life. important to remember how others have contribute d to my accomplishments. Many people have given me valuable wisdom throughout my life that has been important to my success. One of my favorite times of the year is Thanksgiving. Part of really enjoying something good is being t hankful for that thing. support. I feel grateful for the education I have received. Gratitude Gordon, A. M., Impett, E. A., Kogan, A., Oveis, C., & Keltner, D. (2012). To have and to hold: gratitude promotes relationship maintenance in intimate bonds. Journal of Personality and Social Psychology, 103, 257 - 274. Appreciative subscale I tell my partner often that s/he is the best I often tell my partner how much I appreciate her/him. At times I take my partner for granted (R). I appreciate my partner. I make sure my part ner feels appreciated. My partner sometimes says that I fail to notice the nice things s/he does for me. (R) I acknowledge the things that my partner does for me, even the really small things. I am sometimes stuck with a sense of awe and wonder when I thi nk about my partner being in my life. Appreciated subscale My partner makes sure I feel appreciated. When I am with my partner, sometimes s/he will look at me excitedly and tell me how much s/he appreciates me. My partner often tells me the things that s /he really likes about me. At times, my partner takes me for granted (R). My partner often expresses My partner makes me feel special. Gratitude Adler, M. G., & Fagley, N. S. (2005). Appreciation: Individual Differences in Finding Value and Meaning as a Unique Predictor of Subjective Well Being. Journal of Personality, 73 , 79 - 114. 181 Multifactorial Appreciation Scale Gratitude Subscale (10) I notice the sacrifices that my friends make for me. Food, clothing, and shelter are basic needs that I do not need to be grateful for because I am entitled to them. I acknowledge when people go out of their way for me. s my appreciation for their help. I am very fortunate for the opportunity to receive an education. I value the sacrifices that my parents (or guardians) have made (and/or make) for me. Anything that my parents (or guardians) have done for me can be attribu ted to their responsibility as parents (or guardians), and I do not need to be thankful because that was their job. Anger Fuqua, D. R., Leonard, E., Masters, M. A., Smith, R. J., Campbell, J. L., & Fischer, P.C. (1991). A structural analysis of the stat e - trait anger expression inventory. Educational and Psychological Measurement, 51 (2), 439. State Anger Subscale Furious Irritated Angry Feel like yelling Feel like breaking Mad Feel like banging Feel like hitting Burned up Feel like swearing 182 Appendix O : Item statistics of All Main Study Items Table 33 Item statistics of all main study items M SD N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1. LMXSC1 4.47 1.54 451 - 2. LMXSC2 3.81 1.87 452 .49 - 3. LMXSC3 4.03 1.54 452 .74 .49 - 4. LMXSC4 4.35 1.49 451 .77 .50 .75 - 5. LMXSC5 3.91 1.60 452 .69 .51 .76 .78 - 6. LMXSC6 4.13 1.61 452 .73 .45 .72 .75 .78 - 7. LoCRBSSelf1 3.52 .94 453 .49 .23 .43 .45 .37 .37 - 8. LoCRBSSelf2 3.61 .95 453 .55 .29 .48 .50 .43 .44 .65 - 9. LoCRBSSelf3 3.07 .98 453 .36 .20 .34 .37 .34 .34 .59 .54 - 10. LoCRBSSelf4 3.7 1.02 453 .43 .26 .39 .42 .35 .34 .58 .53 .43 - 11. LoCRBSSelf5 3.43 .99 453 .42 .18 .36 .38 .32 .33 .53 .51 .48 .61 - 12. LoCRBSSelf6 3.62 .92 453 .43 .21 .42 .46 .36 .40 .58 .54 .50 .53 .52 - 13. LoCRBSSup1 3.29 .93 453 - .17 - .08 - .05 - .10 - .03 - .08 - .21 - .19 - .17 - .24 - .18 - .19 - 14. LoCRBSSup2 3.72 .84 453 - .07 - .02 .00 - .07 - .02 - .01 - .14 - .01 - .12 - .14 - .12 - .08 .44 - 15. LoCRBSSup3 3.31 .90 453 - .19 - .11 - .09 - .16 - .07 - .10 - .28 - .25 - .19 - .29 - .20 - .27 .62 .45 - 16. LoCRBSSup4 3.38 .93 453 - .09 - .06 .01 - .06 .00 - .03 - .15 - .14 - .15 - .11 - .14 - .12 .54 .39 .52 17. LoCRBSSup5 3.22 .98 453 - .07 - .02 - .02 - .02 .02 .03 - .10 - .10 - .06 - .10 - .02 - .11 .47 .38 .44 18. LoCRBSSup6 3.51 .91 453 - .14 - .11 - .04 - .08 - .10 - .09 - .20 - .19 - .15 - .17 - .13 - .13 .57 .42 .53 19. LMX1 3.91 .95 453 .41 .29 .31 .38 .30 .33 .25 .34 .12 .28 .25 .30 - .12 - .06 - .12 20. LMX2 3.64 1.08 452 .43 .21 .37 .41 .32 .34 .32 .39 .22 .28 .27 .36 - .17 - .06 - .16 21. LMX3 3.66 1.10 450 .51 .36 .44 .51 .44 .46 .38 .43 .21 .40 .36 .38 - .15 - .10 - .16 22. LMX4 3.57 1.04 452 .44 .28 .38 .40 .34 .36 .32 .40 .18 .34 .32 .39 - .15 - .06 - .18 23. LMX5 2.98 1.10 453 .49 .36 .44 .44 .43 .42 .29 .43 .21 .26 .27 .33 - .13 - .07 - .15 24. LMX6 3.67 1.03 452 .48 .29 .38 .44 .38 .38 .40 .47 .26 .40 .39 .41 - .20 - .09 - .24 25. LMX7 3.57 .83 453 .58 .38 .49 .54 .47 .49 .41 .52 .25 .41 .39 .45 - .20 - .09 - .21 183 M SD N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 26. InterpersonalJ.1 4.35 .82 401 .33 .12 .24 .25 .20 .19 .18 .27 .10 .22 .24 .27 - .17 - .12 - .24 27. InterpersonalJ.2 4.36 .83 399 .38 .18 .32 .35 .24 .28 .27 .36 .18 .31 .32 .32 - .17 - .11 - .23 28. InterpersonalJ.3 4.32 .86 399 .41 .19 .33 .37 .29 .29 .32 .36 .19 .31 .36 .33 - .18 - .15 - .22 29. InterpersonalJ.4 4.29 .93 400 .15 .06 .10 .12 .05 .05 .15 .21 .08 .15 .21 .14 - .10 - .08 - .14 30. InformationalJ.1 4.07 .88 401 .32 .14 .23 .27 .20 .22 .26 .30 .20 .17 .27 .32 - .07 - .04 - .17 31. InformationalJ.2 3.84 .94 401 .24 .08 .20 .21 .19 .17 .23 .24 .19 .19 .27 .29 - .08 - .03 - .12 32. InformationalJ.3 3.99 .95 400 .19 .05 .15 .14 .10 .11 .15 .26 .10 .12 .23 .25 - .11 - .01 - .10 33. InformationalJ.4 3.89 .97 401 .24 .10 .23 .22 .18 .18 .27 .24 .18 .25 .29 .36 - .12 - .11 - .18 34. InformationalJ.5 3.62 1.08 401 .36 .16 .30 .31 .33 .28 .28 .33 .20 .26 .39 .32 - .10 - .10 - .20 35. Pride1 3.95 1.14 401 .47 .21 .40 .40 .34 .30 .35 .44 .22 .29 .37 .37 - .16 - .12 - .18 36. Pride2 4.03 1.16 401 .42 .20 .34 .38 .32 .30 .35 .43 .25 .28 .31 .33 - .13 - .17 - .20 37. Pride3 4.06 1.17 400 .45 .21 .37 .41 .36 .33 .36 .44 .25 .31 .33 .35 - .19 - .19 - .23 38. Pride4 4.16 1.08 400 .41 .21 .37 .38 .31 .30 .34 .40 .19 .28 .29 .33 - .18 - .16 - .24 39. Pride5 4.17 1.05 399 .44 .24 .38 .43 .32 .31 .33 .38 .20 .26 .29 .33 - .14 - .10 - .19 40. Pride6 3.73 1.21 401 .46 .24 .39 .43 .38 .35 .38 .45 .27 .26 .31 .36 - .12 - .12 - .19 41. Pride7 3.98 1.13 400 .42 .18 .36 .41 .33 .31 .34 .47 .22 .31 .34 .37 - .14 - .12 - .18 42. Guilt1 1.49 .99 399 - .19 - .01 - .11 - .12 - .06 - .06 - .20 - .33 - .15 - .17 - .24 - .18 .21 .12 .17 43. Guilt2 1.49 1.00 399 - .25 - .01 - .19 - .19 - .13 - .14 - .25 - .37 - .18 - .24 - .30 - .27 .20 .11 .17 44. Guilt3 1.7 1.14 400 - .17 - .03 - .11 - .13 - .11 - .11 - .20 - .27 - .17 - .15 - .21 - .18 .17 .15 .14 45. Guilt4 1.43 .89 400 - .01 .02 .01 .01 .07 .05 - .07 - .08 - .05 - .08 - .12 - .07 .07 .06 .00 46. Guilt5 1.37 .83 395 - .11 .02 - .09 - .12 - .02 - .05 - .13 - .16 - .03 - .16 - .15 - .17 .10 .06 .08 47. Shame1 1.35 .90 400 - .20 - .02 - .11 - .12 - .01 - .09 - .22 - .27 - .15 - .18 - .24 - .23 .17 .09 .16 48. Shame2 1.51 1.00 399 - .23 - .07 - .17 - .18 - .14 - .12 - .22 - .29 - .16 - .22 - .22 - .22 .16 .19 .18 49. Shame3 1.2 .67 401 - .11 - .01 - .06 - .09 - .01 - .01 - .15 - .23 - .10 - .17 - .22 - .14 .08 .05 .07 50. Shame4 1.33 .89 399 - .20 - .04 - .12 - .13 - .06 - .08 - .23 - .28 - .14 - .21 - .21 - .20 .21 .16 .18 51. Shame5 1.29 .81 399 - .15 .03 - .08 - .09 - .03 - .04 - .16 - .27 - .12 - .16 - .23 - .14 .20 .11 .17 52. Shame6 1.3 .81 396 - .25 - .05 - .19 - .18 - .14 - .13 - .27 - .35 - .15 - .20 - .24 - .22 .15 .12 .10 53. Shame7 1.66 1.20 397 - .27 - .08 - .21 - .24 - .19 - .20 - .29 - .43 - .23 - .28 - .27 - .24 .14 .16 .15 184 M SD N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 54. Gratitude1 3.24 1.46 399 .34 .16 .30 .36 .26 .27 .30 .33 .23 .28 .33 .29 - .11 - .08 - .16 55. Gratitude2 2.87 1.44 398 .34 .22 .32 .31 .31 .24 .22 .29 .21 .24 .28 .29 - .10 - .07 - .10 56. Gratitude3 3.81 1.25 400 .44 .17 .38 .39 .35 .30 .37 .47 .28 .33 .39 .40 - .12 - .13 - .23 57. Gratitude4 3.7 1.36 399 .43 .22 .39 .39 .31 .28 .32 .45 .23 .33 .32 .37 - .11 - .07 - .19 58. Anger1 1.43 .98 401 - .25 - .04 - .15 - .17 - .10 - .12 - .27 - .29 - .18 - .27 - .27 - .25 .27 .13 .21 59. Anger2 1.57 1.13 401 - .29 - .07 - .18 - .23 - .16 - .16 - .27 - .36 - .20 - .32 - .32 - .31 .21 .12 .20 60. Anger3 1.41 .95 401 - .28 - .02 - .20 - .20 - .14 - .16 - .29 - .32 - .18 - .26 - .31 - .30 .18 .07 .20 61. Anger4 1.49 1.03 395 - .34 - .07 - .25 - .25 - .18 - .19 - .33 - .39 - .24 - .32 - .37 - .40 .19 .10 .25 62. PA1 3.27 1.07 399 .45 .33 .47 .45 .43 .38 .36 .45 .30 .28 .35 .38 .02 .05 - .05 63. PA2 2.47 1.21 400 .41 .32 .41 .38 .41 .35 .30 .36 .32 .22 .29 .35 .05 .06 - .02 64. PA3 2.83 1.26 400 .45 .35 .42 .43 .39 .35 .30 .33 .25 .25 .23 .31 - .04 - .03 - .06 65. PA4 2.85 1.24 401 .41 .26 .38 .43 .35 .35 .39 .44 .30 .31 .33 .38 - .09 .00 - .10 66. PA5 2.94 1.26 400 .50 .38 .44 .46 .41 .39 .39 .46 .29 .33 .28 .40 - .09 - .01 - .14 67. PA6 3.16 1.21 400 .17 .18 .16 .18 .19 .15 .12 .14 .20 .07 .08 .17 .03 .11 .02 68. PA7 2.79 1.32 401 .47 .35 .42 .46 .43 .39 .37 .43 .30 .31 .30 .39 - .08 .01 - .08 69. PA8 3.32 1.21 400 .41 .30 .35 .40 .34 .36 .34 .40 .24 .34 .31 .37 - .15 - .04 - .17 70. PA9 3.55 1.06 400 .29 .23 .24 .29 .23 .23 .23 .30 .18 .21 .19 .25 .00 .08 - .04 71. PA10 3.07 1.13 399 .34 .29 .33 .37 .27 .23 .33 .32 .26 .22 .22 .28 .04 .06 - .04 72. NA1 1.61 .98 401 - .23 - .07 - .14 - .14 - .11 - .12 - .18 - .29 - .12 - .21 - .18 - .17 .21 .13 .17 73. NA2 1.48 .95 400 - .26 - .07 - .17 - .19 - .14 - .11 - .26 - .36 - .21 - .29 - .28 - .22 .22 .18 .20 74. NA3 1.24 .61 401 .00 .03 .04 .06 .14 .07 - .03 - .07 .06 - .05 - .03 - .03 .06 .00 .04 75. NA4 1.21 .59 400 - .10 .02 - .05 - .08 .01 - .02 - .11 - .16 - .02 - .13 - .15 - .11 .17 .11 .16 76. NA5 1.33 .79 401 - .22 - .02 - .17 - .22 - .13 - .12 - .25 - .32 - .21 - .29 - .29 - .29 .22 .14 .26 77. NA6 1.62 1.01 400 - .31 - .07 - .27 - .26 - .22 - .22 - .32 - .34 - .21 - .32 - .32 - .31 .14 .11 .18 78. NA7 1.21 .59 401 - .03 .07 .02 - .01 .08 .01 - .06 - .10 .01 - .03 - .07 - .07 .07 .00 .09 79. NA8 1.65 .96 401 - .15 - .07 - .14 - .16 - .08 - .07 - .15 - .21 - .10 - .12 - .14 - .15 .15 .11 .11 80. NA9 1.46 .88 401 - .15 - .04 - .14 - .11 - .08 - .08 - .15 - .19 - .06 - .17 - .17 - .10 .09 .11 .08 81. NA10 1.29 .72 401 - .08 .01 - .06 - .05 - .01 .00 - .11 - .16 - .04 - .15 - .15 - .10 .16 .07 .14 185 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 17. LoCRBSSup5 .51 - 18. LoCRBSSup6 .53 .57 - 19. LMX1 - .06 - .10 - .12 - 20. LMX2 - .10 - .11 - .11 .54 - 21. LMX3 - .08 - .11 - .10 .56 .63 - 22. LMX4 - .11 - .13 - .10 .46 .58 .57 - 23. LMX5 - .12 - .09 - .15 .40 .51 .50 .66 - 24. LMX6 - .14 - .14 - .19 .44 .59 .62 .67 .61 - 25. LMX7 - .13 - .14 - .16 .71 .81 .82 .83 .78 .83 - 26. InterpersonalJ.1 - .06 - .12 - .18 .36 .44 .42 .46 .37 .49 .53 - 27. InterpersonalJ.2 - .12 - .07 - .13 .44 .49 .47 .51 .37 .55 .59 .77 - 28. InterpersonalJ.3 - .12 - .07 - .14 .41 .49 .51 .52 .41 .54 .61 .75 .87 - 29. InterpersonalJ.4 - .12 - .01 - .11 .21 .24 .24 .23 .20 .28 .30 .51 .52 .54 - 30. InformationalJ.1 - .03 - .04 - .05 .35 .47 .41 .39 .33 .41 .49 .46 .46 .50 .32 - 31. InformationalJ.2 - .06 - .08 - .06 .37 .49 .36 .37 .31 .41 .49 .46 .49 .51 .35 .45 - 32. InformationalJ.3 - .06 - .12 - .06 .36 .48 .36 .44 .32 .44 .50 .52 .55 .58 .42 .46 .69 - 33. InformationalJ.4 - .06 - .12 - .10 .37 .52 .41 .45 .32 .48 .54 .50 .51 .56 .33 .50 .68 .68 - 34. InformationalJ.5 - .09 - .03 - .14 .29 .43 .41 .49 .45 .48 .54 .50 .57 .62 .39 .42 .61 .57 .63 - 35. Pride1 - .11 - .05 - .11 .43 .51 .54 .52 .47 .53 .63 .57 .64 .69 .37 .47 .51 .52 .54 .58 36. Pride2 - .13 - .11 - .17 .43 .52 .52 .51 .43 .51 .61 .57 .67 .72 .39 .46 .51 .51 .50 .51 37. Pride3 - .19 - .13 - .14 .38 .49 .52 .49 .42 .54 .59 .56 .68 .68 .35 .45 .48 .45 .50 .54 38. Pride4 - .18 - .12 - .15 .39 .43 .47 .46 .39 .48 .55 .53 .61 .63 .34 .39 .44 .44 .45 .52 39. Pride5 - .14 - .18 - .13 .39 .45 .50 .49 .43 .55 .59 .57 .60 .63 .37 .43 .46 .49 .43 .46 40. Pride6 - .11 - .12 - .13 .42 .51 .55 .53 .47 .54 .63 .51 .57 .61 .34 .53 .50 .49 .51 .54 41. Pride7 - .19 - .14 - .14 .42 .50 .57 .48 .44 .53 .62 .51 .59 .64 .37 .44 .46 .47 .51 .52 42. Guilt1 .15 .20 .19 - .34 - .35 - .31 - .25 - .21 - .32 - .37 - .44 - .43 - .43 - .30 - .30 - .31 - .44 - .38 - .32 43. Guilt2 .14 .12 .14 - .35 - .40 - .33 - .32 - .25 - .35 - .42 - .46 - .51 - .52 - .37 - .34 - .37 - .46 - .40 - .45 44. Guilt3 .11 .11 .15 - .36 - .37 - .34 - .27 - .23 - .32 - .39 - .36 - .49 - .45 - .28 - .31 - .37 - .38 - .35 - .37 45. Guilt4 .03 .07 .04 - .29 - .19 - .22 - .16 - .07 - .20 - .23 - .22 - .37 - .30 - .20 - .20 - .27 - .36 - .31 - .21 186 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 46. Guilt5 .07 .09 .09 - .32 - .30 - .28 - .22 - .20 - .28 - .33 - .35 - .44 - .38 - .26 - .21 - .28 - .31 - .33 - .26 47. Shame1 .13 .10 .15 - .41 - .25 - .28 - .24 - .18 - .32 - .35 - .38 - .54 - .46 - .34 - .28 - .36 - .43 - .37 - .31 48. Shame2 .17 .16 .16 - .36 - .36 - .36 - .34 - .24 - .36 - .42 - .41 - .52 - .58 - .34 - .38 - .41 - .45 - .42 - .43 49. Shame3 .07 .06 .10 - .33 - .16 - .22 - .21 - .14 - .24 - .27 - .33 - .45 - .43 - .35 - .22 - .25 - .40 - .32 - .23 50. Shame4 .18 .14 .23 - .29 - .30 - .28 - .30 - .24 - .37 - .37 - .36 - .51 - .49 - .31 - .30 - .30 - .37 - .37 - .35 51. Shame5 .16 .16 .21 - .25 - .25 - .22 - .26 - .23 - .33 - .32 - .36 - .51 - .48 - .36 - .26 - .26 - .41 - .30 - .34 52. Shame6 .12 .09 .14 - .28 - .29 - .31 - .28 - .22 - .31 - .35 - .43 - .53 - .55 - .40 - .32 - .31 - .37 - .37 - .37 53. Shame7 .14 .07 .18 - .29 - .37 - .38 - .30 - .31 - .41 - .43 - .39 - .51 - .55 - .32 - .38 - .36 - .40 - .38 - .45 54. Gratitude1 - .12 - .08 - .10 .29 .40 .38 .40 .39 .43 .48 .39 .43 .48 .27 .35 .32 .38 .34 .38 55. Gratitude2 - .08 - .03 - .09 .24 .34 .36 .37 .41 .37 .44 .35 .40 .44 .20 .33 .35 .38 .36 .42 56. Gratitude3 - .13 - .08 - .10 .38 .50 .49 .49 .46 .54 .60 .56 .64 .67 .39 .48 .53 .51 .53 .60 57. Gratitude4 - .12 - .06 - .07 .35 .48 .45 .46 .46 .54 .58 .53 .58 .61 .36 .45 .41 .49 .49 .52 58. Anger1 .23 .16 .23 - .25 - .29 - .34 - .37 - .32 - .40 - .41 - .39 - .51 - .48 - .35 - .30 - .33 - .40 - .38 - .38 59. Anger2 .23 .14 .23 - .33 - .36 - .36 - .36 - .33 - .43 - .45 - .45 - .53 - .52 - .39 - .29 - .35 - .42 - .40 - .40 60. Anger3 .16 .13 .21 - .28 - .35 - .34 - .37 - .31 - .44 - .44 - .50 - .55 - .50 - .37 - .33 - .37 - .44 - .44 - .44 61. Anger4 .21 .18 .23 - .32 - .39 - .36 - .39 - .32 - .48 - .48 - .47 - .56 - .53 - .34 - .36 - .41 - .48 - .43 - .43 62. PA1 - .01 .05 .02 .31 .44 .42 .47 .47 .51 .55 .44 .47 .51 .24 .40 .40 .38 .43 .52 63. PA2 .04 .10 .06 .21 .32 .27 .34 .37 .35 .39 .25 .28 .35 .11 .31 .32 .28 .36 .43 64. PA3 - .04 .03 - .04 .31 .35 .40 .41 .41 .39 .47 .32 .32 .37 .16 .30 .35 .28 .34 .44 65. PA4 .02 .04 - .03 .32 .46 .41 .46 .45 .48 .54 .35 .42 .47 .15 .34 .38 .36 .45 .48 66. PA5 - .06 - .02 - .07 .37 .49 .47 .53 .50 .55 .61 .40 .47 .51 .24 .40 .37 .39 .44 .50 67. PA6 .04 .03 - .05 .18 .13 .14 .17 .16 .09 .18 .00 .04 .03 .04 .12 .13 .05 .13 .11 68. PA7 - .01 .04 - .05 .34 .47 .46 .47 .45 .49 .56 .37 .40 .46 .23 .34 .40 .35 .44 .50 69. PA8 - .06 - .04 - .09 .39 .38 .43 .42 .40 .42 .51 .31 .37 .41 .20 .30 .35 .34 .37 .36 70. PA9 .04 .02 - .01 .33 .32 .36 .39 .35 .34 .43 .26 .32 .30 .22 .26 .26 .29 .32 .31 71. PA10 .07 .07 .03 .26 .30 .28 .30 .27 .28 .36 .16 .25 .26 .13 .26 .30 .23 .27 .29 187 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 72. NA1 .11 .14 .17 - .34 - .37 - .32 - .34 - .30 - .46 - .44 - .54 - .56 - .55 - .35 - .28 - .37 - .45 - .35 - .42 73. NA2 .16 .11 .19 - .30 - .36 - .36 - .36 - .30 - .44 - .44 - .50 - .57 - .60 - .41 - .32 - .35 - .49 - .41 - .43 74. NA3 - .02 .08 .06 - .26 - .15 - .13 - .14 - .06 - .19 - .19 - .29 - .34 - .29 - .23 - .15 - .21 - .32 - .21 - .12 75. NA4 .13 .10 .16 - .28 - .21 - .22 - .23 - .16 - .31 - .29 - .38 - .44 - .43 - .32 - .21 - .29 - .36 - .30 - .28 76. NA5 .17 .17 .23 - .24 - .34 - .32 - .38 - .28 - .44 - .42 - .54 - .55 - .53 - .40 - .31 - .39 - .45 - .43 - .40 77. NA6 .10 .10 .14 - .33 - .43 - .40 - .41 - .34 - .48 - .50 - .61 - .65 - .65 - .35 - .37 - .48 - .51 - .54 - .53 78. NA7 - .01 .05 .08 - .21 - .19 - .16 - .19 - .08 - .20 - .21 - .35 - .44 - .35 - .29 - .23 - .30 - .34 - .31 - .21 79. NA8 .08 .08 .11 - .31 - .37 - .27 - .32 - .23 - .35 - .38 - .42 - .47 - .48 - .32 - .28 - .37 - .41 - .33 - .36 80. NA9 .08 .09 .12 - .24 - .27 - .26 - .27 - .19 - .31 - .32 - .39 - .44 - .47 - .32 - .27 - .24 - .35 - .32 - .29 81. NA10 .12 .11 .17 - .25 - .22 - .19 - .24 - .19 - .28 - .28 - .36 - .42 - .42 - .32 - .24 - .27 - .38 - .28 - .25 188 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 36. Pride2 .82 - 37. Pride3 .78 .81 - 38. Pride4 .75 .75 .81 - 39. Pride5 .72 .74 .72 .76 - 40. Pride6 .75 .73 .73 .64 .68 - 41. Pride7 .78 .73 .72 .70 .66 .74 - 42. Guilt1 - .44 - .44 - .47 - .47 - .41 - .41 - .44 - 43. Guilt2 - .51 - .51 - .54 - .51 - .44 - .46 - .51 .83 - 44. Guilt3 - .44 - .47 - .49 - .45 - .37 - .41 - .43 .68 .68 - 45. Guilt4 - .25 - .25 - .23 - .25 - .22 - .20 - .27 .48 .40 .58 - 46. Guilt5 - .38 - .40 - .42 - .42 - .31 - .28 - .40 .56 .54 .60 .62 - 47. Shame1 - .46 - .48 - .48 - .51 - .45 - .36 - .44 .63 .60 .56 .55 .68 - 48. Shame2 - .50 - .53 - .55 - .54 - .44 - .46 - .52 .60 .61 .61 .52 .56 .64 - 49. Shame3 - .38 - .37 - .35 - .38 - .33 - .28 - .38 .52 .48 .43 .50 .55 .71 .60 - 50. Shame4 - .46 - .45 - .51 - .48 - .39 - .38 - .48 .59 .58 .57 .44 .57 .69 .64 .71 - 51. Shame5 - .46 - .46 - .47 - .47 - .43 - .36 - .47 .59 .57 .57 .42 .53 .68 .59 .73 .77 - 52. Shame6 - .52 - .53 - .54 - .51 - .40 - .37 - .51 .56 .58 .47 .37 .51 .66 .65 .69 .73 .66 - 53. Shame7 - .53 - .54 - .58 - .53 - .43 - .51 - .54 .57 .61 .63 .39 .47 .56 .66 .50 .63 .62 .64 - 54. Gratitude1 .49 .48 .45 .41 .45 .48 .51 - .29 - .30 - .27 - .10 - .17 - .24 - .30 - .21 - .28 - .27 - .26 - .34 55. Gratitude2 .49 .47 .43 .38 .37 .49 .49 - .22 - .27 - .24 - .11 - .18 - .21 - .26 - .14 - .23 - .23 - .23 - .32 56. Gratitude3 .75 .72 .76 .65 .64 .72 .72 - .43 - .50 - .45 - .22 - .31 - .40 - .48 - .29 - .41 - .40 - .41 - .55 57. Gratitude4 .66 .61 .65 .57 .57 .63 .62 - .41 - .46 - .39 - .19 - .30 - .38 - .40 - .27 - .37 - .38 - .35 - .52 58. Anger1 - .48 - .45 - .48 - .43 - .41 - .38 - .40 .52 .52 .51 .44 .42 .52 .48 .42 .60 .56 .53 .54 59. Anger2 - .55 - .52 - .54 - .49 - .43 - .42 - .51 .58 .63 .58 .38 .54 .59 .59 .54 .65 .62 .62 .63 60. Anger3 - .49 - .49 - .51 - .46 - .43 - .40 - .43 .51 .57 .50 .32 .45 .52 .47 .48 .57 .52 .55 .52 61. Anger4 - .55 - .55 - .53 - .48 - .48 - .45 - .48 .55 .61 .55 .33 .49 .59 .56 .47 .58 .55 .57 .57 189 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 62. PA1 .57 .52 .52 .46 .47 .58 .57 - .24 - .32 - .21 - .10 - .18 - .23 - .30 - .20 - .24 - .22 - .27 - .34 63. PA2 .46 .39 .39 .32 .31 .50 .41 - .11 - .21 - .11 - .03 - .07 - .11 - .16 - .06 - .11 - .10 - .12 - .21 64. PA3 .50 .45 .44 .42 .39 .52 .47 - .17 - .24 - .16 - .08 - .13 - .14 - .20 - .07 - .17 - .12 - .13 - .26 65. PA4 .55 .49 .49 .44 .43 .53 .53 - .22 - .32 - .23 - .11 - .16 - .21 - .28 - .13 - .25 - .20 - .22 - .32 66. PA5 .58 .53 .57 .49 .48 .67 .59 - .26 - .35 - .25 - .12 - .21 - .25 - .30 - .15 - .26 - .21 - .23 - .37 67. PA6 .12 .15 .15 .16 .13 .16 .16 - .02 - .02 .01 .07 .00 - .06 .01 - .04 - .01 .02 - .01 .02 68. PA7 .53 .50 .50 .44 .42 .53 .54 - .25 - .29 - .22 - .06 - .18 - .19 - .25 - .10 - .20 - .16 - .20 - .31 69. PA8 .45 .46 .45 .45 .45 .45 .51 - .27 - .30 - .22 - .11 - .22 - .26 - .26 - .20 - .23 - .24 - .23 - .27 70. PA9 .38 .39 .36 .34 .36 .36 .38 - .20 - .24 - .14 - .11 - .16 - .22 - .22 - .23 - .20 - .21 - .22 - .21 71. PA10 .38 .36 .33 .32 .31 .42 .36 - .11 - .13 - .08 - .04 - .05 - .11 - .08 - .09 - .08 - .06 - .11 - .17 72. NA1 - .51 - .52 - .57 - .53 - .51 - .43 - .45 .54 .59 .57 .34 .47 .51 .54 .37 .53 .47 .48 .55 73. NA2 - .54 - .55 - .55 - .48 - .51 - .46 - .47 .57 .63 .53 .36 .42 .54 .51 .49 .62 .54 .58 .61 74. NA3 - .21 - .28 - .24 - .23 - .28 - .19 - .16 .47 .39 .39 .44 .41 .54 .37 .46 .46 .41 .41 .39 75. NA4 - .34 - .38 - .40 - .39 - .35 - .33 - .34 .51 .48 .45 .43 .47 .58 .48 .56 .55 .55 .52 .46 76. NA5 - .47 - .51 - .51 - .45 - .51 - .44 - .43 .50 .53 .44 .27 .43 .49 .46 .41 .46 .45 .50 .49 77. NA6 - .62 - .61 - .62 - .54 - .53 - .55 - .55 .52 .59 .54 .35 .48 .50 .57 .37 .50 .44 .53 .59 78. NA7 - .36 - .42 - .39 - .38 - .35 - .28 - .30 .45 .42 .43 .40 .60 .62 .42 .49 .46 .52 .51 .40 79. NA8 - .43 - .46 - .46 - .46 - .39 - .38 - .38 .51 .49 .55 .38 .46 .48 .57 .41 .52 .47 .51 .50 80. NA9 - .37 - .42 - .41 - .41 - .34 - .36 - .41 .52 .46 .52 .36 .44 .45 .57 .47 .54 .54 .53 .49 81. NA10 - .34 - .39 - .36 - .37 - .34 - .30 - .32 .52 .48 .48 .38 .47 .57 .48 .55 .53 .57 .54 .48 190 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 55. Gratitude2 .68 - 56. Gratitude3 .58 .56 - 57. Gratitude4 .56 .57 .80 - 58. Anger1 - .35 - .29 - .41 - .40 - 59. Anger2 - .36 - .32 - .49 - .45 .77 - 60. Anger3 - .36 - .29 - .44 - .42 .76 .79 - 61. Anger4 - .38 - .34 - .51 - .46 .71 .87 .81 - 62. PA1 .41 .46 .58 .56 - .24 - .36 - .35 - .39 - 63. PA2 .28 .43 .46 .46 - .08 - .15 - .16 - .20 .70 - 64. PA3 .31 .41 .51 .49 - .18 - .22 - .23 - .24 .59 .61 - 65. PA4 .44 .45 .56 .54 - .28 - .28 - .30 - .31 .67 .71 .63 - 66. PA5 .41 .52 .66 .66 - .26 - .34 - .32 - .37 .70 .68 .72 .73 - 67. PA6 .11 .12 .12 .08 - .04 - .07 - .06 - .08 .31 .28 .31 .28 .25 - 68. PA7 .46 .51 .60 .58 - .23 - .31 - .28 - .32 .63 .67 .63 .72 .76 .30 - 69. PA8 .34 .35 .50 .49 - .28 - .28 - .30 - .34 .55 .48 .58 .60 .62 .41 .63 - 70. PA9 .25 .30 .36 .37 - .23 - .23 - .26 - .25 .46 .44 .46 .47 .47 .57 .49 .65 - 71. PA10 .24 .32 .40 .34 - .15 - .14 - .16 - .20 .53 .56 .58 .54 .54 .49 .58 .60 .64 - 72. NA1 - .30 - .28 - .50 - .46 .53 .60 .61 .58 - .32 - .17 - .25 - .33 - .36 .04 - .30 - .25 - .19 - .11 73. NA2 - .36 - .32 - .51 - .45 .64 .67 .60 .64 - .30 - .15 - .19 - .31 - .33 .02 - .28 - .26 - .19 - .15 74. NA3 - .08 - .02 - .17 - .12 .38 .32 .36 .38 .02 .04 - .01 - .01 - .06 - .02 - .01 - .09 - .13 - .05 75. NA4 - .20 - .14 - .31 - .23 .50 .45 .50 .49 - .16 - .05 - .15 - .18 - .20 .01 - .12 - .19 - .14 - .12 76. NA5 - .33 - .27 - .46 - .43 .61 .63 .72 .68 - .31 - .15 - .19 - .28 - .30 - .01 - .27 - .28 - .21 - .14 77. NA6 - .39 - .34 - .58 - .55 .59 .63 .68 .69 - .43 - .28 - .29 - .42 - .42 .01 - .38 - .32 - .22 - .19 78. NA7 - .15 - .13 - .28 - .20 .41 .44 .46 .50 - .08 .00 - .07 - .09 - .14 - .05 - .07 - .18 - .19 - .08 79. NA8 - .25 - .20 - .40 - .34 .41 .46 .47 .45 - .22 - .14 - .22 - .26 - .28 .05 - .23 - .17 - .11 - .13 80. NA9 - .26 - .19 - .37 - .29 .42 .46 .45 .42 - .17 - .07 - .16 - .19 - .19 .07 - .17 - .16 - .06 - .03 81. NA10 - .21 - .17 - .31 - .27 .47 .47 .48 .50 - .13 - .04 - .08 - .12 - .17 .03 - .10 - .15 - .10 - .08 191 72 73 74 75 76 77 78 79 80 81 73. NA2 .72 - 74. NA3 .41 .49 - 75. NA4 .53 .58 .64 - 76. NA5 .61 .68 .39 .59 - 77. NA6 .70 .70 .38 .49 .74 - 78. NA7 .44 .46 .68 .58 .54 .49 - 79. NA8 .61 .57 .52 .62 .53 .60 .46 - 80. NA9 .52 .54 .48 .58 .52 .55 .43 .72 - 81. NA10 .49 .60 .62 .78 .57 .48 .60 .61 .60 - 192 Appendix P: Factor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study Items Table 34 Factor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study Items Factor Facto r 1 2 3 4 5 6 7 8 1 - 2 - .61 - 3 - .13 .51 - 4 - .31 .53 .59 - 5 .24 - .24 - .10 - .28 - 6 - .40 .55 .19 .31 - .09 - 7 .65 - .58 - .26 - .45 .30 - .43 - 8 - .52 .63 .28 .34 - .19 .52 - .50 - Note. Pairwise deletion was used, item N s =395 - 453. This table shows the factor intercorrelations. The largest correlation was between the guilt/ shame factor and anger factor (factors 1 and 7, r =.65). The weakest correlation was between the supervisor - attributed locus of causality for RBS factor and the informational justice factor (i.e., factors 5 and 6, r = - .09). Several other moderately - sized correlations (i.e. r s =.40 - .60) emerged between fac tors. The guilt/shame factor (factor 1) showed negative relationships with the positive emotions ( factor 2, r = - .61), informational justice (factor 6, r = - .40), and interpersonal justice factors (factor 8, r = - .52). The positive emotion factor (factor 2) exhi bited positive relationships with the LMXSC (factor 3; r =.51), self - attributed locus of causality for RBS (factor 4; r =.53), informational justice (factor 6; r =.55), and interpersonal justice factors (factor 8, r =.63), and negative relationships with the a nger factor (factor 7; r = - .58). The LMXSC factor (factor 3) showed a positive relationships with the self - attributed locus of causality for RBS factor (factor 4; r =.59). The self - attributed locus of causality for RBS factor (factor 4) also showed a 193 moderat ely - sized negative relationship with the anger factor (factor 7; r = - .45). The informational justice factor (factor 6) was negatively related to the anger factor (factor 7; r = - .43), and a positively related to the interpersonal justice factor (factor 8; r =. 52). The anger and interpersonal justice factors were negatively related (factors 8 and 9; r = - .50). 194 Appendix Q: Factor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study Predictor Items Table 35 Fa ctor Correlation Matrix Based on Promax Rotated Maximum Likelihood Extraction Factor Analysis of Main Study Predictor Items Factor Factor 1 2 3 4 1 - 2 .62 - 3 - .10 - .3 0 - 4 .38 .4 3 - .2 3 - Note. Pairwise deletion was applied, item N s =399 - 453. This table shows the factor correlations. The highest correlation emerged between the LMXSC and self - attributed locus of causality for RBS factors (factors 1 and 2 ; r =.62), indicating that participants who perceived they have a relatively better q uality relationship with the supervisor than their peers were also more likely to view their relationship quality with their supervisor as driven by their own actions. Self - attributed locus of causality was also positively related to interpersonal justice (factors 2 and 4, r =.43), indicating those who perceive interactions with the supervisor as considerate also tended to think their actions and characteristics heavily influenced their relationship quality with the supervisor. 195 Appendix R : Descriptive Stat istics of All Main Study Items by Race/Ethnicity Table 36 Descriptive Statistics of All Main Study Items by Race/Ethnicity White Black Asian - PI Multiracial Hispanic Total N=340/290 N=30/26 N=38/33 N=16/15 N=25/22 N=449/386 M SD M SD M SD M SD M SD M SD LMXSC 4.09 1.40 4.01 1.17 4.09 1.08 4.20 1.44 4.41 1.21 4.11 1.35 LoC for RBS Self 3.49 .76 3.62 .80 3.50 .57 3.48 1.05 3.41 .74 3.49 .76 LoC for RBS Supervisor 3.38 .69 3.54 .78 3.44 .63 3.71 .66 3.33 .78 3.40 .70 LMX 3.59 .85 3.51 .85 3.36 .71 3.39 .93 3.81 .73 3.57 .83 Interpersonal J. 4.32 .77 4.41 .66 4.27 .64 4.35 .68 4.33 .83 4.32 .75 Informational J. 3.86 .79 4.02 .89 3.86 .82 3.97 .62 3.91 .79 3.88 .79 Pride 4.00 1.03 3.93 1.12 3.96 .83 4.10 .73 4.15 1.08 4.00 1.01 Guilt 1.49 .81 1.52 .78 1.36 .71 1.51 .82 1.55 .88 1.49 .80 Shame 1.37 .74 1.28 .49 1.24 .62 1.46 1.04 1.60 1.09 1.37 .76 Gratitude 3.36 1.20 3.59 1.15 3.19 1.06 3.95 1.05 3.51 1.13 3.40 1.18 Anger 1.49 .94 1.27 .66 1.22 .70 1.42 1.05 1.74 1.29 1.46 .93 Positive Affect 3.00 .94 3.12 1.09 2.78 .84 3.23 .82 3.15 .82 3.01 .93 Negative Affect 1.41 .61 1.28 .53 1.28 .46 1.68 1.04 1.50 .87 1.41 .64 Note. N listed first reflects the N for measures used in the first survey (LMXSC - LMX), N listed second reflects the N for measures used in the second survey (Interpersonal Justice - Negative Affect). Asian - PI= Asian and Pacific Islander. LMXSC was rated on a 7 - poin t Likert scale; all other scales were rated on a 5 - point Likert scale . LMXSC was rated on a 7 - point Likert scale; all other scales were rated on a 5 - point Likert scale . No mean differences were found across groups. 196 Appendix S : Descriptive Statistics of All Main Study Items by Workgroup Size Table 37 Descriptive Statistics of All Main Study Items by Workgroup Size 1 - 3 Peers N = 44 - 51 4 - 10 Peers N =180 - 208 10 - 20 Peers N =99 - 116 20 - 60 Peers N =52 - 61 60 or more Peers N =14 - 17 Mean differences LMXSC M 4.34 4.21 3.87 4.14 3.94 F (4,447)= 1.63, p =.17 SD 1.27 1.35 1.32 1.34 1.68 Self - attributed Locus of Causality for RBS M 3.42 3.55 3.47 3.51 3.15 F (4,448 )=1.36, p =.25 SD .79 .72 .79 .73 .99 Supervisor - attributed Locus of Causality for RBS M 3.30 3.44 3.40 3.30 3.63 F (4,448 )=1.23, p =.30 SD .69 .70 .71 .66 .69 LMX M 3.64 3.63 3.46 3.62 3.14 F (4,448 )=2.07, p =.08 SD .82 .82 .86 .77 1.07 Interpersonal Justice M 4.42 4.31 4.29 4.41 4.04 F ( 4, 384)= .98, p =.42 SD .81 .73 .80 .68 .63 Informational Justice M 3.82 3.97 3.74 3.95 3.56 F ( 4,384)=2.12, p =.08 SD .77 .78 .78 .74 1.01 Pride M 4.07 4.08 3.88 3.96 3.84 F ( 4, 384)=.83, p =.51 SD 1.01 .94 1.07 1.07 1.07 Guilt M 1.40 1.41 1.61 1.45 2.10 F ( 4,383)=3.30, p =.01 SD .72 .71 .95 .66 1.16 Shame M 1.27 1.35 1.51 1.18 1.68 F ( 4,384)=2.49, p =.04 SD .54 .73 .94 .35 1.07 Gratitude M 3.46 3.45 3.34 3.37 3.00 F ( 4,384)=.60, p =.67 SD 1.20 1.11 1.21 1.26 1.43 197 Anger M 1.34 1.45 1.55 1.33 2.03 F ( 4,384)=2.00, p =.09 SD .77 .91 1.04 .73 1.31 PA M 2.92 3.11 2.94 2.89 2.99 F ( 4,384)=.98, p =.42 SD .92 .93 .93 .87 1.17 NA M 1.30 1.39 1.53 1.29 1.52 F ( 4,384)=1.90, p =.11 SD .42 .62 .81 .39 .61 198 Appendix T : Descriptive Statistics of All Main Study Items by Supervisor Relationship Tenure Table 38 Descriptive Statistics of All Main Study Items by Supervisor Relationship Tenure 0 - 6 mo. N = 36 - 43 6 - 12 mo. N = 42 - 47 1 - 2 years N = 71 - 85 2 - 5 years N = 140 - 162 5+ years N = 73 - 83 Mean differences LMXSC M 3.72 3.80 4.07 4.30 4.33 F ( 4,415)=2.82, p =.03 SD 1.18 1.49 1.19 1.35 1.48 Self - attributed Locus of Causality for RBS M 3.34 3.35 3.51 3.57 3.52 F ( 4,416)=1.32 , p =.26 SD .85 .98 .65 .71 .76 Supervisor - attributed Locus of Causality for RBS M 3.38 3.60 3.36 3.39 3.40 F ( 4,416)=.98, p =.42 SD .71 .72 .60 .75 .69 LMX M 3.33 3.33 3.42 3.63 3.87 F ( 4,416)=5.63, p <.01 SD .95 1.02 .75 .79 .77 Interpersonal Justice M 4.55 4.08 4.20 4.37 4.40 F ( 4,357)=2.79, p =.03 SD .54 1.02 .68 .71 .71 Informational Justice M 4.03 3.78 3.78 3.88 3.95 F ( 4,357)=.95, p =.44 SD .70 .97 .78 .74 .77 Pride M 4.08 3.91 3.79 4.04 4.15 F ( 4,357)=1.36, p =.25 SD .90 1.19 1.03 1.00 .95 Guilt M 1.33 1.62 1.51 1.50 1.43 F ( 4,356)=.72, p =.58 SD .64 .87 .83 .81 .79 Shame M 1.22 1.52 1.37 1.35 1.38 F ( 4,357)=.84, p =.50 SD .40 .86 .70 .70 .92 Gratitude M 3.45 3.48 3.04 3.46 3.68 F ( 4,357)=2.94, p =.02 SD 1.12 1.32 1.20 1.15 1.06 Anger M 1.26 1.62 1.53 1.45 1.40 F ( 4,357)=.90, p =.46 SD .56 1.19 .94 .92 .85 PA M 2.92 3.08 2.78 3.06 3.22 F ( 4,357)=2.23, p =.07 SD .83 1.00 .83 1.00 .87 NA M 1.23 1.53 1.45 1.42 1.34 F ( 4,357)=1.45, p =.22 SD .30 .76 .56 .66 .63 199 REFERENCES 200 REFERENCES Ambrose, M. L., & Schminke, M. (2003). Organization structure as a moderator of the relationship between procedural justice, interactional justice, perceived organizational support, and supervisory trust. Journal of Applied Psychology , 88 (2), 295. Anand, S. (2012). Multi - level Examination of Idiosyncratic Deals: Antecedents and Consequences . University of Illinois at Chicago. Aquino, K., Lewis, M. & Bradfield, M. (1999). Justice Constructs, Negative Affectivity, and Employee Deviance: A Proposed Model and Empirical Test. Journal of Organizational Behavior, 20: 1073 - 91. Baldwin, M. W., & Baccus, J. R. (2004). Maintaining a focus on the social goals underlying self - conscious emotions. Psychological Inquiry , 139 - 144. Bandalos, D. L. (2002). The effects of item parceling on goodness - of - fit and parameter estimate bias in structural equation modeling. Structural Equation Modeling , 9 (1), 78 - 102. Barclay, L. J., Skarlicki, D. P., & Pugh, S. D. (2005). Exploring the Role of Emotions in Injustice Perceptions and Retaliation. Journal of Applied Psychology , 90(4), 629 - 643. Bies, R. J., & Moag, J. S. (1986), Interactional Justice: Communication Criteria of Fairness. In Research on Negotiation in Organizations , vol.1, Greenwich, CT: JAI Press, 43 - 55. Boies, K., & Howell, J. M. (2006). Leader member exchange in teams: An examination of the interaction between relationship differentiation and mean LMX in explaining team - level outcomes. The Leadership Quarterly , 17 (3), 246 - 257. Bo lino, M. C., & Turnley, W. H. (2009). Relative deprivation among employees in lower - quality leader - member exchange relationships. The Leadership Quarterly , 20 (3), 276 - 286. Boswell, W. R., Shipp, A. J., Payne, S. C., & Culbertson, S. S. (2009). Changes in newcomer job satisfaction over time: examining the pattern of honeymoons and hangovers. Journal of Applied Psychology , 94 (4), 844. Brockner, J., & Wiesenfeld, B. M. (1996). An integrative framework for explaining reactions to decisions: interactive effect s of outcomes and procedures. Psychological Bulletin , 120 (2), 189. Brown, D. J., Ferris, D. L., Heller, D., & Keeping, L. M. (2007). Antecedents and consequences of the frequency of upward and downward social comparisons at work. Organizational Behavior a nd Human Decision Processes , 102 (1), 59 - 75. 201 Brown, J. D., & Marshall, M. A. (2001). Self - esteem and emotion: Some thoughts about feelings. Personality and Social Psychology Bulletin , 27 (5), 575 - 584. Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's Mechanical Turk a new source of inexpensive, yet high - quality, data?. Perspectives on Psychological Science , 6 (1), 3 - 5. Burton, J. P., Taylor, S. G., & Barber, L. K. (2014). Understanding internal, external, and relational attributions for abusive supervision. Journal of Organizational Behavior . Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, narcissism, self - esteem, and direct and displaced aggression: Does self - love or self - hate lead to violence?. Journal of personality and social psychology , 75 (1), 219. Bushman, B. J., & Whitaker, J. L. (2010). Like a Magnet Catharsis Beliefs Attract Angry People to Violent Video Games. Psychological Science , 21 (6), 790 - 792. Buss, A. H., & Durkee, A. (1957). An inventor y for assessing different kinds of hostility. Journal of Consulting Psychology , 21(4), 343. Buunk, B. P., Collins, R. L., Taylor, S. E., VanYperen, N. W., & Dakof, G. A. (1990). The affective consequences of social comparison: either direction has its ups and downs. Journal of Personality and Social Psychology , 59 , 1238 - 1249. Chen, Y., Yu, E., & Son, J. (2014). Beyond leader member exchange (LMX) differentiation: An indigenous approach to leader member relationship differentiation. The Leadership Quarterl y , 25 (3), 611 - 627. Choi, D. (2013). Differentiated leader - member exchange and group effectiveness: a dual perspective. University of Iowa. Colquitt, J. A. (2001). On the dimensionality of organizational justice: a construct validation of a measure. Journ al of Applied Psychology , 86 (3), 386. Cook, D. R. (1989). Internalized Shame Scale (ISS). Menomonie, WI: University of Wisconsin Stout. Coulter, R. H., & Pinto, M. B. (1995). Guilt appeals in advertising: what are their effects?. Journal of Applied Psych ology, 80 , 697 - 705. Crosby, F. (1976). A model of egoistical relative deprivation. Psychological Review , 83 (2), 85. Dancy, C. P., & Reidy, J. (2002). Statistics without maths for psychology: Using SPSS for Windows. Pearson: Prentice Hall. 202 Dansereau Jr, F., Graen, G., & Haga, W. J. (1975). A vertical dyad linkage approach to leadership within formal organizations: A longitudinal investigation of the role making process. Organizational Behavior and Human Performance , 13 (1), 46 - 78. Dulebohn, J. H., Bommer , W. H., Liden, R. C., Brouer, R. L., & Ferris, G. R. (2012). A meta - analysis of antecedents and consequences of leader - member exchange integrating the past with an eye toward the future. Journal of Management , 38 (6), 1715 - 1759. Dykema, J., Bergbower, K., Doctora, J. D., & Peterson, C. (1996). An Attributional Style Questionnaire for General Use. Journal of Psychoeducational Assessment , 14 (2), 100 - 108. Emmons, R. A., & Crumpler, C. A. (2000). Gratitude as a human strength: Appraising the evidence. Journal of Social and Clinical Psychology , 19 (1), 56 - 69. Emmons, R. A., & McCullough, M. E. (2003). Counting blessings versus burdens: an experimental investigation of gratitude and subjective well - being in daily life. Journal of Personality and Social Psycholog y , 84 (2), 377. Erdogan, B., & Bauer, T. N. (2010). Differentiated leader member exchanges: The buffering role of justice climate. Journal of Applied Psychology , 95 (6), 1104. Erdogan, B. (2002). Leader - member exchange differentiation fairness: Evidence for a new construct. ProQuest Dissertations & Theses. Festinger, L. (1954). A theory of social comparison processes. Human relations , 7 (2), 117 - 140. Firebaugh, G. (1980). Groups as contexts and fr og ponds. In K. H. Roberts & L. Burstein (Eds.), Issues in aggregation (pp. 43 52). San Francisco: Jossey - Bass. Ford, L. R., & Seers, A. (2006). Relational leadership and team climates: Pitting differentiation versus agreement. The Leadership Quarterly , 17 (3), 258 - 270. Friedman, R., Anderson, C., Brett, J., Olekalns, M., Goates, N., & Lisco, C. C. (2004). The positive and negative effects of anger on dispute resolution: evidence from electronically mediated disputes. Journal of Applied Psychology , 89, 36 9. Fuqua, D. R., Leonard, E., Masters, M. A., Smith, R. J., Campbell, J. L., & Fischer, P.C. (1991). A structural analysis of the state - trait anger expression inventory. Educational and Psychological Measurement, 51 (2), 439. George, J. A. (1999). The impact of team leader behavior on team effectiveness: A leader - member exchange theoretic perspective. Dissertation Abstracts International: Section B: The Sciences and Engineering , 59 (9 - B). 203 Gerrard, M., Gibbons, F. X., & McCoy, S. B. (1993). Emotional inh ibition of effective contraception. Anxiety, Stress and Coping , 6 (2), 73 - 88. Gerstner, C. R., & Day, D. V. (1997). Meta - Analytic review of leader member exchange theory: Correlates and construct issues. Journal of applied psychology , 82(6), 827. Gibson, D. E., & Callister, R. R. (2010). Anger in organizations: Review and integration. Journal of Management , 36 (1), 66 - 93. Gordon, A. M., Impett, E. A., Kogan, A., Oveis, C., & Keltner, D. (2012). To have and to hold: gratitude promotes relationship maintenan ce in intimate bonds. Journal of Personality and Social Psychology, 103, 257 - 274. Graen, G., & Cashman, J. (1975). A role making model of leadership in formal organizations: A developmental approach. In J. G. Hunt & L. L. Larson (Eds.), Leadership Frontiers (pp. 143 - 165). Graen, G. B., & Scandura, T. A. (1987). Toward a psychology of dyadic organizing. In L. L. Cummings & B. M. Staw (Eds.), Research In Organizational Behavior (Vol. 9, pp. 175 - 208). JAI Press. Graen, G., & Uhl - Bien, M. (1995). Rela tionship - based approach to leadership: Development of leader - member exchange (LMX) theory of leadership over 25 years. Leadership Quarterly , 6 (2), 219 - 247. Greenberg, J. (1993). The social side of fairness: Interpersonal and informational classes of organ izational justice. Justice in the workplace: Approaching fairness in human resource management. (pp. 79 - 103) Lawrence Erlbaum Associates, Inc, Hillsdale, NJ. Greenberg, J., Ashton - James, C. E., & Ashkanasy, N. M. (2007). Social comparison processes in org anizations. Organizational Behavior and Human Decision Processes , 102 , 22 - 41. Han, J. H. (2014). Admiration and envy as an impetus: Joint effects of leader - member exchange differentiation and group incentive pay on group affective climates, coordination, and performance. University of Maryland, College Park. Han, H. G., & Bai, Y. (2012). In need of each other: the moderator of task interdependence between LMX variability and justice. Journal of Nursing Management . Harris, T. B., Li, N., & Kirkman, B. L. (2014). Leader member exchange (LMX) in context: How LMX differentiation and LMX relational separation attenuate LMX's influence on OCB and turnover intention. The Leadership Quarterly , 25 (2), 314 - 328. Harrison, D. A., & Klein, K. J. (2007). What's the d ifference? Diversity constructs as separation, variety, or disparity in organizations. Academy of Management Review , 32 (4), 1199 - 1228. 204 Haynie, J. J., Cullen, K. L., Lester, H. F., Winter, J., & Svyantek, D. J. (2014). Differentiated leader member exchange , justice climate, and performance: Main and interactive effects. The Leadership Quarterly , 25 (5), 912 - 922. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Henderson, D. J., Liden, R. C., Glibkowski, B. C., & Chaudhry, A. (2009). LMX differentiation: A multilevel review and examination of its antecedents and outcomes. The leadership quarterly , 20 (4), 517 - 534. Henderson, D. J., Wayne, S. J., Shore, L. M., Bommer, W. H., & Tetrick, L. E. (2008). Leader -- member exchange, diff erentiation, and psychological contract fulfillment: a multilevel examination. Journal of Applied Psychology , 93 (6), 1208. Hooper, D. T., & Martin, R. (2008). Beyond personal leader member exchange (LMX) quality: The effects of perceived LMX variability o n employee reactions. The Leadership Quarterly , 19 (1), 20 - 30. Hooper, D., & Martin, R. (2006). Beyond personal LMX quality: The effects of perceived LMX variability on employee reactions. In A. I. Glendon, B. Myors & B. M. Thompson (Eds.), Advances in organizational psychology: An Asia - Pacific perspective. Brisbane: Australian Academic Press. House, R. J., Hanges, P. J., Javidan, M., Dorfman, P. W., & Gupta, V. (2004). Culture, leadership, and organizations. The GLOBE study of 62 societies. Thousand Oa ks, CA: Sage Publications. Judge, T. A., & Bono, J. E. (2001). Relationship of core self - evaluations traits self - esteem, generalized self - efficacy, locus of control, and emotional stability with job satisfaction and job performance: A meta - analysis. Journ al of Applied Psychology , 86 (1), 80. Kelley, H. K. (1968). Two functions of reference groups. In H. H. Hyman & E. Singer (Eds.), Readings in reference group theory and research (pp. 77 83). New York: Free Press. (Reprinted from Readings in social psycholo gy, pp. 410 414, by G. E. Swanson, T. M. Newcomb, & E. L. Hartley, Eds., 1952, New York: Holt, Rinehart, & Winston). Kelley, H. H., & Michela, J. L. (1980). Attribution theory and research. Annual review of psychology , 31 (1), 457 - 501. Keltner, D., & Lern er, J. S. (2010). Emotion. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of Social Psychology (pp. 317 - 352). Klass, E. T. (1987). Situational approach to assessment of guilt: Development and validation of a self - report measure. Journal of Psychopathology and Behavioral Assessment , 9 , 35 - 48. 205 Kline, R. B. (2004). Principles and practice of structural equation modeling, 2nd ed. New York: Guildford Krehbiel, P. J., & Cropanzano, R. (2000). Procedural Justice, Outcome Favorability and Emotion. Social Justice Research , 13 (4), 339 - 360. Kugler, K., & Jones, W. H. (1992). On conceptualizing and assessing guilt. Journal of Personality and Social Psychology , 62(2), 318. Kwak, W. J. (2011). Multi - level investigation of empowering leadership, leader - member exchange, and subordinate empowerment. Purdue University. Landers, R. N., Behrend, T. S. (2014). An inconvenient truth: Arbitrary distinctions between organizational, mechanical turk, and other convenience samples. Industrial and Organizational Ps ychology , 8 (2). Lau, R. S. Y. (2008). Integration and Extension of Leader - Member Exchange and Organizational Justice at Individual - and Group - Levels of Analysis . Virginia Polytechnic Institute and State University. Le Blanc, P. M., & González - Romá, V. (2012). A team level investigation of the relationship between Leader Member Exchange (LMX) differentiation, and commitment and performance. The Leadership Quarterly , 23 (3), 534 - 544. Lewis, H. B. (1971). Shame and Guilt in Neurosis. Psy choanalytic Review , 58 (3), 419 - 438. Li, A. N., & Liao, H. (2014). How do leader member exchange quality and differentiation affect performance in teams? An integrated multilevel dual process model. Journal of Applied Psychology , 99 (5), 847. Liao, H., Liu, D., & Loi, R. (2010). Looking at both sides of the social exchange coin: A social cognitive perspective on the joint effects of relationship quality and differentiation on creativity. Academy of Management Journal , 53 (5), 1090 - 1109. Liden, R. C., Erdogan , B., Wayne, S. J., & Sparrowe, R. T. (2006). Leader member exchange, differentiation, and task interdependence: implications for individual and group performance. Journal of Organizational Behavior , 27 (6), 723 - 746. Liden, R. C., & Maslyn, J. M. (1998). M ultidimensionafity of leader - member exchange: an empirical assessment through scale development. Journal of management , 24 (1), 43 - 72. Liden, R. C., Sparrowe, R. T., & Wayne, S. J. (1997). Leader - Member Exchange Theory: The Past and Potential for the Futur e. Research in Personnel and Human Resources Management , 15 , 47 - 119. 206 Lind, E. A., & Tyler, T. R. (1988). Two Models of Procedural Justice. In The Social Psychology of Procedural Justice (pp. 221 - 242). Springer US. Ma, L., & Qu, Q. (2010). Differentiatio n in leader member exchange: A hierarchical linear modeling approach. The Leadership Quarterly , 21 (5), 733 - 744. Marschall, D. E., Sanftner, J., & Tangney, J. P. (1994). The State Shame and Guilt Scale. George Mason University, Fairfax, VA. Mascolo, M. F., & Fischer, K. W. (1995). Developmental transformations in appraisals for pride, shame, and guilt. Self - conscious emotions: The psychology of shame, guilt, embarrassment, and pride. (pp. 64 - 113). Maslyn, J. M., & Uhl - Bien, M. (2001). Leader member exchange and its dimensions: Effects of self - effort and other's effort on relationship quality. Journal of Applied Psychology , 86 (4), 697. cal Turk. Behavior research methods , 44 (1), 1 - 23. Mayer, D. M. (2005). Are you in or out? A group - level examination of the effects of LMX on justice and customer satisfaction. University of Maryland, College Park. McAuley, E., Duncan, T. E., & Russell, D . W. (1992). Measuring causal attributions: The revised causal dimension scale (CDSII). Personality and Social Psychology Bulletin , 18 (5), 566 - 573. McClane, W. E. (1991). Implications of Member Role Differentiation Analysis of a Key Concept in the LMX Mod el of Leadership. Group & Organization Management , 16 (1), 102 - 113. McCullough, M. E., Tsang, J. A., & Emmons, R. A. (2004). Gratitude in intermediate affective terrain: links of grateful moods to individual differences and daily emotional experience. Jour nal of Personality and Social Psychology , 86 (2), 295. Mosher, D. L. (1966). The development and multitrait - multimethod matrix analysis of three measures of three aspects of guilt. Journal of Consulting Psychology , 30(1), 25. Naidoo, L. J., Scherbaum, C. A., Goldstein, H. W., & Graen, G. B. (2011). A longitudinal examination of the effects of LMX, ability, and differentiation on team performance. Journal of Business and Psychology , 26 (3), 347 - 357. Niehoff, B. P., & Moorman, R. H. (1993). Justice as a Mediator of the Relationship between Methods of Monitoring and Organizational Citizenship Behavior. Academy of Management Journal , 527 - 556. 207 Nishii, L. H., & Mayer, D. M. (2009). Do inclusive leaders help to reduce turnover in dive rse groups? The moderating role of leader member exchange in the diversity to turnover relationship. Journal of Applied Psychology , 94 (6), 1412. Otterbacher, J. R., & Munz, D. C. (1973). State - trait measure of experiential guilt. Journal of Consulting and Clinical Psychology, 40, 115. 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 (5), 879. Pretzer, J., Epstein, N., & Fleming, B. (1991). Marital Attitude Survey: A measure of dysfunctional attributions and expectancies. Journal of Cognitive Psychotherapy , 5 (2), 131 - 148. Rotter, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs: General and Applied , 80 , 1 - 28. Rupp, D. E., McCance, A. S., Spencer, S., & Sonntag, K. (2008). Customer (In) Justice and Emotio nal Labor: The Role of Perspective Taking, Anger, and Emotional Journal of Management , 34 (5), 903 - 924. Rupp, D. E., & Spencer, S. (2006). When customers lash out: the effects of customer interactional injustice on emotional labor and the med iating role of discrete emotions. Journal of Applied Psychology , 91(4), 971. Russell, J. A. (2003). Core Affect and the Psychological Construction of Emotion. Psychological Review , 110 , 145 - 172. Shoss, M. K., Eisenberger, R., Restubog, S. L. D., & Zagenc zyk, T. J. (2013). Blaming the organization for abusive supervision: The roles of perceived organizational support and supervisor's organizational embodiment. Journal of Applied Psychology , 98 (1), 158. Smith, H. J., Pettigrew, T. F., Pippin, G. M., & Bial osiewicz, S. (2012). Relative Deprivation A Theoretical and Meta - Analytic Review. Personality and Social Psychology Review , 16 (3), 203 - 232. Soper, D.S. (2015). A - priori Sample Size Calculator for Multiple Regression [Software]. Available from http://www.d anielsoper.com/statcalc Spielberger, C. D. (1988). State Trait Anger Expression Inventory . John Wiley & Sons, Inc.. Stewart, M. M., & Johnson, O. E. (2009). Leader Member exchange as a moderator of the relationship between work group diversity and team p erformance. Group & Organization Management , 34 (5), 507 - 535. 208 Stoeber, J., Hutchfield, J., & Wood, K. V. (2008). Perfectionism, self - efficacy, and aspiration level: Differential effects of perfectionistic striving and self - criticism after success and failu re. Personality and Individual Differences , 45(4), 323 - 327. Tangney, J. P., & Dearing, R. L. (2003). Shame and guilt . Guilford Press. Tangney, J. P., Miller, R. S., Flicker, L., & Barlow, D. H. (1996). Are shame, guilt, and embarrassment distinct emotions ?. Journal of Personality and Social Psychology , 70 (6), 1256. Tay, C., Ang, S., & Van Dyne, L. (2006). Personality, biographical characteristics, and job interview success: a longitudinal study of the mediating effects of interviewing self - efficacy and th e moderating effects of internal locus of causality. Journal of Applied Psychology , 91 (2), 446. Tordera, N., & González - Romá, V. (2013). Leader - Member Exchange (LMX) and Innovation Climate: the Role of LMX Differentiation. The Spanish Journal of Psycholog y , 16 , E83. Tracy, J. L., & Robins, R. W. (2006). Appraisal antecedents of shame and guilt: Support for a theoretical model. Personality and Social Psychology Bulletin , 32 (10), 1339 - 1351. Tracy, J. L., & Robins, R. W. (2004). Putting the Self Into Self - Conscious Emotions: A Theoretical Model. Psychological Inquiry , 15 (2), 103 - 125. Tyler, T. R. (2006). Why people obey the law . Princeton University Press. Tyler, T. R., & Blader, S. L. (2003). The group en gagement model: Procedural justice, social identity, and cooperative behavior. Personality and Social Psychology Review, 7, 349 - 361 Vidyarthi, P. R., Liden, R. C., Anand, S., Erdogan, B., & Ghosh, S. (2010). Where do I stand? Examining the effects of lead er member exchange social comparison on employee work behaviors. Journal of Applied Psychology , 95 (5), 849. Watkins, P. C., Woodward, K., Stone, T., & Kolts, R. (2003). Gratitude and happiness: Development of a measure of gratitude and relationships with subjective well - being . Social Behavior and Personality , 31 , 431 - 452. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scale. Journal of Personality and Social Psycholo gy, 54 , 1063 - 1070. Webster, J. M., Duvall, J., Gaines, L. M., & Smith, R. H. (2003). The roles of praise and social comparison information in the experience of pride. The Journal of Social Psychology , 143 (2), 209 - 232. 209 Weiner, B. (2000). Intrapersonal and Interpersonal Theories of Motivation from an Attributional Perspective. Educational Psychology Review , 12 (1), 1 - 14. Weiner, B. (1985). An Attributional Theory of Achievement Motivation and Emotion. Psychological Review , 92 (4), 548 - 573. Weiner, B., Friez e, I. H., Kukla, A., Reed, L., Rest, S., & Rosenbaum, R. M. (1971). Perceiving the causes of success and failure . General Learning Press. Weiss, H. M., Suckow, K., & Cropanzano, R. (1999). Effects of justice conditions on discrete emotions. Journal of Ap plied Psychology , 84 (5), 786. Williams, L. A., & DeSteno, D. (2008). Pride and perseverance: the motivational role of pride. Journal of personality and social psychology, 94(6), 1007. Willis, S. (2014). Toward a model of diversity leadership: Examining creating an inclusive workplace. The Claremont Graduate University. Wood, J. V. (1996). What is social comparison and how should we study it?. Personality and Social Psychology Bulletin , 22 (5), 520 - 537. Wood, J. V. (1989). Theory and research concerning social comparisons of personal attributes. Psychological Bulletin , 106 (2), 231.