UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mood, motivation, and task me Zerbe, Wilfred Joachim 1987

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1988_A1 Z47.pdf [ 13.18MB ]
Metadata
JSON: 831-1.0098296.json
JSON-LD: 831-1.0098296-ld.json
RDF/XML (Pretty): 831-1.0098296-rdf.xml
RDF/JSON: 831-1.0098296-rdf.json
Turtle: 831-1.0098296-turtle.txt
N-Triples: 831-1.0098296-rdf-ntriples.txt
Original Record: 831-1.0098296-source.json
Full Text
831-1.0098296-fulltext.txt
Citation
831-1.0098296.ris

Full Text

MOOD, MOTIVATION, AND TASK MEMORY  by  WILFRED J . ZERBE B.A.  1978, U n i v e r s i t y Of B r i t i s h  Columbia  M.A.  1982, U n i v e r s i t y Of B r i t i s h  Columbia  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE  REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY  in THE  FACULTY OF GRADUATE STUDIES  Department Of Commerce And B u s i n e s s  Administration  We a c c e p t t h i s t h e s i s a s conforming to t h e r e q u i r e d  THE  standard  UNIVERSITY OF BRITISH COLUMBIA November 1987  ©  Wilfred J .  Zerbe, 1987  In  presenting  degree freely  at  this  the  available  copying  of  department publication  of  in  partial  fulfilment  University  of  British  Columbia,  for  this or  thesis  reference  thesis by  this  for  his thesis  and  study.  scholarly  or for  her  of I  I further  purposes  gain  shall  requirements  agree  that  agree  may  representatives.  financial  the  It not  be is  that  of  D a t e  DE-6(3/81)  March 4, 1988  by  understood  be  Commerce and Business Administration  The University of British C o l u m b i a 1956 Main Mall Vancouver, Canada V6T 1Y3  Library  an  allowed  advanced  shall  permission for  granted  permission.  Department  the  for  the that  without  make  it  extensive  head  of  copying my  my or  written  ii  Abstract Theorists i n organizational behavior have generally ignored determinants of behavior.  A task of t h i s d i s s e r t a t i o n was  emotional  to extend the  use  of emotions for understanding organizational behavior i n general and work motivation Two  in particular.  theories, expectancy theory and network theory, are used to make  predictions about the relationship between mood and perceptions relationship between e f f o r t and performance.  of the  According to expectancy theory,  the e f f o r t that people choose to expend at tasks i s a function of their b e l i e f about the degree to which e f f o r t and performance covary.  Network theory  predicts that memories are connected by a network of associations.  The  a c c e s s i b i l i t y for r e c a l l of a memory i s a function of the a c t i v a t i o n of these associations.  In t h i s way  p o s i t i v e events are more accessible for r e c a l l when  individuals are i n a p o s i t i v e mood state because of associations based on a f f e c t i v e valence of memories.  Such a c c e s s i b i l i t y of events for r e c a l l  been shown to be a determinant of p r o b a b i l i t y judgements.  the  has  On t h i s basis i t  was predicted that mood would bias i n d i v i d u a l s ' judgements of the p r o b a b i l i t y that s p e c i f i c l e v e l s of e f f o r t lead to s p e c i f i c levels of performance. other words, that mood a f f e c t s expectancy.  S p e c i f i c a l l y , i t was  In  predicted  that individuals i n an elated mood would report higher expectancy than individuals i n a depressed mood. f e e l i n g state. how  Two  Mood was defined as a self-evaluative  other hypotheses were formed: that mood would influence  cause for behavior i s attributed, and that i n d i v i d u a l differences i n self  esteem would moderate the r e l a t i o n s h i p between mood and expectancy. Three studies were performed to provide a foundation for the testing of these hypotheses.  In a fourth study they were tested.  Study One  assessed the  psychometric properties of measures of mood states, i n d i v i d u a l differences,  iii  and task perceptions. Study Two concerned the experimental induction of mood. Mood manipulations used i n the experimental l i t e r a t u r e were reviewed and one, a musical procedure, was chosen.  The v a l i d i t y of t h i s manipulation was then  tested by having participants l i s t e n to the music of an elated, neutral, or depressed mood induction procedure.  The results of Study Two provided strong  evidence f o r the v a l i d i t y of the manipulation.  Both self-report measures of  mood and an unobtrusive behavioral measure were s i g n i f i c a n t l y a f f e c t e d . The results of Study Two also showed the u t i l i t y of a conceptualization of mood as comprising two components: arousal and pleasure.  I t was shown that depression  i s characterized by low arousal and displeasure, and elation by high arousal and pleasure. Study Three reviewed the conceptualization and measurement of expectancy. It was argued that expectancy  i s properly conceptualized as the perceived  covariation between e f f o r t and performance.  This requires measurement of the  r e l a t i o n s h i p between multiple levels of e f f o r t and multiple levels of performance and c a l c u l a t i o n from these measures of an index of perceived effort—performance covariation.  Most p r i o r measurement has only considered  the relationship between high e f f o r t and high performance.  Further, i t was  argued that such appropriate measurement allows predictions to be made about expectancy across individuals, i n contrast to the argument that  expectancy  theory i s a within-subjects theory. Previous authors have used such an approach to measure expectancy but have not demonstrated i t s v a l i d i t y .  Study Three undertook such v a l i d a t i o n .  Participants completed one of two experimental tasks: one with high objective expectancy,  the other with low objective expectancy.  As predicted, scores on  the perceived covariation measure of expectancy were s i g n i f i c a n t l y higher i n  iv  the high objective expectancy  task.  Measures of related constructs were  influenced i n a manner consistent with t h i s f i n d i n g .  It was concluded that  strong support for the expectancy measure existed. On the foundation of Studies One, Two, and Three, Study Four undertook to test the formal hypotheses of the d i s s e r t a t i o n .  In each of three experimental  sessions, participants completed a business decision-making  task, underwent  either an e l a t i o n , neutral mood, or depression induction procedure, and then completed measures of their mood state, expectancy,  and other task  perceptions. The r e s u l t s of Study Four indicated that s i g n i f i c a n t differences i n mood resulted from the manipulation. were supported.  However, none of the experimental hypotheses  Mood d i d not influence expectancy or task a t t r i b u t i o n s .  A  number of alternate explanations for t h i s finding were considered, including f a i l u r e of the mood manipulation, measurement error, and lack of s t a t i s t i c a l power.  Of these, i t was concluded that while Study Four lacked power to  detect a large e f f e c t , this d i d not f u l l y explain the f a i l u r e to support the experimental hypotheses.  Also compelling was the argument that the mood  manipulation was not s u f f i c i e n t l y powerful.  V  Table of Contents  Abstract L i s t Of Tables  i i .v  L i s t Of Figures  xii  Acknowledgement  xiii  I. II.  III.  INTRODUCTION  1  WHAT IS EMOTION?  5  Causes Of Emotion  6  Fundamental Emotions And The Evolutional Perspective  7  William James And The Primacy Of The Periphery  7  C e n t r a l i s t Theories  10  C o n f l i c t Theories  11  Structural Theories  12  Reconciling Theories Of Emotion  17  What Is Mood?  18  The E f f e c t s Of Mood On Thought And Behavior  21  Affect And Behavior  22  Mood And Thought  24  Network Theory  28  Mood And Arousal  30  Arousal And Memory  32  EMOTION AND ORGANIZATIONS  34  Emotion As Dependent Variable  35  Job S a t i s f a c t i o n As Emotion  35  Emotional Expression As Part Of The Work Role  37  Emotion As An Independent Variable  41  vi  IV.  The Expression Of Affect In Decision-making  42  Job S a t i s f a c t i o n As Mood  43  Mood And Performance Appraisal  45  AFFECT AND MOTIVATION Expectancy Theory  47  Measuring Expectancy  49  Mood And Expectancy  54  Research Hypotheses  V.  VI.  47  59  Hypothesis One: Mood And Expectancy  59  Hypothesis Two: Mood And Causal A t t r i b u t i o n s  63  Hypothesis Three: Self-esteem And Mood  64  Proposed Research  71  STUDY ONE: PSYCHOMETRIC PROPERTIES OF MEASURES  73  Overview  73  Subjects  73  Procedure  74  Measures And Results  75  Individual Differences: Self-esteem And Impression Management  75  Task Perceptions  76  Causal A t t r i b u t i o n s  82  Multiple Affect Adjective Checklist  85  Pleasure And Arousal  89  Dimensionality Of Mood Measures  89  P a r a l l e l Scale Construction  94  Correlations Between Measures  95  STUDY TWO: WHAT IS MOOD? Manipulation Of Mood  103 103  vii  Method Procedure Measures  .110 Ill  Adjective Checklist  Ill  Semantic D i f f e r e n t i a l  Ill  Response Latency  Ill  Results  VII.  110  113  Evaluation Of Assumptions  113  Multivariate Analyses Of Variance  116  Summary  122  Discussion: What Is Mood?  124  STUDY THREE: WHAT IS EXPECTANCY?  127  Conceptualizing And Measuring Expectancy  127  Validating Expectancy Measurement  132  Method  134  Subjects And Design  134  Manipulation Of Expectancy  135  Perceptual-motor Task  135  Cognitive Reasoning Task  135  Measures  137  Effort-performance Covariation  137  Control Over Performance  139  Perceived Correlation  140  Expected Performance  140  Task Perceptions And Causal A t t r i b u t i o n s Results Evaluation Of Assumptions  ...140 140 141  viii  Multivariate Analysis Of Variance Discussion VIII.  143 149  STUDY FOUR: MOOD AND EXPECTANCY  151  Overview  151  Experimental Design  151  Method  154  Subjects  154  Procedure  154  Decision-making  Task  157  Mood Manipulation  162  Dependent Measures  163  Sessions Two And Three  168  Results  169  Evaluation Of Assumptions  169  Analysis Of Manipulation Checks  172  Results Of Manipulation Checks  174  Measures Of Expectancy And Task Perceptions  184  Individual Differences Discussion: Alternate Explanations  ..190 192  Manipulation F a i l u r e  192  Measurement Error  197  Lack Of Experimental Control  198  Lack Of S t a s t i c a l Power  199  The Null Hypothesis Is True  204  Conclusion  206  APPENDIX A: Experimental Materials, Study One  209  APPENDIX B: Musical Selections  224  ix  APPENDIX C: Experimental Materials For Study Three  226  APPENDIX D: Consent Form And Questionnaire, Study Four  230  APPENDIX E: Verbal Expectancy Items, Study Four  237  APPENDIX F: C e l l Means, A l l Dependent Variables, Study Four  238  References  249  X  L i s t of Tables  Table 1.  Fundamental Or Primary Emotions Listed By Three Leading  Theorists (from Mandler, 1984, Table 2.  P.  36)  8  Emotional Reactions Resulting From Structural Evaluations,  According To Roseman (1984)  14  Table 3.  Items Assessing Task Perceptions, Grouped By Constuct  Table 4.  Factor Loadings, Communalities ( h ) , Percent Of Variance For 2  Factor Analysis On Task Perception Items Table 5.  78  ....81  Descriptive S t a t i s t i c s And R e l i a b i l i t y Estimates For Measures  Of Individual Differences, Task Perceptions, And Causal A t t r i b u t i o n s . 83 Table 6.  Items Used From MAACL Short Form, Study One  Table 7.  Means, Standard Deviations And R e l i a b i l i t y Estimates For  Measures Of Mood State, Study One Table 8.  91  Factor Loadings, Communalities For Maximum Likelihood Factor  Extraction On Semantic D i f f e r e n t i a l Items Table 10.  88  Factor Loadings, Communalities For Maximum Likelihood Factor  Extraction On MAACL Items Table 9.  87  93  Means, Standard Deviations And Internal Consistencies Of The  S p l i t - h a l f MAACL Scales, Study One  96  Table 11.  Correlations Between Dependent Variables, Study One  Table 12.  Correlations For Dependent Variables, Study Two  118  Table 13.  Univariate And Stepdown F-tests, Study Two  119  Table 14.  Means And Standard Deviations, Treatment And Gender Groups,  Study Two Table 15.  Univariate And Stepdown F-tests, Study Three.  99  120 145  xi  Table 16. Means And Standard Deviations For Dependent Variables, Study Two  146  Table 17.  Correlations Between Dependent Variables, Study Three  Table 18.  Univariate And Stepdown F-tests, Order By Session (Mood)  148  Effect On Manipulation Checks, Study Four Table 19.  175  Pooled Within-cell Correlations For Manipulation Checks, Study  Four  176  Table 20. Means And Standard Deviations For Manipulation Checks, By Mood E f f e c t , Study Four  ....178  Table 21. Univariate And Stepdown F-tests, Session Effect On Manipulation Checks, Study Four Table 22.  181  Means And Standard Deviations For Manipulation Checks, By  Session E f f e c t , Study Four  ".  182  Table 23. Means And Standard Deviations For Dependent Variables By Mood E f f e c t , Study Four Table 24.  186  Univariate And Stepdown F-tests On Dependent Variables,  Session E f f e c t , Study Four  187  Table 25. Means And Standard Deviations For Dependent Variables By Sessions, Study Four  >  Table 26. Within-cells Correlation Matrix, Study Four  188 189  xii  L i s t of Figures  Figure 1.  Structural Model Of Emotions, From Kemper  16  Figure 2.  Circumplex Model Of Emotions  31  Figure 3.  Contingency Table Model Of The Relationship Between E f f o r t And  Performance  61  Figure 4.  Causal A t t r i b u t i o n Measures, Study One  Figure 5.  Instructions For The MAACL  112  Figure 6.  Terminal Display For The Item "Active"  112  Figure 7.  Assessment Of Relationship Between E f f o r t And Performance,  Study Two Figure 8.  84  138  S p l i t - P l o t Design, Mood Treatment By Session And Order  Conditions  155  xiii  Acknowledgement  I owe my appreciation to the many people who completion.  Most central i s my  i n t e l l e c t and personal support  supervisor, Dr.  guided t h i s manuscript to Ralph Stablein, whose  I c a l l e d on often.  For shepherding me through  the research that i s reported here, as well as through my years i n the doctoral program at the University of B r i t i s h Columbia, I am most g r a t e f u l . I would also l i k e to thank the members of my E r i c Eich, Dr.  Peter Frost, and Dr.  supervisory committee, Dr.  Craig Pinder, for their honest c r i t i c i s m  and encouragement. A d i s s e r t a t i o n i s but one of the steps i n most doctoral programs. Appreciation and thanks are therefore due not only to those who t h i s manuscript to f r u i t i o n , but also to those who the work that came before i t .  helped bring  supported and encouraged  I would l i k e , therefore, to thank my fellow  students, and the f a c u l t y i n the I n d u s t r i a l Relations and Management Area at U.  B.  C.  F i n a l l y , I would l i k e to thank my parents, Edmund and Alma Zerbe, and most importantly, my wife, Rose-Marie Jaeger, for helping me, and supporting me i n my  education.  motivating  me,  1  I.  "Thinking,  INTRODUCTION  f e e l i n g , and acting."  For 150 years the idea that human  experience could be understood to include these three realms has dominated psychological thinking (Hilgard, 1980).  Some proponents of this t r i p a r t i t e  conception intended that every event be represented i n a l l three spheres (Hilgard, 1980;  Isen, 1984).  Thus, an event i n the thinking domain would  influence thinking, f e e l i n g and acting; an emotion or f e e l i n g would influence thinking, f e e l i n g and acting; an action would influence thinking, f e e l i n g and acting: i n t h i s conception each i s inexorably linked to the others.  Others,  the Faculty Psychologists of the 18th century, divided psychological experience into separate realms of cognition, emotion, and motivation.  Modern  theorists have rejected Faculty Psychology, but have nonetheless tended to segregate the influences of the three domains (Hilgard, 1980). we now  For example,  commonly t a l k about thinking without also t a l k i n g about f e e l i n g .  Thoughts and feelings are seen as quite separate.  This separation, and  emphasis on the study of cognition, has paved the way of emotion.  for the modern neglect  We regularly examine cognitive explanations  considering the r o l e of emotion.  an  of behavior without  We have, i n general, ignored emotional  determinants of behavior and the r e l a t i o n s h i p between f e e l i n g and  thinking.  Of late, though, we have been c a l l e d to attend to the r o l e of emotion. For example: What i s the r o l e of emotion i n cognition? We leave i t to the poet, the playwright, the n o v e l i s t . As people we delight i n art and music. We f i g h t , get angered, have joy, g r i e f , happiness. But as students of mental events we are ignorant of why, how? (Norman, 1980)  Emotions have received renewed attention i n the study of  organizations  2  and work settings.  Researchers and theorists have examined, among other-  issues, how work settings establish rules for emotional f e e l i n g and expression (Hochschild, 1983; Rafaeli St Sutton, 1987; Sutton & R a f a e l i , 1986); the effects of emotional expression on the performance of decision-making groups (Guzzo & Waters, 1982); the r o l e emotions play i n forming organizational culture and f a c i l i t a t i n g organizational development (Kahn, 1986; Van Maanen & Kunda, 1985; Weick, 1985); the effect of mood on performance appraisal (Park, Sims & Motowidlo, 1986; S i n c l a i r , i n press), and have attempted to use the construct of mood to shed l i g h t on the relationship between job s a t i s f a c t i o n and performance, p a r t i c u l a r l y of prosocial behaviors (Bateman & Organ, 1983; Brief & Motowidlo, 1986; Motowidlo, 1984). A task of the proposed d i s s e r t a t i o n i s to extend investigation of the implications of emotion for understanding organizational behavior. S p e c i f i c a l l y , the question at hand concerns how mood influences the processing of information about jobs, such as i n d i v i d u a l s ' expectations about the l i n k between e f f o r t and performance, perceptions about the l i k e l i h o o d that tasks w i l l be successfully accomplished, the a t t r i b u t i o n of r e s p o n s i b i l i t y for job performance, task s a t i s f a c t i o n , and so on.  The d i s s e r t a t i o n w i l l attempt to  show how feelings can influence such job-related cognitions.  It w i l l also  discuss, though not examine empirically, where and when emotions a r i s e i n organizational settings, and how emotions can and should be managed. Extensions of the work proposed w i l l also be discussed, including the influence of emotion on variables relevant to behavior i n organizations other than those at hand, and how a program of research building on the foundation of the d i s s e r t a t i o n might proceed. There i s growing acknowledgement of the a f f e c t i v e q u a l i t y of workplaces. In a survey of emotional experiences i n everyday l i f e , Scherer and Tannenbaum  3  (1986) found that work was second only to family as a source of emotions. Work was the largest source of p o s i t i v e experiences and the second largest source of negative experiences.  Such emotional experiences are l i k e l y to have  s i g n i f i c a n t implications f o r individuals i n organizations and for organizations themselves.  Yet we know l i t t l e about such impact.  What, f o r  example, i s the p r a c t i c a l impact of emotion on, say, performance?  Although  emotion i s often conceptualized as motivating, emotion i s absent i n most theories of work motivation.  However, i f we acknowledge the inherent  connections between thinking, f e e l i n g , and acting, i t i s evident that by studying emotion, we can add to what we know about motivation, to our a b i l i t y to predict e f f o r t , attention, and persistence i n job settings.  Such study of  emotions has t h e o r e t i c a l as well as p r a c t i c a l importance for organizations. Increasing our understanding of the determinants of organizational behavior i s i n t r i n s i c a l l y of value.  The goal of t h i s d i s s e r t a t i o n i s to increase our  understanding of the relationship between emotion and motivation.  The following chapters contain the foundation and j u s t i f i c a t i o n for the development of s p e c i f i c research hypotheses, hypotheses. issues.  and present a test of those  Chapter Two reviews theories of emotion and considers conceptual  Approaches which have examined the consequences of emotion are  distinguished from those seeking to establish i t s causes.  The p r i n c i p a l  theories of'the causes of emotion are reviewed and a d e f i n i t i o n of mood i s presented.  Theory and empirical findings about the effects of mood on thought  and behavior are reviewed. In Chapter Three the study of emotions i n organizations i s examined. Chapter Four defends the choice of motivation as an area i n which further investigation of the r o l e of mood should be productive and poses s p e c i f i c hypotheses r e l a t i n g mood to i n d i v i d u a l s ' perceptions of the relationship  4  between e f f o r t and performance. Chapters Five through Seven describe a series of studies which provide the foundation for the testing of these hypotheses. tested d i r e c t l y .  In Chapter Eight they are  In each of these chapters the implications of findings are  discussed. The d i s s e r t a t i o n that follows addresses the following questions: F i r s t , what i s mood?  To understand t h i s , i t i s h e l p f u l to examine approaches to the  study of emotion.  Having examined how emotions have been studied i n  organizational behavior, mood and motivation are chosen as the domain of i n t e r e s t , and the r e l a t i o n s h i p between mood and expectancy as the s p e c i f i c research question to be addressed. therefore, What i s expectancy? appropriately measured?  The second question addressed i s  How i s i t conceptualized and how i s i t  With the answers to these two questions, the ultimate  question i s addressed: Is there a r e l a t i o n s h i p between mood and expectancy?  5  II.  WHAT IS EMOTION?  What i s emotion? James  (1884)  Our understanding of mood i s c l o s e l y related to what we know about emotion.  To provide a background for a discussion of mood and i t s  consequences i n l a t e r sections, i n t h i s section theories of emotion are reviewed. On can, i n fact, evade e x p l i c i t d e f i n i t i o n of emotion by allowing people to  supply their own d e f i n i t i o n .  t e l l you what emotion " i s " .  Most people have such a d e f i n i t i o n , they can  Yet when you ask academic students of emotion  "what i s emotion?", you discover disagreement and lack of consensus. (1980)  lists  literature. list  28  Plutchik  d i f f e r e n t d e f i n i t i o n s of emotion i n the psychological  Kleinginna and Kleinginna  (1981)  list  92  definitions.  They also  9 skeptical statements (e.g., "There does not seem to be a s a t i s f a c t o r y  way to define emotion, aside from i t s manifestations i n act or verbal statements of f e e l i n g " , Cofer,  1972).  Arnold  (I960)  summarized  inquiry i n t o  emotion: " i t i s one of the most d i f f i c u l t and confused f i e l d s i n the whole of psychology" (pp.  10-11).  (Hillman, 1961, p.  There i s "a curious and overwhelming confusion"  5) i n the theory of emotion.  This confusion i s l i k e l y  because d i f f e r e n t theoretical perspectives have fostered d i f f e r e n t d e f i n i t i o n s and theoretical emphases.  Some writers emphasize expressive reactions, others  instrumental behavior, some i d e n t i f y emotions primarily as b i o l o g i c a l processes, others as social processes.  Some emphasize the r o l e of the central  nervous system, some the autonomic nervous system, some emphasize cognitive processes and others f e e l i n g states. How then should we proceed?  How can we wade into the sea of theories of  6  emotion without drowning or muddying the waters further?  One way i s by  recognizing that t h e o r e t i c a l disagreement i s i n part a r e s u l t of d i f f e r e n t approaches and t r a d i t i o n s , and i s not inherent to the topic.  Thus a f i r s t  step toward reconciling t h e o r e t i c a l differences and choosing an appropriate course i s to review these approaches and t r a d i t i o n s . The study of emotion can, for example, be simply divided into two approaches: emotions studied as dependent variables and as independent variables.  That i s , there are theories of emotion that t r y to specify the  determinants of emotion, what i t s causes or antecedents are, and there are theories of the consequences of emotion, what i t s effects are, and how i t influences thought and behavior. reviewed. emotion.  In t h i s chapter, these two approaches are  Presented f i r s t are the many and varied theories of the causes of Also presented i s an attempt at reconciling differences between  these theories.  F i n a l l y , mood i s defined and the effects of mood on thought  and behavior are reviewed.  Causes of Emotion Many writers have distinguished between traditions i n the study of emotion.  Fraisse (1968) described two "faces" of emotion.  The f i r s t ,  "mental" face sees physiological events as the consequences of psychic events. In contrast, the "organic" face views physiological events, rather than thoughts, as the precursors of emotion.  Hochschild (1983) makes a similar  d i s t i n c t i o n between " i n t e r a c t i o n i s t " and "organismic" approaches.  Others  (Mandler, 1984; Schachter & Singer, 1962) have distinguished between "central" theories concerned with central nervous system mechanisms, and "peripheral" theories concerned with peripheral reactions and autonomic nervous system responses.  Candland's (1977) system of description has three parts: the  7  b i o l o g i c a l , evolutional, and cognitive t r a d i t i o n s .  Fundamental Emotions and the Evolutional Perspective Berscheid  (1983) points out that emotion was  one of the f i r s t  psychological phenomena "to be removed from the arena of philosophical speculation" (p. was  119).  This introduction of emotion to s c i e n t i f i c inquiry  accomplished by Darwin (1872), who  emotional expression. Izard, 1971, Berscheid  1977;  explored the adaptive  A number of theorists have continued  Plutchik, 1962;  Tomkins, 1962,  1963,  significance of this work (e.g.,  1980).  According  to  (1983), Darwin has l e f t two legacies:  The f i r s t was an insistence that no theory of emotion would be l i k e l y to be v a l i d i f i t d i d not illuminate what purpose emotion serves for the survival of the species. Today, v i r t u a l l y a l l theorists of emotion agree that the experience and expression of emotion has served, and probably continues to serve, an important function i n the survival of humans. From the evolutionary perspective, then, emotion i s not an " i r r a t i o n a l " and f r i v o l o u s component of human behavior. To the contrary, i t s r a t i o n a l i t y and d i g n i t y derive d i r e c t l y from i t s service i n the l i f e and death struggle of each i n d i v i d u a l to survive and to survive as comfortably and as happily as possible (Berscheid, 1983, p. 120). The second legacy was Darwin's recognition that the expression of emotion was  associated with strong "nervous e x c i t a t i o n . "  This was  taken up, as  we  s h a l l see, by the advocates of the b i o l o g i c a l t r a d i t i o n . Contemporary proponents of the evolutional perspective postulate that emotional expression and experience r e s u l t s from innate neural processes that produce d i s c r e t e , fundamental emotions.  Although there i s much agreement on  the processes leading to fundamental emotions, there i s less agreement over their number and kind.  Table 1 i l l u s t r a t e s t h i s .  William James and the Primacy of the  Periphery  Before James, theories of emotion were largely c e n t r a l i s t i c ; that i s ,  TOMKINS  IZARD  PLUTCHIK  Fear Anger Enjoyment Disgust Interest Surprise Contempt Shame  Fear Anger Joy Disgust Interest Surprise Contempt Shame Sadness  Fear Anger Joy Disgust Anticipation Surprise Sadness  Distress Guilt Acceptance  Table 1. Fundamental or Primary Emotions Listed by Three Leading Theorists (from Mandler, 1984, p.  9  they posited a mind that interpreted events, the i n t e r p r e t a t i o n providing the emotional f e e l i n g (Candland, 1977).  James took up i n v e s t i g a t i o n of the  association between e x c i t a t i o n and emotion that Darwin suggested and theories of emotion toward physiological explanations.  turned  As Candland states,  though, t h i s movement toward physiology was as l i k e l y a r e s u l t of the increasing use of postulative-deductive behavior.  The novelty of James  1  s c i e n t i f i c procedures i n studying  theory i s that i t s h i f t e d attention from the  central mind to the peripheral structures of the body.  According  to James  (1884), our natural way of thinking about...emotions i s that the mental perception of some fact excites the mental a f f e c t i o n c a l l e d the emotion, and that t h i s l a t t e r state of mind gives r i s e to the bodily expression. My thesis on the contrary i s that the bodily changes follow d i r e c t l y the PERCEPTION of the exciting fact,and that our f e e l i n g of the same changes as they occur IS the emotion (p. 189). The sequence James proposed was perception, then arousal or emotion, then cognition.  This has generally come to be c a l l e d the James-Lange theory,  recognizing the similar contribution of Lange (1885). i n fact, more extreme.  Lange's p o s i t i o n was,  Lange i n s i s t e d that emotions were the consequences of  c e r t a i n "vaso-motor e f f e c t s . " Mandler (1984) points out that James has frequently been interpreted as having said that emotions followed v i s c e r a l changes and nothing more.  In fact, James d i d not confine himself to v i s c e r a l  antecedents and argued that Lange had placed too much emphasis on the vasomotor f a c t o r .  Mandler recognizes  theory i s problematic  that i n t e r p r e t a t i o n of the James-Lange  with respect to how  produce the bodily e f f e c t s .  the perceptions of external events  He c i t e s James' insistence that  10  environmental events can give r i s e to bodily, v i s c e r a l changes without any awareness of the meaning of these environmental events and without any interpretation of them. For the modern reader, who sees nothing strange or surprising i n perceptual, cognitive processes occurring i n response to environmental events without conscious accompaniment, the central argument about the v i s c e r a l emotional sequence that James and Lange asserted seems to f a l l apart at that point (Mandler, 1984, p.22). The allowance that complex cognitive processes can intervene between the environment and bodily reaction, and hence that f e l t emotion r e s u l t s i n part from some i n t e r p r e t i v e event that generates the bodily response, undoes the James-Lange p o s i t i o n that bodily responses have primacy.  C e n t r a l i s t Theories Much research has been devoted to testing the sequence proposed by James. Prominent i s the work of Cannon (1927).  His empirical findings d i d not  support the equality of bodily change and v i s c e r a l f e e l i n g as presumed by biological tradition.  the  Cannon showed that d i f f e r e n t emotions are accompanied  by similar, not d i f f e r e n t bodily states; that separation of the v i s c e r a and central nervous system does not preclude emotional behavior; that  the  autonomic nervous system responds slowly and d i f f u s e l y i n contrast to more rapid emotional reactions; and that manipulation of bodily states through a r t i f i c i a l means (e.g., drugs) does not necessarily a f f e c t emotional- states. More recently, Schachter and his colleagues Singer, 1962)  proposed a c o g n i t i o n — a r o u s a l  emotion, i n the t r a d i t i o n of Cannon.  (Schachter, 1971;  Schachter &  or l a b e l l i n g explanation  Schachter and his colleagues  of  argue that  emotional state results from the interaction of physiological arousal and a cognition as to the cause of the arousal.  Meyer (1956), who  work of Schachter, concluded that physiological reactions  anticipated  the  are  undifferentiated, becoming d i f f e r e n t i a t e d into a s p e c i f i c emotional experience only as a r e s u l t of cognition of the stimulus s i t u a t i o n .  Arousal does not  11  evoke any s p e c i f i c emotion, i t serves only to i n t e n s i f y emotional states while cognitions determine their q u a l i t y .  Both physiological arousal and cognitive  evaluation are necessary for the production of emotional states but neither i s sufficient. causes.  Unexplained arousal also stimulates a need to evaluate i t s  If t h i s search i d e n t i f i e s an emotional stimulus as the l i k e l y cause,  then emotional experience w i l l follow.  Most of the empirical tests of  Schachter's theory have examined the consequences of this unexplained arousal. C e n t r a l i s t theories, such as those of Cannon and Schachter, emphasize the importance of structures i n the central nervous system, i . e . the mind, as determining the q u a l i t y of emotional experience.  Physiological reactions are  granted to be necessary but not s u f f i c i e n t f o r emotion to follow.  C o n f l i c t Theories Part of the mental or c e n t r a l i s t p o s i t i o n has adopted a p a r t i c u l a r explanation for the source of autonomic nervous system arousal necessary for emotion, namely c o n f l i c t or interruption.  C o n f l i c t theories "concern  themselves with s p e c i f i c mechanisms whereby current behavior i s interrupted and  'emotional' responses are substituted....when an important a c t i v i t y of the  organism i s blocked, emotion follows" (Hunt, 1941).  Paulhan's  (1930) main  thesis was that we observe the same f a c t — t h e arrest of tendency—whenever any a f f e c t i v e phenomena take place.  Even p o s i t i v e emotions are the result of  disconfirmed expectations or arrested actions.  The p a r t i c u l a r emotion  experienced depends on the tendency that i s arrested and the conditions under which i t occurs. Meyer (1956), for example, stated that emotion i s "aroused when a tendency to respond i s arrested or i n h i b i t e d . "  Meyer gave credit to Dewey  (1894, 1895) for having fathered the c o n f l i c t theory of emotion, though he  12  says that Paulhan's work predates Dewey's. emotion when actions run t h e i r course.  Dewey said that there i s no  Emotions a r i s e only when other  irrelevant reactions r e s i s t integration with those on t h e i r way  to completion.  "Such resistance means actual tensions, checking, interference, i n h i b i t i o n , or conflict....(Such) c o n f l i c t constitutes the emotion..without such c o n f l i c t there i s no emotion, with i t there i s " (Angier,  1927).  In contrast to Dewey and Paulhan, Mandler (1984) does not assume that c o n f l i c t has s p e c i f i c emotional consequences, but rather that i t has undifferentiated v i s c e r a l consequences. The emotional content—the q u a l i t y of the emotion that f o l l o w s — i s set by s p e c i f i c cognitive circumstances of the interruption and possibly i t s consequences. Thus, both p o s i t i v e and negative emotions are seen as following interruption, and the same i n t e r r u p t i v e event may indeed produce d i f f e r e n t emotional states or consequences, depending on the surrounding s i t u a t i o n a l and intrapsychic cognitive context (p. 46). Mandler's theory, as most recently expressed (Mandler, 1984), i s the most comprehensive c o n f l i c t theory of emotion.  Mandler incorporates  modern  theories of attentional and cognitive structures, such as schemas, into his analysis of the process by which individuals appraise stimuli and meaning to their f e e l i n g states.  assign  He also posits interruption i n the realm of  thinking and perception  ( i n addition to behavior) as sources of arousal.  for example, discrepant  perception  expectations and  schema and  currently a c t i v e schema.  i s that which does not f i t our  so interrupts the dominant expectation,  S i m i l a r l y , discrepant  So,  perceptual the  thoughts such as p r i o r  engagements just remembered can interrupt the ongoing stream of thought and mental a c t i v i t i e s .  Structural Theories Mandler proposes very few sources of arousal, interruption being the most  13  important f o r emotion.  Cognitive evaluations, however, can have many sources,  of which he posits three classes: innate preferences, personal and c u l t u r a l predication, and structural value.  These are the sources of value, the other  half of the cognition-plus-arousal-equals-emotion formula. Of most interest here are structural theories of a f f e c t .  As Mandler  describes, the cognitive s t r u c t u r a l i s t s start with a small set of cognitive variables, usually including goals, and then construct possible outcomes f o r various combinations of these cognitive states. The outcomes of these cognitive constellations are then mapped, often with some h e s i t a t i o n , onto the putative emotional labels of the common language. These constructions eventuate i n a r e l a t i v e l y large number of emotional or a f f e c t i v e outcomes, ranging from a dozen or so to sixteen or to as yet indeterminate larger numbers (p. 211). Roseman (1984), f o r example, presents f i v e dimensions which are used to evaluate a r e a l or imagined determines  stimulus.  The r e s u l t of t h i s evaluation  the q u a l i t y of the emotional reaction. Roseman's dimensions are  (1) whether the event i s consistent with the motives of the i n d i v i d u a l , that i s , whether i t i s p o s i t i v e or negative; (2) whether the stimulus i s appetitive (oriented toward giving or withholding rewards) or aversive (oriented toward giving or withholding punishments); (3) whether the event or stimulus i s perceived to be caused by the s e l f , by an other, or by circumstances; (4) whether the occurrence of the event i s c e r t a i n or uncertain; and (5) whether the i n d i v i d u a l i s weak or strong r e l a t i v e to the stimulus or cause.  Liking,  for example, i s the r e s u l t of a stimulus that i s p o s i t i v e , that i s consistent with the i n d i v i d u a l ' s motives, and i s other-caused, and i s either certain or uncertain, and i n r e l a t i o n to which the i n d i v i d u a l i s either strong or weak. Table 2 shows how combinations  of levels of these dimensions are mapped onto  natural language labels f o r emotional  states.  Kemper (1978, 1984) has put forward a s o c i a l i n t e r a c t i o n a l theory of  14  CircumstanceCaused  Positive  Negative  Mot ive-Con s i s t ent Appetitive | Aversive  Motive-Inconsistent Appetitive | Aversive  Surfi r i s e  Unknown  iar  He)pe  Uncertain  Discomfort Certain  Joy  Sorrow Disgust  He>pe  Uncertain  Certain  Relief  Frustration Joy  Strong  Relief  Other-Caused Uncertain  Disliking  Weak  Anger  Strong  Certain Liking Uncertain Certain Self-Caused Uncertain Shame, Guilt  Weak  Regret  Strong  Certain Pride Uncertain Certain  Table 2. Emotional reactions resulting from structural evaluations, according to Roseman (1984).  15  emotion based on the structural q u a l i t i e s of the s o c i a l environment. i s a s o c i o l o g i c a l theory, i t examines " r e l a t i o n a l or organizational of "properties  Kemper's concepts"  of relationships or groups" as determinants of emotion. The  starting point for t h i s theory i s the postulate  that human relationships  generally occur within a context of interdependence and d i v i s i o n of labor between actors.  Further, these s o c i a l relationships have two fundamental  dimensions of interaction content: power and status. Granted t h i s view of relationship, the following proposition i s possible: A very large class of human emotions r e s u l t s from r e a l , anticipated, recollected, or imagined outcomes of power and status r e l a t i o n s . This means that i f we wish to predict or understand the occurrence of many human emotions we must look at the structure and process of power and status r e l a t i o n s between actors (Kemper, 1978, p. 371, emphasis i n o r i g i n a l ) . Kemper's theory proposes three classes of emotions: s t r u c t u r a l emotions, anticipatory emotions, and consequent emotions.  Structural emotions r e s u l t  from the r e l a t i v e l y stable power and status relations between an actor and an other.  Power and status can be excess, adequate, and i n s u f f i c i e n t . Figure 1  depicts  s t r u c t u r a l emotions.  G u i l t , for example, i s the emotion that results  from a s t r u c t u r a l position of excess power over another i n an i n t e r a c t i o n . Anticipatory emotions are responses to how the actor views the future state of the r e l a t i o n s h i p : whether the future i s viewed with optimism or pessimism and with confidence or lack of confidence. r e s u l t of i n t e r a c t i o n episodes.  F i n a l l y , consequent emotions are the  They conclude the chain that l i n k s structural  and anticipatory emotions and i n t e r a c t i o n outcomes.  Power and status can  increase, decrease, or stay the same as a r e s u l t of i n t e r a c t i o n . The combinations of l e v e l s of s t r u c t u r a l , anticipatory, and consequent factors i n Kemper's scheme produce 1701 outcome c e l l s .  While t h i s suggests  that a s o c i a l i n t e r a c t i o n a l approach has the r e q u i s i t e v a r i e t y for analysis of emotion, i t also presents a daunting task of mapping i n t e r a c t i o n outcomes onto  16  Guilt, anxiety  Security  Power  Fear—anxiety  Own  fi* Status  Adequate  -Shame  Happy  'idem•Depression Fear—anxiety  Power  Other's Anger, contempt, shame  Status  Happy  Guilt—shame, anxiety  Figure 1.  Structural model of emotions, from Kemper (1978).  17  emotion l a b e l s .  By means of simplifying assumptions, though, Kemper i s able  to reduce the number of possible outcomes to less than 170.  Reconciling Theories of Emotion The theories reviewed i n t h i s chapter may have spanned the domain of theories but have by no means exhausted  it.  It i s f a i r to say that there are  as many theories of emotion as there are t h e o r i s t s .  The interested reader i s  referred to Strongman's (1977) detailed review of the positions of p a r t i c u l a r emotional t h e o r i s t s . Given these multiple conceptions of emotion, how should we proceed, how can they be reconciled?  It i s tempting to choose one that i s "best", that  maximizes some empirical c r i t e r i o n , say, l i k e predicting more of the variance i n some measure of emotionality.  F e l l (1977) observes, however, that each  theory omits some c r u c i a l aspect of emotion, yet strives to be the foundation from which others are derived and from which the others must situate their data.  For each theory the adoption of a p a r t i c u l a r c r i t e r i o n and d e f i n i t i o n  of emotion determines aspects of emotion.  that i t w i l l best explain a p a r t i c u l a r set of the Thus every theory i s the best explanation of i t s own  c r i t e r i o n and no theory i s the best explanation of a l l c r i t e r i a . The multiple conceptions of emotion can be reconciled by recognizing that at present a universal theory of emotion may not be r e a l i z a b l e .  A view held  by a number of theorists (e.g., A v e r i l l , 1980; Izard, 1977; Leventhal, 1979; Plutchik, 1980), i s that emotion i s a syndrome— a pattern of co-occurring responses.  Cognitive processes, physiological changes, emotional displays,  f e e l i n g states, are parts of a phenomenon that has many components and many expressions.  Each i s appropriate, but each i s only one indicator of emotion.  Therefore, rather than seeking a universal theory of a phenomenon with  18  multiple, disparate conceptualizations that do not j u s t i f y such generality, we should accept e x i s t i n g theories of emotion as middle-range theories. range theories attempt to explain only segments of the universe  Middle-  (Pinder &  Moore, 1979, 1980). The theory that i s best, therefore, depends on the question at hand. I t i s the one that best explains the segment of the universe of facets of emotion that one wishes to explain.  I f , for example, we are interested i n how power  and status a f f e c t s emotion i n organizations we might consult Kemper's theory. If, on the other hand, we are interested i n the s t r e s s f u l effects of emotional c o n f l i c t on worker health, we would be wise to use a t h e o r e t i c a l approach that includes physiological aspects of emotion.  F i n a l l y , i f we are interested i n  the consequences of mood states on the thought and behavior of individuals i n organizations, we should consult theories of mood.  What i s Mood? Apart from or as part of the concept of emotion what do we mean by "mood"? I propose f i r s t of a l l , the use of the term " a f f e c t " to denote the general concept of emotion.  Although some authors use the term a f f e c t to mean  less intense or less s p e c i f i c states (e.g., Mandler, 1984), I concur with Berscheid  (1983) who says that affect i s "an a n t i s e p t i c term, but one that  encompasses without prejudice the entire range of q u a l i t y and i n t e n s i t y of human emotion and f e e l i n g , from mild i r r i t a t i o n to raging hatred to blinding joy to p l a c i d contentment" (p. 110). "Affect" and "emotion" both represent the general concept, although "emotion" i s a more evocative term, suggesting a more r i c h l y colored and varied construct. And what i s this general concept?  Notwithstanding the recognition that  19  defining emotion i s fraught with hazards, and that no one d e f i n i t i o n i s l i k e l y to be s u f f i c i e n t for a l l purposes, the following p o s i t i o n i s adopted, from Mandler  largely  (1984): Affect i s the evaluative aspect of human information  processing, as opposed to i t s ideational or d e s c r i p t i v e aspects or what we usually c a l l cognition.  Affect i s the evaluation of information as good or  bad, p o s i t i v e or negative, pleasant or unpleasant, arousing or nonarousing, for example.  Cognition, on the other hand, i s the d e s c r i p t i v e component of  information, such as i t s color, size, shape, temperature, etc.  We commonly  d i f f e r e n t i a t e b e l i e f s , which are ideational concepts, from attitudes, which are a f f e c t i v e concepts, although they are not pure emotion.  Attitudes are an  evaluation of the goodness of a state of a f f a i r s , as well as a description of it. At the outset, we commented on the d i v i s i o n of psychological experience into thinking, f e e l i n g , and acting, and implied that such separation i s inappropriate.  Indeed, we have presented and proposed for adoption a  conception of emotion that emphasizes cognitive processes as sources of affective quality.  And yet there i s vocal debate over the primacy of affect  over cognition, which i s very much a rewording of the debate about the sequence of peripheral and central structures as determinants of emotion, and a demonstration of the separation of thinking and f e e l i n g .  Zajonc (1980), for  example, argues for the primacy of a f f e c t (or "preferences need no inferences") against Lazarus (1982) who defends the r o l e of cognition. Mandler  (1984) says that the heart of t h i s debate i s how preferences (affect)  and inference (cognition) a t t a i n consciousness.  He suggests that Zajonc's  taunt be rephrased as "conscious preferences need no conscious inferences." Candland (1977), on the other hand, questions the very usefulness of arguments about whether emotion or cognition "comes f i r s t " :  20  there i s no special advantage i n assuming that there i s a temporal sequence among the three processes, and much waste has come from believing that the appropriate way to untangle our confusions about emotion i s to sort out the temporal sequence. We have created a monstrous problem by assuming that explanation requires uncovering a temporal sequence....(p. 66) Candland supports a conception  i n which there i s a feedback loop among the  elements of emotional experience.  Through cognitive and physiological  components emotions can feedback and modify perceptions  and thus emotional  reaction; emotional experience can modify cognitive appraisal and thus modify continuing and future emotional experience, and s i m i l a r l y emotional experience can modify p h y s i o l o g i c a l reactions and thus modify continuing experience. It i s appropriate to emphasize that, l i k e i n the intended conceptualization of the t r i p a r t i t e mind, emotion and cognition are intertwined. mind.  Cognition i n humans i s the processing of information i n the  Unless we believe that emotions are determined outside of the mind or  do not involve the processing of information then we must agree that emotion has cognitive-aspects.  Again, the c r u c i a l d i s t i n c t i o n i s between the  ideational or d e s c r i p t i v e aspects of human information processing and the a f f e c t i v e or evaluative or emotional aspects (Mandler, 1984).  Most often when  we contrast cognition and emotion we intend to contrast ideation and evaluation.  Both evaluation and ideation occur i n the mind and are,  therefore, cognitive processes.  Emotion and evaluation do, however, have  d i s c e r n i b l e effects on thought and behavior. Moods are self-evaluative emotional states. experience, moods are less intense values.  On the spectrum of emotional  Berscheid describes moods as  " l i t t l e emotions", l i k e emotions i n tone but with less intense or no physiological arousal. We can d i f f e r e n t i a t e moods from strong emotion by defining moods as non-  21  interrupting states.  The effects of moods have been argued to r e s u l t from  automatic processes.  More intense emotion demands greater attention, i s  associated with more s p e c i f i c response patterns and can interrupt ongoing behavior (Simon, 1967; Mandler, 1984).  Moods do not have s p e c i f i c targets,  they have pervasive effects, influencing how evaluations are formed and are r e f l e c t e d i n the tone of other evaluations, coloring judgements i n a congruent way.  Moods are temporally limited, they s h i f t over time.  They are manifested  i n subjective experience and i n self-reports. In sum, moods are f e e l i n g states that are r e f l e c t e d i n subjective experience and self-reports, and that broadly and pervasively a f f e c t the character of ongoing behavior and evaluations without interrupting them.  The Effects of Mood on Thought and Behavior The consequences of p o s i t i v e and negative mood f o r thought and behavior are widespread and s i g n i f i c a n t , as we s h a l l see. The effects of mood d i f f e r from those of strong emotion.  Strong p o s i t i v e or negative emotion can a l t e r  which behavior or thoughts are continued: i t can i n t e r f e r e with or disrupt ongoing thought and behavior. count (Young, 1961; c . f .  Indeed emotion has often been maligned on t h i s  Arnold, 1970; Easterbrook, 1959; Leeper, 1970).  Mood, on the other hand, does not change which behaviors or thoughts are continued but may a f f e c t their tone.  So, for example, being i n an anxious  mood may not disrupt one's choice to engage i n a task but may change one's focus of attention.  Isen (1984) argues that such effects of mood are more  pervasive and of more everyday consequence than the effects of intense emotion.  The pervasiveness of the effects of mood i s of obvious consequence  to organizational settings.  While the interrupting e f f e c t s of emotion are  worthy of study i n work settings, interruption i s more obvious.  Moods, on the  22  other hand, may change the frequency, duration, or i n t e n s i t y of work behavior without changing i t s occurrence.  Intense emotional experience may cause  individuals to withdraw from work, such withdrawal i s important.  Also  important i s the effect of influences on the character and tone of ongoing work.  In the following sections, we w i l l review i n b r i e f the empirical  evidence surrounding the effects of mood on thought and behavior.  Much of  t h i s material i s drawn from Blaney (1986) and Isen (1984), to which the interested reader i s also referred.  Affect and Behavior Positive a f f e c t can have broad effects on behavior. to make one more l i k e l y to help others.  Feeling good tends  Finding a coin i n a public telephone  or succeeding on a test or thinking about p o s i t i v e events i s associated with a tendency to help others, with an important q u a l i f i c a t i o n : recent studies have noted that mood protection can occur.  Someone f e e l i n g good may be less l i k e l y  to help someone else i f helping w i l l damage their p o s i t i v e f e e l i n g state (Isen & Simmonds, 1978).  S i m i l a r l y , people who f e e l good may be more l i k e l y to  behave as they please, helping more when they want to but less when there i s a reason to avoid i t (Forest, Clark, M i l l s & Isen, 1979).  P o s i t i v e affect i s  also associated with a tendency to reward oneself (Mischel, Coates & Raskoff, 1968) and to display increased preference for p o s i t i v e than negative information about the self (Mischel, Ebbesen & Zeiss, 1973, 1976).  People who  f e e l good are more w i l l i n g to i n i t i a t e conversations with others (Batson, Coke, Chard, Smith & T a l i a f e r r o , 1979; Isen, 1970) express greater l i k i n g and hold more p o s i t i v e conceptions of others and be more receptive to persuasive communications (Veitch & G r i f f i t t , 1976).  Positive a f f e c t i s associated with  greater tendency to take s l i g h t r i s k s , but not large ones (Isen, Means,  23  Patrick, & Nowicki, 1982).  When r i s k i s substantial subjects who are f e e l i n g  good tend to be more conservative (Isen & Patrick, 1983). Negative states have less consistent influences on behavior.  Sometimes a  symmetry with p o s i t i v e a f f e c t i s found such that negative a f f e c t has opposite e f f e c t s , sometimes no effect i s found, sometimes an effect similar to that for p o s i t i v e a f f e c t i s found.  Isen (1984) suggests that the assumed symmetry  between p o s i t i v e and negative a f f e c t i s more a convention of language than a r e f l e c t i o n of their function.  Negative states seem to evoke c o n f l i c t i n g  tendencies, either to engage i n thought and behavior that i s compatible with negative mood, or tendencies to change or eliminate the .unpleasantness, including by engaging i n affect-incompatible (positive) behavior. For example, negative a f f e c t sometimes reduces helping behavior (e.g. C i a l d i n i & Kenrick, 1976; Weyant, 1978), and sometimes increases i t (e.g. C i a l d i n i , Darby & Vincent, 1973; Weyant, 1978).  Sometimes negative states  seem to have no effect at a l l , as i f the competing tendencies had neutralized each other.  Researchers who have found that negative feelings increase  p o s i t i v e behavior have proposed that this i s the result of a tendency f o r negative a f f e c t to cause attempts at mood improvement or repair, p a r a l l e l to a tendency toward mood protection under p o s i t i v e a f f e c t ( C i a l d i n i et a l . , Isen, Horn & Rosenhan, 1973; Weyant, 1978).  1973;  Mood improvement may have more  complex e f f e c t s because tendencies toward p o s i t i v e behavior oppose negative a f f e c t , while mood protection effects are generally compatible with p o s i t i v e affect.  Isen (1984) points out that both mood repair and protection are  compatible with self-regulation theory (Bandura,  1977), which proposes that  "experiencing p o s i t i v e a f f e c t gives r i s e to strategies designed to maintain that desirable state, and that negative a f f e c t results i n strategies aimed at changing the undesirable state" (Isen, 1984, p.  198).  24  Another explanation for "mood repair" effects i s that they represent controlled rather than automatic processes. these e f f e c t s are not mood effects at a l l .  By the d e f i n i t i o n proposed here, According to t h i s explanation, the  negative a f f e c t which results i n increased helping rather than decreased helping has been of s u f f i c i e n t i n t e n s i t y to s h i f t processing from automatic to controlled.  Rather than influence the character of helping behavior by  reducing i t , thought may be interrupted such that controlled behavior occurs aimed at improving one's emotional state.  Mood and Thought Mood has been argued to influence not only behavior but also the way information i s processed.  C r e a t i v i t y , for example, has been shown to be  f a c i l i t a t e d by p o s i t i v e mood (Isen, Daubman & Gorgoglione, i n press; Mednick, 1962; Mednick, Mednick, & Mednick, 1964).  Positive mood has been shown to  increase the use of an i n t u i t i v e strategy or h e u r i s t i c (Isen, Means, Patrick, & Nowicki, 1982), and the use of an e f f i c i e n t simplifying decision strategy (Isen & Means, 1983).  Whether such strategies are e f f e c t i v e depends on the  nature of the decision making task.  Isen and colleagues (1982) found that  performance was impaired on the tasks they used.  Using a more complex  decision-making task, Isen and Means (1983) found that decisions made under p o s i t i v e a f f e c t were made more quickly and e f f i c i e n t l y , although the f i n a l decisions of persons made to f e e l good d i d not d i f f e r from those of control subjects.  Three mental processes describe the effects of mood on thought.  These  are (1) mood congruence at r e t r i e v a l , (2) mood congruence at encoding, and (3) affect-state-dependent learning. The f i r s t process, mood congruence at r e t r i e v a l , refers to improved  25  r e c a l l for material which has an a f f e c t i v e tone congruent with mood during recall.  Underlying this process i s the argument that mood serves as a  r e t r i e v a l cue for memory, l i k e category names or other organizing units.  A  large number of studies using varied mood inductions have shown that p o s i t i v e words w i l l be r e c a l l e d more e a s i l y during p o s i t i v e moods and negative words w i l l be r e c a l l e d more e a s i l y during negative moods (Blaney, 1986; Isen, 1984). Judgement, evaluation, expectations, decision-making, and behavior follow from these processes. unexpected  For example, p o s i t i v e f e e l i n g associated with receiving an  g i f t results i n subsequent raised reports of the performance and  service records of consumer goods (Isen, Shalker, Clark & Karp, 1978).  Being  i n a p o s i t i v e mood state causes people to express higher expectations f o r future success (Feather, 1966).  Sometimes, though, while p o s i t i v e a f f e c t at  r e t r i e v a l improves r e c a l l of p o s i t i v e material, negative a f f e c t at r e t r i e v a l does not improve r e c a l l of negative material (e.g.  Isen et a l . , 1978;  Teasdale & Fogarty, 1979; Teasdale & Taylor, 1981; Nasby & Yando, 1982; Natale & Hantas, 1982). The second process, mood congruence at encoding, refers to the formation of stronger associations i n memory when the a f f e c t i v e tone of material to be remembered matches the mood state at encoding.  So, f o r example, p o s i t i v e  words w i l l be r e c a l l e d better when learned while i n a p o s i t i v e mood and negative words w i l l be r e c a l l e d better when learned i n a negative mood (e.g., Teasdale & Russell, 1983).  Bower, G i l l i g a n , and Montiero (1981) found that  r e c a l l for f a c t s compatible with a p o s i t i v e state was superior when individuals were happy while learning the material, and likewise individuals who learned material while sad showed superior r e c a l l of information compatible with a negative a f f e c t i v e state. The effects of negative a f f e c t on cognitive processes are, l i k e those on  26  behavior, not always e n t i r e l y c l e a r .  Bower (1981; Bower et a l . , 1981)  reported that sadness at the time of encoding f a c i l i t a t e d the r e c a l l of sad material; Nasby and Yando (1982) d i d not f i n d such an e f f e c t . The t h i r d process i s affect-state-dependent-learning, which refers to improved r e c a l l for a f f e c t i v e l y neutral material when mood at r e c a l l matches mood during encoding.  So, neutral words learned while someone i s i n a  p o s i t i v e mood are more l i k e l y to be remembered when that person i s again i n a p o s i t i v e mood than when he or she i s i n a d i f f e r e n t mood (e.g., Eich & Metcalfe, i n press).  Affect-state-dependent  learning i s the tendency for  material learned when an individual i s i n a s p e c i f i c mood state to be r e c a l l e d better when the subject i s again i n that state.  For example, Bower, Montiero  and G i l l i g a n (1978) found r e c a l l of a l i s t of words to be better when mood during learning was the same as mood during r e c a l l .  Other authors have,  however, f a i l e d to r e p l i c a t e the effect (Isen, Shalker, Clark & Karp, 1978; Nasby & Yando, 1982).  Isen and colleagues have argued that a f f e c t - s t a t e -  dependent learning effects are a t t r i b u t a b l e to the e f f e c t of a f f e c t i v e state on the r e t r i e v a l of mood congruent material. learning effect i s responsible.  That i s , a r e t r i e v a l rather than  Bower and colleagues argue the reverse, that  r e t r i e v a l effects are the result of state-dependent-learning.  Teasdale and  Russell (1983) have, however, shown that r e t r i e v a l effects can occur that are not a t t r i b u t a b l e to a f f e c t during learning.  They attempt to resolve the  discrepancy i n the l i t e r a t u r e by noting differences between studies i n the tone of target words, suggesting that the r e t r i e v a l effect might be statedependent a f t e r a l l . dependent-learning  Nevertheless,  Isen (1984) concludes  that state-  as an effect of induced mood should be considered to be a  specialized phenomenon, d i s t i n c t from r e t r i e v a l e f f e c t s . In a review of the l i t e r a t u r e on a f f e c t and memory, Blaney (1986)  27  concludes that evidence for a f f e c t - state-dependency  i s mixed.  A number of  researchers have f a i l e d to r e p l i c a t e Bower and colleagues' (1978) r e s u l t s . That i s , they have not been able to show experimentally that r e c a l l i s improved when mood at r e t r i e v a l matches mood at encoding or learning.  In  contrast, autobiographical studies do tend to demonstrate such an e f f e c t . Individuals have better r e c a l l for p o s i t i v e l i f e events when i n a p o s i t i v e mood and s i m i l a r l y have better r e c a l l for negative l i f e events when i n a negative mood (e.g.  Bower, 1981).  Such autobiographical studies support the  affect-state-dependent learning hypothesis while experimental studies appear not to. al.,  Bower has since questioned the v a l i d i t y of h i s o r i g i n a l (Bower et  1978) finding, suggesting that i t was a " s t a t i s t i c a l accident" (Bower,  1985).  He has, however, proposed the hypothesis that affect-state-dependent  learning does occur when there i s "causal belonging" between the emotion and the event.  That i s , " i n order to establish e f f e c t i v e associations between the  emotion and the event to be remembered, the subject would have to causally a t t r i b u t e h i s emotional reaction to that event" (Bower, 1985, pp.  16-17).  According to t h i s hypothesis, the r e c a l l of a f f e c t i v e l y neutral material would not be affected when mood occurs only as a background during learning. Eich and Metcalfe ( i n press) have proposed a more general hypothesis of affect-state-dependent learning.  They argue that state-dependent  recall  occurs only for items that individuals generate themselves, as opposed to items that they read.  Eich and Metcalfe demonstrated that a mismatch between  mood state at the time of encoding and r e t r i e v a l hampered the r e c a l l of s e l f generated items but not read items:  28  events that are generated through i n t e r n a l processes such as reasoning, imagination, and thought are more c l o s e l y connected to or colored by one's current mood than are those that derive from external sources, and as a consequence, internal events are more apt to be rendered inaccessible for r e t r i e v a l i n the t r a n s i t i o n from one mood state to another than are external events (Eich and Metcalfe, i n press, p. 24).  Network Theory The dominant theoretical perspective r e l a t i n g a f f e c t and memory and explaining mood congruence and state dependence i s network theory (Blaney, 1986).  Bower's statement of i t i s as follows:  The semantic-network approach supposes that each d i s t i n c t emotion...has a s p e c i f i c node or unit i n memory that c o l l e c t s together many other aspects of emotion that are connected to i t by associative pointers....Each emotion unit i s also linked with propositions describing events from one's l i f e during which that emotion was aroused....These emotion nodes can be activated by many s t i m u l i — b y physiological or symbolic verbal means. When activated above a threshold, the emotion unit transmits excitation to those nodes that produce the pattern of autonomic arousal and expressive behavior commonly assigned to that emotion....Activation of an emotion node also spreads a c t i v a t i o n throughout the memory structures to which i t i s connected, creating subthreshold excitation at those event nodes....Thus...excitation (of) the sadness node...will maintain a c t i v a t i o n of that emotion and thus influence l a t e r memories retrieved (Bower, 1981, p. 135).  Blaney (1986) points out that while Bower's statement favors statedependent  effects and i s focused on mood at input, i t can, with modification,  describe the effects of mood congruence.  A relevant concern i s whether the  three seemingly separate processes of mood-state dependence, mood congruence at encoding, and mood congruence at r e t r i e v a l can be related within a single framework.  The r e v i s i o n of the conditions under which state dependence  occurs, as suggested by Bower (1985) and Eich and Metcalfe ( i n press), help us do t h i s . Mood congruence effects at r e t r i e v a l can be modelled by a semanticnetwork theory as follows: a f f e c t i v e l y toned (non-neutral) mental contents are  29  connected t o congruent emotion nodes. associated  So, f o r example, t h e word "good" i s  w i t h t h e emotion f o r happy and t h e word "bad" i s a s s o c i a t e d  with  the node f o r s a d . Mood s t a t e , as a r e s u l t o f n a t u r a l o r e x p e r i m e n t a l induction,  a c t i v a t e s t h e emotion node, s p r e a d i n g a c t i v a t i o n t o connected  memory u n i t s , c r e a t i n g more a c c e s s i b l e . p o s i t i v e mood.  s u b t h r e s h o l d e x c i t a t i o n , and making t h o s e memory u n i t s  Thus, I am more l i k e l y t o remember t h e word "good" when i n a A t encoding, a c t i v a t i o n o f emotion nodes and consequent  s p r e a d i n g a c t i v a t i o n s t r e n g t h e n s t h e development  of a s s o c i a t i o n s .  l i k e l y t o l e a r n t h e word "good" when i n a p o s i t i v e mood because  I am more  t h a t memory  unit i s activated. State-dependent Bower's " c a u s a l  l e a r n i n g can be understood, e s p e c i a l l y i n l i g h t o f  b e l o n g i n g " h y p o t h e s i s , and E i c h and M e t c a l f e ' s o f i n t e r n a l  g e n e r a t i o n , as t h e p r o c e s s by which p r e v i o u s l y a f f e c t i v e tone o r v a l e n c e .  neutral material  a c q u i r e s some  When mental events a r e generated i n t e r n a l l y a x  c o n n e c t i o n between t h e a c t i v e emotion node and t h e event i s e s t a b l i s h e d . if  I e x p e r i e n c e a p a r t i c u l a r event as p l e a s a n t , t h e elements o f t h a t  become a s s o c i a t e d  w i t h p o s i t i v e emotion.  event  I f I generate a word a s s o c i a t i o n  w h i l e i n a good mood t h a t memory u n i t becomes l i n k e d t o t h e congruent node and takes on a p o s i t i v e a f f e c t i v e tone. mood t o my performance  So,  Or, i f I a t t r i b u t e my p o s i t i v e  on a t a s k then t h a t performance  emotion nodes congruent w i t h p o s i t i v e mood.  emotion  becomes connected t o  R e c a l l f o r the previously  neutral  a s s o c i a t i o n o r event i s subsequently f a c i l i t a t e d by t h e p r o c e s s o f mood congruent r e t r i e v a l . mood-state-dependent encoded  The fundamental  l e a r n i n g i s t h e extant v a l e n c e o f t h e m a t e r i a l  or r e t r i e v e d .  congruence  d i f f e r e n c e between mood congruence and  i t i s not.  In s t a t e dependence p r o c e s s e s i t i s n e u t r a l ,  t o be i n mood  30  Mood and Arousal Network theory proposes that mental contexts are connected on the basis of their a f f e c t i v e tone or valence, b a s i c a l l y whether they are p o s i t i v e or negative.  Thus p o s i t i v e moods prime p o s i t i v e material and negative moods  prime negative material.  Implicit i n such a discussion i s the idea that moods  vary on a continuum from p o s i t i v e to negative.  I t i s also assumed that t h i s  hedonic valence i s the only continuum on which moods l i e and the only basis by which priming i n memory occurs.  In fact, there i s t h e o r e t i c a l and empirical  support f o r the argument that moods also vary along a continuum from arousal t o nonarousal, or of a r o u s a l — s l e e p i n e s s .  Most theories of emotion e x p l i c i t l y  include, or at the least acknowledge, a r o l e for p h y s i o l o g i c a l arousal. Arousal i s often seen as a necessary part of emotional experience. Further, Russell  (1979,  1980)  has argued that the expression of emotion  i s best characterized by two bipolar dimensions.  The f i r s t dimension  corresponds to what we have c a l l e d valence, Russell terms i t p l e a s u r e — displeasure.  The second dimension i s a r o u s a l — s l e e p i n e s s .  Russell  (1980)  showed that the relationships between terms describing the complete range of emotion f a l l meaningfully two dimensions.  around the perimeter of the space defined by these  That i s , they form a circumplex, as shown i n Figure 2. In  other words, the complete range of human emotional expression can be described i n terms of these two dimensions, a r o u s a l — s l e e p i n e s s and p l e a s u r e — displeasure.  A c o r o l l a r y of t h i s assertion i s that both dimensions are  necessary to describe mood states.  Depression and anxiety, f o r example, are  similar i n valence: they are both unpleasant.  Yet they are evidently d i s t i n c t  mood states, a difference which i s captured by recognizing that anxiety i s characterized by more arousal and that depression sleepiness.  i s characterized by  AROUSAL  DISTRESS  EXCITEMENT  MISERY  •*  DEPRESSION  PLEASURE  CONTENTMENT  SLEEPINESS  Figure 2.  Circumplex model of emotions, from Russell (1980).  32  Arousal and Memory Network theory proposes that mental contents are mutually associated to the degree that they are similar i n valence or p l e a s u r e — d i s p l e a s u r e .  It also  posits that mental contents are primed by mood states with which they have congruent valence.  I f , however, moods also vary on an arousal dimension,  mental contents might also be primed by mood states with which they are congruent i n terms of arousal.  That i s , i f a p a r t i c u l a r unit of memory i s  associated with a mild amount of arousal, the presence of a mood state characterized by mild arousal may prime that unit, making i t more accessible to r e c a l l .  Just such a mechanism has been proposed by Clark (1982).  Clark does not take a p o s i t i o n on whether arousal i s necessary to, separate from, or an i n t e g r a l part of what emotion i s . Rather, she assumes that "when a person feels good or bad as a r e s u l t of some event or thought, that person often experiences  increased autonomic arousal" (p.263).  This'  implies that the valence and arousal dimensions of mood coincide, both good and bad mood are accompanied by the presence of arousal.  Russell argues that  arousal and valence are orthogonal, although he makes no claims about the physiological concomitants of his arousal dimension.  Both good and bad mood  can be accompanied by the absence as well as the presence of arousal. According to Russell, the arousal dimension i s necessary to d i s t i n g u i s h between moods, between a state of pleasant calm and e l a t i o n , f o r example, or between anxiety and depression.  Thus, whereas Russell argues f o r separation  of the two dimensions, Clark argues f o r t h e i r combination.  Clark's conclusion  may be a r e s u l t of a bias i n the l i t e r a t u r e she reviews toward mood inductions that tend to produce excitement rather than calm and anxiety rather than depression. In any case, i t i s evident that mood involves both arousal and valence.  33  What i s of interest here i s the role of arousal i n priming r e c a l l of mental contents.  As Clark points out, the concept of arousal has been l e f t out of  most discussions of the influence of mood.  She contends that arousal  i n t e n s i f i e s the priming of memory by p o s i t i v e or negative mood that i s s i m i l a r l y toned.  In a study of the effects of arousal or the f a v o r a b i l i t y of  judgements, Clark (1981) found that arousal caused by exercise increased the effect of a p o s i t i v e mood induction on evaluations. a f f e c t ratings i n the absence of pleasant mood.  Arousal did not, however,  Thus, i t seems that  physiological arousal did not have an independent e f f e c t on r e c a l l . increased arousal did not increase ratings independently  That i s ,  of mood valence.  Clark points out, though, that these r e s u l t s are "consistent with the idea that experiencing arousal i n the present w i l l help to prime material previously stored i n memory linked with arousal" (p. 280). That i s , although i t was not demonstrated, arousal may, by i t s e l f , cue congruent material i n memory, or cue material that i s associated with arousal. In summary, then, i t i s possible that both components of mood, a r o u s a l — sleepiness and pleasure—displeasure, may serve to prime associated material i n memory.  Evidence for an effect of valence i s strong.  Certainly mental  contents can be more e a s i l y placed along the l a t t e r dimension than the former. Memories can be pleasant, unpleasant, or neutral.  More d i f f i c u l t and more  i d i o s y n c r a t i c , i s the c l a s s i f i c a t i o n of memories as associated with arousal or sleepiness.  34  III.  EMOTION AND  ORGANIZATIONS  Emotions themselves are not foreign to organizations or to organizational behavior.  As Park, Sims and Motowidlo (1986) point out, a f f e c t i v e variables  such as s a t i s f a c t i o n (Locke, 1976), valence preference  (Cyert and March, 1963)  understanding of organization.  (Vroom, 1964;  M i t c h e l l , 1974)  and  have h i s t o r i c a l l y been central to our  The early h i s t o r y of the Human Relations  movement included discussion of emotion but, following trends i n psychology and sociology, this was of behavior. behavior  Now,  replaced by an emphasis on the cognitive determinants  again following i t s i n t e l l e c t u a l parents, organizational  i s beginning  to refocus attention on emotion.  The study of emotion i n organizational behavior  can, l i k e the study of  emotion i n general, be s i m p l i s t i c a l l y categorized as representing approaches.  two  F i r s t i s the treatment of emotion as a dependent v a r i a b l e caused  by, or a function of, the work s e t t i n g .  One  such conception of emotions i s as  intrapsychic outcomes that indicate employee well-being and happiness (Rafaeli & Sutton, 1985).  Stress research and, more obviously, the job s a t i s f a c t i o n  l i t e r a t u r e are examples.  Another conception of emotion as a dependent  v a r i a b l e focuses on displayed rather than f e l t emotions, p a r t i c u l a r l y as f u l f i l l m e n t s of work r o l e expectations. (1983) and R a f a e l i and Sutton  Exemplified by the work of Hochschild  (1987), t h i s approach examines how  emotional  display comes to form part of the work r o l e and what the consequences of such managed emotion are.  This approach adopts a s o c i o l o g i c a l approach to the  generation of the rules that govern emotional expression.  It does grant,  though, that displayed emotion can influence psychological emotional experience. The second approach treats emotion as an independent v a r i a b l e , one that  35  influences processes or variables of organizational consequence.  Rather than  treat emotion as an outcome of work or as a required r o l e display, emotion i s an independent variable that influences thought and behavior. Conceptual frameworks drawn from cognitive psychology and u t i l i z i n g a f f e c t have been applied to performance appraisal (Park, Sims, & Motowidlo, 1986; S i n c l a i r , i n press), task perceptions and performance (Motowidlo, Packard & Manning,  1985;  Park, Sims & Motowidlo, 1984), organizational c i t i z e n s h i p (Bateman & Organ, 1983), and turnover (Motowidlo & Lawton, 1984).  Emotion as dependent variable Rafaeli and Sutton (1985) contend that emotion has most often been examined as a dependent variable i n organizational behavior, predominantly i n the vast l i t e r a t u r e on job s a t i s f a c t i o n .  Locke (1976), for example, defines  job s a t i s f a c t i o n as "a pleasurable or p o s i t i v e emotional state r e s u l t i n g from the appraisal of one's job or job experiences" (p.  1300).  A large l i t e r a t u r e  has addressed the causes of job s a t i s f a c t i o n , such as work rewards, supervision and coworkers, working conditions, task c h a r a c t e r i s t i c s , individual needs and values, and others.  Job s a t i s f a c t i o n as emotion. It i s misleading, however, to equate job s a t i s f a c t i o n with emotion.  Job  s a t i s f a c t i o n i s also, perhaps most often, defined as a "person's attitude toward work" (e.g. Tosi, Rizzo & C a r r o l l , 1986).  Attitudes are a f f e c t i v e  constructs, that i s , they have an evaluative, " f e e l i n g " component. are also ideational constructs, representing b e l i e f s about jobs. s a t i s f a c t i o n i s at least p a r t l y such an ideational construct.  But they Job  For example,  job s a t i s f a c t i o n can include comparison of aspects of one's job with some  36  standard.  Such a comparison can be quite d i s t i n c t from how one f e e l s at work.  When asked i f I am s a t i s f i e d with my salary I may t e l l you that I am not, because I believe that similar jobs are better paid elsewhere.  Yet t h i s  b e l i e f , t h i s d i s s a t i s f a c t i o n may be quite independent of my emotional state: I may be happy at work.  On the other hand, I may be unhappy because I am  s a t i s f i e d with my work. Job s a t i s f a c t i o n can be an inappropriate indicator of emotion because of t h i s confounding of a f f e c t and ideation. variable influences job s a t i s f a c t i o n .  Suppose that we f i n d that some  Does t h i s mean that emotions have been  affected, b e l i e f s have been affected, or both?  In the absence of unconfounded  measurement we cannot t e l l . In summary, i t i s l i k e l y that job s a t i s f a c t i o n as an organizational outcome taps some of the emotional response to work.  But because job  s a t i s f a c t i o n i s an impure measure of emotional response i t i s unclear exactly what job s a t i s f a c t i o n t e l l s us about emotions on the job. are the sources of emotions on the job? satisfaction?  For example, what  Do they match the dimensions of job  That i s , are pay, promotions, supervision, coworkers, or the  work i t s e l f sources of emotions at work?  Are some job dimensions more l i k e l y  to arouse an emotional response than others?  When we examine the dimensions  of job s a t i s f a c t i o n i t seems evident that we should exclude non-work causes. The same may not be true for job emotions. individual's  It makes sense that an  f e e l i n g state o f f the job i s l i k e l y to a f f e c t h i s or her f e e l i n g  state on the job.  The l i t e r a t u r e on job stress recognizes t h i s : non-work-  related s t r e s s f u l incidents can have effects on the job.  It i s clear that  while a f f e c t i v e variables such as job s a t i s f a c t i o n have received much study, they cannot t e l l as much as we might l i k e about emotions on the job.  Future  organizational research on emotion per se would help answer these and other  37  questions.  Emotional expression as part of the work r o l e The influence of r o l e expectations on the expression of emotion i s a second example of the treatment of emotion as a dependent v a r i a b l e i n organizations.  Although the expression of feelings i s affected by rules i n  a l l contexts, t h i s approach has p a r t i c u l a r relevance to organizational behavior.  The workplace contains unique and powerful interpersonal,  organizational, and economic forces that author such " f e e l i n g r u l e s " . Hochschild (1983) c a l l s the management of emotion as a condition of. employment "the commercialization of human f e e l i n g " .  She examined the suppression and  d i s t o r t i o n of f e l t emotions among f l i g h t attendants. Most recently, Rafaeli and Sutton (1987) have proposed a conceptual framework f o r the expression of emotion as part of the work r o l e .  This  framework i s limited to the d i s p l a y or expression of emotion i n contrast to the subjective experience of i n d i v i d u a l f e e l i n g s .  The emotional displays that  Rafaeli and Sutton analyze are "control moves" (Goffman, 1969), or "the intentional e f f o r t of an informant to produce expressions that he thinks w i l l improve h i s s i t u a t i o n i f they are gleaned by the observer" (Goffman, 1969, p. 12).  R a f a e l i and Sutton point out that employees express emotions to promote  the interests of others as well, such as c l i e n t s , superiors, coworkers and subordinates. Rafaeli and Sutton's framework has three parts: the sources of r o l e expectations, the c h a r a c t e r i s t i c s of displayed emotions, and the outcomes of displayed emotions for the organization and the i n d i v i d u a l .  They propose two  sources for the r o l e expectations that create, influence and maintain the expression of emotion.  These they c a l l organizational context and emotional  38  transactions.  The organizational context comprises  the organizational  practices that influence the d i s p l a y of emotion: recruitment and selection, s o c i a l i z a t i o n , and rewards and punishments.  The feedback and cues that  employees receive from the "targets" of emotional expression, or the emotional transactions themselves,  are another source of r o l e expectations.  Customers,  c l i e n t s , superiors, coworkers, and subordinates can provide verbal and nonverbal cues that influence expression. The expression of emotion i t s e l f , as a consequence of r o l e expectations, i s the second component of Rafaeli and Sutton's framework. dimensions of emotions conveyed on the job.  They offer two  F i r s t , they posit that expressed  emotions can be placed on a continuum from p o s i t i v e through neutral to negative.  Smiling i s how p o s i t i v e affect i s expressed, negative a f f e c t  includes frowning.  Second, emotions are said to vary i n the extent to which  "they enhance the self-esteem of the target.  That i s , jobs may require the  expression of emotion to support others ( s o c i a l workers, for example), other jobs require n e u t r a l i t y (judges and referees), and some ask employees to degrade the self-esteem of others (e.g., d r i l l sargeants,  immigration  officials). The f i n a l component of the framework concerns the outcomes of expressed emotion.  Both p o s i t i v e and negative outcomes are possible for the  organization and for the i n d i v i d u a l .  Organizations benefit, f o r example, when  the p o s i t i v e emotional expression results i n increased sales.  Hochschild  (1983), f o r example, emphasizes the negative impact of emotion work on psychological and physical well-being.  She argues that displayed emotions are  harmful when they are inconsistent with true f e e l i n g s .  The r e s u l t can be a  loss of emotional control, physical i l l n e s s , substance abuse, and absenteeism. The l i t e r a t u r e on burn out (Maslach, 1978) also supports the costs of  39  emotional labor.  Rafaeli and Sutton point out, though, that expressing r o l e  appropriate emotion can have p o s i t i v e e f f e c t s .  Learning how to express  appropriate emotion may protect physicians, f o r example, from the negative e f f e c t s of f e l t emotion.  Workers who learn p o s i t i v e emotion management on the  job may benefit o f f the job. F i n a l l y , Rafaeli and Sutton provide a provocative discussion of the relationship between expressed emotion and experienced emotion, assessing the consequences of the mismatch between expression, experienced or "true feelings", and f e e l i n g r u l e s . Rafaeli and Sutton's contribution to the examination of emotion i n organizations i s s i g n i f i c a n t .  They have delineated how emotional display can  a r i s e from organizations and why: emotional display has outcomes that are salient to the organization and to individuals i n organizations.  Future  research can b u i l d on the foundation they have provided, as has already begun (e.g., Sutton & Rafaeli, 1986; Denison & Sutton, i n press), and should lead to refinement of their framework. At present, though, and i n a n t i c i p a t i o n of more empirical development, t h e o r e t i c a l refinements to the conceptual framework proposed by Rafaeli and Sutton are possible. clarification.  Of greatest immediate importance i s conceptual  Role required emotional display i s best understood to be those  actions that are under a person's v o l i t i o n a l control, the emotions that one chooses to display.  This i s c e r t a i n l y part of the conception of expressed  emotion as control moves or "intentional e f f o r t " , as offered by Rafaeli and Sutton.  Such a conception implies that emotional display i s functionally  independent  of emotional experience and that one can choose which emotion to  express separate from one's f e e l i n g s . At times i n their discussion, however, Rafaeli and Sutton do not maintain  40  the d i s t i n c t i o n between displayed and f e l t emotion.  For example, they say  that expressed emotions can be characterized as p o s i t i v e or negative depending on how they appear.  They also say, though, that emotions with the same  appearance can be either p o s i t i v e or negative depending on how they are experienced by the r o l e actor.  They c i t e t h i s ambiguity as a l i m i t a t i o n of  the p o s i t i v e — n e g a t i v e continuum.  By adhering to a conceptualization that  separates emotional expression as i t appears, from emotional experience as i t f e e l s , t h i s ambiguity can be avoided.  Rather than confusing the  conceptualization of the construct of expressed emotion by allowing for i t to have d i f f e r e n t meaning depending on the experience of the encoder, c l a r i t y can be gained by constraining the construct to mean emotions as expressed to and ascribed meaning by a decoder.  The point of view of the object of expression  rather than the subject should be taken. The second dimension of expressed emotion proposed by Rafaeli and Sutton, whether the display enhances or degrades the esteem of the object of the emotion, takes such a point of view.  R a f a e l i and Sutton say that a l i m i t a t i o n  of t h i s dimension i s that individual differences moderate the effect on a person's self-esteem. A more serious c r i t i c i s m of this second dimension i s to ask "why t h i s distinction?"  What i s special about the effects of displayed emotion on an  observer's self-esteem?  Other factors could be named, such as those proposed  by Kemper (1978, 1984) as the essential dimensions of an interpersonal theory of emotion: power and status. effects on the r e c i p i e n t .  Displayed emotion i s l i k e l y to have many  In the absence of an exhaustive l i s t of such  effects or a rationale for choosing one, I suggest that no single effect of displayed emotion be put forward as e s p e c i a l l y c h a r a c t e r i s t i c . How, then, should the intentional display of role-required emotions be  41  characterized?  As R a f a e l i and Sutton note, f a c i a l expressions can be labelled  with great r e l i a b i l i t y and accuracy  (Ekman, 1984; Leathers and Emigh, 1980).  A l i s t of d i s c r e t e emotions associated with such expression might, therefore, serve as c h a r a c t e r i s t i c .  A number of researchers have proposed between s i x  and twelve monopolar factors of a f f e c t , such as sadness, anger, anxiety, e l a t i o n , and tension (e.g., Borgatta, 1961; Clyde, 1963; McNair & Lorr, 1964; Nowlis, 1965).  More recently, however, methodological  refinements have  supported the conclusion that a f f e c t , including f a c i a l expression, can be represented by the dimensions of pleasure/displeasure and arousal/sleepiness. These could serve as c h a r a c t e r i s t i c dimensions of expressed emotion.  That i s ,  the range of required emotional display could be expressed as combinations of these two dimensions.  Emotion as an independent variable The e f f e c t of emotion on variables i n organizations has been studied i n only a limited way.  As with the study of emotion as a dependent v a r i a b l e ,  organizational scholars may be unsure what emotion i s . studies, which represent  In the following  some of the studies of the e f f e c t s of emotion i n  organizations, we s h a l l see emotion conceptualized loosely as expression or display, similar to that discussed i n the previous section, and with  little  consideration for the process by which such expression has an e f f e c t .  We  s h a l l also review studies that have attempted to use the construct of mood to shed l i g h t on the consequences of job s a t i s f a c t i o n for performance, and which use job s a t i s f a c t i o n as a proxy for mood.  As suggested above, such use i s  inappropriate because job s a t i s f a c t i o n confounds a f f e c t with ideation. F i n a l l y , we w i l l review the few studies that have used the construct of mood directly.  These have examined i t s implications for performance appraisal i n  42  organizations.  The expression of a f f e c t i n decision-making Guzzo and Waters (1982) examined the expression of a f f e c t i n d e c i s i o n making groups.  The considered what happens when groups are c a l l e d on to make  decisions about problems that are l i k e l y to e l i c i t emotional responses.  They  c i t e past propositions that emotional expression i s detrimental to e f f e c t i v e decision-making: that high l e v e l s of emotion hinder clear thinking and communication (Maier, 1963), that group members may lack the s k i l l s necessary for dealing e f f e c t i v e l y with a f f e c t i n groups (Argyris, 1966), and that norms against expressing emotion may suppress the sharing of information important for decision-making  (Argyris, 1966; Maier, 1963).  Guzzo and Waters examined how the timing of the expression of a f f e c t influenced decision-making performance.  Maier (1963) prescribed the  expression of a f f e c t early i n the decision-making process.  Guzzo and Waters  found, however, that groups that followed instructions to delay expression made better decisions than those that expressed i t early.  The unregulated  expression of affect among control subjects led to decisions of intermediate quality.  Guzzo and Waters speculated that early expression of a f f e c t may  drain the group of productive energy that might otherwise be spent generating alternatives, or that the group's focus of attention may be r e s t r i c t e d to a narrow range of a l t e r n a t i v e s . Of additional interest i s Guzzo and Waters' finding that group participants instructed not to express a f f e c t apparently ignored these instructions.  Guzzo and Waters concluded that expression of a f f e c t i s  unavoidable when an emotionally arousing problem exists, that some expression of f e e l i n g i s essential to decision-making, and that norms against such  43  expression cannot be completely enforced i n some settings. It should be noted that Guzzo and Waters used a decision problem l i k e l y to create negative a f f e c t , and about which individuals could be expected to f e e l anxiety and displeasure. Negative mood has been related to narrowed focus of attention (Easterbrook, 1959), thus delay of i t s expression may have been b e n e f i c i a l to the generation of a l t e r n a t i v e problem solutions. mood, i n contrast, can f a c i l i t a t e decision-making  Positive  (e.g., Isen, Daubman, &  Gorgoglione, i n press; Mednick, 1962; Mednick, Mednick, & Mednick, 1964).  Had  the decision problem been one involving the expression of p o s i t i v e emotion early rather than l a t e expression may have led to better decisions.  Isen and  colleagues point out though, that while mood influences the strategies used to process information, whether such strategies are e f f e c t i v e depends on the nature of the task (Isen, Means, Patrick, & Nowicki, 1982; Isen & Means, 1983).  Job s a t i s f a c t i o n as mood Some researchers have attempted  to show evidence of a job s a t i s f a c t i o n —  performance linkage by equating job s a t i s f a c t i o n with mood.  According to  Motowidlo: Although mood i s conceptually and operationally d i f f e r e n t from job s a t i s f a c t i o n the two constructs are s t i l l intimately related. People who f i n d their work situations s a t i s f y i n g should generally be i n more p o s i t i v e moods than people who do not. Accordingly, the a f f e c t i v e response i m p l i c i t i n job s a t i s f a c t i o n would have causal e f f e c t s on behaviors at work (p. 911). Motowidlo (1984) found that job s a t i s f a c t i o n was associated with the performance of considerate job behaviors.  In a similar vein, Bateman and  Organ (1983) argue that p r o s o c i a l behaviors are most l i k e l y to occur when a person experiences a generalized mood state characterized by p o s i t i v e a f f e c t . To the extent that job s a t i s f a c t i o n , as conventionally  44  measured, r e f l e c t s t h i s p o s i t i v e a f f e c t i v e state, i t i s l i k e l y that more s a t i s f i e d persons display more of the p r o s o c i a l , c i t i z e n s h i p behaviors (p. 588). Bateman and Organ found s i g n i f i c a n t relationships between s a t i s f a c t i o n and prosocial organizational behavior, although the results of a cross-lagged panel regression analysis did not support the predicted d i r e c t i o n of causality.  Bateman and Organ speculate as to why a causal connection between  "mood" and c i t i z e n s h i p behavior was not found, including the p o s s i b i l i t y of methodological weakness or a common antecedent v a r i a b l e .  I t should be noted  that cross-lagged analyses have been c r i t i c i z e d as not v a l i d l y demonstrating causal effects (Rogosa, 1980). Two a d d i t i o n a l explanations are possible. may not r e f l e c t mood.  F i r s t i s that job s a t i s f a c t i o n  As we said, job s a t i s f a c t i o n confounds emotional and  ideational reactions to work. generally conceptualized  Further, mood, as defined e a r l i e r and as  i n studies of i t s e f f e c t on behavior, is*understood  to be r e l a t i v e l y variable over time.  Moods s h i f t and change.  Mood  measurement, by adjective c h e c k l i s t s or rating scales, usually specify current mood or, for example, "your feelings right now".  Job s a t i s f a c t i o n , on the  other hand, i s generally understood to be f a i r l y stable.  One would not expect  job s a t i s f a c t i o n to vary from one part of the day to another. A second explanation  for Bateman and Organ's r e s u l t i s suggested by the  f i n d i n g that p o s i t i v e mood does not always lead to prosocial behavior.  Recall  that while i n d i v i d u a l s i n p o s i t i v e moods tend to perform p o s i t i v e behaviors they may avoid behaviors that threaten  to disrupt t h e i r good mood (Forest,  Clark, M i l l s , & Isen, 1979; Isen & Simmonds, 1978).  Thus the r e l a t i o n s h i p  between mood and organizational c i t i z e n s h i p behavior may be moderated by the perceived  consequences of those behaviors f o r the mood of the i n d i v i d u a l . The  obvious implication for organizations  i s that i t i s i n s u f f i c i e n t to promote  45  organizational c i t i z e n s h i p by improving the a f f e c t i v e state of workers when the consequences of organizational c i t i z e n s h i p are aversive.  Mood and performance appraisal Perhaps because of the l i n k s between the performance appraisal l i t e r a t u r e and theories of s o c i a l cognition, examination of the e f f e c t s of mood on performance appraisal have been more appropriately conceptualized.  Park,  Sims, and Motowidlo (1986) reviewed models of the r e l a t i o n s h i p between a f f e c t and cognition and considered  implications for organizational behavior.  In  p a r t i c u l a r , they focused on the r o l e of a f f e c t i n performance appraisal processes.  Because performance appraisal involves a complex of memory-based  cognitive tasks, a f f e c t i s l i k e l y to be an important f a c t o r .  They draw on  models such as Bower's (1981), .in which mood biases the encoding and r e c a l l of information.  For example, individuals i n p o s i t i v e moods are said to be more  l i k e l y to attend to and remember p o s i t i v e information. mood of a rater w i l l influence how This was,  This implies that the  performance i s rated.  i n f a c t , what S i n c l a i r ( i n press) demonstrated.  Sinclair  provided raters with information about the performance of a target i n d i v i d u a l , experimentally  manipulated the mood of the raters, and then asked them to  evaluate the target i n d i v i d u a l . Raters i n a p o s i t i v e mood provided more favorable evaluations and r e c a l l e d more p o s i t i v e information than did raters i n a negative or neutral mood.  However, consistent with the argument that  elated individuals make broader categorizations (Easterbrook,  1959;  Isen &  Daubman, 1984), raters i n a depressed mood made more accurate evaluations displayed less halo error. Many of our theories of i n d i v i d u a l l e v e l organizational behavior are  and  46  based on i n d i v i d u a l s as a c t i v e , gatherers and processors.  although perhaps bounded, information  The effects of mood on performance appraisal are  suggestive of a variety of s i g n i f i c a n t implications for s o c i a l judgements i n organizations.  S i m i l a r l y , differences i n the r e c a l l and use of information as  a result of mood has obvious implications for such topics as organizational decision making, the use of personal interviews i n personnel selection, setting,  and cognitive theories of motivation.  goal  C l e a r l y , the effect of emotion  i n general and mood i n p a r t i c u l a r i n these areas i s worthy of i n v e s t i g a t i o n .  47  IV.  AFFECT AND MOTIVATION  Individual l e v e l job performance continues to be the v a r i a b l e of primary interest to the micro side of organizational behavior.  As Staw (1984) states,  theories of work motivation can be considered most d i r e c t l y relevant to performance, which they are devoted p r i m a r i l y to p r e d i c t i n g .  Following  Campbell and Pritchard (1976) and M i t c h e l l (1982), motivation i s defined as the psychological processes that explain the d i r e c t i o n , amplitude,  and  persistence of purposeful voluntary behavior.  Expectancy  Theory  Expectancy or valence—instrumentality—expectancy (VIE) formulations have dominated theories of work motivation.  Expectancy  formulations agree  that motivational force i s the product of (a) the expectancy  that e f f o r t w i l l  lead to task accomplishment, (b) the'instrumentality between task accomplishment and i t s outcomes, and (c) the valence of the outcomes. Although c r i t i c i z e d on conceptual and empirical grounds, expectancy continues to have the support of motivation theorists.  theory  This support may  be  because i t has strong i n t u i t i v e appeal, or because the formulation of VIE theory makes i t resistant to empirical r e b u t t a l .  For example, although  Campbell and Pritchard (1976) report a maximum c o r r e l a t i o n of about  .30  between motivation and independent ratings of effort i n c o r r e l a t i o n a l f i e l d and experimental studies of the f u l l v a l e n c e — i n s t r u m e n t a l i t y — expectancy models, because VIE theory does not specify the content of outcomes to be considered such weak results can be defended on the grounds that researchers have not included i n the study a l l the outcomes relevant to the subjects (Mitchell, 1974).  On the other hand, the use of other experimental designs or  48  dependent measures can y i e l d highly supportive r e s u l t s .  Wanous, Keon & Latack  (1983) revealed an average within-subjects c o r r e l a t i o n between valence times instrumentality and organizational attractiveness of .72 across 16 withinsubjects expectancy studies.  Campbell and Pritchard (1976) conclude that  while the expectancy framework w i l l continue to have strong h e u r i s t i c value, i t s p r a c t i c a l value would be enhanced by greater attention to i t s t h e o r e t i c a l and empirical d i f f i c u l t i e s .  In p a r t i c u l a r they argue for i n depth  investigation of the i n d i v i d u a l components of VIE theory rather than tests of a " f u l l " model with " s u p e r f i c i a l measures of poorly understood v a r i a b l e s " (p. 95).  Further, they urge more research i n t o the process by which i n d i v i d u a l s  choose to expend a c e r t a i n l e v e l of e f f o r t .  What, they ask, i s expectancy and  how does i t r e l a t e to other variables? The goal of this d i s s e r t a t i o n i s to begin to answer t h i s question with emphasis on the emotional expectancy related?  state of i n d i v i d u a l s .  That i s , how are mood and  There i s c e r t a i n l y theoretical speculation supporting the  proposition that mood influences expectancy. Staw (1984), f o r example, points out that as a model of individual decision making expectancy theory i s subject to well known l i m i t a t i o n s on human information processing.  The careful screening of alternatives and  assessment of rewards that expectancy theory assumes may occur i n special circumstances  only.  Staw advocates increased emphasis on examination of the  h e u r i s t i c s and biases that influence decisions i n routine situations.  One of  these i s the a v a i l a b i l i t y h e u r i s t i c (Tversky and Kahneman, 1973), which proposes that, under conditions of uncertainty and limited information processing capacity, i n d i v i d u a l s estimate p r o b a b i l i t i e s of events occurring based on how e a s i l y the event comes to mind.  In other words, subjective  judgements of the p r o b a b i l i t y of an event depend on how a v a i l a b l e instances of  49  the event are i n memory.  To the extent that mood state cues or primes  material i n memory, making i t more accessible, mood should influence estimates of p r o b a b i l i t y .  Included among these i s the perceived p r o b a b i l i t y that e f f o r t  leads to task outcomes, that i s : expectancy.  Measuring Expectancy The construct expectancy i s , i n Vroom's (1964) o r i g i n a l formulation, a subjective p r o b a b i l i t y .  S p e c i f i c a l l y , i t i s the perceived p r o b a b i l i t y  associated with the strength of the relationship between i n d i v i d u a l e f f o r t and job outcomes, such as task performance.  According to Vroom, i t i s the  perceived covariation between e f f o r t and performance.  E f f o r t i s , of course,  only one possible manifestation of motivation. Vroom's formulation has also been applied to job choice and job persistence.  The discussion here i s  constrained to the case of e f f o r t , although similar l i n e s of argument can be developed for choice and persistence. Vroom described expectancy as the degree to which a person believed that high e f f o r t would lead to high performance.  This description may explain the  high degree of consensus i n how expectancy has been measured (Mitchell, 1974). For example, Garland (1984) measured "expectancy" by asking individuals to estimate the l i k e l i h o o d that they could beat a specified performance standard. Similar measurements of "expectancy" have asked subjects to indicate the p r o b a b i l i t y that they could perform an intended behavior (Snyder, Howard, & Hammer, 1978).  Others have asked for ratings of the relationship between  working hard and performing well (Muchinsky,  1977).  In spite of apparent consensus, t h i s way of measuring expectancy i s not true to the o r i g i n a l concept.  Expectancy i s , and should therefore be measured  as, the covariation between e f f o r t and performance.  That i s ,  expectancy  50  should comprise the perceived p r o b a b i l i t y that high l e v e l s of e f f o r t lead to high l e v e l s of performance, as well as that low levels of e f f o r t lead to low levels of performance, and that low levels of e f f o r t do not lead to high levels of performance or v i c e versa.  In other words, expectancy measurement  should span the rows and columns of the perceived  effort-performance  covariance matrix. This i s not a new  admonition.  Hollenbeck (1984) proposes a "matrix  method" for expectancy research that includes measurement of expectancy at several l e v e l s of e f f o r t and performance.  Hollenbeck's method also proposes,  moreover, that instrumentality be measured as the r e l a t i o n s h i p between l e v e l s of performance and l e v e l s of relevant outcomes.  Instrumentality  should  comprise the l i n k between each of high, medium, and low performance and  high,  medium, and low l e v e l s of outcomes, such as wages and feelings of accomplishment.  Further, valence should be measured for-each l e v e l of  relevant outcomes.  Matrix m u l t i p l i c a t i o n can then be applied to calculate the  motivational force associated with each l e v e l of the relevant outcomes. Hollenbeck's method makes at least two  s i g n i f i c a n t contributions.  First  i s the recognition that expectancy should be measured at multiple l e v e l s . Second i s the demonstration that motivational force exists for l e v e l s of relevant outcomes rather than some o v e r a l l outcome state.  In other words, the  resultant of expectancies between levels of e f f o r t and performance, instrumentalities between levels of performance and outcomes, and valences of l e v e l s of outcomes, i s a motivational force score, or propensity to d i r e c t e f f o r t , toward each of a number of performance l e v e l s .  Hollenbeck points out  that force scores are nothing more than the valences of p a r t i c u l a r e f f o r t levels.  Unfortunately,  no t h e o r e t i c a l rationale exists for r e l a t i n g e f f o r t to  these multiple force scores.  Say,  for example, that a person has force scores  51  of  10 and 9 for high e f f o r t and low e f f o r t .  The t r a d i t i o n a l maximization  predicts that people " w i l l exhibit the single l e v e l of e f f o r t that to the single strongest force score" (Hollenbeck, 1984,  p.  586).  rule  corresponds Yet  we  would be u n l i k e l y to predict the same behavior for a person with respective scores of 10 and 1.  Hollenbeck points out that "just as the valence of high  performance has no p a r t i c u l a r meaning u n t i l i t i s compared to the valence of low performance, the force of high e f f o r t has no p a r t i c u l a r meaning u n t i l i t i s compared to the force of low e f f o r t " (p.  586).  Hollenbeck  suggests a  p r o b a b i l i s t i c interpretation of the l i n k between force and e f f o r t , i n which individuals exert e f f o r t i n proportion to the r e l a t i v e attractiveness of d i f f e r e n t levels of e f f o r t . d i s t r i b u t i o n of e f f o r t over  Multiple force scores thus predict the time.  Kennedy, Fossum, and White (1983) describe the procedure of obtaining motivational force scores for several e f f o r t l e v e l s as the "choice model": i t predicts that individuals choose the e f f o r t l e v e l with the highest product of expectancy, instrumentality, and valence.  They c i t e a number of reasons  supporting the choice model as the model of choice. Primarily, the choice model i s the one Vroom intended.  Kennedy and  colleagues c i t e Vroom's example of three i n d i v i d u a l s : the f i r s t has high expected payoffs for both high and low e f f o r t , the second has high  expected  payoffs for high e f f o r t but not for low e f f o r t , and the t h i r d has high expected payoffs for neither high nor low e f f o r t .  "The second person i s the  only one predicted to choose high e f f o r t , because only he benefits from putting forth high e f f o r t .  Choice and single alternative (high e f f o r t )  predictions would be contradictory, since the single-alternative model would have had the f i r s t person also choosing high e f f o r t " (Kennedy et a l . , 1983, 125).  This issue i s at the heart of the argument that expectancy i s a  p.  52  "within-subjects theory" (Mitchell, 1974).  Expectancy theory predicts how  individuals w i l l choose from among a set of e f f o r t levels at which l e v e l of e f f o r t he or she desires to work.  Kennedy and colleagues add that the  strength of a person's motivation at one l e v e l of e f f o r t i s only meaningful when compared to that person's motivation f o r other e f f o r t l e v e l s .  This i s  the same as saying that, a l l other things being equal, one person's expectancy for one l e v e l of e f f o r t must be measured alongside h i s or her expectancies f o r other e f f o r t l e v e l s .  Expectancy theory thus requires assessment within  i n d i v i d u a l s of the degree to which each of a set of e f f o r t levels i s perceived to lead to each l e v e l of set of outcomes (Mitchell, 1974; Muchinsky, There has been much debate about this "within/between issue".  1977). Some  researchers have argued that a l l between-subjects tests of expectancy theory are i n v a l i d .  This position rests on two objections: that between-subjects  studies do not take into account i n d i v i d u a l differences, and that betweensubjects studies do not assess the choices that individuals make between e f f o r t l e v e l s , tasks, or outcomes.  The f i r s t objection i s , I propose, not  founded on a requirement of expectancy theory but i s instead one of r e l a t i v e methodological advantage.  I t i s an issue about the fact that factors which  influence motivation may not be included i n the expectancy theory model and so, unless stated, measured, and accounted for are included as error variance. Within-subjects analyses use each subject as his or her own control, thus c o n t r o l l i n g for such unmeasured v a r i a t i o n .  However, a between-subjects study  which completely measured such i n d i v i d u a l difference variables and controlled for them would, i n p r i n c i p l e , y i e l d the same r e s u l t s as a comparable withinsubjects analysis. measurement.  Note that the fundamental issue here i s one of  A within-subjects study which i m p l i c i t l y matches subjects across  levels of the variable of interest w i l l always outperform a between-subjects  53  study i n which a l l the factors relevant to the variable of interest are not measured.  Empirical comparisons of within- versus across-subjects designs  have generally revealed greater effects for the former (e.g. & White, 1983; Muchinsky, 1977; Wanous, Keon & Latack, 1983).  Kennedy, Fossum But this r e s u l t  w i l l hold for most theories, and as such does not i n v a l i d a t e between-subjects tests of expectancy theory.  I t merely demonstrates that when v a r i a t i o n  a t t r i b u t a b l e to factors extraneous to the theory i s reduced, the theory predicts more of the measured v a r i a t i o n . We should also note that the finding that within-subjects tests of expectancy theory outpredict between-subjects tests implies that the model i s underspecified.  That i s , i t omits e x p l i c i t statement of components, probably  related to i n d i v i d u a l differences, that have predictive v a l i d i t y .  (Few  theories are completely specified, though, and most would be too unwieldy i f we attempted i t . ) However, some w i t h i n — s u b j e c t s tests of expectancy have used performance rather than motivation as the c r i t e r i o n . proports to predict motivation, not performance.  Expectancy theory  It i s generally held that  performance i s the resultant of motivation and other factors, including ability.  So when performance i s used as the predictive c r i t e r i o n i n  expectancy theory tests, within-subjects studies are bound to outperform between-subjects studies.  Individual differences i n a b i l i t y , among other  factors not included i n expectancy theory, are being accounted f o r .  In using  a performance c r i t e r i o n as the proxy for motivation within-subjects studies are improperly ascribed some t h e o r e t i c a l ascendance. The second objection to between-subjects studies, that they v i o l a t e assumptions about the nature of expectancy theory i s , I propose, founded on improper conceptualization and operationalization of the expectancy construct. As we have discussed, expectancy should be conceptualized as e f f o r t -  54  performance covariation, and should be measured as the perceived p r o b a b i l i t y , within i n d i v i d u a l s , that e f f o r t leads to performance across multiple levels of e f f o r t and performance.  This captures the essence of expectancy theory that  people choose high effort as opposed to low e f f o r t when the former leads to performance.  By understanding and measuring expectancy as within-subjects  effort-performance covariation the within-subjects aspect of expectancy theory, that individuals compare choices about actions and their i s maintained.  consequences,  It i s then possible to make between subjects predictions, such  as that individuals i n work roles where objective effort-performance covariation i s higher w i l l perceive greater expectancy (and hence be more motivated) than w i l l other individuals i n work roles with lower objective effort-performance covariation.  Without the c a p a b i l i t y of making predictions  across people, expectancy theory seems to have few implications for job design.  Mood and Expectancy Empirical results suggest a r e l a t i o n s h i p between mood and expectancy. The f i r s t such suggestion comes from the l i t e r a t u r e on depression rather than mood.  A l l o y and Abramson (1982) compared the judged contingency between  responses and outcomes of self-selected depressed and non-depressed individuals.  Students were asked to judge the degree of control their  responses exerted over outcomes.  A l l o y and Abramson found that subjective  representations of contingency generally mirror objective contingencies. However, when responses and outcomes are noncontingently related, nondepressed i n d i v i d u a l s f a l s e l y i n f e r control from reinforcement.  In other  words, non-depressed individuals overestimate the contingency between their actions and outcomes.  Depressed individuals on the other hand, are r e a l i s t i c  55  i n their judgements whether or not reinforcement i s noncontingently high or low.  Apparently non-depressed  individuals f a l s e l y take c r e d i t f o r t h e i r  "success". While d i s p o s i t i o n a l l y depressed individuals are l i k e l y to experience depressed mood, i t i s also possible that they possess d i s p o s i t i o n a l differences i n cognitive s t y l e . mood states, may be responsible outcome r e l a t i o n s h i p s .  These, rather than differences i n transient for differences i n judgements of response-  To assess t h i s , A l l o y , Abramson and Viscusi (1981)  tested the impact of induced mood on contingency judgements.  They found that  induction of depressive a f f e c t i n n a t u r a l l y non-depressed individuals resulted i n reduced judgements of control over non-contingent outcomes.  Induction of  elation among naturally depressed individuals resulted i n increased of control.  judgements  Thus transient mood states do influence the r e l a t i o n s h i p between  objective and perceived response-outcome contingency when no contingency exists. Alloy, Abramson, and V i s c u s i (1981) do not specify the mechanism by which mood state influences outcomes.  Differences  judgements of the contingency between responses and i n the a c c e s s i b i l i t y of events for r e c a l l provides one  such mechanism (Jennings, Amabile, & Ross, 1982).  As we have said already,  Kahneman and Tversky propose that people judge the p r o b a b i l i t y of events based on how e a s i l y exemplars of the event come to mind, that i s , how available they are i n memory.  Research on how people construct  judgements of the contingency  between variables suggests that, when the event i s desirable (e.g., successful task performance), and when co-occurrences of the variables are presented sequentially ( i . e . , rather than i n tabular form), individuals display a bias i n favor of the o v e r a l l p r o b a b i l i t y of the outcome (Allan & Jenkins, 1980; A l l o y & Abramson, 1979; A l l o y & Tabachnik, 1984; Jenkins & Ward, 1965;  56  Wasserman, Chatlash & Neunaber, 1983).  That i s , estimates of the covariation  between a response and an outcome are based on the perceived l i k e l i h o o d of the outcome occurring.  Thus, to the extent that mood state primes the r e t r i e v a l  from memory of desirable task outcomes, i n d i v i d u a l s ' judgements of the contingency between responses (such as e f f o r t ) and task performance w i l l be affected. Studies of self-esteem also suggest an effect of mood on expectancy. C l i n i c a l theories of depression (Beck, 1967) central component of depressive symptoms.  include low self-esteem as a  At the same time, people with low  self-esteem have been shown to have lower expectancies Campbell & Fairey, 1985; what combination  (Brockner,  1979;  Coopersmith, 1967), but i t i s unclear whether or i n  this effect i s due to the ideational as opposed to emotional  components of self-esteem.  That i s , individuals with high d i s p o s i t i o n a l s e l f -  esteem may believe themselves to be more competent, more able to accomplish goals, or they may be predisposed to more favorable mood states, which may bias estimates of their competence.  Conversely, people with low self-esteem  may be more susceptible to negative mood and subsequently have lowered expectations for success-.  The Beck depression inventory (Beck, 1967), which  measures self-esteem (Brockner & Guare, 1983) perceived f a v o r a b i l i t y of future events.  taps mood state as well as  In sum,  the influence of self-esteem  on expectancies i s consistent with a mood effect but may be the r e s u l t of a non-affective cognitive mechanism. More d i r e c t evidence comes from studies that have induced mood.  Brown  (1984) found that subjects i n a p o s i t i v e mood state were more confident of future task success.  After task performance that was  successful or  nonsuccessful this expectation remained s i g n i f i c a n t l y higher i n p o s i t i v e a f f e c t subjects and was higher for subjects experiencing success.  57  Furthermore, the interaction of mood and task outcome was s i g n i f i c a n t .  After  success individuals i n the elated condition had more p o s i t i v e a n t i c i p a t i o n for future success than individuals who f a i l e d , while individuals i n a negative mood d i d not show t h i s difference. Teasdale and Spencer (1982, 1984) investigated  the effects of mood on  estimates of past success and p r o b a b i l i t y of future success.  Subjects  completed 72 t r i a l s of an "unconscious-decision-making task", i n which they had  to choose one of a pair of words and received  50% random feedback as to  whether they had chosen "correctly" or " i n c o r r e c t l y " . depressed or elated mood was induced. of the number of successful  After the task  Unhappy subjects gave lower estimates  t r i a l s than did elated subjects.  Similarly,  estimates of the p r o b a b i l i t y of future success were lower f o r subjects i n whom depressed mood had been induced. Wright and Mischel (1982) also found that p o s i t i v e mood increased expectations of success on tasks as well as s a t i s f a c t i o n with previous performance and self-rated a b i l i t y .  Individuals  described themselves i n more p o s i t i v e terms.  i n a p o s i t i v e mood also  A successful  task outcome also  improved expectations f o r the future and s a t i s f a c t i o n with performance. Although the interaction effect of mood and outcome on s a t i s f a c t i o n f e l l  just  short of significance, i t was apparent that negative mood state had l i t t l e effect on s a t i s f a c t i o n after success, but exacerbated d i s s a t i s f a c t i o n after failure.  It i s l i k e l y that successful performance helped counteract the  effect of negative mood on s a t i s f a c t i o n .  Wright and Mischel point out though,  that feedback about success did not overcome the effect of mood on performance expectations.  Despite repeated feedback i n d i c a t i n g good performance, negative  a f f e c t subjects repeatedly underestimated expected performance..  Similarly,  p o s i t i v e - a f f e c t subjects did not adjust their expectations to match the  58  negative feedback they received. A l l o y , Abramson & V i s c u s i , 1981)  In contrast to A l l o y and Abramson (1982; negative a f f e c t subjects were not more  accurate than non-depressed subjects.  This incongruity may,  differences between the measures of expectations  though, r e f l e c t  used.  Wright and Mischel's findings are consistent with the operation of an a c c e s s i b i l i t y mechanism as underlying mood e f f e c t s , perhaps e s p e c i a l l y under conditions of uncertainty.  They found that the estimates of the frequency of  p o s i t i v e task outcomes of p o s i t i v e - a f f e c t subjects were s i g n i f i c a n t l y higher than those of negative-affect subjects. r e c a l l of successful outcomes.  P o s i t i v e - a f f e c t f a c i l i t a t e d the  Subjects i n the successful outcome condition  also r e c a l l e d more successful outcomes.  In p a r t i c u l a r , though, a f t e r  experiencing p o s i t i v e outcomes, negative-affect subjects underrecalled their p o s i t i v e outcomes.  S i m i l a r l y , a f t e r f a i l u r e p o s i t i v e - a f f e c t subjects  overrecalled their p o s i t i v e outcomes. In summary, mood appears to be an important influence on i n d i v i d u a l judgements under uncertainty. e f f e c t on expectancies  While i t thus warrants examination for i t s  i t has not received such attention as yet.  Feather  (1984a), i n summarizing the present status and future d i r e c t i o n s of expectancy theory, says that while a f f e c t has not been neglected i t c e r t a i n l y needs increased consideration.  "When a f f e c t does appear i n expectancy-value models  i t often seems to appear almost as an appendage or epiphenomenon, rather than as an i n t e g r a l part of the theory"  (1984a, p.  406).  Most often affect i s  considered as a coincidental outcome of achievement, such as i n Weiner and colleagues' analysis of the a f f e c t i v e consequences of success and (Weiner, Russell & Lerman, 1978,  1979;  c.f., Feather,  1984b).  failure  59  Research Hypotheses  The preceding review of theory and research, supports investigation of the effect of mood on memory and motivation.  In the following sections,  s p e c i f i c hypotheses about the relationship between mood states, task perceptions, and individual differences are proposed.  P r i n c i p a l among these  i s the effect of mood on subjective effort-performance covariation, i . e . , expectancy.  Hypothesis One: Mood and Expectancy Individuals i n a pleasant mood w i l l report higher subjective e f f o r t performance covariation than w i l l individuals i n a negative mood.  The' studies reviewed above (e.g., A l l o y & Abramson, 1979; Alloy, Abramson & V i s c u s i , 1981; Brown, 1984; Wright & Mischel, 1982) are consistent with and highly suggestive of an effect of mood on expectancy. used the term "expectancy",  Although they have even  they have not examined expectancy as  conceptualized i n theories of work motivation, that i s , as the subjective b e l i e f that performance and e f f o r t covary.  A l l o y and Abramson (1979) and  A l l o y , Abramson and Viscusi (1981), for example, examined the effect of mood on the perceived contingency between pressing or not pressing a button and the onset of a l i g h t .  Brown (1984) and Wright and Mischel (1982) assessed mood  effects on expectations that task performance would be successful, as opposed to on expectations that task performance was a function of task e f f o r t . The mechanism proposed to underlie t h i s mood effect i s enhanced a c c e s s i b i l i t y i n memory: the mood state of an individual cues for r e c a l l material i n memory that i s a f f e c t i v e l y congruent with that mood state.  When  60  i n a p o s i t i v e mood, individuals are more l i k e l y to remember p o s i t i v e task outcomes, such as successful task performance.  To the extent that individuals  perceive themselves as working hard, t h i s increased a c c e s s i b i l i t y of r e c o l l e c t i o n s of high performance w i l l increase i n d i v i d u a l s ' perception of the connection between working hard and performing well. To state i t another way, consider a model of individuals' representations of  the r e l a t i o n s h i p between e f f o r t and performance as a contingency  l i k e that i n Figure 3.  matrix,  Each c e l l contains the number of "co-occurrences" of  the marginals, such as the number of times that high performance and high e f f o r t co-occur, or each c e l l can be taken to contain the p r o b a b i l i t y that the marginals w i l l coincide.  Thus, c e l l a, for example, i s the p r o b a b i l i t y that  high e f f o r t and high performance coincide.  Covariation between e f f o r t and  performance increases as the size of c e l l s a and d increase r e l a t i v e to c e l l s b and c. Thus perfect covariation i s when low e f f o r t only and always leads to low performance and high e f f o r t only and always leads to high performance. According to the mechanism we have described, i n which mood influences the r e c a l l of congruent memories, when individuals are i n a p o s i t i v e mood, they are more l i k e l y to remember the occurrences i n c e l l s a and b, or to r a i s e their estimates of the p r o b a b i l i t y that the marginals of those c e l l s coincide. Conversely, when i n a negative mood, individuals w i l l r a i s e their estimates of the magnitude of c e l l s c and d.  When individuals see themselves as working  hard, then, p o s i t i v e mood i n f l a t e s c e l l a.  Negative mood, by the same  reasoning, i n f l a t e s c e l l c. Objectively, we know that covariation depends on how much bigger c e l l a i s than c e l l c, although also on how much bigger c e l l d i s than c e l l b.  Research evidence suggests that individuals combine  contingencies to obtain subjective estimates of covariation i n two ways.  One  way i s by comparing the incidence of c e l l s a and d, the "confirming cases", to  61  Effort HIGH  LOW  HIGH  LOW  Figure 3. Contingency table model of the relationship between e f f o r t and performance.  62  that of c e l l s b and c, the "disconfirming cases" (Shaklee and Tucker, 1980). In t h i s way, a l l four c e l l s are used as i n the c a l c u l a t i o n of r e a l or objective covariation. d, the confirming  A l t e r n a t i v e l y , i n d i v i d u a l s focus only on c e l l s a and  cases (Ward and Jenkins,  1965), and perhaps even only on  c e l l a (Smedslund, 1963). Which of these ways best describes how subjective perceptions of covariation are formed has yet to be resolved.  But i n each of them  covariation depends on the r e l a t i v e magnitude of c e l l a.  To the extent then  that p o s i t i v e mood i n f l a t e s c e l l a by influencing the a c c e s s i b i l i t y of memories of high performance, perceived  e f f o r t performance covariation w i l l  increase. The implications of mood for job performance are based on the presumption that expectancy i s prospective.  That i s , i t i s forward looking.  Individuals  are hypothesized to make choices about future levels of e f f o r t based on t h e i r evaluations of expectancy, instrumentality, and valence.  When i n a good mood,  individuals are more l i k e l y to remember past successes and increase t h e i r estimates of the covariation between e f f o r t and performance.  Given that  valued rewards and performance also covary (instrumentality and valence are high), higher expectancy w i l l r e s u l t i n i n d i v i d u a l s choosing a higher l e v e l of effort. The influence of mood may extend to other task cognitions, such as task i n t e r e s t , d i f f i c u l t y , s a t i s f a c t i o n , and perceived  task e f f o r t .  These  influences w i l l also be tested by including measures of these cognitions as dependent variables i n the studies undertaken below.  Other measures, used i n  the past to measure "expectancy" but which do not measure effort—performance covariation, may also be influenced by mood.  Therefore,  measures such as the  subjective p r o b a b i l i t y of success and the perceived c o r r e l a t i o n between e f f o r t  63  and performance w i l l also be used as dependent variables.  Hypothesis Two: Mood and Casual Attributions Individuals i n a p o s i t i v e mood state w i l l a t t r i b u t e successful and unsuccessful performance more to i n t e r n a l causes than w i l l individuals i n a negative mood state.  Weiner, Russell and Lerman (1978) have proposed that a f f e c t i v e reactions to success or f a i l u r e depend on how cause i s attributed.  Some emotional  reactions to success and f a i l u r e are posited to be attribution-dependent, i . e . , they depend on how cause i s attributed, and other reactions are outcomedependent but a t t r i b u t i o n independent.  For example, Weiner, Russell, and  Lerman (1979) report that happiness i s outcome dependent but a t t r i b u t i o n independent: happiness i s a r e l i a b l e r e s u l t of success no matter to what success i s attributed.  S i m i l a r l y , sadness i s a r e l i a b l e r e s u l t of f a i l u r e .  In contrast, pride requires that success be attributed to i n t e r n a l causes. Gratitude r e s u l t s when success i s attributed to external causes.  Porac,  Nottenburg, and Eggert (1981) have extended t h i s approach to an organizational context. In t h i s study, we are not interested i n effects of a t t r i b u t i o n s on mood but rather i n the effects of mood on other variables.  The present research  provides an opportunity to examine how mood influences causal S p e c i f i c a l l y , i t i s hypothesized that p o s i t i v e mood increases  ascriptions. internal  a t t r i b u t i o n s for both success and f a i l u r e r e l a t i v e to negative mood. The r a t i o n a l e for t h i s hypothesis i s related to the explanation for s e l f serving biases i n a t t r i b u t i o n theory.  This body of research has shown that  individuals tend to make i n t e r n a l , self-enhancing a t t r i b u t i o n s for p o s i t i v e outcomes and external s e l f - p r o t e c t i v e attributions for negative outcomes, (see  64  Bradley, 1978, for a review). blame for f a i l u r e .  That i s , they take c r e d i t for success and avoid  Positive mood may have two influences on t h i s process: i t  may increase the perception of success and thus increase the tendency to take credit f o r i t .  Positive mood might also increase i n t e r n a l i t y of a t t r i b u t i o n s  by biasing expectancy or effort-performance covariation.  If i n d i v i d u a l s  perceive a stronger relationship between e f f o r t and performance when i n a good mood then they are more l i k e l y to a t t r i b u t e success or f a i l u r e to e f f o r t i n t e r n a l cause.  an  Negative mood, on the other hand, may increase the perceived  magnitude of f a i l u r e and thus the avoidance of blame.  Negative mood may also  bias effort-performance expectancy downward so that success or f a i l u r e i s attributed to factors other than e f f o r t .  In fact, biased expectancies may  represent a component of the mechanism by which cause i s a t t r i b u t e d .  So, i t  i s hypothesized that p o s i t i v e mood w i l l be associated with i n t e r n a l a t t r i b u t i o n s for performance.  Hypothesis Three: Self-esteem and Mood D i s p o s i t i o n a l self-esteem w i l l moderate the effect of mood on expectancies. S p e c i f i c a l l y , the e f f e c t of mood on expectancies w i l l be stronger for individuals lower i n self-esteem. There i s reason to believe that the effect of mood on expectancy may depend i n part on an i n d i v i d u a l s ' self-esteem.  The f i r s t suggestions f o r such  an effect stems from the finding that global self-esteem i s p o s i t i v e l y related to p r i o r expectations of success (Campbell Shrauger, 1972).  That i s , people who are more self-accepting believe that  they w i l l perform better. expectancy.  & Fairey, 1985; Coopersmith, 1967;  This, i n turn, i s l i k e l y to result i n higher  People with higher self-esteem may report higher e f f o r t -  performance covariation.  This may be the result of some ideational process,  i n which high self-esteem i s associated with more p o s i t i v e s e l f - b e l i e f , or of  65  a predisposition among individuals with high self-esteem toward more favorable temporary mood states.  Either way, self-esteem i s p o s i t i v e l y associated with  absolute levels of expectancy.  This may have the r e s u l t of imposing a c e i l i n g  on the relationship between mood and expectancy among high self-esteem individuals, reducing the strength of the r e l a t i o n s h i p . The second effect i s also possible.  It i s based on the finding that  individuals with high self-esteem may be resistant to mood manipulations. High self-esteem individuals have been shown to be less sensitive to a negative mood manipulation  (Brockner, H j e l l e & Plant, 1985).  Whether low  self-esteem individuals would be less sensitive or more sensitive to p o s i t i v e mood inductions i s not evident.  That i s , they may be more or less e a s i l y put  into a good mood. Nevertheless, high self-esteem individuals may be c h r o n i c a l l y i n a better mood, and may also be resistant to temporary mood inductions and their effects.  Thus the effect of mood on expectancies i s l i k e l y to be weaker f o r  individuals with high self-esteem. Self-esteem i s of p a r t i c u l a r relevance to the study of mood and expectancy because i t has been demonstrated to be related to both mood and expectancy. theory.  Other variables also of interest are those related to expectancy  These include locus of control, or a generalized b e l i e f about the  source of rewards or reinforcement, and the concept of s e l f - e f f i c a c y . In a recent examination  of conceptual and methodological  issues  surrounding self-esteem. Demo (1985) i d e n t i f i e d two separate dimensions of self-esteem: experienced  self-regard measured by self-report and presented  self-regard measured by s p e c i f i c others. behaviors consistent  The l a t t e r , presented s e l f , involves  with r o l e requirements  necessarily consistent  and s i t u a t i o n a l demands but not  with the actual or the desired  self.  I t focuses on the  66  l e v e l of self-regard communicated to others.  Experienced  self-esteem, i n  contrast, i s one's a t t i t u d e toward oneself, degree of self-respect and perception of self as a person of worth.  Experienced  self-esteem i s most often viewed as a global p o s i t i v e or  negative self-assessment.  In t h i s view, self-esteem i s a personality t r a i t  that i s r e l a t i v e l y stable over time and situations. Although there are apparently multiple measures of self-esteem and approaches to i t s measurement, Demo (1985) showed that these cluster around the concepts of experienced as opposed to presented self-esteem. The methodological question of how best to measure self-esteem i s thus p r i m a r i l y dependent on which conceptual dimension i s of i n t e r e s t . the Rosenberg Self-Esteem Scale (Rosenberg, 1979) Esteem Inventory experienced  (Coopersmith,  self-esteem.  1967)  Demo concluded that  and the Coopersmith Self-  were shown to be v a l i d measures of  The former appeared, of the two, to have the better  reliability. It i s experienced proposed here.  self-esteem that i s of interest i n the investigation  Our p r i n c i p a l variables are perceptions held by individuals,  such as effort-performance covariation or expectancy, and experienced mood state.  We are interested i n the influence of i n d i v i d u a l differences i n global  experienced self-esteem on these perceptions and the r e l a t i o n s between them. Experienced  self-esteem has also been linked t h e o r e t i c a l l y and empirically to  expectancy percepts.  In h i s expectancy theory model of motivation, Lawler  (1971) says that individual b e l i e f s about the p r o b a b i l i t y that e f f o r t w i l l lead to performance are influenced by "the subject's self esteem .  .  ., that  i s , h i s general (SR) b e l i e f s about his a b i l i t y to cope with and control his environment (p.  107)."  Studies by Brockner (1979), Campbell and Fairey  (1985), Coopersmith (1967), Lied and Pritchard (1976) and Schrauger (1972)  67  have found that high self-esteem individuals generally have higher performance expectancies than do low self-esteem i n d i v i d u a l s .  Individuals with more  favorable self-regard are more l i k e l y to expect that their e f f o r t w i l l be successful. S e l f - e f f i c a c y i s a construct that i s related to the components of expectancy theory, although i t i s not, s t r i c t l y speaking, a stable i n d i v i d u a l difference l i k e self-esteem.  According to s e l f - e f f i c a c y theory, the behavior  of individuals, including e f f o r t expended on and persistance at tasks, i s determined by t h e i r sense of personal mastery or e f f i c a c y (Bandura, 1982;  1986).  1977;  "An e f f i c a c y expectation i s the conviction that one can  successfully execute the behavior required to produce (certain) outcomes" (Bandura, 1977, p.193).  E f f i c a c y expectancy i s contrasted with outcome  expectancy, which i s "defined as a person's estimate that a given behavior w i l l lead to (the) outcomes" (Bandura, 1977, p.193). The correspondence  between these concepts and the components of  expectancy theory i s s t r i k i n g .  (Especially so i n l i g h t of seeming claims to  innovation on the part of e f f i c a c y theorists (e.g., Goldfried & Robins, 1982).) Outcome expectancies are obviously very similar to instrumentality beliefs.  E f f i c a c y expectancies are s i m i l a r l y very close to effort-performance  expectancy as conceptualized by Vroom (1964).  The difference, perhaps,  lies  i n Vroom's emphasis on the determinants of e f f o r t , whereas Bandura allows that successful performance i s a function of a b i l i t y ( s k i l l ) as well as e f f o r t . Theories of work motivation assume that performance of the behavior i s possible and are concerned with motivating e f f o r t .  S e l f - e f f i c a c y theory grew  out of therapeutic e f f o r t s to reduce i n d i v i d u a l fears about an i n a b i l i t y to perform certain behaviors.  Nonetheless, both approaches come together i n  positing that behavior i s a function of b e l i e f that the behavior i s possible.  68  Bandura d i f f e r e n t i a t e s between s e l f - e f f i c a c y and self-esteem i n at least two ways. mastery.  F i r s t , he argues for the independence of self-worth from a sense of Second, he says that "judgements of personal e f f i c a c y do not operate  as d i s p o s i t i o n a l determinants 411).  independently of conceptual factors" (1978, p.  Let us consider each of these arguments i n turn. Bandura says that self-esteem and s e l f - e f f i c a c y represent two d i f f e r e n t  aspects of self-referent thought.  "Self-esteem pertains to the evaluation of  self-worth, which depends on how the culture values the a t t r i b u t e s one possesses and how well one's behavior matches personal standards of worthiness.  Perceived s e l f - e f f i c a c y i s concerned with the judgement of  personal c a p a b i l i t i e s . uniform r e l a t i o n .  Judgments of self-worth and s e l f - c a p a b i l i t y have no  Individuals may regard themselves as highly efficacious i n  an a c t i v i t y from which they derive no pride ( s k i l l e d combat soldier) or judge themselves i n e f f i c a c i o u s at an a c t i v i t y without (e.g., inept skater).  suffering a loss of self-worth  However, i n many of the a c t i v i t i e s people pursue, they  c u l t i v a t e s e l f - e f f i c a c i e s i n what gives them a sense of self-worth.  Thus,  both self-esteem and s e l f - e f f i c a c y contribute i n their own way to the q u a l i t y of human l i f e (Bandura, 1986, p.410).  D i s p o s i t i o n a l self-esteem i s not  necessarily associated with judgements of s e l f - e f f i c a c y , and s e l f - e f f i c a c y does not necessarily influence self-esteem.  However, a sense of mastery and  competence i n a valued a c t i v i t y does contribute to self-regard and individuals are generally able to s e l f - s e l e c t a c t i v i t i e s to pursue. Unlike global self-esteem, perceived s e l f - e f f i c a c y i s not a d i s p o s i t i o n to view the self i n a p a r t i c u l a r way.  For Bandura, e f f i c a c y must be assessed  at a microanalytic l e v e l , with a degree of s p e c i f i c i t y i n the assessment of s e l f - b e l i e f s that matches the s p e c i f i c i t y of the behavior. understand  tendencies to perform  If we wish to  s p e c i f i c levels of a p a r t i c u l a r behavior we  69  must measure b e l i e f s about the conviction that those s p e c i f i c levels of that p a r t i c u l a r behavior can be successfully  executed.  Although s e l f - e f f i c a c y  b e l i e f s can generalize i n a limited way, Bandura decries the notion of generalized s e l f - e f f i c a c y .  " I t i s no more informative to speak of s e l f -  e f f i c a c y i n global terms than to speak of non-specific s o c i a l behavior" (1986, p.411). Nevertheless, the idea that individuals  possess stable differences i n  their degree of s e l f - e f f i c a c y across a v a r i e t y of situations  i s appealing.  Individuals who have a history of numerous and varied mastery experiences are l i k e l y to have p o s i t i v e  s e l f - e f f i c a c y expectations i n general.  Scherer and  colleagues (Scherer, Maddux, Mercandante, Prentice-Dunn, Jacobs & Rogers, 1982) and  predicted "that i n d i v i d u a l differences i n general s e l f - e f f i c a c y exist that these differences have behavioral correlates.  An individual's  experiences with success and f a i l u r e i n a v a r i e t y of situations  past  should result  i n a general set of expectations that the individual carries into new situations.  These generalized expectancies should influence the individual's  expectations of mastery i n the new situations" prediction,  (1982, p.664).  To test  their  Scherer and colleagues constructed a self-report measure of  general s e l f - e f f i c a c y and correlated of other personality c h a r a c t e r i s t i c s .  scores on this instrument with measures Among other r e s u l t s , they found that  general s e l f - e f f i c a c y was strongly associated with self-esteem (r = .51). While they took this as evidence of convergent v a l i d i t y , the magnitude of t h i s correlation  raises concerns about divergent v a l i d i t y , especially i n view of  Bandura's d i s t i n c t i o n between s e l f - e f f i c a c y and self-esteem.  Scherer and  colleagues also found a s i g n i f i c a n t association between s o c i a l l y desirable responding (r = .43) and general s e l f - e f f i c a c y , r a i s i n g the spectre of response bias, although a b e l i e f i n one's efficacy i s l i k e l y to be seen as a  70  s o c i a l l y desirable c h a r a c t e r i s t i c .  These findings, and the novelty of this  measure, caution against i t s unbridaled use. what i t does or does not measure. not part of Bandura's theory.  As yet, evidence i s sparse on  The concept of generalized s e l f - e f f i c a c y i s  Were Bandura to posit a stable i n d i v i d u a l  difference related to s e l f - e f f i c a c y expectations, i t i s more l i k e l y that he would name i t self-esteem. A construct which was conceptualized as a stable i n d i v i d u a l difference r e f l e c t i n g a generalized expectation i s locus of control.  Rotter (1966)  conceived of locus of control as a person's generalized expectancy to perceive reinforcement along a continuum from i n t e r n a l to external.  Internal locus of  control i s the perception that reinforcement i s contingent on one's own behaviors as opposed to a perception of an external locus of control, that reinforcement i s the r e s u l t of forces beyond one's control and due to chance, fate, or powerful others.  The conception of perceived control was part of  Rotter's s o c i a l learning theory.  According to this theory, a person's actions  are predicted by expectations and values.  S p e c i f i c a l l y , the p o t e n t i a l of a  person to perform a certain behavior i n a situation i s a function of the expectancy that the behavior w i l l be reinforced i n that situation and the value of reinforcement i n that s i t u a t i o n .  Locus of control as a d i s p o s i t i o n a l  factor or stable i n d i v i d u a l difference i s the b e l i e f that reinforcement i s as a r e s u l t of "behavior, s k i l l s , or i n t e r n a l d i s p o s i t i o n s " (Rotter, 1966, Rotter's theory i s an expectancy theory.  p.4).  In contrast to Vroom, however,  Rotter d i d not d i s t i n g u i s h between performance expectancy and outcome expectancy.  Rotter's focus was c l e a r l y on the concept of reinforcement.  to the extent that locus of control represents d i s p o s i t i o n a l determinants  Yet of  perceived expectancy, expectancy for behavior as well as reinforcement, i t i s also involved with generalized expectations for the performance of behavior.  71  This i s evident from the content of some of the items of Rotter's measure, which refer to e f f o r t , hard work, and the a b i l i t y to do things.  As well,  elaboration of the components of locus of control has lead to development of subscales that make e x p l i c i t effort-performance expectancy.  Lefcourt's  (1981)  Multidimensional- M u l t i a t t r i b u t i o n a l Causality Scale includes assessment of b e l i e f s regarding s p e c i f i c a t t r i b u t i o n s of causality to a b i l i t y and e f f o r t . Paulhus and C h r i s t i e s ' (1981) Spheres of Control measure separates the three behavioral  domains of personal e f f i c a c y , interpersonal  s o c i o p o l i t i c a l control.  control, and  Personal e f f i c a c y , or control i n the sphere of  personal achievement i s related to b e l i e f s about a b i l i t y and e f f o r t . The locus of control construct  thus includes both expectancy for control over  reinforcement and expectancy for control over behavior.  In summary, while the concept of s e l f - e f f i c a c y i s related to e f f o r t performance expectancy, i t i s neither conceptualized as a stable i n d i v i d u a l difference, nor has substantial empirical evidence been developed to suggest that i t i s a stable i n d i v i d u a l difference.  In contrast,  self-esteem i s  related to both mood and expectancy and might be expected to influence the r e l a t i o n s h i p between them.  S i m i l a r l y , locus of control i s related to the  expectancy and instrumentality  components of expectancy theory.  Therefore, i t  i s proposed that locus of control also be measured and i t s possible moderating effects investigated,  i n addition to testing the hypothesized effects of s e l f -  esteem.  Proposed Research The previous sections explicate three novel hypotheses r e l a t i n g mood and expectancy.  To test these and related hypotheses a laboratory  undertaken.  This i s described  below as Study Four.  experiment was  To provide a foundation  72  for Study Four, three other studies were completed.  In Study One, measures of  constructs i d e n t i f i e d i n the previous chapter were chosen and administered. This allowed evaluation of their psychometric properties and examination of their i n t e r - r e l a t i o n s h i p .  In Study Two mood induction procedures that have  been used i n experimental settings are reviewed and t h e i r s u i t a b i l i t y f o r testing the hypotheses posed here i s evaluated. A s p e c i f i c induction procedure was thereby chosen and i t s v a l i d i t y was then tested.  In Study  Three, the issue of v a l i d measurement of expectancy or the covariation between e f f o r t and performance i s discussed. To effect such v a l i d a t i o n , perceptions of perceived covariation were c o l l e c t e d from participants i n two tasks which d i f f e r e d i n objective expectancy.  F i n a l l y , Study Four u t i l i z e d the measures  developed and validated i n Studies One and Three, and the manipulation chosen in Study Two, to test the p r i n c i p a l hypotheses of t h i s d i s s e r t a t i o n .  These  four studies can be summarized as c o l l e c t i v e l y addressing the following questions: What i s mood?, What i s expectancy?, and How are they related?  73  V.  STUDY ONE: PSYCHOMETRIC PROPERTIES OF MEASURES  Overview Study One was undertaken to assess the i n t e r r e l a t i o n s h i p s between and dimensionality  and r e l i a b l i t y of measures of task perceptions,  and i n d i v i d u a l differences.  mood states,  Participants completed measures of i n d i v i d u a l  differences, undertook a target task, and completed measures of perceptions of that task and of their mood state. In the following sections, the nature of the participants i n Study One and i t s procedure are described more f u l l y . i s then described  Each measure or set of measures  separately with the r e s u l t s of the study for that measure.  Subjects Participants were second year business students, recruited to p a r t i c i p a t e i n a study of goal-setting and productivity.  Participants met i n group  settings where they were asked f i r s t to complete a "questionnaire  about  themselves" which contained the i n d i v i d u a l difference measures, then "to p a r t i c i p a t e i n a simulated q u a l i t y control task", and f i n a l l y to complete a questionnaire  to assess t h e i r "reactions to and perceptions of the task".  This f i n a l questionnaire  contained the remaining measures.  Participants were  t o l d that they were part of a baseline, no goal-setting treatment. Participants were 99 males and 89 females, ranging i n age from 19 to 35 (median age = 20).  Each received course c r e d i t for p a r t i c i p a t i n g .  C o n f i d e n t i a l i t y of responses was promised and demonstrated by having participants place their completed questionnaires  i n unmarked envelopes.  Informed consent forms were completed and c o l l e c t e d separately. procedure was made clear at the outset of the study.  This  The informed consent  74  form i s contained i n Appendix A. debriefed and dehoaxed. was  Following the session, p a r t i c i p a n t s were  That i s , the purpose of the study and i t s procedures  elaborated and the absence of any deception was emphasized.  procedures  The  used i n this study were reviewed and approved by an appropriate  i n s t i t u t i o n a l e t h i c a l review committee.  Procedure Participants f i r s t completed the questionnaire containing the measures of self-esteem and impression management.  When a l l questionnaires had been  completed, they were placed by the respondent i n an unmarked envelope.  This  questionnaire i s shown i n Appendix A. Participants were then asked to undertake a task described as a simulated q u a l i t y control task.  They were t o l d that i n q u a l i t y control tasks "a product  must be matched against a standard".  Participants were provided with sheets  of numbers, asked to check the number at the l e f t of each row on each sheet, and then to c i r c l e each number i n the row that matched the number at the l e f t . Participants worked at the task for a series of 10 one-minute periods.  The  purpose of t h i s task was to provide a basis for the measures of task perceptions that followed.  The instructions for the task are shown i n  Appendix A. Following the task, p a r t i c i p a n t s completed a questionnaire containing measures of task perceptions, namely s a t i s f a c t i o n , d i f f i c u l t y , interest, challenge, performance, e f f o r t , and i n t e r n a l work motivation.  Also contained  i n t h i s questionnaire were two measures of mood states, an adjective checklist and a semantic d i f f e r e n t i a l measure.  The complete questionnaire i s contained  i n Appendix A. The entire procedure,  from introduction to debriefing, took  approximately  75  90 minutes.  Measures and Results In the following sections, each measure i s described and i t s psychometric properties are assessed. reliability.  Two properties are of i n t e r e s t : dimensionality and  Because the measures used were adopted i n whole or i n part from  e x i s t i n g scales, the concern with respect to dimensionality i s one of confirmation, as opposed to exploration. For example, rather than determining how many factors appear to underlie the 48 items i n the Multiple Affect Adjective Checklist, we are interested i n confirming that the three factors hypothesized by Zuckerman and Lubin (1965) are present. The r e l i a b i l i t y of a measure i s i t s freedom from measurement error, s p e c i f i c a l l y from v a r i a t i o n over t r i a l s  (Guttman, 1945).  Measurement error  attenuates correlations between variables; unreliable measures degrade an analysis while r e l i a b l e ones enhance i t . acceptable r e l i a b i l i t y i s 0.70, scales used i n research.  In t h i s study, the c r i t e r i o n for  following Nunnally's (1978) r u l e of thumb for  Two estimates of r e l i a b i l i t y are reported: Internal  consistency or c o e f f i c i e n t alpha ( a ) i s reported because i t i s most commonly used.  Also reported i s the highest of Guttman's (1945) six estimates of  reliability. reliability,  Guttman showed that each of these estimates i s a lower bound on therefore true test-retest r e l i a b i l i t y i s at least equal to the  highest estimate.  This highest estimate w i l l be referred to as the "Guttman  lower bound" or "GLB". Individual differences: Self-esteem and Impression Management.  The  measure of self-esteem chosen for use was the Rosenberg Self-Esteem Scale (RSE) (Rosenberg,  1965).  The measure of impression management chosen was  the  Impression Management scale of the Balanced Inventory of Desirable Responding  76  (BIDR-IM) developed by Paulhus (1984). for the BIDR-IM and the RSE.  A five-point response format was used  Individuals were asked to indicate for each  statement whether their usual attitude, f e e l i n g or behavior was best r e f l e c t e d by the word rarely, occasionally, sometimes, frequently, or usually. As i n , for  example, the item "I  f e e l that I have a number of good q u a l i t i e s . "  The RSE consists of ten items which measure self-acceptance, or l i k i n g and approving of s e l f . v a l i d i t y for the RSE, 0.85  Robinson and Shaver (1973) c i t e high r e l i a b i l i t y  and  including a test-retest c o r r e l a t i o n over two-weeks of  (Silber & Tippet, 1965).  Most recently, Demo (1985) substantiated the  r e l i a b i l i t y and v a l i d i t y of the RSE i n a multimethod, m u l t i t r a i t study. this study, the internal consistency of the RSE was 0.90. bound on true r e l i a b i l i t y was  In  The Guttman lower  0.91.  The BIDR-IM was developed to measure impression management.  Impression  management or "other deception" i s self-presentation that the presenter knows to be f a l s e (Paulhus, 1984;  Zerbe & Paulhus, 1987).  items about s o c i a l l y desirable but s t a t i s t i c a l l y  The BIDR-IM contains 20  infrequent behaviors.  A high  score indicates a tendency to act i n a s o c i a l l y acceptable manner, to present a favorable, though f a l s e , impression. In this study, one item was deleted from the BIDR-IM because i t created undesirable reactions i n a p i l o t sample of students (the item was  "I sometimes  pick my nose").  In this study the i n t e r n a l consistency of the remaining  items was 0.68.  The Guttman lower bound was  0.75.  19  Paulhus (1987) reports  that measures of impression management are u n l i k e l y to have very high i n t e r n a l consistency when respondents  are unsure as to exactly what impression they  should present. Both the RSE and BIDR-IM exceeded Nunnally's c r i t e r i o n for r e l i a b i l i t y , supporting their use i n the research that follows.  77  Task perceptions.  Sixteen items were chosen to measure task  s a t i s f a c t i o n , task performance, task d i f f i c u l t y , task challenge, task i n t e r e s t , task e f f o r t , and i n t e r n a l work motivation.  Some of these items were  adapted from the Job Diagnostic Survey (Hackman and Oldham, 1980), others were written for t h i s study.  Participants were asked to indicate their agreement  with each item using a five-point scale from "Strongly Agree" to "Strongly Disagree".  The sixteen items used are shown i n Table 3.  To confirm that the seven concepts l i s t e d above were contained i n the sixteen items, their underlying factor structure was investigated v i a factor a n a l y t i c procedures using the P4M program i n the BMD series (Dixon, 1981).  statistical  software  Before reporting the results of the factor analyses,  however, the p r a c t i c a l issues underlying the procedure are considered. S p e c i f i c a l l y , factor analysis i s sensitive to problems created by outlying cases and variables with skewed d i s t r i b u t i o n s .  (Tabachnik & F i d e l l , 1983).  Univariate o u t l i e r s were defined as cases with standard scores i n excess of 2.58.  Of the 16 items across 188 p a r t i c i p a n t s , 24 scores exceeded this  criterion.  This represents 0.8 %, under the 1 % of scores we would expect  to chance.  Therefore, no cases were excluded as univariate o u t l i e r s .  due  Multivariate o u t l i e r s were i d e n t i f i e d by examining the Mahlanobis distance (D ) of each case to the centroid of the sample, computed by BMDPAM. 2  The  Mahlanobis distance i s d i s t r i b u t e d as a chi-square variable, and so a c r i t i c a l value for extreme cases can be computed. value for D  2  Eight cases exceeded the c r i t i c a l  at p 5 .01 and were excluded from subsequent analyses.  f i n a l sample comprised  Thus the  180 cases.  Inspection of the c o r r e l a t i o n matrix revealed several sizable correlations, making factor analysis appropriate. multiple correlations (SMC)  Examination of the squared  of each v a r i a b l e with a l l other variables revealed  S a t i s f a c t i o n with Task Performance: 1. Generally speaking, I am unsatisfied with my performance on the proofreading task. 2. A l l i n a l l , I am very s a t i s f i e d with my performance on the proofreading task. 3. Compared to other people, I don't think I d i d very well on the proofreading task. 4. My performance on the proofreading task was high. Task D i f f i c u l t y and Challenge: 5. 6. 7. 8.  I found the proofreading task d i f f i c u l t . I found the proofreading task to be easy. The proofreading task was challenging. I didn't f i n d the proofreading task very challenging.  Task Interest: 9. 10.  The proofreading task was not very i n t e r e s t i n g . I enjoyed working on the proofreading task.  Task E f f o r t : 11. I expended a high l e v e l of e f f o r t on the proofreading task. 12. Overall, I didn't t r y very hard on the proofreading task. Internal Work Motivation: 13. 14. 15. 16. Table 3.  I f e e l a great sense of personal s a t i s f a c t i o n when I do well. My opinion of myself goes up when I do well. My own feelings generally are not affected much one way or another by how well I d i d on the proofreading task. I f e e l bad and unhappy when I've done poorly. Items assessing task perceptions, grouped by construct.  79  that 12 of the 16 items shared variance i n excess of 40% with the other variables.  The remaining four items, a l l assessing  i n t e r n a l work motivation,  had SMC's of less than .25 i n d i c a t i n g that they contributed analysis.  In i t s e l f  less to the  t h i s does not warrant elimination of these v a r i a b l e s .  It  does, however, suggest that care be taken i n subsequent analysis. P r i n c i p a l components extraction was factors.  used to estimate the number of  Five factors had eigenvalues greater than 1.  ( C a t t e l l , 1966)  suggested four or f i v e f a c t o r s .  The  scree test  Inspection of the varimax  rotated factor pattern for the f i v e factor p r i n c i p a l components solution revealed  simple structure with items assessing  together for a l l constructs  a similar construct  except i n t e r n a l work motivation.  factors extraction of f i v e factors revealed defined nor i n t e r n a l l y consistent.  The  loading  Principal  that the f i f t h factor was  not  squared multiple c o r r e l a t i o n of  factors predicted from scores on the observed variables did not reach  well  the .70,  Tabachnik and F i d e l l ' s (1983) suggested c r i t e r i o n for i n t e r n a l consistency. The  i n s t a b i l i t y of t h i s factor was demonstrated further by the r e s u l t s of  maximum l i k e l i h o o d factor extraction, i n which the f i f t h factor was poorly  very  structured.  On the basis of these r e s u l t s , the four items assessing motivation were excluded from subsequent analyses.  i n t e r n a l work  P r i n c i p a l components  extraction of the remaining 12 items showed four factors with eigenvalues greater than one.  The  scree test supported four factors as well.  Inspection  of the varimax rotated factor structure from p r i n c i p a l components analysis revealed  simple structure for a l l f a c t o r s .  Comparison of the r e s u l t s of  p r i n c i p a l components, p r i n c i p a l factors, and maximum l i k e l i h o o d factor analysis revealed  l i t t l e v a r i a t i o n i n rotated structure, strong evidence for  the s t a b i l i t y of the solution.  As indicated by SMC's, a l l factors were  80  i n t e r n a l l y consistent and well-defined by the variables; the lowest SMC for factors from variables was .785. Further, variables were well defined by the factor solution, communality values, shown i n Table 4, tended to be high, and a l l variables p a r t i c i p a t e d i n the solution.  Inspection of the matrix of  residual correlations revealed a l l to be near-zero.  Maximum l i k e l i h o o d  extraction was chosen for the f i n a l solution because of i t s appropriateness to cases where the common factor model i s held to be true, as i s the case here where the objective i s the discovery of few common f a c t o r s . Also, maximum l i k e l i h o o d factor analysis i s generally considered to be superior to p r i n c i p a l factors extraction. normality. 1983).  Maximum l i k e l i h o o d factor analysis assumes multivariate  Multivariate normality i s d i f f i c u l t to test (Tabachnik & F i d e l l ,  In the present circumstance  the l i k e l i h o o d that the data are  multivariate normal i s increased by the finding that the data are more or less univariate normal. Inspection of Table 4 reveals that the two items assessing s a t i s f a c t i o n and the two items assessing performance define the f i r s t factor, suggesting that these constitute a single construct.  Similarly, the items assessing task  d i f f i c u l t y and those assessing challenge load together on the second f a c t o r . The t h i r d factor i s defined by two items assessing task interest and the f i n a l factor by two items assessing task e f f o r t .  No items are complex, that i s ,  none show a factor loading of more than .45 on more than one f a c t o r .  This  represents a less than 20% variance overlap, the cutoff f o r i n c l u s i o n of an item i n the d e f i n i t i o n of a factor proposed by Comrey (1973). The conclusion drawn from t h i s analysis i s that the set of items assessing task perceptions taps four dimensions, corresponding  to (1)  s a t i s f a c t i o n with task performance, (2) task d i f f i c u l t y and challenge, (3) task interest, and (4) task e f f o r t .  Four scales were constructed accordingly.  FACTOR  I  II  III  IV  h  2  Item 1. 2. 3. 4.  Satl Sat2 Perfl Perf2  .64 .85 .61 .71  -.09 -.03 -.07 -.11  -.02 .07 -.02 -.07  .02 .08 .23 .32  .42 .74 .42 .62  5. 6. 7. 8.  Dif'fl Diff2 Chall Chal2  -.18 -.09 .06 -.02  .58 .64 .75 .77  -.09 .09 .38 .24  .01 .02 .17 .17  .37 .43 .73 .67  9. I n t r l 10. Intr2  -.09 .06  .30 .05  .68 .99  .07 .07  .57 .99  11. E f f l 12. E££2  .20 .21  .13 .12  .10 .06  .79 .78  .69 .67  Percent of Variance  18.0  17.0  14.6  12.0  Table 4. Factor loadings, communalities ( h ) , percent of variance for fourfactor maximum l i k e l i h o o d factor extraction on task perception items. Factor loadings subject to Varimax r o t a t i o n . 2  82  A f i f t h scale, comprised of the i n t e r n a l work motivation items, was also constructed, although these items do not appear to form a coherent dimension as indicated by the factor analyses above and the r e l i a b i l i t y estimation below.  Table 5 shows the means, standard deviations, internal consistency  c o e f f i c i e n t s , and Guttman lower bounds for these scales.  With the exception  of the i n t e r n a l work motivation scale, these measures have good r e l i a b i l i t y . The lack of coherency i n the i n t e r n a l work motivation scale i s again shown by i t s low r e l i a b i l i t y .  On the basis of these r e s u l t s , i t was decided that the  four r e l i a b l e scales would be retained for subsequent use and the f i f t h scale discarded. Causal a t t r i b u t i o n s .  The Causal Dimension Scale (CDS) (Russell,  1982)  was used to assess how individuals perceived the cause of their performance on the  proofreading task.  The CDS contains 9 items that are presented i n a  semantic d i f f e r e n t i a l format as shown i n Figure 4. of i n t e r n a l i t y , s t a b i l i t y , and c o n t r o l l a b i l i t y .  Three items measure each  These three subscales  correspond to the causal dimensions described by Weiner (1979).  Internal  locus of c a u s a l i t y refers to whether the cause i s attributed i n t e r n a l l y , to something about the attributor or to some external cause, outside the attributor.  S t a b i l i t y i s defined as whether the cause i s constant over time  or variable over time.  C o n t r o l l a b i l i t y refers to whether the cause could be  changed by the a t t r i b u t o r or by someone else.  Russell (1982) recommended that  individuals be provided with a reason or cause for an outcome and then be asked to use the CDS to rate the cause.  He presented evidence for the  v a l i d i t y of the CDS: causes that contained objective combinations of i n t e r n a l i t y , s t a b i l i t y and c o n t r o l l a b i l i t y were accurately r e f l e c t e d i n ratings on the CDS.  Further, the CDS measure d i f f e r e n t i a t e d between these  three dimensions of causal a t t r i b u t i o n s .  Russell (1982) reported r e l i a b i l i t y  Items  Mean  Scale  Stand. Dev.  a  GLB  Self-esteem  10  43.40  5.95  .904  .913  Impression-Management  19  73.60  7.44  .681  .745  4  14.18  2.60  .788  .789  4  9.49  3.29  .781  .800  2 2 4  4.86 8.23 15.48  2.01 1.32 1.86  .820 .831 .240  .820 .831 .256  3 3 3  19.72 14.03 19.59  3.92 5.86 4.00  .686 .806 .474  .707 .807 .484  Task perceptions: S a t i s f a c t i o n with performance D i f f i c u l t y and challenge Task Interest Task E f f o r t Internal Work motivation Causal  attributions:  Internality Stability Controllability T  Table 5. Descriptive s t a t i s t i c s and r e l i a b i l i t y estimates for measures of i n d i v i d u a l differences, task perceptions, and causal a t t r i b u t i o n s .  84  The next set of questions ask you to consider the reasons behind your actual performance on the proofreading task. For each scale, c i r c l e the number that best describes your impression or opinion of the cause of your performance. Is the cause of your performance something that : Reflects an aspect of yourself  9  8  7  6  5  4  3  2  1  Reflects an aspect of the situation  5  4  3  2  1  Uncontrollable by you or other people  Is the cause of your performance: Controllable by you or other people  9  8  7  6  Is the cause of your performance something that i s : Permanent  9  8  7  6  5  4  3  2  1  Temporary  2  1  Unintended by you or other people  Is the cause of your performance something: Intended by you or other people  9  8  7  6  5  4  3  Is the cause of your performance something that i s : Outside of you  9  8  7  6  5  4  3  2  1  Inside of you  Is the cause of your performance something that i s : Variable over time  9  8  7  6  5  4  3  2  1  Stable over time  5  4  3  2  1  Something about others  Is the cause'of your performance: Something about you  9  8  7  6  Is the cause of your performance something that i s : Changeable  9  8  7  6  5  4  3  2  1  Unchanging  Is the cause of your performance something f o r which: No one i s responsible  Figure 4.  9  8  7  6  5  4  3  2  Causal A t t r i b u t i o n Measures, Study One  1  Someone i s responsible  85  c o e f f i c i e n t s of 0.87,  0.84,  and 0.73  for the locus of i n t e r n a l i t y , s t a b i l i t y  and c o n t r o l l a b i l i t y scales, respectively. In this study respondents were not asked to name a cause for t h e i r performance but rather to "consider the reasons behind [their] actual performance on the proofreading their impression  task."  They then used the CDS  of the cause of their performance.  to describe  The i n t e r n a l  consistency  c o e f f i c i e n t s and Guttman bounds for the i n t e r n a l i t y , s t a b i l i t y , and c o n t r o l l a b i l i t y scales are shown i n Table 5.  The s t a b i l i t y scale has  acceptable r e l i a b i l i t y , while that of the i n t e r n a l i t y scale i s marginal and that of the c o n t r o l l a b i l i t y scale i s unacceptable.  It i s possible that  Russell (1982) found higher r e l i a b i l i t y because he provided e x p l i c i t causes which respondents were asked to rate.  Because i t was his aim to show  convergent and divergent v a l i d i t y , i s i s l i k e l y that the causes he  provided  were more e a s i l y r e l i a b l y c l a s s i f i e d than the causes for performance which subjects a t t r i b u t e d i n this study, which were not e x p l i c i t l y provided  nor  identified. On the basis of Study One,  i t was decided to r e t a i n the i n t e r n a l i t y and  s t a b i l i t y scales and discard the c o n t r o l l a b i l i t y scale i n further Studies. Multiple Affect Adjective Checklist. administered.  The f i r s t was  Two measures of mood state were  the brief version of the Multiple Affect  Adjective Checklist (MAACL) (Zuckerman & Lubih, 1965). f u l l version, an 89 item measure comprising depression,  and h o s t i l i t y .  The MAACL i s , i n i t s  three scales : anxiety,  The MAACL measures emotional state by means of  verbal reports, that i s , by asking respondents to report whether adjectives .describe their feelings "now."  Zuckerman and Lubin  (1965) report numerous  demonstrations of the v a l i d i t y of the component scales, such as relationships with other measures, differences between normal and c l i n i c a l populations,  and  86  differences r e s u l t i n g from s i t u a t i o n a l inducements. Zuckerman and Lubin have also constructed a b r i e f version of the MAACL with anxiety, depression and h o s t i l i t y scales of 10, 24 and 14 items, respectively.  The items which best discriminated between the f u l l scales were  chosen for the b r i e f version. for  Zuckerman and Lubin report v a l i d i t y evidence  the b r i e f scales close to that of the f u l l scales.  Correlations between  the b r i e f and f u l l scales are reported by Zuckerman and Lubin as 0.82, and 0.92  for the anxiety, depression and h o s t i l i t y scales respectively.  items comprising  0.92,  and  for f u l l scale measures of state anxiety, depression and h o s t i l i t y i n a  sample of college students. (0.15  The  the brief scales are shown i n Table 6.  Zuckerman and Lubin report s p l i t - h a l f r e l i a b i l i t i e s of 0.79, 0.90  0.93  to 0.21)  Retest r e l i a b i l i t i e s over seven days were low  r e f l e c t i n g the f l u c t u a t i n g nature of mood i n members of the  normal population. While Zuckerman and Lubin (1965) used a yes/no, " c h e c k l i s t " response format, i n t h i s study a four-point response scale from " d e f i n i t e l y do not f e e l " to " d e f i n i t e l y do f e e l " was  used, as suggested by Russell (1979).  This  four-point response format for s e l f - d e s c r i p t i v e adjectives has been used by Meddis (1972), and shown by Russell (1979) to r e s u l t i n a better response d i s t r i b u t i o n than a l t e r n a t i v e formats, and to avoid biases due to to check most or few adjectives.  tendencies"  In this study, participants were asked to  indicate, using the four-point scale, how described t h e i r "feelings r i g h t now."  well each of the 48 adjectives  (One item, "gay", was  changed to  "happy", i n keeping with more contemporary usage.) The means, standard deviations and i n t e r n a l consistencies of the b r i e f scales of anxiety, depression,  and h o s t i l i t y i n this study and their Guttman  lower bounds are shown i n Table 7.  As shown, the r e l i a b i l i t i e s for the  Anx  Plus afraid(1) fearful(1) frightened(2) nervous(1) panicky(2) shaky(1) tense(2) u p s e t ( 1) worryi ng(2)  ety  Depres  Mi nus calm(2)  T a b l e 6. I t e m s u s e d f r o m MAACL o r p a r a l l e l V e r s i o n 2.  Plus alone(1) awful(2) blue(1) discouraged(2) forlorn(2) g l o o m y ( 1) hopeless(2) 1onely(1) lost(2) low(1) mi s e r a b l e ( 1 ) rejected(2) suffering(1) sunk(2) t e r r i b l e ( 1) tormented(2) unhappy(1) wi 1 t e d ( 2 )  s h o r t form.  Numbers  5 i on  Hosti  Mi nus  Plus  act i ve(1) alive(2) fined ) happy(2) heal thy(1) merry(2)  i n parentheses  Mi nus  angry(2) cruel(1) di sagreeable(1) mad(2)  refer  t o items  i ty  used  agreeable(1) am i a b l e ( 2 ) c o o p e r a t i v e ( 1) kindly(2) p o l i te(1 ) sympathet i c ( 1 ) tender(2) understanding(1) devoted(2) warm(2)  in parallel  Version 1  Items  Mean  St.  Dev.  a  GLB  MAACL Anxiety Depression Hostility  10 24 14  16.48 40.63 27.79  4.72 10.97 4.99  .827 .931 .765  .850 .951 .801  6 6  35.54 27.63  7.40 8.33  .826 .840  .828 .845  Semantic Differential Pleasure Arousal  Table 7. Means, Standard Deviations and R e l i a b i l i t y Estimates for measures mood state, Study One.  89  h o s t i l i t y and anxiety scales exceed 0.80, depression scale exceeding Pleasure and Arousal.  with the r e l i a b i l i t y for the  0.90. The second measure of mood state administered was  Russell and Mehrabian's (1977) measures of arousal-sleepiness and pleasuredispleasure.  Each of these i s measured by 6 nine-point semantic d i f f e r e n t i a l  scales on which respondents are asked to indicate the point that best describes t h e i r "feelings right now."  Table 7 shows the means, standard  deviations, and internal consistencies of the pleasure and arousal scales. The r e l i a b i l i t y estimates of both scales exceed  Dimensionality of Mood Measures.  .80.  Both sets of mood measures used i n this  study are extant measures with a history of use i n v a l i d i t y and other studies. Zuckerman and Lubin have, for example, demonstrated convergent v a l i d i t y of the MAACL scales.  the divergent and  It would not be appropriate,  therefore, to rearrange the items comprising the MAACL scales, or those of the Mehrebian and Russell semantic d i f f e r e n t i a l measure, on the basis of this study.  However, the responses of participants do permit us to confirm the  dimensionality of the MAACL and semantic d i f f e r e n t i a l measures.  Factor  a n a l y t i c techniques were used to do t h i s . Again, before factor analysis proceeded the s u i t a b i l i t y of the data set was evaluated.  Inspection of the d i s t r i b u t i o n of scores on the 48 items  comprising the MAACL revealed 89 scores across 188 cases for which the standard score was i n excess of 2.58  standard units above or below the mean.  This represents 0.98%, or about the 1% of scores, as would be expected due to chance.  Therefore, no cases were excluded as univariate o u t l i e r s .  Multivariate o u t l i e r s , those cases with s i g n i f i c a n t (p < .01) distances, were i d e n t i f i e d using BMDPAM.  Mahlanobis  Seventeen cases were so i d e n t i f i e d  90  and excluded from the analyses.  Thus the f i n a l sample comprised 171 cases.  Inspection of the c o r r e l a t i o n matrix revealed many sizeable  correlations.  Each variable shared at least 40% variance with a l l other variables, as indicated by SMC's.  Therefore, the factor analysis proceeded.  P r i n c i p a l components extraction revealed 10 factors with eigenvalues greater than one.  The scree test indicated three or four f a c t o r s .  Because of  the presence of three a p r i o r i scales, a three factor solution was selected. Maximum l i k e l i h o o d factor extraction revealed that a l l factors were distinguishable  and well defined by the items, the lowest of the SMC's f o r  factors from items was .83. The results of p r i n c i p a l components, p r i n c i p a l factors, and maximum l i k e l i h o o d factor analysis were very similar, an indicator of the s t a b i l i t y of the solution.  Maximum l i k e l i h o o d extraction was  chosen as most appropriate. The f i t of the solution to the a p r i o r i three factor structure can be assessed by examining Table 8.  The items have been ordered according to the  scale that Zuckerman and Lubin intended they be part o f . Overall, the f i t i s only f a i r .  As can be seen, while most of the depression items load on the  f i r s t factor, most of the anxiety items load on the second, and most of the h o s t i l i t y items load on the t h i r d factor, some of the loadings are r e l a t i v e l y small and a number of items do not load on any factor.  Using a factor  loading  c r i t e r i o n of .45 for inclusion of an item i n a factor, of the 24 items intended to form the depression scale, 19 were included factor ("hits"), 1 item was included  i n the Depression  i n the t h i r d , H o s t i l i t y factor ( a "mis-  h i t " ) while the remaining 4 items were "misses" rather than "mis-hits": were not included 2 were "mis-hits",  i n any factor.  they  S i m i l a r l y 7 of 10 anxiety items were " h i t s ,  loading on the Depression factor, and 1 was a miss.  Eight  of the 14 h o s t i l i t y items loaded together on the t h i r d factor, 2 were mis-  91 I Depressiorl  II Anxiety  III Hostility  h  2  1 4 5 8 9 15 17 18 20 21 22 23 25 26 27 29 30 34 36 37 41 42 44 47  active alive alone awful blue discouraged fine forlorn gloomy happy healthy hopeless lonely lost low merry miserable rejected suffering sunk terrible tormented unhappy wilted  0.38 0.41 0.52 0.74 0.79 0.64 0.40 0.53 0.75 0.44 0.42 0.64 0.63 0.65 0.79 0.37 0.79 0.69 0.64 0.70 0.76 0.60 0.67 0.60  -0.29 -0.31 0.22 0.06 0.05 0.31 0.14 0.00 0.09 -0.08 -0.10 0.25 0.31 0.32 0.11 0.00 0.21 0.21 0.31 0.21 0.20 0.36 0.23 0.18  0.27 0.35 0.01 0.07 -0.04 0.10 0.32 0.01 0.10 0.44 0.36 0.18 -0.03 0.00 0.05 0.49 0.14 -0.08 0.02 0.11 0.08 0.09 0.13 -0.02  0.32 0.39 0.32 0.56 0.63 0.52 0.28 0.28 0.58 0.40 0.32 0.50 0.50 0.54 0.64 0.37 0.69 0.52 0.51 0.54 0.62 0.49 0.53 0.40  2 10 16 19 31 32 35 40 45 48  afraid calm fearful frightened nervous panicky shaky tense upset worrying  0.28 0.07 0.33 0.31 0.12 0.14 0.22 0.22 0.69 0.46  0.60 0.29 0.57 0.53 0.69 0.61 0.52 0.53 0.34 0.45  -0.15 0.04 -0.12 -0.05 -0.04 0.03 0.04 0.14 0.05 -0.06  0.46 0.09 0.45 0.38 0.50 0.40 0.32 0.34 0.60 0.42  3 6 7 11 12 13 14 24 28 33 38 39 43 46  agreeable amiable angry cooperative cruel devoted disagreeable kindly mad polite sympathetic tender understanding warm  0.13 0.17 0.47 0.09 0.41 -0.08 0.44 0.15 0.57 0.03 -0.18 -0.12 -0.04 0.12  -0.10 -0.05 0.26 0.05 0.14 -0.15 0.07 -0.04 0.27 0.04 0.08 0.09 0.07 0.06  0.53 0.36 0.06 0.50 0.13 0.42 0.19 0.61 0.17 0.48 0.52 0.63 0.63 0.68  0.31 0.16 0.29 0.26 0.20 0.21 0.23 0.40 0.42 0.24 0.31 0.42 0.40 0.47  T  Table 8. Factor loadings, communalities ( h ) for maximum l i k e l i h o o d factor extraction on MAACL items. Factor loadings are subject to Varimax rotation. 2  92  h i t s , loading on the Depression factor, and the remaining four were misses. Thus, o v e r a l l , 34 out of 48 items were included i n the proper factor, 5 were included i n some other factor, and 9 items were not included i n the solutions. As Table 8 shows, communalities for these 9 misses were low, as one would expect. A similar set of analyses was performed for the 12 items that make up the semantic d i f f e r e n t i a l measures of pleasure and arousal. and Mehrebian  According to Russell  (1977) these are orthogonal components of mood state.  A two  factor solution i s thus anticipated. Of the 12 scores for each of 188 cases, 7 scores were i d e n t i f i e d as extreme, well below 1% of cases.  Twelve multivariate outlying cases were  i d e n t i f i e d using BMDPAM and excluded from further analyses.  Examination of  the c o r r e l a t i o n matrix and SMC's among items supported factor analysis, the smallest SMC was .36. Having met the p r a c t i c a l l i m i t a t i o n s of the technique, factor analysis proceeded. P r i n c i p a l components extraction revealed two factors with eigenvalues greater than one.  A scree test also supported at least two factors.  Again,  the r e s u l t s of p r i n c i p a l components, p r i n c i p a l factors and maximum l i k e l i h o o d extractions were very s i m i l a r .  Maximum l i k e l i h o o d extraction was chosen.  Inspection of the SMC's for factors from items revealed that each factor was well defined, the lowest SMC was .87. Factor loadings are shown i n Table 9. Using a loading of .45 as a c r i t e r i o n for inclusion of an item i n a factor, i t i s evident that each item loads on only one factor and loads with other items i n the same scale.  That i s , the pleasure items load together, defining the  f i r s t factor, and the arousal items load together, defining the second factor. Russell and Mehrebian's proposed structure i s thus very well supported.  93  I Pleasure Unhappy—Happy P1ea sed—Annoyed Unsatisfied—Satisfied Contented—Melancholic Despairing—Hopeful Relaxed—Bored Relaxed—Stimulated Excited—Calm Sluggish—Frenzied Jittery—Dull Sleepy—Wide Awake Aroused—Unaroused  II Arousal  h  2  0.71 0.83 0.75 0.72 0.60 0.56  0.14 0.01 0.07 0.12 0.06 0.10  0.52 0.69 0.57 0.53 0.36 0.33  -0.16 -0.06 0.26 0.09 0.42 0.32  0.73 0.70 0.71 0.66 0.67 0.75  0.56 0.49 0.57 0.44 0.62 0.67  Table 9. Factor loadings, communalities ( h ) for maximum l i k e l i h o o d factor extraction on Semantic D i f f e r e n t i a l items. Factor loadings are subject to Varimax rotation. 2  94  P a r a l l e l scale construction The measures used i n t h i s Study were administered  so that their  psychometric properties could be evaluated p r i o r to their use i n l a t e r studies.  Of most concern i s the use of self-report measures of mood state as  a check on the effectiveness of the experimental manipulation of mood. The high r e l i a b i l i t i e s of the adjective c h e c k l i s t and semantic d i f f e r e n t i a l scales reported above c e r t a i n l y support their use. In Study Four below, however, i t w i l l be desirable to measure mood twice during one experimental session: immediately following manipulation of mood and later following the measurement of the p r i n c i p a l dependent v a r i a b l e s . Measurement at the l a t t e r instant as well as at the former checks both the fact and the duration of the manipulation.  However, repeating the 60 mood  items would constitute a very lengthy manipulation check indeed.  Therefore,  construction of shorter, p a r a l l e l measures of each mood state was undertaken. The c r i t e r i a for a c c e p t a b i l i t y of these p a r a l l e l scales i s that they correlate highly with each other, and that the scales of most i n t e r e s t meet Nunnally's (1978) r e l i a b i l i t y c r i t e r i o n .  In Study Four, the mood induction to be used  w i l l manipulate mood along a depression-elation dimension.  Therefore, the  scales of most interest for checking t h i s manipulation are the MAACL depression  scale, and both the arousal and pleasure scales of the semantic  d i f f e r e n t i a l measure. P a r a l l e l anxiety, depression,  and h o s t i l i t y scales were constructed for  the MAACL by including alternate items, subject to the constraint that the versions contain equal proportions of p o s i t i v e l y and negatively scored items. Thus the p a r a l l e l forms contained  5, 12, and 7 items respectively.  i d e n t i f i e s which MAACL items were included i n each version.  Table 6  Parallel  pleasure  and arousal scales were s i m i l a r l y constructed for the semantic d i f f e r e n t i a l  95  scales.  Each scale thus contained 3 items.  The responses of participants i n Study One scales.  were used to evaluate these  Table 10 shows the means, standard deviations, and r e l i a b i l i t i e s for  each version, the Guttman s p l i t - h a l f c o e f f i c i e n t s , and the  correlations  between forms. Because of the decrease i n items per scale the r e l i a b i l i t i e s can expected to drop.  Fortunately,  the r e l i a b i l i t y c o e f f i c i e n t s of the depression  scales remains quite high, exceeding .85. also high.  be  The c o r r e l a t i o n between forms i s  While the r e l i a b i l i t i e s of the anxiety and h o s t i l i t y scales  to reach .70 i n three of four instances,  fail  they are not of central concern.  Therefore, the p a r a l l e l forms of the MAACL scales were adopted. Because the pleasure and arousal  scales contain only 6 items each,  construction of shorter, p a r a l l e l scales was  not completed.  The reduction  in  items per scale to 3 i s l i k e l y to impact s i g n i f i c a n t l y on r e l i a b i l i t y and,  in  any event, use of the f u l l scales w i l l not increase  the questionnaire  length  substantially.  Correlations between measures While not the purpose of Study One, correlations between measures. measures of mood?  How  i t i s of interest to examine the  For example, what are the correlations between  do they compare with the i n t e r c o r r e l a t i o n s between  MAACL scales reported by Zuckerman and Lubin? between the MAACL and  What are the  relationships  semantic-differential measures of mood?  Hypothesis  Two  states that individuals i n a p o s i t i v e mood state w i l l a t t r i b u t e performance to internal rather than external causes.  We can examine the c o r r e l a t i o n between  mood and a t t r i b u t i o n s to shed some l i g h t on t h i s issue.  Hypothesis Three i s  based p a r t l y on the notion that individuals with high self-esteem may  be more  Version  One  Version  Two  i  Guttman (S o r r e l a t i o n S t Dev Items Mean  Alpha  Guttman  Mean  St.Dev  Alpha  Guttman  Spl i t Hal f  Between Forms  2 56 6 09 2 74  . 729 .883 .651  .734 .902 .678  8.77 20. 15 14.27  2.50 5 . 19 2.74  .661 .852 .559  .680 .869 .606  . 848 .934 . 797  . 736 .888 .662  3 77 4 75  .738 .674  .739 .678  17.48 14.21  4 . 36 4 .09  .725 .749  .731 .750  . 784 . .867  MAACL Anx i e t y D e p r e s s i on Host i 1 i t y  5 12 7  7.72 20.48 13.52  S e m a n t i c - D i f f e r e n t is il P1easure Arousal  3 3  18 .06 13.43  .653 .774  T  Table  10.  Means,  standard  d e v i a t i o n s and i n t e r n a l  c o n s i s t e n c i e s of the S p l i t - H a l f  MAACL  s c a l e s . S t u d y One.  97  resistant to mood induction.  The c o r r e l a t i o n between self-esteem and s t a t i c  mood states i s therefore of i n t e r e s t . Also of interest are the relations between task perceptions, and between task perceptions and mood states. f i n a l l y , the c o r r e l a t i o n between impression  And  management and other measures i s  an i n d i c a t i o n of their s u s c e p t i b i l i t y to s o c i a l l y desirable responding. In the following sections the p r o b a b i l i t i e s associated with i n d i v i d u a l c o r r e l a t i o n c o e f f i c i e n t s are not reported.  This i s because the p r o b a b i l i t y  associated with each c o e f f i c i e n t taken singly i s an inappropriate r e f l e c t i o n of the p r o b a b i l i t y associated with the c o e f f i c i e n t given that a large number of tests have been performed.  That i s , i n d i v i d u a l p r o b a b i l i t i e s do not  control experimentwise error rate.  In such a post hoc analysis of very many  c o e f f i c i e n t s , c o n t r o l l i n g f o r experimentwise error would diminish power to such an extent that no c o e f f i c i e n t would be declared  stastical  significant.  The relationships discussed below must be taken as highly speculative and subject to the cautions that accompany such a " f i s h i n g  expedition".  Zuckerman and Lubin (1965) found very large correlations between their f u l l scales of anxiety, depression and h o s t i l i t y , and smaller but s i g n i f i c a n t ones between their b r i e f scales.  S p e c i f i c a l l y , i n a sample of 40 c l i n i c a l and  n o n - c l i n i c a l respondents, they found correlations between the b r i e f scales of anxiety and depression of .82, between anxiety and h o s t i l i t y of .31 and between depression  and h o s t i l i t y of .47.  Recall that Zuckerman and Lubin  u t i l i z e d a c h e c k l i s t rather than the rating scale response format used here. Checklist response formats may be more susceptible to method variance a r t i f a c t s than r a t i n g scale responses and thus i n f l a t e c o r r e l a t i o n s .  In t h i s  sample of 188 students, the correlations between rating scales were more moderate, .54 between anxiety and depression,  .15 between anxiety and  h o s t i l i t y and .41 between depression and h o s t i l i t y .  Note that the r e l a t i v e  98  magnitude of these correlations matches those reported by Zuckerman and Lubin. Examination of the correlations between the semantic-differential adjective c h e c k l i s t scales reveals that anxiety i s negatively  and the  correlated with  pleasure (r = -.34) and p o s i t i v e l y with arousal (r = .28) as one would expect. Depression i s highly negatively  correlated with arousal (r = -.20).  agrees with the location of depression i n the Russell as low i n both pleasure and arousal.  This  (1980) circumplex model  The h o s t i l i t y scale, which i s less well  defined by Zuckerman and Lubin i n terms of i t s conceptual relationship to anxiety or depression, i s negatively  correlated with both pleasure (r = -.49)  and with arousal (r = -.21). Support for Hypothesis Two, that p o s i t i v e mood i s associated  with  a t t r i b u t i o n to i n t e r n a l as opposed to external causes, can be sought using the pleasure and depression scales as indicators of p o s i t i v e mood and the i n t e r n a l i t y scale of Russell's negatively  associated  associated  (r = .18).  (1982) CDS.  As Table 11 shows, depression i s  with i n t e r n a l i t y (r = -.12), and pleasure i s p o s i t i v e l y A similar pattern i s present for s t a b i l i t y of  attributions (r = -.12; r = .21, r e s p e c t i v e l y ) .  Controllability i s  s i g n i f i c a n t l y related to pleasure (r = .22) although not to depression ( r = .08).  These r e s u l t s are consistent  with Hypothesis Two.  p o s i t i v e mood make more internal a t t r i b u t i o n s .  Individuals  in a  Stated i n another way,  individuals who a t t r i b u t e t h e i r performance to internal causes tend to be i n a good mood.  The design of Study One precludes conclusions as to causality, so  we cannot determine whether a t t r i b u t i o n s influence moods, as Weiner, Russell, and  Lerman (1979) have shown, or whether moods influence a t t r i b u t i o n s , as  posed i n Hypothesis Two. It should also be noted that i n t e r n a l i t y , s t a b i l i t y , and c o n t r o l l a b i l i t y of a t t r i b u t i o n s are p o s i t i v e l y related to s a t i s f a c t i o n with performance (r =  (1) (1) (2) (3) (4) (5) (6) (7) (8) (9) ( 10) (11) (12) (13) (14) (15)  Table  Self-esteem Impression Management S a t i s f a c t i o n with task performance Task d i f f i c u l t y and cha11enge Task i n t e r e s t Task e f f o r t I n t e r n a l work motivation Perceived i nterna1i ty stabi1i ty controllabi1ity Anxiety Depression Hostility Pleasure Arousal  11.  (2)  (3)  (4)  (5)  (6)  (7)  (8)  (9)  (10)  (11)  (12)  (13)  (14)  . 239 . 172  - .078  - .061 .056 .051  .013 . 146 - .015  -.118 - .038 . 364  .416 .229  . 184  -.151  -.092  - .014  - .015  .046  .060 . 178 .040 -. 247 -.411 - . 146 .334 - .078  - .009 . 160 - .006 - .001 -. 137 - .229 . 194 .091  .268 .240 .207 - . 216 -.265 .050 .209 .035  - .032 -. 167 -.038 .113 .007 - .017 .093 .261  C o r r e l a t i o n s between Dependent  .220  - .002 . 179 - .045 . 198 . 150 . 171 .021 .009 - . 156 -.231 -.311 -.115 .340 .274 .369 ' .298  V a r i a b l e s , S t u d y One.  .318 .060 .068 . 198 .097 - .057 - .027 .181  .409 .208 - .079 - . 122 - .028 . 184 . 100  .082 - .094 - . 122 - .067 . 206 .038  - . 136 - .085 - .075 .219 .019  .538 . 145 - .409 -.336 --.677 . 284 • - .200  -- .491 -- .209  .213  100  .27, .24, .21, r e s p e c t i v e l y ) .  Taking c r e d i t for performance  by making an  i n t e r n a l a t t r i b u t i o n i s related to the whether that performance or not, as well as to mood.  i s successful  In Study One, the mean s a t i s f a c t i o n with  performance was 14.2, rated on a scale from 4 to 20. Overall, then, successful performance was related to i n t e r n a l i t y of a t t r i b u t i o n s and to p o s i t i v e mood, i n keeping with Weiner et a l . ' s model. To investigate whether the relationship between a t t r i b u t i o n s and mood d i f f e r s as a function of s a t i s f a c t i o n with performance, was performed.  a subgroup analysis  The sample was divided between those scoring above and below  the midpoint of the s a t i s f a c t i o n with performance measure.  The correlations  between a t t r i b u t i o n s and mood among those individuals l i k e l y to have experienced their performance as successful  (N = 146) were then compared to  those who had l i k e l y experienced i t as less than successful  (N = 42).  In the  former group, i n t e r n a l i t y , s t a b i l i t y and c o n t r o l l a b i l i t y of a t t r i b u t i o n s were p o s i t i v e l y related to pleasure (r = .16, .16, .19, respectively) but were unrelated  to depression (r = -.02, -.09, -.06, r e s p e c t i v e l y ) .  people who are successful, i n t e r n a l i t y i s associated least as measured by the pleasure scale.  Thus, among  with p o s i t i v e mood, at  For people scoring below the  midpoint on s a t i s f a c t i o n , the only s i g n i f i c a n t association was between pleasure and c o n t r o l l a b i l i t y (r = .26). for f a i l u r e were unrelated  In t h i s sample, then, attributions  to mood, individuals who were i n a good mood were  no more or less l i k e l y to report that they were responsible  for their low  performance. Overall, then, support i s evident for Hypothesis Two when performance i s successful. supported.  When s a t i s f a c t i o n with performance  i s low, Hypothesis Two i s not  Also not supported i s Weiner, Russell and Lerman's contention that  i n t e r n a l a t t r i b u t i o n s for f a i l u r e are associated  with negative mood.  Thus,  101  further investigation i s appropriate, to test the d i r e c t i o n of causality inherent i n Hypothesis Two and to evaluate the effect of mood on a t t r i b u t i o n s without reference to success or f a i l u r e . Although s t a t i c correlations between measures of mood and self-esteem cannot answer the question of whether self-esteem a f f e c t s resistance to mood change, the correlations i n Table 11 do show that mood and self-esteem are related.  As we would expect, high self-esteem i s p o s i t i v e l y associated with  p o s i t i v e mood, negatively associated with negative mood ( anxiety, r = -.25; depression, r = -.41, p < .001; h o s t i l i t y , r = -.15) and unrelated to arousal (r = -.08). With respect to task perceptions, s i g n i f i c a n t c o r r e l a t i o n s are present between self-reported e f f o r t and s a t i s f a c t i o n with performance (r = .36), task d i f f i c u l t y and challenge (r = .23) and task interest (r = .18).  Task interest  was s i g n i f i c a n t l y related to d i f f i c u l t y and challenge (r = .42) but unrelated to s a t i s f a c t i o n .  S a t i s f a c t i o n with task performance was marginally negatively  associated with task d i f f i c u l t y and challenge (r = -.12). The pattern of relationships between mood measures and task perceptions is interesting.  Satisfaction with performance was s i g n i f i c a n t l y negatively  associated with anxiety and depression, p o s i t i v e l y associated with pleasure, and unassociated with h o s t i l i t y or arousal.  Task d i f f i c u l t y and challenge was  related only to arousal, and the association was p o s i t i v e .  Task interest was  negatively related to depression and h o s t i l i t y , and p o s i t i v e l y related to pleasure and arousal.  Task e f f o r t was negatively associated with depression  and p o s i t i v e l y related to pleasure and arousal. Drawing broadly on the labels attached to the mood measures to describe the task perceptions, the pattern of correlations suggests the following conclusions: s a t i s f a c t i o n with performance i s moderately related to the  102  pleasantness of mood, whereas task d i f f i c u l t y and challenge i s related to i t s arousal component.  Task interest and e f f o r t are associated with both p o s i t i v e  mood and arousal. F i n a l l y , a precautionary note.  In t h i s study, sizeable correlations were  present between impression management and measures of task i n t e r e s t (r = .15), s t a b i l i t y of a t t r i b u t i o n s (r = .16), h o s t i l i t y (r = .23) and pleasure (r = .19, p < .01).  Impression management, which i s the conscious  presentation of a s o c i a l l y desirable front, was measured.  Relationships  between scales which correlate highly with i t are subject to special scrutiny as subject to measurement a r t i f a c t s (Zerbe & Paulhus, 1987).  Problems  associated with response bias are not, however, a necessary result of such zero order correlations (Ganster, Hennessey, & Luthans, 1983).  From Study One we can conclude that most of the measures used have good psychometric properties.  A few have unacceptable i n t e r n a l consistency and  have been discarded; they w i l l not be used further. acceptable r e l i a b i l i t y .  The remainder have  Further, shorter versions of scales which were  constructed for the adjective checklist mood measure were shown to have good reliability.  F i n a l l y , investigation of the associations between measures  revealed that most were related i n the way we would expect and that while some support was evident for hypotheses, further research was c l e a r l y necessary.  103  VI.  STUDY TWO: WHAT IS MOOD?  Study Two was undertaken to demonstrate the experimental manipulation of mood.  A v a l i d and r e l i a b l e mood induction i s required to test the  hypothesized effects of mood on expectancy b e l i e f s and task perceptions i n an experimental context.  This chapter reviews the use of mood manipulations i n  the l i t e r a t u r e and evaluates their s u i t a b i l i t y for the present research. A manipulation i s chosen for use, namely a musical mood manipulation, and an experiment  undertaken to show that the manipulation has appropriate effects on  various measures of mood.  Manipulation of Mood The increasing interest i n the relationship between emotion and thought or behavior has lead to the development of a number of mood induction procedures.  These have, for the most part, been developed and validated  within the context of a p a r t i c u l a r investigation.  T y p i c a l l y a manipulation i s  chosen which has face v a l i d i t y (e.g., a "funny" f i l m , f a i l u r e on a task, "sad" music) or for which a t h e o r e t i c a l rationale for an effect on mood exists (e.g., reading self-evaluative statements, remembering or imagining l i f e events).  The induction i s supported by the administration of a f f e c t - s e n s i t i v e  measures, such as self-reports (e.g., adjective c h e c k l i s t s , bipolar adjective scales), or psychomotor tasks (e.g., writing and counting speed).  These  manipulation checks are themselves constructed on the basis of t h e o r e t i c a l or empirical differences between mood states.  For example, individuals with  depressed mood r e l i a b l y exhibit slower psychological and motor responding, and c l i n i c a l l y depressed individuals, individuals i n a neutral mood state, and elated individuals d i f f e r i n the extent to which they endorse a f f e c t i v e  104  adjectives as s e l f - d e s c r i p t i v e .  Across d i f f e r i n g studies, mood manipulations  show convergent v a l i d i t y : the effects of d i f f e r e n t inductions generally agree, and the e f f e c t s of inductions on a f f e c t - s e n s i t i v e tasks match differences between c l i n i c a l and normal populations. The procedures employed i n the experimental l i t e r a t u r e i n order to induce mood have included providing feedback as to success and f a i l u r e (e.g. Feather, 1966;  Isen, 1970), hypnosis (e.g.  1982), providing an unexpected music (Clark, 1983;  g i f t (e.g.  Bower, 1981; Natale & Hantas, Isen, Shalker, Clark & Karp, 1978),  Sutherland, 1982), asking individuals to imagine or  remember emotion producing events (Nasby & Yando, 1982), watching funny or sad films (Isen & Gorgoglione, 1983), and the Velten technique, which involves reading statements designed to induce mood (Velten, 1968)  (for a review of  mood induction procedures see Goodwin & Williams, 1982). Some of these procedures are more suitable than others to the investigation of expectancy.  S i m i l a r l y , some manipulations are more suitable  to p a r t i c u l a r research designs.  Within-subjects designs, i n which each  subject receives a l l experimental treatments, are vulnerable to hypothesis guessing and demand c h a r a c t e r i s t i c s .  Because multiple treatments are  experienced by participants, they can more e a s i l y guess what a study i s about and perhaps behave to confirm the hypothesis they have formed.  A mood  induction procedure employing d i f f e r e n t films could be susceptible to such a r t i f a c t s because of the films' obvious content. unexpected  S i m i l a r l y , providing an  g i f t might be a r e l i a b l e way to produce p o s i t i v e mood.  Such a  procedure has an organizational analogue i n the provision of bonuses (c.f., Boggiano & Hertel, 1983). obvious.  However, a comparable negative induction i s not  Taking away a g i f t might have unpredictable effects on mood, as  l i k e l y producing anger as sadness.  105  In general the c r i t e r i a for a desirable mood induction i s that i t produces mood, that i s has good construct v a l i d i t y , and that i t does not foster threats to i n t e r n a l v a l i d i t y .  Inducing mood through hypnosis, f o r  example, i s u n r e l i a b l e : some individuals are r e s i s t a n t to hypnotic suggestions,  leading to a special v a r i e t y of s e l f - s e l e c t i o n .  Hypnosis has  also been c r i t i c i z e d as a r t i f a c t u a l l y producing mood-maintenance e f f e c t s (Isen, 1984). Success or f a i l u r e on a task has high face v a l i d i t y as a source of mood effects i n organizational settings.  In an experimental exploration of the  effects of mood on task expectancies such a procedure i s undesirable, however, because i t confounds induced mood with information about the p r o b a b i l i t y of task accomplishment.  As Kavanagh and Bower (1985) put i t ,  "feedback about one's success or f a i l u r e at a task can modify one's emotional state as well as provide 'cognitive information' about one's c a p a b i l i t y at that s p e c i f i c task" (p.510). In the Velten technique, used most often to experimentally subjects read a set of s e l f - r e f e r e n t statements.  induce mood,  Those i n the e l a t i o n  induction condition, for example, progress from neutral to elated. For instance, from "Today i s neither better nor worse than any other day", to "Things w i l l be better and better today", and "God, I f e e l great". depression  In the  induction condition statements progress from neutral to depressive.  Although the Velten technique produces a close analogue to n a t u r a l l y occurring mood (Clark, 1983), i t has been c r i t i c i z e d as containing strong demands to act elated or depressed (Buchwald, Strack, & Coyne, 1981; Polivy & Doyle, 1980), and as f a i l i n g to induce mood i n some people (Clark, 1983). A more damaging c r i t i c i s m of the Velten procedure i s that i t has a large cognitive aspect  (Isen, 1982).  The o r i g i n a l statements include many phrases  that concern b e l i e f s about success (e.g., "I know I've got what i t takes to  106  succeed", and "I am discouraged and unhappy about myself").  C l e a r l y the  p o t e n t i a l for ideational as well as a f f e c t i v e changes i s present i n such statements.  In defence of t h i s c r i t i q u e , Isen points out that effects on  thought and behavior of the Velten, hypnosis, success-failure, and g i f t procedures generally agree.  However, when the dependent v a r i a b l e of interest  i s related to achievement or task performance, as i n the proposed study, the cognitive statements i n the Velten technique represent an unacceptable confound. A mood induction procedure that i s gaining prominence as an a l t e r n a t i v e to the Velten technique i s a musical induction procedure.  This involves the  use of suggestive music alone ( P i g n a t i e l l o , Camp, & Rasar, 1986) or as a background to subjects' own e f f o r t s to develop a p a r t i c u l a r mood, as i s often done (e.g., Clark & Teasdale, 1982; Eich & Metcalfe, i n press; Richards, 1981; Sutherland, Newman & Rachman, 1982; Sutton & Teasdale, 1982; Teasdale & Spencer, 1982, 1984).  The musical induction procedure has been shown to have  good construct v a l i d i t y , produce stronger s h i f t s i n mood than the Velten procedure, and to r e l i a b l y produce mood i n most people. Clark (1983) evaluated the Velten and musical mood induction procedures by comparing their effects to those of n a t u r a l l y occurring depressed mood. For example, musical induction of sadness r e s u l t s i n higher levels of s e l f reported sadness or despondency r e l a t i v e to the elation induction (Clark & Teasdale, 1982; Sutherland, Newman & Rachman, 1982 Sutton & Teasdale, 1982; Teasdale & Spencer, r e l a t i v e to e l a t i o n .  1982, 1984), as does naturally occurring depression The musical induction procedure has also been shown to  influence measures of psychomotor retardation, as occurs i n naturally occurring depression, such as counting out loud (Clark & Teasdale, 1982; Richards, 1981; Sutton & Teasdale, 1982; Teasdale & Spencer,  1982) and writing  107  speed (Richards, 1981).  Subjects i n the sadness condition of the musical  induction procedure r e c a l l more negative words than p o s i t i v e words, while subjects i n the elation condition r e c a l l more p o s i t i v e words (Clark & Teasdale, 1982).  Teasdale and Spencer  (1982, 1984) used the musical induction  procedure to show that depressed subjects gave lower estimates of the p r o b a b i l i t y of future success and lower r e c a l l for past successes than elated subjects.  Eich and Metcalfe ( i n press) have recently used the musical  induction procedure to show mood dependent memory e f f e c t s .  Subjects who  generated word associations i n a p a r t i c u l a r musically induced mood condition were more l i k e l y to r e c a l l them when i n the same mood than when their mood was altered. The Velten and musical procedures have been d i r e c t l y compared.  Clark  (1983) combined the data from Clark and Teasdale (1982), who used the musical induction procedure, and Teasdale and Russell (1982), who used the Velten procedure.  These two studies came from the same laboratory, used the same  self-report measures and drew subjects from the same student population. Clark reported s i g n i f i c a n t l y greater e f f e c t s of the musical procedure on ratings of despondency and happiness.  Eich and Metcalfe ( i n press) report  that i n a p i l o t study, the Velten technique produced s i g n i f i c a n t but only short l i v e d s h i f t s i n s e l f - r a t i n g s .  When using the Velten technique they  found no evidence of mood effects on memory.  In contrast, when using the  musical technique they were able to show mood e f f e c t s . F i n a l l y , the music induction procedure appears to produce mood more r e l i a b l y than the Velten procedure.  Sutherland et a l .  (1982) found that  100%  of subjects who underwent a musical mood induction met a predetermined mood change c r i t e r i o n compared to only 68% i n a study employing the Velten procedure.  Clark (1983) c i t e s Clark and Teasdale's (1982) finding that 87% of  108  subjects report, i n a post-experimental questionnaire, that they experienced a genuine change of mood as a r e s u l t of the musical procedure, as compared to Polivy and Doyle's (1980) finding that only 50% of subjects reported genuine mood change as a r e s u l t of the Velten technique. In the studies reviewed above, the musical mood induction procedures employed used music i n addition to verbal instructions to "to try to develop" the specified mood, usually by "thinking about past events".  Clearly  specifying a desired mood state raises the p o s s i b i l i t y that demand c h a r a c t e r i s t i c s are produced, as Clark (1983) warns.  He also says, though,  that the music technique might not be e f f e c t i v e without such additional effort. P i g n a t i e l l o , Camp, and Rasar (1986) have, however, developed a musical mood induction that does not include such i n s t r u c t i o n s . They used n o n - l y r i c a l selections from c l a s s i c a l , popular and musical  score soundtracks.  These were  rated by judges, including music therapists, according to how depressing/elating they were. high i n i n t e r r a t e r r e l i a b i l i t y .  Three inductions were created using selections The selections chosen were ordered such that  each induction condition started with the same neutral selection and then became either successively more e l a t i n g , more depressing,  or remained neutral.  This ordering was based on two p r i n c i p l e s used i n therapeutic settings to a l t e r the mood of i n d i v i d u a l s : the "Iso" p r i n c i p l e and "vectoring" (Shatin, 1970).  The "Iso" p r i n c i p l e refers to the process of matching the mood of the  music to the mood of the subject i n order to a l t e r mood with gradual changes i n the music.  These gradual changes are termed "vectoring", and are the  directed movement of music towards the desired goal, such as from sadness to cheerfulness, or as i n the case of P i g n a t i e l l o and colleagues' induction, from a neutral to a happy mood.  109  P i g n a t i e l l o and colleagues report two assessments of the v a l i d i t y of their induction.  In the f i r s t experiment a s i g n i f i c a n t e f f e c t of mood on  self-reported depression was found. writing speed or time estimation. writing speed pretest. reported depression.  No influence was found on a measure of A second experiment, however, included a  Again, s i g n i f i c a n t differences were found f o r s e l f When pretest writing speed was used as a covariate a  s i g n i f i c a n t e f f e c t of mood on t h i s psychomotor task was revealed. on time estimation were found.  No effects  F i n a l l y , when subjects who had indicated that  they had guessed that the experiment dealt with mood were excluded from the analysis, the same results were evident. In summary, the music induction i s the preferred method f o r experimentally  inducing mood (Clark, 1983; Eich & Metcalfe, i n press).  I t has  been shown to produce a strong analog to n a t u r a l l y occurring mood, has produced e f f e c t s on a f f e c t - s e n s i t i v e tasks, and when used i n the absence of verbal instructions to achieve a p a r t i c u l a r mood state, i s less open to the c r i t i c i s m that demand c h a r a c t e r i s t i c s are present. for use i n the present  study.  It i s therefore proposed  S p e c i f i c a l l y , the induction developed by  P i g n a t i e l l o and colleagues w i l l be used.  To further demonstrate the v a l i d i t y  of this manipulation i n the subject population of i n t e r e s t , Study Three was undertaken. S p e c i f i c a l l y , i t i s predicted that individuals i n the Depression condition of the musical mood induction w i l l self-report greater than w i l l individuals i n the E l a t i o n condition.  depression  The scores of individuals i n  the Neutral condition are expected to be intermediate.  In keeping with the  circumplex model of emotion (Russell, 198x), i t i s predicted that the E l a t i o n induction w i l l r e s u l t i n high scores on pleasure and arousal and that the Depression induction w i l l r e s u l t i n low scores on self-reported pleasure and  110  arousal.  Method Procedure Twenty-five second year business students p a r t i c i p a t e d i n the study. Each was recruited as part of a study of "Music i n the Workplace", and received course c r e d i t for h i s or her p a r t i c i p a t i o n .  Of the p a r t i c i p a n t s , 13  (52%) were men, 12 (48%) were women, and their median age was 20 years.  Each  participant was randomly assigned to one of three mood induction conditions. On a r r i v a l to the experimental sessions, each participant was seated i n a comfortable chair facing two loudspeakers and provided with the following instructions: Music i n the Workplace Informed Consent Form The study i n which you are being asked to p a r t i c i p a t e i s part of an investigation of the use of music i n the workplace. Many organizations play music as a background to work. The study you are i n i s examining the use of music during work and people's perceptions of the use of music. Later t h i s term, we w i l l be asking participants to help with two parts of t h i s study. In the f i r s t part they w i l l be asked to complete a simulated business decision-making task on a computer terminal while music i s played i n the background. In the second part, we would l i k e p a r t i c i p a n t s to l i s t e n to some music and then we w i l l ask them some questions about the music. In today's session, we would l i k e your help with this second part only. So, f i r s t we would l i k e you to s i t back and l i s t e n to some music that might be used at work, and then we w i l l ask you some questions about your reactions to the music. The music you w i l l l i s t e n to w i l l probably evoke d i f f e r e n t reactions i n d i f f e r e n t people. These honest reactions are what we are interested i n . Your p a r t i c i p a t i o n i n this study i s voluntary, you are free to discontinue p a r t i c i p a t i o n at any time without penalty. Your responses w i l l be used only for the purposes of this study and w i l l be kept c o n f i d e n t i a l . If you wish to participate,, please sign below, i n d i c a t i n g that you have, read this form and give your informed consent to p a r t i c i p a t e i n the study.  Ill  Each p a r t i c i p a n t then l i s t e n e d t o one d e v e l o p e d by P i g n a t i e l l o et a l (1986). musical  of the t h r e e mood i n d u c t i o n tapes  Each r e c o r d i n g  l a s t e d 20 minutes.  s e l e c t i o n s i n c l u d e d i n each i n d u c t i o n a r e l i s t e d i n Appendix  The  B.  Measures Following administered students, The monitor.  the mood i n d u c t i o n , s e l f - r e p o r t measures of mood s t a t e were  on a computer t e r m i n a l .  business  f a m i l i a r w i t h the use of the computer keyboard. dependent measures were p r e s e n t e d item by For  w i t h an example.  of the measures  item.  Following  each response, the  the next item p r e s e n t e d .  The  the response t o each item as i t was  the i n s t r u c t i o n s t o the f i r s t  entered.  set of items and  provided  and  screen would  program a d m i n i s t e r i n g Figures  the  video  described  shown i n s t r u c t i o n s f o r the measure and  P a r t i c i p a n t s were then asked t o c l e a r the d i s p l a y  respond t o the f i r s t c l e a r e d and  item on the computer  each set of items c o r r e s p o n d i n g t o one  below, p a r t i c i p a n t s were f i r s t  captured  A l l p a r t i c i p a n t s were, as  be  items  5 and  6 show  the d i s p l a y of the f i r s t  item,  "active", respectively.  Adjective Checklist.  The  first  measure of s e l f - r e p o r t e d mood was  the  b r i e f v e r s i o n of the M u l t i p l e A f f e c t A d j e c t i v e C h e c k l i s t (MAACL) (Zuckerman & Lubin,  1968), as d e s c r i b e d  Semantic D i f f e r e n t i a l . Mehrebian and  Russell's  pleasure-displeasure  and  Response l a t e n c y .  i n Study  The  (1970) 12  One.  second measure of item  s e l f - r e p o r t e d mood  was  semantic d i f f e r e n t i a l measure of  arousal-sleepiness,  also described  In a d d i t i o n t o c a p t u r i n g  i n Study  One.  the n u m e r i c a l response t o  each s e l f - r e p o r t item, the computer program timed the l a t e n c y of r e s p o n s e . each item was  p r e s e n t e d , the computer r e c o r d e d  the number of m i l l i s e c o n d s  As from  112  Music Reaction  Questionnaire  The following set of questions are designed to measure your reactions to the music you just heard. We would l i k e you to use the words and scales you w i l l see to describe your feelings right now. You w i l l be presented with words that describe f e e l i n g s . Please use the following scale to i n d i c a t e how well these words describe your f e e l i n g s . 1 2 definitely do not f e e l  do not feel  3 slightly feel  4 definitely feel  If the word d e f i n i t e l y describes how you f e e l at the moment you read i t , respond with a 4. . If the word only s l i g h t l y describes how you f e e l at the moment, respond with a 3, and so on. Work quickly, do not spend a long time on one word. Ready to go?  Figure 5.  Instructions for the MAACL  active 1 definitely do not f e e l  2 do not feel  3 slightly feel  How well does active describe the way you feel?  Figure 6 .  Terminal display f o r the item "Active".  4 definitely feel  113  completion of the item display to completion of the subject's response.  This  constituted an unobtrusive measure of depressed mood. Placed between the two self-report mood measures were 5 bogus items intended to reinforce the cover story.  These asked about the s u i t a b i l i t y of  the music to various kinds of work settings.  Results A multivariate analysis of variance (MANOVA) was performed on the data using the SPSSX s t a t i s t i c a l software (SPSS Inc., 1983).  Before presenting the  r e s u l t s of the MANOVA procedure, the v a l i d i t y of assumptions underlying the technique w i l l be evaluated.  Evaluation of Assumptions. The assumptions of MANOVA are (1) multivariate normality, (2) homogeneity of variance-covariance matrices, (3) l i n e a r i t y of relationships among dependent variables, and (4) freedom from m u l t i c o l l i n e a r i t y and s i n g u l a r i t y . The mathematical model underlying MANOVA i s based on the multivariate normal d i s t r i b u t i o n .  It requires that the sampling d i s t r i b u t i o n of the means  of the various dependent variables i n each c e l l are normally d i s t r i b u t e d .  In  univariate analysis, the sampling d i s t r i b u t i o n of means can be expected to approach normality for large samples.  MANOVA has also been shown to be robust  to modest v i o l a t i o n of normality i f the v i o l a t i o n i s created by skewness rather than by o u t l i e r s (Mardia, 1971).  Tabachnik and F i d e l l (1983) say that  robustness i s ensured with a sample size that produces 20 degrees of freedom for error i n the univariate case so long as sample sizes are equal and two t a i l e d tests are used. In t h i s study robustness was assured by equating the sample size i n each  114  c e l l of the design.  One male participant was discarded randomly from the  Depression treatment group.  This resulted i n a t o t a l sample size of 24, 4  participants i n each combination of three mood treatments and 2 gender conditions.  Nevertheless, normality was assessed by comparing the skew of  each d i s t r i b u t i o n of scores to the standard error of skewness. six dependent variables was  s i g n i f i c a n t l y skewed.  None of the  Concomitantly,  examination  of extreme scores revealed no values beyond 3 standard deviations from the mean of each group (to mitigate the e f f e c t s of low power, the search for o u t l i e r s and skew was also conducted  i n the three groups formed by combining  gender conditions.) The presence of multivariate o u t l i e r s was evaluated by examining the Mahlanobis distance (D ) from each case to the centroid of i t s z  group.  The Mahlanobis distance i s d i s t r i b u t e d as a chi-square variable,  therefore a c r i t i c a l value for extreme cases can be computed.  No s i g n i f i c a n t  multivariate o u t l i e r s were found. Significance tests i n MANOVA are robust to both heterogeneity of variance and non-normality for  i f sample sizes are equal and exceed 20 degrees of freedom  error i n the univariate case (Hakstian, Roed and Lind, 1979; Tabachnik and  F i d e l l , 1983).  In this study robustness to v i o l a t i o n of the assumption of  homogeneity of variance-covariance matrices was guaranteed by equal sample size.  Examination  of the r a t i o of largest to smallest variances for each  dependent v a r i a b l e across groups revealed some ratios i n excess of Tabachnik and F i d e l l ' s (1983) suggested c r i t e r i o n of 20:1.  None of the univariate  homogeneity of variance tests across the six groups was s i g n i f i c a n t , although these were based on small samples.  Homogeneity of variance tests across the  three treatment groups only, a more powerful test, also showed no s i g n i f i c a n t r e s u l t , although that for h o s t i l i t y approached significance (p= .062).  The  multivariate test of homogeneity of dispersion across the three mood treatment  115  groups was not s i g n i f i c a n t (Box's M = 72.43, F(42,1397) = 1.00). The MANOVA model assumes that the i n t e r r e l a t i o n s h i p s among a l l DV's are linear within each c e l l .  Deviation from l i n e a r i t y w i l l reduce the power of  s t a t i s t i c a l tests i n that linear combinations of dependent variables w i l l not show maximum relationship with the independent variables.  The significance  test produced by MANOVA are tests of linear relationships.  They do not  capture nonlinear relationships but at the same time provide unbiased tests of linear r e l a t i o n s h i p s .  Therefore, non-linearity w i l l r e s u l t i n weaker  significance tests, i t biases tests i n the d i r e c t i o n of greater conservativeness. versus predicted  Linearity was assessed by examining the plots of observed residuals.  No evidence of gross c u r v i l i n e a r i t y was found.  M u l t i c o l l i n e a r i t y occurs when two variables are p e r f e c t l y or nearly p e r f e c t l y correlated and show a similar pattern of correlations with other dependent v a r i a b l e s .  Singularity occurs when one score i s a linear or nearly  linear combination of others. problems for multivariate unstable matrix inversion.  M u l t i c o l l i n e a r i t y and s i n g u l a r i t y pose similar  analyses.  S p e c i f i c a l l y , they prohibit or render  Portions of the multivariate  solution that  m u l t i p l i c a t i o n by an unstable inverted matrix are also unstable.  follow  One method  of dealing with m u l t i c o l l i n e a r i t y or s i n g u l a r i t y i s to delete the offending variable or variables.  Because one variable i s a linear combination of  others, t h i s does not r e s u l t i n any loss of information. and  Multicollinearity  s i n g u l a r i t y were assessed by examining the squared multiple  (SMC)  of each variable predicted  by a l l others.  correlation  No SMC's approached 1.0.  S i m i l a r l y , the determinant of the w i t h i n - c e l l s c o r r e l a t i o n matrix was nonzero. Homogeneity of regression,  or the assumption that the slope of the  116  regression of dependent variables on covariables i s equal across c e l l s , i s a requirement of analysis of covariance and stepdown analysis i n MANOVA. In t h i s study, test of homogeneity of regression were run for a l l variables, since each serves as covariates for a l l others i n stepdown F-tests. Homogeneity of regression was achieved for a l l components. Given succesful evaluation of the assumptions underlying the procedure, MANOVA proceeded.  Multivariate analyses of variance. A 2x3 multivariate analysis of variance was performed on the six dependent v a r i a b l e s : anxiety latency.  , depression,  h o s t i l i t y , pleasure, arousal and  Independent variables were mood treatment ( e l a t i o n , neutral, and  depression) and gender (male and female).  Total N was 24 with an equal number  of subjects i n each c e l l . Wilk's c r i t e r i o n indicated a s i g n i f i c a n t effect of the mood treatment on the combined dependent variables, F(12,26)= 5.96, of gender was not s i g n i f i c a n t , F(6,13)= 1.11,  p < .001.  p > .05.  i n t e r a c t i o n was also not s i g n i f i c a n t , F(12,26)= 2.02,  The main e f f e c t  The gender by group p > .05.  To simplify  the discussion that follows, and to provide increased degrees of freedom for subsequent analyses because the gender e f f e c t was non-significant, the data for males and females were combined.  The multivariate test of significance  for the one-way effect of treatment group produced an o v e r a l l F value of 6.02 (d.f= 12,32; p < .001).  This r e f l e c t e d a strong association between the  e f f e c t of the mood induction treatment and the combined dependent variables (Wilk's A= 0.094;  7? = 2  .906).  Because the omnibus MANOVA shows s i g n i f i c a n t multivariate e f f e c t s , the univariate effects can be examined.  S i g n i f i c a n t univariate effects of the  117  mood induction treatment were found for the measures of latency, pleasure, depression, and h o s t i l i t y .  arousal,  The univariate test of significance for  the six dependent variables are shown i n Table 12.  These results indicate  that the mood manipulation had effects on the variables of i n t e r e s t , namely response latency, self-reported depression, pleasure and arousal.  The  pattern  of means, shown i n Table 13, i s as hypothesized. Univariate procedures do not, however, take into account correlations between the dependent variables.  Examination of the pooled w i t h i n - c e l l  c o r r e l a t i o n matrix, shown i n Table 14 reveals, as expected, that the measures of mood are l a r g e l y i n t e r c o r r e l a t e d . was  various  B a r t l e t t ' s test of sphericity  s i g n i f i c a n t (x (15)= 43.46, p<.001) supporting r e j e c t i o n of the hypothesis 2  that the c o r r e l a t i o n matrix i s an i d e n t i t y matrix or that the variables are independent.  Therefore, stepdown analysis i s  appropriate.  Stepdown analysis determines the contribution of each dependent variable while c o n t r o l l i n g for i t s relationship to other variables. also controls for Type I error rate.  Stepdown analysis  In t h i s study, the dependent variables  were entered into the analysis i n the following order: latency, pleasure, depression, anxiety and h o s t i l i t y .  arousal,  In the stepdown analysis  procedure, the f i r s t variable i s tested i n an univariate ANOVA.  The  second  variable i s then tested i n an analysis of covariance with the f i r s t variable as the covariate. covariates, and  The t h i r d variable i s tested with the f i r s t two  as  so on.  S p e c i f i c a t i o n of the order of entry or variables into stepdown analysis i s on the basis of the importance of the dependent variables to the hypotheses of i n t e r e s t , as determined by the researcher (Tabachnik & F i d e l l , 1983). Clearly, some orders w i l l r e s u l t i n a larger number of s i g n i f i c a n t dependent variables than others.  Because stepdown analysis accounts for correlations  118  Latency  Arousal  Pleasure  Depression  Anxiety  Latency Arousal  .028 -.201  .337  Depression  .280  -.128  -.544  Anxiety  .311  .451  -.301  .503  Hostility  .037  .203  -.594  .383  Pleasure  Table 12.  Correlations for dependent variables, Study Two.  .577  119  Univarie ite F  d.f.  P  4.37 9.95 22.12 45.09 0.69 11.20  2/21 2/21 2/21 2/21 2/21 2/21  .026 .001 .000 .000 .513 .000  Dependent Variable LatencyArousal Pleasure Depression Anxiety Hostility  Stepc [own F  4.37 7.43 8. 86 6. 10 1.95 0.62  T  Table 13.  Univariate and Stepdown F-tests, Study Two.  d.f.  P  2/21 2/20 2/19 2/18 2/17 2/16  .026 .004 .002 .009 .173 .552  .29 .43 .50 .40 .19 .07  120  Elal :ion  Neuti-al  Depresssion  Mean  St. Dev.  Mean  St. Dev.  Mean  St. Dev.  16 .38 34 .63 25 .00 41 .88 29 .87 234 .47  4 .44 5 .40 5 .81 4 .32 8 .29 38 .67  14 .50 39 .25 23 .37 40 .38 19 .75 258 .40  4 .47 5 .29 2 .20 6 .30 5 .87 46 .13  17 .37 66 .25 32 .62 25 .62 15 .37 312 .04  5 .88 9 .91 3 .70 5 .39 5 .50 71 .05  15 .75 38 .25 22 .75 42 .75 25 .50 212 .74  5 .68 4 .78 4 .03 2 .75 5 .26 32 .49  13 .50 40 .25 24 .25 40 .25 20 .25 231 .64  4 .12 6 .39 2 .87 6 .13 8 .22 35 .91  17 .25 66 .75 32 .00 23 .75 14 .75 341 .48  2 .06 5 .97 3 .16 6 .50 6 .18 79 .54  17 .00 31 .00 27 .25 41 .00 34 .25 256 .21  3 .56 3 .16 6 .99 5 .83 9 .03 34 .25  15 .50 38 .25 22 .50 40 .50 19 .25 285 .15  5 .19 4 .65 1 .00 7 .42 3 .50 42 .03  17 .50 65 .76 33 .25 27 .50 16 .00 282 .59  8 .73 13 .89 4 .57 4 .04 5 .59 56 .04  Group Measure Combined (n=8) Anxiety Depression Hostility Pleasure Arousal Latency Males (n=4) Anxiety Depression Hostility Pleasure Arousal Latency Females (n=4) Anxiety Depression Hostility Pleasure Arousal Latency T  Table 14. Means and Standard Deviations, Treatment and Gender Groups, Study Two.  121  between variables and sets experimentwise Type I error rate, the procedure permits such s p e c i f i c a t i o n . The r e s u l t s of the stepdown analysis are shown i n Table 12. Response latency was strongly associated with mood induction treatment, stepdown F(2,21) = 4.37, p < .05. Strength of association (rj ) between the independent 2  v a r i a b l e and response latency, as indicated by the r a t i o of hypothesis to t o t a l sum of squares, was .29. As Table 13 shows, mean latency of response was greatest i n the depression  induction group (312.0) and least i n the  e l a t i o n induction group (234.5).  After the pattern of differences measured by  latency was entered, a difference was also found for self-reported arousal, stepdown F(2,20) = 7.43, p < .01, rj =.43. 2  Self-reported arousal was greatest  i n the e l a t i o n group (29.9) and least i n the depression  group (15.4).  The  t h i r d v a r i a b l e entered, self-reported pleasure, was s i g n i f i c a n t l y related to the mood treatment, stepdown F (2,19) = 8.86, p < .01, T} = .50. 2  Self-reported  pleasure was highest i n the e l a t i o n induction condition (41.88), although not very much more so than i n the neutral condition (40.38) while the depression group was much lower (25.62).  Self-reported depression,  the fourth variable  entered, also made a unique contribution to the o v e r a l l difference between mood treatment groups, stepdown F (2,18)= 6.10, p<.01, T J = .40. As i n the 2  case of self-reported pleasure,  the mean for the e l a t i o n condition was at one  extreme (34.63) but d i d not d i f f e r greatly from the neutral condition (39.25) while the depression (66.25).  induction group reported the greatest depressed mood  Anxiety, which would not have contributed to the difference between  groups i n a univariate context, was not s i g n i f i c a n t i n the stepdown analyses. F i n a l l y , the measure of h o s t i l i t y , which showed a s i g n i f i c a n t univariate F, was not s i g n i f i c a n t when variance i t shared with variables entered into the stepdown analysis before i t was accounted f o r .  122  Summary It i s evident from the r e s u l t s of the multivariate analysis of variance that s i g n i f i c a n c e differences i n self-reported mood state resulted from the musical mood induction procedures. between males and females.  No differences i n responses were found  I t was also evident from the r e s u l t s that the  association between the mood treatment and scores on the combined mood measures was very strong.  The v a l i d i t y of the manipulation i s strongly  supported. Examination of group means and of the r e s u l t s of stepdown analyses revealed that the mood induction treatment had the expected e f f e c t on the measures.  The pattern of responses r e f l e c t e d greatest negative a f f e c t i n the  depression  induction as indicated by self-reported depression, pleasure and by  response latency, a behavioral measure of psycho-motor depression. negative a f f e c t was indicated i n the e l a t i o n induction condition. mood induction produced intermediate  Least The neutral  scores on these measures, although they  were closer to those i n the e l a t i o n condition.  This suggests that while there  i s strong evidence that the e l a t i o n , neutral, and depression  induction  conditions produce an ordering of mood from unhappy to happy, the mood induced by the neutral treatment was not unlike that of the e l a t i o n condition. Conversely, i t could be argued that i t was more d i f f i c u l t to produce e l a t i o n r e l a t i v e to neutral mood, than i t was to produce depressed mood. The r e s u l t s f o r the three measures not yet discussed, anxiety,  hostility,  and arousal, both confirm the v a l i d i t y of the musical mood manipulation and suggest an elaboration.  F i r s t , the induction had no e f f e c t on the self-report  measure of anxiety and, after accounting  for the contribution of other  variables, had no effect on the measure of h o s t i l i t y . discriminant v a l i d i t y .  This i s evidence of  In combination with evidence that the mood  123  manipulation does influence convergent measures, this provides for the construct v a l i d i t y of the mood manipulation.  strong support  That i s , the induction  does produce r e l a t i v e elation and depression. In our sample, however, i t appears that the mood produced by the neutral induction condition does not d i f f e r very much i n terms of self-reported depression  from that i n the e l a t i o n condition.  The results for the measure of  self-reported arousal suggest an elaboration, however. semantic d i f f e r e n t i a l measures of pleasure-displeasure  Recall that the and arousal-sleepiness  come out of a t h e o r e t i c a l framework that proposes that these dimensions are orthogonal and that descriptors of moods form a circumplex marked by these two dimensions.  E l a t i o n , i n this framework, i s characterized by high pleasure and  high arousal.  Depression i s characterized by low pleasure and low arousal.  The results for the elation and depression  induction condition f i t t h i s model.  Viewed i n this way, the neutral induction condition i n this study i s shown to have been neutral i n terms of arousal-sleepiness, but somewhat pleasant.  That  i s , examining the mean scores on the arousal and pleasure measures we f i n d that the mean score for self-reported arousal f o r the neutral induction i s intermediate between that of the e l a t i o n and depression s l i g h t l y closer to depression  than e l a t i o n .  conditions, being  On the pleasure dimension,  however, the mean pleasure score of the neutral group i s close to that of the e l a t i o n group.  In other words, rather than producing a mood state that i s  neutral i n a l l respects, the neutral induction i n this study produced a state that Russell would c a l l "calm"; unaroused but pleasant. This finding affirms Russell's contention that i n order to describe mood states, two dimensions are necessary; one i s not s u f f i c i e n t .  Most discussions  and most examinations of the v a l i d i t y and impact of mood inductions only differences on a dimension of pleasantness.  consider  124  In summary, i t i s evident that the musical mood induction procedure developed  by P i g n a t i e l l o and colleagues produces mood states of elation and  depression, as indicated by self-report and behavioral measures. Alternative explanations, such as demand c h a r a c t e r i s t i c s or hypothesis guessing are u n l i k e l y explanations of the e f f e c t on the unobtrusive, behavioral latency measure, or of the intermediate impact of the neutral induction.  Further use  of t h i s manipulation i s therefore appropriate.  Discussion: What i s Mood? In Chapter Two, which reviewed theories of and approaches to emotions, the position of the fundamental emotion theorists was outlined.  This p o s i t i o n  holds that there are a d i s c r e t e number of q u a l i t a t i v e l y d i f f e r e n t emotions. Joy, for example, i s a d i f f e r e n t emotion than sadness. circumplex model o f emotions.  In contrast i s the  By holding that the r e l a t i o n s h i p s between  emotions can be mapped on a two-dimensional space defined by pleasure and arousal, Russell (1980) places joy and sadness on one continuum.  Distress and  contentment are not q u a l i t a t i v e l y d i f f e r e n t , but are d i f f e r e n t ends of one dimension, representing d i f f e r e n t combinations of pleasure and arousal. Study Two showed that the musical mood manipulation d i d produce the hypothesized  states of e l a t i o n and depression.  circumplex model. inductions.  I t also affirmed the  Both pleasure and arousal measures were affected by the  In other words, both dimensions were needed to characterize the  mood induced i n Study Two; both pleasure and arousal are necessary components of mood. Mood researchers sometimes adopt a unitary approach to mood, overlooking the arousal component of mood, although i t can illuminate otherwise research r e s u l t s .  cloudy  Consider, f o r example, two studies of mood and performance  125  appraisal.  In the f i r s t ( S i n c l a i r , i n press), raters were provided with  information about a target i n d i v i d u a l . The mood of raters was  then  manipulated using the Velten procedure, which produces happy and sad mood. Raters were then asked to evaluate the target i n d i v i d u a l . S i n c l a i r found that raters i n a p o s i t i v e mood were more p o s i t i v e i n their evaluations and r e c a l l e d more p o s i t i v e information about the target than did raters i n a negative mood. He also found that raters i n a depressed mood made more accurate  evaluations  and displayed greater dispersion of ratings across dimensions and hence less halo error, consistent with the argument that elated i n d i v i d u a l s make broader categorizations (Easterbrook,  1959;  Isen & Daubman, 1984).  The second study, by Srinivas and Motowidlo (1987), investigated the effect of stress on performance ratings. a s t r e s s f u l event produced negative mood.  The authors predicted and found that Although they did not f i n d an  e f f e c t of mood of the f a v o r a b i l i t y of ratings, Srinivas and Motowidlo d i d f i n d that ratings made by people who  underwent a s t r e s s f u l experience showed less  dispersion across performance dimensions.  In other words, i n d i v i d u a l s i n a  more negative mood displayed more halo, they made narrower categorizations. On the surface, then, the findings of S i n c l a i r oppose those of Srinivas and Motowidlo.  In the former study negative mood was associated with greater  dispersion of ratings, i n the l a t t e r negative mood was dispersion.  associated with less  This seeming contradiction can be resolved i f we consider  r e l a t i v e l e v e l of arousal i n the two studies.  the  In the f i r s t study the Velten  procedure produced e l a t i o n and depression, depression being characterized i n the circumplex model as low pleasure and low arousal. depression  induction contains statements l i k e "I'm  Indeed the Velten  so t i r e d . "  In contrast, the negative condition i n the Srinivas and Motowidlo study  126  i s characterized by high arousal.  On the face of i t s t r e s s f u l experiences are  more arousing than non-stressful ones.  Further, Srinivas and Motowidlo report  stronger associations between their treatment and h o s t i l i t y or anxiety than depression. It i s possible, then, that the differences i n dispersion i n the two studies were a result of the arousal component of mood.  The greater  dispersion i n S i n c l a i r ' s negative mood group may have been a function of the lower arousal r e l a t i v e to the p o s i t i v e mood group.  In the same way the  greater dispersion i n Srinivas and Motowidlo's p o s i t i v e mood, low stress group may have been a r e s u l t of lower arousal.  In sum, then, arousal i s a possible  explanation for what otherwise are contradictory r e s u l t s . So, mood i s not a unitary construct.  By assuming that i t varies only  along one dimension from p o s i t i v e to negative we overlook much of the domain of mood i t s e l f , as well as much of the domain.to which i t might be relevant. Future research i n organizational behavior would be wise to acknowledge a more complete conception of mood, recognizing that the manipulation of mood produces many effects and that measures which capture these effects should be employed.  127  VII.  STUDY THREE: WHAT IS EXPECTANCY?  Study Three was undertaken to develop and v a l i d a t e a measure of perceived effort-performance covariation or expectancy.  In the following sections, the  t h e o r e t i c a l and empirical background to expectancy measurement i s reviewed.  Conceptualizing and measuring expectancy Expectancy, as conceptualized by Vroom (1964), i s the perceived covariation between e f f o r t and performance.  As we discussed e a r l i e r , to  measure t h i s construct i t i s necessary to assess not only a person's b e l i e f that high e f f o r t w i l l lead to high performance, but also his or her b e l i e f about the extent to which low e f f o r t leads to high performance, high e f f o r t leads to low performance, and so on.  Expectancy can be measured i n keeping  with Vroom's conceptualization by assessing individual's perceptions of the p r o b a b i l i t y that multiple l e v e l s of e f f o r t are associated with multiple levels of performance, and then combining those multiple perceptions i n a way that r e f l e c t s covariation. This approach to measurement of expectancy was pioneered by Ilgen, Nebecker & Pritchard (1981).  However, i t has not been  validated, as we shall see. Ilgen and colleagues point out that while Vroom's conception of expectancy was one of effort-performance covariation, many studies have overlooked this i n favor of single p r o b a b i l i t y estimates of the l i n k between high e f f o r t and high performance.  Ilgen, Nebecker and Pritchard's approach  was to measure the degree of association between a l l l e v e l s of e f f o r t and performance.  They asked respondents to indicate the p r o b a b i l i t y that exerting  a given l e v e l of e f f o r t would result i n a given l e v e l of performance.  Ilgen,  Nebecker and Pritchard also measured the association between e f f o r t and performance by asking for estimates of the frequency of  effort—performance  128  combinations and by using verbal indicators of p r o b a b i l i t y . For  each of these three ways of measuring the l i n k between levels of  e f f o r t and performance two composite scores were constructed.  The f i r s t  was  an "expected value index", r e f l e c t i n g the number of work units an i n d i v i d u a l expected to complete.  Given a set of n x m p r o b a b i l i t i e s p ( i , j ) r e l a t i n g each  of m levels of effort C(j) to each of n l e v e l s of performance R(i) where C(j) and R(i) are the values or weights assigned to each l e v e l of e f f o r t or performance, the expected value index (EVI) i s calculated as follows: EVI = Z R ( i ) Z p ( i , j ) C ( j ) (Z Z p ( i , j ) C ( j ) ) The second composite score was a c o e f f i c i e n t r e f l e c t i n g "the degree of linear covariation" between e f f o r t and performance.  This uses the same matrix  of responses, treating the values as frequencies i n a b i v a r i a t e d i s t r i b u t i o n . Based on these frequencies, covariation between the two variables was calculated i n the same way that a c o r r e l a t i o n c o e f f i c i e n t i s calculated. Using the notation from above, and where E(R) and E(C) are the weighted mean levels of perceived performance R and e f f o r t C, respectively, the perceived covariation measure (COV) i s calculated as follows: COV = Z Z (R(i) - E(R)) (C(j) - E(C)) p ( i , j ) • Z (R(i) - E(R))  2  p ( i ) • Z (C(j) - E ( C ) )  2  p(j)  Consider, for example, the assessment of the perceived covariation between three levels of e f f o r t and four l e v e l s of performance. are  Respondents  asked twelve questions about the p r o b a b i l i t y that a specified l e v e l of  e f f o r t w i l l lead to a specified l e v e l of performance, for each combination of e f f o r t and performance l e v e l s .  These are the p ( i , j ) .  Unit weights are  129  assigned to e f f o r t and performance l e v e l s .  That i s , the lowest e f f o r t and  lowest performance levels are assigned a weight of 1, the next highest levels a weight of 2, and so on. A l t e r n a t i v e l y ,  real values can be used as weights,  as i n the study by Ilgen, Nebecker, and Pritchard. R(i),  These are the C(j) and  respectively. It i s important to note that c a l c u l a t i o n of the product moment  correlation between two variables  from a matrix of p r o b a b i l i t i e s requires that  the c e l l entries are joint p r o b a b i l i t i e s .  That i s , p r o b a b i l i t i e s of the joint  occurrence of the specified levels of the variables, high e f f o r t and high performance co-occur. however, are conditional  probabilities.  such as for example, that  The p r o b a b i l i t i e s p ( i , j ) above,  They represent the p r o b a b i l i t y of a  specified l e v e l of performance occurring given a specified l e v e l of e f f o r t . The perceived covariation  c o e f f i c i e n t i s not,  to a correlation c o e f f i c i e n t . conditional  therefore, s t r i c t l y equivalent  The two c o e f f i c i e n t s are equal when the  p r o b a b i l i t i e s and the joint p r o b a b i l i t i e s are equal, which occurs  when the marginals are equal, that, i s , when each l e v e l of e f f o r t i s perceived to be equally  likely.  By asking individuals  to report conditional  p r o b a b i l i t i e s we are asking  for the joint p r o b a b i l i t i e s that would hold i f the marginal p r o b a b i l i t i e s were equal, i f high, medium, and low e f f o r t were equally probable.  Variation i n  the marginal p r o b a b i l i t i e s results i n r e s t r i c t i o n of the range of the correlation c o e f f i c i e n t .  For example, i f the perceived probability of low  e f f o r t i s zero then the range of possible c o r r e l a t i o n between e f f o r t and performance i s much less than i f the perceived p r o b a b i l i t y of e f f o r t i s much more than zero.  Different  individuals may hold d i f f e r e n t perceptions of the  p r o b a b i l i t y of expending low e f f o r t or high e f f o r t .  Yet these same  individuals may have similar perceptions of the degree to which low or high  130  e f f o r t lead to high performance, or i n general of the r e l a t i o n s h i p between e f f o r t and performance.  By constraining i n d i v i d u a l s to report conditional  p r o b a b i l i t i e s t h i s problem i s avoided.  This approach ensures that the  perceived covariation c o e f f i c i e n t has the same possible range for each respondent. E f f o r t Level 1 2 3 Performance Level  1 2 3 4  40 25 15 5  10 20 25 10  0 10 20 35  For the matrix shown above, t r e a t i n g the c e l l values as p r o b a b i l i t i e s i n a b i v a r i a t e d i s t r i b u t i o n , f i r s t d i v i d i n g each value by the sum of a l l values so that the t o t a l p r o b a b i l i t y i s unity, the strength of r e l a t i o n s h i p between e f f o r t and performance i s .61.  This score r e f l e c t s the degree of linear  covariation between e f f o r t and performance.  That i s , i t r e f l e c t s expectancy.  Inspection of the matrix reveals that low e f f o r t i s most l i k e l y to lead to low performance and not l i k e l y to lead to high performance.  High e f f o r t e f f o r t i s  most l i k e l y to lead to high performance and i s not perceived to be at a l l l i k e l y to lead to low performance. E f f o r t Level 1 2 3 Performance Level  1 2 3 4  25 25 25 10  10 20 25 10  10 15 20 20  For the second matrix shown above, i n which i t i s evident that a low l e v e l of e f f o r t i s equally l i k e l y to r e s u l t i n the three lowest levels of performance and high e f f o r t i s perceived to be equally l i k e l y to lead to the highest and next to highest l e v e l of performance and may even lead to low performance, the strength of r e l a t i o n s h i p between e f f o r t and performance i s  131  .22. In contrast to the covariation index, the expected value index i s not a measure of covariation.  It does not r e f l e c t perception of the extent to which  e f f o r t and performance covary, or the degree to which high e f f o r t leads to high performance and not low performance, or the degree to which low e f f o r t leads to low performance and not high performance.  Consider the case where an  i n d i v i d u a l believes with complete confidence (100% p r o b a b i l i t y ) that he or she w i l l achieve a high level of performance i f he or she exerts a high or medium or low l e v e l of e f f o r t , as follows: 0 0 100  0 0 100  0 0 100  In this case e f f o r t and performance do not covary.  High performance i s  associated with a l l e f f o r t levels, and expectancy should have a value" of zero. In this case the expected value index i s numerically equal to the value assigned to the high l e v e l of performance, ( i f unit weights are used, 3). Consider a second case where a person believes with complete confidence that no matter what l e v e l of effect i s expended that he or she w i l l achieve a medium level of performance, as follows: 0 100 0  0 100 0  0 100 0  Again, subjective effort-performance covariation i s absent, yet i n t h i s case the expected value index i s d i f f e r e n t  from that i n the previous case; i t i s  equal to the value assigned to the medium l e v e l of performance, ( i f unit weights are used, 2).  Consider yet a t h i r d case where subjective-effort  performance covariation i s perfect: an i n d i v i d u a l believes that the p r o b a b i l i t y i s 100% that i f he or she expends a low l e v e l of e f f o r t , a low  132  l e v e l (and only a low level) of performance w i l l result and that i f he or she expends a medium l e v e l of e f f o r t that a medium l e v e l of performance w i l l r e s u l t , and that high e f f o r t w i l l result i n high performance, as follows: 100 0 0  0 100 0  0 0 100  In this t h i r d case the expected value index i s equal to that of the second case, yet they are obviously quite d i f f e r e n t .  The expected value index  constructed by Ilgen, Nebecker and Pritchard i s just that, a measure of expected performance. or  expectancy.  It i s not a measure of e f f o r t - performance covariation  The c o r r e l a t i o n a l measure, i n contrast does capture the  covariation between e f f o r t and performance.  It i s t h i s measure that we should  seek to v a l i d a t e .  Validating expectancy measurement In general, the construct v a l i d i t y of a measure i s supported by demonstrations that the measure f i t s the nomological net surrounding i t .  That  i s , that i t converges with other measures of the same construct, diverges from measures of other constructs, i s responsive to changes i n the presence or absence of the construct and due to v a r i a t i o n s i n persons or settings, and i s i n s e n s i t i v e to changes i n the presence of unrelated constructs.  Ilgen,  Nebeker and Pritchard (1981) sought to demonstrate the v a l i d i t y of a measure of  expectancy by comparing scores i n two settings: one with low d i f f i c u l t y and  another with high d i f f i c u l t y . meet a standard.  They manipulated the amount of work required to  In the f i r s t condition workers were required to complete  f i v e items i n a unit of work as compared to seven items per unit i n the second condition.  Ilgen, Nebecker and Pritchard argued that the low d i f f i c u l t y  condition had objectively high expectancy and that the high d i f f i c u l t y  133  condition had objectively low expectancy. objective expectancy i s inappropriate.  This operationalization of  If expectancy i s effort-performance  covariation then objective expectancy should be higher i n jobs i n which the connection between e f f o r t and performance i s stronger.Tasks i n which performance i s primarily a function of e f f o r t should have high expectancy. Jobs with low expectancy are jobs where performance i s a matter of a b i l i t y , or luck, or how well the person on the assembly l i n e ahead of you does h i s or her job.  Low expectancy i s where performance i s high or low independent of  effort.  Task d i f f i c u l t y can sometimes increase d i f f i c u l t y and sometimes  decrease i t .  A task may be so d i f f i c u l t that performance i s uniformly low.  Or a task may be so easy that performance i s uniformly high. instances, expectancy i s constrained.  In both these  S i m i l a r l y , two tasks may d i f f e r i n  d i f f i c u l t y yet have the same effort-performance r e l a t i o n s h i p . A, i n which 10 units of e f f o r t produce 10 units of performance. e f f o r t to 15 units increases performance to 15 units.  Consider task Increasing  In task B, an easier  task, 10 units of e f f o r t produce 20 units of performance and increasing e f f o r t to 25 units increases performance to 25 units.  In each task a 5 unit increase  i n e f f o r t i s associated with a 5 unit increase i n performance.  The  effort—  performance relationship i s i d e n t i c a l i n the two tasks and yet task A i s more d i f f i c u l t than task B.  The expected performance given 10 or 15 units of  e f f o r t i s lower i n task A. Ilgen, Nebecker and Pritchard's manipulation i s l i k e our hypothetical Tasks A and B.  In their high task d i f f i c u l t y , "low expectancy" condition,  each unit of work required the completion of seven items.  In the low  d i f f i c u l t y , "high expectancy" condition, only f i v e items were required per unit.  The underlying relationship between e f f o r t and performance, however,  was e s s e n t i a l l y unchanged.  134  Ilgen and colleagues found that the d i f f i c u l t y manipulation s i g n i f i c a n t l y affected scores on the expected value index.  Expected performance  was  s i g n i f i c a n t l y higher i n the low d i f f i c u l t y , "high expectancy" condition.  In  contrast, their c o r r e l a t i o n a l index "showed l i t t l e responsiveness to the expectancy manipulation" (p.215).  They interpreted these results as evidence  for the v a l i d i t y of the expected value index as a measure of expectancy, over and above the c o r r e l a t i o n a l index.  As we have discussed here, however, Ilgen,  Nebecker and Pritchard's operationalization d i d not create differences i n objective effort—performance covariation.  They did manipulate d i f f i c u l t y ,  which was r e f l e c t e d i n the perceptions of participants that they could expect to complete more units of work when fewer items were required per u n i t .  The  lack of responsiveness of the c o r r e l a t i o n a l index i s an i n d i c a t i o n that they did not manipulate expectancy, rather than a sign that the measure lacks validity.  In f a c t , this lack of responsiveness i s an i n d i c a t i o n of divergent  validity. In sum,  then, Ilgen, Nebecker and Pritchard have developed a measure of  expectancy that i s true to Vroom's conceptualization.  At present, however,  evidence as to the v a l i d i t y of such a multiple-level c o r r e l a t i o n a l index of effort—performance covariation i s lacking.  Study Two was undertaken to  address t h i s .  Method Subjects and Design Participants i n the study were 221 second year business students. received course credit for p a r t i c i p a t i n g .  Each  Participants completed one of two  tasks, a cognitive reasoning and decision making task or a perceptual motor s k i l l s task.  The perceptual motor task was one i n which a strong objective  135  effort-performance connection existed whereas the cognitive reasoning task had a weak objective e f f o r t - performance connection.  Following completion of the  task measures of the perceived link between e f f o r t and performance were taken.  Manipulation of  Expectancy  Perceptual-motor  task.  The high objective expectancy condition was  created by using a task r e l y i n g on the perceptual and motor s k i l l s of participants.  Specifically,- the task chosen was a "proofreading or q u a l i t y  control" task, b r i e f l y described above i n Study One.  Participants were t o l d  that " i n q u a l i t y control a product must be matched against a standard." Participants were provided with a booklet, each page of which contained 25 rows of d i g i t s between 0 to 9, each row having 20 d i g i t s . chosen from a random number table.  These numbers were  In the l e f t margin of each page was a  column of d i g i t s corresponding to the rows.  The task assigned to the  participants was to check the number at the l e f t margin of each row, c i r c l e each number i n the row that matched i t .  then  Following a practice session,  participants worked at the task for a series of 10 work periods which averaged one minute i n length.  They were asked to do their best but to work c a r e f u l l y  as "only c o r r e c t l y completed rows count."  At the end of each period  participants were asked to draw a l i n e under the l a s t row completed, count the number of rows completed and write the number of rows completed i n the right hand margin. page.  They were then instructed to begin again at the top of the next  The complete instructions and a sample of the task provided are shown  i n Appendix C.  Cognitive reasoning task.  The low objective expectancy condition was  created by using a task r e l y i n g on reasoning a b i l i t y and conceptual understanding.  The task was  the "Brand Managers' A l l o c a t i o n Problem", or  136  "Marketing Game" (Mclntyre 1979, 1982).  This task involved the a l l o c a t i o n of  a fixed promotional budget across three markets, with the objective of maximizing the t o t a l p r o f i t earned.  Each market was represented by an  independent response function that related promotional expenditures to p r o f i t . Participants were t o l d that they were the newly hired brand manager and had to decide how to divide their promotional budget among three markets for each of 10 periods.  Before beginning, they were shown the market a l l o c a t i o n s and  market p r o f i t results of the previous manager for f i v e previous periods, although the previous budget was not f i x e d and promotion was allocated evenly across markets. So, f o r example, a participant may have been assigned a promotional budget of $54,000 f o r each of 10 periods.  Before beginning, they were shown  the p r o f i t that resulted i n each market from equal d i v i s i o n of d i f f e r e n t promotional budgets over each of f i v e previous periods.  They were then asked  to divide the $54,000 among the three markets, were shown the p r o f i t that resulted i n each market, and the procedure was repeated.  The instructions  provided to participants are shown i n Appendix C. Performance on the Proof-reading task was p r i m a r i l y a function of two aspects of e f f o r t : attention to the number provided as standard and c r i t e r i o n , and energy directed at completing as many rows as possible.  Performance was  not a function of a b i l i t y or understanding of how to achieve high but rather of e f f o r t expended.  performance  In contrast, performance on the Marketing Game  depended on p a r t i c i p a n t s ' marginal analysis s k i l l s and a b i l i t i e s ; on examining past a l l o c a t i o n decisions and p r o f i t r e s u l t s , i n f e r r i n g the r e l a t i o n s h i p between promotion and p r o f i t , and applying that understanding to the next decision.  Unlike the Proof-reading task performance on the Marketing Game  depended on i n d i v i d u a l differences i n s k i l l s and a b i l i t i e s .  Performance was  137  thus constrained and dependent on factors over and above e f f o r t .  E f f o r t at  the Marketing Game was present primarily i n the form of concentration and time spent on a decision.  Thus the degree to which performance was a result of  e f f o r t was lower i n the Marketing Game than i n the Proof-reading task.  Measures Three measures of the relationship between e f f o r t and performance were taken: a multiple e f f o r t and performance l e v e l measure of covariation, a single item measure of control over performance, and a single item measure of the perceived relationship between working hard and performing well. Participants' expected performance was also measured.  F i n a l l y , measures of  task perceptions and causal a t t r i b u t i o n s were taken. Perceived covariation.  Participants were asked, for each combination of  f i v e levels of performance and'three levels of e f f o r t , to estimate the p r o b a b i l i t y that they would achieve the specified l e v e l of performance i f they were to expend the specified l e v e l of e f f o r t .  For both the proofreading task  and the Marketing Game the l e v e l of performance chosen divided the range of possible performance into f i v e equal i n t e r v a l s .  For example the performance  i n t e r v a l s for the proofreading task were 1 to 10 rows, 11 to 20 rows, 21 to 30 rows, 31 to 40 rows, and more than 40 rows completed.  For the Marketing Game,  the i n t e r v a l s were $1400 or less p r o f i t , $1400 to $1600 p r o f i t , $1601 to $1800 p r o f i t , $1801 to $2000, and more than $2000 p r o f i t . levels s p e c i f i e d were "high", "medium", and "low".  For both tasks the e f f o r t Figure 7 shows the  assessment of the perceived p r o b a b i l i t y that high e f f o r t would lead to $1801 to $2000 p r o f i t i n the Marketing Game. For each participant, then, 15 perceived p r o b a b i l i t i e s were recorded. These were combined into a composite score r e f l e c t i n g the covariation between  138  The next set of questions are about your p a r t i c i p a t i o n i n the Marketing Game. We'd l i k e you to think about what would happen i f you were to do the Game again, under similar circumstances. We would l i k e you to estimate the p r o b a b i l i t y of achieving d i f f e r e n t amounts of t o t a l p r o f i t . Remember that you earned somewhere between $1000 and about $2000 p r o f i t each period. The following questions w i l l ask you to estimate the p r o b a b i l i t y that you could earn on average per period : $ 1401 or between $ between $ between $ more than  less 1401 and $ 1600 1601 and $ 1800 1801 and $ 2000 $ 2000  The following questions also ask you to think about what would happen i f you were to expend a HIGH, MEDIUM, or LOW degree of e f f o r t . We are not interested i n the actual amount of e f f o r t you expend, but i n what would happen i f you were to do the game again under similar circumstances. If you were to expend a HIGH l e v e l of e f f o r t , what i s the p r o b a b i l i t y that you would earn between $ 1801 and $ 2000 on average per period? (Answer i n percentage terms)  Figure 7.  Assessment of relationship between e f f o r t and performance, Study Two.  139  e f f o r t and performance by treating the values i n the matrix as frequencies i n a bivariate distribution.  Unit weights were used for the three l e v e l of  e f f o r t (e.g., 1, 2, 3) and for the f i v e performance levels (e.g., 1, 2, 3, 4, 5).  The composite score was then transformed using the Fisher r to z  transformation to normalize the d i s t r i b u t i o n of c o e f f i c i e n t s . transformation the d i s t r i b u t i o n was  s i g n i f i c a n t l y skewed, skew = -.684, p <  .01, following transformation skew was Control over performance.  Prior to  .092, p >  .05.  A second measure was constructed using the  concept of "control", as described by A l l o y and Abramson (1979).  A l l o y and  Abramson argue that the r e l a t i o n between a response and an outcome, or the dependence of an outcome on a response, controllability.  i s best construed as one of  Most importantly, control conveys the technical meaning of  covariation i n natural, everyday language (Jenkins and Ward, 1965). Participants were provided with the following i n s t r u c t i o n s : On the following scale, i n d i c a t e your judgement of the amount of control you had over your performance on the [Proofreading task/ Marketing Game], at 100 i f you had complete control and at 0 i f you had no c o n t r o l . Complete control means that the [ number of rows you complete/ p r o f i t you achieve] i s determined by how hard you t r y . No control means that how hard you t r y or don't t r y has nothing to do with your performance. Another way to look at having no control i s that the [number of rows completed/profit achieved ] i n any period i s t o t a l l y determined by factors such as chance or luck, rather than the e f f o r t you expended. Intermediate control means that your e f f o r t has some influence but does not completely determine the p r o f i t you achieve.  Underneath these instructions was displayed a scale with endpoints of 0 and 100 with increments  of 10 also indicated.  140  Perceived c o r r e l a t i o n . studies of expectancy  A t h i r d measure mirrored those used most often i n theory.  It asked participants to indicate, on a 10  point-scale with end-points of 0 and 9 l a b e l l e d "no r e l a t i o n s h i p " and "strong relationship", the number that best represented "the r e l a t i o n s h i p between working hard on the task and performing w e l l " .  Expected performance.  The f i f t e e n p r o b a b i l i t y estimates c o l l e c t e d to  measure perceived covariation, above, were used to calculate the expected value index developed by Ilgen, Nebecker and Pritchard, as described above. Task Perceptions and Causal A t t r i b u t i o n s .  Measures were also completed  of four task perceptions and two causal a t t r i b u t i o n s .  The task perceptions  measured were (1) s a t i s f a c t i o n with performance, (2) task d i f f i c u l t y and challenge, (3) task interest, and (4) task e f f o r t .  The Causal Dimension Scale  (Russell, 1982) was used to measure two dimensions of causal a t t r i b u t i o n s : i n t e r n a l i t y and s t a b i l i t y ,  he measures of task perceptions and causal  attributions were those described i n Study One.  As shown i n Study One,  these  measures have good r e l i a b i l i t y .  Results In this Study, we w i l l consider the following pattern of r e s u l t s to be i n support of the v a l i d i t y of the perceived covariation measure of expectancy: It i s hypothesized that scores on the measures of perceived covariation, control over performance and perceived c o r r e l a t i o n w i l l be s i g n i f i c a n t l y higher i n the high expectancy condition. The scores on the remaining measures do not bear as d i r e c t l y on the v a l i d i t y of the measure, although we would predict that cause would be attributed more to e f f o r t i n the high expectancy condition. That i s , a t t r i b u t i o n s of greater i n t e r n a l i t y and less s t a b i l i t y should be  141  made. Before proceeding with the multivariate analyses, the variables used w i l l be evaluated with respect to p r a c t i c a l l i m i t a t i o n s of the technique.  Evaluation of Assumptions The analysis procedure appropriate to the case of one independent variable with two levels and multiple dependent variables i s Hotelling's T . 2  Hotelling's T  2  i s a special case of multivariate analysis of variance  (MANOVA), which i s designed to test hypotheses about group differences on a set of measures.  The following discussion w i l l be described i n terms of  MANOVA. In this study, the data set comprised 188 subjects i n the  high-expectancy  task condition and 33 subjects i n the low-expectancy task condition .  The  d i f f e r e n t i a l ease of administering the paper-and-pencil motor s k i l l s task i n group settings, as opposed to the i n d i v i d u a l l y , computer-administered cognitive-reasoning task, accounted for t h i s large d i f f e r e n c e i n group size. Twelve of the subjects i n the high expectancy group d i d not provide p r o b a b i l i t y estimates for a l l 15 effort—performance  l e v e l combinations.  The  perceived covariation and expected value measures'could therefore not be computed, so data for these p a r t i c i p a n t s were discarded.  Thus scores for 176  subjects were a v a i l a b l e for analysis i n the high expectancy group. Inspection of the d i s t r i b u t i o n s of scores on the ten dependent variables revealed s i g n i f i c a n t skew on only one measure: that of perceived c o n t r o l . Before taking action to resolve t h i s , the presence of o u t l i e r s investigated. scores. or 0.4%  was  Univariate o u t l i e r s were i d e n t i f i e d by examining extreme z-  Of the 2090 observations (10 dependent variables time's 209 cases), 8 had z-scores i n excess of 3.09.  This was  s l i g h t l y more than chance  142  expectation.  The most extreme score d i d not, however, exceed 3.75  deviations from i t s mean.  Nine cases were i d e n t i f i e d as  standard  multivariate  o u t l i e r s , f i v e i n the high-expectancy group and four i n the low-expectancy group. skewed.  Following deletion of these cases, no variable was s i g n i f i c a n t l y The assumption of multivariate normality was  thus  met.  As a preliminary check for homogeneity of variance-covariance matrices, sample variances for each of the 10 dependent variables were examined. ' For  no  DV d i d the r a t i o of largest to smallest variance across the two groups approach 20:1,  the c r i t e r i o n suggested by Tabachnik and F i d e l l .  homogeneity of variance a=  .01  t e s t s , however, revealed three s i g n i f i c a n t results at  (Bartlett-Box F), for the measures of perceived  and c o r r e l a t i o n .  Univariate  covariation, control,  A multivariate test of homogeneity of dispersion was,  surprisingly, also s i g n i f i c a n t (Box's M=  115.4, p < .01).  For the  not  perceived  covariation measure the larger variance belonged to the larger group, thus the significance test would have been more conservative.  For the other  two  variables, however, the larger variance belonged to the smaller group.  The  d i r e c t i o n of bias produced i n the test of multivariate significance by t h i s heterogeneity i s therefore unknown. To resolve this uncertainty,  equality of sample sizes was  achieved by  randomly selecting 29 cases for analysis from the high-expectancy group, to equal the number of cases i n the low expectancy group.  Significance tests i n  MANOVA are robust to both heterogeneity of variance and non-normality i f sample sizes are equal and exceed 20 degrees of freedom for error i n the univariate case (Hakstian,  Roed and Lind, 1979;  Tabachnik and F i d e l l ,  1983)  Evidence of non-linearity of relationships among dependent variables sought by examining the plots of observed values versus standardized for each dependent v a r i a b l e .  No evidence of gross c u r v i l i n e a r i t y was  was  residuals found,  143  therefore no transformation of variables was undertaken. In t h i s study, m u l t i c o l l i n e a r i t y was assessed by examining the squared multiple c o r r e l a t i o n (SMC) variables.  between each variable and a l l other dependent  No variables showed a SMC approaching 1.0.  S i m i l a r l y , the  determinant of the w i t h i n - c e l l c o r r e l a t i o n matrix was non-zero. F i n a l l y , an assumption of homogeneity of regression, or that the slope of the regression of dependent variables on covariables made i n stepdown analysis i n MANOVA.  i s equal across c e l l s , i s  In t h i s study, tests of homogeneity of  regression were run for a l l variables, since each serves as a covariate for a l l others i n the Stepdown F-tests that follow.  Homogeneity of regression was  achieved for a l l components.  Multivariate analysis of variance. A multivariate analysis of variance was performed on the ten dependent variables: perceived  covariation, perceived  control, perceived c o r r e l a t i o n ,  expected performance, s a t i s f a c t i o n with task performance, task d i f f i c u l t y and challenge,  task e f f o r t , task i n t e r e s t , and perceived  of the cause of performance.  i n t e r n a l i t y and s t a b i l i t y  The independent variable was objective task  expectancy (high and low). The MANOVA procedure i n SPSSX was used for the analyses.  Because samples  sizes were equal (n=29), no adjustment for nonorthogonality was necessary. Total N was 58 following the deletion of o u t l i e r s and the sampling procedure described  above.  Hotelling's T  2  c r i t e r i o n showed a highly s i g n i f i c a n t difference between  the two groups, F(10,  47) = 14.21, p < .001.  These r e s u l t s r e f l e c t e d a strong  association between objective task expectancy and the combined dependent variables (A = .25; r) = 0.75). 2  144  The influence of the difference i n objective effort-performance r e l a t i o n s h i p between the two groups on each dependent v a r i a b l e was examined next.  The univariate test of significance for the eleven dependent variables  are shown i n Table 15.  S i g n i f i c a n t univariate e f f e c t s are present for  perceived covariation, perceived control, perceived c o r r e l a t i o n ,  expected  performance, task e f f o r t , task i n t e r e s t , and i n t e r n a l i t y of a t t r i b u t i o n s . Table 16 shows, the average perceived covariation was  .41 i n the proofreading,  high expectancy group, compared to .28 i n the marketing game, low group.  expectancy  By i t s e l f , t h i s result confirms the v a l i d i t y of the composite measure  of expectancy.  In keeping with this r e s u l t , mean perceived control was also  s i g n i f i c a n t l y higher i n the high expectancy group than i n the low group.  As  expectancy  Mean expected task performance was, however, s i g n i f i c a n t l y lower i n  the high expectancy group.  This means that the hypothesis formulated by  Ilgen, Nebecker and Pritchard, that the expected value index measures expectancy,  i s disconfirmed.  Instead of showing that scores on the expected  value index were higher i n the high expectancy group, which would be evidence of v a l i d i t y , or that scores were unaffected by objective expectancy,  which i s  the n u l l hypothesis, Study Three supports r e j e c t i o n of the n u l l hypothesis i n the opposite d i r e c t i o n to that supposed. The univariate results also showed that participants rated the cause of their performance as s i g n i f i c a n t l y more internal i n the high  expectancy  condition than i n the low expectancy condition. Overall, these results provide strong evidence for the v a l i d i t y of the expectancy manipulation and the perceived covariation measure. high expectancy,  F i n a l l y , task e f f o r t was higher i n the  Proof-reading task, but participants found the Marketing Game  more i n t e r e s t i n g . Although the univariate s t a t i s t i c s are of i n t e r e s t , they overlook  145  Urlivariat :e  SU jpdown  F  d.f.  P  F  7 .15 8 .19 5 .56  1/56 1/56 1/56  .016 .006 .022  7 .15 7 .23 .32  1/55 1/55 1/54  .127 .131 .571  .071 .071 .006  70 47  1/56  .000  54 .32  1/53  .000  .506  Perceived internality stability  4 .83 .34  1/56 1/56  .032 .558  3 .93 2 .93  1/52 1/51  .053 .092  .070 .054  S a t i s f a c t i o n with task performance  0 59  1/56  .444  .18  1/50  .674  .003  Task d i f f i c u l t y and challenge  3 .61  1/56  .062  2 .54  1/49  .117  .049  Task e f f o r t  21 10  1/56  .000  8 .33  1/48  .006  .148  Task interest  36 19  1/56  .000  4 .60  1/47  .037  .089  d.f.  P  Measure Perceived covariation control correlation Expected performance  T  Table 15.  Univariate and Stepdown F-tests,  Study Three.  146  Proofreading Task (High Expectancy) Measure Perceived covariation  Mean  St. Dev.  Marketing Game (Low Expectancy) Mean  St. Dev.  0.41  0.23  0.28  0.16  81.72  16.49  70.93  11.85  Perceived c o r r e l a t i o n  7.17  1.60  6.10  1.84  Expected performance  2.12  0.38  2.98  0.39  20.89  3.07  18.83  4.02  14.72  5.76  13.86  5.36  S a t i s f a c t i o n with task performance  14.10  2.67  13.58  2.43  Task d i f f i c u l t y and challenge  10.00  2.55  11.20  2.27  Task e f f o r t  8.48  1.18  6.82  1.54  Task interest  5.13  1.72  7.62  1.40  Perceived control  Perceived internality stability  Table 16. Means and standard deviations f o r dependent variables, Study Three.  147  relationships between the dependent variables.  The matrix of correlations  between dependent variables, shown i n Table 17, reveals i n t e r c o r r e l a t i o n s among many of the measures.  B a r t l e t t ' s test of s p h e r i c i t y was s i g n i f i c a n t  (x (45) = 140.4, p < .001) supporting r e j e c t i o n of the hypothesis that the 2  c o r r e l a t i o n matrix i s an i d e n t i t y matrix and hence that the variables are independent.  To evaluate the contribution of each dependent variable with the  effects of i t s c o r r e l a t i o n with other variables controlled, a stepdown analysis was performed.  In t h i s study, the measures r e l a t i n g e f f o r t and  performance were of most i n t e r e s t and were entered into the analysis f i r s t , followed by the measures of expected performance, causal a t t r i b u t i o n s , and task perceptions,  as shown i n Table 15. The f i r s t variable entered was the  composite measure of perceived  effort-performance covariation.  Scores on t h i s  measure were s i g n i f i c a n t l y associated with objective task expectancy, stepdown F ( l , 56)= 7.15, p < .02. As indicated by the r a t i o of hypothesis sum of squares to error plus hypothesis sum of squares, the strength of association was  .13. After the pattern of differences i n perceived  entered, a s i g n i f i c a n t difference remains for perceived 55) = 7.23, p < .01, n  2  covariation was control, stepdown F ( l ,  = .13. Given i t s high c o r r e l a t i o n with  perceived  control, i t i s not surprising that the unique contribution of the perceived c o r r e l a t i o n measure i s non-significant, stepdown F ( l , 54) = .325, p > .05. The fourth variable entered, expected performance, made a strong unique contribution to the multivariate difference between expectancy conditions, stepdown F (1,53) = 54.33, p < .001, rj = .51. Stepdown analysis 2  that i n t e r n a l i t y of a t t r i b u t i o n s made a marginally  revealed  s i g n i f i c a n t unique  contribution, stepdown F = 3.93, p < .053, 77 = .07. As stated above, the 2  mean i n t e r n a l i t y score was higher i n the high expectancy condition.  Of the  remaining measures, only task e f f o r t and task interest made s i g n i f i c a n t unique  (1) (1) (2) (3) (4) (5) (6) (7) (8) (9) ( 10)  Table  17.  Perceived covariation Perceived control Perceived correlation Expected performance S a t i s f a c t i o n with task performance Task d i f f i c u l t y and c h a l 1 e n g e Task e f f o r t Task i n t e r e s t Perceived i nternali ty s t a b i 1 i ty  Correlations  . 1 14 .249 -.243  (2)  .523 -.252  (4)  (3)  (5)  (6)  (7)  - .277  . 332  . 126  - .056  - . 106 .340 - . 133  - .088 .207 - . 145  - . 104 .231 .009  .097 - .412 .525  - .253 .383 - .023  . 297 .461  .096 .027  .640 .457  .291 .373  - .055 - .044  . 399 .533  -. 194 - . 189  variables.  (9)  - . 170  - .042  between dependent  (8)  Study  Three.  .215 .200  -.071 .093  . 584  ( 10)  149  contributions to the o v e r a l l e f f e c t .  Task e f f o r t was  the high expectancy condition, stepdown F = 8.34, interest was = 4.60,  s i g n i f i c a n t l y higher i n  p < .01, 77 = .15. 2  Task  s i g n i f i c a n t l y higher i n the low expectancy condition, stepdown F  p < .05, 7? = 2  .09.  Discussion Study Three provided strong evidence for the v a l i d i t y of the c o r r e l a t i o n a l index of perceived covariation as a measure of expectancy.  The  immediate implication of this i s that t h i s measure i s not only appropriate to further research but that i t should be used i n place of other measures.  The  perceived covariation measure i s closest to the conceptualization of expectancy as intended by Vroom and i s responsive to a manipulation that matches that conceptualization of expectancy. . The f i r s t task of future research should be to r e p l i c a t e Study Three using similar tasks and measures and, more importantly, using other tasks.  A key demonstration would be to  show that the perceived covariation measure i s responsive to differences i n objective expectancy within a single task.  Such a difference might be created  by manipulating task interdependency or the r o l e of chance.. It i s not clear, however, to what extent expectancy can be manipulated without changing the essential nature of tasks. the objective expectancy of tasks malleable? i n expectancy only exist across tasks. jobs.  That i s , to what extent i s  It may be that gross differences  This question i s key to the design of  Perceived effort-performance covariation i s an important factor i n  determining the e f f o r t that individuals choose to expend at work, and objective expectancy i s probably the p r i n c i p a l determinant of percepts.  expectancy  It follows then, that to design jobs to motivate people we should  know more about manipulating objective expectancy.  Better understanding of  150  the concept and how  to measure i t i s only one step i n that d i r e c t i o n .  Study Three showed that self-reported task e f f o r t was higher i n the high expectancy  task.  significantly  This cannot be taken as evidence of the  v a l i d i t y of the expectancy manipulation, the expectancy measure, or theory because instrumentality and valence are unknown.  expectancy  The obvious question  i s whether e f f o r t and expectancy are s t i l l p o s i t i v e l y associated when the other components of expectancy theory are held constant.  That i s , does the  perceived covariation measure r e l i a b l y predict e f f o r t ?  An affirmative answer  would lend support to both the measure and the theory.  A negative answer  would cast doubt on both the theory and the measure. expectancy  One could argue that  theory as conceptualized by Vroom has as yet not been tested  because appropriate measures have not been used. expectancy theory i s not i n .  Clearly, the f i n a l word on  151  VIII.  STUDY FOUR: MOOD AND EXPECTANCY  Overview Study Four tested the p r i n c i p a l hypotheses about the influence of mood on expectancy. participants.  A laboratory experiment  was conducted with students as  In each experimental session students p a r t i c i p a t e d i n a  simulated business task, underwent a mood induction procedure, and then completed measures of mood state, expectancy, and task cognitions. The experimental design chosen was a within-subjects design i n which each participant received a l l levels of the independent v a r i a b l e .  Further, the  order of presentation of the three l e v e l s of the mood manipulation was counterbalanced by creating six between-subjects complete design was a mixed between-within,  order conditions. Thus, the  or s p l i t - p i o t design (Kirk, 1968).  The design has p a r t i c u l a r advantages and disadvantages and poses some unique analysis issues.  Experimental Design Analysis of variance (ANOVA) procedures employed to test differences among mean scores can conveniently be summarized i n terms of how the sums of squared differences between the scores of individuals and their group means" are p a r t i t i o n e d .  For example, i n an experimental design where each subject i s  assigned to one of a number of treatment groups (a one-way between-subjects design), the t o t a l sum-of-squares can be partitioned into those associated with differences between groups and those associated with differences within groups, e.g. the sum-of-squares t o t a l equals sum-of-squares groups plus sumof -squares subjects-within groups.  The F test of the significance of the  152  difference between groups i s a function of the r a t i o of the between-group to subjects-within-group  sum-of-squares.  The within-group  sum-of-squares i s used  to estimate the population (error) variance. In within-subjects ANOVA, each subject receives more than one treatment. This i s also known as a repeated measures design.  The source of differences  among means for levels of the independent variable i s the same as the betweensubjects design, above. e.g.  The error term can, however, be p a r t i t i o n e d further,  sum-of-squares t o t a l equals sum-of-squares groups plus sum-of-squares  subjects plus sum-of-squares subjects-by-group.  Differences between  individuals( sum-of-squares subjects) can be viewed as a systematic source of variance i n scores.  If i n d i v i d u a l s are measured repeatedly, these i n d i v i d u a l  differences can be examined.  Since this v a r i a t i o n i s systematic, i t can be  subtracted from the error term.  Or, stated d i f f e r e n t l y , the best estimate of  population (error) variance i s the group by subject i n t e r a c t i o n .  This error  term i s smaller than that i n a between-subjects design to the degree that the v a r i a t i o n due to subjects i s d i f f e r e n t from zero.  That i s , i f the sum-of-  squares subjects i s non-zero then the error term i n a within-subjects analysis i s smaller than i n a between-subjects a n a l y s i s . A smaller between-group effect i s therefore needed to achieve a s i g n i f i c a n t F - r a t i o , provided, of course, that the loss of degrees of freedom i n the within-subjects case does not o f f s e t any gain. Within-subjects designs have disadvantages that may mitigate any gains i n s t a t i s t i c a l power.  F i r s t , there i s the p o t e n t i a l for effects due to the fact  of repeated measurement, such as learning effects or e f f e c t s of hypothesis guessing.  Second, the degrees of freedom for error are reduced from those i n  a between-subjects design. freedom may  This loss i n power from diminished degrees of  overcome gains i n power from reduced error variance.  The  relative  153  e f f i c i e n c y of the within-subjects design depends on the degree to which v a r i a t i o n between subjects within groups has been reduced r e l a t i v e to the v a r i a t i o n between groups.  We can conceptualize each subject i n a within-  subjects design as serving as his or her own control.  The r e l a t i v e power of a  within-subjects design i s a function of the degree to which such control f o r i n d i v i d u a l differences i s relevant. In t h i s study both the dependent and independent variables are perceptions held by i n d i v i d u a l s : of q u a l i t i e s of tasks and of their own f e e l i n g state.  These are l i k e l y to be highly i n d i v i d u a l .  That i s , people are  l i k e l y to vary greatly i n the perceptions they have of tasks and of t h e i r own mood.  Across individuals t h i s v a r i a t i o n may be much larger than differences  over occasions within i n d i v i d u a l s .  Controlling for these differences across  levels of the independent variable i s therefore highly advantageous. within subjects design these between-subjects  In a  effects are estimated and  removed from the error term, increasing the power to detect a betweentreatment group difference.  A further reason for using a within-subjects design i s the p o s s i b i l i t y of additional error variance due to d i f f e r e n t i a l success or f a i l u r e experiences between subjects. mood.  Task success and f a i l u r e are r e l i a b l e manipulations of  Some participants w i l l experience the experimental task as more  successful than w i l l others.  Such an inadvertant mood manipulation would add  to error variance i n a between-subjects  design.  however, t h i s effect i s held constant.  The d i f f e r e n t tasks used on repeated  sessions are of equal d i f f i c u l t y .  In a within-subjects design,  Thus each participant i s l i k e l y to have the  same success or f a i l u r e experience i n each of the three mood treatment sessions.  In a between-subjects  design the experience of d i f f e r e n t  154  individuals i s l i k e l y to be d i f f e r e n t even i f task d i f f i c u l t y i s held constant. Because each participant i n a within-subjects design receives each l e v e l of the independent variable, the p o s s i b i l i t y of session e f f e c t s i s created. That i s , scores may vary systematically from the f i r s t to subsequent experimental sessions.  Further, the order of presentation of levels of the  independent variable may influence scores over sessions.  To control for these  concerns i n this study, each participant was randomly assigned to one of the six possible orders of presentation of the three mood treatments.  The  complete design i s thus a mixed-between-within, or s p l i t - p l o t design.  This i s  diagrammed i n Figure 8.  Method Subjects. students.  Participants i n the study were 94 UBC second-year business  Each received course credit i n return for h i s or her p a r t i c i p a t i o n .  Sixteen participants were excluded from the analysis for a number of reasons. Data from six participants was not complete due to their f a i l u r e to attend sessions or because of computer data-capturing malfunctions.  The data for s i x  additional participants were discarded because of procedural errors i n which they received the same mood treatment during two sessions. thus c o l l e c t e d for 82 participants over the three sessions.  Complete data was Four of these  cases were discarded such that the f i n a l design contained 13 cases i n each of the 6 order conditions.  Procedure.  That i s , the f i n a l sample comprised  78 p a r t i c i p a n t s .  Invitations to p a r t i c i p a t e i n the study were made to students  during regularly scheduled course time.  This i n v i t a t i o n described the study  as an investigation of music i n the workplace and informed participants as to the benefits of p a r t i c i p a t i o n , the commitment required, and what procedures  155  Session One  Two  Three  1  Elation  Depression  Neutral  1  Elation  Neutral  Depression  1  Depression  Elation  Neutral  1  Depression  Neutral  Elation  1  Neutral  Elation  Depression  1  Neutral  Depression  Elation  Subjects nOl Order 1  nl3  nl4 Order 2  n26 n27 Order 3  n39 Order 4  n40 n52  n53 ' Order 5  n65 Order 6  n66  n78  Figure 8.  S p l i t - P l o t Design, Mood Treatment by Session and Order Conditions.  156  would be employed.  I t a l s o ensured t h a t c o n f i d e n t i a l i t y o f responses would be  m a i n t a i n e d and t h a t p a r t i c i p a t i o n was v o l u n t a r y .  P a r t i c i p a n t s were asked t o  s i g n a form i n d i c a t i n g t h a t they had gave t h e i r informed consent t o participate. At  The form used i s shown i n Appendix  D.  t h e time t h e i n v i t a t i o n t o p a r t i c i p a t e was i s s u e d , A l l p r o c e d u r e s  employed i n Study Four were judged e t h i c a l by i n s t i t u t i o n a l review,  the  q u e s t i o n n a i r e c o n t a i n i n g measures o f i n d i v i d u a l d i f f e r e n c e v a r i a b l e s was d i s t r i b u t e d f o r c o l l e c t i o n t h e next c l a s s .  Arrangements were made t o c o n t a c t  each p e r s o n a t a l a t e r time t o s c h e d u l e h i s o r h e r p a r t i c i p a t i o n i n t h e experimental sessions.  P a r t i c i p a n t s were t e s t e d i n d i v i d u a l l y i n each of t h e  three experimental sessions.  These  s e s s i o n s were scheduled on d i f f e r e n t  so t h a t mood i n d u c e d i n one s e s s i o n would n o t c a r r y over t o t h e n e x t . s c h e d u l i n g c o n s t r a i n t s imposed  days  Other  were t h a t a l l t h r e e s e s s i o n s occur w i t h i n  seven  days, t o reduce a t t r i t i o n , " a n d t h a t a l l t h r e e s e s s i o n s f o r each p a r t i c i p a n t take p l a c e b e f o r e noon or a f t e r noon.  T h i s l a t t e r c o n s t r a i n t was imposed t o  reduce v a r i a b i l i t y due t o t i m e - o f - d a y e f f e c t s .  A l t h o u g h these l a t t e r two  c o n s t r a i n t s were adhered t o as much as p o s s i b l e , t h e y were r e l a x e d where n e c e s s a r y t o ensure a s u b j e c t ' s p a r t i c i p a t i o n . The p r o c e d u r e f o r each s e s s i o n was l a r g e l y i d e n t i c a l .  Each o f t h e  p a r t i c i p a n t s completed a b u s i n e s s d e c i s i o n - m a k i n g t a s k , then underwent a mood i n d u c t i o n , and f i n a l l y completed measures o f t h e m a n i p u l a t i o n checks and dependent  variables.  The major e x c e p t i o n t o t h i s procedure was t h a t t h e f i r s t  session included a p r a c t i c e period i n the decision-making task. S e s s i o n One.  On a r r i v a l a t t h e f i r s t  were g r e e t e d by one o f t h r e e e x p e r i m e n t e r s .  experimental session,  participants  The t h r e e experimenters were  male, between 20 and 30 y e a r s of age and were r e h e a r s e d t o ensure t h a t t h e i n t e r a c t i o n o f each s u b j e c t w i t h each experimenter was  comparable.  157  Participants were then seated i n front of a computer terminal and with a description of the procedure for Session One,  presented  as follows:  Music i n the Workplace SESSION ONE The study i n which you are being asked to p a r t i c i p a t e i s part of an investigation of the use of music i n the workplace. Many organizations play music as a background to work. The study you are i n i s examining the use of music during work and people's perceptions of the use of music. We would l i k e your help with two parts of this study. In the f i r s t part you w i l l be asked to complete a simulated business decision-making task on a computer terminal while music i s played i n the background. In the second part of today's session we would l i k e you to l i s t e n to some music and then we w i l l ask you some questions about the music. Both parts of the session w i l l provide us with baseline information about the decision-making task and about people's reactions to the music. Because we are interested i n the use of d i f f e r e n t types of music we would l i k e you to come back twice more to l i s t e n to other selections. At that time we would again l i k e to c o l l e c t baseline information on another version of the decision-making task. Today's session w i l l include a p r a c t i s e period and so should take about one hour, the following two sessions should take less than an hour each. The decision-making task i s based on a r e a l i s t i c business problem. The music you w i l l l i s t e n to w i l l probably evoke d i f f e r e n t reactions i n d i f f e r e n t people. These honest reactions are what we are interested i n . Your p a r t i c i p a t i o n i n this study i s voluntary, you are free to discontinue p a r t i c i p a t i o n at any time without penalty. Your responses w i l l be used only for the purposes of this study and w i l l be kept c o n f i d e n t i a l .  Decision-making  task.  To provide a basis for measurement of  participant's expectancies they were asked to complete an experimental  task.  Expectancy research has used experimental tasks such as processing catalogue orders (e.g., Ilgen, Nebecker & Pritchard, 1981), and has asked individuals to rate the expectancy associated with educational or occupational choices (e.g., Snyder, Howard & Hammer, 1978; Muchinsky, 1977).  Studies of the influence of  mood on memory have usually used much more a r t i f i c i a l tasks, such as mental rotation, concept formation, or contingency  judgement tasks (Alloy, Abramson &  158  V i s c u s i , 1981; Brown, 1984; Wright & Mischel, 1982).  As a study cf  organizational motivation, i t was desirable to use a more r e a l i s t i c task with g e n e r a l i z a b i l i t y to organizational settings i n t h i s study.  Further, based on  findings that mood effects were most l i k e l y when the objective contingency between behavior and outcomes was low (Alloy & Abramson, 1979; Alloy, Abramson & V i s c u s i , 1981), a task with low expectancy was preferable. A task which met these requirements was the "Brand Manager's A l l o c a t i o n Problem" (Mclntyre, 1979, 1982), the same task used i n the low expectancy condition of Study Two.  This task involves the a l l o c a t i o n of a fixed  promotional budget across three t e r r i t o r i e s with the objective of maximizing the resultant t o t a l p r o f i t .  Each of the three t e r r i t o r i e s i s represented by  an independent response function that relates promotional expenditures to p r o f i t , of the following form:  P r o f i t = t min + ts * (aX ** b / ( l + aX ** b)) - e - X  where, X = promotional d o l l a r s allocated to the t e r r i t o r y , t min = sales i n the t e r r i t o r y when X = 0, ts = saturation sales minus t min i n the t e r r i t o r y , a, b = parameters, e = random error term (uniformly d i s t r i b u t e d between -20 and 20%). The decision-making task was to a l l o c a t e the fixed promotional budget across the three t e r r i t o r i e s to maximize t o t a l p r o f i t .  Participants were t o l d  that they would be evaluated based on the t o t a l p r o f i t earned over .the ten decision periods.  Each period was independent, that i s , there was  no  promotional carryover from one period to another, of which participants were  159  informed.  After each a l l o c a t i o n decision the subjects would automatically  receive an updated "History Display".  This report showed the promotional  a l l o c a t i o n s made to each t e r r i t o r y , p r o f i t by t e r r i t o r y , and t o t a l p r o f i t f o r each decision.  Before making the f i r s t decision, subjects received a f i v e -  period history report.  These f i v e decisions were presumably made by the  previous manager whom the subject was replacing.  In the i n i t i a l five-period  h i s t o r y report the budget varied by plus or minus 10% from period to period but was always allocated equally between the t e r r i t o r i e s . Following the p r a c t i s e session, any further questions were answered. Participants were then t o l d that they would now begin the actual decisionmaking simulation.  It was emphasized that the task was i d e n t i c a l but that the  p a r t i c u l a r problem was a d i f f e r e n t one: participants were t o l d that they had been given three new markets and a d i f f e r e n t promotional budget.  The actual  simulation was 10 periods long.  with  Participants were again provided  a l l o c a t i o n s and p r o f i t r e s u l t s for f i v e "previous" periods made by the "previous manager". To enhance p a r t i c i p a n t s ' b e l i e f s that the study was about "music i n the workplace" so that the subsequent mood induction would be p l a u s i b l e but disguised, music was also played i n the background during the task. Participants were t o l d they were to "concentrate part of their normal job".  on the task as i f i t were  A selection of n o n - l y r i c a l , neutral music  (Ackerman, 1981; "Passage: Pieces for guitar") was played at low volume. The r e s u l t s of a pretest indicated that this music was perceived as neutral and had no s i g n i f i c a n t e f f e c t on the mood of p a r t i c i p a n t s . The task was presented i n t e r a c t i v e l y on a computer terminal video monitor.  Participants were t o l d that the computer would record their p r o f i t  earned for each a l l o c a t i o n made.  The p r o f i t r e s u l t s shown a f t e r each  160  a l l o c a t i o n decision were the actual p r o f i t earned i n those markets, given their response functions and the degree of randomness ("external market factors") present.  The instructions provided to participants were as follows: THE MARKETING GAME: Session One Practice  You w i l l play the r o l e of a newly hired brand manager f o r National Foods. Your job i s to decide how to spend money on promoting your company's product i n your region. You w i l l be given a promotional budget and have to decide how best to divide that budget between the three markets i n your region. The more of the promotional budget you spend i n a market the more p r o f i t you w i l l earn i n that market. But i n some markets the same amount of promotion earns more p r o f i t . Your goal i s to earn as much p r o f i t i n your region as you can over the next ten periods. So you must decide f o r each of these periods, how to allocate or d i v i d e your budget between the three markets to earn as much p r o f i t as possible. The way you w i l l do t h i s i s by making your a l l o c a t i o n f o r one period, then you can look at the r e s u l t s of that a l l o c a t i o n before you go on to make the a l l o c a t i o n for the next period. To help you get started i n your new job, you w i l l be shown the "history report" of the previous manager. This report w i l l look l i k e t h i s : H I S T O R Y D I S P L A Y ALL F I G U R E S IN PERIOD:  1  2  3  4  $ 000'S  5  PROMOTION MARKET 1: MARKET 2: MARKET 3:  7 7 7  10 10 10  8 8 8  11 11 11  9 9 9  PROFIT MARKET 1: MARKET 2: MARKET 3:  221 97 443  315 103 634  244 99 494  332 102 686  279 105 563  837 1120  947  TOTAL PROFIT 761 1052 HIT RETURN TO CONTINUE This shows that i n period One the previous manager divided a budget of $21 ( a l l figures are i n thousands) evenly between Market 1, Market 2, and Market 3. This resulted i n a p r o f i t of $221 i n Market 1, $97 i n Market 2, and $443 i n Market 3 for a t o t a l of $761. In period 2 the previous manager divided a budget of $30 evenly among the three markets.  161  The previous manager had a d i f f e r e n t budget i n each period. YOUR budget w i l l be EQUAL for each period. During t h i s practice session you w i l l have $54 to a l l o c a t e each period to the three markets i n your region. You must spend a l l of this budget each period. The p r a c t i c e session w i l l be THREE periods long. Your performance w i l l be evaluated on the basis of the t o t a l p r o f i t you earn over the next three periods. During t h i s practice session and the r e a l session l a t e r , the computer w i l l prompt you each period. It w i l l show you the h i s t o r y report, including the decisions and r e s u l t s of the previous manager, then i t w i l l ask you to make a decision. The prompt w i l l look l i k e t h i s : YOUR BUDGET IS: $ 54 WHAT IS YOUR ALLOCATION TO MARKET 1, MARKET 2, MARKET 3 FOR  PERIOD:  Enter your decision by entering three numbers on the same l i n e , by spaces. For example: ?20 14  6  separated  20  The computer w i l l then show you the updated h i s t o r y report with the p r o f i t r e s u l t s of your decision and go on to the next period. A recent market survey, commissioned by National Foods, has determined that although other market factors also influence p r o f i t , the single most important decision that you can make i s how to divide your promotional budget. This survey has also determined that although other factors may influence p r o f i t within a period, the periods are independent. There i s no carryover of promotion from one period to the next. Any  questions?  Following an opportunity to have any questions about the procedure or the Marketing Game answered, each participant was choice questions was  to ensure understanding of the decision-making task.  taken that these questions  evaluation.  asked a series of multiple Care  were posed for c l a s s i f i c a t i o n rather than  Any misunderstandings were corrected by the experimenter.  When  the participant was prepared to proceed, the experimenter i n i t i a t e d the practice version of the decision-making task, ensured that the program  was  162  running properly and l e f t the room to allow the participant to complete the practise.  The practice session included three decision periods and took  participants from f i v e to ten minutes. Four versions of the decision-making task were used.  Each version had a  d i f f e r e n t promotional budget (e.g., 54, 56, 62 or 75 thousands of d o l l a r s ) , and a d i f f e r e n t response function r e l a t i n g promotion to p r o f i t .  One of these  cases was used for a l l participants during the practice session. The remaining three cases were assigned randomly to participants such that i n each session, a given participant received a new case and such that a l l possible orderings of these three cases occurred an equal number of times.  The three  cases used i n the experimental sessions were of equal objective d i f f i c u l t y , as defined by Mclntyre (1979) as the degree to which the p r o f i t r e s u l t i n g from the optimal budget a l l o c a t i o n d i f f e r e d from that r e s u l t i n g from an equal a l l o c a t i o n , as a proportion of the l a t t e r .  That i s ,  D i f f i c u l t y = ( P r o f i t optimal - P r o f i t equal)/ P r o f i t equal Mood manipulation.  Following the decision-making task, the subjects were  t o l d that the second part of the session would begin.  The following  instructions were read to each p a r t i c i p a n t : We are interested i n the use of d i f f e r e n t types of music i n the workplace and i n how people react to music. What we would l i k e you to do next i s s i t back and l i s t e n to some music that might be used at work. Listen c a r e f u l l y and think about the images that the music brings to mind. Afterward we w i l l ask you what you think of the music, and what images the music brings to mind. Upon receiving these instructions, the subjects were seated i n a comfortable lounge chair, which was placed i n front of two stereo loudspeakers.  Through these speakers was played, at a comfortable l i s t e n i n g  volume, one of the elated, neutral, or depressed mood induction audio tapes  163  developed by P i g n a t i e l l o et a l .  (1986).  Each induction took twenty minutes.  The p a r t i c u l a r mood induction used i n a given session was  predetermined  randomly so that equal numbers of subjects were assigned to each order condition and so that subjects received each The experimenter was kept b l i n d to the treatment condition The treatment condition to be used i n each session was not known by the experimenter.  This was achieved through the use of  multiple audiotapes which were given a unique code number and assigned to participants by a t h i r d party. Dependent Measures.  Following the musical induction, participants  completed manipulation checks, measures of the p r i n c i p a l dependent variables, and bogus measures designed to support the "music i n the workplace" for  the study.  The dependent measures were presented item by item on the  computer terminal. for  rationale  T y p i c a l l y , the participants would be shown instructions  each measure and provided with an example i f necessary.  clear the display screen and respond to the f i r s t item. be cleared and the following item presented.  They could then  The screen would then  Details of these measures follow  i n the order they were administered.  1.  F i r s t manipulation check: Checklist.  Subjects were presented with  instructions for the MAACL, as shown i n Figure 5.  The computer then  presented, one item at a time, the 24 items comprising the f i r s t  split-half  version of the MAACL short form, measures of anxiety, depression and h o s t i l i t y as described i n Study One.  Participants were asked to indicate on a four  point scale how well the words described their current f e e l i n g s . presented i n alphabetical order.  Items were  An example of the display for the item  "active" i s shown i n Figure 6.  2.  F i r s t manipulation check: Semantic d i f f e r e n t i a l .  Participants were  164  asked to complete the 12-item semantic d i f f e r e n t i a l measure of p l e a s u r e — displeasure and a r o u s a l — s l e e p i n e s s (Russell & Mehrabian, 1977) to describe t h e i r current f e e l i n g s .  3.  Latency.  In addition to capturing the numerical response to each  mood measure item, the latency of response was also captured, as described i n Study Two. 4.  Bogus measures.  Participants were then asked a number of questions  to enhance the face v a l i d i t y of the mood manipulation procedure and the cover story.  Subjects were asked to rate, on a seven point scale from "Very  Unsuitable" to "Very Suitable", the s u i t a b i l i t y of the music to (1) work i n general, (2) r e p e t i t i v e work, (3) i n t e r e s t i n g work, (4) work requiring concentration, and (5) work done i n groups.  They were also asked to rate  their l i k i n g of the music played during the decision-making task, and how much they f e l t that the music enhanced or detracted from their performance on the task.  5.  Perceived c o r r e l a t i o n .  The measure of perceived c o r r e l a t i o n  described i n Study Three was used.  It asked participants to indicate on a 10-  point scale the relationship between working hard and performing- well. 6.  Expectancy.  Participants were then presented with 15 items (each  combination of 3 levels of e f f o r t and 5 levels of performance) to assess their expectancy, as described i n Study Three.  These 15 scores were combined into a  composite measure of expectancy by treating them as frequencies i n a b i v a r i a t e d i s t r i b u t i o n and c a l c u l a t i n g a c o r r e l a t i o n c o e f f i c i e n t .  The combined score  was subjected to a Fisher r to z transformation to normalize the d i s t r i b u t i o n of scores.  165  7.  Recall.  Participants' r e c a l l of outcomes i n the task was assessed by  asking them to indicate how many periods out of the 10 i n which they participated that their p r o f i t was within each of f i v e ranges of performance. The levels of performance corresponded to those for the expectancy measure. These f i v e scores were combined into a composite measure using the same weights as f o r the expectancy measure.  8.  Expectations for future performance.  Participants were asked to  indicate what they expected their performance would be i f they were to make decisions f o r 10 more periods and were to expend the same amount of e f f o r t . They were asked how many times their p r o f i t would be i n each of the f i v e ranges of performance.  These f i v e scores were combined into one composite  score. 9.  Perceived control over performance.  Participants were then asked  some questions "about their p a r t i c i p a t i o n i n the Marketing Game."  They were  f i r s t asked to indicate their judgement of the amount of control they had over t h e i r performance, as described i n Study Two. 10.  Causal a t t r i b u t i o n s .  The Causal Dimension Scale (Russell, 1982)  described i n Study Three was used to assess perceived i n t e r n a l i t y ,  stability,  and c o n t r o l l a b i l i t y of a t t r i b u t i o n s . 11.  Task perceptions.  Participants were presented with 24 items  assessing task e f f o r t , interest, d i f f i c u l t y , s a t i s f a c t i o n , performance, challenge and motivation as described i n Study One.  Included among these  items also were 8 items which were verbal (as opposed to numerical) statements of the concept of effort-performance covariation. shown i n Appendix E.  These eight new items are  Following the r e s u l t s of the factor analysis performed  166  i n Study One, the four items measuring i n t e r n a l work motivation were discarded.  The twelve items exclusive of the verbal expectancy items were  combined into the following scales: (1) s a t i s f a c t i o n with task performance, (2) task d i f f i c u l t y and challenge, (3) task e f f o r t , and (4) task i n t e r e s t . Thus, with the verbal expectancy measure, f i v e task perception scales were administered.  12.  Second manipulation check: C h e c k l i s t .  To ensure that the mood  manipulation had endured the measurement of the dependent variables a second manipulation check was administered.  This consisted of the second p a r a l l e l  form of the MAACL b r i e f version, as described i n Study One. 13.  Second manipulation check: Semantic d i f f e r e n t i a l .  Participants were  again presented with the 12-item measure of p l e a s u r e — d i s p l e a s u r e and arousal—sleepiness. 14.  Response latency.  Again, the latency of response to the measures  constituting the second manipulation check was captured. 15.  Imagery.  At the end of the second and t h i r d experimental sessions  participants were asked to verbally report the images the musical mood manipulation brought to mind.  These were recorded by the experimenter, i n  part to continue the cover story for the manipulation, and i n part to assess what, i n fact, participants thought of during the manipulation.  Imagery  concerned with achievement settings, f o r example, could induce a change i n ideational set that might represent a possible confound. 16.  Recall Accuracy.  During the experimental task the computer recorded  the actual p r o f i t performance for each t r i a l .  The number of times p r o f i t was  i n each of f i v e ranges of performance, corresponding to the expectancy and  167  r e c a l l measures, was counted.  These f i v e scores were combined into one  composite measure of performance which was to obtain a measure of r e c a l l accuracy.  subtracted from the r e c a l l measure  An accuracy score of zero represents  perfect r e c a l l , p o s i t i v e scores represent optimistic r e c a l l and  negative  scores represent pessimistic r e c a l l . The questionnaire administered p r i o r to the experimental contained the following measures.  sessions  These were used as covariates i n the  analyses that follow. 1.  Self-esteem.  Global self-esteem was measured using Rosenberg's  (1965) 10-item Self-esteem Scale, as described i n Study 2.  Impression Management.  Paulhus' (1984) Balanced  Impression management was measured by  Inventory of Desirable Responding-Impression  Management Scale (BIDR-IM) as described i n Study One. i n t e r n a l consistency of the BIDR was 3.  One.  Locus of control.  .733,  In this study the  i t s Guttman lower bound was  .811.  Locus of control was measured using the Spheres of  Control Battery (Paulhus & C h r i s t i e , 1981).  The conceptual model underlying  t h i s battery holds that perceived control spans three domains of i n t e r a c t i o n : personal e f f i c a c y , interpersonal control, and s o c i o p o l i t i c a l control. Accordingly, the Spheres of Control Battery has three subscales, each has items.  Higher  of control.  10  scores on these scales indicates greater perceived i n t e r n a l i t y  Paulhus and C h r i s t i e report alpha r e l i a b i l i t i e s of .75,  .77,  and  .81 for the three scales, respectively, and substantial evidence as to the v a l i d i t y and u t i l i t y of the subscales. and Guttman lower bounds of .57 and scale, .81 and  In t h i s study i n t e r n a l consistencies  .61 were found for the personal e f f i c a c y  .82 for the interpersonal control scale, and  .76 and  .79 for  168  the s o c i o p o l i t i c a l control scale.  Inspection of the item i n t e r c o r r e l a t i o n  matrix revealed no obvious reason for the low r e l i a b i l i t y of the personal e f f i c a c y scale i n this sample.  Because i t f a i l e d to meet the c r i t e r i o n for  r e l i a b i l i t y of .70 adopted here, i t was not used further. At the end of each session each participant was asked to wait for a few minutes while the experimenter ostensibly checked to make sure that the computer program captured the participant's responses.  During this time the  music from the "elated" mood induction procedure used by Eich and Metcalfe ( i n press) to manipulate mood was played. selection of magazines to read.  Participants were also provided with a  This procedure was employed to moderate the  possible depressed post-experimental state of p a r t i c i p a n t s .  Before being  allowed to leave, each participant was thanked and reminded of his or her next appointment.  Because of the length of session one the imagery questionnaire  was not administered.  Sessions Two and Three The second and t h i r d sessions were i d e n t i c a l to Session One, with the following exceptions. F i r s t , the practice session on the decision-making task was omitted i n sessions Two and Three.  Participants were t o l d that they would begin the  actual decision-making problem without a practice session but that, i n order to ensure that they remembered the essential aspects of the task, they would be asked a number of questions about the task. as i n Session One.  The same questions were used  This allowed some time to pass between the time the  participant entered the experimental session and when the task began, and ensured that this time was spent i n a similar way by a l l p a r t i c i p a n t s .  This  increased the l i k e l i h o o d that participants began the task i n Sessions Two and  169  Three i n a similar neutral mood state. Second, i t was emphasized that although the decision making task i n a l l sessions was i d e n t i c a l , the decision problem was a d i f f e r e n t one. A d i f f e r e n t response function was used, of equal d i f f i c u l t y but with a d i f f e r e n t budget. Third, the mood induction condition employed was, of course, d i f f e r e n t i n each session. Fourth, the shorter length of Sessions Two and Three allowed the openended measure of mood imagery to be administered. F i n a l l y , at the end of each participant's t h i r d session, the experimenter asked a set of questions to assess participant's suspicions about the study, whether they had guessed the true hypothesis or some other hypothesis, and whether any inter-subject contamination had occurred.  Each p a r t i c i p a n t was  thanked and f u l l y debriefed v i a a written summary once a l l subjects had participated.  Results Before proceeding with the multivariate analyses, the variables used w i l l be evaluated with respect to p r a c t i c a l l i m i t a t i o n s of the technique.  Evaluation of Assumptions Studies i n which measures of few dependent variables are obtained f o r each of several levels of the independent v a r i a b l e may be analyzed by multivariate or univariate procedures.  That i s , scores can be conceptualized  as the same measure on multiple occasions and analyzed by repeated-measures Analysis of variance (ANOVA) procedures.  A l t e r n a t i v e l y , scores can be  conceptualized as separate tests of the same subject where each dependent variable i s measured on the same scale.  M u l t i v a r i a t e ANOVA (MANOVA) can then  170  be performed on the between-subjects independent variables for the set of within-subjects dependent measures.  While the univariate procedures are less  complex, they also require a highly r e s t r i c t i v e set of assumptions concerning population treatment variances and covariances, (e.g.  the assumption of  homogeneity of covariance, or that the correlations among levels of the within-subjects variable are constant over a l l combinations of l e v e l s ) . MANOVA procedures are preferable where such assumptions are unwarranted.  More  importantly, MANOVA procedures protect against Type I error i n testing multiple dependent measures, as i s the case here. As described e a r l i e r , MANOVA i s robust to modest v i o l a t i o n s of the assumption of multivariate normality insofar as the v i o l a t i o n i s created by skewness rather than by o u t l i e r s (Mardia, 1971).  In t h i s study, the data set  comprised 13 subjects i n each of 6 between-subjects order c e l l s . subject, one observation was missing.  For one  A multiple regression procedure  (BMDPAM) was used to predict the missing value from complete cases. The missing observation was replaced with t h i s estimated value.  Thus sample sizes  are equal. With 13 cases i n each of 6 between-subjects c e l l s , the degrees of freedom for error i n the univariate case i s 14, below Tabachnik and F i d e l l ' s suggested c r i t e r i o n for robustness. Therefore, the extent of skewness and the presence of o u t l i e r s was tested f o r . The skewness of the d i s t r i b u t i o n each variable i n standard units was computed.  Of the 25 dependent variables i n each of  eighteen order by session c e l l s , i n 11 instances (or 2.4% of the 468 instances examined) skewness was i n excess of ± 2.58 standard units. same variable skewed i n d i f f e r e n t order c e l l s .  In no case was the  Thus skewness was not  concentrated on any one variable across order c e l l s .  Univariate o u t l i e r s were  i d e n t i f i e d by examining extreme z-scores for variables on a casewise basis.  171  Of the 5850 observations examined (25 variables on 3 occasions for 78 cases), 56 or about 1% had z-scores i n excess of ± 2.58. therefore, are not numerous.  Univariate o u t l i e r s ,  The presence of multivariate o u t l i e r s was  evaluated by examining the Mahlanobis distance of each case to the centroid of i t s group.  No cases exceeded the c r i t i c a l value for Mahlanobis distance at  p < .01. Overall, the d i s t r i b u t i o n of scores for the dependent variables can be summarized as moderately skewed for a small proportion of cases but containing no o u t l i e r s .  Transformations might be appropriate for those variables with  skewed d i s t r i b u t i o n s .  This would, however, make interpretation more complex  and would be required for not just skewed variables, but their measure counterparts. demonstration  repeated-  In l i g h t of the absence of o u t l i e r s and the  by Mardia (1971) that MANOVA i s robust to v i o l a t i o n of normality  when caused by skewness rather than o u t l i e r s , then, no variables were transformed.  Examination of the r e s u l t s of MANOVA following square-root  transformation of skewed variables resulted, i n f a c t , i n r e s u l t s i d e n t i c a l to those from the analysis of untransformed v a r i a b l e s . Analysis of the d i s t r i b u t i o n s of variables following square-root transformation showed that this eliminated s i g n i f i c a n t skewness. As a preliminary check of homogeneity of variance-covariance matrices, sample variances for each of the 78 dependent variables were examined across the 6 order groups.  For no dependent variable d i d the r a t i o of the largest to  smallest variance approach 20:1, F i d e l l (1983).  the c r i t e r i o n suggested by Tabachnik and  Univariate homogeneity of variance tests on the 78 variables  revealed f i v e s i g n i f i c a n t r e s u l t s at p = .05, including one s i g n i f i c a n t result, at p = .01 (Bartlett-Box F ) . of tests performed.  Such a result i s to be expected given the number  In any event, the robustness of significance tests i n  172  MANOVA i s guaranteed for equal sample sizes (Hakstian, Roed & Lind, 1979). Evidence of non-linearity of relationships among dependent variables was sought by examining the plots of observed values versus standardized residuals for  each dependent variable.  No evidence of gross c u r v i l i n e a r i t y was found,  therefore variables were retained i n o r i g i n a l form.  In the analyses that follow, MANOVA i s performed  f i r s t on each set of  manipulation check measures, and separately on the remaining measures. The presence of m u l t i c o l l i n e a r i t y and s i n g u l a r i t y was investigated by examining the SMC of each variable with a l l other variables within each of these analyses.  In no instance was a SMC close enough to 1.0 to warrant  concern.  Inspection of the eigenvalues and determinants of the w i t h i n - c e l l s c o r r e l a t i o n matrix revealed that they were s u f f i c i e n t l y d i f f e r e n t from zero such that neither m u l t i c o l l i n e a r i t y not s i n g u l a r i t y was judged to be a problem.  Analysis of Manipulation Checks Recall that the manipulation checks consisted of six measures: s e l f reported anxiety, depression, h o s t i l i t y , pleasure, arousal, and a behavioral measure of response latency.  Recall also that p a r a l l e l versions of these  measures were administered before and a f t e r the other dependent measures. Thus the manipulation checks consist of two sets of 6 v a r i a b l e s . be referred to as the " f i r s t " and "second" manipulation checks.  These w i l l Overall,  multivariate analysis of variance would be appropriate to a test of differences between a l l 12 measures.  However, the interpretation of  subsequent analyses, p a r t i c u l a r l y stepdown F-tests, would be problematic. Because stepdown analyses p a r t i a l out the effects of other variables, and because we know that the p a r a l l e l versions of each mood manipulation check  173  measure are very highly correlated, i f we included both sets of manipulation checks i n a single stepdown analysis, we would not expect to f i n d any effects for  the second set of v a r i a b l e s .  checks are considered separately. experimentwise  Therefore the f i r s t and second manipulation However, to guard against increased  error inherent i n such multiple testing, multivariate tests  using a l l twelve measures simultaneously are also reported. The experimental design, as discussed above, included the within-subjects session variable, which was whether the measure was administered during the first,  second, or t h i r d experimental session, and the between-subjects  order  variable, denoting which of the six possible orderings of the mood manipulation that the participants experienced over the three sessions. MANOVA was thus a 6 x 3 mixed, between-within variables.  Each  analysis of 6 dependent  In both SPSSX and BMDP this type of design i s treated as a  repeated-measures analysis of 18 dependent variables, with the dependent variables grouped to capture the session v a r i a b l e . It i s important to note that the influence of the mood treatment on the dependent measures i s captured i n the order by session i n t e r a c t i o n .  Because  the order e f f e c t represents i n which session the participant receives each l e v e l of the mood induction, the effect of the mood induction on dependent measures i s that of the order by session i n t e r a c t i o n . One consequence of t h i s i s to complicate inspection of group means. There are 78 order by session c e l l s for each dependent variable.  Each of  these c e l l means for each dependent variable i s presented i n Appendix F. To f a c i l i t a t e inspection of main and i n t e r a c t i o n e f f e c t s , the tables i n the text that follows have been collapsed according to the treatment e f f e c t of interest and show the mean score for the 78 participants f o r each variable i n each treatment.  174  Results of Manipulation Checks. Wilk's c r i t e r i o n indicated a non-significant effect of order on the f i r s t and second manipulation checks taken simultaneously, F(60,289.42) = 0.79, p > .05.  Order of mood treatment presentation over the three experimental  sessions d i d not s i g n i f i c a n t l y a f f e c t the response of p a r t i c i p a n t s to the mood manipulations, as indicated by the 12 manipulation checks. The effect of interest, the order by session i n t e r a c t i o n , was s i g n i f i c a n t for  the f i r s t and second manipulation checks taken simultaneously, as  indicated by Wilk's c r i t e r i o n , F (120, 1046.95) = 2.16, p< .001, rj = .82. 2  When the f i r s t manipulation checks are considered alone, the effect of order i s non-significant, F(30, 270)= 0.69, p> .05. The order by session i n t e r a c t i o n was s i g n i f i c a n t , F(60, 733.32)= 3.11, p< .001, r? = .70. 2  S i m i l a r l y , the second manipulation checks, considered separately, reveal a non-significant effect of order, F(30,270)= .69, p> .05, and a s i g n i f i c a n t order by session interaction F(60, 733.32)= 2.22, p< .001,  7? = 2  .59. Thus, the  mood manipulation had a s i g n i f i c a n t effect on the manipulation checks administered both before and a f t e r the other dependent measures.  The r e s u l t s  also indicate a strong association between the mood treatment and the mood measures. To evaluate the contribution of each mood measure to the o v e r a l l session by order interaction, stepdown F-tests were performed.  The results of  univariate analyses are shown i n Table 18. Among the f i r s t manipulation checks, s i g n i f i c a n t univariate order by session i n t e r a c t i o n effects were found for  self-reported anxiety, depression, h o s t i l i t y , pleasure, and arousal. The  latency measure was not s i g n i f i c a n t . correlations between the v a r i a b l e s .  Such univariate analyses overlook Examination of the pooled,  averaged  within c e l l s correlations, shown i n Table 19, reveals that the measures of  175  Univariate Dependent Variable  Stepdown  d.f.  d.f.  F i r s t Manipulation Check: Latency Arousal Pleasure Depression Anxiety Hostility  .53 11.66 4.41 11.40 4.62 2.17  10/144 10/144 10/144 10/144 10/144 10/144  .863 .000 .000 .000 .000 .023  .53 11.58 1.99 3.29 1.25 2.00  10/144 10/143 10/142 10/141 10/140 10/139  .863 .000 .039 .001 .262 .037  10/144 10/144 10/144 10/144 10/144 10/144  .187 .016 .006 .000 .074 .080  1.39 2.27 2.29 5.02 1.06 1.59  10/144 10/143 10/142 10/141 10/140 10/139  .187 .017 .016 .000 .397 .116  Second Manipulation Check: Latency Arousal Pleasure Depression Anxiety Hostility  1.39 2.29 2.59 7.37 1.76 1.73  Table 18. Univariate and Stepdown F-tests, Order by Session (Mood) Effect on manipulation checks, Study Four.  Latency  Arousal  Pleasure  Depression  Anxiety  F i r s t Manipulation Checks:  Arousal  -.130  Pleasure  -.129  .308  Depression  .103  -.432  -.702  Anxiety  .028 .  .014  -.369  .430  Hostility  .075  .178  .214  -.177  .192  Second Manipulation Checks :  Arousal  -.168  Pleasure  -.043  .022  Depression  .250  -.295  -.472  Anxiety  .112  .072  -.358  .460  Hostility  .110  -.068  -.390  .515  ,.297  Table 19. Pooled w i t h i n - c e l l correlations for manipulation checks, Study Four.  177  mood are i n t e r c o r r e l a t e d .  B a r t l e t t ' s test of sphericity was s i g n i f i c a n t for  the c o r r e l a t i o n matrices of the f i r s t and second manipulation checks ( x (15)= 2  192.9, p< .001; x of  2  (15)= 158.4, p< .001, respectively), supporting r e j e c t i o n  the hypothesis that the matrices are i d e n t i t y matrices and hence that the  variables are independent. The r e s u l t s of the stepdown analyses are shown i n Table 18. The order of entry into analysis was as follows: latency, arousal, pleasure, depression, anxiety, and h o s t i l i t y .  Entered f i r s t , the stepdown test f o r response latency  i s equivalent to the univariate test, and so showed no e f f e c t .  The second  variable entered, self-reported arousal, showed a s i g n i f i c a n t e f f e c t , stepdown F(10, 143)= 11.58, p< .001. The strength of association between the order by session e f f e c t and arousal, as indicated by the r a t i o of hypothesis sum of squares to hypothesis plus error sum of squares, was 0.45. Table 20 shows a steady trend from highest arousal i n the elation treatment  (31.0) to least  arousal i n the depression treatment (19.5), with the score f o r the neutral treatment  (23.6) f a l l i n g closer to the depression treatment.  After the  pattern of differences measured by latency and arousal were entered, a difference was found f o r self-reported pleasure, stepdown F(10, 142)= 1.99, p< .05, r> = .11. Pleasure was greatest i n the elation condition (37.1), s l i g h t l y 2  less i n the neutral condition (36.0), and least i n the depression condition (32.2) The fourth variable entered, depression showed a s i g n i f i c a n t e f f e c t , stepdown F(10, 141)= 3.29, p< .001, rj = .19. Self-reported depression was 2  greatest i n the depression treatment group (26.4) and least i n the e l a t i o n group (19.1).  The mean score f o r the neutral group (21.9) was, as i n the case  for pleasure, closer to that of the e l a t i o n condition. The f i f t h variable entered, anxiety, was not s i g n i f i c a n t l y affected, stepdown F ( l , .05.  140)= 1.25, p>  F i n a l l y , h o s t i l i t y was s i g n i f i c a n t l y affected, stepdown F ( l ,  139)= 2.00,  178  Elation  Mean  St. Dev.  Neutral Mean  St. Dev.  Depression  Mean  St. Dev.  F i r s t Manipulation Check: Anxiety Depression Hostility Pleasure Arousal Latency  7.03 19.12 17.00 37.14 31.02 144.18  1.95 4.58 1.40 6.48 10.34 51.22  7.19 21.98 16.70 36.06 23.65 145.63  2.19 6.58 1.75 7.20 8.16 57.17  8.11 26.41 16.53 32.21 19.51 146.03  2.31 7.31 1.56 8.66 6.18 52.56  8.02 21.65 15.73 31.56 28.15 110.21  2.23 5.86 1.58 3.27 3.33 27.94  8.53 23.92 16.10 31.00 27.88 107.69  2.075 6.365 1.884 3.797 3.154 24.751  Second Manipulation Check: Anxiety Depression Hostility Pleasure Arousal Latency  8.19 19.91 15.92 31.76 29.46 107.85  2.13 4.39 71 73 35 27.73  Table 20. Means and Standard Deviations f o r Manipulation Checks, by Mood Effect, Study Four.  179  p< .05. Scores on the h o s t i l i t y scale were highest i n the e l a t i o n condition (17.0), lower i n the neutral condition (16.7) and least i n the depression condition (16.5).  Although  s t a t i s t i c a l l y s i g n i f i c a n t , these effects are  small. The mood treatment had a much smaller effect on the second set of manipulation checks.  Significant univariate effects were evident for arousal,  pleasure and depression.  In stepdown analyses, these three measures were  shown to contribute uniquely to the o v e r a l l order by session i n t e r a c t i o n . Latency, as before, d i d not s i g n i f i c a n t l y d i f f e r e n t i a t e between treatment groups.  Self-reported arousal was uniquely and s i g n i f i c a n t l y related to  treatment,  stepdown F(10, 143)= 2.27, p< .05, r) = .14, as was pleasure, 2  stepdown F(10, 142)= 2.29, p< .05, r) = .14, and depression, stepdown F(10, 2  141)= for  5.02, p< .001, 7) = .26. Table 20 shows a congruent pattern of scores z  the three v a r i a b l e s , pleasure and arousal were highest i n the e l a t i o n  group (31.8 and 29.5, respectively), intermediate i n the neutral group (31.6 and 28.1) and lowest i n the depression group (31.0 and 27.9).  The scores f o r  depression mirrored these, being highest i n the depression treatment intermediate i n the neutral treatment treatment  (19.9).  (23.9),  (21.6) and lowest i n the e l a t i o n  These mean differences, although s t a t i s t i c a l l y  significant,  were notably smaller than those among the f i r s t set of manipulation  checks.  Analysis of the session variable revealed a s i g n i f i c a n t multivariate effect on the combined dependent variables for both the f i r s t and second set of manipulation checks, F(12, 278) = 31.76, P < .001, and F(12, 278) = 9.07, p < .001, respectively. Wilk's c r i t e r i o n indicated that a strong association existed between the session variable and the combined dependent variables f o r the manipulation checks administered before the other measures (T? = .82), 2  while a more moderate association existed for the second set of manipulation  180  checks (r) =  .48).  2  Examination of the univariate tests reveals s i g n i f i c a n t session effects among the f i r s t manipulation checks for latency, arousal, depression hostility.  These are shown i n Table 21.  and  C e l l means, shown i n Table 22., show  that the latency of p a r t i c i p a n t s ' responses dropped markedly over the three sessions, from 203.5 to 125.1  to 107.3  increased somewhat ( c e l l means of 22.3,  seconds. 25.4,  Self-reported arousal  and 26.5  for sessions 1, 2,  3) while depression and h o s t i l i t y decreased (23.9, 21.9, depression;  17.0,  16.7,  and 16.5  for h o s t i l i t y ) .  and 21.7  and  for  However, when stepdown  analyses were performed on these variables, the r e s u l t s showed that once the pattern of differences measured by the latency variable had been entered, remaining variables did not contribute to the session e f f e c t .  the  The stepdown  r e s u l t s are also shown i n Table 21. A similar pattern was  evident for the second manipulation checks, but  while the univariate test on self-reported arousal almost exceeds a significance c r i t e r i o n of .05, the only s i g n i f i c a n t r e s u l t was latency.  for response  Again, though, stepdown analysis revealed that once latency  was  entered into the analysis, the remaining variables d i d not make a unique contribution.  Examination of c e l l means showed that, as for the f i r s t  latency  measure, response latency decreased markedly over the three sessions (means = 123.6, 107.2, and 95.0). Overall, the results for the manipulation checks indicate that the mood induction procedure produced s i g n i f i c a n t s h i f t s i n the mood states of individuals as predicted. depression  As was  found i n Study Two,  the e l a t i o n and  inductions were characterized by self-reports of high and  low  pleasure, respectively, high and low arousal, respectively, and low and depression,  respectively.  high  The neutral induction r e s u l t s were most l i k e those  181  Univariate Dependent Variable  Stepdown  d.f.  d.f.  F i r s t Manipulation Check: Latency Arousal Pleasure Depression Anxiety Hostility  314.15 6.74 .89 4.33 1.18 3.40  2/144 2/144 2/144 2/144 2/144 2/144  .000 .002 .412 .015 .310 .036  314.15 0.05 0.53 0.72 1.03 0.97  2/144 2/143 2/142 2/141 2/140 2/139  .000 .951 .587 .487 .356 .380  2/144 2/144 2/144 2/144 2/144 2/144  .000 .055 .962 .286 .378 .082  63.08 0.11 0.13 0.49 0.28 1.17  2/144 2/143 2/142 2/141 2/140 2/139  .000 .894 .875 .609 .755 .311  Second Manipulation Check: Latency Arousal Pleasure Depression Anxiety Hostility  63.08 2.95 0.03 1.26 0.97 2.54  Table 21. Univariate and Stepdown F-tests, Session Effect on manipulation checks, Study Four.  182  Session One  Mean  Std. Dev.  Session Two  Mean  Std. Dev.  Session Three Mean  Std. Dev.  F i r s t Manipulation Check: AnxietyDepression Hostility Pleasure Arousal Latency  7.58 23.84 17.02 34.50 22.26 203.48  2.16 7.06 1.55 8.09 8.92 46.09  7.23 21.98 16.73 35.12 25.42 125.10  2.16 7.00 1.66 8.12 10.35 26.22  7.52 21.69 16.48 35.79 26.50 107.26  2.29 6.59 1.51 7.08 9.16 23.18  8.21 21.75 15.82 31.38 28.48 107.18  2.05 5.80 1.50 3.74 3.52 21.29  8.10 21.39 15.73 31.44 29.03 95.02  2.09 5.58 1.56 2.85 3.11 19.86  Second Manipulation Check: Anxiety Depression Hostility Pleasure Arousal Latency  8.43 22.33 16.20 31.50 27.97 123.55  2.31 6.08 2.05 3.27 3.33 30.06  Table 22. Means and Standard deviations for manipulation checks, by Session E f f e c t , Study Four.  183  for the e l a t i o n induction for measures of pleasure and depression, but more l i k e those for the depression induction on the arousal measure.  The size of  these effects was smaller than i n Study Two, however, and no effect on the behavioral response latency measure was found. S i g n i f i c a n t effects were evident primarily for the manipulation checks administered f i r s t , d i r e c t l y after the musical induction.  Those administered  second, after the other measures had been c o l l e c t e d , were smaller. strength of association between the treatment and the combined  The  dependent  variables was lower for the second set of measures ( T J = .59 vs. 2  n  2  = .82),  and mean differences were very small. Informal analysis of the content of the imagery questionnaires completed revealed that i n no instances d i d any participant report images concerned with task evaluation of performance or that might i n any way represent a confound to the study of expectancy.  Most often images concerned the content of the  music, such as i n reports of images of "marching bands", "busy c i t i e s " , " s t r i n g quartets", "pastoral scenes", "rainy days", and "funerals". Participants often used words d e s c r i p t i v e of moods and feelings.  There was a  high degree of agreement between the a f f e c t i v e tone of the descriptions and the mood treatment which the participant had just completed.  This further  supports the effectiveness of the mood manipulation procedure. Worthy of comment i s the finding that the behavioral response latency measure was not s i g n i f i c a n t , unlike the finding of Study Two.  That i s , no  s i g n i f i c a n t difference between mood treatment groups was evident on the psycho motor measure.  One possible reason for this may be the increased f a m i l i a r i t y  and experience with the computer keyboard over the three experimental sessions for p a r t i c i p a n t s i n Study Four.- Participants i n Study Four used the keyboard for the experimental task as well as the dependent measures.  The s e n s i t i v i t y  184  of the latency measure to the effects of depression may have been overshadowed by the effects of p r a c t i c e .  Evidence for t h i s i s found i n the multivariate  effect of the session variable on the manipulation checks which was revealed by stepdown analyses to be due to the latency measure.  Over the three  experimental sessions, the response latency of participants decreased by almost 50%.  Measures of Expectancy and Task Perceptions A multivariate, repeated-measures analysis of variance was performed on the 13 remaining dependent measures.  These consisted of the p r i n c i p a l e f f o r t -  performance covariation measure, measures of perceived control and c o r r e l a t i o n , measures of r e c a l l , accuracy, and expected performance, of causal a t t r i b u t i o n s and task perceptions. The multivariate test of significance of the order effect was not s i g n i f i c a n t , F (65,287.49) = 1.12, p > .05. The order of presentation of the mood effect had l i t t l e apparent effect on the combined dependent measures. The means and standard deviations of each measure for a l l order c e l l s are shown i n Appendix F. As before, the mood treatment effect i s tested by the session by order interaction effect.  For the combined dependent measures, the o v e r a l l  multivariate test was not s i g n i f i c a n t , F (130, 1072) = .838, p >.05. r e s u l t s do not support Hypothesis One or Two. mood treatment on expectancy  Thus the  That i s , there was no effect of  (Hypothesis One), nor on i n t e r n a l i t y of  a t t r i b u t i o n s (Hypothesis Two). It i s possible that adjustment  for i n d i v i d u a l  differences might reveal some effect, therefore Hypothesis Three w i l l be tested below.  Although the absence of a s i g n i f i c a n t omnibus MANOVA precludes  most further analysis, the reader might be interested to know that for not one  185  of the t h i r t e e n dependent variables was significant.  The effect of mood was,  the univariate order by session test  apparently,  not present or not  detected.  The means and standard deviations of the dependent variables, aggregated by mood treatment, are shown i n Table 23. The averaged multivariate e f f e c t of the session v a r i a b l e on the combined dependent variables was  s i g n i f i c a n t , F(26, 264)  = 2.35,  p < .001,  r? =  .34.  2  Examination of the univariate tests revealed s i g n i f i c a n t e f f e c t s for perceived control, verbal expectancy, performance r e c a l l , r e c a l l accuracy, task e f f o r t , task i n t e r e s t , and task d i f f i c u l t y and challenge. shown i n Table 24.  The univariate r e s u l t s are  Inspection of the means i n Table 25 shows that mean  perceived control over performance declines over sessions as does the verbal measure of expectancy. l a t e r sessions and, sessions.  Individuals r e c a l l higher levels of performance i n  i n fact, t h i s r e c a l l i s more accurate i n subsequent  Self-reported task e f f o r t declined over the three sessions  and  individuals reported the task to be less d i f f i c u l t and challenging, and less i n t e r e s t i n g i n l a t e r sessions. To investigate the unique e f f e c t s of the session main e f f e c t on the i n d i v i d u a l dependent variables, a stepdown analysis was  performed.  Examination of the averaged w i t h i n - c e l l s c o r r e l a t i o n matrix, shown i n Table 26, reveals that the dependent measures are i n t e r c o r r e l a t e d . B a r t l e t t ' s test of s p h e r i c i t y was  s i g n i f i c a n t , F (13,144) = 10692.9, p < .001,  supporting  r e j e c t i o n of the hypothesis that the matrix i s an i d e n t i t y matrix and that the variables are independent.  The a p r i o r i ordering of the importance of the  dependent variables was as follows: (1) perceived covariation, (2) control, and  (3) c o r r e l a t i o n , (4) verbal expectancy, (5) performance r e c a l l , (6) r e c a l l  accuracy, (7) performance expectation,  (8) i n t e r n a l i t y , (9) s t a b i l i t y , of  causality, (10) task e f f o r t , (11) task i n t e r e s t , (12) s a t i s f a c t i o n with task  186  Elated Measure  Mean  Perceived Covariation Control Correlation Satisfaction with performance  Neutral  Depressed  Std.Dev.  Mean  Std.Dev.  Mean  .22 52.24 4.79  .18 19.41 2.15  .24 53.59 4.92  .16 20.06 1.95  .24 53.98 4.83  .18 21.42 2.23  13.29  2.03  13.12  2.21  13.48  2.39  6.84  1.33  6.65  1.60  6.73  1.53  11.94  2.74  11.62  2.69  11.67  2.58  6.93  1.52  6.64  1.53  6.66  1.63  Expectancy-verbal  28.10  5.13  27.16  5.43  27.89  4.58  Attributed Internality Stability Controllability  15.87 12.23 17.92  4.47 4.30 3.98  16.05 12.15 17.64  4.60 4.20 3.69  15.43 12.30 17.28  4.79 3.98 3.64  Performance recall expectation  3.45 3.56  .51 .66  3.46 3.59  .60 .47  3.42 3.60  .61 .51  .22  .39  .22  .57  .21  .52  Effort Task d i f f i c u l t y and challenge Task Interest  Recall accuracy  Std.Dev.  Table 23. Means and standard deviations f o r dependent variables by mood e f f e c t , Study Four.  187  Univariaite F  d.f.  Measure  Stepdc>wn P  F  d.f.  P  Perceived covariation control correlation  1.56 8.37 .77  2/144 2/144 2/144  .215 .000 .463  1.99 10.44 1.44  2/144 2/143 2/142  .140 .000 .241  .027 .127 .018  Verbal expectancy  4.56  2/144  .012  1.21  2/141  .301  .017  Performance recall expectation  3.11 .51  2/144 2/144  .048 .603  1.59 1.26  2/140 2/139  .206 .288  .022 .018  Recall  accuracy  6.54  2/144  .002  .05  2/138  .950  .000  Attributions of internality stability  2.17 .33  2/144 2/144  .117 .720  2.31 2.61  2/137 2/136  .735 .077  .043 .037  11.74  2/144  .000  3.32  2/135  .039  .047  Task i n t e r e s t  6.86  2/144  .001  4.75  2/134  .010  .066  S a t i s f a c t i o n with task performance  1.99  2/144  .140  .02  2/133  .980  .000  Task d i f f i c u l t y and challenge  3.05  2/144  .050  .99  2/132  .372  .815  Task e f f o r t  T  Table 24. Univariate and Stepdown F-tests on dependent variables, Session e f f e c t , Study Four.  188  Session One Mean Perceived covariation Perceived control Perceived correlation  Std.Dev.  Session Two Mean  Std.Dev.  Session Three Mean  Std.Dev.  .25  .14  .22  .18  .22  .20  58.60  18.47  51.85  20.55  49.36  20.73  5.00  2.16  4.67  2.17  4.87  1.99  28.67  4.42  27.58  5.03  26.89  5.55  3.33  .72  3.46  .56  3.54  .37  3.54  .64  3.61  .56  3.60  .44  .35  .64  .21  .50  .09  .23  Internality  16.01  4.34  16.12  4.83  15.21  4.65  Stability  12.10  4.03  12.44  4.48  12.14  3.95  Task E f f o r t  7.19  1.25  6.74  1.48  6.29  1.57  Task Interest  7.05  1.55  6.64  1.57  6.55  1.54  S a t i s f a c t i o n with task performance  13.57  2.02  13.34  2.23  12.98  2.36  Task D i f f i c u l t y and challenge  12.15  2.99  11.53  2.45  11.56  2.50  Verbal Expectancy Performance recall Performance expectation Recall accuracy  Table 25. Means and Standard Deviations for dependent variables by Sessions, Study Four.  (1)  (1) S a t i s f a c t i o n w i t h task performance (2) Task e f f o r t (3) Task d i f f i c u l t y and c h a l 1 e n g e (4) Task i n t e r e s t (5) V e r b a l e x p e c t a n c y (6) P e r c e i v e d i n t e r n a l i t y s t a b i 1 i ty (7) covariation (8) P e r c e i v e d control (9) correlat ion ( 10) (11) Performance r e c a l l ( 12) P e r f o r m a n c e e x p e c t a t i o n (13) Recal1 a c c u r a c y  Table  26.  Within-cells  (2)  (3)  (4)  (5)  (6)  (7)  (8)  (9)  (10)  (11)  (12)  .485 .051 .395 .513 .447 .095 .225 - .049  . 137 .207 . 522 . 358 . 154 .209 - . 107  .010 .081 .009 .027 . 104 .004  . 193 . 229 - .009 .052 .039  .514 . 125 .203 .008  .026 .080 .060  . 6 16 -.833  - . 360  . 1 16 - . 177 .111 -.052 .060 - .003 - . 152 . 321 . 206 . 189 .311 . 104  correlation  .274 . 4 18 . 191 .200 . 108 . 103 .238 .229 .098 - .020 - .072  . 336 . 153 .077 - .077 . 128 .077 . 108 - .057 - . 129 .017  m a t r i x , Study  Four.  .209 . 133 - . 194 . 136 .218 . 192 .026 - .030 .016  -  190  performance, and  (13) task d i f f i c u l t y and challenge.  stepdown analysis are shown i n Table 24. shown i n Table  The r e s u l t s of the  Unadjusted session group means are  25.  Table 24 shows that, when the pattern of differences measured by variables entered e a r l i e r i n t o the stepdown analyses are accounted f o r , s i g n i f i c a n t unique contributions to the multivariate test are made by perceived c o n t r o l , performance r e c a l l , and task e f f o r t .  These session effects  are l i k e l y the r e s u l t of increased experience with the experimental reduced novelty of the experimental  task and  procedure.  Individual Differences To test whether individual d i f f e r e n c e variables moderated the effect of mood on self-esteem a sub-group analysis was performed.  That i s , p a r t i c i p a n t s  were divided i n t o those with high scores and those with low scores, and t h i s factor was added as an independent v a r i a b l e to the multivariate analysis of variance.  The use of a within-subjects design precluded the treatment of the  individual d i f f e r e n c e measures as continuous  covariates.  Because the  i n d i v i d u a l d i f f e r e n c e measure i s assumed to be stable, i t i s measured only once and i s therefore constant across levels of the within- subjects f a c t o r . So no covariate effect on the mood factor i s possible. Within each order condition group the six participants highest on an i n d i v i d u a l d i f f e r e n c e variable were placed i n one subgroup and the six lowest scorers were placed i n the other. order group was discarded. maintained  Data for the remaining p a r t i c i p a n t i n each  Ties were broken randomly.  This procedure  equality of c e l l size, thus avoiding problems caused by  orthogonality and v u l n e r a b i l i t y to v i o l a t i o n of assumptions.  non-  The r e s u l t i n g  tests are very conservative because of the power loss inherent i n adding a  191  factor, and because some members of a "high" subgroup i n one order condition may a c t u a l l y have lower i n d i v i d u a l difference scores than members of a "low" subgroup i n another order condition, and v i c e versa. The self-esteem  subgroup analysis revealed no s i g n i f i c a n t e f f e c t s . The  multivariate main effect of self-esteem i n t e r a c t i o n e f f e c t s with order, representing  the mood e f f e c t .  was not s i g n i f i c a n t , neither were the  session or order by session, the l a t t e r Thus i n d i v i d u a l differences i n self-esteem  not have any influence on the mood —  expectancy r e l a t i o n s h i p .  did  Hypothesis  Three received no support. A similar analysis for the interpersonal control scale of the Spheres of Control measure revealed no main effects or simple interactions, however the interpersonal control by order by session i n t e r a c t i o n was marginally s i g n i f i c a n t , F(130,1170) = 1.23, p = .051 by P i l l a i s ' c r i t e r i o n , F(130,880) = 1.26,  p = .035 by Wilks' c r i t e r i o n .  The presence'of a multivariate e f f e c t  suggests at least one s i g n i f i c a n t univariate e f f e c t : exactly one was found. The univariate three way i n t e r a c t i o n was s i g n i f i c a n t f o r the perceived c o r r e l a t i o n measure.  Inspection of the c e l l means indicated that the form of  the order by session i n t e r a c t i o n (the mood effect) was d i f f e r e n t for each control subgroup i n the following way: For individuals high i n interpersonal control, scores on the perceived c o r r e l a t i o n measure were high i n the elation and depression  conditions and lower i n the neutral condition.  For individuals  low i n interpersonal control, scores were low i n the e l a t i o n and depression conditions and higher i n the neutral condition. The lack of any other e f f e c t s and the marginal nature of the multivariate significance test suggest that l i t t l e meaning should be attached result.  to t h i s  I t i s , quite possibly, due to chance alone.  F i n a l l y , no effects were found i n the analysis of subgroups formed on the  192  s o c i o p o l i t i c a l control measure.  Discussion: Alternate Explanations A number of a l t e r n a t i v e explanations exist for the f a i l u r e of Study Four to demonstrate a relationship between mood and expectancy.  Among these are  f a i l u r e of the musical induction procedure to successfully manipulate mood, i n s e n s i t i v i t y of the expectancy measure, lack of experimental control, i n s u f f i c i e n t s t a t i s t i c a l power to detect a relationship or the absence of any true r e l a t i o n s h i p .  Support for each of these p o s s i b i l i t i e s w i l l be considered  i n turn and ways of surmounting them w i l l be explored.  Manipulation Failure It i s possible that a r e l a t i o n s h i p exists between mood and expectancy, as hypothesized, but that i t was not revealed because mood was not successfully manipulated.  In other word, there was no mood effect because there was no  mood change. The a p r i o r i support for this alternate explanation i s mixed.  Recall  that Study Two showed that the manipulation was successful i n influencing both self-reported mood and an unobtrusive, behavioral response latency indicator of depression.  The self-report measures have a long h i s t o r y of use and  substantial support for their v a l i d i t y .  Taken alone, Study Two suggests that  we can have confidence i n the success of the manipulation. Four contained a task that was not present i n Study Two.  However, Study The self-report  manipulation checks used there were s i g n i f i c a n t l y influenced by the manipulation, but the latency measure was not.  Further, the size of the  manipulation effect was much smaller on the measures taken after the p r i n c i p a l dependent measures.  From these r e s u l t s we can conclude f i r s t that a  193  s i g n i f i c a n t difference i n self-reported mood d i d exist following the mood induction.  Second, while the difference i n self-reported mood at the time of  the second manipulation check was also s t a t i s t i c a l l y s i g n i f i c a n t , the mean differences were small.  Overall, the mood induction procedure appeared to  change mood as hypothesized but t h i s manipulation was greatly diminished over the measurement period. These mood s h i f t s , although s t a t i s t i c a l l y s i g n i f i c a n t , may not have been of s u f f i c i e n t magnitude to cause the e f f e c t s hypothesized.  Eich and Metcalfe  (in press) found that the Velten induction d i d not produce mood state dependent learning, although i t evidently influenced manipulation checks.  The  magnitude of the mood differences induced i n Study Four, although s i g n i f i c a n t l y d i f f e r e n t across conditions, may not have been s u f f i c i e n t to drive the memory e f f e c t s .  The complexity of the experimental procedure may  have contributed to t h i s .  Consider, for instance, differences between Study  Two and Study Four.  Study Two employed only the musical induction and  subsequent manipulation checks.  Study Four also employed a r e a l i s t i c business  decision making task, which may have served to d i s t r a c t participants and moderate their attention to the induction.  Comparison of the mean mood scores  of Study Four (Table 20) with those of Study Two (Table 14) (adjusting for the fact that the MAACL scales i n Study Four had half as many items) shows that the former were smaller than the l a t t e r .  For example, the mean scores for  pleasure i n the E l a t i o n and Depression conditions of Study Two were 41.8 and 25.6, respectively, compared to 37.1 and 32.2 i n Study Four.  Comparable  depression scores i n the E l a t i o n and Depression conditions were 17.3 and 33.1 i n Study Two, versus 19.1 and 26.4 i n Study Two.  It should be noted that  there were other differences between Studies Two and Four, such as the between— versus w i t h i n — s u b j e c t s design.  Nevertheless, i t i s possible that  194  the additional elements of Study Four moderated the mood manipulation.  The  additional cognitive requirements of the decision- making task may have served as a d i s t r a c t i o n .  A l t e r n a t i v e l y , the repeated exposure to the mood inductions  procedure, though not to a p a r t i c u l a r treatment condition may have reduced the novelty and hence interest and attention to the musical selections. To test the p o s s i b i l i t y that carry over effects from session to session interfered with the mood manipulation or i t s e f f e c t s , a  between-subjects  analysis was performed on the two sets of manipulation checks and on the set of dependent measures.  The r e s u l t s indicated a s i g n i f i c a n t o v e r a l l effect of  the mood manipulation on the f i r s t set of manipulation checks, F(12,140) = 2.87,  p < .001, n  2  = '36. Examination of the univariate tests showed  s i g n i f i c a n t e f f e c t s for scores on depression, F(l,75) = 10.27, p < .001, and on arousal, F(l,75) = 6.83, p < .002. The means i n the E l a t i o n , Neutral, and Depression conditions were 20.54, 22.73, and 28.27, respectively, f o r depression, and 26.31, 22.69, and 17.81, respectively, for arousal. The o v e r a l l effect of the induction of the second manipulation checks was not s i g n i f i c a n t , F(12,140) = 1.64, p < .088. For the combined dependent variables the multivariate effect was not s i g n i f i c a n t , F(26,126) = .836, p = .694. None of  the univariate tests on the dependent variables were s i g n i f i c a n t .  These  r e s u l t s indicate that the presence of carry-over effects i s not a l i k e l y explanation for f a i l u r e to f i n d an effect of mood i n Study Four.  When  presented with only one induction for a f i r s t time, individuals reported changes i n t h e i r mood as measured before the dependent measures, but d i d not evidence any e f f e c t s on the dependent measures nor on the second set of manipulation checks.  Thus some additional evidence as to the weakness of the  mood manipulation i s also shown. power of the between-subjects  It should be noted that the nominal  relative  analysis, above, and the mixed-between-within  195  analysis i n Study Four i s v i r t u a l l y i d e n t i c a l .  In the within-subjects case,  however, the e f f e c t size might be expected, a p r i o r i , to be larger.  It i s ,  nonetheless, rather low. The preceding sections considered why  the complete experimental procedure  employed i n Study Four f a i l e d to demonstrate an e f f e c t .  Further insight into  the impact of the mood induction might be gained from i n t e r n a l analyses. Analyses, that i s , performed only on those p a r t i c i p a n t s who were, according to their mood self-reports, influenced by the manipulation i n a predicted pattern.  Such analyses have the potential to reveal effects otherwise d i l u t e d  by the f a i l u r e of some p a r t i c i p a n t s to respond to the manipulation.  However,  such selection l i m i t s our a b i l i t y to make causal statements about the effect of the indpendent v a r i a b l e . The responses of the p a r t i c i p a n t s i n Study Four to the mood induction were examined on the basis of their scores on the f i r s t depression, arousal, and pleasure manipulation checks.  These variables are  the best indicators of the success of the mood induction. pattern of scores over the three sessions was examined.  Each p a r t i c i p a n t ' s  Participants whose  scores f i t the pattern predicted by their order treatment were noted.  For  example, a participant whose scores on depression were highest i n the session i n which they recieved the depression induction, lowest after the elation induction, and intermediate following the neutral induction were c l a s s i f i e d as f i t t i n g the predicted pattern for that measure. each of the three manipulation checks.  Each pattern was examined for  Of the 78 participants, 44  evidenced  f i t for the depression measure, 43 for the arousal measure, and 30 for the pleasure measure. participants who  Two  sets of analyses were performed: One  set on the 44  reported the predicted pattern of scores on the depression  measure, and one set on the 18 p a r t i c i p a n t s who a l l three manipulation checks.  f i t the predicted pattern on  These represent the least and most stringently  196  selected i n d i v i d u a l s .  The r e s u l t s of both sets of analyses were similar and  support the r e s u l t s of e a r l i e r analyses.  In each internal analysis the  multivariate test on the f i r s t set of manipulation checks was s i g n i f i c a n t , obviously i n part due to the selection procedure, F(60,345.61) = 3.58, p = .000 f o r N = 41, and F(60,94.12) = 3.08, p = .000 for N = 18. The mean scores for depression i n the Elation, Neutral, and Depression conditions were, respectively, 17.95, 22.19, and 29.68 for N = 41, and 18.00, 22.06, and 32.41 for N = 18. The mean scores for arousal i n the Elation, Neutral, and Depression conditions were, respectively, 33.90, 22.17, and 17.41 f o r N = 41, and 37.23, 23.47, and 16.35 f o r N = 18. The mean scores f o r pleasure i n the E l a t i o n , Neutral, and Depression conditions were, respectively, 38.54, 36.54, and 28.78 f o r N = 41, and 39.88, 34.58, and 22.71 f o r N = 18. Despite the selection, however, for both analyses no multivariate effects of mood on the dependent measures was found, F(130,480) = .959, p = .606 f o r N = 41, and F(130,96) = .984, p = .538 f o r N = 18. For none of the dependent measures was their a s i g n i f i c a n t univariate e f f e c t .  S i m i l a r l y , their were no multivariate  or univariate e f f e c t s of mood on the 15 p r o b a b i l i t i e s that comprise the composite covariation measure.  It i s clear, then, that the mood manipulation  as evident i n the self-reported responses to the manipulation checks, d i d not have an effect on the dependent measures. In l i g h t of a l l of the above, what can be reasonably concluded about the manipulation of mood i n Study Four?  It i s evident that differences i n s e l f -  reported mood greater than chance were found.  That i s , s t a t i s t i c a l l y  s i g n i f i c a n t e f f e c t s of mood on the manipulation checks were present. these differences may not be p r a c t i c a l l y s i g n i f i c a n t or meaningful.  However, In  support of this argument i s the finding of small differences among the mean mood scores.  Countering this argument i s the fact that the s i g n i f i c a n t mood  197  e f f e c t s were not, as we s h a l l discuss more completely i n a l a t e r section, the r e s u l t of excess power.  Further, a review of the size of mood effects found  i n the experimental l i t e r a t u r e reveals studies whose e f f e c t sizes as indicated on mood scales of the sort used here were smaller than found here yet which were s u f f i c i e n t to produce hypothesized e f f e c t s (e.g., Sutherland, Newman, & Rachman, 1982; Teasdale & Russell, 1983).  In sum their i s evidence f o r both  the p o s i t i o n that the mood induction i n Study Four was  sufficent to produce  the hypothesized effects as well as for the position that the effects produced were of l i t t l e meaningful  significance.  Future research might seek to strengthen the mood manipulation, perhaps by adding measures designed to focus i n d i v i d u a l s ' attention.  Study Four  included a step i n this d i r e c t i o n by asking participants to think of the images that the music brought to mind and subsequently asking about those images.  Strengthening the manipulation i n t h i s and other ways might also  increase the spectre of hypothesis guessing by p a r t i c i p a n t s .  Alternatively a  procedure l i k e that used by Eich and Metcalfe (1987) could be employed.  They  required that subjects meet a mood change c r i t e r i o n before proceeding with tests of mood e f f e c t s .  The effectiveness of the manipulation could then be  v e r i f i e d by using unobtrusive measures taken before and a f t e r the dependent measures.  Measurement Error A relationship between mood and expectancy may not have been found because the expectancy measure was not sensitive to r e a l changes i n perceived expectancy.  That i s , there may have been expectancy change which was not  measured. L i t t l e support exists for t h i s explanation.  Study Three showed that  198  scores on the expectancy measure matched objective differences i n e f f o r t performance covariation between tasks. valid.  The expectancy measure was shown to be  Further, Study Four included a verbal, rating scale measure of  p a r t i c i p a n t s ' agreement with statements i n d i c a t i v e of effort-performance covariation.  This measure was also not influenced by mood.  In order to determine whether mood might have had an effect on the i n d i v i d u a l components of the matrices used to compute the covariation measure, analysis of variance was performed on the scores for the 15 c e l l s .  The  analysis was performed on the non-normalized scores ( i . e . , p r i o r to d i v i d i n g them by their t o t a l so that they sum to u n i t y ) .  The o v e r a l l multivariate  effect was not s i g n i f i c a n t , F(150,1108.22) = .886, p = .825. Inspection of the 15 univariate tests revealed no s i g n i f i c a n t e f f e c t s . evident from analyses of the normalized scores.  The same r e s u l t was  Mood d i d not, apparently,  influence judgements of the p r o b a b i l i t y associated with any c e l l of the matrix, just as i t had no influence on the composite measure derived from those p r o b a b i l i t i e s . In sum, then, there i s good evidence from Study Three for the v a l i d i t y of the expectancy dependent v a r i a b l e .  R e c a l l , though, that this conclusion was  q u a l i f i e d by other differences between the conditions i n Study Three, and by the need for r e p l i c a t i o n of Study Three.  To the extent that the perceived  covariation measure of expectancy i s v a l i d , the effect of the mood induction on perceived effort-performance covariation would have been r e f l e c t e d i n the expectancy measure.  Lack of Experimental Control When an experiment succeeds from the researcher's point of view, or hypothesized differences are found to r e l i a b l y exist, then i t i s incumbent on  199  the researcher to consider alternative explanations for the effect or e f f e c t s . Of p a r t i c u l a r concern are confounded  alternative causes, or experimental  procedures that covary with the intended treatment, such that effects could be a t t r i b u t a b l e to either or both.  Experimental procedures that are e s s e n t i a l l y  random, which do not vary systematically with the treatments, are ignored because they serve to i n f l a t e error variance. to f i n d an e f f e c t .  When an experiment  They reduce the study's a b i l i t y  succeeds i t has apparently done so  despite such error, and the error i s of no i n t e r e s t . When an experiment f a i l s , the sources of random error are of i n t e r e s t . Unfortunately, they are by nature random and so are d i f f i c u l t to i d e n t i f y .  In  this study, although three experimenters were used, the gender of the experimenter was controlled.  Care was taken to ensure that experimenter-  participant interaction was minimized and standardized.  I t i s u n l i k e l y that  this contributed substantially to error variance. In the within-subjects, repeated-measures design used, participants completed the decision making task three times.  A d i f f e r e n t decision problem  was used each time but the decision problems were of e s s e n t i a l l y equal difficulty.  The p a r t i c u l a r decision problem used i n each session was  determined randomly to prevent them from being confounded with the mood manipulation or session. Therefore, differences i n the perceptions of the task due to the decision problem would have been included as error variance. However, i t i s unlikely that the p a r t i c u l a r decision problem had an effect on expectancy. In sum, few obvious, uncontrolled sources of random error were present i n the study.  Future research would f i n d l i t t l e upon which to improve.  Lack of S t a t i s t i c a l Power  200  It i s possible that a relationship  exists between mood and  expectancy  which was not detected because the experimental design lacked s t a t i s t i c a l power. null  S t a t i s t i c a l power i s defined as the p r o b a b i l i t y of rejecting a f a l s e  sypotheses. S t a t i s t i c a l power has three codeterminants:  (1) the  significance  c r i t e r i o n , or alpha l e v e l chosen by the researcher as the p r o b a b i l i t y of rejecting a true n u l l hypothesis, (2) the r e l i a b i l i t y of sample r e s u l t s , which i s dependent on the size of a sample, and (3) the effect size or magnitude of the phenomenon i n a population (Cohen, 1977; Mazen, Graf, Kellogg & Hemmasi, 1987).  In other words, the larger alpha i s (e.g., .10 vs.  l i k e l y a f a l s e n u l l hypothesis w i l l be rejected  .05), the more  and the higher the power.  The  larger the sample size i s , the greater the r e l i a b i l i t y , and the higher the p r o b a b i l i t y of rejecting a f a l s e n u l l hypothesis.  The larger the effect size,  the more l i k e l y the effect w i l l manifest i t s e l f and be detected. If these three parameters are known, power can be p r e c i s e l y  determined.  Conversely, the sample size, effect size, and alpha necessary to set power at a certain l e v e l can be determined.  As Mazen et a l .  say, the i d e a l  procedure  would be to a p r i o r i set alpha and the l e v e l of power, estimate effect size, and then solve for the necessary sample size. calculated  Power can, of course, be  after the fact of an experiment.  Most d i f f i c u l t to estimate among the three parameters i s effect size. Often past, well-conceived research from which the proportion of variance explained can be gleaned does not e x i s t .  Cohen has for t h i s reason, provided  conventions for effect size, corresponding to small, medium, and large effects.  For example, for the standardized difference  three values are .20,  .50, and  between two means, the  .80.  The u t i l i t y of power analysis  i s that i t reveals the l i k e l i h o o d of  201  finding an e f f e c t i n a p a r t i c u l a r research contest.  Cohen (1977) recommends  that the p r o b a b i l i t y of f a i l i n g to reject a f a l s e n u l l hypothesis l i k e l i h o o d of Type II error) be set at .20, .80.  (the  and hence that power be set at  Investigations should, according to Cohen, have at least an 80% chance  of r e j e c t i n g a f a l s e n u l l .  Power analysis  research, however (Brewer, 1972; Power analysis  Chase & Chase, 1976;  can also be used to sustain  power exists (say,  .90)  i s not the norm i n present  day  Mazen et a l . , 1987).  the n u l l hypothesis.  If s u f f i c i e n t  to f i n d an effect so small that the researcher i s -  w i l l i n g to declare i t of no meaning, and  t h i s effect i s not found, then i t i s  possible to conclude that the l i k e l i h o o d of not finding an effect where one exists i s s i g n i f i c a n t l y small and can be rejected. however.  thus the hypothesis that an effect  Such a conclusion i s limited to the p a r t i c u l a r  exists  study,  Possibly for t h i s reason, and because accepting the n u l l runs  counter to the logic of f a l s i f i c a t i o n as usually employed, power analysis seldom used i n t h i s way,  is  when i t i s performed at a l l .  Cohen (1977) has published tables which show power for most sample sizes, alpha values, and  effect sizes.  types of significance tests and parameter so that correct research designs. main and  He provides indices of effect size for most formulae which adjust the sample size  estimates of power are made i n the case of complex  In a mixed between-within design, such as Study Four, the  interaction effects have d i f f e r i n g degrees of freedom and hence  d i f f e r i n g power. univariate  The  analysis  index of effect size for significance  tests i n  of variance (he does not consider multivariate ANOVA) i s  T? , the proportion of sample variance explained by an e f f e c t . 2  e f f e c t size corresponding to small, medium and  large for r\ are 2  The  levels of  .01,  .06,  and  .14. In Study Four, s t a t i s t i c a l power was  .94  to detect a large mood (session  202  by order interaction) effect, .50 to detect a medium mood e f f e c t , and .10 to detect a small mood e f f e c t .  These values were determined f o r a 6 by 3,  between by within design, where alpha = .05 and c e l l size i s 13.  Power to  detect a small, medium, and large order effect was .08, .33, and .75, respectively.  For the session effect corresponding power values were .17,  .78, and .99. Mazen and colleagues surveyed the power values i n a sample of the 1984 management l i t e r a t u r e , which we can use to provide a context f o r the power of Study Four.  They found that the average power to detect small e f f e c t s was  .31, although the median power was .25 and the mode .14.  The corresponding  values for medium effects were .77, .89 and .99, and for large effects .91, .99 and .99.  Compared to the management l i t e r a t u r e , then, Study Four had  lower than t y p i c a l power.  It should be noted, though, that only about 10% of  the significance tests included i n Mazen et a l . ' s survey were F-tests.  Most  of the remainder were tests of c o r r e l a t i o n or regression c o e f f i c i e n t s .  It i s  l i k e l y that sample sizes i n the experimental designs surveyed by Mazen et a l . were smaller than those i n other designs, and hence the average power of experimental designs may be lower than the average they reported.  The  s t a t i s t i c a l power of Study Four i s thus l i k e l y closer to the average i n the experimental management research l i t e r a t u r e . Further, surveys of s t a t i s t i c a l power i n other d i s c i p l i n e s report average power lower than that i n the management l i t e r a t u r e .  Cohen (1962) found  average power to detect small, medium and large effects of .18, .48, and .83 among a sample of studies i n the 1960 Journal of Abnormal and Social Psychology.  This may be because other d i s c i p l i n e s , such as psychology, employ  experimental designs proportionately more often. recent studies have higher power —  Or i t may be that more  the survey reported i n the psychological  203  l i t e r a t u r e i s much older. Nevertheless,  the s t a t i s t i c a l power of Study Four was below that t y p i c a l  i n management research and psychology, even though i t was close to that i n the psychological l i t e r a t u r e .  It was not s u f f i c i e n t to permit us to declare that  we had high power to detect a small but meaningful effect and hence that the n u l l hypothesis i s sustained. How  could the power of Study Four be improved?  be increased.  In order to achieve  F i r s t , sample size could  .80 power to detect a small, medium and  large effect of mood, given the same alpha of .05, 136, 23, and 9 subjects would be respectively required. e f f e c t size. variance.  Second, steps could be taken to increase  E f f e c t size i s a function of both mean differences and error  As discussed above, steps were taken to minimize random error.  Differences between the means of treatment groups depend on the treatment e f f e c t and the r e a l r e l a t i o n s h i p between the treatment e f f e c t and the r e a l relationship between the treatment and dependent measures. used i n Study Four was  shown to be v a l i d by Study Two.  The mood induction  A within-subjects  design was chosen to control for factors that might moderate the r e a l r e l a t i o n s h i p between mood and expectancy.  It seems, then, that increasing  e f f e c t size much over Study Four might not be possible. .80, with sample s i z e equal to 13 and alpha =.05 ?7= .101 would be necessary. 2  To achieve power of  an effect size equivalent to  This corresponds to Cohen's convention for a  large effect s i z e . F i n a l l y , power might be increased by relaxing alpha.  The p r o b a b i l i t y of  r e j e c t i n g a true n u l l hypothesis i s conventionally limited to 5%. could be argued that this i s too stringent. research on new  However, i t  E s p e c i a l l y i n the case of  topics, using new measures and variables, a less stringent  c r i t e r i o n might be appropriate to declare that a r e a l , non-chance effect  204  exists at alpha = .10. Unfortunately, meet this c r i t e r i o n .  the findings of Study Four would not  The power of Study Four to detect a small, medium and  large e f f e c t where alpha = .10, and sample size of 13 would be .18, .63, and .97 respectively.  Clearly, relaxing alpha improves power, e s p e c i a l l y for  small and medium e f f e c t sizes, but does not increase i t .80, the l e v e l sought. In summary, lack of s t a t i s t i c a l power i s a plausible explanation for the f a i l u r e of Study Four to f i n d an e f f e c t . increased i n the ways described above. employ a d i f f e r e n t design.  With the same design, power could be A l t e r n a t i v e l y , future research  could  A between subjects design with the same c e l l  sample size of 13 and alpha =.05, would have power of .08, .25, and .57 to detect a small, medium and large d i f f e r e n c e .  A between subjects design with  the same number of participants as Study Four, namely 78, with 26 i n each treatment would have corresponding power of .11, .48, and .89. F i n a l l y , a between subjects design with the same number of observations," namely 234,  with  78 subjects i n each treatment, would have respective power of .26, .94, and more than .995.  The Null Hypothesis i s True It i s possible that mood does a f f e c t expectancy judgements but that the conditions of Study Four d i d not f a c i l i t a t e demonstrating such a r e l a t i o n s h i p . For example i t might be that expectancy was measured too soon a f t e r task performance.  If we understand mood to influence subjective perceptions  i t follows that i t w i l l emerge when objective influences are weaker.  then  When  r e c a l l i s for a task farther i n the past, mood may have an influence on memories of performance and hence expectancy judgements.  I t might also be the  case that the l e v e l of objective expectancy was too high i n the Marketing Game.  That i s , studies of the influence of mood on contingency judgements  205  reveal that mood matters when objective contingency i s low. It may be that the objective e f f o r t performance r e l a t i o n s h i p i n Study Four was not low enough.  When no r e l a t i o n s h i p exists then individuals i n a p o s i t i v e mood may  perceive one. Explanations  l i k e the preceding,  which speculate as to the conditions  under which an e f f e c t might be found, must be considered study of organizational motivation.  i n the context of a  That i s , although we might be able to  f i n d an e f f e c t , i n doing so we are moving farther away from the conditions found i n organizations.  We can remove expectancy measurement temporally  performance of an experimental task. perceptions  from  In organizations, however, expectancy  a r i s e i n the midst of ongoing work.  experimental task that has no expectancy.  We can produce an  However, work tasks i n  organizations do have expectancy. The f i n a l explanation for the f a i l u r e of Study Four to show a r e l a t i o n s h i p between mood and expectancy i s that the n u l l hypothesis i s true, that no r e l a t i o n s h i p e x i s t s .  The cognitive processes involved i n the  formation of expectancy percepts may be resistant to mood e f f e c t s .  It may be  that estimates of the r e l a t i o n s h i p between e f f o r t and performance are robust to changes i n mood state.  People experience success and f a i l u r e i n tasks,  they do not abandon tasks because f a i l u r e s cause them to believe that no success i s possible.  In t h i s way i t i s possible that expectancy i s robust to  changes i n performance.  If so, i t i s u n l i k e l y that transient mood states can  influence expectancy judgements. Related to this i s the p o s s i b i l i t y that the process of forming expectancy judgements requires so much attention and cognitive e f f o r t that mood i s destroyed  or severely diminished  when the questions  covariation variable are presented.  that comprise the  Studies which have used t r i v i a l  206  experimental tasks may have been able to show an e f f e c t because of that triviality.  When a person i s bored at a task their mood might be e a s i l y  manipulated and might influence their task perceptions.  But when involved i n  a more i n t e r e s t i n g experimental task t h e i r mood and perceptions might be immune to manipulation.  I t may be an irony that more i n t e r e s t i n g , emotionally  involving tasks are thus less influenced by previously induced mood. The t h e o r e t i c a l chain which we described i n Chapter Four i s long. be too long.  It may  Recall that we postulated that i f mood influences r e c a l l of  p o s i t i v e and negative events, and r e c a l l influences p r o b a b i l i t y judgements, and p r o b a b i l i t y judgements combine to form expectancy, then mood should influence expectancy. connection  In a chain l i k e t h i s each link i s required for the  between mood and expectancy to be made.  Any weakness i n  r e l a t i o n s h i p between parts of the chain, as a r e s u l t of t r u l y weak or limited relationships, or of experimental imprecision, threatens the v a l i d i t y of the model and our a b i l i t y to support i t experimentally.  Conclusion If Study Four had shown a s i g n i f i c a n t relationship between mood and expectancy, then we would be discussing the implications of such a r e l a t i o n s h i p for organizations.  We would also be discussing threats to the  i n t e r n a l v a l i d i t y of Study Four and implications for future research. Instead, we are l e f t with uncertainty as to whether an e f f e c t might be detected i n another study, whether mood influences expectancy under only very specialized conditions which might not e a s i l y generalize to organizational settings, or whether mood does not a f f e c t expectancy at a l l . So what does this d i s s e r t a t i o n mean for the study of emotions i n organizations and for the study of expectancy?  F i r s t , i t has taken a  207  theoretical framework from psychology and applied i t to organizational behavior.  That i s , the network model of how mood influences memory and  judgement has been used to make predictions about effects on variables of interest to organizational researchers.  The lack of success i n the domain  chosen does not negate the potential a p p l i c a b i l i t y i n other domains.  Future  research could apply similar theoretical frameworks to other variables, such as performance appraisal, employee selection, risk-taking, organizational decision-making, or prosocial organizational behavior. Second, i t has shown that mood can be successfully manipulated i n an experimental context applicable to studying organizations and that mood can be v a l i d l y measured i n such contexts.  Similar experimental procedures, or  procedures using d i f f e r e n t mood manipulations can be applied to future research questions.  The measures of mood used can be applied i n experimental  settings l i k e those used here, or to non-experimental  research designs  including questionnaire research. Third, t h i s d i s s e r t a t i o n has provided a rationale for the measurement of expectancy i n a manner consistent with Vroom's conceptualization, and has validated such measurement.  In doing so i t has examined the relationship  between objective task expectancy and task d i f f i c u l t y and shown them to be unrelated contrary to much of the l i t e r a t u r e .  It has also argued that  multiple e f f o r t and performance l e v e l measurement of expectancy allows comparison of expectancies to be made across i n d i v i d u a l s .  Expectancy  i s shown to not be limited only to "within-subjects" predictions.  theory  The  question of how judgements of the contingency between e f f o r t and performance are formed are raised a l s o .  That i s , what i s i t about tasks that causes  individuals to perceive high or low expectancy?  While much research i s  proceeding on this topic i n psychology, l i t t l e attention i s being directed to  208  i t i n organizational behavior, although i t has d i r e c t relevance to job design. Conceptualizing expectancy as covariation and measuring i t accordingly opens a number of d i r e c t i o n s f o r expectancy research. F i n a l l y , although i t showed only that mood could be manipulated Study Four provided a basis for future investigations of mood and expectancy.  A  number of possible reasons f o r the f a i l u r e to f i n d an effect were explored and a l t e r n a t i v e approaches considered.  On t h i s foundation i t may yet be  demonstrated that mood does unequivocally a f f e c t expectancy or, unequivocally, that i t does not.  210  PART ONE Remember that the information c o l l e c t e d here w i l l be c o n f i d e n t i a l and used only for the purposes of the study. Age: Sex: Please read each of the following statements. Where there i s a blank , decide what your normal or usual attitude, f e e l i n g or behavior would be: 1 RARELY  2 OCCASIONALLY  3 SOMETIMES  4 FREQUENTLY  5 USUALLY  Write the number that describes your usual attitude or behavior i n the blank. For example: A.  I  t e l l others that Vancouver i s a f i n e c i t y .  If you RARELY t e l l others t h i s , write a 1 i n the blank. If you OCCASIONALLY t e l l others t h i s , write a 2 i n the blank, and so on. Answer a l l the items. If you have d i f f i c u l t y with one do not leave i t blank, answer as best you can. 1.  When faced with a problem I  2.  I  3.  I need frequent encouragement from others for me to keep working at a d i f f i c u l t task.  4.  I f e e l that I'm a person of worth, at least on an equal basis with others.  5.  I own work.  6.  I have received too much change from a cashier and  throw my l i t t e r  t r y to forget i t .  into waste paper baskets on the street.  l i k e jobs where I can make decisions and be responsible f o r my not said  anything. 7.  I  8.  When I hear people t a l k i n g p r i v a t e l y I  9.  If I want something I  10.  I  change my opinion when someone I admire disagrees with me. avoid l i s t e n i n g .  work hard to get i t .  f e e l that I have a number of good q u a l i t i e s .  211  1 RARELY  11.  I  2 OCCASIONALLY  3 SOMETIMES  4 FREQUENTLY  5 USUALLY  prefer to learn the facts about something from someone else  rather than have to d i g them out for myself. 12.  I have  taken things that didn't belong to  me.  13.  I will  accept jobs that require me to supervise others.  14.  I  t e l l l i e s i f I have to.  15.  I  have a hard time saying "no" when somone t r i e s to s e l l  me  something I don't want. 16.  A l l i n a l l , I am  17.  I  18.  I keep my promises, no matter how inconvenient i t might be to do so. I consider the d i f f e r e n t sides of an issue before making any decisions. I take a sick-leave from work or school even though I wasn't  19. 20.  really  i n c l i n e d to think that I am a f a i l u r e .  l i k e to have a say i n any decisions made by any group I'm i n .  sick.  21.  What other people think  has a great influence on my  22.  I am  23.  Whenever something good happens to me I  able to do things as well as most other  behavior.  people.  f e e l i t i s because I've  earned i t . 24.  I  l i k e to gossip about other people's business.  25.  I  enjoy being i n a p o s i t i o n of leadership.  212  RARELY 26. 2 7  •  I have 1  2 8  •  1  2 9  •  1  3 0  •  OCCASIONALLY  FREQUENTLY  USUALLY  done things that I don't t e l l other people about.  someone else to p r a i s e my work before I am s a t i s f i e d with what I've done. n  e  e  d  f e e l I do not have much to be proud of. am  1  31.  SOMETIMES  sure enough of my opinions to t r y and influence others. s a  y only good things about my friends behind their backs.  When something i s going to a f f e c t me I  learn as much about i t as  I can. 3 2  •  1  put o f f u n t i l tomorrow what I should do today.  3 3  •  1  decide to do things on the spur of the moment.  • 35. 3 4  1  take a p o s i t i v e a t t i t u d e toward myself. For me, knowing I've done something well i s i s  more important  than being praised by someone else. 3 6  •  1  3 7  •  1  declare everything at customs. l e t  other peoples' demands keep me from doing things I want to  do. 3 8  •  1  think I have some pretty awful habits.  3 9  •  1  s t i c k to my opinions when someone disagrees with  40.  On the whole, I am  s a t i s f i e d with myself.  me.  213  1 RARELY  2 OCCASIONALLY  3 SOMETIMES  4 FREQUENTLY  5 USUALLY  41.  I • to do.  do what I f e e l l i k e doing not what other people think I ought  42.  I  t e l l the truth.  43.  I  get discouraged when doing something that takes a long time to  achieve r e s u l t s . 44.  I am _ _ _ _ _  late f o r appointments.  45.  When part of a group I  prefer to l e t other people make a l l the  decisions. 46.  I  wish I could have more respect for myself.  47.  When I have a problem I  follow the advice of friends or  relatives. 48.  I  obey t r a f f i c laws even i f I'm u n l i k e l y to get caught.  49.  I  enjoy trying to do d i f f i c u l t tasks more than I enjoy trying to  do easy tasks. 50.  When I was a c h i l d I  51.  I  obeyed my parents.  prefer situations where I can depend on someone else's a b i l i t y  rather than just my own. 52.  I  53.  Having someone important t e l l me I d i d a good job i s more important than f e e l i n g I've done a good job. When I'm involved i n something I t r y to f i n d out a l l I can about  54.  f e e l useless at times.  what i s going on even when someone else i s i n charge. 55.  I am  56.  I  57.  I have  p o l i t e to others including my friends and family. think I am no good at a l l . cheated on a test or assignment  i n any way.  214  PART TWO The task we would l i k e you to complete contains aspects of a proofreading or q u a l i t y control task. In q u a l i t y control a product must be matched against a standard. In the task you w i l l complete you w i l l be shown rows of numbers. Your task i s to check the number at the l e f t of each row, and then c i r c l e each number i n the row that matches the number at the l e f t of the row. For example:  3  4  6  4  6  9  I  3  9  0  4  4  9  9  I  9  2  I  I  7  9  4  3  0  5  2  6  7  6  2 6  6 2  4 5  Please work c a r e f u l l y , only c o r r e c t l y completed rows count. following rows yourself:  I 9  8 A  0  Try the  0  9  7  1  5  0  0  6  4  5  6  8  1  6  0  7  8  k  k  6  9  6  1  8  7  1  1  8  7  9  1  7  7  7  3  1  5  1  2  6  7  0  8  7  9  9  5  7  k  2  6  0  7  3  8  0  8  7  6  6  5  5  8  7  6  6  3  0  0  0  7  2  5  6  9  8  8  k  9  7  5  6  1  7  0  8  6  3  2  k  8  8  0  7  2  2  2  2  3  6  0  8  6  8  k  6  3  7  9  3  1  6  7  6  0  3  8  6  5  8  5  5  7  1  9  1  9  8  8  0  0  6  9  0  1  1  6  5  7  9  5  9  5  8  7  6  5  5  2  9  3  7  2  2  2  0  2  A  2  5  8  2  0  k  2  2  0  6  3  8  1  8  1  7  2  6  2  0  2  7  6  2  7  3  7  6  8  7  6  6  4  3  9  7  9  1  9  0  0  1  k  6  9  2  Next you w i l l be asked to work at this task f o r a series of 10 work periods. The periods w i l l be a minute long on average, although some may be longer than others. Do your best during each of these periods but remember to work c a r e f u l l y . Only c o r r e c t l y completed roes count. For each period, you w i l l be t o l d when to begin and when to stop working. At the end of each period, draw a l i n e under the l a s t l i n e you completed and count the number of rows you completed during that period. Write the number of rows you completed i n the right margin and wait u n t i l you are instructed to begin again. When you are t o l d to begin working again turn to the next page and start immediately at the top. Any questions?  215  PART THREE For the next few questions we would l i k e you to estimate the p r o b a b i l i t y of c e r t a i n outcomes. We would l i k e you to answer i n percentages and make sure that your answers add up to 100%. For example, i f asked to estimate the chances of i t raining i n Vancouver i n the next month on less than 7 between 7 and 13 between 14 and 20 on more than 21  days?: days?: days?: days?:  You might answer i n the following  way:  on less than 7 days?: between  7  and 13  days?:  between 14  and 20  days?:  on  more than 21 days?:  Note that the answers add up to 100%.  Any questions?  We would l i k e you to think about the proofreading task. We are not interested i n your actual l e v e l of e f f o r t or your actual performance. Instead we would l i k e you to think about what would happen i f you were to repeat the simulation under similar circumstances. If you were to complete 10 more periods, and you were to expend a HIGH degree of e f f o r t , what are the chances that you would complete on average each of the following numbers of rows! between  1  and 10  rows?:  between 11  and 20  rows?:  between 21  and 30  rows?:  between 31  and 40  rows?:  more than 40  rows?: 100%  216  If you were to complete 10 more periods, and you were to expend a MEDIUM degree of e f f o r t , what are the chances that you would complete on average each of the following numbers of rows:  between  1  and 10  rows?:  between 11  and 20  rows?:  between 21  and 30  rows?:  between 31  and 40  rows?:  more than 40  rows?: 100%  If you were to complete 10 more periods, and you were to expend a LOW degree of e f f o r t , what are the chances that you would complete on average each of the following numbers of rows:  between  1  and 10  rows?:  between 11  and 20  rows?:  between 21  and 30  rows?:  between 31  and 40  rows?:  more than 40  rows?: 100%  217  On the following scale, indicate your judgement of the amount of control you had over your performance, at 100 i f you had complete control and at 0 i f you had no c o n t r o l . Complete control means that the number of rows you complete i s determined by how hard you t r y . No control means that how hard you t r y or don't t r y has nothing to do with your performance. Another way to look at having no control i s that the number of rows completed i n any period i s t o t a l l y determined by factors such as chance or luck, rather than by the e f f o r t you expended. Intermediate control means that your e f f o r t has some influence but does not completely determine the number of rows you complete. C i r c l e the number that best represents your 0  :  i0  :  20  :  30  :  40  :  50  :  60  judgement: :  70  :  80  :  90  :  100  If you were to complete 10 more periods, for how many of those periods would you complete on average each of the following numbers of rows: Note that your answer should be i n numbers of periods and should add up to 10. between  1  and 10  rows?:  between 11  and 20  rows?:  between 21  and 30  rows?:  between 31  and 40  rows?:  more than 40  rows?: 10  218  For the following descriptions please use the folowing scale to indicate how a t t r a c t i v e each i s to you. C i r c l e the number that best describes your f e e l i n g . For example:  A l l things considered, how a t t r a c t i v e would i t be to you  ...  ...to take part i n outdoor a c t i v i t i e s . 1 Very Unattractive  2  3  4 Neutral  5  6  7 Very Attractive  If this i s Very Unattractive to you, c i r c l e the 1. If this i s Somewhat Unattractive to you, c i r c l e the 2, and so on. Any questions?  Please begin.  A l l things considered, how a t t r a c t i v e would i t be to you  ...  ...to expend a HIGH l e v e l of e f f o r t on the proofreading task? 1 Very Unattractive  2  3  4 Neutral  5  6  7 Very Attractive  ...to complete between 1 and 10 rows on the proofreading task? 1 Very Unattractive  2  3  4 Neutral  5  6  7 Very Attractive  ...to complete between 31 and 40 rows on the proofreading task? 1 Very Unattractive  2  3  4 Neutral  5  6  7 Very Attractive  219  Using the following scale, c i r c l e the number that best represents the relationship between working hard on the proofreading task and performing well. For example, i f you think there i s no relationship between working hard on the proofreading task and performing well, c i r c l e the 0. If you think there i s a very strong relationship, c i r c l e the 9. If you think the r e l a t i o n s h i p i s somewhere i n between, c i r c l e the appropriate number.  0 1 No Relationship  2  3  4  5  6  7  8  9 Strong Relationship  As best you can r e c a l l , how many rows d i d you complete i n each period of the proofreading task? How many rows did you complete i n the  FIRST period?:  How many rows did you complete i n the  SECOND period?:  How many rows did you complete in the  THIRD period?:  How many rows did you complete i n the  FOURTH period?:  How many rows did you complete in the  FIFTH period?:  How many rows did you complete in the  SIXTH period?:  How many rows did you complete i n the SEVENTH period?: How many rows did you complete in the  EIGHTH period?:  How many rows did you complete i n the  NINTH period?:  How many rows did you complete i n the  TENTH period?:  220  For the next set of questions, we are interested i n how much you agree with the statements that follow. Using the following scale as a guide, write a number beside each statement to indicate how much you agree with i t . 1 Strongly Disagree 1. . 2.  2 Disagree  3 Neutral  4 Agree  5 Strongly Agree  I f e e l a great sense of personal s a t i s f a c t i o n when I do w e l l . I expended a high l e v e l of e f f o r t on the proofreading task.  3.  The proofreading task was not very i n t e r e s t i n g .  4.  I found the proofreading task d i f f i c u l t .  5.  Generally speaking, I am unsatisfied with my performance on the proofreading task.  6.  The proofreading task was challenging.  7.  Compared to other people, I don't think I d i d very well on the proofreading task.  8.  My opinion of myself goes up when I do well.  9.  Overall, I didn't t r y very hard on the proofreading task.  10.  I enjoyed working on the proofreading task.  11.  My own feelings generally are not affected much one way or another by how well I d i d on the proofreading task. A l l i n a l l , I am very s a t i s f i e d with my performance on the  12.  proofreading task. 13.  I didn't f i n d the proofreading task very challenging.  14.  My performance on the proofreading task was high.  15.  I f e e l bad and unhappy when I've done poorly.  16.  I found the proofreading task to be easy.  221  The next set of questions ask you to consider the reasons behind your actual performance on the proofreading task. For each scale, c i r c l e the number that best describes your impression or opinion of the cause of your performance. Is the cause of your performance something that: Reflects an aspect of yourself  9  8  7  6  5  4  3  2  1  Reflects an aspect of the situation  5  4  3  2  1  Uncontrollable by you or other people  Is the cause of your performance: Controllable by 9 you or other people  8  7  6  Is the cause of your performance something that i s : Permanent  9  8  7  6  5  4  3  2  1  Temporary  2  1  Unintended by you or other people  Is the cause of your performance something: Intended by you or other people  9  8  7  6  5  4  3  Is the cause of your performance something that i s : Outside of you  9  8  7  6  5  4  3  2  1  Inside of you  Is the cause of your performance something that i s : Variable over time  9  8  7  6  5  4  3  2  1  Stable over time  5  4  3  2  1  Something about others  Is the cause of your performance: Something about you  9  8  7  6  Is the cause of your performance something that i s : Changeable  9  8  7  6  5  4  3  2  1  Unchanging  Is the cause of your performance something for which: No one i s 9 responsible  8  7  6  5  4  3  2  1  Someone i s responsible  222  indicate no» well J ^ese f - o r d f ^ f f quickly, not spend a l o t o f t S e on S e ^ o r l o £  d  6 7 0  o  definitely do not f e e l  1.  active  _  2.  afraid  _  3.  agreeable  _  4.  alive  _  5.  alone  6.  amiable  7.  angry  8.  awful  9.  blue  10.  calm  11.  cooperative  12.  cruel  13.  devoted  2 do not feel  slightly feel  _ 17.  fine  _ 18.  forlorn  _ 19.  frightened  _ 20.  gloomy  _ 21. happy _ 22. healthy  9 3 0 1 1 o £  " ^  t h e  o r d s  "  b e l m  ° » to ' ° * W  definitely feel  _ 33.  polite  _ 34.  rejected  _ 35.  shaky  _ 36.  suffering  _ 37.  sunk  _ 38.  sympathetic tender  23.  hopeless  _ 39.  24.  kindly  . 40. tense  25.  lonely  . 41. t e r r i b l e  26.  lost  27.  low  28.  mad  29.  merry  14.  disagreeable  30.  miserable  15.  discouraged  31.  nervous  16.  fearful  32.  panicky  r  . 42. tormented . 3. 4  understanding  44.  unhappy  45.  upset  46.  warm  47.  wilted  48.  worrying  223  For the following pairs of words, please indicate the point on the scale that describes your current feelings. For each pair c i r c l e the number on the scale that best describes your feelings right now. 1 Unhappy  2  3  4  5  6  7  8  9 Happy  1 Relaxed  2  3  4  5  6  7  8  9 Stimulated  1 Pleased  2  3  4  5  6  7  8  9 Annoyed  1 Excited  2  3  4  5  6  7  8  9 Cain  1 Unsatisfied  2  3  4  5  6  7  8  9 Satisfied  1 Sluggish  2  3  4  5  6  7  8  9 Frenzied  1 Contented  2  3  4  5  6  7  8  9 Melancholic  1 Jittery  2  3  4  5  6  7  8  9 Dull  1 Despairing  2  3  4  5  6  7  8  9 Hopeful  1 Sleepy  2  3  4  5  6  7  8  9 Wide Awake  1 Relaxed  2  3  4  5  6  7  8  9 Bored  1 Aroused  2  3  4  5  6  7  8  9 Unaroused  That's i t . Thanks a l o t .  224  APPENDIX B: Musical Selections Elated Tape: (20 minutes and 11 seconds) (1)  "Intermezzo", Leopold Stokowski conducts the National Philharmonic Orchestra, Great performances Carmen and L'Arlesienne Suites, CBS, MY 37260. (time: 2:58)  (2)  "An American in Paris" (Gershwin), Leonard Bernstein conducts the New York Philharmonic Symphony, Columbia Records, M 31804. ( f i r s t 3:14)  (3)  "Ode to Joy" (Schiller), Karl Bohm conducts the Weiner Philharmoniker, Beethoven's Symphonie No. _9, Duetsche Grammophon, 2707073, started recording 3:37 into piece, recorded for 2:23.  (4)  "Guadalcanal March" (Rodgers), Robert Bennett conducts, Victory at sea, RCA, VCS-7064. (time: 2:58)  (5)  "Le Basque", (Galway), Annie's song and other Galway favorites, ARL1-3061. (time: 1:50)  (6)  "Les Torreadors", (Bizet), Carmen Suite, Mercury, MG 50374. 2:14)  (7)  "Overture", (Conti), Rocky II, United A r t i s t s , LA 972-1. (omit f i r s t 0:18, record 1:36, then omit u n t i l break at 4:41, resume recording for 1:42).  RCA,  (time:  Depressed Tape: (19 minutes and 39 seconds) (1)  "Intermezzo" (same as for elated tape)  (2)  "Egmont Overture", (Beethoven), Josef Krips conducts the London Symphony Orchestra, Everest, 3119. ( f i r s t 1:16)  (3)  "A Song to the Evening Star", (Wagner), Young Listener's Library, ( L i l l i a n Baldwin, ed.), Sound Book Press Society, Inc., MSB 33103B. (time: 3:15).  (4)  "Overture-Fantasy" from Romeo and J u l i e t , (Tchaikovsky), Scheherzade rhapsodic mood music, Charles Gerhardt conducts, RCA. ( f i r s t 2:26)  (5)  "Introduction" from Scottish Fantasy, (Bruch) Op. 46, S i r Malcolm Sargent conducts the New Symphony Orchestra of London featuring Heiffetz as v i o l i n i s t , RCA, LSC-2603. ( f i r s t 2:28)  (6)  "Sonata No. 7 i n D major", Op. 10, No. No.3, second movement, Beethoven's "Piano Sonatas" (Vol. 3), Orpheus, B 118. ( f i r s t 2:37)  (7>  "Marche Funebre", Sonata No.2 i n Bb minor. Op. 35, (Chopin), (50th anniversary complete ed.), Westminister, XWN 18882. ( f i r s t 2:15)  225  (8)  "Symphony No.6 i n Bm", (Pathetique), Op. 74, 4th movement, Otto Klemperer conducts the Philharmonica Orchestra, Angel, 35787. (last 2:09)  Neutral Tape: (19 minutes and 36 seconds) (1)  "Intermezzo" same as for the elated and depressed  tapes  (2)  "Canon i n D major", (Pachelbel), Jean-Francois Paillard conducts the Jean-Francois Chamber Orchestra, Musical Heritage Society, Inc., 1060. ( f i r s t 4:08)  (3)  "Symphonic Variations for Piano and Orchestra", (Franck), Massimo Freccia conducts, RCA. ( f i r s t 3:10)  (4)  "Othello Overture", (Dworak), Op. 93, Istvan Kertesz conducts the London Symphony Orchestra, London, CS 6527. (omit f i r s t 3:20, record 2:53)  (5)  "Les Parfums de l a Nuit", (DeBussy), Iberia, Lorin Maazel conducts the Cleveland Orchestra, London, CS 7128. (time: 3:48)  (6)  "The Homecoming", (Hardy), courtesy of WAJY-FM.  (time:  2:28)  226  PART TWO The task we would l i k e you to complete contains aspects of a proofreading or quality control task. In quality control a product must be matched against a standard. In the task you w i l l complete you w i l l be shown rows of numbers. Your task i s to check the number at the l e f t of each row, and then c i r c l e each number i n the row that matches the number at the l e f t of the row. For example:  5  3  2  4  6  9  3  9  0  4  4  9  9  1  9  2  1  2  6  4  1  8  4  1  1  7  9  4  3  0  5  2  6  7  6  6  2  5  9  4  0  Please work c a r e f u l l y , only correctly completed rows count, following rows yourself:  Try the  4  3  9  9  7  2  2  2  2  0  9  7  1  5  0  0  6  4  5  6  8  7  9  1  4  0  2  4  2  4  1  6  0  7  8  4  4  6  9  6  1  8  9  0  0  5  8  2  0  4  7  1  1  8  7  9  1  7  7  7  3  4  1  1  4  2  2  0  6  3  5  1  2  6  7  4  0  8  7  9  9  5  4  7  6  8  1  8  1  7  4  2  6  0  7  4  3  8  0  8  7  6  6  5  5  2  6  2  0  2  8  7  6  6  3  0  0  0  7  2  5  6  9  8  8  4  7  6  2  7  9  7  5  6  1  7  0  8  6  3  2  4  8  8  0  7  2  3  7  6  2  2  2  3  6  0  8  6  8  4  6  3  7  9  3  1  6  7  8  7  6  0  3  8  6  5  8  5  5  7  7  9  1  9  8  8  0  0  6  6  6  9  0  1  1  6  5  7  9  5  9  5  8  7  6  5  5  2  9  3  Next you w i l l be asked to work at this task for a series of 1 0 work periods. The periods w i l l be a minute long on average, although some may be longer than others. Do your best during each of these periods but remember to work c a r e f u l l y . Only correctly completed roes count. For each period, you w i l l be told when to begin and when to stop working. At the end of each period, draw a l i n e under the last l i n e you completed and count the number of rows you completed during that period. Write the number of rows you completed i n the right margin and wait u n t i l you are instructed to begin again. When you are told to begin working again turn to the next page and start immediately at the top. Any questions?  227  THE MARKETING GAME: Session One Practice You w i l l play the r o l e of a newly h i r e d brand manager f o r N a t i o n a l Foods. Your job i s t o decide how t o spend money on promoting your company's product i n your r e g i o n . You w i l l be given a promotional budget and have t o decide how best to d i v i d e that budget between the three markets i n your r e g i o n . The more of the promotional budget you spend i n a market the more p r o f i t you w i l earn i n that market. But i n some markets the same amount o f promotion earns more p r o f i t . Your goal i s t o earn as much p r o f i t i n your r e g i o n as you can over the next ten p e r i o d s . So you must d e c i d e f o r each o f these p e r i o d s , how to a l l o c a t e or d i v i d e your budget between the three markets t o earn as much p r o f i t as p o s s i b l e . The way you w i l l do t h i s i s by making your a l l o c a t i o n f o r one p e r i o d , then you can look a t the r e s u l t s of that a l l o c a t i o n b e f o r e you go on t o make the a l l o c a t i o n f o r the next p e r i o d . To h e l p you get s t a r t e d i n your new job, you w i l l be shown the " h i s t o r y r e p o r t " of the p r e v i o u s manager. T h i s r e p o r t w i l l look l i k e this:  H 1 S T 0 R Y D I S P L Jl Y A L L F I G U R E S IN PERIOD:  1  2  3  4  $ 000'S  5  PROMOTION MARKET 1: MARKET 2: MARKET 3:  7 7 7  10 10 10  8 8 8  11 11 11  9 9 9  PROFIT MARKET 1: MARKET 2: MARKET 3:  221 97 443  315 103 634  244 99 494  332 102 686  279 105 563  837 1120  947  TOTAL PROFIT 761 1052 HIT RETURN TO CONTINUE  T h i s shows that i n p e r i o d One the p r e v i o u s manager d i v i d e d a budget o f $21 ( a l l f i g u r e s a r e i n thousands) evenly between Market 1, Market 2, and Market 3. T h i s r e s u l t e d i n a p r o f i t o f $221 i n Market 1, $97 i n Market 2, and $443 i n Market 3 f o r a t o t a l o f $761. In p e r i o d 2 the p r e v i o u s manager d i v i d e d a budget o f $30 e v e n l y among t h e three markets.  228  The previous manager had a d i f f e r e n t budget i n each p e r i o d . YOUR budget w i l l be EQUAL f o r each p e r i o d . During t h i s p r a c t i c e s e s s i o n you w i l l have $54 t o a l l o c a t e each p e r i o d t o the three markets i n your r e g i o n . You must spend a l l o f t h i s budget each p e r i o d . The p r a c t i c e s e s s i o n w i l l be THREE p e r i o d s long. Your performance w i l l be evaluated on t h e b a s i s o f t h e t o t a l p r o f i t you earn over t h e next t h r e e p e r i o d s . During t h i s p r a c t i c e s e s s i o n and t h e r e a l s e s s i o n l a t e r , the computer w i l l prompt you each p e r i o d . I t w i l l show you t h e h i s t o r y r e p o r t , i n c l u d i n g t h e d e c i s i o n s and r e s u l t s o f the previous manager, then i t w i l l ask you t o make a decision. The prompt w i l l look l i k e t h i s : YOUR BUDGET IS: $ 54 WHAT IS YOUR ALLOCATION TO MARKET 1, MARKET 2, MARKET 3 FOR PERIOD: ?  6  Enter your d e c i s i o n by e n t e r i n g t h r e e numbers on t h e same l i n e , separated by spaces. F o r example: ?20 14 20  The computer w i l l then show you t h e updated h i s t o r y r e p o r t w i t h the p r o f i t r e s u l t s o f your d e c i s i o n and go on t o the next p e r i o d . A recent market survey, commissioned by N a t i o n a l Foods, has determined t h a t a l t h o u g h other market f a c t o r s a l s o i n f l u e n c e p r o f i t , t h e s i n g l e most important d e c i s i o n t h a t you can make i s how t o d i v i d e y o u r p r o m o t i o n a l budget. T h i s survey has a l s o determined t h a t although o t h e r f a c t o r s may i n f l u e n c e p r o f i t w i t h i n a p e r i o d , t h e periods a r e independent. There i s no c a r r y o v e r o f promotion from one p e r i o d t o t h e next. Any  questions?  230  Remember that t h e i n f o r m a t i o n c o l l e c t e d here w i l l be c o n f i d e n t i a l and used o n l y f o r t h e purposes o f t h e study. Age: Sex: Student Number: ( f o r i d e n t i f i c a t i o n purposes o n l y ) Please read each o f t h e f o l l o w i n g statements. For each statement i n d i c a t e the degree t o which you agree w i t h t h e statement, u s i n g t h e f o l l o w i n g s c a l e : 1 STRONGLY DISAGREE  2 MODERATELY DISAGREE  3 SLIGHTLY DISAGREE  4 NEUTRAL  5 SLIGHTLY AGREE  6 MODERATELY AGREE  7 STRONGLY AGREE  W r i t e t h e number t h a t d e s c r i b e s how much you agree w i t h t h e statement i n t h e b l a n k . For example: A.  Vancouver i s t h e f i n e s t c i t y i n North America. I f you STRONGLY AGREE w i t h t h i s , w r i t e a 1 i n t h e b l a n k . I f you MODERATELY AGREE w i t h t h i s , w r i t e a 2 i n t h e blank, and so on.  Answer a l l t h e i t e m s . answer as best you c a n .  I f you have d i f f i c u l t y w i t h one do n o t l e a v e i t blank,  1.  When I g e t what I want i t ' s u s u a l l y because I worked hard f o r i t .  2.  I always throw my l i t t e r i n t o waste paper baskets on t h e s t r e e t .  3.  Even when I'm f e e l i n g s e l f c o n f i d e n t about most t h i n g s , I s t i l l seem to l a c k t h e a b i l i t y t o c o n t r o l i n t e r p e r s o n a l s i t u a t i o n s .  4.  I have r e c e i v e d t o o much change from a c a s h i e r and n o t s a i d anything.  5.  By t a k i n g an a c t i v e p a r t i n p o l i t i c a l and s o c i a l a f f a i r s we, t h e people, can c o n t r o l w o r l d events.  6.  I f e e l t h a t I'm a person o f worth, a t l e a s t on an e q u a l b a s i s w i t h others.  7.  When I make p l a n s I am almost c e r t a i n t o make them work.  8.  When I hear people t a l k i n g p r i v a t e l y I a v o i d l i s t e n i n g .  9.  I have no t r o u b l e making and keeping f r i e n d s .  10.  I have taken t h i n g s t h a t d i d n ' t belong  t o me.  231  1 STRONGLY DISAGREE  2 MODERATELY DISAGREE  3 SLIGHTLY DISAGREE  4 NEUTRAL  5 SLIGHTLY AGREE  6 MODERATELY AGREE  7 STRONGLY AGREE  11.  The average c i t i z e n can have an influence on government decisions.  12.  I feel that I have a number of good q u a l i t i e s .  13.  I prefer games involving some luck over games requiring pure s k i l l .  14.  I sometimes t e l l l i e s i f I have to.  15.  I'm not good at guiding the course of a conversation with several others.  16.  I always keep my promises, no matter how inconvenient i t might be to do so.  17.  It i s d i f f i c u l t for people to have much control over the things politicians do i n o f f i c e .  18.  A l l i n a l l , I am inclined to feel that I am a f a i l u r e .  19.  I can learn almost anything i f I set my mind to i t .  20.  I have taken a sick-leave from work or school even though I wasn't really sick.  21.  I can usually establish a close personal relationship with someone I find sexually attractive.  22.  I like to gossip about other people's business.  23.  This world i s run by the few people i n power and there i s not much the l i t t l e guy can do about i t .  24.  I am able to do things as well as most people.  25.  My major accomplishments are entirely due to hard work and intelligence.  26.  I have done things that I don't t e l l other people about.  27.  When being interviewed I can usually steer the interviewer toward the topics I want to talk about and away from those I wish to avoid.  28.  I say only good things about my friends behind their backs.  29.  With enough effort we can wipe out p o l t i c a l corruption.  232  1  STRONGLY DISAGREE  2  MODERATELY DISAGREE  3  SLIGHTLY DISAGREE  4  NEUTRAL  5  SLIGHTLY AGREE  6  MODERATELY AGREE  7  STRONGLY AGREE  30.  I feel I do not have much to be proud o f .  31.  I usually don't make plans because I have a hard time following through on them.  32.  I sometimes put off u n t i l tommorrow what I should do today.  33.  If I need help in carrying out a plan of mine, i t ' s usually d i f f i c u l t to get others to help.  34.  I always declare everything at customs.  35.  One of the major reasons we have wars i s because people don't take enough interest in p o l i t i c s .  36.  I take a positive attitude toward myself.  37.  Competition encourages  38.  I think I have some pretty awful habits.  39.  If there's someone I want to meet I can usually arrange i t .  40.  I always t e l l the truth.  41.  There i s very l i t t l e we, as consumers, can do to keep the cost of  excellence.  living from going higher. 42.  On the whole, I am s a t i s f i e d with myself.  43.  The extent of personal achievement i s often determined by chance.  44.  I am sometimes late for appointments.  45.  I often find i t hard to get my point of view across to others.  46.  I always obey t r a f f i c laws even i f I'm unlikely to get caught.  47.  When I look at i t carefully I r e a l i z e that i t i s impossible to have any r e a l l y important influence over what p o l i t i c i a n s do.  48.  I wish I could have more respect for myself.  49.  On any sort of exam or competition I l i k e to know how well I do relative to everyone else.  233  1 STRONGLY DISAGREE  2 MODERATELY DISAGREE  3 SLIGHTLY DISAGREE  4 NEUTRAL  5 SLIGHTLY AGREE  6 MODERATELY AGREE  7 STRONGLY AGREE  50.  I have never cheated on a test or assignment i n any way.  51.  In attempting to smooth over a disagreement I usually make i t worse.  52.  When I was a c h i l d I obeyed my parents.  53.  I prefer to concentrate my energy on other things rather than i n solving the world's problems.  54.  I certainly f e e l useless at times.  55.  Despite my best efforts I have few worthwhile accomplishments.  56.  I am always p o l i t e to others including my friends and family.  57.  I find i t easy to play an important part i n most group situations.  58.  In the long run we, the voters, are responsible for bad government on a national as well as a local l e v e l . At times I think I am no good at a l l .  59.  234  APPENDIX E: Verbal expectancy items, Study Four. 1.  If I t r i e d harder on the Marketing Game my performance would improve  2.  Whether I do better or worse does not depend on trying harder.  3.  When doing the Marketing Game, i f I increase my e f f o r t , my performance i s l i k e l y to go up.  4.  There's a strong connection between my level of e f f o r t and my level of performance.  5.  I'm l i k e l y to do as well or better on the Marketing Game even i f I were to reduce my effort.  6. When i t comes to my performance l e v e l , i t really doesn't matter much whether I work hard or not. 7.  My performance on the Marketing Game would go down i f I were to decrease my effort.  8.  My performance on the Marketing Game wouldn't be affected much i f I t r i e d harder.  APPENDIX F: C e l l means, a l l dependent variables, Study Four. F i r s t Manipulation Variable;  Checks  Anxiety Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev,  EDN END DEN DNE NED NDE  6.92 8.07 8.53 8.30 7.00 6.69  1.25 2.49 2.29 2.52 2.04 1.70  8.30 8.00 6.84 6.07 6.38 7.76  2.56 2.08 1.57 1.32 1.93 2.58  7 .53 7 .23 7 .84 6 .30 8 .53 7 .69  2.18 1.87 3.18 1.43 2.14 2.39  7.58  2.16  7.23  2.16  7 .52  2.29  Entire sample  Variable:  Depression Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev,  EDN END DEN DNE NED NDE  19.84 21.23 27.15 29.38 22.46 23.00  4.66 2.31 7.25 9.00 5.28 7.64  27.38 22.61 17.53 18.00 20.30 26.07  6.14 6.04 5.09 5.53 4.93 8.21  22.07 20.69 23.76 16.30 27.76 19.53  7.49 4.46 6.78 4.40 5.94 4.42  23.84  7.06  21.98  7.00  21.69  6.59  Entire sample  Variable:  Hostility Session One  Order EDN END DEN DNE NED NDE Entire sample  Mean  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std. Dev,  17.15 17.15 17.00 16.92 17.30 16.61  1.40 1.34 1.29 1.65 1.54 2.14  15.92 15.92 16.92 17.61 17.23 16.76  1.44 2.01 1.11 1.55 1.92 1.30  15.76 16.53 17.00 16.46 16.07 17.07  1.23 1.94 1.35 1.45 1.65 1.18  17.02  1.55  16.73  1.66  16.48  1.51  Variable:  Pleasure Session One  Order  Mean  EDN END DEN DNE NED NDE  36.84 34.69 32.84 31.00 35.76 35.84 34.50  i t i r e sample  Variable:  Std. Dev.  Session Two Mean  Std. Dev.  Mean  Std. Dev.  5. 14 6. 93 8. 16 11. 43 5.80 9. 41  28.07 35.07 37.76 39.38 35.61 34.84  8 .29 7 .17 6 .84 7 .28 5 .72 9 .47  36.07 35.53 34.23 39.61 31.00 38.30  5 .89 6 .60 7 .41 6 .35 6 .35 7 .59  8.09  35.12  8 .12  35.79  7 .08  Arousal Session One Mean  EDN END DEN DNE NED NDE  25.53 27.07 18.69 16.92 23.46 21.92  11 .20 10 .08 6 .70 5 .21 7 .06 8 .89  17.61 23.00 37.69 27.00 27.53 19.69  22.26  8 .92  25.42  i t i r e sample  Std. Dev.  Session Two  Order  Variable:  Session Three  Mean  Std. Dev.  Session Three Mean  Std. Dev.  6 .25 6 .70 6 .57 10 .87 11 .24 6 .07  25.23 23.84 21.30 33.84 20.30 34.46  7 .57 7 .32 7 .30 8 .44 3 .25 9 .24  10 .35  26.50  9 .16  Response Latency Session One  Order EDN END DEN DNE NED NDE Entire sample  Mean  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std. Dev.  194.13 206.38 209.36 191.69 204.52 214.81  45.81 32.59 46.95 48.18 33.47 66.26  128.61 125.03 121.81 113.71 134.64 126.80  30.15 26.83 27.19 16.74 18.77 33.81  109.40 104.78 106.31 98.99 114.92 109.15  20.26 22.97 21.30 24.82 19.86 29.87  203.48  46.09  125.10  26.22  107.26  23.18  Second Manipulation Checks Variable:  Anxiety Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev  EDN END DEN DNE NED NDE  8.38 9.23 8.23 8.30 8.38 8.07  2.10 2.38 2.04 2.32 3.15 1.93  8.92 8.69 8.38 6.69 8.23 8.38  1.97 1.97 2.56 1.18 1.96 2.02  8.23 8.00 8.07 7.07 9.38 7.84  1.73 1.41 2.75 1.49 2.56 1.95  8.43  2.31  8.21  2.05  8.10  2.09  Entire sample  Variable:  Depression Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev.  EDN END DEN DNE NED NDE  21.46 21.23 23.84 24.07 20.53 22.84  3.77 4.90 6.86 8.32 5.41 6.49  25.46 22.69 19.23 18.00 21.30 23.84  4.70 6.34 4.78 3.76 4.23 7.49  23.00 20.38 22.84 16.84 25.92 19.38  5.95 3.50 6.10 3.10 5.70 4.17  22.33  6.08  21.75  5.80  21.39  5.58  Entire sample  Variable:  Hostility Session One  Order EDN END DEN DNE NED NDE Entire sample  Mean  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std. Dev  17.38 16.30 15.84 16.61 15.76 15.30  2.43 1.65 1.95 2.69 1.36 1.60  16.76 15.69 15.46 15.53 15.69 15.76  1.53 1.60 1.56 1.50 .85 1.69  16.00 15.15 16.07 15.23 16.46 15.46  1.47 1.57 2.06 1.16 1.39 1.45  16.20  2.05  15.82  1.50  15.73  1.56  Variable:  Pleasure Session One  Session Two  Order  Mean  EDN END DEN DNE NED NDE  30.92 31.84 31.07 30.07 32.61 32.46  2.28 3.15 3.68 4.64 1.75 3.20  29.53 31.46 32.15 32.46 31.15 31.53  4.48 3.52 3.07 4.57 2.99 3.55  31.15 32.69 29.23 33.00 31.07 31.53  2.60 3.25 2.48 2.64 2.72 2.10  31.50  3.27  31.38  3.74  31.44  2.85  Entire sample  Variable:  Std. Dev.  Mean  Mean  Std. Dev.  Arousal Session One  Session Two  Order  Mean  EDN END DEN DNE NED NDE  28.23 29.00 28.69 26.15 27.84 27.92  2.71 4.67 3.70 2.37 3.18 2.66  27.23 28.38 30.15 28.23 29.38 27.53  27.97  3.33  28.48  Entire sample  Variable:  Std. Dev.  Session Three  Std. Dev.  Mean  Std. Dev.  Session Three Mean  Std. Dev.  3.26 4.03 3.15 3.65 3.92 2.72  28.69 29.38 27.84 30.07 28.30 29.92  3.42 4.07 3.50 2.36 1.60 2.98  3.52  29.03  3.11  Response Latency Session One  Order EDN END DEN DNE NED NDE i t i r e sample  Mean  Session Two  Std. Dev.  Mean  Std. Dev.  Session Three Mean  Std. Dev.  131.96 117.18 118.25 120.48 119.36 134.10  35.59 27.90 26.03 26.99 23.08 38.88  114.45 108.59 98.76 103.52 114.42 103.34  23.24 22.06 22.41 20.87 16.00 21.55  100.62 90.74 95.08 92.65 98.90 92.15  21.58 20.94 21.70 17.59 17.81 21.08  123.55  30.06  107.18  21.29  95.02  19.86  Dependent Variables Variable:  S a t i s f a c t i o n with Task Performance Session One  Order  Mean  EDN END DEN DNE NED NDE E n t i r e sample  Variable:  Std. Dev.  Mean  Session Three  Std. Dev.  Mean  Std. Dev  12.84 13.61 13.30 15.30 13.76 12.61  1.67 2.02 2.46 1.18 1.58 2.10  13.15 12.38 13.38 14.23 13.38 13.53  1.72 2.66 2.32 2.27 1.60 2.60  12 .15 12 .84 13 .61 13 .38 12 .76 13 .15  1.67 2.40 2.32 2.32 3.00 2.44  13.57  2.02  13.34  2.23  12 .98  2.36  Internal Work Motivation Session One  Order  Mean  EDN END DEN DNE NED NDE E n t i r e sample  Variable:  Session Two  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std . Dev  16.92 16.23 16.92 16.84 16.53 17.69  1.89 2.04 1.11 2.26 1.19 1.97  16.61 16.84 16.92 17.00 17.07 17.92  1.75 1.86 1.25 1.95 1.80 1.70  16.69 17.38 17.07 16.84 17.07 17.23  1 .70 2 .43 1 .38 2 .76 1 .80 2 .77  16.85  1.79  17.06  1.73  17.05  2 .14  Task Effort Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  EDN END DEN DNE NED NDE  7.00 6.84 7.23 7.46 7.23 7.38  1.15 1.28 1.09 1.71 1.09 1.26  6.15 6.38 6.84 6.84 7.07 7.15  1.21 1.66 1.46 1.72 1.11 1.67  5.76 5.92 6.30 6.38 6.46 6.92  1.48 1.49 1.93 1.66 1.50 1.38  7.19  1.25  6.74  1.48  6.29  1.57  E n t i r e sample  Std. De  Variable:  Task D i f f i c u l t y and Challenge Session One Mean  EDN END DEN DNE NED NDE  11.84 11.76 12.15 11.92 12.15 13.07  3.36 3.65 2.67 2.98 2.57 3.01  11.07 10.53 11.23 12.00 11.69 12.69  2.66 1.85 2.74 2.27 2.62 2.32  11.00 10.30 11.00 12.15 11.92 13.00  2.67 1.75 3.18 2.03 2.72 1.77  12.15  2.99  11.53  2.45  11.56  2.50  Variable:  Task  Mean  Session Three  Order  Entire sample  Std. Dev.  Session Two  Std. Dev. Mean  Std. Dev.  Interest Session One Mean  EDN END DEN DNE NED NDE  6.76 6.69 6.69 7.38 6.92 7.84  1.09 1.43 2.01 1.89 1.49 1.06  5.84 6.38 6.38 6.76 7.00 7.46  .98 1.60 1.98 1.53 1.77 1.05  5.92 6.00 6.00 7.15 6.61 7.61  1.03 1.52 1.73 1.67 1.60 .86  7.05  1.55  6.64  1.57  6.55  1.54  Variable:  Mean  Session Three  Order  E n t i r e sample  Std. Dev.  Session Two  Std. Dev. Mean  Std. Dev.  Verbal Expectancy Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  EDN END DEN DNE NED NDE  28.23 27.61 29.53 28.92 27.53 30.23  5.05 4.29 3.61 4.53 4.35 4.76  26.46 26.84 27.38 25.61 29.30 29.92  5.02 5.38 4.55 5.33 5.40 3.83  25.38 24.92 27.38 25.69 27.61 30.38  5.83 4.49 6.31 7.01 4.53 3.45  28.67  4.42  27.58  5.03  26.89  5.55  Entire sample  Std. Dev. Mean  Std. Dev.  Variable:  Internality Session One  Session Two  Order  Mean  EDN END DEN DNE NED NDE  13 .84 15 .92 15 .69 16 .61 16 .84 17 .15  5 .32 3 .75 3 .66 4 .87 3 .91 4 .29  14. 61 15. 15 16. 61 17. 30 16. 92 16. 15  6.44 4.91 4.01 3.54 4.44 5.44  14.00 14.00 15.84 15.23 15.53 16.69  5.61 3.91 4.98 5.35 4.33 3.72  16 .01  4.34  16. 12  4.83  15.21  4.65  i t i r e sample  Variable:  Std. Dev. Mean  Std. Dev.  Session Three Mean  Std. Dev.  Stability Session One  Mean  Std. Dev.  Session Three  Order  Mean  EDN END DEN DNE NED NDE  11 .69 11 .84 10 .84 12 .07 12 .30 13 .84  5 .51 3 .82 2 .91 3 .42 4 .81 3 .26  12.53 11.69 11.84 12.61 13.30 12.69  4.78 3.56 4.39 3.96 4.49 6.01  10.69 12.23 11.76 12.38 13.46 12.30  4.30 3.29 5.15 3.33 2.75 4.64  12 .10  4 .03  12.44  4.48  12.14  3.95  i t i r e sample  Std. Dev.  Session Two  Mean  Std. Dev.  Variable: C o n t r o l l a b i l i t y Session One Order EDN END DEN DNE NED NDE E n t i r e sample  Mean  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std. De  17.00 18.23 17.38 18.23 15.92 19.00  6.37 3.67 3.59 3.26 2.53 3.62  17.92 17.15 19.15 17.92 16.76 18.23  4.19 3.28 2.07 3.49 3.56 4.16  17.53 16.15 18.30 17.23 15.76 19.15  5.34 3.10 3.19 3.56 3.29 3.46  17.62  4.02  17.85  3.50  17.35  3.80  Variable:  Perceived Covariation Session One  Order  Mean  EDN END DEN DNE NED NDE Entire sample  Variable:  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std. D€  .17 .28 .30 .26 .23 .26  .16 .12 .16 .14 .15 .08  .21 .27 .25 .19 .18 .20  .23 .10 .18 .22 .17 .18  .22 .21 .25 .16 .19 .28  .22 .20 .19 .26 .16 .13  .25  .14  .22  .18  .22  .20  Perceived Control Session One  .Session Two  Session Three  Order  Mean  EDN END DEN DNE NED NDE  54 .07 50 .38 59 .23 70 .84 53 .84 63 .23  17 .91 22 .02 16 .93 10 .30 20 .22 16 .60  53.23 43.46 54.23 58.30 49.23 52.69  24 .67 19 .72 17 .54 14 .10 22 .53 23 .68  46 .87 43 .84 55 .84 51 .53 44 .07 54 .00  21 .53 19 .80 23 .59 19 .51 20 .76 19 .78  58 .60  18 .47  51.85  20 .55  49 .36  20 .73  i t i r e sample  Variable:  Std. Dev. Mean  Std. Dev.  Mean  Std. Dev.  Perceived Correlation Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev  EDN END DEN DNE NED NDE  4.84 4.92 5.07 4.92 5.46 4.76  2.70 1.84 2.46 2.39 1.89 1.87  4.92 4.38 4.30 4.76 4.46 5.23  2.43 2.06 2.46 1.58 2.10 2.55  4.76 4.61 5.38 4.46 4.23 5.76  2.31 1.85 2.10 2.06 1.87 1.64  5.00  2.16  4.67  2.17  4.87  1.99  Entire sample  Variable:  Performance Recall Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev,  EDN END DEN DNE NED NDE  3.34 3.10 3.21 3.73 3.03 3.58  .53 .85 .70 .29 1.12 .37  3.18 3.45 3.46 3.63 3.53 3.52  1.08 .47 .45 .31 .31 .28  3.37 3.54 3.71 3.62 3.36 3.66  .46 .27 .30 .30 .46 .27  3.33  .72  3.46  .56  3.54  .37  Entire sample  Variable:  Performance Expectation Session One  Session Two  Session Three  Order  Mean  Std. Dev.  Mean  Std. Dev.  Mean  Std. Dev.  EDN END DEN DNE NED NDE  3.51 3.43 3.47 3.84 3.40 3.60  .50 .95 .71 .26 .58 .66  3.65 3.64 3.56 3.72 3.35 3.72  .35 .46 .52 .29 1.04 .39  3.43 3.57 3.76 3.77 3.36 3.73  .41 .51 .31 .28 .65 .24  3.54  .64  3.61  ,56  3.60  .44  Entire sample  Variable: Recall Accuracy Session One Order EDN END DEN DNE NED NDE Entire sample  Mean  Std. Dev.  Session Two Mean  Std. Dev.  Session Three Mean  Std. Dev,  .35 .50 .39 .06 .72 .07  .28 .71 .50 .19 1.15 .21  .40 .26 .25 .05 .13 .15  1.06 .44 .30 .21 .23 .16  .17 .08 .03 .04 .18 .00  .32 .19 .06 .16 .32 .11  .35  .63  .21  .50  .08  .22  244  References  A l l a n , L.G., & Jenkins, H.M.  (1980) The judgement of contingency and  the nature of the response alternative.  Canadian Journal of  Psychology, 34, 1-11. A l l o y , L.B.  & Abramson, L.Y.  (1979) Judgement of contingency i n  depressed and nondepressed  students: Sadder but wiser?  Journal of  Experimental Psychology: General, 108, 441-485. A l l o y , L.B.  Sc Abramson, L.Y.  (1982) Learned helplessness, depression,  and the i l l u s i o n of c o n t r o l .  Journal of Personality and S o c i a l  Psychology, 42, 1114-1126. A l l o y , L.B., Abramson, L.Y. i l l u s i o n of control.  St V i s c u s i , D.  (1981) induced mood and the  Journal of Personality and Social Psychology,  41, 1129-1140. Alloy, L.B., St Tabachnik, N.  (1984) Assessment of covariation by humans  and animals: The joint influence of prior expectations and current s i t u a t i o n a l information. Psychological Review, 91, 112-149. Angier, R.P.  (1927) The c o n f l i c t theory of emotion.  American Journal  of Psychology, 39, 390-401. Argyris, C.  (1966) Interpersonal barriers to decision making.  Harvard  Business Review, 44, 84-97. Arnold, M.B.  (1960) Emotion and personality (Volumes 1 and 2).  New  York: Columbia University Press. Arnold, M.B. M.B.  (1970) Perennial problems i n the f i e l d of emotion.  Arnold (Ed.), Feelings and Emotion (pp.169-186).  Academic Press.  In  New York:  245  A v e r i l l , J.R.  (1980) A constructionist view of emotion.  Plutchik & H. experience Bandura, A.  In R.  Kellerman (Eds.), Emotion: Theory, research and  (Vol.  1). New York: Academic Press.  (1977) S e l f - e f f i c a c y : Toward a unifying theory of  behavioral change. Bateman, T.S.  Psychological Review, 84, 191-215.  & Organ, D.W.  (1983) Job s a t i s f a c t i o n and the good  soldier: The relationship between a f f e c t and employee "citizenship".  Academy of Management Journal, 26, 587-595.  Batson, CD., Coke, J.S., Chard, F., Smith D.  & T a l i a f e r r o A.  (1979)  Generality of the "Glow of Goodwill": E f f e c t s of mood on helping and information a c q u i s i t i o n . S o c i a l Psychology Quarterly, 42, 176179. Beck, A.T.  (1967) Depression:  aspects. Berscheid, E.  (1983) Emotion.  Blaney, P.  In H.H Kelley, E.  Harvey, T.  Peplau, & D.R.  168).  and t h e o r e t i c a l  New York: Harper & Row.  Christensen, J . A.  C l i n i c a l , experimental  Huston, G.  Peterson  Berscheid, A.  Levinger, E.  McClintock,  (Eds.), Close Relationships (pp.  110-  San Francisco, C a l i f . : Freeman. H.  (1986) Affect and memory: A Review.  Psychological  Bulletin,99, 229-246. Boggiano, A.K.  & Hertel, P.T.  i n memory. Borgatta, E.  I.  (1983) Bonuses and bribes: Mood e f f e c t s  Social Cognition, 2, 49-61. (1961) Mood, personality and i n t e r a c t i o n . Journal of  General Psychology, 64, 105-137.  246  Bower, G.H.  (1981) Mood and memory.  American Psychologist, 36, 129-  148. Bower, G.H.,  G i l l i g a n , S.G.  & Montiero, K.P.  learning caused by a f f e c t i v e states.  (1981) S e l e c t i v i t y of  Journal of Experimental  Psychology: General, 110, 451-473. Bower, G.H., Montiero, K.P.  & G i l l i g a n , S.G.  a context f o r learning and r e c a l l .  (1978) Emotional mood as  Journal of Verbal Learning and  Verbal Behaviour, 17, 573-585. Bower, G.H.  (1985) Review of research on mood and memory.  Paper  presented at the symposium on Affect and Cognition at the meeting of the Cognitive Psychology section of the B r i t i s h Psychology Association, Oxford. Bradley, G.W.  (1978) Self-serving biases i n the a t t r i b u t i o n process: A  reexamination of the fact or f i c t i o n question.  Journal of  Personality and Social Psychology, 36, 56-71. B r i e f , A.  P.  & Motowidlo, S.  behaviors. Brockner, J .  J.  (1986) Prosocial organizational  Academy of Management Review, 11, 710-725.  (1979) The effects of self-esteem, s u c c e s s — f a i l u r e , and  self-consciousness on task performance.  Journal of Personality and  Social Psychology, 37, 1732-1741. Brockner, J .  (1983) Low self-esteem and behavioral p l a s t i c i t y : Some  implications.  In L.  Wheeler & P.R.  Shaver (Eds.) Review of  Personality and Social Psychology (Vol. Brockner, J . , & Guare, J .  4.). Beverly H i l l s : Sage.  (1983) Improving the performance of low-self-  esteem i n d i v i d u a l s : An a t t r i b u t i o n a l approach. Management Journal, 26, 642-656.  Academy of  247  Brockner, J . , H j e l l e , L., & Plant, R.W.  (1985) Self-focused attention,  self-esteem, and the experience of state depression.  Journal of  Personality, 53, 425-434. Brown, J . (1984) Effects of induced mood on causal attributions for success and f a i l u r e .  Motivation and Emotion, 8, 343-353.  Buchward, A.M., Strack, S.  & Coyne, J.C. (1981) Demand c h a r a c t e r i s t i c s  and the Velten mood induction procedure.  Journal of Consulting and  C l i n i c a l Psychology, 49, 478-479. Campbell, J.P. & Fairey, P.J. (1985) Effects of self-esteem, hypothetical explanations, and verbalization of expectancies on future performance.  Journal of Personality and S o c i a l Psychology,  48, 1097-1011. Campbell, J.P. & Pritchard, R.D.  (1976) Motivation theory i n  i n d u s t r i a l and organizational psychology.  In M.D.  Dunnette  Handbook of Industrial and Organizational Psychology (pp.  (Ed.),  62-130).  Chicago: Rand McNally. Candland, D.K.  (1977) The persistent problems of emotion.  Candland, J.P. F e l l , E. Tarpy (Eds.), Emotion Cannon, W.B.  Keen, A.I.  (pp.1-84).  Leshner, R.  In D.K.  Plutchik & R.M.  Monterey, C a l i f . : Brooks/Cole.  (1927) The James-Lange theory of emotions: a c r i t i c a l  examination and an a l t e r n a t i v e theory.  American Journal of  Psychology, 39, 106-124. C i a l d i n i , R.B., Darby, B., & Vincent, J . altruism: A case for hedonism. Psychology, 9, 502-516.  (1973) Transgression and  Journal of Experimental Social  248  C i a l d i n i , R.B.  & Kenrick, D.T.  (1976) Altruism as hedonism: A s o c i a l  development perspective on the relationship of negative mood state and helping.  Journal of Personality and Social Psychology, 34,  907-914. Clark, M.S.  (1981) Enhancing the l i n k between f e e l i n g states and  judgements through arousal.  Unpublished manuscript, Carnegie-  Mellon University. Clark, M.S.  (1982) A role for arousal i n the link between f e e l i n g  states, judgments, and behavior.  In M.S.  Clark & S.T.  Fiske  (Eds.) A f f e c t and Cognition, London: Erlbaum. Clark, D.M.  (1983) On the induction of depressed mood i n the  laboratory: Evaluation and comparison of the Velten and Musical procedures. Clyde, D.J. Fla.:  Advances i n Behaviour Research Therapy, 5, 27-49.  (1963) Manual for the Clyde Mood Scale.  Coral Gables,  Biometric Laboratory, University of F l o r i d a .  Cofer, C.N.  (1972) Motivation and emotion.  Glenview, I l l i n o i s :  Scott,  Foresman. Coopersmith, S.  (1967) The Antecedents of Self-Esteem.  San Francisco:  Freeman. Cyert, R.M., & March, J.G.  (1963) A behavioral theory of the firm.  Englewood C l i f f s , N.J.: Prentice-Hall. Darwin, C.  (1872) Expression of emotion i n man and animals.  New York:  Appleton. Demo, D.H.  (1985) The measurement of self-esteem: Refining our methods.  Journal of Personality and Social Psychology.  48, 1490-1502.  249  Denison, D.R.  & Sutton, R.I.  ( i n press) Using emotion and meaning i n  organizational diagnosis: Lessons form a team of operating room nurses.  In J.R.  Hackman (Ed.), Groups that Work.  San Francisco:  Jossey-Bass. Dewey, J .  (1894) The theory of emotions: I Emotional attitudes.  Psychological Review, 1, 553-569. Dewey, J .  (1895) The theory of emotions: II The significance of  emotions. Dixon, W.J.  Psychological Review, 2, 13-32.  (Ed.) (1981) BMDP S t a t i s t i c a l Software.  Berkeley:  University of C a l i f o r n i a Press. Easterbrook, J.A.  (1959) The e f f e c t of emotion on cue u t i l i z a t i o n and  the organization of behaviour. Eich, E. vs.  & Metcalfe, J .  Psychological Review, 66, 183-201.  ( i n press) Mood dependent memory for internal  external events.  Journal of Experimental  Psychology:  Learning, Memory & Cognition. Ekman, P.  (1984) Expression and the nature of emotion.  Scherer & P.  Ekman (Eds.) Approaches to emotion.  In K.R. H i l l s d a l e , N.J.:  Erlbaum. Feather, N.T.  (1966) Effects of p r i o r success and f a i l u r e on  expectations of success and subsequence performance.  Journal of  Personality and Social Psychology, 3, 287-298. Feather, N.T.  (1984a) Expectancy-value approaches:  future d i r e c t i o n s .  In N.T.  Actions: Expectancy-Value H i l l s d a l e , N.J: Erlbaum.  Present status and  Feather (Ed.), Expectations and  Models i n Psychology (pp.  395-420).  250  Feather, N.T.  (1984b) Actions i n r e l a t i o n to expected consequences: An  overview of a research program.  In N.T.  Feather (Ed.),  Expectations and Actions: Expectancy-Value Models i n Psychology (pp.  53-96).  F e l l , J.P.  H i l l s d a l e , N.J: Erlbaum.  (1977) The phenomenological approach to emotion.  Candland, J.P.  F e l l , E.  Keen, A.I.  Tarpy (Eds.), Emotion (pp.253-285). Forest, D., Clark, M., M i l l s , J .  Leshner, R.  In  D.K.  Plutchik &  R.M.  Monterey, C a l i f . : Brooks/Cole.  & Isen, A.M.  (1979) Helping as a  function of feeling state and nature of the helping behaviour. Motivation and Emotion, 3, 161-169. Fraisse, P.  (1968) Les Emotions.  T r a i t e de Psychologie Experimentale,  5, Paris: Presses U n i v e r s i t a i r e s . Ganster, D.C.,  Hennessey, H.W.,  Luthans, F.  (1983) Social D e s i r a b i l i t y  Response e f f e c t s : Three A l t e r n a t i v e Models.  Academy of Management  Journal, 26, 321-331. Garland, H.  (1984) Relation of effort-performance expectancy to  performance i n goal-setting experiments.  Journal of Applied  Psychology, 69, 79-84. Goffman, E.  (1969) Strategic Interaction.  Philadelphia: University of  Pennsylvania Press. Goodwin, A.W.  & Williams, J.M.G.  (1982) Mood induction research: Its  implications for c l i n i c a l depression. Behaviour Research and Therapy, 20, 373-382. Guttman, L.  (1945) A basis for analyzing t e s t - r e t e s t  Psychometrika, 10, 255-282.  reliability.  251  Guzzo, R.A.  & Waters, J.A.  ( 1 9 8 2 ) The expression of a f f e c t and the  performance of decision-making groups. Psychology, 6 7 ,  Journal of Applied  67-74.  Hackman, J.R., & Oldham, G.R.  ( 1 9 8 0 ) Work Design.  Reading, Mass.:  Addison-Wesley. Hakstian, A.R.,  Roed, J.C., Si Lind, J.C.  (1979) Two-sample T  2  procedure  and the assumption of homogeneous covariance matrices. Psychological B u l l e t i n ,  86,1255-1263.  ( 1 9 8 0 ) The t r i l o g y of mind: Cognition, a f f e c t i o n and  Hilgard, E.R. conation.  Journal of the History of the Behavioural Sciences, 16,  107-117.  ( 1 9 6 1 ) Emotion: A comprehensive phenomenology of theories  Hillman, J .  and their meanings for therap.  Evanston 1 1 1 . : Northwestern  University Press. Hochschild, A.  ( 1 9 7 9 ) Emotion work, f e e l i n g rules and s o c i a l structure.  performance. Hochschild, A. Feeling.  R.  American Journal of Sociology, 8 5 ,  ( 1 9 8 3 ) The Managed Heart: Commercialization of Human  University of C a l i f o r n i a Press: Berkeley.  Hollenbeck, J .  ( 1 9 8 4 ) A matrix method for expectancy research.  of Management Review, 4 , Hunt, W.A.  Academy  579-587.  ( 1 9 4 1 ) Recent developments i n the f i e l d of emotion.  Psychological B u l l e t i n , 3 8 , Ilgen, D.R.,  551-575.  Nebecker, D.B.,  249-276.  Si Pritchard, R.D.  ( 1 9 8 1 ) Expectancy theory  measures: An empirical comparison i n an experimental simulation. Organizational Behavior and Human Performance, 2 8 ,  189-223.  252  Isen, A.M.  (1970) Success, f a i l u r e , attention and reactions to others:  The warm glow of success.  Journal of Personality and S o c i a l  Psychology, 15, 297-301. Isen, A.M.  (1984) Toward understanding the r o l e of a f f e c t i n cognition.  In R.S.  Wyer J r .  Cognition Isen, A.M.,  (Vol.  and T.K. 3) (pp.  & Daubman, K.A.  categorization.  S k r u l l (Eds.), Handbook of S o c i a l  179-236).  H i l l s i d e , N.J.: Erlbaum.  (1984) The influence of a f f e c t on  Journal of Personality and S o c i a l Psychology, 47,  1206-1217. Isen, A.M., Daubman, K.A.  & Gorgoglione, J.M.  (In press) The influence  of positive affect on cognitive organization.  In R.  Snow & M.  Farr (Eds.), Aptitude, Learning and Instruction: A f f e c t i v e and Conative Processes. Isen, A.M.,  H i l l s d a l e , N.J.: Erlbaum.  & Gorgoglione, J.M.  (1983) Some s p e c i f i c e f f e c t s of four  a f f e c t induction procedures.  Personality and S o c i a l Psychology  . B u l l e t i n , 9, 136-143. Isen, A.M., Horn, N., & Rosenhan, D.L. f a i l u r e on children's generosity.  (1973) E f f e c t s of success and Journal of Personality and  Social Psychology, 27, 239-247. Isen, A.M.  & Means, B.  (1983) The influence of p o s i t i v e a f f e c t on  decision making strategy.  Social Cognition, 2, 18-31.  Isen, A.M., Means, B., Patrick, R. a f f e c t and decision making. Affect and Cognition.  & Nowicki, G. In M.S.  (1982) P o s i t i v e  Clark & S.  H i l l s d a l e , N.J.: Erlbaum.  Fiske (Eds.),  253  Isen, A.M.  & Patrick, R.  (1983) The effects of positive feelings on  r i s k - t a k i n g : When the chips are down.  Organizational Behavior and  Human Performance, 31, 194-202. Isen, A.M.,  Shalker, T., Clark, M.  & Karp, L.  (1978) Affect,  a c c e s s i b i l i t y of material i n memory and behaviour: A cognitive loop?  Journal of Personality and Social Psychology, 36, 1-12.  Isen, A.M.  & Simmonds, S.F.  (1978) The effect of f e e l i n g food on a  helping that i s incompatible with good mood.  Social Psychology,  41, 345-349. Izard, C E .  (1971) The face of emotion. New York: Plenum.  Izard, C E .  (1977) Human Emotions. New York: Plenum.  James, W.  (1884) What i s emotion?  Jenkins, H.M.,  & Ward, W.C  (1965) Judgement of contingency between  responses and outcomes. No.  Mind, 9, 188-205.  Psychological Monographs, 79, (1, Whole  594).  Jennings, D.L., Amabile, T.M.,  & Ross, L.  (1982) Informal covariation  assessment: Data-based versus theory-based judgements. Kahneman, P.  Slovic, & A.  In D.  Tversky (Eds.), Judgement under  uncertainty: Heuristics and biases (pp.  211-230).  Cambridge:  Cambridge University Press. Kahn, W.A.  (1986) Towards a sense of organizational humor: Implications  for organizational diagnosis.  Paper presented at the annual  meeting of the Academy of Management, Chicago. Kavanagh, D.J., & Bower, G.H.  (1985) Mood and s e l f - e f f i c a c y : Impact of  joy and sadness on perceived c a p a b i l i t i e s . Research, 9, 507-525.  Cognitive Therapy and  254  Kemper, T.D.  (1978) A Social Interactional Theory of Emotions.  Wiley:  New York. Kemper, T.D.  (1984) Power, status, and emotions: A S o c i o l o g i c a l  contribution to psychophysiological domain. Ekman (Eds.), Approaches to emotion (pp.  In K.R.  Scherer & P.  369-384), H i l l s d a l e ,  N.J.: Erlbaum. Kennedy, C.W.,  Possum, J.A.  & White, B.J.  (1983) An empirical  comparison of within-subjects and between-subjects expectancy theory models.  Organizational Behaviour and Human Performance, 32,  124-143. Kirk, R.E.  (1968) Experimental Design: Procedures for the Behaviour  Sciences.  Belmont, C a l i f o r n i a : Brooks Cole.  Kleinginna, P.R.,  & Kleinginna, A.M.  (1981) A categorized l i s t of  emotion d e f i n i t i o n s , with suggestions for a consensual d e f i n i t i o n . Motivation and Emotion, 5, 345-379. Lange, C.  (1885) Om Sindsbevaegelser.  Lazarus, R.S.  (1982) Thoughts on the r e l a t i o n s between emotion and  cognition. Leathers, D.G.,  Copenhagan.  American Psychologist, 37, 1019-1024. & Emigh, T.H.  test with decoding norms.  (1980) Decodong f a c i a l expressions: A new Quarterly Journal of Speech, 66, 418-  436. Leeper, M.R.  (1970) The motivational and perceptual properties of  emotions i n d i c a t i n g t h e i r fundamental character and r o l e . Arnold (Ed.), Feelings and Emotions (pp.151-168). Academic  Press.  In M.B.  New York:  255  Leventhal, H.  (1979) A perceptual-motor processing model of emotion.  In P.Pliner, K.  Blankstein, & I.M.  Spigel (Eds.), Advances i n the  Study of Communication and Affect (Vol. 5). New York: Plenum. Locke, E.A.  (1976) The nature and causes of job s a t i s f a c t i o n .  Dunnette  (Ed.) Handbook of Industrial and Organizational  Psychology. Maier, N.R.F.  Skokie, 111.: Rand McNally.  (1963) Problem solving discussions and conference:  Leadership methods and s k i l l s . Maier, N.R.F. groups. Mandler, G.  In M.  New York: McGraw-Hill.  (1970) Problem solving and c r e a t i v i t y : In individuals and Monterey, C a l i f . : Brooks/Cole. (1984) Mind and Body.  Mardia, K.V.  New York: Norton.  (1971) The effect of nonnormality on some multivariate  tests and robustness to nonnormality i n the linear model. Biometrika, 58, 105-121. Maslach, C.  (1978) How people cope.  Mclntyre, S.H. models.  Public Welfare, Spring, 56-58.  (1979) The leverage impact of judgement based marketing  Unpublished Doctoral Dissertation, Stanford University  Graduate School of Business, Stanford, C a l i f o r n i a . Mclntyre, S.H.  (1982) Experimental study of the impact of judgement  based marketing models. McNair, D.N.  & Lorr, M.  Management Science, 28, 17-33.  (1964) An analysis of mood i n neurotics.  Journal of Abnormal and Social Psychology, 69, 620-627. Meddis, R.  (1972) Bipolar factors i n mood adjective c h e c k l i s t s .  B r i t i s h Journal of Social and C l i n i c a l Psychology, 11, 178-184.  256  Mednick, S.A. (1962) The associative basis of creative process. Psychological Review, 69, 220-232. Mednick, M.T. Mednick, S.A. S. Mednick, E.V.  (1964) Incubation of  creative performance and s p e c i f i c a s s o c i a t i v e priming.  Journal of  Abnormal and Social Psychology, 69, 84-88. Meyer, L.B.  (1956) Emotion and Meaning i n Music.  Chicago: University  of Chicago Press. Mischel, W., Coates, B. & Raskoff, A. f a i l u r e on s e l f - g r a t i f i c a t i o n .  (1968) Effects of success and  Journal of Personality and S o c i a l  Psychology, 10, 381-390. Mischel, W., Ebbesen, E. & Zeiss, A.  (1973) Selective attention to the  s e l f : Situational and d i s p o s i t i o n a l determinants.  Journal of  Personality and Social Psychology, 27, 129-142. Mischel, W., Ebbesen, E. & Zeiss, A. memory about the s e l f .  (1976) Determinants of s e l e c t i v e  Journal of Consulting and C l i n i c a l  Psychology, 44, 92-103. Mitchell, T.R. (1974) Expectancy models of job s a t i s f a c t i o n , occupational preference  and e f f o r t : A t h e o r e t i c a l , methodological,  and empirical a p p r a i s a l .  Psychological B u l l e t i n , 81, 1053-1077.  Mitchell, T.R. (1982) Motivation: New d i r e c t i o n s for theory, and p r a c t i c e . Motowidlo, S.J.  research,  Academy of Management Review, 7, 80-88.  (1984) Does job s a t i s f a c t i o n lead to consideration and  personal s e n s i t i v i t y ?  Academy of Management Journal, 27, 910-915.  Motowidlo, S.J. & Lawton, G.W.  (1984) A f f e c t i v e and cognitive factors  i n soldier's reenlistment decisions. , 69, 157-166.  Journal of Applied Psychology  257  Motowidlo, S.J., Packard, J.S. & Manning, M.R. stress: Its causes and consequences  (1985) Occupational  for job performance.  Unpublished manuscript, Department of Organizational Behavior, Pennsylvania State University. Muchinsky, P.M.  (1977) A comparison of within- and across-subjects  analyses of the expectancy-valence model for predicting e f f o r t . Academy of Management Journal, 20, 154-158. Nasby, W.  & Yando, R.  (1982) Selective encoding and r e t r i e v a l of  a f f e c t i v e l y valent information: Two cognitive consequences of children's mood states.  Journal of Personality and Social  Psychology, 43, 1244-1253. Natale, M.  & Hantas, M.  (1982) Effect of temporary mood states on  selective memory about the s e l f .  Journal of Personality and Social  Psychology, 42, 927-934. Nelson, R.E.  & Craighead, W.E.  (1977) Selective r e c a l l of p o s i t i v e and  negative feedback, s e l f - c o n t r o l behaviours, and depression. Journal of Abnormal Psychology, 86, 379-388. Norman, D.A.  (1980) Twelve issues for Cognitive Science. Cognitive  Science, 4, 1-32. Nowlis, V.  (1965) Research with the mood adjective c h e c k l i s t .  Tomkins & C E .  In S.S.  Izard (Eds.), Affect, cognition, and personality.  New York: Springer. Nunnally, J . C  (1978) Psychometric Theory.  New York: McGraw-Hill.  Park, O.S., Sims, H.P., J r . , & Motowidlo, S.J. (1984) Mood and cognition: Variations and causal relationships i n the achievement context'.  Unpublished manuscript, Department of Organizational  Behavior, Pennsylvania State University.  258  Park, O.S.,  Sims, H.P.,  & Motowildo, S.L.  (1986) Affect i n  organizations: How feelings and emotions influence managerial judgement.  In H.P.  Organization (pp. Paulhan, F.  Sims & D.A. 215-237).  Gioia (Eds.), The Thinking  San Fransisco: Jossey-Bass.  (1930) The Laws of Feeling.  (Translated by C.K.  Ogden.)  London: Kegan Paul, Trench, Trubner. Paulhus, D.L.  (1984) Two-component models of s o c i a l l y desirable  responding.  Journal of Personality and Social Psychology, 46, 598-  609. Paulhus, D.L.  (1987) Personal communication.  P i g n a t i e l l o , M.F.  Camp, C.J., & Rasar, L.A.  (1986) Musical mood  induction: An alternative to the Velten technique.  Journal of  Abnormal Psychology, 95, 295-297. Pinder, C.C.,  & Moore, L.F.  the development behavior. Pinder, C C .  (1979) The resurrection of taxonomy to a i d  of middle range theories of organizational  Administrative Science Quarterly, 24, 99-118. & Moore, L.F.  (Eds.) (1980) Middle Range Theory and The  Study of Organizations. Boston: Martinus N i j h o f f . Plutchik, R.  (1962) The Emotions: Facts, Theories and a New Model.  New  York: Random House. Plutchik, R.  (1977) Cognitions i n the service of emotions: An  evolutionary perspective. A.I.  Leshner, R.  189-210).  In D.K.  Plutchik & R.M.  Candland, J.P.  F e l l , E.  Keen,  Tarpy (Eds.), Emotion (pp.  Monterey CA: Brooks Cole.  259  Plutchik, R.  (1980) Emotion: A Psychoevolutionary Synthesis.  Harper & Polivy, J .  New  York:  Row.  & Doyle, C.  (1980) Laboratory induction of mood states  through the reading of self-referent mood statements: A f f e c t i v e changes or demand characteristics.  Journal of Abnormal Psychology,  89, 286-290. Porac, J.F., Nottenburg, G.  St Eggert, J .  (1981) On extending Weiner s 1  a t t r i b u t i o n a l model to organizational contexts.  Journal of Applied  Psychology, 66, 124-126. R a f a e l i , A.  & Sutton, R.J.  the work r o l e . Richards, C.  (1987) The expression of emotion as part of  Academy of Management Review, 12, 23-37.  (1981) The use of music to induce mood.  year undergraduate project.  Dept.  Unpublished 3rd  of Experimental Psychology,  University of Oxford. Robinson, J.P.  St Shaver, P.R.  attitudes. Roseman, I . J . theory.  Survey Research Center I n s t i t u t e for S o c i a l Research. (1984) Cognitive determinants of emotion: A structural  In P.  Shaver (Ed.) Review of personality and s o c i a l  psychology (Vol. Rosenberg, M.  (1973) Measures of s o c i a l psychological  5).  (pp.  11-36).  Beverly H i l l s : Sage.  (1965) Society and the adolescent self-image. Princeton,  N.J.: Princeton University Press. Russell, D.  (1982) The causal dimension scale: A measure of how  i n d i v i d u a l s perceive causes. Psychology, 42, 1137-1145.  Journal of Personality and Social  260  Russell, J.A.  (1979) A f f e c t i v e space i s b i p o l a r .  Journal of  Personality and Social Psychology, 37, 345-356. Russell, J.A.  (1980) A circumplex model of a f f e c t .  Journal of  Personality and Social Psychology, 39, 1161-1178. Russell, J.A., & Mehrebian, A. theory of emotions.  (1977) Evidence f o r a three factor  Journal of Research i n Personality, 11, 273-  294. Schachter, S.  (1971) Emotion, Obesity and Crime.  New York: Academic  Press. Schachter, S., Singer, J.E. (1962) Cognitive, s o c i a l and physiological determinants of emotional state.  Psychological Review, 69, 379-  399. Scherer, K.  R.  & Tannenbaum, P.  H.  everyday l i f e : A survey approach.  (1980) Emotional experiences i n Motivation and Emotion, 10, 295-  314. Shaklee, H., & Tucker, D.  (1980) A r o l e analysis of judgements of  covariation between events. Shatin, L.  Memory & Cognition, 8, 459-467.  (1970) A l t e r a t i o n of mood v i a music: A study of the  vectoring e f f e c t .  Journal Psychology, 75, 81-86.  Shrauger, J.S. (1972) Self-esteem and reactions to being observed by others. S i l b e r , E.  Journal of Personality and Social Psychology, 23, 192-200. & Tippett, Jean.  (1965) Self-esteem: C l i n i c a l  and measurement v a l i d a t i o n . Simon, H.  A.  assessment  Psychological Reports, 16, 1017-1071.  (1967) Motivational and emotional controls of cognition.  Psychological Review, 74, 29-39.  261  S i n c l a i r , R.C ( i n press) Mood, categorization breadth, and performance appraisal: effects of information a c q u i s i t i o n and a f f e c t i v e state on halo, accuracy, information r e t r i e v a l and evaluations. Personality and Social Psychology B u l l e t i n . Snyder, R.A., Howard, A., St Hammer, T.H.  (1978) The predictive power of  within- versus across-subjects scores i n expectancy research. Journal of Psychology, 100, 285-292. Staw, B.M.  (1984) Organizational behaviour: A review and reformulation  of the f i e l d ' s outcome variables.  Annual Review of Psychology, 35,  627-666. Strongman, K.T.  (1978) The psychology of emotion (2nd Ed.).  Chichester: Wiley. Sutherland, G., Newman, B., & Rachman, S.  (1982) Experimental  investigations of the relations between mood and i n t r u s i v e , unwanted cognitions.  B r i t i s h Journal of Medical Psychology, 55,  127-138. Sutton, R.I. & Rafaeli, A. performance.  (1986) Emotion work and f i n a n c i a l  Unpublished manuscript.  Sutton, L . & Teasdale, J.D.  Stanford University.  (1982) Unpublished manuscript.  Department  of Psychiatry, University of Oxford. Tabachnik, B.G., & F i d e l l , L.S. (1983) Using multivariate s t a t i s t i c s . New York: Harper & Row. Teasdale, J.D. & Fogarty, S.J.  (1979) D i f f e r e n t i a l e f f e c t s of induced  mood on r e t r i e v a l of pleasant and unpleasant events from episodic memory.  Journal of Applied Psychology, 88, 248-257.  262  Teasdale, J.D.  & Russell, (1983) D i f f e r e n t i a l aspects of induced mood  on the r e c a l l of p o s i t i v e , negative, and neutral words.  British  Journal of C l i n i c a l Psychology, 22, 163-171. Teasdale, J.D.  & Spencer, P.  (1982) E f f e c t s of induced e l a t i o n -  depression on estimates of past successes and subjective p r o b a b i l i t y of future success.  Unpublished manuscript, Department  of Psychiatry, University of Oxford. Teasdale, J.D. success.  & Spencer, P.  (1984) Induced mood and estimates of past  B r i t i s h Journal of C l i n i c a l Psychology, 23, 149-150.  Teasdale, J.D.  & Taylor, R.  (1981) Induced mood and a c c e s s i b i l i t y of  memories: An effect of mood state or of induction procedure? B r i t i s h Journal of C l i n i c a l Psychology, 20, 39-48. Tomkins, S.  (1962) Affect, Imagery, and Consciousness.  Positive Affects. Tomkins, S.  Tomkins, S.  New York: Springer.  (1963) Affect, Imagery, and Consciousness.  Negative Affects.  V o l . I: The  Vol.11: The  New York: Springer.  (1980) Affect as amplification: Some modifications i n  theory.  In R.  Pultchik and H.  Emotion (Vol.1) (pp. 141-163).  Kellerman (Eds.), Theories of New York: Academic Press.  T o s i , H.L., Rizzo, J.R., & C a r r o l l , S.J. (1986) Managing Organizational Behavior. Tversky, A.  Marshfield, MA.: Pitman.  & Kahneman, D.  (1973) A v a i l a b i l i t y : A h e u r i s i t i c f o r  judging frequency and p r o b a b i l i t y . 232.  Cognitive Psychology, 5, 207-  263  Van Maanen, J .  & Kunda, G.  (1985) "Real f e e l i n g s " : Organizational time  outs and the expressions of emotions.  Paper presented at the  Annual Meeting of the Academy of Management, San Diego. Veitch, R.,  & G r i f f i t t , W.  (1976) Good news—bad news: A f f e c t i v e and  interpersonal effects.  Journal of Applied Social Psychology,  6,  69-75. Velten J r . , E.  (1968) A laboratory task for induction of mood states.  Behavior Research and Therapy, 6, 473-482. Vroom, V.H.  (1964) Work and Motivation.  Wanous, J.P., Keon, T.L.  s> Latack, J.C.  New  York: Wiley.  (1983) Expectancy  occupational/organizational choices: A review and  theory and  test.  Organizational Behavior and Human Performance, 32, 66-86. Ward, W.C.,  & Jenkins, H.M.  (1965) The d i s p l a y of information and the  judgement of contingency.  Canadian Journal of Psychology,  19,  231-  241. Wasserman, E.A.,  Chatlosh, D.L.,  & Neunaber, D.J.  (1983) Perception of  causal relations i n human factors a f f e c t i n g judgements of responseoutcome contingencies under free-operant procedures.  Learning and  Motivation, 14, 406-432. Weick, K.E.  (1985) The Emotions of Organizing.  Paper presented at the  Annual Meeting of the Academy of Management, San Diego. Weiner, B.  (1979) A theory of motivation f o r some classroom  experiences.  Journal of Educational Psychology,  Weiner, B., Russell, D.  & Lerman, D.  of causal ascriptions.  In J.H.  (1978).  71,  A f f e c t i v e consequences  Harvey, W.J.  Ickes & R.F.  (Eds.), New Directions i n A t t r i b u t i o n Research (Vol. H i l l s d a l e , N.J.: Erlbaum.  3-25.  2).  Kidd  264  Weiner, B., R u s s e l l , D.  & Lerman, D.  (1979) The cognition-emotion  process i n achievement-related contexts.  Journal of Personality  and Social Psychology, 37, 1211-1220. Weyant, J.M.  (1978) Effects of mood states, costs, and benefits of  helping.  Journal of Personality and Social Psychology, 36, 1169-  1176. Wright, J .  & Mischel, W.  (1982) Influence of a f f e c t on cognitive  social learning person variables.  Journal of Personality and  Social Psychology, 43, 901-914. Young, P.T. Zajonc, R.B.  (1961) Motivation and Emotion. New York: Wiley. (1980) Feeling and thinking: Preferences  inferences. Zerbe, W.J.,  need no  American Psychologist, 35, 151-175.  & Paulhus, D.L.  (1987) S o c i a l l y desirable responding i n  organizational behavior: A reconception.  Academy of Management  Review, 12, 250-264. Zuckerman, M., & Lubin, B. Adjective C h e c k l i s t .  (1965) Manual for the Multiple Affect San Diego, CA.: Educational Testing Service.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0098296/manifest

Comment

Related Items