QI ‘ _; 5 IEII', WI” (III? 1 .-I I IR": IIIII'I. I'II 7.3. " ‘ ~ .. .wIFIPIcII 2:91: .1 .. “ . I. : :f-fixgagfii -_. ..‘-.31.. 3‘3. 1 ' A.--“ 1'0 ' 'IIII'IIffzII:IIII :9 I) II III. . l I}. 1 . I911, . 3 IL. in. £11,121 IIM1 Iago- II. (II-I1..,.. " I 3m gag . .. ..I I? -IIII'1 ‘fi‘IIlu’ In ”III “III I In“. IIIEII I'I1‘:I'I:3U'¥I~}IIJ :i_|' I w I ‘ :1 'fl 1 :r“ I',‘ 1‘t\~”:”"‘ll1l {$13 IIIIA‘Y'J" IIIIIGE'IJI'I I. I' " “mm VII'L', IlIn'IIIIII fig. I-‘II f-II' 155% IN I . ._ “C- I III 7‘3“”... I-I'Ifig ;_- _'.'"E;‘I J55! .._ I I:\ 7”: _ II' It I"If“ LNIE 51%é1.‘II'-Ibg\;ru_ \ : ‘ ‘ f2”: |I‘:‘3#%r“'{.u _ - hm“ I J‘rki-S mrfigfl‘b: Q I 451‘ I} I' t‘o I‘ 3.; ..‘;éil if"; ’I I I: .‘I .3 I‘ III' ,1- 2’ Iv,“ ‘ti .1 -. 1, 1',:.v-. ,I‘.’ L‘. -‘ I’uI‘I IIII‘I‘IQIRI II‘IZIILw? .1 {II I; - I I I .: : . . .: a. ' r ‘!.. 'Ifi" 21:15... ~ , .I .I. .55; I"'~I.. ‘~~ 73'“. III .I I) 'I!‘ I'I I I IIIIIIII IIIPII II II-I. .."IIII IIIIII'5 '. I IIUIIJ'I' ‘ ' .EII‘IEI IIIIII 'II ‘ w III1; 1:1 I! :11; \‘I...1,1I 1.11 1 ‘ .11.,Iw'1: ‘I'I II I I I 4.9 ‘ .I . 'III II ‘1 I1? 1.1!! II'I11I 1‘ 313,51 'IiI,‘ ..1; i\11' . .1 1 I .. v I ,1 fit"! .1. 1.1. IIIII‘III “:11-‘111311111'3‘ “I1 l"... 1‘". 1.1:“,- 15‘ 1 . 1 I: I I! :1,': I ”1y. I I . f: “'1 0". 1 1 q I." “I I” 1 ' 1 ‘ [:l -f “A I n I HI}- I; 1 .- III-1 I I 14.7 'I 1.1 . - ,I1'. 1 "1 .3". ‘.. 1'13““; . . :11 ..I 1_ (11.1". ”1'31“; I 1:31,“..in . -1 .._I1 ..’ 1 1 . . . I 1 .1 .1 II _ .. ‘l ‘1 . -- r' I 3» -:.‘ ”-1111... 'I1 '1 I I 1' I '1‘ .1 I 1 AlgliIIl'd) ' I. II’“I III.I»II I“ -- Ifi'ézfiumx' , . II" V.qu'fl%6 ’i1' AVI- u11u1 :. I‘ IH’n 2.: IIH .3uv I . MII InfrII J‘IIIufl 4;??Iifl'wHII-I3wIiII'I ,. . I ' II r‘ I I I II " I? I :III. I II'..' II‘\I I ' I '° ' .III III' ':.I :' 'fI’I‘. I I II'III' o .' ."b I I 1 1 1 )I II ! ‘(- 1 ..‘. |I‘ I II .II. ‘II.' .I I. 1" - I1I‘I I1I II‘I‘I‘II . “MEI. 1'III I ‘I '_‘ ' 1 I1" II : I ‘1 ‘IIII I . 1 '1 .f I II ' I" 1 1 '1’ 1111....11 .1 I. I‘l‘I: '1 I ' . 11 1 1' . ..1 1 A I .11 "II" 1"?» !'Il‘1I.' 1', 1..“ IIIIII1‘II'II 1 II.. 1 "11:1. l‘ .1: ..‘.I ' ..‘; :1 II1.;.‘V..H:!I1I’;I1}I.‘,I II’ ' Iv: III'III'I 1 :II I . I II’ ”II II I .I' II.‘ III ‘II‘JIIW. III 'I...I'I1 .I I ..1I‘I. I‘PI'II'I,‘ I‘IIII. ‘II II '1:\I'l II'IIIDI II’I'II1III‘III-'I‘:‘I'. 'I ‘,;:I I‘ V I; 5:41”: III“ II" .‘(I I; 1 1‘. I? 1] . II “:11: II' 1.11.0? 1' r1“; ”.1 ”'I :1II 11"111'1.‘II, H II:II1 1' . “11““111101. “1(2-II‘. ..‘ ‘ I.|'IIII H . I. I I: ‘ I ’ II . .f I 'I'II 'II III II 1‘ I:‘:If .!‘ '1 ‘ . .‘I, I: II’: 19115 : il\11.1:'13111 I' 'III'I'IE; I I‘ 'I I" I ' l‘ I I . '..-'.' II‘.- I. “I 'I..' "I II " . ‘I'. 1I'111I I' 4 . II 1. . 1 1 I 1 11 ”1| 1 1 1 .I . ., ”31-: :I1I:_"'I 1.1.1] I: 1' H: .1 I': .1 ‘I‘. ’ I:..;.: . ‘ I II 1111.11... . II». ., 1:1“ 1 . 1 - .' «1-21 .'..I1-. .II -I' III I1}.1I,...111I”11 II' .1I.I',I1I ,:. Ir,‘ II" I'I"(, I.‘.II1-'I' III: 11¢ I, . ”II IIIIIIIII, 1.1.6 3!” I.” .‘,I._ I.II,'.. '11. 1111.11 "II“ 114.3111” .'. ., . II 'I I‘iru' 1 I" I H11, {'19} 'Ahifi’ 1 . | ' 1 I'I' I1?) :IIIIII' II II In": 1.11;... . . I‘1.I’I I {I‘ 1:” .Im1'I 1::1’11 $114111: I‘I'JIF III.- I“ I . 11'1‘ gnévl.” $.Jg1‘ .21 ‘3 ‘I .' II ' IIIII '-' . ‘I -I* I. I: " II' ‘1. 1.1... ‘I;II II‘ 'I I I II IIIIII‘III II I. I III III -IIII III III IIIIII III I ”IIIII'II II III III IIIIIIIIIIIII III I IIIIII IIIIIIIII I'I I'I I'” IIIIIIIII EIIIIIiI |I‘” I II I' ‘I I' I I I. 'I II'I'I : 1,. ”‘0' I1- “‘1 | :I I' :, I. 1‘ 'I I 1' '.I_I'II I: IIIII. I II' III, II. I‘IIIII I. III "1' IVIIIIIII'II II. .' ..' :III... ’II ”II‘" I| I':1I I. 'IJIII"_1 I’:':‘ I - IIIIII 'III I'I I1 II I I!" IIII.I ' .I"III"..IJ..III' ' .I.' .. "' II iI’IIIII 'IIIIIIII .':...III'IIIIIII1III.I.::III III . G II'IIIII II IIIIIIIIII l ‘ I II I.I‘II ! II'IIII II I I I III I It IIII“ {NIIII ‘QIIIIIIIIIIw II I ~ I’ ‘. I;- “III“ .‘I' III III IN :y, I I 'IIII‘I . I III: II'IIIII I I'III'WIIII' -.I,I’IIIIII .. IIIIIIII 1&13‘1' .1 ..’“. IIIIIIIIHUI' (IIIIIII‘ 1:111! ELIIII" . .v1IIIII..(I 1' 1,1 I; I . I 1 I} I'I-I‘IIIIIJL .M‘I‘ II/II’I'I-IIIIIIIIII. IIII'II' II'I III \‘I‘Ilk-WIIIII.II.#II ..(JflIIIII IIUVIIIMBIIII # \« ’ ’m- m-v-y. , f r r , A. W-fif'gf; I “ ll '3 er? r O ‘ ‘ ’tv‘ ‘ ‘ II g“ 3:I; IT‘S '«ts'v '1 ;,.V ‘ «i L . ‘T . N "'2' -* i 4 . . ‘. “P.v fih'. "" \- M 1 H 7 \ I WE‘V— / This is to certify that the thesis entitled BURNOUT: CAUSATION AND MEASUREMENT presented by Beth Rubin has been accepted towards fulfillment of the requirements for M.A. degree in Psychoiogy £1» /\ _ ALP/Xx {/21 {II/\ Major professor I / I ‘, ’7 #\ Date [A r/IK ,/&:z(/C 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution IIIIIIIIIIIII IIIIIIIIIII IIIIIIIII'III 31293 00867 6623 )V153‘_} RETURNING MATERIALS: Place in book drop to “33511155 remove this checkout from .—J_. your record. FINES W11] be charged if book is returned after the date stamped beiow. IW8 0 7 193.1 ' 3*r313 “ 1 1M 2519-“?5“. BURNOUT: CAUSATION AND MEASUREMENT BY Beth Rubin A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1982 ABSTRACT BURNOUT: CAUSATION AND MEASUREMENT BY Beth Rubin This thesis examines the literature on worker burnout and presents an integrative definition of the construct. A general model of the antecedents and consequences of burn— out is explained. Six hypotheses of the relationships of burnout and timing control, role strain and job satisfac- tion are developed, and combined in a process model of burnout. An affectively oriented measure of burnout is developed by exploratory factor analysis, and its structural integrity and reliability are supported by confirmatory factor analy- sis on a second sample. The six hypotheses are tested and supported. Burnout, role strain, job dissatisfaction and lack of timing control are significantly intercorrelated. Burnout and job dissatisfaction appear to be overlapping constructs. Path analysis of the process model suggests that burn- out results from role strain, job dissatisfaction and lack of control over the timing of work. ACKNOWLEDGEMENT S I would like to express my gratitude to all the people without whom this thesis would not have been completed. First, I wish to thank all the members of my family for their unceasing belief in me, and their consistent support, both financial and emotional. I also want to thank the members of my committee, Ben, Neal, Mary and Mary, for their guidance, direction, and the knowledge they shared with me. This thesis would never have been completed were it not for all my friends, who listened to me, consoled me, and encouraged me to continue. Likewise, I wish to thank Mr. Rob Allemier and the Lansing American Tackwondo Associa- tion; the ATA was one of the greatest sources of strength I had, and inspired "perseverance" and "indominable spirit" in this endeavor. Most of all, I want to thank Mary Van Sell for all she has done, but the words needed to describe her gifting and my feelings would not fit onto this page. Suffice to say that, Mary, I am Galatea to your Pygmalion; this erstwhile lump of marble has shape, form and breath in this academic world due to your skill, effort, belief and caring. Thank you for all that you have done. --Beth Rubin ii TABLE OF CONTENTS Page LIST OF TABLES............................. ........ ... v LIST OF FIGURES ..... .... ....... ....................... Vii I. INTRODUCTION.......................... ...... ...... 1 Importance of Burnout .......................... 1 Purpose of this Study..... ..... ................ 3 Summary................. ........... . ......... ‘.. 8 II. REVIEW OF THE LITERATURE .......................... 9 The Burnout Construct..................... ..... 9 Conceptual Confusion ........................ 9 Multi-stage Conceptulaizations.............. 13 The Synthesis of Multiple Perspectives on Burnout.................................. lS COping Strategies.................. ......... l7 Burnout Defined............................. 18 Antecedents and Consequences of Burnout........ 20 Theoretical Antecedents............... ...... 20 Theoretical Consequences.................... 23 Empirical Correlates of Burnout... .......... 26 A Causal Model of Burnout................... 31 Control Over Timing.......... ....... ... ........ 32 Control Over the Work Setting............... 32 General Situational Control.............. 33 Situational Control in Autonomy.......... 35 Summary.................................. 38 Control Over Time in Flexible Working Hours. 39 Coping Mechanisms and Control Over Time..... 43 Summary of Time Control Literature.......... 45 Role Strain.............................. ..... . 46 Relationship of Role Strain with Other Salient Variables.................. ...... 47 Summary..................................... 51 Job Satisfaction......................... ...... 52 Relationship of Job Satisfaction with Other Salient Variables................. ....... 54 iii TABLE OF CONTENTS - continued Page Summary and Hypotheses. ....................... 56 Causal Model of Burnout ....................... 57 Summary........... ........... . ..... . .......... 58 III. METHODOLOGY ...................................... 60 Subjects. ................................. .... 60 Procedure......... ................. . ........ .. 61 Measures........ ......... . .................... 61 The Measurement of Burnout ................. 61 Development of the Affective Burnout Measure.. 64 Other Measures ....... . .............. ... ....... 71 Analyses ...................................... 72 IV. RESULTS.......................................... 77 Confirmatory Factor Analysis ................. p. 77 Reliability of the Measures .......... . ........ 80 Tests of the Hypotheses ...... . ................ 82 Initial Path Analyses. .................. . ..... 85 Revised Path Models ........................... 89 Summary ....................................... 100 V. DISCUSSION ....................................... 101 The Measurement of Burnout .................... 101 Theoretical Implications ...... . ........ . ...... 104 Job Satisfaction ....................... .... 104 Burnout ........ ............ ..... ........... 106 The Relationship of Job Satisfaction and Burnout ................... ..... ..... .... 107 Role Strain..... ........ ..... ..... .. ....... 109 Timing Control........... ......... ......... 111 The Relationship of Role Strain and Timing Control....... ......... .... ......... .... 112 Summary of Theoretical Implications.. ...... 113 Implications for Practice ...... . ....... . ...... 114 Directions for Future Research ................ 115 Summary ....... . .......... . .................... 120 REFERENCES ........................................... 122 APPENDIX--Questionnaire .............................. 136 I iv LIST OF TABLES TABLE 120 13. 14. 15. 16. Definitions of Burnout......... ..... ... ......... Theoretical Antecedents of Burnout. ....... . ..... Theoretical Consequences of Burnout............. Empirical Correlates of Burnout................. Factor Loadings and Reliabilities of Burnout measure scales0000000000000 ..... 0000.00.00 000000 Factor Loadings from Confirmatory Factor AtlaIYSis00000000000000.000000000000000 0000000000 Intercorrelations of Burnout Factors ............ Reliability, Means and Standard Deviations of All Measures. 0000000000000000 0000000000000 000000 Intercorrelations of A11 Scales................. Matrix of Path Coefficients for Initial Model... Input and Reproduced Correlation Matrix of Initial MOdel00000000000000000 000000 0000000 00000 Matrix of Path Coefficients of Initial Revised MOdelO00000000000000000000000000.0000. 000000 0000 Reproduced Correlation Matrix of Initial ReVisedMOde1000000000000000000000000000 0000000 0 Difference Between Observed and Predicted Correlations of Initial Revised Model..... ...... Matrix of Path Coefficients of Secondary Revised Model (#1)....,................... ...... Reproduced Correlation Matrix of Secondary Revised Model (#1)..... ........ ......... ........ Page 22 24 27 68 78 79 84 86 86 91 91 92 95 95 LIST OF TABLES - continued TABLE 17. 18. 19. 20. 21. 22. Difference Between Observed and Predicted Correlations of Secondary Revised Matrix (#1).... Matrix of Path Coefficients of Secondary Revised Model (#2).................... ........... Reproduced Correlation Matrix of Secondary Revised Model (#2).. ............ ... ...... . ....... Difference Between Observed and Predicted Correlations of Secondary Revised Model (#2) ..... Matrix of Path Coefficients of Secondary Revised Model (#3)... .............. . ......... . ........... Reproduced Correlation Matrix of Secondary Revised Model (#3)......... ............ . ....... -. vi Page 96 97 97 98 99 99 LIST OF FIGURES Strategies for coping with stress... ....... .... General model of antecedents and consequences Of burnout00000000000000000000 00000 000000000000 Combined hypotheses of burnout relationships... Influence matrice for the burnout model.. ...... Path diagram of initial model ............ . ..... Initial revised path model. ....... . ........... 2. Secondary revised path model (#1) ..... . ........ Secondary revised path model (#2). ............. Secondary revised path model (#3) .............. vii Page 19 21 58 76 87 89 93 93 93 I. INTRODUCTION This thesis examines the phenomenon of worker burnout and tests causative models of it. The literature on burnout is reviewed, and a definition and description based on previ- ous research is presented. A general model of the potential antecedents and results of burnout is explained. A specific causal model is presented, and tested by path analysis. This section introduces the construct of worker burnout, and gives an overview of the conceptual and measurement issues that will be explored in this thesis. Burnout is defined, and its relationships with other variables of inter- est are presented. The importance of studying burnout and the specific purpose and contribution of this study are described. Finally, the other variables salient to the process of burnout are introduced. Importance of Burnout The subject of worker burnout has received increasing attention and concern in both practitioner and academic publications in recent years. Since the introduction of the term in the mid-1970's, the number of articles published in mass-appeal journals has mushroomed. Such varied publications as Law and Order, Child Care Quarterly, and Journal of Occupational Behavior have featured articles on the subject. Workshops on burnout and its avoidance have become increasingly popular (Cherniss, 1980, 1981); more and more managers and executives in human service fields are showing an awareness of and interest in the problem (Cherniss, 1981; Minnehan and Paine, 1981). In 1981, the first National Conference on Burnout was held, which brought together for the first time the pre- eminent researchers and students of burnout. The papers presented there represented the range of orientations toward the analysis of burnout, from the "Basic Economic and Legal Consequences of Burnout" (Minnehan and Paine, 1981) to the development of an "ecological framework" of burnout (Caroll and White, 1981). This conference represents an attempt to unify the concepts and terms which have been examined in the study of burnout. It also indicates the widespread interest, both on the part of business practitioners and academics, and underscores the importance of understanding the problem. Regardless of any "faddish" element which may be present in the study of burnout, it remains an important topic for research. It is strongly related, both theoret- ically and empirically, to job stress, job dissatisfaction and job performance. The construct may help to clarify the relationship among the three variables. This relationship has received much attention with few conclusive results (Van Sell, Brief and Schuler, 1981). In many cases, especially in samples of human service workers, the link between job Stress and job dissatisfaction as antecedents to poor job performance may be explained by the occurrence of burnout. In addition to the theoretical importance of burnout, the construct has great practical importance: burnout is very costly for the individual, the organization, and, where applicable, the clients. These costs range from the con- crete, monetary expenses of health insurance, sickness, accidents, turnover, decreased performance, lawsuits and administrative costs, to the abstract, personal costs of conflict with family members, emotional angst, depression and loss of self esteem (Minnehan and Paine, 1981). Purpose of this Study This study reviews the multiple theoretical and method- olOgical conceptualizations of burnout, and unifies several related definitions into one. It then describes in detail the elements that comprise this concept. Individual, inter- personal, job, task, and organizational variables which correlate with it are reviewed, and organized into a general model of the antecedents and consequences of burnout. Several specific variables which have theoretical and/or empirically supported relationships to burnout are described, along with the relevant literature. These variables, role strain, global job satisfaction, and control over the timing of work, form the basis of a process model of burnout. Specific hypotheses about their relationships are developed. Methods of measuring and testing both the model and the hypotheses concerning the process of burnout are then described. The need for a unifying definition and conceptualiza- tion of burnout has been discussed in detail by Maslach (1981), and is obvious from even a casual perusal of the literature. Table 1 lists a few of the many definitions which have been used in published analyses of burnout (Maslach, 1981). Because each researcher defines the phenomenon of burnout differently, and most of the research on burnout has been in the form of case studies (Freuden- berger, 1975; Maslach, 1976, 1978; Kahn, 1978), the studies vary widely in content. There are no hard definitional lines between a case of burnout and one of job dissatisfac- tion, emotional breakdown, or inappropriate career choice. Due to the overabundance of case studies, sweeping generalizations about burnout have been made from as small a sample size as one. This study is one of the few to empirically examine an objectively measured conceptualiza- tion of burnout through correlational and regression TABLE 1 DEFINITIONS OF BURNOUT "A syndrome of emotional exhaustion, depersonalization, and reduced personal accomplishment that can occur among in- dividuals who do 'people-work'." (Maslach, 1981) "A process in which a previously committed professional disengages from his/her work in response to stress and strain experienced on the job." (Cherniss, 1980) "A state of physical, emotional, and mental exhaustion marked by physical depletion and chronic fatigue, feelings of helplessness and hopelessness, and the development of a negative self-concept and negative attitudes toward work. life, and other people." (Pines, 1981) “An ongoing process that varies both in its severity and in the number of times the cycle repeats itself...a series of predictable stages...enthusiasm...stagnation... frustration...apathy...burnout means apathy." (Edelwich & Brodsky, 1981) "A malaise of the spirit. A loss of will, an inability to mobilize interest and capabilities." "A pervasive mood of anxiety giving way to depression and despair." "To deplete oneself. To exhaust one's physical and mental resources..." analyses. It is also one of the few to use large samples of employees of several organizations. This paper also presents the evolution of an affectively oriented measure of burnout. Several other concrete, paper and pencil measures of burnout have been developed; one of these has extensive reported reliability and validity coeffi- cients, and has been used by several researchers in the field (Maslach and Jackson, 1981; Golembiewsky and Munzenrider, 1981). This measure, the Maslach Burnout Inventory (MBI), focuses on cognitive reactions to the work situation, the worker's performance, and relationships with clients. This is congruent with Maslach and Jackson's definition of the burnout construct, but not with the construct as defined in this thesis. The measure of burnout presented here focuses on the affective, rather than the behavioral or cognitive, aspects of burnout. This paper examines whether these affec- tive aspects of burnout parallel the cognitive aspects tapped by the MBI in their relationships with antecedent variables. This study endeavors to combine empirical research with the multiple approaches to burnout found in the literature. It describes burnout theoretically and operationally, derived from an empirical testing of the multiple dimensions of burnout described by researchers with varied orientations and conceptualizations of the construct. Like many of the existing studies, it goes beyond the definition of the concept to posit the process through which burnout evolves; unlike other existing studies, it empirically tests compet— ing hypotheses with regression analyses. The next two sections present the construct of burnout in detail, including further description of the conceptual confusion underlying the research. Several views of the construct are combined in a factor analysis of multiple hypothesized aspects of burnout, to produce a single unify- ing concept. Many of the variables which have been posited as causing or resulting from burnout are reviewed, and are organized into a general model of antecedents and conse- quences. These variables are categorized as individual factors, interpersonal factors, job and task factors, and organizational factors. The specific variables of role strain, job satisfaction and control of the timing of work are selected from the literature for analysis. Time control is reviewed in depth, and its relationships to stress coping strategies are described. The relationships of these variables and burnout are explored, and specific hypotheses are derived. These form the basis of the causal model of burnout which is pre- sented. The methods for testing the models are then de- scribed, along with the subjects studied and the procedures used. The evolution of an affective paper-and-pencil measure- ment of burnout is described. The other measures are also presented, along with specific hypotheses and statistical analyses used to test them. Conclusions are drawn from these results, and suggestions are offered for future research. Summary This section introduced the concepts which will be analyzed in this study: burnout and its relationships in the role strain, job satisfaction, and time control. It described the importance of understanding burnout, and the unique contribution which the proposed study will make to the present knowledge about the area. It briefly out- lined the purpose and procedures of this research, and then previewed the model that is tested. II. REVIEW OF THE LITERATURE This section begins with a literature review of the burnout concept, and organizes the variables which have already been researched into a general framework of burnout's antecedents and consequences. The confusion surrounding the concept of burnout is described, and a solution that inte- grates many of the definitions is suggested. Several vari- ables which have been related to burnout are examined, and the research on their interrelationships is reviewed. A causal model and specific hypotheses suggested by the research are presented. The Burnout Construct Conceptual Confusion The present state of burnout research is one of confused definitions and operationalizations (Maslach, 1981). The confusion is due to the recency of the topic, the overwhelm- ing number of case studies, and the flashy nature of the name "burnout" itself. The word has multiple meanings in everyday speech; everyone has an idea of what it is like to be "burned out", but this may refer to a particularly hard day, week, time in one's life, or way of life. 10 According to Maslach (1981), the only elements common to all definitions of burnout are the following: 1) burnout is an individual-level event, process or syndrome; 2) burnout is an internal process of a psychological nature; and 3) burnout involves negative consequences, problems, distress, discomfort, dysfunction, or some combination thereof. Some authors speak of burnout as a process (Cherniss, 1980; Edelwich and Brodsky, 1981; Freudenberger, 1975, 1977, 1981), some as a syndrome (Maslach, 1981; Kahn, 1978), and some as a state (Pines, 1981). Some include behavioral factors, such as personal accomplishment (Maslach, 1981a, Maslach and Pines, 1977, Maslach and Jackson, 1979, 1980) and physical exhaustion (Pines, 1981) as a part of the burn— out construct. Most researchers, however, describe atti- tudinal and/or affective characteristics such as "negative self-concept and ... attitudes toward work life and other people" (Pines, 1981), "emotional withdrawal" (Cherniss, 1980). "alienation from the job" (Daley, 1979), and "cynicism and negativism and a tendency to be inflexible" (Freudenberger, 1977). Other definitional contrasts are evident in the domain of potential victims of burnout. Some theorists include only "individuals who do 'people-work'" (Maslach, 1981), or pro- fessionals (Cherniss, 1980), "workers in the world of helping institutions" (Freudenberger, 1977) or "human service 11 professionals" (Daley, 1979). Others do not limit the domain of victims to any one particular type of job or organization (Freudenberger, 1975; Pines and Aronson, 1981; Golembiewski et a1., 1981). Another element that sometimes appears is a hypothesized cycle of burnout and recovery (Edelwich and Brodsky, 1981), as Opposed to a linear progression that results in severely burned-out workers who have a higher probability of leaving the work environment than other workers (Cherniss, 1980; Shinn, 1981). Some of the many elements that have been used to charac- terize burnout are: emotional, mental, or physical exhaus- tion (Pines and Kafry, 1981), apathy (Edelwich and Brodsky, 1981), withdrawal (Cherniss, 1980), anomie (Chamberlin, 1978), decreased perceived personal achievement, depersonal- ization (Maslach, 1981), negativism, inflexibility, conde- scending attitude (Freudenberger, 1977), inappropriate attitudes (Kahn, 1978), and changes in attitudes towards work and clients (Cherniss, 1980). There is somewhat more agreement on the idea that burn- out is the result of job stresses and their resultant strain on the worker (Cherniss, 1980; Golembiewski et a1., 1981, Maslach, 1978; Kahn, 1978; Pines and Kafry, 1978; Sweeney, 1981; Shinn, 1981). Although burnout and job stress have been found to correlate (e.g., Maslach, 1980; Jackson and Maslach, 1981), there is no empirical evidence for a causal relationship. Job stress is usually defined as the state 12 "when environmental demands tax or exceed the resources of the person" (Lazarus and Launier, 1978). Strain has been defined as "any deviation from normal responses in [a] person" (Caplan et a1., 1975); an alternative definition used is "the immediate, short-term emotional response to ... imbalance, characterized by feelings of anxiety, tension, fatigue, and exhaustion" (Cherniss, 1980). Job strains have also been operationalized as job dissatisfaction, boredom, depression, and somatic complaints; these overlap with several of the definitions of burnout. Many characteristics of individuals, tasks, environments and interpersonal relationships have been hypothesized to relate to burnout. Due to the lack of consensus on a defini- tion of what exactly burnout is, however, many of these characteristics are posited by different researchers as being elements of the burnout construct, causes of burnout, moderators of burnout, and the result of burnout. For example, Cherniss (1980) presents exhaustion as a part of job strain, which he posits as a cause of burnout; Pines and Kafry (1981) define burnout as the combination of three types of exhaustion: mental, emotional and physical. They measure those three aspects as an index of burnout. Maslach (1978) refers to "emotional exhaustion resulting from the stress of interpersonal contact" as burnout. This factor is one of the three subscales in the Maslach Burnout Inventory. Kahn refers to exhaustion as an 13 "uncomfortable physical symptom" that is "often associated" with burnout (Kahn, 1978). Multi-stage Conceptualizations Several theorists have suggested multi-stage models of burnout, wherein the worker proceeds from one to the next in chronological order (Cherniss, 1980; Golembiewski and Munzenrider, 1980; Edelwich and Brodsky, 1981). Cherniss (1980) presents a three-step process, the culmination of which he defines as burnout. The first step is an imbalance between resources and demands (stress). This is followed by a short-term emotional response, characterized by anxiety, tension, fatigue and exhaustion, which he defines as strain. The third and last stage is a set of changes in attitude and behavior, that include emotional detachment, withdrawal, cynicism and rigidity. He does not present empirical evidence to support this theory, however. Edelwich and Brodsky (1981) conceive of burnout as a cyclical process, in which the worker progresses from one stage to the next, and runs through the cycle several times in his/her career. The first stage is one of enthusiasm and positive affect. This is followed by a period of stagnation. People go through a series of predictable stages in relation- ship to their work. The first is enthusiasm, a period of high hopes, high energy, unrealistic expectations, and oversimplification with the job. The second is stagnation, in which personal, financial, and career development needs begin to be felt. 14 This is followed by frustration, in which one ques— tions one's effectiveness and the value of one's efforts in the face of obstacles to meaningful accom- plishment. Frustration ... can lead either back to enthusiasm ... or down to the fourth stage of apathy, and abyss of chronic indifference that defies most efforts at intervention. (Edelwich and Brodsky, 1981, p. 202) Edelwich and Brodsky specifically define the last stage of the four as being burnout: "Frustration is not burnout. Burnout means apathy. Frustration is the experience of learning to cope with limitations.... Apathy, although common, is not normal." (Edelwich et a1., p. 202.) Golembiewski and Munzenrider (1980) adapt Maslach's conceptualization of burnout, but turn it from a single-stage model to a multi-stage one. They theorize that Maslach's three elements of burnout occur in a sequential order. Workers first become alienated from clients, viewing their clients in depersonalized ways, as non-people; they then perceive a lack of personal accomplishment; finally, they become emotionally exhausted. Golembiewski and Munzenrider (1980) performed correlational analyses, with paired com- parisons of levels of different MBI subscales. The results support the hypotheses, such that subjects who scored highly on more "advanced" stages of burnout scored higher on other measures of burnout and strain. There are several aspects of this study which decrease its validity asaatrue test of a multi-stage model. First, the pair comparisons indicate that the more of the 15 subcategories of burnout a worker scores highly on, the more overall burnout s/he seemed to experience. This is not proof that s/he moves from one stage of burnout to the next, although it is interpreted as such. Second, the authors used Maslach's measurement of burnout, which is aimed specifically at human service workers and theoretically limited to that population; however, they administered it to employees who were not human service workers. The deperson- alization scale of the MRI refers specifically to deperson- alizing clients. Golembiewski and Munzenrider (1980) use a sample of workers in "a product-line division" in an "industrialized setting", and use depersonalization in reference to co-workers. This ignores the concept of responsibility for clients who are no longer viewed as people, with the attendant guilt and decreased performance that was intrinsic to the original measure. No other multi-stage model has received any type of test; no longitudinal or path analyses have been performed on any process model of burnout. Both the theoretical and empirical support for multiple stages of burnout are minimal, leading to the rejection of the multi-stage conceptualization. The Synthesis of Multiple Perspectives on Burnout The present conceptualization of burnout, used as the focus of the causal model of burnout, is primarily adapted from Cherniss (1980), Maslach (1981), and Pines and Aronson 16 (1980). Cherniss describes "a transactional process that begins with job stress. Stress contributes to strain, and efforts by individuals to cope with that strain lead to ... burnout." Cherniss, however, conceptualizes burnout as this process itself; the present conceptualization defines burnout as the syndrome resulting from this process. Cherniss defines burnout in terms of emotional detachment and withdrawal; the present model defines other affective and attitudinal reactions. Lastly, Cherniss limits burnout to occurring in "human service" workers; the present model does not. Cherniss describes in detail the severe stresses that exist for human service workers; case study after case study documents the difficulties involved in their work (e.g., Maslach, 1976; Daley, 1979; Kermisch and Kushin, 1969; Freudenberger, 1975). However, there are undeniably other occupations typified by extreme levels of stress, such as air traffic controllers. Workers in these professions, however, may not have as large a decrease in the quality of their performance due to the affective and attitudinal changes involved in burnout. The negative affective changes, while uncomfortable and disturbing, would not automatically decrease the performance level of a secretary or an air traffic controller as much as it might that of a teacher, a therapist or a police officer. Others may be no less burned out, but the effects may not be so noticeable due to the 17 type of work that is performed. Human service work depends on good communication, openness, and trust, which are very likely to suffer from the negative affect involved in burn- out. Maslach refers to burnout as a syndrome of negative attitudes and affects, which she lists as depersonalization, emotional exhaustion and decreased personal competence. Golembiewski describes burnout as "implying inadequate coping with job stressors and their derivative strain ... which ... surfaces in 'inappropriate attitudes toward clients and toward self; often associated with uncomfortable ... emotional symptoms'" (Golembiewski, 1981, quoting Kahn, 1978). He also includes physical symptoms along with the emotional ones in his conceptualization of burnout; the present model portrays physical effects as second level results of burnout, rather than as part of the construct it- self. A worker can be burned out and not be physically exhausted. CopipgfiStrategies "Coping refers to ... efforts taken to manage demands and conflicts which tax or exceed a person's resources.... Coping may be cognitive, behavioral, or a combination" (Cherniss, 1980, p. 45). Lazarus and Launier (1978) suggest that the type of situation in which a person operates affects what type of coping behavior s/he will tend to use. 18 Situations of high conflict, ambiguity or helplessness will result in intrapsychic, passive coping strategies; either the person is seeking information, s/he is incapable of taking effective action to reduce stresses in the environ- ment, or the situation is too unclear for effective action to be identified. Pines and Aronson (1980) empirically examine multiple means of coping with stress, and find that direct and active c0ping strategies have a lower correlation with Pines' (1978) Tedium measure of burnout. The strategies with the highest correlations with measured burnout are indirect and passive ones, such as taking drugs or using alcohol. Pines and Aronson develop a two-by-two matrix that categorizes all the strategies along two orthogonal, dichotomous variables; activity/passivity and directness/indirectness. This matrix is presented in Figure 1. Direct strategies are defined as those applying to the external environment, as opposed to indirect ones which apply to one's own behavior or emotions. Active strategies are those which entail confronting or attempting to change the source of stress or oneself, as opposed to passive ones which entail avoidance or denial by cognitive or physical means. Burnout Defined One of the more commonly presented causes of the burn- out construct which has received correlational support is 19 ACTIVE PASSIVE *Changing the source *Ignoring source of of the stress of the stress DIRECT *Confronting the source *Avoiding source *Adopting a positive *Leaving attitude *Taking about the source *Alcohol or drugs of stress *Getting ill INDIRECT *Changing self *Collapsing *Getting involved in other activities Figure 1. Strategies for coping with stress (Pines and Aronson, 1980). the existence of job stressors, such as overload, difficult client population, difficult environment, conflicting demands, red tape, paper work, role conflict and ambiguity (Pines, 1981; Maslach and Jackson, 1979, 1980), which result in worker strain. There is also evidence supporting the idea that the relationship between strain and burnout is moderated by the type and success of the coping strategies used (Shinn, 1981; Pines and Aronson, 1981). Following the research, a preliminary definition of burnout is evolved: burnout is the syndrome of negative changes in affects and attitudes which result from the ineffective coping with the strain produced by job stresses. The specific affects and attitudes which undergo nega— tive changes are specified later in this thesis; they were determined by empirical analysis of this definition. 20 Antecedents and Consequences of Burnout Many factors have been posited as relating to burnout, and can be categorized into antecedents and consequences of burnout. Although many of the factors may act as moderators, there is not enough agreement as to the definition of the burnout construct itself to make such a complicated analysis fruitful. The antecedents of burnout are organized into a categorization system based on levels of analysis: individ- ual, interpersonal, job/task and organizational factors. The consequences are organized into the categories of inter- personal variables and three types of individual variables: affective/attitudinal, health-related, and work behaviors. These combine to produce organizational level variables, such as organizational costs (Minnehan and Paine, 1981). This categorization system is presented in Figure 2. Theoretical Antecedents According to the literature, burnout results from many different factors. A partial listing of these is presented in Table 2. Some of the individual characteristics which have been proposed to be associated with burnout are single marital status, female gender, inadequate training, and hav- ing unrealistic job expectations (Edelwich and Brodsky, 1981; Cherniss, 1981; Wilder and Plutchik, 1981). Most hypothesized antecedents of burnout are gathered under the .usocusn mo moososvmmcoo can mucmbmooucm wo Hmpoe Hmumcmw .m musmflm 21 muouomm HMGOHDMNflcmmHO 14 mammmumm< muouomm HMHOH>mnmm Umuwamulxuoz 1 uuuuuuuuuuuuuu 1 muouomm umumamuusuammm can Hwoammnm muouomm m>nuommm¢ can Hmcnnsuflbua /I‘{\ muouomm anaemummuwusH saouenTesuoa IeAeq Ienptnrpul Hm>mq Hmaoa>uoau UDOGHDQ MO m OOGQQUOmCOU muouomm Hmsea>ueaH muouomm HmcomummumusH usosusm muouomm xmme AQIII. can now mnouomm HMCOADMNHcmmHO usosusm mo mucmpmomusfl 22 TABLE 2 THEORETICAL ANTECEDENTS OF BURNOUT ORGANIZATIONAL FACTORS Formalization Centralization Bureaucratization JOB AND TASK FACTORS Control of work environment Ability to take "time-outs" from stressful task Availability of training in job Workload Specific task performed Physical environment (noise level, crowdedness) Autonomy Task variety Task identity Task significance Closeness of supervision Competence of supervision Feedback from the job itself Feedback from supervisors or peers INTERPERSONAL FACTORS Communication with peers or supervisors Support from peers or supervisors INDIVIDUAL FACTORS Bureaucratic/professional orientation Stress tolerance Personality type (A vs. B) Amount of education Neurotic anxiety Locus of control Perceived role conflict Perceived role ambiguity Flexibility Age Sex Marital status Self esteem Perceived competence 23 umbrella of job stresses; however, these vary in perspective from author to author. Role strains have been suggested, such as role overload (Carroll and White, 1981), and various aspects and dimensions of role conflict (Carroll and White, 1981; Shinn, 1981; Golembiewski, 1981). Task factors such as insufficient autonomy, flexibility in job and scheduling, ability to take time-outs, feedfack, participation in decision making, control over environment, boundary spanning positions, and type of client (Edelwich and Brodsky, 1981; Golembiewski, 1981; Shinn, 1981; Carroll and White, 1981; and Pines, 1981) have all been discussed as causing burnout. Interpersonal factors such as relations with co-workers, climate of trust, quality of supervision, and distance from role sender (White, 1981; Shinn, 1981; Golembiewski, 1981; and Pines, 1981) have been suggested as antecedents of burn- out. Organizational factors, such as amount of red tape, bureaucratization, centralization of decision making, opportunity for career advancement, shape of organization, and equitability of resource sharing (Edelwich and Brodsky, 1981; Golembiewski, 1981; and Pines, 1981) have also been suggested. Theoretical Consequences Almost as many consequences of burnout as causes have been suggested in the literature. A partial listing of these is presented in Table 3. On an individual/affective 24 TABLE 3 THEORETICAL CONSEQUENCES OF BURNOUT INDIVIDUAL LEVEL ATTITUDINAL AND AFFECTIVE FACTORS Apathy Alienation Nervousness Irritability Depression Cynicism Intention to quit Job involvement (low) Job dissatisfaction Perceived ineffectiveness Tension PHYSICAL AND HEALTH-RELATED FACTORS Use of aIcohol Use of drugs Use of tranquilizers Accident rates Exhaustion Insomnia High blood pressure Coronary heart disease Headaches Backaches Weight loss Weight gain WORK-RELATED BEHAVIORS Decreased productivity (quality and quantity) Absenteeism Turnover Tardiness Task completion Early retirement Employee theft ORGANIZATIONAL FACTORS Low morale Departmental relation- ships (hostility, competition, lack of cooperation, lack of communica- tion) Organizational effec- tiveness ' Incidence of involun- -tary turnover INTERPERSONAL FACTORS Family conflict Trust in supervisors or peers 25 level, distancing, paranoia, depression, martyrdom, tension, anger, decreased emotional control and apathetic attitudes have been posited as "indicating" burnout; whether these are part of the syndrome, coincidental with it, or resulting from it is never made clear (Minnehan and Paine, 1981; Edelwich and Brodsky, 1981). Freudenberger includes the perceptions of worthlessness, helplessness, hopelessness, depression, and martyrdom in his clinical description of burned-out cases (Freudenberger, 1981; Minnehan and Paine, 1981). Attitudes that are hypothesized to increase with burnout are boredom, cynicism, distrust of management and peers, decreased flexibility, and decreased tolerance for ambiguity (Freudenberger, 1981; Minnehan and Paine, 1981). Interpersonal factors that have been suggested to result from burnout include an increase in isolation from peers and clients, and an increase in interpersonal conflicts with staff members and with family (Minnehan and Paine, 1981; Maslach and Jackson, 1981). A great number of health-related factors are hypothe- sized to result from burnout. Most of these are empirical correlates of job stress. They include: fatigue, illness, headaches, sleep disturbances, sudden gains or losses in weight, injuries, muscular pain, premenstrual tension, psychosomatic illnesses, and use of alcohol and drugs (Minnehan and Paine, 1981; Edelwich and Brodsky, 1981). 26 There are multiple job behavior variables that have also been posited to result from burnout. These include turnover, absenteeism, tardiness, failure to perform required work, and decreased quality and quantity of performance (Edelwich and Brodsky, 1981; Pines, 1981; Minnehan and Paine, 1981). Empirical Correlates of Burnout Recently evolved paper-and-pencil measures of burnout have allowed correlational studies examining the relationship of burnout and individual, interpersonal, organizational and task characteristics. Although no longitudinal designs or other tests of causality have been performed for any of these factors, some of them have been empirically correlated with objective measures of burnout. These include: task/ job variables such as autonomy, variety and feedback; job context factors such as the ability to take short work breaks, the ability to change the physical environment, and the physical space of the work setting; interpersonal variables such as relations with peers and supervisors, and support from peers and supervisors; and individual variables of four types: attitudes/affects such as job satisfaction, health- related problems such as use of alcohol and tranquilizers, demographic variables such as age and sex, and work-related behaviors such as absenteeism and turnover (Maslach and Jackson, in press; Pines, 1981). Empirical correlates of burnout are presented in Table 4. 27 TABLE 4 EMPIRICAL CORRELATES OF BURNOUT JOB AND TASK VARIABLES Inability to take "time-outs" Lack of technical support Low autonomy Lack of variety in work environment Overload (quantitative and qualitative) Difficult type of work (direct contact with clients, severity of client problems) Low task significance Low feedback (from the job itself; feedback about performance) Poor physical environment (noise levels, crowding, architectural dysfunction) Conflicting demands Red tape Paper work INTERPERSONAL VARIABLES Communication problems Administrative interference in goal achievement Poor relations with peers Poor relations with supervisors Problems with family and friends INDIVIDUAL VARIABLES ATTITUDES AND AFFECT WORK-RELATED BEHAVIORS Intentions to quit Absenteeism Low job satisfaction (with work, Turnover life, oneself, peers) Coping stretegies Low need satisfaction (active/passive, (with existence, growth, direct/indirect, and relatedness needs) instrumental/ Hopelessness palliative) Loss of idealism PHYSICAL AND HEALTH-RELATED DEMOGRAPHIC VARIABLES VARIABLES Age Use of alcohol Sex Use of tranquilizers Poor health Sleep problems Headaches Backaches Stomach aches Loss of appetite Nervousness Marital status Amount of education 28 Maslach and Jackson (1981), and Jackson and Maslach (In press) found significant correlations between the emotional exhaustion subscale of the Maslach Burnout Inventory and spouse ratings of policemen's being upset or angry, tense or anxious, physically exhausted and complaining about work. All three subscales were significantly correlated with case- load. Emotional exhaustion correlated with spending time in direct client contact. All three subscales correlated nega- tively with feedback from the job itself and job satisfac- tion; the Involvement subscale correlated positively with spending time with others. Antecedent demographic variables that were correlated with burnout included sex; women scored higher than men on emotional exhaustion, while men scored higher on depersonal- ization and personal achievement (the latter of which indi- cates a lack of burnout). Age correlated with all three sub- scales such that younger people experienced more burnout. Married people scored lower on emotional exhaustion than single or divorced people. Level of education was related such that the more education, the greater emotional exhaus- tion, and the less depersonalization; high degrees of educa- tion scored highest on personal accomplishment, followed by low degrees, followed by median (college) degrees. Other variables that correlated with the MBI included subscales of the Job Diagnostic Survey: "growth satisfac- tion" was negatively correlated with all three subscales, as 29 was "knowledge of results" and "peer and co-worker satisfac- tion". Burnout was correlated with intention to quit; absen- teeism was correlated with the depersonalization subscale only. Depersonalization was correlated with reports of fewer friends, being absent from family celebrations, not sharing feelings or friends with the spouse, and being emotionally distant from children. The more burnout on all subscales,the more reports of the worker getting angry at his/her spouse or children. Health-related variables that correlated with the MBI included insomnia and drinking: these correlated with the emotional exhaustion subscale. Use of tranquilizers corre- lated (negatively) with the personal accomplishment subscale. Golembiewski and Munzenrider (1981) found significant correlations between burnout and job involvement, job tension, trust in supervisor and fellow employees, and participation; they do not mention the directionality of these correlations, however. Significant relationships were found for the Job Descriptive Index subscales of satisfaction with work, super- vision, and co-workers, as well as with the Job Diagnostic Survey measures of meaningfulness of work, responsibility for work, general satisfaction, internal work motivation, growth satisfaction, job security, co-workers and supervision. Pines (1981) presents a review of her research, which utilized her Tedium measure. This operationalizes burnout as 30 physical, emotional and mental exhaustion. She found signifi- cant correlations with turnover rates, tardiness, intention to quit, poor physical health, sleep problems, amount of alcohol consumed, on-duty headaches, loss of appetite, nervousness, backaches, and stomach aches. Tedium was nega- tively correlated with satisfaction from work, life and one- self, and positively with hopelessness and loss of idealism about work. Antecedents of burnout that were significantly corre- lated with Tedium included: task antonomy, variety, overload and significance; perceived actualization and growth; the structure, noise and spacing of the physical environment, the flexibility to change elements of the physical environ- ment, and the ability to take "time-outs". The number and problems of the clients were also related to her measure of burnout. Interpersonal elements included co-worker work relations and support from co-workers. Rewards, support, challenge and feedback from supervisors were also related. Organizational variables included perceived amounts of red tape, paper work, communication problems, amount of rules and regulations, participation in decision making, and ability to influence policy. Lastly, role conflict and ambiguity correlated with burnout, as did status disorder. All of these studies are correlational in nature. Causal models of burnout have been developed, but have not been tested empirically through use of longitudinal or path 31 analysis. Therefore, no causal inferences can be drawn on the basis of the correlational results reported. Some vari- ables, such as the demographics, are clearly antecedent vari- ables; most, however, can only be categorized as antecedent or consequential on the basis of logic. The use of a variety of measures which tap different conceptualizations of burnout also makes explicit categorization difficult. The only conclusive statement possible is that these variables are related to burnout; the state of knowledge about burnout is insufficient for more than logical guesswork about the exact form and direction of the relationships. A Causal Model of Burnout The review of the burnout literature illustrates the absence of an empirically supported causal process leading to burnout, despite the number and variety of variables that correlate with it. Several elements have been chosen for this study from the theoretical explanations and data avail- able in the literature. These were selected on the basis of their theoretical importance, their importance in interviews with human service and clerical employees, and their inclu- sion in the synthesized definition of burnout. These vari- ables are: control of work timing, role strain, and global job satisfaction. The next section examines these variables, reviews their relationships with burnout, and derives hypotheses concerning their interrelationships. These are combined into a causal model of burnout. 32 Control Over Timing Control over the timing of work can be viewed from several perspectives, each having an impact on burnout. First, timing control can be seen as a type of control over the work setting. It includes two relevant areas of the literature: general situational control, and controlling the work setting as a subset of autonomy. Second, timing control appears as a part of the flexitime research, and can be viewed as a constraint on stress coping strategies avail- able to people. The research on each of these aspeCts will be briefly reviewed, and the impact of timing control assessed. Control Over the Work Setting Several researchers describe lack of control of the work situation as one of the causes of burnout (Cherniss, 1980; Carroll and White, 1981; Golembiewski, 1981; Pines, 1981). One way to control the work situation is through self-determination of which aspect of work is being performed at any given time, and through the ability to remove oneself from the work setting. This type of situation control has been studied as the general control over one's environment and the specific control over the work setting; control of work scheduling is a frequently used operationalization of autonomy. 33 General Situational Control The importance of control over one's immediate sur— roundings has been shown in many studies of both humans and animals. Lack of control over one's environment has been related to depression (Rehm, 1977; Abramson, Seligman and Teasdale, 1978), illnesses (Suls and Mullen, 1981), decreased motivation and depressed affect (Glass, Singer and Friedman, 1969; Rehm, 1977; Abramson et a1., 1978; Langer and Rodin, 1976; and Lefcourt, 1973), and decreased happiness, activity, and involvement (Langer et a1., 1976). Lack of control over oneself and one's own actions is also related to depression (Rehm, 1977), which, like burnout, is marked by negative changes in affect. Langer and Rodin (1976) explored the behavioral effects of situational control of nursing home residents. They verbally emphasized the decision-making ability and options available to one group of residents, while giving them complete control over and responsibility for the care of a plant. A comparison group of residents received a comparable verbal message, without the emphasis on decision-making, and received a plant that was taken care of by a nurse. The experimental group members had a significantly greater increase in self-reported happiness and activity level than the comparison group, and greater increases in objective measures of their time spent in proactive interactions. 34 Control over negative as well as positive events is an important factor in depression and illness. Suls and Mullen (1981) present evidence that perceived control of negative, life changing events is negatively related to the number of illnesses subsequently contracted by college undergraduates. Perceived control over both positive and negative events is presented as the causal factor in the "learned helplessness" theory of depression (Seligman, 1972; Abramson, Seligman and Teasdale, 1978). In the modified depression model, the real- ization that "the probability of an outcome is the same whether or not a given response occurs" (Abramson et al., p. 51) results in helplessness and depression. Depression has four elements: deficient motivation, cognitive processing, self esteem and depressed affect. The more certain the sub- ject is of his/her helplessness (lack of control), the greater the deficits in each of the four elements (Abramson et a1., 1978). Rehm (1977) adds a new element in the helplessness model of depression by stressing the importance of self-control. His formulation differentiates between the depression caused by perceived helplessness and that caused by perceived incompetence. In all of these models, an inability to control both positive and negative events of personal importance results in depression and its attendant behavioral and psychological deficits. 35 Lefcourt's (1973) review of the helplessness literature focuses on the perception of control of events; the greater the perceived control, the less acute the depressive symptoms and the less aversive the perception of negative conditions. In summary, the research on general situational control has related it to illness and depression (Suls and Mullen, 1981; Rehm, 1977; Abramson, Seligman and Teasdale, 1978; Glass, Singer and Friedman, 1969; Lefcourt, 1973; Langer and Rodin, 1976). One of the primary symptoms of depression is depressed affect, which is similar to the negative changes in affect that constitute burnout. Exactly which feelings change for the worse in depression are not identified; those which change for the worse in burnout are described in the next section of this thesis. Situational Control in Autonomy Control of one's situation and actions have beenwexamined imithe Industrial/Organizational Psychology and Organizational Behavior literature under the heading of autonomy. Autonomy originally referred to that control over task content which led to perceived responsibility by the employee (Hackman and Oldham, 1975). Recently, the specific aspect of control over the context of the work situation has been included in measures of autonomy; in some cases, it has been separated out of the broader measure for specific examination. Control as a subset of autonomy on a job has been re- lated to job performance (Birchell and Wild, 1976; Marshall and COOper, 1978; Karasek, 1979) and stress levels (Karasek, 36 1979; Marshall and Cooper, 1978). Autonomy as a general construct has been related to increased job satisfaction, motivation, productivity, employee growth, life satisfaction, self esteem, and depression (Beehr, 1976; Hackman and Lawler, 1971; Hackman and Oldham, 1976; Argyris, 1964; Likert, 1961; McGregor, 1960). In their study of the relationships of perceived job attributes and objective performance measures, Birchell and Wild (1976) broke autonomy down into "worker responsibility" and "control over work activity". Both of these aspects of autonomy were significantly correlated with improved job performance, as was "self-actualization". Birchell and Wild, however, emphasize that it is the workers' perceptions of autonomy which lead to increased performance, rather than objective autonomy. As Lefcourt (1973) made clear, the "illusions of control and freedom" have a very powerful impact on actions. Several studies have operationalized the context control subset of autonomy as control over work scheduling and timing. Control over the timing of work includes exactly Epigp hours are worked, but not the overall number of hours. Control over the number of hours is having control over the demands of a job, as opposed to the decision latitude (Karasek, 1979). Controlling the number of hours worked means controlling the goals set for the employee, and the amount of overload; these are parts of the job content, and 37 are sources of job stress and strain on the worker. Control over the timing of work is context control; it acts as a modifier on the relationship between strain and burnout, rather than as a causal factor in producing strain. Time control has been related to burnout, stress and job satisfaction in a few studies. Marshall and Cooper (1978) looked at job characteristics that differentiated between the job satisfaction and job stress on lower, middle and upper-level managers. The stressors were grouped into four factors: job-intrinsic qualities, work role (including "making important decisions"), behavioral restrictions due to being "trapped" in a large organization, and the "work- home interface". The latter factor, which was empirically separated from the job-responsibility aspects of overall job autonomy, contained items describing "more pressure from working long hours, having to spend leisure time on work, business travel and the conflict of work with home demands." All the patterns of stress did not increase linearly with the level of management; both the upper two groups reported equal degrees of spending leisure time on work, and middle managers reported more pressure from longer hours. One of the consistent correlates of burnout is an in- ability to take brief "time-outs", during which an employee can get away from a stressful task for a short while (Maslach and Pines, 1977; Pines, 1980). Karasek (1979) presents a cross-cultural study which combines time pressures and 38 limitations with the amount of work required to create a scale of "job demands". He contrasts these with "decision latitude“‘ and "decision authority" (including "freedom as to how to work"). He examines their relationships with depression, exhaustion, job and life satisfaction and job strain. In samples within both the United States and Sweden, Karasek found that the combination of high job demands (little time and many pressures) and low decision latitude and authority (constraints on autonomy, both in job intrinsic and job process decisions), had the highest correlation with reports of depression, exhaustion, and anxiety, and job strain. A worker's time is divided among leisure, work and family, which may result in role conflict, alienation, and a host of other negative affects and perceptions such as ten- sion and poor familial relationships (Kanter, 1977; Korman and Korman, 1980; Haavio-Mannila, 1971; Willmott, 1971). Control over when the employee works should allow him/her to make the interface between work and family life more smooth; the ability to alter one's schedule enough to take children to school, spend a long weekend away, or make appointments for special occasions may go a long way in decreasing role conflict (Kanter, 1977; Ronen, 1981). Summary In summary, control over timing has been examined as an aspect of control over the work setting. This has been 39 especially apparent in the research on autonomy. Control over the work setting is related to low stress levels and job performance (Karasek, 1979; Marshall and Cooper, 1978; Birchell and Wild, 1976). Control over timing of work is related positively with job satisfaction, and negatively with job stress, depression, burnout, role conflict, tension and poor familial relations (Marshall and Cooper, 1978; Karasek, 1979; Maslach and Pines, 1977; Kanter, 1977; Haavio-Mannila, 1971; Willmott, 1971; Korman and Korman, 1980; Ronen, 1981). Control Over Time in Flexible Working Hours There is one set of research findings which examines the results of increased control over one's time, without changing the amount of time that is worked; it examines a change in control over context without changing the content of the work, the goals, the demands or the workload. This is the research on flexible time systems, a recent work life intervention that has become increasingly popular with organ- izations in the last five to ten years. There are several variations of the flexitime system, but its basis consists of a set of "core" hours from mid- morning to mid-afternoon, during which all employees must be on the job. Before and after the core are a range of two to three hour periods, during which employees begin and leave work. The total number of hours worked each pay period 40 remains unchanged. Some systems have constant lunch hours, and others have variable ones; some allow daily fluctuations in beginning and leaving times, while others have employees pick a set of hours for each month; some allow hours to be credited to employees over a period of weeks, to be taken later as an extra day off, while others do not. Some of the proponents of flexitime claim increases in job satisfaction, employee morale, work climate, quality of leisure time, tardiness, absenteeism, lost work hours, and quality of the work-family interface (Donahue, 1975; Hopp anui Sommerstad, 1975; Walker, Fletcher and McLeod, 1975). Some of these factors, particularly the affective and subjective ones, have been empirically related to the use of flexitime (Schein, Maurer and Novak, 1977; Golembiew- ski, Hilles and Kagno, 1974; Ronen and Primps, 1980; Evans, 1973; Ronen, 1981). Schein, Maurer and Novak (1977) examined supervisors' responses to flexitime in twelve units of an insurance company. They factor analyzed the responses, and reported large positive reactions to the effect of flexitime on the factors of employee productivity and effectiveness, overtime and scheduling, employee honesty and morale, administrative concerns, employee time handling, and employee work habits. Evans (1973) compared the levels of satisfaction with both work and leisure time along five categories: "using capabilities, accomplishment, prestige, social activity, and 41 leisure". He examined the differences between the actual and preferred levels of these qualities in a flexitime and a control group of non-supervisory personnel. There were significant differences between the flexitime and control groups along all five categories in work satisfaction, but not in leisure satisfaction. Golembiewski, Hilles and Kagno (1974) examined atti- tudes about work in a longitudinal study of non-supervisory personnel. Using two flexitime groups and one control, they found that after six months and after a year, flexitime employees reported greater satisfaction with work hours, greater ease in handling personal business, fewer problems with traffic congestion, and improved "impact of work-hour policy on personal productivity". Orpen (1981) used an experimental design to examine the changes in satisfaction and productivity after six months in randomly assigned flexitime and control groups of clerical workers. He found significant differences in the change in overall job satisfaction. Ronen and Primps (1980) describe the results of twenty-five studies performed in various organizations, describing both objective and subjective data in four general areas. In studies using objective data, seven out of twelve found a decrease in absenteeism, and nine out of nine found a decrease in tardiness. In studies using subjective (attitudinal) data, fourteen out of seventeen reported improvements in the control of work hours and quiet 42 time. Seventeen out of fifteen found a positive attitude about flexitime. Subjective data on absenteeism and tardi- ness resulted in six studies out of six showing improvement in the former, and five out of five showing improvement in the latter. All nine studies that examined attitudes about individual usage of time found improvements. In general, the research on the use of the flexible time systems supports the hypothesis that it causes increased job satisfaction, even over long time periods. The preponder- ance of data also supports the conclusion that flexitime decreases absenteeism and tardiness, but does not decrease productivity; it increases perceived control of work hours, and improves attitudes about use of time. The reason for both these affective and behavioral changes, however, is not clear. The reports of "improved control over work scheduling and work process" (Ronen and Primps, 1980), "a feeling of freedom" (Walker, Fletcher and McLeod, 1975), and increased "degree of participation in decisions about work assignments" (Golembiewski, Hills and Kagno, 1974) all indicate an increased sense of control over the work situation. Research in altering work hours without giving employees control over them, such as is done with a four—day, forty- hour work week, indicates that there is a short-term improve- ment in satisfaction and stress-anxiety levels, but no differ- ence in absenteeism or job performance. Long term 43 examination, however, has found no difference from a com- parison group in satisfaction, stress-anxiety, absenteeism or performance (Ivancevich and Lyon, 1976). These findings further support the causative function of the increased control of the work environment in the increases in satis— faction and decreases in absenteeism and tardiness. In summary, the flexitime research shows a correlation between use of flexitime and increased job satisfaction, morale and work climate, and decreased tardiness and absen- teeism. There is strong support for the importance of timing control in creating these positive responses to flexi- time systems: workers report increased control, and the positive changes do not last when hours are changed but the employees have no control over them. Coping Mechanisms and Control Over Time The amount of control workers have over the timing of their jobs may have two effects. First, it can be a type of decision latitude: it is a way for workers to control their environment, and perceive themselves as in control rather than helpless. Second, it affects the types of coping strategies that may be used to deal with job stress. According to the research previously described, the perception of control over the work environment should lead to increased job performance and job satisfaction, and to decreased absenteeism, incidence of depression, exhaustion, 44 anxiety and job strain (Karasek, 1979; Birchell and Wild, 1976; Ronen and Primps, 1980; Ronen 1980; Evans, 1973; Orpen, 1981). Active, direct strategies are the most effective in coping with strains, but there are limits on the ability of workers to enact effective, outer-directed coping tech- niques. One major set of limits is the timing of work. As has been suggested, controlling this can allow an employee to actively cope with strains such as inter-role conflict by smoothing the interface between his/her work and family or private lives (Ronen, 1980). This has been empirically supported by the significant correlations found between incidence of family problems and burnout (Maslach and Jackson, 1980). One aspect of employee control of timing is the ability to take brief "time-outs", during which time the worker can leave the immediate setting. These allow the direct/passive coping response of temporarily leaving the field to avoid the sources of stress for a while. As has been mentioned, Maslach and Jackson (1980) found a strong negative correla- tion between the ability to take "time-outs" of about 10 minutes duration, and incidence of burnout. According to the definition of burnout, workers who experience a great deal of role strain may experience burnout. They will do so if they employ coping strategies that do not effectively deal with the strain they experience; that is, 45 if they use passive, indirect coping strategies. Having less control over the timing of work will decrease the number of direct and active strategies a worker is able to use. This will increase the probability that she/he will cope in- effectively, and experience burnout. Having less control over timing will also give the worker less control over the context of his/her work, which has deleterious effects on work and life satisfaction, depression, alcohol and drug abuse. Summary of Time Control Literature Overall, the various aspects of timing control show relationships with job stress, job dissatisfaction and burnout. Lack of control over one's life situation leads to depression, which shares the negative affect of burnout. Lack of control over the work setting, and specifically the timing of work, is related to job stress, role strain, depression, burnout, and job dissatisfaction. The control over timing that characterizes flexitime is related to job satisfaction and decreased withdrawal behaviors. Control over timing also affects the stress c0ping strategies that are available, which directly impacts on the incidence of burnout. From this research, specific hypotheses of the relation- ships of timing control, job satisfaction, role strain and burnout can be derived. 46 Hypothesis 1: The more control over the timing of work an employee has, the less burnout he/she will experience. Hypothesis 2: The more control over the timing of work an employee has, the more job satisfac- tion he/she will experience. Hypothesis 3: The more control over the timing of work an employee has, the less role strain he/she will experience. Role Strain There are many ways of viewing the job-related stresses that employees are subject to. One viewpoint which has received increasing attention in recent years is the concept of job encumbents as occupying multiple roles, which may give rise to role conflicts and/or role ambiguity (Katz and Kahn, 1966). This seems to be especially salient when look- ing at job stresses from a burnout perspective, because so many of the case studies describe workers who are frustrated by performing tasks that they did not expect to, or are torn between conflicting responsibilities, or who find themselves acting contrary to their expectations of themselves. These all seem to involve sets of expectations which conflict with one another. A role is "a set of expectations applied to the incum- bent of a particular position by the incumbent and by role senders within and beyond an organization's boundaries" (Van Sell, Brief and Schuler, 1981, p. 43). Role strain can 47 be caused by two types of situations, role conflict and role ambiguity. Role conflict occurs when there is "an incon- gruity of the expectations associated with a role (Van Sell et a1., p. 44)." Several types of role conflict have been identi- fied: a) intra-sender role conflict--incompatible expectations from a single role sender; b) inter-sender role conflict--expectations from one role sender which are incompatible with those from another role sender; c) person-role conflict--incompatibility between the expectations held by the role incumbent and the expec- tations otherwise associated with his/her position; d) inter-role conflict--role pressures stemming from one position; and e) role overload--expecting the role incumbent to engage in several role behaviors, all of which may be mutually compatible in the abstract, with- in too short a time period.... (Kahn et a1., 1964.) ... Generally, role ambiguity has been defined as the degree to which clear information is lacking regarding a) the expectations associated with a role, b) methods for fulfilling known role expectations, and/or c) the consequences of role performance (Kahn et a1., 1964; Van Sell et a1., 1981). Relationship of Role Strain with Other Salient Variables Role conflict and ambiguity have been found to relate to many of the same variables which are correlated with burnout. Many studies have examined the relationships of individual, interpersonal, job-related, and organizational variables with these two role stresses, both as antecedent and consequential factors. There are obvious similarities in the variables related to burnout and those related to role strain. This provides circumstantial support for the hypothesis that burnout results from role strain that is 48 ineffectively c0ped with. On an individual level, job tenure has been negatively correlated with role conflict (Corwin, 1961) and role ambiguity (Organ and Greene, 1974). The interpersonal variables of social support from supervisor and co-workers have been negatively related to role ambiguity (Caplan, Cobb, French, Van Harrison and Pinneau, 1975). The vast majority of the research on antecedent factors in role stress examines job and task characteristics. Autonomy has been negatively correlated with role conflict (Brief and Aldag, 1976), as has involvement in decision making (Belasco and Alutto, 1969). The timing of work has been examined extensively. Work load, variance in workload, and the number of hours of unwanted overtime correlate posi— tively with role ambiguity; the latter is also correlated with role conflict (Caplan et a1., 1975). The best supported consequence of role conflict and ambiguity is job dissatisfaction. Job dissatisfaction has been correlated with role conflict (Beehr, Walsh and Taber, 1976; Brief and Aldag, 1976; Caplan, Cobb, French, Van Harrison and Pinneau, 1975; Gross, Mason and McEachern, 1958; Hall and Gordon, 1973; House and Rizzo, 1972; Johnson and Stinson, 1975; Kahn, Wolfe, Quinn, Snoeck and Rosenthal, 1964; Miles, 1976; Rizzo, House and Lirtzman, 1970; Sorensen and Sorensen, 1974; SzilagyianuiSims, 1975; Tosi and Tosi, 49 1970), and with ambiguity (Beehr, 1976; Beehr, Walsh and Taber, 1976; Caplan et a1., 1975; French and Caplan, 1972; Greene, 1972; Hamner and Tosi, 1974; House and Rizzo, 1972; Johnson and Stinson, 1975; Kahn et a1., 1964; Lyons, 1971; Miles, 1976; Miles and Petty, 1975; Rizzo et a1., 1970; Szilagyi and Sims, 1975). Other attitudinal and affective outcomes of role con- flict and ambiguity include tension and anxiety. Tension has been correlated with role conflict (Beehr et a1., 1976; Brief and Aldag, 1976; French and Caplan, 1972; Gross et a1., 1958; Kahn et a1., 1964; Miles, 1976; Rizzo et a1., 1970), and with role ambiguity (Beehr et a1., 1976; Brief and Aldag, 1976; Caplan and Jones, 1975; Ivancevich and Donelly, 1974; Kahn et a1., 1964; Lyons, 1971; Miles, 1976; Miles and Petty, 1975; Rizzo et a1., 1970). Anxiety has been posi- tively correlated with role conflict (Brief and Aldag, 1976; Caplan et a1., 1975; Caplan and Jones, 1975; Gross et a1., 1958; House and Rizzo, 1972; Rizzo et a1., 1970; Tosi, 1971), and with role ambiguity (Brief and Aldag, 1976; Caplan et a1., 1975; Caplan and Jones 1975; Rizzo et a1., 1970; Wispe and Thayer, 1957). Effective outcomes include an increased sense of futil- ity, related to role conflict (French and Caplan, 1972) and role ambiguity (Kahn et a1., 1964). Depression has been correlated with role ambiguity (Beehr, 1976; Caplan et a1., 1975) and with role conflict (Caplan et a1., 1975; Getzels 50 and Guba, 1954). Irritation has been positively correlated with role conflict (Caplan et a1., 1975) and happiness nega- tively correlated with it (Hall and Gordon, 1973). Several health-related variables have been correlated with role conflict and ambiguity. These include fatigue (Beehr, Walsh and Taber, 1976), heart rate (Caplan and Jones, 1975; French and Caplan, 1972), physical stress (Ivancevich and Donelly, 1974) and somatic complaints (Caplan et a1., 1975). Intent to quit has been positively correlated with role conflict (Sorenson and Sorenson, 1974); propensity to leave has been correlated positively with both role conflict and ambiguity in one study (Brief and Aldag, 1975), and with role ambiguity alone in several others (House and Rizzo, 1972; Ivancevich and Donnelly, 1974; Lyons, 1971; and Rizzo, House and Lirtzman, 1970). Actual turnover has been related to role conflict (Johnson and Graen, 1973) and to role ambiguity (Johnson and Graen, 1972; Lyons, 1971). On an interpersonal level,aifew variables have been examined in relation to conflict and ambiguity. These in- clude friction, which correlates with role conflict (Haas, 1964), and group involvement, which correlates with role ambiguity (Raven and Rietsma, 1957). On an organizational level, group productivity has been correlated negatively with role ambiguity (Smith, 1969), and perceptions of organiza- tional effectiveness have been negatively correlated with 51 both aspects of role strain (House and Rizzo, 1972). A series of studies have also examined variables that moderate the relationship of role conflict and ambiguity and its various outcomes. The most clear moderators seem to be qualities of the task or job. Autonomy reduces the relationships between role ambiguity and both job dissatis- faction and depression (Beehr, 1976). Participation in decision making moderates the relationships of both role strains with job satisfaction and performance (Schuler, 1977). Organizational level has been found to have the same moderating effect (Schuler, 1977; Silagyi and Sims, 1975), and these two moderators were also found to have an inter- active effect on the relationships of both role strains with satisfaction and performance (Schuler, 1977). These find- ings are significant for the burnout research; both of the moderating variables allow for more control over a greater selection of coping strategies. Burnout could well be an "intervening" variable here, acting as the connector between role strain and decreased performance. Summary The literature on role strain has been examined in depth in order to highlight its similarity to burnout. Nearly all of the antecedents and consequences of role strain are also related to burnout, as previously described. The similar patterns of these relationships indicate a close 52 link between the two variables. This supports the hypothe- ses that burnout results from role strain. From the above literature review, the following hypotheses are derived: Hypothesis 4: The more role strain an employee experiences, the less job satisfaction he/she will experience. Hypothesis 5: The more role strain an employee experiences, the more burnout he/she will experience. Job Satisfaction Job dissatisfaction is one of the elements that is often included in the burnout construct; if not part of burnout, it is correlated highly with it. Unfortunately the conceptual confusion that surrounds burnout is similar to that which surrounds the construct of job satisfaction. Although it is one of the most widely examined concepts in Industrial/Organizational Psychology, there are a multitude of operational and theoretical definitions of it, none of which seem to measure the same thing (Wanous and Lawler, 1972). The Handbook of Industrial and Organizational Psychology defines job satisfaction as "a pleasurable or positive emo- tional state resulting from the appraisal of one's job or job experiences" (Locke, 1976, p. 130). Wanous and Lawler (1972) examine the multiple operational definitions, and 53 conclude that the different measures each tap a different aspect of the construct, and the experimenter should choose carefully based on the variables she/he is going to be relating it to. Burnout is conceptualized as a combination of affec- tive and attitudinal changes that result from c0ping with job strain. As such, it taps a content area highly related to job satisfaction. Burnout results specifically from a pro- cess of coping ineffectively, which differentiates it from job satisfaction. It contains both affects and attitudes, but these affects are more varied than that of satisfaction with one's job and its component parts; as will be seen in section III, they include self-efficacy, fear, perceived support and feeling pressured, which are not only perceptions of job attributes, but self-perceptions. Burnout is characterized by extreme negative affect; burned out workers experience the attitudes and feelings to a very strong degree. This is not true of job satisfaction, which is thought of as a difference in degrees of liking. Job satisfaction makes fine differentiations of a more limited range of feelings, with a negative anchor of "very dissatisfied". The negative anchors of burnout include "frustrated", "bad", "worried", "tense" and "worthless", 54 which indicate greater intensity of feeling than job satis- faction measures. Relationship of Job Satisfaction with Other Salient Variables Job satisfaction is often used as a dependent variable, to describe how various qualities of a job, task, organiza- tion, situation or individual affect a job incumbent's atti- tudes. Job commitment and calculative organizational commit- ment (based on economic gain rather than normative values) were correlated with job satisfaction (Weiner and Vard, 1980), as was holding work as a Central Life Interest (Dubin and Champoux, 1977). Task characteristics that have been correlated with job satisfaction include job scope (Stone, 1976), Motivating Potential and its component parts of autonomy, task identity, skill variety, task significance and feedback (Hackman and Lawler, 1971; Wanous, 1974). Umstot, Bell and Mitchell (1976) describe an experimental study which sup- ports the causal impact of enriched jobs on increased job satisfaction. Other work attributes that have been related to job satisfaction include control over work methods and work pace, opportunity to use skills and abilities, and amount of work (Locke, 1976). As a causal factor, job satisfaction has been related to satisfaction with life (Kornhauser, 1965; Iris and Barrett, 1972; Weitz, 1952). It has been related to several 55 health-related variables, such as fatigue, headaches and ill health (Burke, 1969). In Palmore's (1969) longitudinal study of longevity, the best predictor of length of life was job satisfaction. Kornhauser (1965) also found consist— ent relationships between job satisfaction and mental health. Behavioral consequences of job satisfaction include negative relationships with absenteeism and turnover (Vroom, 1964; Atchison and Lefferts, 1972; Kraut, 1970; Taylor and Weiss, 1972; Waters and Roach, 1973; Hulin, 1968). There is an obvious similarity in the constructs of burnout and job dissatisfaction, based on their relation- ships with the other variables described. Besides being highly correlated (Maslach and Jackson, 1981, Golembiewski and Munzenrider, 1981; Pines, 1981), they show similar patterns of relationships with health and behaviorally related variables. However, the two variables do not totally overlap; one can be dissatisfied with one's job without being burned out, and without going through the process of ineffec- tive coping with strain that produces burnout. In summary, the research literature indicates that job satisfaction is correlated with autonomy, control over work pace, and negatively correlated with burnout. This allows us to derive the following hypothesis: Hypothesis 6: The more job satisfaction an employee experiences, the less burnout he/she will experience. 56 Summary and Hypotheses The construct of burnout has been defined as an intra- psychic process in which the worker copes ineffectively with role strains, and experiences negative shifts in several affects and attitudes. The specific affects and attitudes will be described in the next section. Burnout results from role strain, so is positively correlated with both role con- flict and role ambiguity. It depends on the use of effective coping mechanisms at work; the number of effective strategies available is increased by control over the timing of work. Hence, burnout is negatively correlated with timing control. Role strain leads to job dissatisfaction, and is posi- tively correlated with it. It leads to, and is positively correlated with, burnout. It is negatively correlated with control of timing of work because role conflict is reduced by the ability to ease inter-role conflicts in terms of work and non-work roles. Control of the timing of work is negatively related to role strain, negatively related to job dissatisfaction, and negatively related to burnout. Lastly, job dissatisfaction is positively correlated with burnout. It is caused by lack of control of timing of work, and hence is negatively correlated with that control. It is caused by role strain, and is positively correlated with both role conflict and role ambiguity. 57 These relationships are based on the literature reviews previously presented. They are described in the following hypotheses: Hypothesis 1: The more control over the timing of work an employee has, the less burnout he/she will experience. Hypothesis 2: The more control over the timing of work an employee has, the more job satisfac- tion he/she will experience. Hypothesis 3: The more control over the timing of work on employee has, the less role strain he/she will experience. Hypothesis 4: The more role strain an employee experi- ences, the less job satisfaction he/she will experience. Hypothesis 5: The more role strain an employee experi- ences, the more burnout he/she will experience. Hypothesis 6: The more job satisfaction an employee experiences, the less burnout he/she will experience. Causal Model of Burnout These hypotheses can be combined to describe the causal process of burnout. This forms a recursive model in which role strain and timing control are negatively and non- causally correlated. Role strain produces decreased job satisfaction; it also directly causes increased burnout. Timing control causes increased job satisfaction, and also directly decreases burnout. Job satisfaction directly de— creases burnout. These relationships are depicted in Figure 3. 58 ROLE STRAIN ¢ JOB SATISFACTION ’l BURNOUT] + A TIMING CONTROL Figure 3. Combined hypotheses of burnout relationships. Path analysis of this model will allow non-significant rela- tionships to be dropped from the causal model. The exist- ence of both direct and indirect influence of role strain and timing control on burnout will be checked. Likewise, the negative, noncausal relationship between role strain and timing control will be checked for its importance in explaining burnout. Summary This section has reviewed the literature on burnout and pinpointed several related variables: role strain, time con- trol and job satisfaction. The literature concerning the 59 interrelationships of these four variables was examined, and specific hypotheses describing these relationships were derived. These hypotheses were combined into a causal model of burnout. Section III presents the methods by which this model will be tested. I I I . METHODOLOGY This section presents the methodology used to test the hypotheses previously stated. It describes the subject group that is examined, the procedures used to develop and administer the questionnaires, and the design of the analyses. The specific analyses used to test each hypothe- sis are described, and the instruments are presented. Subjects The subjects who participated in the present study con- sisted of 243 employees of a community mental health center in Michigan. Employees included secretaries and clerical workers, therapists, case managers, supervisors, residence home managers, and middle and upper level administrators. Sixty-three percent of the subjects were female. Their ages ranged from below 20 years old to over 45 years old, with a mean between 25 and 35 years. Fifty-one percent had children. Educational levels ranged from some high school to a doctoral degree, with an average of a bachelor's degree. Tenure with the organization ranged from less than one year to more than ten years, with a mean between one and three years. 60 61 Procedure All data were gathered in questionnaire form. The data for this thesis were gathered as part of a larger project; the other measures on the questionnaire included an organiza- tional diagnostic. A five member research team administered the questionnaires to all employees; this was done through on-site visits over a five week period. Employees were gathered in groups of five to twenty in one room set aside for the purpose, where the researcher explained the directions and was available to answer.ques- tions. Questionnaires were completed during paid work hours, and in the presence of the researcher; virtually all employees filled out the questionnaire. Computer-scanned answer sheets were used; reponses were marked on a five-point Likert-type scale. The instruments used for this study comprised approxi- mately one-third of the entire questionnaire; the other instruments included measures of perceived relative import- ance of various constituents and organizational effective- ness. The entire questionnaire took approximately one hour to complete. Measures The Measurement of Burnout The instrument used to measure burnout was developed in lieu of using any of the previously developed measures of 62 burnout. The vest validated measure up to the preSent time did not fit the present theoretical definition of burnout, nor did it fit the empirical findings from interviews and preliminary questionnaire investigation. Four pencil-and-paper measurements of burnout have been reported in the literature; each one measures a different concept of burnout, and thus are probably not strongly related to one another. The first is the Maslach Burnout Inventory (MBI) (Maslach and Jackson, 1981), and it has the most extensive evidence of reliability and validity of any of the measures. It contains three subscales which measure emotional exhaustion, personal achievement (negatively scored) and depersonalization. Each item is scored in terms of both frequency and intensity of occurrence. The subscales have been correlated significantly with spouse evaluation, amount of direct client contact, and lack of feedback from the job, among other factors. All of the subscales focus on cognitive reactions to the work situation, the worker's performance, and relationships with clients. The emotional exhaustion and depersonalization subscales were not found to be separate factors in the development of the present measure. Conceptually, the exhaustion aspect which Maslach posits as a part of burnout itself, is viewed as a result of burnout. The items on the Maslach scale appear to measure a combination of perceived strain on the employee and emotional exhaustion; these may be pulled 63 together into one factor because both are related to the layperson's interpretation of the term "burnout", which is the main anchoring item in this subscale. The depersonaliza- tion, which Maslach conceptualizes as a part of burnout itself, is seen in the present conceptualization as a common- ly used coping strategy; it produces burnout, because of its deleterious effects on work performance, but it is not part of burnout itself. The second paper-and-pencil measure of burnout is Pines and Kafry's (1981) Tedium scale, which includes three dif— ferent subscales: physical exhaustion, mental exhaustion, and emotional exhaustion. Burnout, according to Pines, is tedium that results from work with people; the measure includes items referring to feeling worthless, disillusioned and resentful about people. Again, this conceptualization is not congruent with that operational in the present research; the items measure an amalgam of parts of the present definition of burnout, results of burnout and causes of burnout (physical exhaustion and ineffective coping strategies, respectively). The third measure is Jones' (1980) Staff Burnout Scale. It includes subscales measuring job dissatisfaction, psycho- logical and interpersonal tension, physical illness and distress, and unprofessional patient relationships. In addi- tion to the conceptual differences between the present and Jones' formulation, this scale is limited to professional 64 human service workers. The last scale, developed by Berkeley Planning Associates (1977), measures burnout as the combina- tion of five types of alienation, with a subscale for each type: alienation from clients, co-workers, projects, the job, and opportunities in the job. This conceptualization has little overlap with the most common elements of the con- cept, and very little with the one examined in this study. Development of the Affective Burnout Measure None of the existing measures taps all of the negative affective and attitudinal changes which constitute the present definition of burnout. A measure of burnout was created which defined the specific affective responses to stress which characterize burnout. These were identified through factor analysis of a wide variety of affective items. The results pinpointed five factors which tap the feelings of: job dissatisfaction/stagnation, incompetence, worry, lack of recognition, and time pressure. These emerged as the five affects which characterize burnout. The items for the burnout measure used in this study were generated to cover the dimensions of burnout suggested in the literature. These included emotional exhaustion, low personal accomplishment, withdrawal, fear, and job dissat- isfaction. The dimensions, and all items constituting them, were chosen through a combination of literature review and interviews. The employees interviewed ranged from thera- pists and case workers in a community mental health center 65 to clerical workers from a large state university in Michigan. Interviews were very loosely structured, and aimed primarily at finding out how workers felt about their jobs and what problems they perceived in their organization. Specific questions of what they thought made them feel "burned out" were also used. Questionnaire items were written in the semantic differ- ential format, which consists of a set of paired adjectival antonyms. This format was used because, when combined with factor analysis, it defines the dimensions of meaning of its focal topic; it describes the focal topic's dimensional qualities and their intensity, which Osgood calls describing its "semantic Space" (Osgood et a1., 1957). This allowed the measurement of more than the "evaluative" description of employees' feelings about their job, which is the subject of most job attitude and satisfaction scales. Evaluation is only one of the factors that are usually identified by use of semantic differential scales. Osgood et al. (1957), found three orthogonal dimensions that underlie people's semantic space in describing many topics: evaluation, potency and activity. However, he con- cludes that ... the functional semantic space is to some degree modifiable in terms of what kinds of concepts are being judged, i.e., the relative importance and relationship among factors may vary with the frame of judgments. Certainly, specific scales may change their meaning, in the factorial composition sense, as a function of 66 the concept being judged. And ... it is clear that what we have called the three dominant factors do not exhaust the dimensions along which meaningful judgments are differentiated. (Osgood et a1., 1957, p. 72.) Thus the semantic differential format of the questionnaire directly measured the employees' perceptions of the dimen- sions of how they feel at work, and the intensity of those dimensions. Responses were marked on a five-point Likert- type scale. Subjects were instructed in filling out a semantic differential, and requested to describe "the way you generally feel at work". Approximately 120 items were created, and these were culled down to create a preliminary form of the measure con- taining 80 items; items were deleted due to similarity of meaning and lack of clarity. The 80-item questionnaire was administered to 825 participants from two organizations; participation was on a voluntary basis. Participants included approximately 350 clerical and technical workers, and 450 undergraduate and graduate stu- dents; also included were 25 employees of a chain of retail furniture stores. Seventy-one percent of the respondants were female. Their ages ranged from below 20 years old to over 57 years old, with a mean between 20 and 29 years. Educational levels ranged from a high school diploma to a doctoral degree, with an average of a high school diploma. Tenure with the organization ranged from less than one year to more than ten years, with a mean between one and five 67 years; this was true for each of the organizations sampled. At the time of test administration the clerical and technical workers were in negotiation for a new contract and anticipating a strike; thus the stresses on them may have been unusually high. The data were factor analyzed using principal factoring with iteration and orthogonal (varimax) rotations. Five factors emerged which accounted for 86.9 percent of the total variance. The measure was then pared down from 80 items to 53 using several criteria. Items were deleted if they did not have factor loadings of .40 or higher. .Items that loaded at .40 or higher on more than one factor were rejected if they did not have at least a .10 difference in those factor loadings. Items with low item-total correla- tions were rejected, as were those whose deletion increased the factors' coefficient alpha. Items were recoded to orient toward the negative descriptor, so as to make all scales and items uniform in interpretation. Factor loadings and measures of reliability are presented in Table 5. All factors which had an eigen- value of greater than 1.0 were examined for contruct validity and internal consistency reliability. The first factor formed a Job Dissatisfaction/Stagnation subscale, which accounted for 54.6% of the variance in the measure. Of the twenty-four items in this subscale, fourteen were reverse scored. The internal consistency reliability (coefficient 68 TABLE 5 FACTOR LOADINGS AND RELIABILITIES OF BURNOUT MEASURE SCALES Dissatisfaction/Stagnation I II III IV V Frustrated .62 Pleased (R) .67 Trapped .56 Excited (R) .68 On a Treadmill .70 Energetic (R) .50 Powerless .50 Disillusioned .69 Committed (R) .46 Idealistic (R) .57 Fulfilled (R) .77 , Rewarded (R) .56 .41 Stagnant .76 Underutilized .42 Dissatisfied .75 Worthwhile (R) .48 Happy (R) .66 Optimistic (R) .66 Enthusiastic (R) .61 Negative .56 Successful (R) .48 Motivated (R) .61 Good (R) .48 Aimless .41 a = .95 N = 24 Incompetence Disorganized .51 Sophisticated (R) .56 Confused .51 Incompetent .62 Productive (R) .42 Destructive .44 Unsure .61 Authentic (R) .44 Responsible (R) .45 Realistic (R) .41 Prepared (R) .52 Passive .40 Dependent .49 Foolish .47 Disoriented .55 g f '87 R = reverse-scored 15 continued 69 TABLE 5 - continued Worry Secure (R) Worried Uneasy .80 3 (1 N Recognition Appreciated (R) .43 Recognized (R) .42 Supported (R) Heard (R) .84 4 a N Time Pressure Underworked (R) Pressured Relaxed (R) Rushed a N .78 4 III .47 .58 .58 R = reverse-scored IV .58 .60 .56 .54 .69 .71 .51 .72 70 alpha) of this subscale was .95. The second factor formed an Incompetence subscale, which accounted for 10.2% of the variance. There were fifteen items in the subscale, six of them recoded; time alpha coefficient for the subscale was .87. The third factor contained three items, and formed a sub- scale measuring Worry. It accounted for 9.1% of the vari- ance, with an alpha coefficient of .80; one of the items was reverse-scored. The fourth subscale had four items in it, and measured Recognition. It accounted for 3.7% of the variance and had an alpha coefficient of .84. All of the items in this factor were reverse-scored. The fifth and last factor contained four items, and measured Time Pres- sure. The coefficient alpha was .78, and two of the items were reverse-scored. It accounted for 3.2% of the variance. This factor analytic procedure clearly identified five factors in the eighty-item questionnaire. These factors were titled: Dissatisfaction/Stagnation, Incompetence, Worry, Recognition and Time Pressure. These factors suggest that employees perceptions of their feelings at work vary along these dimensions. Despite the previous research (Pines, 1981), the inclu- sion of items referring to exhaustion and fatigue produced no evidence of a factor for that construct. A two-item factor of withdrawal was identified, but had an unacceptable reliability (a = .52), and thus was not included in the 71 measure. Due to space limitations, all the items of the measure were not used in this study. The factor structure of the items used is examined in the next section. Other Measures The measure of perceived control over time was in the form of nine questions concerning the amount of employee con- trol over the timing of work, lunch hours, work breaks, and rearranging hours for special needs. Items were scored on a five-point Likert scale, with anchors of "no control", "little control", "some control", "a good deal of control", and "total control". The two time referents of "daily" and “weekly" were used, based on comments from the interviews. Some employees could impose control over timing on a future day, but had little control over the events of each ordinary day. Job satisfaction was measured by the one item General Motors Faces Scale (Kunin, 1955). Both male and female faces were used, although no significant differences were found by using either of the two forms (Dunham and Herman, 1975). Role strain was measured by the short form of the Rizzo, House and Lirtzman Role Conflict and Ambiguity measure (1977). Factor analysis and validity coefficients for multi- L>le samples have been reported for this form of the measure. Zkll of these measures can be found in the Appendix. 72 Analyses The analyses were conducted in four steps. The first two consisted of examining the factor structure of the burn- out measure and the reliability of all the measures; the third involved testing the hypotheses with a correlation matrix, and the fourth consisted of path analysis of the model. The first step was to run a confirmatory factor analysis on the five-factor solution of the burnout measure. Confirm- atory factor analysis involves the testing of the hypothe- sized variables, or factors, which underly a set of data items; it determines whether the hypothesized number of vari- ables is accurate, and whether the specific items group together, as hypothesized, to form those factors (Nunnally, 1978). This assumes that the 53 items in the burnout measure actually tapped into five underlying constructs. These were: dissatisfaction, incompetence, worry, recognition and time pressure. Thus all the items in each scale should have been strongly interrelated, and as a set tapped the underlying variables. Confirmatory factor analysis tested the hypothet- ical underlying relationships. It partitioned the indicated items into clusters, and one factor was defined by each cluster. All of the items in the burnout measure were entered int0 the confirmatory factor analysis program of the PACKAGE £3)(stem of FORTRAN routines (Hunter, Cohen, Gerbing and Nicol, 73 1980). The factors found in the initial development of the burnout measure were treated as hypothetical constructs, and tested by the program. A full correlation matrix of the items and predetermined factors was generated. The corre- lation matrix was generated with communalities in the diagonal, as determined by four iterations of the data. If the factors held up under this testing, the patterns of correlations should have been maintained; each item would load at .40 or higher with its factor, and less than that with all other factors, with at least a .10 difference between the highest and next highest factor loadings. Any items which did not fit these criteria were deleted from the measure. The second step was to analyze the internal consistency reliability of each of the subscales of the burnout measure, as well as the role conflict, role ambiguity, and time con- trol measures. This was done using Cronbach's alpha, as per- formed by the Reliability Subprogram of the Statistical Programs for the Social Sciences set of FORTRAN programs (Nie, Hull, Jenkins, Steinbrenner, and Bert, 1976). This resulted in a measure of how well each of the items in each scale was tapping the same construct as the others. If any items increased the internal consistency reliability when deleted, they were deleted from the measure. The third step was to test the hypotheses by examining the inter-scale correlations. Finally, a series of path analyses were performed, using the summed scores of each 74 scale, along with the single-item job satisfaction measure. Path analysis is "a procedure for systematically com- bining the use of partial multiple correlation to study the causal relations among a set of variables" (Hunter and Gerbing, 1980, p. 35). It is useful when a set of causal relationships between variables is known, and the model thus formed is recursive, and if all the relationships between the variables are linear. These conditions were met by the specific causal model of burnout, as presented at the end of Section II. The analysis was performed using the PATH subprogram of the PACKAGE programs (Hunter et a1., 1980). This program used ordinary least squares to identify the path coefficients, or the amount of change in the value of antecedent variables caused by changes in the value of causally prior variables. It also identified the value of the residuals for each vari- able; these represent the effect of all extraneous causes of the variable, not included in the models. Last, this program reproduced a correlation matrix of all variables based on the path coefficients produced. The closer the reproduced corre- lation matrix was to the original, the better the causal model matched the data. The sum of the squared deviations from the mean was the single-item statistic used to measure the good- ness of fit of the reproduced correlation matrix based on the model. 75 In order to use this program, an influence matrix was input for the causal model. The influence matrix was an alternative representation of the causal model, consisting of columns of variables. Every arrow in the path diagram was replaced by a l in the influence matrix. The influence matrix for the causal model is presented in Figure 4. 76 Role Strain = X1 Timing Control = X2 Job Satisfaction = X3 Burnout = X4 1 0 0 1 l f_‘ . X2 0 0 l 1 X1 1 L 4 Note - X and X are indicated as having a known noncausa relationship. Figure 4. Influence matrix for the burnout model. IV. RESULTS This section presents the results of the data analyses described in Section III. First, the results of the con- firmatory factor analysis on the burnout measure are described. Second, reliability coefficients, means and variances of all scales are reported. Third, the hypotheses presented in Section II of this thesis are addressed. Finally, the path analysis is reported, as well as several post hoc revisions. One model is selected from the second revision process which most closely fits the data. Confirmatory Factor Analysis A confirmatory factor analysis of the burnout measure subscales was performed to determine the factor loadings of each item on the estimated true scale scores, and the inter— correlations of these scores; this procedure used communati- ties in the diagonal of the correlation matrix. Table 6 presents the factor loadings of the scale items from the confirmatory factor analysis. Table 7 presents the intercorrelations of their estimated scale scores. The factor loadings of the items indicates that all items do load more highly on the factor to which they belong than 77 78 TABLE 6 FACTOR LOADINGS FROM CONFIRMATORY FACTOR ANALYSIS ITEM Dissatisfaction Stagnant Negative *Happy *Optimistic Disillusioned *Enthusiastic *Fulfilled *Pleased Frustrated Trapped Disoriented Unsure *Responsible *Prepared Incompetent Destructive Disorganized Insecure Worried Uneasy *Appreciated *Recognized *Heard Pressured Rushed 0L * = Recoded I .62 .80 .79 .78 .72 .74 .77 .84 .75 .r;;. .36 .35 .22 .41 .13 .36 .30 .34 .41 .58 .60 .63 .68 .31 .93 (afiln II .44 .22 .55 .26 .35 .45 .30 .26 .29 .42 .41 .68 .75 .38 .55 .63 .76 .70 .54 .43 .60 .23 .22 .30 .06 .05 .83 III .52 .25 .59 .42 .43 .55 .46 .22 .42 .60 .50 .54 .68 .19 .39 .44 .46 .48 .33 .32 .42 .38 .12 .78 IV .61 .50 .52 .65 .70 .61 .57 .65 .73 .53 .49 .23 .19 .30 .32 -.02 .20 .16 .29 .28 .41 .81 .84 .75 .22 .10 .84 .36 .10 .30 .27 .25 .22 .27 .24 .29 .39 .28 -.01 .13 -.17 .20 -.04 .07 .19 .08 .40 .30 .22 .15 .17 .7' .71 .66 79 oo.H whammoum mafia mm. oo.H :oHu Iflcmooomk m>Hucmoc ou cocoomm n « mm. mo. om. whammmum oEHB me. am. om. cowuflsmoooma oo.H an. ow. ammo: oo.H mv. mocmuomfioocH oo.H aoflumammbm \aoflbommmflummmflo GOADMGmmum wuuoz mosmummfioosH \cofluommmflummmflo mmOBudm BDOZMDm ho mZOHfifiqmmmOUMMBZH h mamdfi 80 they do on any other factor, confirming the factor structur- ing. The items in factor one, however, also load highly on factor four, and vice versa. Consonant with the selec- tion criteria for the factor elements, the differences between the loadings are at least .10. However, this indi- cates a strong correlation between factors one and four; the factor intercorrelation matrix (Table 7) identifies this correlation as .80. Several of the items in factor two load highly on factor three, and vice versa. Most of these items also differ at a .10 level in their factor loadings with the exception of "Unsure/Sure", which has a .07 differential in its loading. The factor intercorrelation matrix identifies the relation- ship between factors two and three as a .71 correlation. Because of the high intercorrelations of the burnout subscales, they were summed and treated as one variable for the following path analyses. Reliability of the Measures Estimates of internal consistency reliability were obtained for each of the measures, with the exception of the single-item global job satisfaction measure. The SPSS reli- ability subprogram was used to determine Cronbach's alpha coefficient for these measures. The scale means, standard deviations, number of items, and reliability estimates are reported in Table 8. 81 TABLE 8 RELIABILITY, MEANS AND STANDARD DEVIATIONS OF ALL MEASURES 0‘ 13112. s_1_>_ N items Time Control .87 2.89 .76 10 Role Strain .85 2.79 .60 1h Role Ambiguity .83 2.47 .68 6 Role Conflict .80 3.02 .71 8 Job Involvement .74 2.50 .72 h Global Job Satisfaction 2.h8 .98 1 Burnout Scales: Dissatisfaction/ Stagnation .93 3.24 .81 11 Incompetence .83 3.98 .62 8 Worry .78 3.h7 .86 3 *Recognition .8h 3.21 1.00 3 Time Pressure .66 2.39 .90 2 Summed Burnout Scale .93 3.h2 .60 27 * = all items recoded to reverse of factor title 82 All scales had a high value of 5.0 and a low of 1.0. Items were recorded in the direction of the scale name; i.e., 5.0 indicated total control over time, high role ambiguity, high role conflict, high job involvement and high job satis- faction. The reliability measures were of a satisfactory nature for all scales, and ranged from .66 to .93. Tests of the Hypotheses The hypotheses described on page 57 were tested by correlational analysis. Zero-order correlations were com- puted among each of the scales and subscales. These are pre- sented in Table 9, page 84. Almost all of the variables are significantly intercorrelated at the .05 level. Hypothesis 1: The more control over the timing of work an employee has, the less burnout he/she will experience. This hypothesis was supported by the data. The Time Control scale correlated with the entire burnout measure at the .001 level (r==-.36). Time Control correlated with each of the subscales of the burnout measure at the .001 level, with the exception of the Time Pressure subscale (r = .05, NS). Hypothesis 2: The more control over the timing of work an employee has, the more job satisfaction he/she will experience. This hypothesis was supported by the data. The Time Control scale correlated with Job Satisfaction at the .001 83 level (I = .27). Hypothesis 3: The more control over the timing of work an employee has, the less role strain he/she will experience. This hypothesis was supported by the data. Time Control correlated with Role Strain at the .001 level (r==-.32). Time Control correlated with each of the subscales of Role Strain, Role Ambiguity and Role Conflict, at the .001 level (r = -. 31 and - .25, respectively). Hypothesis 4: The more role strain an employee experi- ences, the less job satisfaction he/she will experience. This hypothesis was supported by the data. The Role Strain measure correlated with Job Satisfaction at the .001 level (r = -.54). The two subscales of Role Strain, Role Ambiguity and Role Conflict, correlated with Job Satisfaction at the .001 level (r = -.53 and -.43, reSpectively). Hypothesis 5: The more role strain an employee experi— ences, the more burnout he/she will experience. This hypothesis was supported by the data. The Role Strain measure correlated with burnout at the .001 level (r = .66). Role Strain correlated with all the burnout sub- scales at the .001 level: the Role Strain subscales, Role Conflict and Role Ambiguity, correlated with burnout and all of its subscales at the .001 level, with the exception of Role Ambiguity and Time Pressure; they correlated at the .01 level (r = .14). 84 Ho.vm um usmowmecmflm mo.vm aw unmofimwcmflm H «« oouoom omum>om u m Hoo.vm um unmoHMHsmwm u ear oo.H vm. Nb. an. mm. mm. mm. mm. mm. mm.: mn.l usosusm .HH «as are «xx «xx «as «a« era are are «as oc.a 5H. hm. mo. mm. Hm. hm. vH. vo.s HN.I musmmmum mEHB .oa « ««« «as «xx «xx k« «as oo.a mm. mm. mm. mm. mv. mm. om.| No.l Amv cofiuacmoomm .m «at car are «as «as «re «as are oo.H om. am. He. Hm. mm. om.l hm.l anuoz .m «k* «as «as «r« «as era «at oo.H mv. ca. mm. mv. vm.n vm.l mocmuomfioosH .5 ask «as «r« a«« «xx «xx oo.H No. av. mm. mm.l mh.l sowumcmdum .w I «as axe sax are are \GOHuommmHUMmmHQ oo.H om. ow. mm.l vm.l :Hmuum maom .m are arr «ex «ea oo.a mv. mm.n mv.| UOHHMGOU oaom .v #11! an“! alii oo.H Hm.u mm.- suuaauns< maom .m arr arr oo.H hm. Houucou meB .N «as oo.H coauommmfiumm now .H Ha oa m m n m m v m N H mMAdUm AA¢ m0 mZOHfiéqmmmOUmMBZH m mqmfla 85 Hypothesis 6: The more job satisfaction an employee experiences, the less burnout he/she will experience. This hypothesis was supported by the data. Job satis- faction correlates with burnout at the .001 level (r = -.73). Job Satisfaction also correlates with all the Burnout sub- scales at the .001 level. All of the hypotheses established in Section II of this thesis were supported by the data. Initial Path Analyses Path analyses were performed to test the model described in Section III. The matrix of beta weights, or path coeffi- cients, is presented in Table 10. The path diagram formed from this matrix is presented in Figure 5. The reproduced correlation matrix based on the beta weights is listed in Table 11. There is no difference between the reproduced correlation matrix and the observed one because the model being tested is exactly identified; in other words, the model includes exactly enough information to make unique estimates of each parameter. The path coefficients from Role Strain were -.50 to Job Satisfaction and .35 to Burnout. The co- efficients of time control were .11 to Job Satisfaction and -.11 to Burnout. The coefficient of the Job Satisfaction- Burnout path was -.51. 86 TABLE 10 MATRIX OF PATH COEFFICIENTS FOR INITIAL MODEL X1 x2 x3 X4 X1 0 0 0 0 X2 0 O 0 0 X3 - 50 .11 0 0 X4 .35 -.11 -.51 0 TABLE 11 INPUT AND REPRODUCED CORRELATION MATRIX OF INITIAL MODEL x1 X2 x3 X4 X1 1.00 .32 -.54 .66 X2 .32 1.00 .27 .36 X3 -.54 .27 1.00 .73 X4 .66 -.36 -.73 1.00 X1 = Role Strain X2 = Time Control X3 = Job Satisfaction X4 = Burnout 87 .35 I Role Strain Job Satisfactionyj -.51 Time Control I *4.» Burnout ‘. Figure 5. Path diagram of initial model. 88 Because of the small size of several of the beta weights, tests of significance were performed for them. The standard errors of the beta weights were derived by the formula: where N = number of subjects 2 2 1-R2 1 R ‘ Blrl + B2r2 o = x . _ 2 l = time control - B l r N-2 . . . xy job satisfaction 2 = role strain - job satisfaction The 95% confidence interval was established around the beta- weight for the time control-job satisfaction relationship, and was found to range from .01‘ Role Job .___.. Control -.32 Strain -.54 Satisfaction -.52 lBurnout Figure 6. Initial revised path model. 90 Because of the change in causal ordering, the designa- tions "X1" and "X2" were reversed, such that Xl became time control and x2 became role strain. The matrix of path coef- ficients for the revised model is found in Table 12. The correlation matrix reproduced from the beta weights is pre- sented in Table 13. The difference between this reproduced matrix and the original input matrix is listed in Table 14. The matrix of the difference between the original and reproduced correlations indicated that there was a stronger relationship between time control and the dependent variables than this revised model can account for. The correlation with job satisfaction is underestimated by .10, and the corre- lation with burnout is underestimated by .15. This finding led to a second stage revision of the model. A direct causal link was added from time control to the dependent variables, although time control remained a causal factor in role strain. Three parallel models were tested. In the first, time control was considered a direct cause of job satisfaction; this is shown in Figure 7. In the second, time control had a direct causal influence on burnout, which appears in Figure 8. This third model contained direct causal links between time control and both dependent variables, and is presented in Figure 9. The matrices of path coefficients, reproduced correla- tions, and differences between input and reproduced 91 TABLE 12 MATRIX OF PATH COEFFICIENTS OF INITIAL REVISED MODEL X1 x2 x3 x4 x1 0 o o 0 x2 - 32 o o 0 x3 0 -.54 o 0 x4 0 .38 -.52 0 TABLE 13 REPRODUCED CORRELATION MATRIX OF INITIAL REVISED MODEL X1 X2 X3 X4 X1 1.00 -.32 18 - 21 X2 -.32 1.00 - 54 .66 x3 .18 -.54 1.00 -.73 X4 -.21 .66 -.73 1.00 Sum of Squared Deviations = .03 = Time Control Role Strain Job Satisfaction ><><><>< b UJN F4 Burnout 92 TABLE 14 DIFFERENCE BETWEEN OBSERVED AND PREDICTED CORRELATIONS OF INITIAL REVISED MODEL X1 X1 0.0 x2 0.0 x3 .10 x4 -.15 Sum of Squared Deviations 0.0 x3 X4 .10 o o 0.0 -.15 o o 0.0 o o 0.0 = .03 x1 x2 x3 x4 Time Control Role Strain Job Satisfaction Burnout Time Control -.32 Figure 7. 93 Role 4 Strain '50 .11 .38 Job Satisfaction -.52 J, Secondary revised path model (#1). 1 Burnout I .35 Time _ 32 Role _ 54 Job _ sib- Burnout] Control ' Strain ' Satisfaction ' , -.11 I Figure 8. Secondary revised path model (#2). . 35 1 Time __ 3 2 Role W Job 75—1? Burnout Control ' Strain 8’ Satisfaction ' .11 I -.11 Figure 9. Secondary revised path model (#3). 94 correlations follow. Tables 15 through 17 describe the first of these models, Tables 18 through 20 describe the second, and Tables 21 through 22 describe the third. Examination of the difference tables (Table 17 and Table 20) indicates that the first model, where time control directly causes job satisfaction, appears to fit the data better than the second model, where time control directly causes burnout. In the first model, only the time control- burnout relationship is not correctly predicted; it is under- predicted by r = .09. In the second model, the time control and job satisfaction was underestimated by .10, the Correla- tion between time control and burnout by .05, and the corre- lation between job satisfaction and burnout by .01. The sum of the squared deviations from the mean are identical (.01) for both models,however. The sum of the squared deviations from the mean is a statistical criteria of the goodness of fit of the linear curve predicted by the model. It is calculated by squaring the horizontal devia- tions of each data point to the predicted line (Freund and Williams, p. 304-305). Thus, this statistic indicates that the first and second revised models fit the data equally well. The last model, like the first, unrevised model, is exactly identified; thus the observed and reproduced correla- tion matrixes are the same. This model fits the data such that all the zero-order correlations are predicted, and the strength of the time control-role strain correlation is 95 TABLE 15 MATRIX OF PATH COEFFICIENTS OF SECONDARY REVISED MODEL (#1) x1 x2 X3 X4 x1 0 o o 0 x2 -.32 o o 0 x3 .11 -.so 0 0 x4 0 38 - 52 0 TABLE 16 REPRODUCED CORRELATION MATRIX OF SECONDARY REVISED MODEL (#1) 1 2 4 X1 1.00 -.32 .27 -.27 X2 - 32 1.00 - 54 .66 X3 .27 -.54 1.00 -.73 X -.27 .66 -.73 1.00 96 TABLE 17 DIFFERENCE BETWEEN OBSERVED AND PREDICTED CORRELATIONS OF SECONDARY REVISED MODEL (#1) x1 x2 X3 X4 X1 0 0 0 - 09 X2 0 0 0 0 X3 0 0 0 0 X4 - 09 0 0 0 Sum of squared deviations = .01 X1 = Time Control X2 = Role Strain X3 = Job Satisfaction X = Burnout 97 TABLE 18 MATRIX OF PATH COEFFICIENTS OF SECONDARY REVISED MODEL (#2) x1 x2 x3 X4 0 o 0 o _ 32 0 0 0 o -.54 o o -.11 .35 -.51 0 TABLE 19 REPRODUCED CORRELATION MATRIX OF SECONDARY REVISED MODEL (#2) 1 2 3 4 l 1.00 - 32 18 -.31 2 -.32 1.00 -.54 .66 3 .18 1.54 1.00 -.72 4 -.31 .66 -.72 1.00 X1 = Time Control X2 = Role Strain X3 = Job Satisfaction X4 = Burnout 98 TABLE 20 DIFFERENCE BETWEEN OBSERVED AND PREDICTED CORRELATIONS OF SECONDARY REVISED MODEL (#2) X1 X2 X3 X4 X1 0 0 .10 -.05 X2 0 0 0 0 X3 .10 0 0 -.1 X4 -.05 0 -.l 0 Sum of squared deviations = .01 = Time Control Role Strain Job Satisfaction ><><><>< h»tu Riva Burnout 99 TABLE 21 MATRIX OF PATH COEFFICIENTS OF SECONDARY REVISED MODEL (#3) X X X X 1 2 3 4 X1 0 0 0 0 X2 -.32 0 0 0 X3 .11 -.50 0 0 X4 -.11 .35 -.51 0 TABLE 22 REPRODUCED CORRELATION MATRIX OF SECONDARY REVISED MODEL (#3) l 2 3 4 X1 1.00 -.32 .27 -.36 X2 -.32 1.00 -.54 .66 X3 .27 -.54 1.00 -.73 X4 -.36 .66 -.73 1.00 Sum of squared deviations = .00 = Time Control Role Strain Job Satisfaction >< X >< X a to u>:4 Burnout 100 incorporated into the model. At the same time, the smaller but still significant relationships of time control and the two dependent variables are accounted for by the model. Summary This section presented the results of the analyses performed in this thesis. The confirmatory factor analysis of the affective burnout measure supported the validity of the measure's factor structure. Internal consistency relia- bility of all measures were reported at acceptable levels. The six hypotheses were supported by data from a zero-order correlation matrix of all scales and subscales. The results of the initial path analysis were presented. A series of modifications of the original path model were described, each including another relationship which improved the fit of the model to the data. A final version was established which contained all the elements of the original model, with the addition of a causal relationship between time control and role strain. V. DISCUSSION This section presents the conclusions drawn from the results of the described study of burnout. First the method of measuring burnout is examined, and the robustness of the factor structure of the affective measure is supported. Next the findings on each variable are presented, along with the theoretical implications of those findings. Future directions for research are suggested, and limitations of the present study are described. The Measurement of Burnout Burnout is defined as the negative changes in affects and attitudes that result from ineffective coping with job stress. Previous research has described measures which focus on cognitive aspects of burnout (Maslach, 1981) and exhaustion (Pines, 1981). An affectively oriented measure of burnout was developed, which consisted of 53 semantic differential items describing emotions and attitudes. These items loaded highly on five factors in an exploratory factor analysis; each factor described a feeling or attitude in response to work. This measure was administered to the 243 subjects of this study, and confirmatory factor analysis was used to examine the robustness of the factor structure. 101 102 Due to space limitations, all of the 53 items in the measure could not be included in this study. Twenty-six semantic differential items were used to tap the five factors. Eleven items came from the dissatisfaction/- stagnation factor, seven items came from the incompetence factor, three items from the worried factor, three from the recognition factor, and two from the time pressure factor. All items were found to load more highly on the factors from which they were drawn than on any other factor, reproducing the original factors in the new sample. Each of the factors formed a subscale which deScribed a different affective reaction to work. The internal con- sistency reliabilities of the factors were all acceptable; the two—item "Time Pressure" scale had the lowest of these at a = .66, while the other scales ranged from .78 to .93. All of the burnout scales were highly intercorrelated. The intercorrelations of the estimated scale scores, corrected for attenuation, ranged from .08 to .80, with a mean of .40. The most closely related scales were Dissatisfaction/Stagna- tion and Recognition (r=-.69), Incompetence and Worry (r=.60), and Dissatisfaction/Stagnation and Worry (r=.51). The scale that is the most independent of the others is Time Pressure. It correlated significantly with Dissatisfaction/Stagnation (r=.29, p:<.001), with Worry (r=.27, p:<.001) and with Recog- nition (r=-.l7, p:<.01); it did not signicantly correlate with Incompetence. 103 There are several potential reasons for these high scale intercorrelations. First, there is probably a sub- stantive amount of method bias occurring. All items were presented in the same Likert format. The items were inter- mingled in their order of presentation, such that items from different scales were adjacent to one another. Second, the items all measure feelings in response to work. Although they describe different feelings, they measure the same type of variable. The average scale inter- correlation found in this study is not especially high for this type of rating set. For example, when satisfaction with five aspects of work is measured with the common format of the faces scale, the interrelations range from .16 to .59, and the mean intercorrelation is .42 (Locke, Smith, Kendall, Hulin and Miller, 1964). Third, the scales with the highest correlations also are logically related. If an employee feels recognized and appreciated, he/she is more likely to feel satisfied, posi- tive, happy and forward-moving, while an employee who feels ignored and unappreciated is more likely to be dissatisfied, unhappy, negative, and trapped. Likewise, an employee who feels incompetent has more reason to be worried and insecure about his/her job. Last, the conceptualization of burnout assumes that these feelings will all change together. Burnout is a 104 negative change in several affects and attitudes in response to job stress; if the stress is not effectively coped with, many feelings and attitudes take on a negative cast. There are no theoretical grounds to assume that some feelings change while others do not; therefore, these feelings should covary. As stress increases and the employee fails to cope with it, the entire syndrome of negative changes occurs (Cherniss, 1980; Maslach, 1978, 1981). For these reasons, the measure of burnout is taken as the average of responses on all the items from all the subscales. The internal consistency reliability of this summed measure was a =.93. The five factors which emerged from the initial analysis were confirmed in this sample; there appear to be five distinct areas of affective reaction to work tapped by this measure. These areas are strongly related, however. Much as one can describe "satisfaction with work" as the summation of satisfaction with several distinct aspects of work, as the Job Descriptive Index does (Locke et a1., 1964), one can describe "burnout" as the sum- mation of several distinct affective reactions to work. Theoretical Implications Job Satisfaction This study_found job satisfaction to be significantly correlated,én:the .001 level, with all other variables 105 examined in the model. It correlated positively with time control (.27), and negatively with role ambiguity (r = -.53), role conflict (r = -.43), role strain (r = -.54), and burnout (r = -.73). It was hypothesized that high role strain and low timing control resulted in low degrees of job satisfaction. These hypotheses are supported by the data. The Path model which most closely fit the data included the direct causal impact of role strain in decreasing job satisfaction. The beta weight for this causal link was -.50, indicating a very strong direct relationship. This Path model also included the direct causal impact of time control on job satisfaction, along with its indirect impact by its contribution to role strain. These results add further support to the previous litera- ture which found a strong link between role strain and job satisfaction. The regression analysis also implies that role strain has a causal effect on job satisfaction. These analyses are not longitudinal, and hence cannot directly support the causal hypothesis and the direction of causality; this is not proof that role strain creates job dissatisfac- tion, rather than vice versa. However, the close fit of the final Path model with the data supports the hypothesis that job satisfaction is decreased by role strain and directly increased by control over the timing of work. 106 Burnout It was hypothesized that burnout is related to low job satisfaction, high role strain and low timing control; all of these relationships are supported by the correlational data. The strong correlations found between the affective measure of burnout and the theoretically related variables of role strain, job dissatisfaction and timing control support the validity of the present measure. The similarity in the correlational patterns of the present measure to those of previous burnout measures, such as the Maslach Burnout Inven- tory, suggests that the present measure does indeed tap affec- tive aspects of burnout. Further validation and cross- validation is needed to specify the exact nature of the relationships between affective and cognitive measures of burnout. The regression analysis reveals that most of the variance in burnout is accounted for by job dissatisfaction. The second greatest direct impact on burnout is role strain; the last element which impacts burnout is timing control. Again, the temporal aspects of the model and the direc- tion of causality cannot be proven by this study; an experi- mental or longitudinal study is required to give specific evidence for this. However, the model of the causal proc- esses which result in burnout was repeatedly modified to create the closest fit with the data; this implies that the 107 assumptions of direct causal impact of role strain, and job satisfaction on burnout are accurate. The Relationship of Job Satisfaction and Burnout Job satisfaction and burnout were hypothesized to be strongly negatively correlated, distinct reactions to work. They were theoretically differentiated by the specific process of ineffective coping with job stress that results in burnout, as opposed to the variety of causes of job dissatisfaction; by the extreme intensity of burnout, as opposed to the more moderate attitudinal and affective reactions that have been measured as job satisfaction; and by the variety of attitudes and affects that constitute burnout, of which job dissatisfaction is one part. The overall correlation of the affective measures of burnout and job satisfaction was -.73. The correlations of job satisfaction with the subscales of burnout varied from .21 to .79. The pattern of correlations indicated that the dissatisfaction/stagnation subscale of the burnout measure tapped the same construct as global job satisfaction; the two were correlated at .79. This subscale contained almost half of the items in the entire burnout measure, which accounted for the high correlation of the summed scale with job satisfaction. Job satisfaction correlated .62 with recognition, .37 with worry, .34 with efficacy, and .21 with time pressure. 108 The stronger relationships with the burnout satisfaction subscale than with the global job satisfaction measure are probably due at least in part to measurement bias. The identical format and interspersed position of the burnout items may account for some similarity of responses, as opposed to the job satisfaction measure, which had a com- pletely different format and appeared in another section of the questionnaire. Despite the somewhat stronger correlations of the burn- out dissatisfaction scale with the other burnout subscales, the pattern of correlations is similar to that between the global satisfaction measure and the four last burnout sub- scales. Both are strongly related to feelings of recogni- tion, moderately related to feelings of efficacy and worry, and slightly related to feeling pressured and rushed. This suggests that job satisfaction is one of the atti- tudes and affects that changes in response to job stress. Despite its history as the central direct measure of affect in the organizational literature, it is not the only reaction that people have to their work. It is strongly correlated with overall burnout, and somewhat less strongly with per- ceptions of being recognized. It correlates moderately with feelings of efficacy and worry, and slightly with time pre— sure. Thus, burnout and job satisfaction appear to be dis- tinct but related constructs. 109 Role Strain Role strain was defined as the sum of role conflict and role ambiguity; it was hypothesized as a type of job stress which leads to job dissatisfaction and burnout. Lack of timing control was hypothesized as a type of job stress and a limitation on coping methods, which lead to dissatisfac- tion and burnout. The results of this study supported these hypotheses, with some modifications. The zero-order correlations of role strain with the dependent variables indicate that it correlated with burnout at .66 and with job satisfaction at -.54. Each of its sub- scales also correlated highly with the dependent measures: role ambiguity correlated at r = -.53 with job satisfaction, and at .62 with burnout, while role conflict correlated at r = .43 and .52, respectively. The role strain subscales had similar but not identical patterns of relationships with the burnout subscales. Role ambiguity was more strongly related to dissatisfaction/- stagnation than role conflict (r = .58 and .59), to efficacy (r = .43 and .28), to worry (r = .38 and .31) and to recogni- tion (r = .52 and .48). The last burnout subscale, time pressure, showed the opposite pattern; it correlated at r = .14 with role ambiguity, and r = .37 with role conflict. These patterns may be understood by a close analysis of the burnout subscales. Time pressure had the lowest correla- tion with other measures of any of the burnout subscales. 110 It also had the lowest correlation of any of the subscales with the overall burnout measure. It may be that time pres- sure was perceived as the direct result of or as an indicator of role conflict. Role conflict is the incompatability of expectations from various sources, one type of which is role overload: "expecting the role incumbent to engage in several role behaviors,... within too short a time period" (Kahn et a1., 1964). The logical result of role overload is the perception of being rushed and pressured. Role overload is not specifically included in the form of the role strain measure used in this thesis, but several of the items have been used elsewhere to measure it, such as, "I receive an assignment without the manpower to complete it" and "I re- ceive an assignment without adequate resources and materials to execute it". This rationale implies that the time pressure subscale of the burnout measure may be used as an indicator of per- ceived role overload. This hypothesis is supported by the fact that time pressure correlated higher with role conflict (r = .37) than with any other measure, including the overall burnout scale (r = .34). The regression analyses used to test the Path model indicate that role strain directly accounts for 50% of the variance in job satisfaction, and 35% of the variance in burnout. This supports the direct causal impact of role strain on job dissatisfaction and burnout, as well as the 111 indirect effect of role strain on burnout through decreased job satisfaction. Support for the model, however, cannot be taken to disprove competing models with alternative causal Paths. Experimental or longitudinal analysis is required to rigorously test such hypotheses. TimingiControl Timing control was defined as the amount of control employees have over which aspects of work they do, and the placement of the hours they work. It was measured by a ten item scale which was originated for this study. Results showed that timing control was independent of perceived time pressure (Table 9). This indicates that the measure did indeed tap the job context factor of when work is done, rather than the job content factor of the amount of work done or overload. The literature suggests that there is a relationship between control and negative affect, although the effect of control over the timing of work has not been singled out for study before. General situational control has been directly related to negative affect and attitudes (Suls and Mullen, 1981; Rehm, 1977; Abramson, Seligman and Teasdale, 1978; Glass, Singer, and Friedman, 1969;Iefcourt,l973; Langer and Rodin, 1976). Likewise, autonomy has been correlated with increased job satisfaction (Hackman and Lawler, 1971); Hackman and Oldham, 1976). This implies a direct relation- 112 ship between timing control and the affective dependent variables. The zero-order correlations of time control with the dependent variables indicate that it correlated with burnout at -.36 and with job satisfaction at .27. The path analyses indicated that time control had a small but significant direct effect on job satisfaction. The initial revised model and the first two models of the second revision were rejected because they did not include direct causal impact of time control on both dependent variables. The last model of the second revision did include small direct links to both dependent variables, and appeared to maximally fit the data. The Relationship of Role Strain and Timing Control Timing control had its greatest impact as a causal factor in role strain. The zero order correlation between the two was .32, and the beta weight of the link in the path model was .32. Previous research indicated a relation- ship between timing control and role strain (Kanter, 1977; Korman and Korman, 1980; Ronen, 1981), as well as a direct relationship between timing control and job stress (Marshall and Cooper, 1978; Maslach and Pines, 1977; Karasek, 1979). This had not been represented in the initial path model, although it was present in all revisions. 113 Summary of Theoretical Implications The results of this study imply that burnout, measured as affective and attitudinal perceptions of one's job, is a construct related to, but differentiated from, global job satisfaction. The affects and attitudes consist of five underlying factors. Of these, dissatisfaction]stagnation and recognition factors are strongly related to global job satisfaction. The incompetence and worry factors are sig- nificantly, but less strongly, correlated. Time pressure is a separate factor which is not significantly correlated with incompetence, and has a low correlation with recognition. Burnout, job satisfaction, role strain and time control are all significantly intercorrelated. Role strain is formed of two related but differentiated constructs, role conflict and role ambiguity. They have similar degrees of relationship with burnout and job satisfaction, and smaller but significant relationships with timing control. The time pressure subscale of the burnout measure may be the result of role overload. Timing control is significantly correlated with all variables except for time pressure. This indicates that the control over the timing of work is separate from feeling pressured and rushed at work. Path analyses imply that burnout is caused by role strain, job dissatisfaction and lack of timing control. Time control decreases the role strain and increases the 114 job satisfaction experienced, while role strain decreases the job satisfaction experienced. Implications for Practice The results of this study are especially significant in terms of their practicality. Many structural elements of organizations and individual difference variables have been found to be correlated with burnout; these are particularly difficult to change in a functioning organization, either due to expense or to the limitations of selection procedures. Two key elements of this model, time control and role strain, are situation and job-related, however; this makes them relatively easy to change in functioning organizations. This study implies that in human service organizations such as mental health institutions, both the mild affect of job dissatisfaction and the extreme affect of burnout may be decreased by decreasing role strain. Arranging work situa- tions so that employees know what is expected of them and how to fulfill those expectations, and so that they operate under compatible expectations may decrease the negative affects. Hence, EH1 intervention which would decrease role conflict or ambiguity will be likely to decrease burnout and increase job satisfaction. Timing control is shown to be significantly related to role strain, job dissatisfaction and burnout. This implies 115 that,in organizations with direct client care, a simple, inexpensive means for affecting role strain and negative affect towards work may be to increase employees' control of the timing of their work hours and the ordering of their work activities within those hours. Timing control emerged in this study as an identifiable factor which seemed to act as a type of control over the work situation and impacted on employee perceptions of role strain, affects, and attitudes toward their jobs. This control of the process of work is especially salient because it does not appear to greatly reduce the amount of work being done, or involve restructur- ing of jobs themselves. Directions for Future Research This study has supported the proposed model of burnout, which includes the elements of role strain, timing control, job satisfacton and burnout itself. There are a multitude of unexplored areas that this study briefly touches on, which invite further analysis. These include other effects of timing control, other variables that cause or moderate burnout, and variables which result from burnout. The variable of timing control has been virtually unexplored in previous literature. It is empirically separate from perceived time pressure, and seems to directly impact on role strain. As an element of work that is not 116 particularly costly for organizations to manipulate, it could be of practical importance in decreasing role strain and increasing job satisfaction. Its interaction with general autonomy and control over the job content seem particularly salient areas for study. It may be that differ- ent types of people value control over the content vs. the context of a job, or it may be that job context control only becomes important when there is a certain amount of job content control. As for the causes of burnout itself, this study examined only a few salient variables. There are many others which may have even greater causal impact on burnout, such as any of those listed on Table 4; this will be known only by further examination. This study utilized path analysis to allow causal infer- ences, but it could not directly test causality, nor could it include any non-recursive variables. This is a dire limitation in burnout research, as there is reason to believe that some of the most focal elements in the occur- rence of burnout have complex interactions and involve feed- back loops. The most obvious of these is job involvement. Most of the case studies and in-depth descriptions of burned out workers, as well as the individuals interviewed for this study, described highly dedicated, job-involved individuals. Maslach (1981), Pines (1981), and Cherniss (1980), limit their definition of burnout to occurring only 117 in human service workers, who seem to be generally highly job involved. Cherniss (1980) describes the high identification with the job that typifies human service professionals; others (Maslach and Pines, 1977; Maslach, 1978; Freudenberger, 1975, 1977; Matingly, 1977; Kahn, 1978), briefly mention the assumption that human service workers are deeply involved with their clients and their jobs. The withdrawal from the job that Cherniss (1980), Freudenberger (1975), and Pines (1981) discuss as part of burnout seems to be even more noticeable due to the extreme initial involvement (Edelwich and Brodsky, 1981). Edelwich and Brodsky place involvement with enthusiasm in the first stage of a process model of burnout. Thus it seems that the variable of job involvement is included in many models of burnout, without being specifically identified as such. Job involvement may be a necessary causal factor in burnout. Its formal identification would allow for the direct measure- ment of the variable, and the integration of previous research on job involvement into the burnout construct. Patchen (1970) found a significant relationship between control over work methods and job involvement. This rela- tionship was increased by a great deal of feedback or job difficulty. Waters, Roach and Batlis (1974) found a signifi- cant correlation between job involvement and work autonomy. Rabinowitz (1975) found a significant correlation between 118 job involvement and Hackman and Lawler's (1971) Job Diag- nostic Survey, measuring "job scope". It has been hypothesized that job involvement is both a cause and result of several factors in the work situation. "Involvement increases as a result of satisfying job experi- ences, and in turn, the more involved a person is, the more effort he or she will exert on the job" (Rabinowitz and Hall, 1981, p. 285). Some empirical support of this view of job involvement has been found (Hall and Foster, cited in Rabinowitz et al.). In the case of burnout, job involvement may be}a complex and important variable. Without a high degree of job in- volvement burnout may not occur; the employee may not care enough about the conditions and occurrences on the job to be severely strained by job stresses (Cherniss, 1980). On the other hand, the negative changes in attitudes that constitute burnout frequently result in decreased job involvement. This can occur through depersonalization of clients (Maslach, 1980, 1981), through emotional exhaustion (Pines, 1981), or simply through withdrawing from the work and reordering the priorities of the aspects of one's life (Korman and Korman, 1980; Cherniss, 1980). The literature provides ample evidence for the import- ance of job involvement in burnout. However, to analyze this variable properly, longitudinal data are needed. 119 Without that, no conclusions can be drawn about the process which creates any relationships that may be found. In order to have a true test of any causal model, including the one presented in this study, experimental or longitudinal data are required. Future research that attempts to determine what causes burnout will have to in- clude such data. Another limitation of this study which should be treated by future research is the need for multiple measures of job satisfaction and burnout. The relationship between these two variables would be clarified by the use of multi- item, multi-faceted measures of job satisfaction, as well as by the use of more behaviorally and cognitively oriented measures of burnout, such as the Maslach Burnout Inventory. The interrelationships of these measures would be very help- ful in understanding where and how job satisfaction and burnout overlap. All the variables in this study were measured by subjec- tive employee perceptions. The very high correlations found may be due in part to the percept-percept limitations of the methodology. Future research must avoid these limitations by including objective measures. These may include objec- tively measured job stresses such as workload, severity of clients, number of hours worked, and time and space limita- tions. Behavioral measures of antecedent and coincidental variables are also needed. Cognitive and physical aspects of 120 burnout have been correlated with turnover and absenteeism; they have been hypothesized as relating to accident rates and performance levels as well. Data on all of these behaviors should be used to define the extent of these relationships, and further delineate the differences between job satisfac— tion and both the cognitive and affective aspects of burnout. Lastly, the results of burnout have been mentioned in this and other studies, but have never been extensively examined. Hypotheses have been raised, costs have been estimated and some correlational data has been gathered, but no complete categorizations and analysis have been attempted. Again, to ascertain true consequences of burnout, longitud- inal data are required. That is the only way to separate out the confusion of which related elements precede burnout, which follow from it, and their potentially intricate interactions. Summary This study reviewed the literature on burnout and its relationships with time control, role strain and job satis- faction. An affectively oriented measure of burnout was developed and its structural integrity and reliability was examined and supported. Six hypotheses of the relationships of the four variables were developed, tested and supported. 121 A process model of burnout was suggested from those relation- ships, and was modified to maximally fit the data. The analyses suggest that burnout results from role strain, job dissatisfaction and lack of control over the timing of work. Timing control decreases role strain, in— creases job satisfaction and decreases burnout. Role strain decreases job satisfaction and increases burnout. Further analysis is required to support the causal assumptions of the model; longitudinal or experimental studies are required to test the specific causal orderings. Multiple measurement, both behavioral and questionnaire, are needed to delineate the constructs of job satisfaction and burnout. The role of job involvement in the process of burnout should be examined longitudinally, in order to examine the sus- pected non-recursive relationships and feedback loops involved. REFERENCES REFERENCES Argyris, C. Integrating the Individual and the Opganization. New York: Wiley, 1964. Abramson, L. Y. Seligman, M. E. P., and Teasdale, J. D. Learned Helplessness in Humans: Critique and Reformu- lation. Journal of Abnormal Psychology, 1978, 81, 49-74. Atchison, T. J., and Lefferts, E. A. The Prediction of Turn- over Using Herzberg's Job Satisfaction Technique. Personnel Psychology, 1972, 25, 53-64. Anderson, C. R. Locus of Control, Coping Behaviors, and Performance in a Srress Setting: A Longitudinal Study. Journal of Applied Psychology, 1977, 62, 446-451. Beehr, T. A. Perceived Situational Moderators of the Rela- tionship Between Subjective Role Ambiguity and Role Strain. Journal of Applied Psychology, 1976, pl, 35-40. Beehr, T. A., Walsh, J. T., and Taber, T. D. Relationship of Stress to Individually and Organizationally Valued States: Higher Order Needs as a Moderator. Journal of Applied Psychology, 1976, Q1, 41-47. Belasco, J. A., and Alutto, J. A. Line Staff Conflicts: Some Empirical Insights. Academy of Management Journal, 1964, 1, 469-477. Birchell, D. W. and Wild R. Perceived Job Attributes, Job Attitudes and the Behaviour of Blue Collar Workers: A Research Note. The Journal of Management Studies, 1976, 13, 191-195. Breaugh, J. A. A Comparative Investigation of Three Measures of Role Ambiguity. Journal of Applied Psychol- Brief, A. P., and Aldag, R. J. Employee Reactions to Job Characteristics: A Constructive Replication. Journal of Applied Psychology, 1975, 60, 182-186. Brief, A. P., and Aldag, R. J. Correlates of Role Indices. Journal of Applied Psychology, 1976, pl, 468-472. 122 123 Brief, A. P., Wallace, M. J., and Aldag, R. J. Linear vs. Non-Linear Models of the Formation of Affective Reac- tions: The Case of Job Enlargement. Decision Sciences, 1976, 1, 1-9. Brodsky, C. M. Long-term Work Stress in Teachers and Prison Guards. Journal of Occupational Medicine. 1977, i2, 133-138. Burke, R. L. Occupational and Life Strains, Satisfaction, and Mental Health. Journal of Business Administration, 1969, 1, 35-41. Capella, J. N. Structural Equqtion Modeling: An Introduction. In P. C. Monge and J. N. Capella (Eds.), Multivariate Techniques in Human Communications Research. New York: Academic Press, 1980. Caplan, R. D., Cobb, 8., French, J. R. P., Van Harrison, R., and Pinneau, S. R. Job Demands and Worker Health: Main Effects and Occupational Differences, Washington, D.C.: U. S. Government Printing Office, 1975. Caplan, R. D., and Jones, N. W. Effects of Work Load, Role Ambiguity, and Type A Personality on Anxiety, Depression, and Heart Rate. Journal of Applied Psychology, 1975, 69, 713-719. Carroll, J. F. X., and White, W. L. Understanding Burnout: Integrating Individual and Environmental Factors within an Ecological Framework. Proceedings of the First National Conference on Burnout. Philadelphia, PA: Gwynedd Mercy College, 1981. Chamberlin, S. Anomie, Burnout and the Wyatt Earp Syndrome. Law and Order, 1978, 48, 20-21. Cherniss, C. Staff Burnout: Job Stress in the Human Services. Beverly Hills, CA: Sage Publications, 1980. Cherniss, C. Burnout: Political, Economic, and Historical Roots. Proceedings of the First National Conference on Burnout, 1981. Corwin, R. G. Patterns of Organizational Conflict. Admin- istrative Scienceguarterly, 1969, 14, 507-520. Cosier, R. A. and Aplin, J. C. Effects of Delegated Choice on Performance. Personnel Psychology, 1980, 33, 581- 593. 124 Cummings, T. G., and Bigelow, J. Satisfaction, Job Involve- ment, and Intrinsic Motivation: An Extension of Lawler and Hall's Factor Analysis. Journal of Applied Psy- chology, 1976, pl, 523-525. Daley, M. Preventing Worker Burnout. Child Welfare, 1979, 58, 443-450. Donahue, R. J. Flexible Time Systems: Flex Time Systems in New York. Public Personnel Management, 1975, 14, 212-215. Dubin, R. Industrial Workers World: A Study of Central Life Interests of Industrial Workers. Social Problems, 1956, 3, 131-142. Dubin, R., and Champoux, J. Central Life Interests and Job Satisfaction. Organizational Behavior and Human Per- formance, 1977, i8, 366-377. Dunham, R. B., and Herman, J. B. Development of a Female Faces Scale for Measuring Job Satisfaction. Journal of Applied Psychology, 1975, 69, 629-631. Edelwich, J. and Brodsky, A. Expectations Versus Limita- tions Linking Training Interventions to Needs On and Off the Job. Proceedings of the First National Confer- ence on Burnout, 1981. Evans, G. Notes on the Impact of Flextime in a Large Insur- ance Company. Occupational Psychology, 1973, 42, 237-240. Fink, E. L. Unobserved Variables Within Structural Equation Models. In P. C. Monge and J. N. Capella (Eds.), Multivariate Techniques in Human Communication Research. New York: Academic Press, 1980. Frankenhauser, M., and Gardell, B. Underload and Overload in Working Life: Outline of a Multi-Disciplinary Approach. Journal of Human Stress, 1976, 2, 35-46. French, R. P., and Caplan, R. D. Organizational Stress and Individual Strain. In A. J. Marrow (Ed.). The Failure of Success. New York: AMACOM, 1972. Freudenberger, H. J. The Staff Burn-out Syndrome in Altern- ative Institutions. Psychotherapy: Theory, Research and Practice, 1975, 12, 73-82. 125 Freudenberger, H. J. Burn-out: Occupational Hazard of the Child Care Worker. Child Carepguarterly, 1977, pg, 90-99. Freudenberger, H. J. Burn-out: The Organization Menace. Training and Development Journal, 1977(a), 5;, 26-27. Freudenberger, H. J. The Burnout-Professional: What Kind of Help? Proceedings of the First National Conference on Burnout, 1981. Freund, J. E. and Williams, F. 8. Elementary Business Statistics: The Modern Approach. Englewood Cliffs, N. J.: Prentice-Hall, 1964. Gannon, M. J. and Hendrickson, D. H. Career Orientation and Job Satisfaction Among Working Wives. Journal of Applied Psychology, 1973, 51, 339-340. Getzels, J. W., and Guba, E. G. Role, Role Conflict, and Effectiveness: An Empirical Study. American Socio- logical Review, 1954, 12, 164-175. Glass, D. C., Singer, J. E., and Friedman, L. N. Psychic Cost of Adaptation to an Environmental Stressor. Journal of Personality and Social Psychology, 1969, .lg, 200-210. Golembiewski, R. T. and Munzenrider, R. Phases in Disaf- fection from Work: Progressive Burnout Mapped by 24 Indicators. Presented at 1981 Annual Meeting, Academy of Management. Golembiewski, R. T. Organization Development (OD) Interven- tions: Limiting Burn-out through Changes in Interaction, Structure, and Policies. Proceedings of the First National Conference on Burnout, 1981. Goodale, J. A. and Aagaard, A. N. Factors Relating to Varying Reactions to the 4-Day Workweek. Journal of Applied Psychology, 1975, fig, 33-38. Greene, C. N. Relationships Among Role Accuracy, Compliance, Performance Evaluation, and Satisfaction within Manager- ial Dyads. Academy of Management Journal, 1972, $5, 205-215. Gross, N., Mason, W. S., and McEachern, A. W. Exploration in Role Analysis. New York: Wiley, 1958. 126 Haas, J. E. Role Conception and Group Consensus. Columbus, Ohio: Bureau of Business Research, 1964. Haavio-Mannila, E. Satisfaction with Family, Work, Leisure and Life Among Men and Women. Human Relations, 1971, 22, 585-601. Hackman, J. R. and Lawler, E. E. Employee Reactions to Job Characteristics. Journal of Applied Psychology, 1971, 55, 259-286. Hackman, J. R., and Oldham, G. R. Development of the Job Diagnostic Survey. Journal of Applied Psychology, 1975, 55, 159-170. Hackman, J. R., and Oldham, G. Motivation Through the Design of Work: Test of a Theory. Organizational Behavior and Human Performance, 1976, 55, 250-279. Hall, D. T., and Gordon, F. E. Career Choices of Married Women: Effects on Conflict Role Behavior, and Satis- faction. Journal of Applied Psychology, 1973, 55, 42-480 Hall, D. T., Schneider, B., and Nygren, H. T. Personal Factors in Organizational Identification. Administra- tive Science Quarterly, 1970, 55, 176-190. Hamner, W. C., and Tosi, H. W. Relationships of Role Con- flict and Role Ambiguity to Job Involvement Measures. Journal of Applied Psychology, 1974, 55, 497-499. Hofstede, G. Alienation at the Top. Organizational Hopp, M. A., and Sommerstad, C. R. Reaction at a Computer Firm: More Pluses than Minuses. Monthly Labor Review, Special Flexitime Reports, 1977, 69-71. House, R. J., Schuler, R. S., and Levanoni, E. Role Con- flict and Ambiguity Scales: Reality or Artifacts? Journal of Applied Psychology (in press). House, R. J., and Rizzo, J. R. Role Conflict and Ambiguity as Critical Variables in a Model of Organizational Behavior. Organizational Behavior and Human Perform- ance, 1972, 1, 467-505. Hrebiniak, L., and Alutto, J. Personal and Role Related Factors in the Development of Organizational Commit- ment. Administrative Science Quarterly, 1972, l5, 555-573. 127 Hulin, C. L. Effects of Changes in Job Satisfaction Levels on Employee Turnover. Journal of Applied Psychology, 1968, 52, 122-126. Hulin, C. L., and Blood, M. R. Job Enlargement, Individual Differences and Worker Responses. Psychological Bulletin, 1968, 55, 41-55. Hunter, J. E., Cohen, S. H., Gerbing, D. W. and Nicol, T. S. "Package 1980: A System of FORTRAN Routines for the Analysis of Correlational Data." Unpublished manu- script, 1980. Hunter, J. E., and Gerbing, D. W. Unidimensional Measure- ment, Second Order Factor Analysis and Causal Models, in B. M. Staw and L. L. Cummings (Eds.). Research in Orgsnizational Behavior, Volume IV. Greenwich, Conn: JAI Press, 1980. Iris, B., and Barrett, G. V. Some Relations Between.Job and Life Satisfaction and Job Importance. Journal of Applied Psychology, 1972, 55, 301-304. Ivancevish, J. M., and Donnelly, J. H. A Study of Role Clarity and Need for Clarity for Three Occupational Groups. Academy of Manggement Journal, 1974, ll, 28-36. Ivancevich, J. M., and Jyon, H. L. The Shortened Work Week: A Field Experiment. The Shortened Applied Psychology, 1977, g3, 34-37. Jackson, S., and Maslach, C. After-effects of Job-related Stress: Families as Victims. Journal of Occupational Behavior (in press). Johnson, T. W., and Graen, G. Organizational Assimilation and Role Rejection. Organizational Behavior and Human Performance, 1973, lg, 72-87. Johnson, T. W., and Stinson, J. E. Role Ambiguity, Role Conflict, and Satisfaction: Moderating Effects of Individual Differences. Journal of Applied Psychology, 1975, 55, 329-333. Jones, J. W. The Staff Burnout Scale: A Validity Study. Paper presented at the 52nd Annual Meeting of the Midwestern Psychological Association. Kahn, R. Job Burnout: Prevention and Remedies. Public Welfare, 1978, 55, 61-63. 128 Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoeck, J. D. and Rosenthal, R. A. Organizational Stress: Studies in Role Conflict and Ambiguity. New York: Wiley, 1964. Kanter, R. M. Work and Family in the United States: A Critical Review and Agenda for Research and Policy. New York: Russell Sage Foundation, 1977. Karasek, R. Job Demands, Job Decision Latitude, and Mental Strain: Implications for Job Redesign. Administrative Science Quarterly, 1970, 54, 285-308. Katz, D. and Kahn, R. L. The Social Psychology of Organiza- tions. New York: Wiley, 1966. Keller, R. T. Role Conflict and Ambiguity: Correlates with Job Satisfaction and Values. Personnel Psychology, 1975, 25, 57—64. Kermish, J. and Kishin, F. Why High Turnover? Social Work Staff Loses in a County Welfare Department. Public Korman, A. K. Environmental Ambiguity and Locus of Control as Interactive Influences on Satisfaction. Journal of Applied Psychology, 1971, 55, 339-342. Korman, A. K., and Norman, R. W. Career Success/Personal Failure: Alienation in Management. Englewood Cliffs, NJ: Prentice-Hall, 1981. Kornhauser, A. The Mental Health of the Industrial Workers. New York: Wiley, 1965. Kraut, A. I. The Prediction of Employee Turnover by Employee Attitudes. American Psychological Association, 1970. Kunin, T. The Construction of a New Type of Attitude Measure. Personnell Psychology, 1955, 5, 65-77. Langer, E. J., and Rodin, J. The Effects of Choice and Enhanced Personal Responsibility: A Field Experiment in an Institutional Setting. Journal of Personality and Social Psychology, 1976, 55, 191—198. Lazarus, R. S., and Launier, R. Stress-Related Transactions Between Person and Environment. In L. A. Pervin and M. Lewis (Eds.). Perspectives in Interactional Psychology. New York: Plenum Presss, 1978. 129 Lefcourt, H. M. Locus of Control: Current Trends in Theory and Research. Hillsdale, NJ: Erlbaum, 1976. Lefcourt, H. M. The Function of the Illusions of Control and Freedom. American Psychologist, 1973, 55, 417-425. Likert, R. New Patterns of Management. New York: McGraw- Hill, 1961. Locke, E. A. The Nature and Causes of Job Satisfaction. In Handbook of Industrial and Organizational Psychology, Marvin D. Dunnette (Ed.), Rand McNally College Publish- ing Co., 1976. Locke, E. A., Smith, P. C., Kendall, L. M., Hulin, C. L., and Miller, A. M. Convergent and Discriminant Validity for Areas and Methods of Rating Job Satisfaction. Journal of Applied Psyghology, 1964, 55, 313-319. Lodahl, T. M., and Kejnes, M. The Definition and Measurement of Job Involvement. Journal of Applied Psychology, 1965, 55, 24-33. Lyons, T. F. Role Clarity, Need for Clarity, Satisfaction, Tension, and Withdrawal. Organizational Behavior and Human Performance, 1971, 5, 99-110. Mannheim, B. A Comparative Study of Work Centrality, Job Rewards and Satisfaction. Sociology of Work and Occupa- tions, 1975, 3, 79-102. Marshall, J. and Cooper, C. L. Work Experiences of Middle and Senior Managers: The Pressure and Satisfactions. Management International Review, 1978, lg, 81-96. Maslach, C. Understanding Burnout: Problems, Progress and Promise. Proceedings of the First National Conference on Burnout, 1981. Maslach, C. Burned-out. Human Behavior, 1978, 5, 17-22. Maslach, C. Job Burnout: How People Cope. Public Welfare, 1978(b), 55, 56-58. Maslach, C. The Client Role in Staff Burn-out. Journal of Social Issues, 1978(c), 55, 11-124. Maslach, C., and Pines, A. The Burn-out Syndrome in the Day Care Setting. Child Care Quarterly, 1977, 5, 100-113. 130 Maslach, C., and Jackson, S. The Measurement of Experienced Burnout. Journal of Occupational Behavior, 1981, 5, 99-113. McGregor, D. The Human Side of Enterprise. New York: McGraw-Hill, 1960. Mechanic, D. Social Structure and Personal Adaptation: Some Neglected Dimensions. In G. V. Coelho, D. A. Hamburg and J. E. Adams (Eds.), Coping and Adaptation. New York: Basic Books, 1974. Miles, R. H. An Empirical Test of Causal Inference Between Role Perceptions of Conflict and Ambiguity and Various Personal Outcomes. Journal of Applied Psychology, 1976, 55, 334-339. Miles, R. H. A Comparison of the Relative Impact of Role Perceptions of Ambiguity and Conflict by Role. . Academy of Management Journal, 1976(b), l5, 35-35. Miles, R. H., and Petty M. M. Relationships Between Role Clarity, Need for Clarity, and Job Tension and Satis- faction for Supervisory and Non-supervisory Roles. Academy of Managpment Journal, 1975, 55, 877-883. Minnehan, F., and Paine, W. S. Burning Money: Analyzing Some Basic Economic and Legal Consequences of Burnout. Proceedings of the First National Conference on Burnout, 1981. Mitchell, T. R. Motivation and Participation: Integration. Academy of Management Journal, , 55, 670-679. Mitchell, T. R., and Pollard, W. E. Instrumentality Theory Predictions of Academic Behavior. Journal of Social Psychology, 1973, 55, 35-45. Newman, J. E., and Beahr, T. A. Personal and Organizational Strategies for Handling Job Stress: A Review of Research and Opinion. Personnel Psychology, 1979, 55, 1-43. Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K., and Bent, D. H. Statistical Package for the Social Sciences, Second Edition. New York: McGraw-Hill, 1975. Nunnally, J. C. Psychometric Theory, Second Edition. New York: McGraw-Hill, 1979. 131 Oldham, G. R., and Hackman, J. R. Relationships Between Organizational Structure and Employee Reactions: Comparing Alternative Frameworks. Administrative Sciencegguarterly, 1981, 55, 66-83. Oldham, G. R., Hackman, J. R., and Pearce, J. L. Conditions Under Which Employees Respond Positively to Enriched Work. Journal of Applied Psychology, 1976, 5l, 395- 403. O'Reilly, C. A., Parlette, G. N., and Bloom, J. R. Perceptual Measures of Task Characteristics: The Bias- ing Effects of Differing Frames of Reference and Job Attitudes. Academy of Management Journal, 1980, 55, 118-131. Organ, D. W., and Greene, C. N. Role Ambiguity, Locus of Control and Work Satisfaction. Journal of Applied Psychology, 1974, 55, 101-102. Orpen, C. Effect of Flexible Working Hours on Employee Satisfaction and Performance: A Field Experiment. Journal of Applied Psychology, 1981, 55, 113-115. Osgood, C. E., Suci, G. J., and Tannenbaum, P. H. The Measurement of Meaning. Urbana, IL: University of Illinois Press, 1957. Palmore, E. Predicting Longevity: A Follow-up Controlling for Age. The Gerontologist, 1969, 5, 247-250. Patchen, M. Participation, Achievement and Involvement on The Job. Englewood Cliffs, NJ: Prentice-Hall, 1970. Pearlin, L. I., and Schooler, C. The Structure of Coping. Journal of Health and Social Behavior, 1978, 55, 2-21. Pines, A., and Kafry, D. Occupational Tedium in the Social Services. Social Work, 1978, 55, 499-507. Pines, A., and Maslach, L. Characteristics of Staff Burnout in Mental Health Settings. Hospital and Community Psychiatry, 1978, 55, 233-237. Pines, A., and Aronson, E. Burnout. Schiller Park, IL: MTI Teleprograms, 1980. Rabinowitz, S. "An Examination of the Influence of Individ- ual Difference Variables and Perceived Job Stimulation on Job Involvement." Unpublished Masters Thesis, Michigan State University, 1975. 132 Rabinowitz, S., and Hall, D. T. Organizational Research ' on Job Involvement. Psycpglogical Bulletin, 1977, 84, 265-288. ‘ ‘— Raven, B. R., and Rietsma, J. The Effects of Differential Clarity of the Group Path and the Group Goal Upon the Individual and His Relationship to the Group. Human Relations, 1957, l5, 29-48. Rehm, L. P. A Self-Control Model of Depression. Behavior Therapy, 1977, 5, 787-804. Rhoads, J. M. Overwork. Journal of the American Medical Association, 1977, 237, 2615-2618. Rizzo, J. R., House, R. J., and Lirtzman, S. I. Role Con- flict and Ambiguity in Complex Organizations. Administrative Science Quarterly, 1970, 55, 150-163. Roberts, K., and Glick, W. The Job Characteristics Approach to Task Design: A Critical Review. Journal of Applied Psychology, 1981, 55, 193-217. Ronen, 8. Flexible Work Hours: An Innovation in the Qual- ity of Work Life. New York: McGraw-Hill, 1980. Ronen, S., and Primps, S. B. The Impact of Flexitime on Performance and Attitudes in 25 Public Agencies. Public Personnel Management Journal, 1981, l5, 201-207. Ruh, R. A., White, J. K., and Wood, R. R. Job Involvement Values, Personal Background, Participation in Decision Making and Job Attitudes. Academy of Management Journal, 1975, 55, 300-312. Saleh, S. D., and Hosek, J. Job Involvement: Concepts and Measurements. Academy of Management Journal, 1976, l5, 213-224. Schein, V. B., Maurer, E. H., and Novak, J. F. Impact of Flexible Working Hours on Productivity. Journal of Applied Psychology, 1977, 51, 463-465. Schuler, R. S. Role Perceptions, Satisfaction, and Perform- ance Moderated by Organizational Level and Participation in Decision Making. Academy of Management Journal, 1977, 55, 159-165. Seligman, M. E. P. Learned Helplessness. Annual Review of Medicine, 1972, 55, 407-412. 133 Shapiro, C. H. Creative Supervision: An Antidote for Burnout. Proceedings of the First National Conference on Burnout, 1981. Shinn, M. Caveat Emptor: Potential Problems in Using Information on Burnout. Proceedings of the First National Conference on Burnout, 1981. Siegel, A. L., and Ruh, R. A. Job Involvement, Participation in Decision Making, Personal Background and Job Behavior. Organizational Behavior and Human Performance, 1973, 5, 318—327. Smith, E. E. The Effects of Clear and Unclear Role Expecta- tions on Group Productivity and Defensiveness. Journal of Abnormal and Social Psychology, 1957, 55, 213-217. Sorensen, J. E., and Sorensen, R. L. The Conflict of Professionals in Bureaucratic Organizations. . Administrative Science Quarterl , 1974, 55, 38-42. Staw, R. M., and Oldham, G. R. Reconsidering Our Dependent Variables: A Critique and Empirical Study. Academy of Management Journal, 1978, 3;, 539-559. Stone, E. F. The Moderating Effect of Work-Related Values on the Job Sc0pe-Job Satisfaction Relationship. Organizational Behavior and Human Performance, 1976, 55, 147-167. Suls, J. and Mullen, B. Life Events, Perceived Control and Illness: The Role of Uncertainty. Journal of Human Stress, 1981, 1, 30-34. Sweeney, D. R. Burnout: Is It Really a Stress Syndrome: Proceedings of the First National Conference on Burnout. Szilagyi, A. D., and Sims, H. P. Locus of Control, Role Dynamics and Job Behavior in a Health Care Institution. In J. G. Hunt (Ed.), Proceedings of the Midwest Division of the Academy of Management, 1975, 104-115. Taylor, K. B., and Weiss, D. J. Prediction of Individual Job Termination from Measured Job Satisfaction and Biographical Data. Journal of Vocational Behavior, 1972, 5, 123-132. Tosi, H. L. Organizational Stress as a Moderator of the Relationship Between Influence and Role Response? Academy of Management Journal, 1971, 55, 7-20. 134 Tosi, H., and Tosi, D. Some Correlates of Role Conflict and Ambiguity Among Public School Teachers. Journal of Human Relations, 1970, l5, 1068-1075. Umstot, R., Bell, E. H., and Mitchell, T. R. Effects of Job Enrichment and Task Goals on Satisfacton and Productiv- ity: Implications for Job Design. Journal of Applied Psychology, 1976, 5l, 379-394. Van Sell, M., Brief, A. P., and Schuler, R. S. “Role Con- flict and Role Ambiguity: A Review and Synthesis of the Literature." Unpublished paper, 1976. Van Sell, M., Brief, A. P., and Schuler, R. S. Role Con- flict and Ambiguity: Integration of the Literature and Directions for Future Research. Human Relations, 1981, 55, 43-71. Vroom, V. H. Work and Motivation. New York: Wiley, 1964. Walker, J., Fletcher, C., and McLeod, D. Flexible Working Hours in Two British Government Offices. Public Personnell Management, 1975, l5, 216-222. Wanous, J. P. Individual Differences and Reactions to Job Characteristics. Journal of Applied Psychology, 1974, 55, 616-622. Wanous, J. P., and Lawler, E. E. Measurement and Meaning of Job Satisfaction. Journal of Applied Psychology, 1972, 55, 95-103. Waters, L. K., and Roach, D. Job Attitudes as Predictors of Termination and Absenteeism: Consistency Over Time and Across Organizational Units. Journal of Applied Psychology, 1973, 51, 341-342. Waters, L. K., Roach, D., and Batlis, N. Organizational Climate Dimensions and Job-Related Attitudes. Personal Psychology, 1974, 11, 465-476. Weissenberg, P., and Gruenfeld, L. W. Relationship Between Job Satisfaction and Job Involvement. Journal of Applied Psychology, 1968, 51, 469-473. Weitz, J. A Neglected Concept in the Study of Job Satisfac- tion. Personnel Psychology, 1952, 5, 201-205. 135 Whispe, L. G., and Thayer, P. W. Role Ambiguity and Anxiety in an Occupational Group. Journal of Social Psychology, 1957, 55, 41-48. Wiener, Y., and Vardi, Y. Relationships Between Job, Organi- zation, and Career Commitments and Work Outcomes--An Integrative Approach. Organizational Behavior and Human Performance, 1980, 15, 81-96. Wilder, J. F., and Plutchik, R. Preparing the Professional: Building Prevention into Professional Training. Proceedings of the First National Conference on Burnout, 1981. Willmott, P. Family, Work and Leisure Conflicts Among Male Employees. Human Relations, 1971, 15, 575-584. APPENDIX QUESTIONNAIRE 136 APPENDIX This part of the survey asks you to describe the way you _generally feel at work. Below are some pairs of words which are opposites. Circle the number between the pair which most closely reflects how you feel at work. EXAMPLE: Calm 1 : 2 : 3 : 4 : 5 Nervous If you generally feel very calm at work, you would circle a "1". If you feel neither calm nor nervous, you would circle a "3". Dissatisfied 1 : 2 : 3 : 4 : 5 Satisfied Appreciated l : 2 : 3 : 4 : 5 Unappreciated Disoriented 1 : 2 : 3 : 4 : 5 Oriented Pressured 1 : 2 : 3 - 4 : 5 Not pressured Stagnant 1 : 2 ° 3 : 4 : 5 Developing Unsure 1 : 2 : 3 : 4 : 5 Confident Recognized 1 : 2 : 3 : 4 : 5 Ignored Rushed 1 : 2 : 3 : 4 : 5 Leisurely Negative 1 : 2 : 3 : 4 : 5 Positive Insecure 1 : 2 : 3 : 4 : 5 Secure Responsible 1 : 2 : 3 : 4 : 5 Not responsible Worried 1 : 2 : 3 : 4 : 5 Unworried Happy 1 : 2 : 3 ° 4 : 5 Unhappy Prepared 1 : 2 : 3 : 4 : 5 Unprepared Underworked 1 : 2 : 3 ° 4 : 5 Overloaded Naive 1 : 2 : 3 : 4 : 5 Sophisticated Wise 1 : 2 : 3 : 4 : 5 Foolish Excited 1 : 2 : 3 : 4 : 5 Bored On a treadmill 1 : 2 : 3 : 4 : 5 Making Progress Energetic 1 : 2 : 3 ° 4 : 5 Exhausted Powerful 1 : 2 : 3 : 4 : 5 Powerless 137 Remember, circle the number which reflects the way you generally feel at work. Uncommitted 1 : 2 : 3 : 4 : 5 Committed Idealistic 1 : 2 : 3 : 4 - 5 Cynical Exploited 1 : 2 : 3 : 4 : 5 Rewarded Utilized 1 : 2 : 3 : 4 : 5 Underutilized Unsuccessful 1 : 2 : 3 - 4 : 5 Successful Active 1 : 2 : 3 : 4 : 5 Passive Uneasy 1 : 2 : 3 - 4 : 5 Comfortable Supported l : 2 : 3 : 4 : 5 Unsupported Optimistic l : 2 : 3 : 4 : 5 Pessimistic Unprepared 1 : 2 : 3 : 4 - 5 Prepared Heard l : 2 : 3 ° 4 ° 5 Not heard Worthwhile 1 : 2 : 3 4 - 5 Worthless Incompetent 1 : 2 : 3 : 4 : 5 Competent Disillusioned 1 : 2 - 3 : 4 ° 5 Hopeful Destructive 1 : 2 : 3 : 4 : 5 Constructive Disorganized 1 : 2 : 3 - 4 : 5 Organized Enthusiastic l : 2 : 3 : 4 : 5 Dejected Dependent 1 : 2 : 3 : 4 : 5 Independent Fulfilled 1 : 2 : 3 : 4 : 5 Unfulfilled Pleased l : 2 : 3 : 4 : 5 Angry Frustrated l : 2 : 3 : 4 : 5 Content Trapped 1 : 2 : 3 : 4 : 5 Free Authentic 1 : 2 : 3 : 4 - 5 Phony Confused 1 : 2 : 3 : 4 : 5 Unconfused Good 1 : 2 : 3 : 4 : 5 Bad 138 Remember, circle the number which reflects the way you generally feel at work. Aimless 1 : 2 : 3 : 4 : 5 Focused Motivated 1 : 2 : 3 : 4 : 5 Unmotivated Incompetent l : 2 : 3 : 4 : 5 Competent Productive 1 : 2 : 3 : 4 : 5 Unproductive Unsure 1 : 2 : 3 : 4 : 5 Sure Relaxed 1 : 2 : 3 : 4 : 5 Tense u N u U) u b u U1 Realistic 1 Unrealistic 139 Please indicate the amount of control that you personally have over the different aspects of your work time, following scale: A = Total Control B = A Good Deal of Control C = Some Control D = Very Little Control E = No Control How much control do you have over: 1. 10. The specific number of hours you work every day? The specific number of hours you work every week? Which specific hours do you work every day? Which specific aspects of your job you work on, at different times during the day? Which specific aspect of your job you work on, at different days of the week? The specific amount of time you take for lunch every day? Which specific time you take lunch every day? The specific amount of time you take for work breaks every day? Which specific time(s) you take work breaks every day? Rearranging your work hours to deal with something special that comes up in your personal or family life? using the 140 JOB SATISFACTION Each pair of faces below has been given a number from 1 through 5. Pick the pair that best describes howpyou feel about youryjob and mark the number of that pair of faces u h 075, 5‘ (Kunin, 1955) JOB INVOLVEMENT Describe your feelings about your job, using the following scale: A = Strongly Disagree B = Disagree C = Neither Agree Nor Disagree D = Agree E = Strongly Agree 1. The major satisfactions of my life come from my job. A B C D E 2. I live, eat and breathe my job. A B C D E 3. I am very much involved personally in my work. A B C D E 4. The most important things that happen to me involve my job. A B C D E (Rabinowitz, 1979, from Lodahl and Kejner, 1966) 141 ROLE CONFLICT AND AMBIGUITY Describe the conditions under which you work, using the following scale: A = Very False B = False C = Neither True Nor False D = True E = Very True 1. I feel certain about how much authority A B I have. 2. There are clear, planned goals and objectives for my job. A B 3. I have to do things that should be done differently. A B 4. I know that I have divided my time properly. A B 5. I receive an assignment without the manpower to complete it. A B 6. I know what my responsibilities are. A B 7. I have to buck a rule or policy in order to carry out an assignment. A B 8. I work with two or more groups who operate quite differently. A B 9. I know what is expected of me. A B 10. I receive incompatible requests from two or more people. A B 11. I do things that are apt to be accepted by one person and not accepted by others. A B 12. I receive an assignment without adequate resources and materials to execute it. A B 13. Explanation is clear of what has to be done. A B 14. I work on unnecessary things. A B (Rizzo, House and Lirtzman, 1977)