. :2. I) I2. .. . .2 3561209222 21-2; 4.. .2. 2... tin? 5222.2 H2. L. .52... . 25523-3222....2222222222 22292 .22.. 4233... 3.3.2... 2.un.2§22.¢u2.222 . n .. I invithuthfiflitl .1. 2.2.2.4120; “.2. .2 .. 5.1%...” 2.1.... 2 . 2.... . l. 1,70! §§l£2.7all~nfi.§ i. .«r T... influx. . V .. 22:22.223223..22h3 Baum”: Nuilzuflg‘; 1 $2555 la! a! , i“:...7:2.= gird... 21.5. Madam.“ “22.x... : 3 5.02.3722»... .5 run; 22.5.3625. .. .. n . saiv.~!|i.r.a$ . . is. . 1...... $.23 35.537... .3 Irv: an. . a 1 ‘ g: In. . It . filial fiiufk“ I lag. 6.210;: Higigkufll? Egvgaa %92» r. «a. a Mun-"22’ 2,2.u.........w...2_ #umamfigzhuwfi; ..2§..2222§ . . 1. 1.. . .2 :. 5267a; _ “frat-L 2..» ..§ “#91 (2‘ ,.2:.......u.1.2.~n, ... figmgmhfiagzs. 2...». . u. .3 2 2322.22.22”... 2 ..2.........2nh.....2..w#3..w:§uw.£u .326»... . .2. s mu 2. 22... 2.42.3 2.5.222. 2&2? 4.3.2.28: 4.. 22.2,2lszit.r.¥7.:.§ ... . t, 2u.21.....:.:. .. 25393.25 b. 2.2 2.. 25 . ..2 ‘ y?! 3:13.!!232‘111! O! 2. 2: . I . . I ‘4... [6:9 2:3) 2 ...2-.fin.u:2u..r§a$um. $22.93!}; ....... 1.. .5 .2 32.32.. .2. 42.9;5332233 l . n IIV‘t- ' . I... I . 5 I In“? . .. P2 . .2? a..." 2 70:15}; 2.2025?! fzf' fl .Il ) gin... gig..-’ 51 §§.r.lall\€.z £0 5:22.254: , LEE. Lair; 2xrllwait§n 2.22322... 2.2. 3.23! it. 2P”... 2.2.. 2......2. rafter .alfi é9§lfiu.a§lbxft.xti:gz 225.. 2 .. 2.2.2.4... - . . 2522.02.21. .43. 2. I A ‘I|.9H.Iunu.£-Iao‘nb~3. . A 2.22349“... 22622 l-.. . .1 lg. 12:354. :r... ifiigihvma 225.394.”. Kg. 1.4...“ .4222. [722.341. .2. 2.51‘213. 2. .. I 8551. gm? 2 I . iiiguggfihfis.i 3522.2... 3 .. . z 2b. . ... 2 12.2.2 . «I! 3 3:! . o .5 3% {in < 4 , if. 2. a 2.. v..pfi.y.§§i.§£urflnuul.ilx{2§i¥ In... .. .1 a , 2 .1. ..s2 .. .. 21.5.... a... 2 s a a v . a923,... .. 9.691.”. 1 .l 22...: .5 , 3 L»... 525...... as. :3:r....2z.:zi¥§ dun-.. .22.... 2...... 4 n. .. .5115. 22... . $2.2. :Lauib...‘ 21.2.3.2. . xtuilrtt I gal-1:5“. ”fquauau - . g . 5-4521... £20224“... 23b .2. 1950a. I. v. '3 .1311 . ~ i ’1’: E‘Lfl . 3' 3". ‘ i Q ’1 . It“; éi LI. ,‘li. G T11}... .. . 225.322.52.227... SNZ¥L<1,&2::W«.H.21 an}: hasn‘t}... Hi222121... % 3.. x . . 5|? >311;£:-7l72€,..r}):gg?§?sf!§3 3,333 gun... , . 1113‘ T .ln‘2g3v.n.42..ill.4flahn.§l! zigaflgla - I 1'7: 51.3 .1! .. I I 1 #2.: . 20...... «£522.23. a..fl...giwv 1 221%.... , 2 mm Ospahw‘tflivl Thanh. “0.0%ng gift?! .. .l l \ z 2 22-95.... 22.2w»-.i§222.¥ it... . . %§H~§.§ 4.5.:- 2:57 - VOA .1" 1: 26.7. v 2.. 2% x, 0 g i .31. ‘- a. a2.“ if I: 22.. . .. . vfifinfahblrxg 5 22.: 2.... 223334.22”... .. 22v: 2 2...... 13% 2.1... .. . 2. 2 7 2.223271;an 4.313.122 5.5....» x . u .. 2. 2 I5 1 .223§.222~m £L22.22§E!hru$ (2...... 20;: 2 l .flflfiyia 2.222. 2...... .. .§.l.r;ll’lét :2 42 2224.212 gaggiififle 2 \n t it‘ll! $2» : Uguflgifllsffflfiizlifl 222.2 gjdzkafibw§2§2§23§ , l‘lli g‘t“ :I.V f f£¥§23.; . ufltggé 3.2.1,; {gulf {5| 2.1.2.;qu 2 .2 .2. ..?222P§H2221§2¥~§522¥., .1217. (r:- fi‘tg 2 I; . lull-2J2 . 7222.28 g;.§. 22222225222305} . tang” , 242.2. (.22 3.2.9 . a. 2 2 if!!! g, a . a: . . . 2. 2 .22. ., 22.? . 71 . «1.. . J 1.... fi , . 2.. . . 1 t . 2 . . 3 2 .2 1 v. . . I! .Iw 2. . .32 2 K . 71 2 . , . 2 , V? )1? . . . l . . . ....u Iu‘l. 1 TI 2. n L.» a 3 sfiflwd‘? g: z 3 1(4th €2.52 .52.: . .§.5L..x .0. : }.-ufl.§!3..fxit . 222 2.. .2 . 2! _ if . . I .2 {#52 .23 a). 2‘. I. '2‘ 129.621 2417?: \ E . . .1 .. I . .2... , 2.2.2.535... . .122. 2.2 '- ugfzigl‘v’ . .uguvillnfisxzflnl .2425)... J I .( £2 2 -2 332.....2... 22:22.22...le222 2: flmfiitflkisi.ft2 . .24 .43... 2 z. . 2 .. \ to .1. . £213,231. , 2. 132.1%? .. s 1.61. .12.... u” A. 10...! .191 I if t. :— g‘kb‘i‘l 2 I . 2 MIN"??? .2 1.13.12.12.32? 3... . .ratf . if}??? V... . 2 .4 ,: lawf22h2yfl22nws 115.1274 E, 121... r 1 v .241. 3532121222.... it! . u . Y. 2. . .. . V . . . "lex K...“I.LII It; .2 a”? ., . .1222? ”H.232._:.1£.x .111"? l E£§Af2.fhu!2§.ffituuv é ;.2H.OMNI‘;§IKI.2222?.¥JH£.(2.2AIIH;2(5‘|N|WI‘I. 2 .24 k 2.14... f§22%.2 3522.25. . .. . fig) .t 7 ii {I It)? in... 42 a: z. 2.42 2 $3.22? 422.22.222.22 2 2 5215:2522 if 132%.: 2 . 2. .i g ~ 2.222(3) llul {(4 I)... 1012.22.19... 2:22Q22H221v4.‘.212..2 (I. 2.3. .. ggsfflgr Hg). . .5 24 .3 :12. 2 25.825)!!! 3.! 72. . I (fighblxilalzng » .. . 7 p7 . {2... .2 222....ww1522125222tm2 222222222222 vb} 22214055552. .. 2.4 . . , .3 . . x I >aa¢§m.h.n.msw..2huhfl$ 23.3.. a . . 2-2. 2cm {Mair}: :7 ..2 . . 2. (fit. 2.. 2... 2 2 r affl}. 2.2 ..2.. . .. . .p,..4| 1.. 2.1;.A\\ 1.. .. a. I I l. y .237... Lllllyi 2 . I...2 b. lllllllllllllllllllllllll\lllllllllzllllllllllll r‘zamu‘fi"°4-JJL;L 1‘ g . . 3 1293 104562 $ I \.-‘ ,5 .’_i t I: l‘ b“ . ‘9, ft ii “- vuuV" wv‘it—vt‘ —- This is to certify that the thesis entitled A“ vacsjhjcfko“ “f GurnovlT ‘IA Hum Cm Semas Uz\u(\°—u5 presented by thmrb L R Mbcu’ has been accepted towards fulfillment of the requirements for M . h - degree in ?SHL\0\0\7 ( J Major professor Date 7 /O/é}1/ / / MS U is an Affirmative Action/Equal Opportunity Institution 0-7639 RETURNING MATERIALS: IV1531_] . Place in book drop to L B remove this checkout from £;!;:;:::;_ your record. FINES will be charged if book is returned after the date stamped below. fie; ' ‘1, l w??? “792‘ 3"" '7 - K '. , . '1, .. .t n .- 45+- ‘W’fl‘n" 0 / ‘4 W fill 37 “F863 w a was MAJ: "$30. "'Riixfi M- 4. “‘ ‘V': w». n"? ‘V ' ’aA'V',‘ . a 15"" l AN INVESTIGATION OF BURNOUT IN HUMAN SERVICES VOLUNTEERS By Richard L. Amdur A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1982 CH7955’ ABSTRACT AN INVESTIGATION OF BURNOUT IN HUMAN SERVICES VOLUNTEERS By Richard L. Amdur The literature on burnout in human services agency workers was reviewed and a model of burnout was proposed and applied to human services volunteers. In order to examine the validity of the model, burnout measures were de- veloped using factor-analytic techniques. Subsequently, relationships between these measures of burnout and volunteer personality, client behavior and type of volunteer training/ supervision were examined, using discriminant analysis. Major findings included significant relationships be— tween: a) initial volunteer personality variables and later volunteer burnout; b) initial client behavior and volunteer burnout. In addition, volunteer burnout was unrelated to client outcome. Finally, it appeared that volunteer person- ality, client behavior and situational variables must be combined in a complex multivariate fashion in order to explain the relationships between these predictors and volunteer burnout. The implications of these findings were explored. The usefulness and validity of the burnout measures and the relative strengths of the predictors were discussed. Suggestions were made for future research. ACKNOWLEDGMENTS I would first like to thank my committee members, William Davidson, Robin Redner and Dozier Thornton, whose guidance and support sustained me though a long and satisfy- ing research experience. Their trust and respect for me are particularly appreciated. Bill Davidson deserves special thanks for helping me clarify my ideas andtranslatetfmmiinto research questions that could be answered empirically. He taught me a great deal about using statistical procedures and then tolerated my need to run countless new analyses with good humor, and approached each new set of results with enthusiasm. My committee members willingly read through numerous earlier drafts of the research proposal and the thesis, and I believe their suggestions resulted in a much—improved final product. Robin Redner, in particular, helped me to express myself in a clear and precise way. Data analysis would probably have taken twice as long and would certainly have been twice as frustrating, had it not been for Lynn Snellman. Her suggestions for managing and using the Adolescent Diversion Project's vast system of files and her help in learning to use the MSU computer were invaluable. She was always willing to help and her enthusiasm was contagious. Many of my early ideas on the subject of burnout were formulated as a result of my experiences as a volunteer and worker at Butler Hospital in Providence. There the relation— ships between administrative policies, staff attitudes and quality of client care became quite clear to me. It was there that I first became aware of the intense feelings (both positive and negative) that can be generated for workers in human services agencies. Many other friends provided support as well as helpful criticism and advice at various points along the way. Neil Bennett was especially generous in this regard. Finally, I wish to thank Daniela Wittmann, who has been helpful in countless ways since the start of this project. Most important to me has been the way our relationship continues to flourish, in spite of the numerous stresses and strains of graduate school. ii TABLE OF CONTENTS LIST OF TABLES CHAPTER I. INTRODUCTION Review of Literature Problems in the Present Burnout Literature A Model of Burnout Burnout in Volunteers Research Questions Measuring Burnout Predicting Burnout From Volunteer Personality, Behavior and Experience The Relationship between Burnout and Changes in Locus of Control The Relationship between Volunteer Burnout and Client Behavior Predicting Burnout from Volunteer and Situational Variables CHAPTER II. METHOD Research Setting Subjects Training and Supervision Measures Measuring Burnout Predicting Burnout Relating Burnout to Other Personality Changes Relating Burnout to Client Behavior Predicting Burnout from Volunteer and Situational Variables CHAPTER III. RESULTS Measuring Burnout Predicting Burnout from Volunteer Variables Self-evaluation Youth—evaluation Peer Ratings Peer Rankings Supervisor Evaluation Course Evaluation iii vi The Relationship between Burnout and Locus of Control The Relationship between Burnout and Client Behavior Client Behavior Prior to Selection Client Behavior During the Project Client Behavior After the Project Change in Client Behavior Predicting Burnout from Volunteer and Situation Variables Self-evaluation Youth-evaluation Peer Rating Peer Ranking Supervisor Evaluation Course Evaluation Summary CHAPTER IV. DISCUSSION Measuring Burnout Predicting Burnout from Volunteer Personality, Behavior and Experience Self-evaluation Youth-evaluation Peer Ratings Peer Rankings Supervisor Evaluation Course Evaluation Summary The Relationship between Burnout and Locus of Control The Relationship between Burnout and Client Behavior Client Behavior (pre—intervention) Client Behavior (during—intervention) Client Behavior (post-intervention) Predicting Burnout from Volunteer and Situational Variables Self-evaluation Youth-evaluation Peer Rating Peer Ranking Supervisor Evaluation Course Evaluation iv 77 80 81 84 85 87 90 93 95 95 95 95 99 102 105 105 106 107 107 107 109 109 110 110 111 113 113 115 117 120 122 122 123 123 124 126 Comments on the Method Discriminant Analysis Calculation of Burnout Scores Comparisons of Predictors Validity of Burnout Scores Directions for Future Research APPENDICES LIST OF REFERENCES 127 127 128 129 130 133 135 146 10. 11. 12. 13. 14. 15. 16. LIST OF TABLES Factors Related to Burnout in Human Service Agency Employees Raw Score Means and Standard Deviations for Burnout Measure Items or Factors Steps in Development of Semantic Differential Factor Analysis of Peer Rating Measure Steps in Development of Peer Rating Factor Analysis of Peer Ranking Measure Steps in Development of Peer Ranking Factor Analysis of Supervisor Evaluation Measure Steps in Development of Supervisor Evaluation Steps in Development of Course Evaluation Factor Analysis of Jackson's PRF Factor Analysis of Behavior Measure Measures Used In This Study Superordinate Factor Analysis Discriminant Analysis Results: Predicting Self—evaluation Burnout From Volunteer Variables Discriminant Analysis Results: Predicting Youth—evaluation Burnout from Volunteer Variables vi 39 41 43 44 46 47 49 50 52 54 57‘ 61 64 71 72 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. Discriminant Analysis Results: Predicting Peer Ratings from Volunteer Variables Discriminant Analysis Results: Predicting Peer Rankings from Volunteer Variables Discriminant Analysis Results: Predicting Supervisor Evaluation from Volunteer Variables Discriminant Analysis Results: Predicting Course Evaluation from Volunteer Variables Correlations Between Burnout Variables and LOC Change Scores Discriminant Analysis Results: Predicting SRD-pre from Burnout Scores Discriminant Analysis Results: Predicting SRD-during from Burnout Scores Discriminant Analysis Results: Predicting Recidivism—post from Burnout Scores Discriminant Analysis Results: Predicting SRD—post from Burnout Scores Discriminant Analysis Results: Predicting PRE—POST SRD Difference from Burnout Scores Discriminant Analysis Results: Predicting Self-evaluation from Volunteer' and Situational Variables Discriminant Analysis Results: Predicting Youth—evaluation from Volunteer and Situational Variables Discriminant Analysis Results: Predicting Peer Ratings from Volunteer and Situational Variables Discriminant Analysis Results: Predicting Peer Rankings from Volunteer and Situational Variables vii 74 75 76 78 79 83 86 88 89 91 94 96 97 98 31. 32. 33. 34. ICU'TJP‘JUOUU Discriminant Analysis Results: Predicting Supervisor Evaluation from Volunteer and Situational Variables Discriminant Analysis Results: Predicting Course Evaluation from Volunteer and Situational Variables Directions of Effects of Volunteer— Related Predictors on Burnout Scores Directions of Effects of Volunteer and Situational Predictors on Burnout Scores APPENDICES Semantic Differential Concepts and Dimensions Peer Rating Items and Factors Peer Ranking Items and Factors Supervisor Evaluation Items and Factors Course Evaluation Items and Factors Jackson PRF Scales and Their Definitions Behavior Measure Items and Factors Self—Report Delinquency Items viii 100 101 108 121 135 136 137 138 140 142 144 145 CHAPTER ONE: INTRODUCTION The problems of low morale at work, high rates of ab- senteeism and turnover, job dissatisfaction and job stress have long been recognized as legitimate areas of study for industrial and organizational psychologists. Attempts have been made to find the causes of these problems and to examine their effects on workers‘ productivity (Vroom, 1964). Recently the focus of this type of research has widened to include workers in human services agencies. Various authors have looked for characteristics of the organizational structure, of the job and of the worker that might be related to one or more of these problems. In addition, an attempt has been made to view all these problems as symptoms of a single syndrome. Researchers have adopted the term ”burn- out" (probably first used by the workers themselves) to describe the syndrome. The snydrome is often believed to include fatigue, hostility, detachment, disillusionment, increased alcohol and/or drug use, physiological symptoms of depression or psychosomatic problems, as well as absen- teeism, turnover, dissatisfaction and other symptoms. This Thesis presents a definition of burnout that inte- grates theories presented in the literature. It also reports the results of a study that attempts to measure burnout, to predict it based on worker, client and situational vari- ables, and to examine its relationship to client outcome. Review of LiteratUre Reports describing the burnout syndrome in nurses and other health professionals (Hall, Gardner & Perl, 1979; Marshall and Kasman, 1980), social workers (Edelwich and Brodsky, 1980; Daly, 1979; Kahn, 1978), policemen (Ellison & Genz, 1978), teachers (Hendrickson, 1979), child care workers (Maslach & Pines, 1977; Freudenberger, 1977), and professional and paraprofessional mental health workers (Pines & Maslach, 1978; Maslach, 1978a) suggest that burnout is a serious problem for human service agency workers. Although no two workers who are burning out show exactly the same symptoms, researchers have identified six major elements which are commonly found: 1) Detachment/Withdrawal: Workers try to maintain a distance between themselves and clients. In doing so, they may also distance themselves from co—workers, friends and family members. They avoid others and are alienated and apathetic. Clients see them as impersonal. They find it difficult to become involved with others or with their work (Freudenberger, 1977; Maslach & Pines, 1977; Kahn, 1978; Maslach, 1978b; Maslach & Jackson, 1978; Daly, 1979; Edelwich and Brodsky, 1980; Marshall & Kasman, 1980; Allesandra, 1981). 2) Hostility/Scapegoating: Blaming the victim, i.e, the client, possibly as a defense against feelings of failure or incompetence. Workers adopt a condescending and intolerant attitude toward clients. They act in ways that are dehumanizing and demeaning and describe clients using derogatory labels. They are chroni— cally hostile and easily angered. This may also be directed at co-workers, supervisors or family and friends, increasing one's isolation and leading to marital conflict. (Freuden- berger, 1977; Maslach & Pines, 1977; Maslach, 1978a, 1978b; Maslach & Jackson, 1978; Daly, 1979; Marshall & Kasman, 1980; Allesandra, 1981). 3) Seeking Security/Inflexibility: Workers are unmotivatedenuiemotionally exhausted. They lack creativity and avoid challenges. They actively fight against change and innovation, seeking security instead. They follow rules mechanically and show rigid and inflexible thinking (Freudenberger, 1977; Maslach & Pines, 1977; Maslach, 1978a, 1978b; Daly, 1979; Edelwich & Brodsky, 1980). 4) Cynicism/Disillusionment: Workers are cynical and nega- tivistic about their own abilities, and the effectiveness of their agencies. They feel helpless and hopeless. They may feel that their abilities are not being fully utilized or that the problems they are supposed to solve are just too immense. Feeling stagnant and disillusioned, they may leave their jobs or change fields (Freudenberger, 1977; Edelwich & Brodsky, 1980; Marshall and Kasman, 1980). 5) Physical/Emotional Symptoms: Minor physical ailments and psychosomatic problems become more common, leading to absenteeism. Drug or alcohol use increases. Workers are frustrated, dissatisfied, and in general, unable to enjoy life. Defensiveness and feelings of resentment and guilt are common, leading workers to seek psychotherapy. They may also become depressed or chronically fatigued (Freuden- berger, 1977; Maslach & Pines, 1977; Kahn, 1978; Maslach, 1978a, 1978b; Maslach & Jackson, 1978; Edelwich & Brodsky, 1980; Marshall & Kasman, 1980; Allesandra, 1981). 6) Over- involvement: Although seemingly contradictory, this facet of burnout is widely reported to occur. It is probably most pronounced in younger workers or those recently hired for a job. These workers are ambitious, idealistic, and en- thusiastic, have high expectations and a lot of energy. They overidentify with clients and become over-involved, so that a client's success or failure (or the success or failure of their agency) has great emotional impact (Maslach & Jackson, 1978; Daly, 1979; Edelwich & Brodsky, 1980; Marshall & Kasman, 1980; Allesandra, 1981). Thus there is agreement on what the observable symptoms of burnout are. But various explanations for these symptoms are proposed. The theoretical explanations fall into three categories: 1) burnout results from a lack of appropriate interpersonal boundaries in the worker-client relationship; 2) burnout results from workers' neediness or ambition or from other personality traits of workers; 3) burnout results from conditions in the work environment that cause stress. Some researchers (Freudenberger, 1977; Maslach & Pines, 1977; Maslach, 1978a, 1978b) feel that burnout is the result of a failed attempt to cope with stressful interpersonal encounters with clients. It is seen as the result of an inability to maintain the proper "therapeutic” attitude to— ward clients. This type of attitude would include a certain amount of empathy, plus the proper amount of distance. Burned out workers substitute mere detachment for what Maslach calls "detached concern.” The reasons for this may be that workers are insufficiently trained in the recognition and productive use of ”countertransference” feelings or that they are in a position where they must try to form relation- ships with clients who, although needy and demanding, ultimately wish to avoid forming relationships. Thus a psychological defense (withdrawal or distancing) is employed which provides temporary relief of tension, but is ulti- mately maladaptive. Another theory proposes that certain personality traits of workers make it likely that these workers will burn out. Ambition (Allesandra, 1981), dedication and the expectation of receiving gratification from work (Freudenberger, 1977) have been mentioned as traits that may lead to burnout. The third theory focuses on a variety of stressors in the work place that may lead to emotional and physical over- load for workers. These stressors include: role conflict (i.e. having to deal simultaneously with conflicting demands of colleagues, supervisors, clients and workers in other agencies) (Kahn, 1978), lack of feedback about effectiveness, unrealistic or inappropriate techniques of measuring client outcomes (Gottheil, 1975), changing technology, rotating staff, interdisciplinary conflict, frequent crises and the need to treat chronic patients (Marshall & Kasman, 1980). A great deal of research has shown statistical relationships between various environmental factors in the work place .(usually some aspect of organizational structure or atmos- phere, or job characteristics) and various aspects of burn— out. For a summary of these factors, see Table 1. It is clear from Table 1 that a large variety of situ- ational factors have been found that relate to some aspect of burnout. Within the group of researchers who study the situational causes of burnout, there is a disagreement about which factors are the most important. The second and third theories of burnout are sometimes seen as mutually exclusive, with some researchers presenting evidence that personal characteristics are more important causes of burnout (Abdel-Halim, 1979), while others show that organizational variables are more important (Zaleznik, deVries & Howard, 1977; Armstrong, 1978; Cherniss & Equatios, 1978; Denton, 1976). There is also a group of researchers who attempt to provide an integrated theory of burnout, combining organi- zational and individual personality factors as precipitants of burnout (Edelwich & Brodsky; 1980, Shinn, 1971; Kamis, 1980; Daly, 1979; Effraty, 1976; Argyris, 1957; Armstrong, 1978). According to this integrated theory, a conflict be— tween workers' needs or desires and the realities of what the work place can provide is at the root of burnout. A a Factors Related to Burnout Factor Table 1 in Human Service Agency Employeesb References Relationship with co—workers, supervisors, administrators Participation in decision- making Ambiguity; Communication; Role- or Value-Conflict Caseload Job Design Opportunity for professional development, promotion Organizational Structure/ Treatment Model: Traditional vs. Innovative Other Factors: 1. Type of Service Provided/ Severity of Client Problems 2. Professional vs. Para- professional Status of Worker Glicken (1977); Price (1977); Armstrong (1978); Effraty (1976); Roehl (1976); Shreisheim and Murphy (1976); Denton (1976); Ray(l975). Cherniss and Egnatios (1978); Bloom and Parad (1976); Armstrong (1978); Gilmour-Barrett (1974); Nogardi (1978); Roehl (1976); Shaefer (1976); Abdel-Halim (1979). Cherniss and Egnatios (1978); Gilmour-Barrett (1974); Hrebniak (1974); Harris (1976); Price (1977); Schreisheim and Murphy (1976); Denton (1976); Tinsley (1974); Mervyn (1978); Shaefer (1976); Roehl (1976); Ray (1975); Hesse (1976). Maslach and Pines (1977); Cherniss and Egnatios (1978); Armstrong (1978). Price (1977); Hrebniak (1974); Armstrong (1978); Sarata and Jappesen (1977); Glicken (1977); Maslach and Pines (1977). Roehl (1976); Ray (1975); Effraty (1976); Price (1977); Glicken (1977). Sarata (1977); Heinstein, Gibbs and Middlestadt (1979); Gilmour-Barrett (1974); Effraty (1976); Lorber and Satow (1977); Maslach and Pines (1977). Quint (1974); Effraty (1976); Cherniss and Egnatios (1978). Sarata and Jeppesen (1977); Effraty (1976); Ray (1975). a . . . . . . Although defined differently in each study, burnout IS often operationally defined as some combination of: dissatisfaction, turnover, absenteeism, alienation, lack of commitment or involvement, psychosomatic symptoms and poor performance. b . . . Most of these studies look at a large number of subjects (generally over 100) either from a single agency or a number of different agencies. A broad range of types of agencies is covered. theoretical model of burnout that takes into account organi— zational, as well as individual personality factors, will be presented later in this section. Problems-in the Present Burnout Literature One important problem in the burnout literature is the confusion of symptoms of burnout with the process of burnout. The pattern of symptoms exhibited by each person who burns out is different (Daly, 1979; Edelwich & Brodsky, 1980, p. 13). The way burnout manifests itself in an individual de— pends on the particular personality traits and situational factors that support the onset and continuation of the pro— cess in that individual. But in order for the concept to be meaningful (i.e. in order for it to be more than an operationalized definition), it must be true that a single, specific underlying psychological process occurs in all cases of burnout. This process will be described in detail in the next section. Through an understanding of this process we can both explain the multitude of forms burnout may take and avoid the confusion of calling every case of dissatis— faction and turnover a case of burnout. Researchers often use operational definitions of burn— out. Thus the dependent variable is often some easily measured trait like turnover or dissatisfaction. Different studies rarely agree on which traits will be used. At present, no valid ”burnout measure” has been published. Without one, it is difficult to integrate results across studies. It is unlikely that different researchers will achieve consistent results in predicting burnout, and in measuring its causes and effects when they are, in effect, comparing apples and oranges. There has been a large gap between theoretical/descrip— tive work, and quantitative work in the area of burnout. A small number of researchers have provided most of the theoretical work so far, while a different group has done related quantitative work. Unfortunately, those doing the theorizing do not often test their theories and those doing the quantitative research are not primarily interested in the concept of burnout. They look for organizational causes for job turnover or dissatisfaction instead. Thus although there is evidence of dissatisfaction or turnover as well as evidence that these are related to role ambiguity or lack of input into decisions, there is very little research show- ing how personality and organizational variables combine to produce burnout. The personality variables that have been studied in relation to burnout, have often consisted of items from SES questionnaires (Denton, 1967; Cherniss & Equatios, 1978). The lack of statistical relationships between these items and various dependent variables have been used to show that organizational variables arenmnmaimportant than personality variables in causing dissatisfaction and turnover. Re— searchers interested in burnout need to define the particular personality traits that predispose one to burnout. Measures 10 of these traits would be more likely to predict burnout than SES variables. 1 The research study to be presented in this Thesis was an attempt to begin to remedy some of these problems. A model of burnout that helps to define the condition will be presented. Operational measures of burnout can then be developed. These measures will be used to examine some of the precipitants to burnout - both in the work situation and in the worker's personality. The effects of worker burn— out on clients will also be examined. A Model of Burnout The model discussed here is a description of the overt signs and symptoms of burnout. It is based on the reports describing burnout in human services workers. A theory of the underlying causes of burnout is also presented. The theory helps to explain why these symptoms, in particular, are manifested. It also explains why certain symptoms are seen early in the process and other later. The theory inte— grates the three explanations for burnout already mentioned: +.& that it results from a loss of boundaries in the worker- client relationship, that it results from workers' personali- ties, that it results from a variety of environmental stressors. After presenting the model of burnout for human services workers, the theory will be applied to human services volun— teers and a new model of burnout in volunteers will be 11 presented. Unfortunately we do not yet have the research tools that would be necessary to test the validity of the theory. In order to do this kind of empirical test, we would need burnout measures that could tell us exactly where in the burnout process the subject was. After the models are presented, a research study that begins the work on develop— ment of such measures will be reported. Several authors have suggested that, rather than being a single emotional state, burnout is a process that occurs in stages (Costello & Zalkind, 1963, pp. 125-129; Edelwich & Brodsky, 1980; Veninga & Spradley, 1981). Each stage is characterized by a specific cluster of symptoms and/or feelings. Similar stages have been described by the various authors. Stage one is characterized by naive idealism and over- involvement in work. In stage one, workers are quite active in their jobs and have high expectations of them- selves, of the agencies in which they work, and of their clients. They are satisfied and feel competent and confi— dent. They are learning new skills and may receive positive feedback from co-workers. Underlying this stage is a process in which the worker focuses on work as the place to meet a variety of personal needs. These may include growth, affiliation, dominance, dependency, succorance, altruism, achievement and security. Stage two is a time of stagnation, unmet expectations and disillusionment, asrhtbecomesapparent to the worker 12 that his or here needs are not being met. This occurs both because of the kinds of needs the worker brings to the work place and because of work-related variables. Job—design variables (like degree of autonomy in scheduling hours, amount of feedback about effectiveness, opportunity for training), organizational structural variables (like amount of participation in agency decision—making, amount and clarity of communication between colleagues, amount of support and cooperation between workers), type of clients and type of goals are some of the work-related variables that may affect the degree to which workers' needs are met. In stage two, unmet needs result in emotional conflict and upset. Workers feel stagnant, disillusioned, less compe— tent, less satisfied than in stage one, and sometimes angry. They may become less involved and less active at work than they were originally, in order to reduce the impact of these unpleasant feelings. This, in turn, leads to greater stagna— tion and dissatisfasction. The third stage of the burnout process istiw wane ammimm- a1 crisis that results from handling stage two ineffectively. Workers who at first had a vague sense that things at work were not quite right may now experience the rude shock of awareness that they are not getting what they want. This experience is often accompanied by a great deal of anger. Where the anger is directed is probably determined by the kinds of needs that are going unfulfilled. The target may be clients, peers, the boss, ”the system" or even family l3 and friends.' The majority of workers who reach stage three feel increasingly tense and dissatisfied. They become less active and involved in work and lower their expectations of themselves, their clients and their agency. They may also begin to experience physiological symptoms of stress. Drug and alcohol use may increase, as one way of coping with these feelings. Workers in stage three often complain to colleagues or family members about the situation at work. They are at their least effective in their work with clients. Attending mainly to their own needs, they are unresponsive to the needs of clients and colleagues as well as friends and family members. Many workers are only able to resolve this crisis through a regressive realignment of needs. This means a descent into stage four: chronic apathy. Those who go the route of chronic apathy manage to reduce the intense dis- satisfaction and tension of stage three. They adjust to the organization in which they work, rather than trying to change it. Although they lose something in the process, they are able to stay at work without feeling tense and miserable. In stage four of burnout, the worker's focus is no longer on needs for growth, altruism or affiliation. Instead, the worker primarily seeks security, and occasionally, nurturance or power. This means workers in stage four resist attempts at change or innovation, they have little desire for involvement with peers or supervisors, for input into 14 decisions, or for learning. They have low expectations for themselves, their clients, their organization and for the system in which they work. They focus attention on money and other material rewards as a substitute for emotional rewards, as Argyris (1957) pointed out. Workers may get involved in unionization activity at this time. Shifting the set of needs that one attempts to gratify in this way produces a basic personality change (Argyris, 1957). It may appear to others that the worker becomes more cynical, more secure at work and more accepting of the status quo. However the underlying and more important change is the realignment of needs. This change helps the worker adapt to an intolerable situation at work. Tension is reduced and satisfaction rises as one makes this shift, because there is less conflict between what the worker needs and what the organization can provide. Workers in stage four are able to function fairly effectively since they are by this time somewhat experienced at their jobs, and are not bothered by tensiOn and dissatisfaction. Their lack of motivation makes it unlikely that they would perform at the level of workers who have reached stage five, however. Along with the regressive realignment of needs, there is an overall decrease in the investment in work as a place to have needs met. This results in a decrease in involve— ment, energy and activity at work and an increase in apathy, during stage four. 15 Stage five is the resolution phase. Workers who reach this point are generally experienced, able to monitor their own reactions and to cope realistically with unmet needs. They do this by readjusting expectations as necessary, by focusing some needs elsewhere (i.e. somewhere other than work), by deciding on a realistic course of action to meet needs that cannot be met elsewhere, and by planning further strategy based on the results of those actions. Workers who reach stage five are able to cope effectively with fluctuating circumstances at work and with their own emotion— al fluctuations. To summarize the burnout stages: Stage One: Workers are very satisfied, involved and active at work, have high expectations of themselves and of clients. They feel effective at work. They may be seen as somewhat naive and idealistic or ambitious by more experienced colleagues. Work becomes a focus for need-fulfillment. Stage Two: Satisfaction, involvement and activity decrease slightly. Workers become somewhat tense or upset. They feel they are less effective than initially, and do not feel they, or their jobs, agencies or clients are living up to their expectations. Stage Three: Satisfaction, involvement and activity decrease dramatically. For some workers, involvement with co—workers may increase, while involvement with clients decreases. Workers are often upset and tense. They feel ineffective at their jobs and are seen as ineffective by peers. This 16 is a state of emotional crisis. Stage Four: Satisfaction is moderate, and is now dependent on salary and job security. Involvement and activity are moderate. Workers are no longer tense or upset, and have lowered expectations dramatically. They are somewhat apathetic and unmotivated, although due to experience at the job, they may function fairly effectively. Stage Five: Workers are satisfied and optimally effective. They are fairly highly involved and active, but not over— involved. They experience tension and emotional upset occassionally, but deal with this effectively, so that it does not chronically impede functioning at work or adversely affect home life. Workers in this stage have realistic expectations. Burnout is resolved. Although several authors (Edelwich & Brodsky, 1980; Veninga & Spradley, 1981) believe that stages similar to these exist, there is disagreement about the order in which they occur. Except for stage one, which all agree is pri- mary, it is likely that different workers progress through the stages in their own unique ways. Stage one (idealism) occurs primarily in individuals with little experience in human service agencies, who are therefore unaware of the kinds and extent of rewards they will get from work. At the same time these individuals have a great emotional investment in work as a place to have their needs met. This investment is a result of personality traits, as well as outside factors: relationships with 17 family, friends, the economic situation, etc. The ability to invest a great deal of energy in work makes these people attractive candidates when new workers are selected. It is possible that more experienced workers will slip back into stage one when they change jobs, take on new responsibi- lities or start working with new clients. In order to reach stage two (disillusionment), it is necessary to pass through stage one. However, there are several ways out of the disillusionment phase. Workers who are able to adjust expectations and set appropriate goals may move toward stage five (resolution). Those who are less able to cope with unmet needs or less able to meet needs outside work, may try even harder to meet all the needs they originally focused on at work. They may push for more learn— ing opportunities, more closeness with fellow staff members, more chances to exert power, to be altruistic or to be nurtured, etc. Because the organization is not set up to grant these wishes, these workers find, once again, that their needs are not met. This time, others are likely to respond less favorably to their requests and they may be seen as ”rocking the boat” or destabilizing the system. Thus they find that rather than achieving what they want, their efforts have a paradoxical effect, and they get even less. Workers who respond this way in stage two go to stage three (emotional crisis). Quitting may be the only way of avoiding further conflict, and is another way of exiting from stage two. 18 Stage three may be reached via stages two or four. Because it is a crisis that often must be resolved through drastic action, it can be a turning point in a worker's life.' Some workers will take the intense emotions that occur at this stage as a sign that something is wrong and will begin to reassess their situation at work. With the help of friends outside the workplace, they may be able to discover the underlying conflict between their needs and what the organization provides, and begin to cope more effectively.- This will occur if they are able to give up the notion that work can satisfy all the needs they originally had in mind for it, if they are able to accept the reality of what it can provide, and if they have friends and activities outside of work that can provide the rest of what they need. Thus they may find a way to resolution of burnout. Workers who are less willing or able to handle the con- flict are likely to quit at this point, feeling frustrated, angry and disillusioned. Other workers (the least fortunate ones) are unable to resolve the conflict, but unwilling to leave work. They will first develop physiological symptoms of depression (sleep disorder, fatigue, lack of appetite and sex drive) or psychosomatic symptoms (headaches, hypertension, gastro— intestinal problems, etc.) Continuing to work in a state of chronic tension and dissatisfaction leads to serious physical problems, drug or alcohol abuse, and eventually to physical or mental breakdown. 19 It can now be seen that to exit stage three into stage four (apathy) is somewhat adaptive. Due to the nature of many human services agencies, for a large number of workers it would be unrealistic to expect satisfaction of higher level needs (growth, achievement, affiliation, altruism) at the workplace. Although reaching stage four has far- reaching implications for the worker, his or her family and (if many workers are apathetic and cynical), for the agency, it is one way of avoiding literally burning oneself out. Although stage four is not necessarily permanent, in fact it often is. Workers who reach this stage sacrifice a great deal in order to remain on their jobs, and are un— likely then to quit. They no longer seekto change things at work or to rise through the ranks and are thus unlikely to reenter stage one. Thus they are quite stable, and are probably valued employees (unlike those in stage three who may be seen as troublemakers). Systems in which pay is based on seniority support this trend. In these systems there is pressure to remain in the chronically burned-out condi— tion. To progress beyond stage four, the worker would probably have to be somewhat ynsuccessful at stifling the original high—level needs (growth, affiliation, achievement, altruism, etc.). If these needs reassert themselves, it may bring back the tension and anxiety of stage three. By this time, the worker may have gained enough experience in a variety of coping techniques to cope more effectively with the per— 20 sonal need-organizational structure conflict. He or she may thus be able to reach stage five. Some workers find themselves going through burnout cycles periodically. Changes in the organizational structure or in the requirements of their jobs may cause work to be less satisfying and rewarding and thus lead to crisis. The burnout of a spouse or close friend in their own job may reduce the amount of support a worker gets from sources out— side of work and may cause an increased investment in work, which again leads to disappointment and emotional crisis. The effects of burnout on families have not been studied, but it would not be surprising to find that chronic apathy and lack of involvement at work generalize to outside relationships as well, causing the burned—out worker to also be a burned out parent and a burned-Out spouse. In addition to making it more likely that the other spouse will burn out at their own job, this lack of emotional rewards at home may add additional force to the focus on material rewards and on acquisition of material goods, mentioned by Argyris (1957). The effects of burnout on the workplace can be deva— stating, severely limiting the agency's ability to function. When burned out workers cease to function effectively or quit, the pressure on their co-workers increases. As staff turnover increases in an agency, it is likely that staff cohesiveness declines. Newer workers are initiated into the "system” by older, more cynical (i.e. chronically burned 21 out) workers. They either learn to "adapt" (by becoming detached and cynical) or they leave. In this way an agency that starts out with excitement and idealism evolves, through a period of crisis, into a rigid, stale bureacracy. Often this means that it is unable to provide the services it was designed for and instead serves only to perpetuate itself. Thus, at the agency level, burnout is a contagious and de— bilitating disease that can be fatal. According to Argyris (1957), who was one of the first authors to notice this problem, workers who reach middle age with limited skills realize that they probably won't ever move up in the hierarchy. They push for constantly increasing wages, security and consumption of material goods instead. Unions fight to increase material rewards. The process is cemented in place as management attempts to in- crease productivity by becoming even more authoritarian, or by acting interested in workers' needs without being willing to make changes. The cycle of high expectations, dissatisfaction, apathy and increased emphasis on material rewards continues. In an era of low productivity and high inflation, this model of burnout has broad implications for American industry and for the American economy. Argyris also believes that the cycle of apathy, lack of ego involvement, frustration and focus on material goods may "permanently modify the basic growth trends” of workers and their families and of entire classes of people. 22 In this model, we can see that burnout is a complex, multi—dimensional phenomenon. It has a variety of causes in the worker's personality, the work environment and in areas of the environment outside the workplace. Its major manifestation is in the worker's personality. Changes here lead to changes in relationships with clients, co—workers, family and friends. The overt symptoms differ for each worker, as does the manner of progressing through the various stages. The symptom-clusters used in this model have been re— ported in the literature as occurring in a wide variety of human services workers. The theory of the underlying changes that explain these symptoms is an integration of theories seen in the literature. The next section will apply this model to human services volunteers. Burnout in-Volunteers Paraprofessionals and volunteers are playing an in— creasing role in providing human services. This is partially a result of the realization on the part of professionals that they alone cannot provide all the mental health care, teaching or medical care that needs to be done. It has also been pointed out that there is little likelihood of there 3233 being sufficient numbers of professionals to provide all the human services that are needed (Cowen and Zax, 1967). At the same time, research, at least in the mental health field, has begun to show that paraprofessionals can often 23 be as effective as (and in some cases more effective than) professionals (Karlsruher, 1974; Durlak, 1979, 1981; Nietzel and Fisher, 1981). Although a great deal of research has been done on job satisfaction, job stress, and burnout in professional and paraprofessional employees of human services agencies, little work has been done with a focus on volunteers. The model of burnout presented here so far is meant to apply to full- time workers. But volunteers differ in some fundamental ways from full—time workers, whether professional or para— professional. This is especially true of the volunteers to be discussed in this study, who are mostly college students, between the ages of 18 and 21. These differences between volunteers and regular workers might be expected to affect the susceptibility of volunteers to burnout. Volunteers usually work few hours per week, often for a specified and limited period of time, and are not paid. This may mean that they will be less involved at work, and that they will develop whatever feelings they do have more slowly than full-time workers. It is thus less likely that they will be found in the fourth stage of burnout. It probably takes a year Or more to reach this stage of chronic apathy. Thus there is some reason to believe that volunteers are less likely than full—time workers to burn out. However, there are also a number of factors that may make it more likely that volunteers will burn out. Volun— teers are likely to be young, inexperienced, idealistic and 24 enthusiastic, and to have unrealistic expectations of what impact they will have and what their rewards might be. Be— cause they often lack training and experience, they have to rely on commitment, motivation and intuition. These qualities may be enough to help them succeed, but initial failure, by decreasing the level of motivation and causing distrust in one's intuitions, may make further failure more likely. If volunteers are young and unmarried, they may find it harder to rely on outside relationships to provide support. The lack of security and tenuousness of their posi- tion may be stressful in itself. Finally, volunteers rarely have responsibility for making final decisions about the way cases will be handled, and are often expected to imple— ment the decisions of others. This makes it likely that they will feel their skills are not being fully utilized. Thus it does seem possible that even volunteers in human service agencies may burn out. Because volunteers are likely to have many of the per— sonal characteristics that predispose one to burn out (little human services experience, investment in work, enthusiasm and idealism); because the needs they seek to satisfy at work do not generally include security, while there might be a bigger-than-usual emphasis on growth and altruism; and because their volunteer activity is temporary, it is likely that burnout will not occur for them in exactly the same way as in full-time workers. Instead of a five-stage process occurring over a period of a year or two, burnout in volun— 25 teers is likely to be seen as a gradual degeneration in motivation and activity, in expectations of self and client, in involvement and empathy with clients and colleagues, in satisfaction and feelings of competence, and possibly as an increase in feelings of tension, emotional upset, and feelings of helplessness. Volunteer burnout can thus be seen as a two-stage process. Stage one is like the over-involvement stage for full—time workers, only more intense, since volunteers are more naive and idealistic, probably have higher initial expectations, more energy, and must make a greater attempt to become involved. Stage two is an extended and more in— tense version of the full-time workers' stagnation and dis— illusionment phase. Volunteers may resist giving up their initial expectations and commitment because these may be the only tools they have. They may resist acknowledging that their needs for growth, affiliation and altruism are not being met, because these may be all they came to the experience for. They are able to experience stage two more fully only as it becomes time for them to leave, and they begin to disengage from the volunteer experience.. Thus they are likely to feel stagnant, dissatisfied, disillusioned and incompetent by the time the experience ends. Burnout in volunteers is serious because it may occur at a time when career choices are being made and when social attitudes and values are beginning to solidify. Since the negative effects reach their peak just as the volunteer 26 leaves, the experience may have a permanent impact. Volunteer burnout is also a serious problem for the agencies that use volunteers. If the personal qualities that lead people to become volunteers, and that agencies look for in selecting volunteers, are the same qualities that predispose one to burn out, many volunteer programs will be doomed to failure from the start. Weinstein, Gibbs and Middlestadt (1979) find that although volunteers in human service agencies don't have much impact on policy, rules or decision-making, they do help to decrease staff workload and improve quality of service. This is especially true in agencies with more non—professional workers (i.e. in 'alternative' agencies). These authors suggest that volunteer programs help relieve staff burnout, especially in undermanned settings, by giving staff a chance to super- vise or teach (which extends staff's sense of competence and influence, as well as taking them away from direct client contact), by reminding them of idealistic attitudes they once had, by stimulating evaluation and rationalization of programs, and by stimulating thinking and planning about incorporating new resources. This suggestion is supported by research showing that supervisors and administrators are less likely to burn out than workers who only provide direct services (Hrebniak, 1974; Price, 1977; Maslach & Pines, 1977). The danger is that, rather than revitalizing staff, volunteers will become the new ”low-man—on—the—totem—pole,” will be given tasks no one else wants, and looked down upon 27 by those above them on the job hierarchy. Volunteers treated this way are unlikely to stay (Ray, 1975; Effraty, 1976; Sarata & Jeppesen; 1977). The effect of a failed volunteer program on an agency that may already have a burnout problem is not known. But we can imagine that a gift given and then taken away would only increase workers' withdrawal and dis— trust. To summarize the burnout model for volunteers: Stage One: Volunteers are motivated and committed, enthusi— astic, involved (possibly over-involved), and empathic, with co-workers and clients. They are idealistic and have high expectations. They are satisfied with the volunteer ex- perience. They are not especially tense or upset, although they may become anxious at times. (Anxiety, however, is not a defining characteristic of this stage, since other new volunteers, who will not burn out later, may also be anxious initially). Stage Two: The volunteer is now dissatisfied with the ex- perience, is less enthusiastic, less active, less involved and empathic, has lower expectations, may feel helpless, and is likely to feel tense and upset. Thus the model developed to explain worker burnout can be applied to human services volunteers. A good deal of empirical work must be done, however, in order to bring this type of discussion beyond the level of armchair theorizing. A study that begins the work on development of burnout measures will now be presented. 28 Research Questions Measuring Burnout The first question to be explored was how burnout could be measured. Several paper-and-pencil measures that seemed to be related to the changes described in the model of volunteer burnout were used. Changes in several dimensions (for example: involvement, hostility, flexibility, expecta— tions) had been described as part of the burnout process in the literature. The factor structure underlying the burn— out measures was examined to find out a) whether a set of factors similar to those described in the literature could be found underlying the burnout measures; b) whether there was in fact more than one dimension underlying the measures; and based on these results, c) how the data in the burnout measures should be combined to produce a set of ”burnout scores” for each volunteer. Predicting Burnout From Volunteer Personality, Behavior and Experience A number of the studies describing burnout mentioned an initial state of naivete, high expectations and involve- ment, which led to burnout. In the model of burnout already presented, this was labeled stage one. However, the debate continues in the quantitative literature on job satisfaction and turnover whether personality variables have any impact at all on worker burnout. The second part of this study examined the relationship between the burnout scores derived earlier and initial volunteer personality, behavior in 29 relation to youth and experience in human services settings. The questions being examined were a) whether burnout could be predicted using volunteer personality, behavior and experience variables alone (that is, without using any predictors measuring situational factors), b) whether the volunteer variables that did predict burnout were those hypo- thesized by the burnout model to compose stage one of burnout. The-Relationship Between Burnout and Changes in Locus of Control It has been suggested that burnout results in, or is accompanied by, significant personality changes. In an attempt to understand the relationship between burnout and other personality changes, the relationship between burnout scores and concurrent changes in the volunteers' locus of control was examined. There were two hypotheses about this relationship. It might be expected that as volunteers begin to feel less competent and in control they would move toward the external pole on the locus of control measure. An alternative hypo- thesis might be that direction of change in locus of control would depend on the volunteer's pre-existing personality structure. Those volunteers who already leaned in the ”external” direction (for example, those who might be described as having a more ”hysterical” personality style) would go further in the ”external" direction as they burned out and felt more out of control. Those volunteers who 30 initially leaned in the "internal” direction (for example, those who might be described as haVing an ”obsessive- compulsive” personality style) would go further in the "internal” direction, as they burned out and became more rigid and controlled. The Relationship Between-Volunteer Burnout and Client Behavior It is often assumed that burnout leads to decreased - effectiveness in work with clients (Marshall and Kasman, 1980; Edelwich and Brodsky, 1980; Maslach and Pines, 1977). It is also generally assumed that workers who have ”tougher” or ”more difficult” clients are more likely to burn out, because tougher clients may change more slowly, providing less positive feedback, and may be more difficult to engage in a therapeutic relationship (Gottheil, 1975; Freudenberger, 1974). However, there is little empirical evidence that either of these assumptions is true. In this part of the study, the relationship between burnout of volunteer workers and client behavior was examined. Predicting Burnout From Volunteer and Situational Variables In this part of the study, two types of questions were asked. The first type of question was concerned with the variables used to predict burnout. The debate in the literature about the relative importance of worker person— ality, versus situational predictors of burnout has already been discussed. In this part of the study, empirical data related to this question were examined. Both volunteer— 31 related and situational variables were used as predictors of burnout. The relative predictive power of each type of variable was examined. The second type of question in this part of the study focused on the burnout measures themselves. It was first asked whether some of the burnout measures were better than others at measuring burnout. This question could be stated: which measures have the best validity, or which measures are the best operational definitions of volunteer burnout, as described in the theoretical model. This could be answered by examining the strength and direction of the relationships between predictors and burnout scores. The burnout scores with most validity were those which had the strongest relationships to predictors operating in the directions specified by the model. The final question was: what are the other ”burnout measures” (i.e. the ones that are not good operational definitions of burnout) measuring? Although this question could not be answered firmly, hypotheses could be arrived at, based on the predictors that were related to these burn- out scores, and their directions of effect. CHAPTER TWO: METHOD Research Setting The research to be presented took place at a diversion project for delinquent youth, funded by the National Institutes of Mental Health. Youth are diverted from normal processing through the juvenile court system and referred instead to the project. Each youth is seen by a trained non—professional volunteer on a regular basis, for a period of approximately six months. Each volunteer works with a single youth. The project has been in existence since 1972. Data from the Fall of 1977 through Spring of 1979 was analyzed. Youth are referred to the study by the court if they have a court petition filed against them by police, school, parents or others, if the court accepts the petition, and if they admit to the charges against them during an inquiry or preliminary hearing. Youth are generally of high school or junior high school age, approximately 15 percent are female and approximately 30 percent are non—white. The majority come from single—parent families. Criminal activi— ties include a wide range of misdemeanors and non-serious felonies. Subjects Selection College students at a major midwestern university inter— ested in participating in a three-term psychology field 32 33 experience sequence attended an initial information session. If still interested, they were asked to fill out several paper-and—pencil measures, and to sign contracts agreeing to participate in the study and to be available for further assessments in one year. Since many more students were interested than could be accommodated in the course, students were informed that selection would be made randomly. Those who attended a second pre-selection meeting were asked to complete a second set of measures. Only those who attended both sessions were eligible to participate. These students were then placed in one of several experimental training conditions or in a waiting/control group, or were rejected. This selection was done randomly, stratified by sex. The few students in experimental conditions who dropped out after this point and before training had begun were replaced by a same sex individual chosen at random from the waiting/ control group. Very few students dropped out after training had begun. These students were not included in the analysis. Fifty-five students participated in one of the experi— mental conditions during the 1977-78 academic year, 36 during the 1978-79 academic year. Most were juniors and seniors majoring in psychology or criminal justice. Most planned a career in a human services field or graduate education in a related discipline. Fifty-four percent of the volunteers were Protestant, 42 percent were Catholic and 4 percent were Jewish. Volunteers came from families with an average of 4.3 children. 34 Thirty-one percent were the oldest child in their family, 23 percent were youngest, 45 percent were in the middle, and 1 percent were only—children. They had participated in an average of 7.6 activities (i.e. clubs, teams, etc.) in high school. Mean grade—point average for the volunteers was 2.9. Most had taken between two and four Psychology courses, one Criminal Justice course and one other related social science class (e.g. Social Work, Sociology, etc.). Eight percent had had a similar field experience course. Politically, the volunteers rated themselves as slightly liberal. Eight percent had been arrested, for shoplifting, possession of marijuana or disorderly conduct. Training and Supervision The project director trained, supervised and consulted with a group of graduate students, and several under— graduates, who were then responsible for training and supervising the undergraduate volunteers. Those students selected to be in one of the experimental training conditions were placed in a small group of students, with either one or two supervisor/trainers. The project employs several different types of training and supervision. The training/supervision groups vary in intensity (number of students in the group and frequency of meetings) and mode of intervention (focus on behavioral contracting, advocacy, therapeutic relationship, or natural volunteer skills). In general, training involves didactic presentations, quizes, homework assignments, roleplays and 35 discussions. The topics covered in different groups include: an introduction to the program, the justice system and court, theories of delinquency, essential ingredients of therapeutic relationships, use of assessment prior to intervention, selecting appropriate interventions, monitoring compliance, as well as training in the specific intervention techniques to be employed. Training/supervision groups meet weekly (high intensity) or monthly (low intensity) for two hours. Supervision focuses on clarifying what the volunteers have learned about their youth and how they feel about this, on planning strategy for intervention, and on developing goals. Students in all training/supervision groups are expected to spend six to eight hours per week with their youth. Measures The model of burnout in volunteers specified that the measures used in studying volunteer burnout should focus on: the volunteers' attitudes and expectations about them— selves, their clients, their agencies; the volunteers' level of involvement, empathy, activity and enthusiasm with colleagues and clients; and their overall satisfction with the volunteer experience. Changes in a direction of reduced expectations, more negative attitudes, less empathy, activity, enthusiasm and satisfaction indicate. burnout. Based on the model, the measures of stage one of volun- teer burnout should examine the volunteer's need or ability to become involved with others, his or her ability to be ammthu: 36 and the strength of needs to achieve, to nurture and to gain the support of nurturance of others. Naivete might be measured by looking at the level of experience in human services work. The measures used in this study were chosen from an extensive set of measures administered at up to five time periods during the nine—month academic year. Five methods were used to measure burnout. Measuring Burnout Semantic Differential. The Semantic Differential was used to measure volunteers' attitudes toward different con— cepts. It was administered at orientation (pre) and at the end of Spring term (post). The original items for this measure consisted of 22 concepts, like ”adolescents,” "parents of adolescents,” ”juvenile justice system,” ”police, ”myself,” and ”volunteers.” For a complete list of semantic differential concepts, see Appendix A. Each concept was rated along six dimensions, such as pleasant—unpleasant, good-bad, and friendly—unfriendly, using a seven-point Likert scale. For a complete list of dimensions, see Appendix A. According to the model of burnout presented here, one indicator of burnout is a more negative evaluation over time of concepts related to oneself and one's clients. This measure had been developed in previous research (Mitchell, Kantrowitz & Davidson, 1980). A principle— components analysis with varimax rotation and Kaiser's criterion produced 23 orthogonal factors, one for each of 37 the 22 original concepts and one for the predictable- unpredictable dimension of evaluation. The "predictable" factor had been discarded. The remaining factor scores had been refactored, using the same technique, producing three orthogonal superordinate factors. Thus 22 concepts, each rated on six dimensions, had been reduced to three factor scores. Concepts are grouped under factor headings in Appendix A. The three factors were Attitudes Toward Self, Attitudes Toward Youth, and Attitudes Toward the System. Only the first two factor scores were used in further analyses in the present study, since the third, Attitudes Toward the System, could easily change in a negative direc- tion for reasons other than burnout (for example, due to training). In the model, burnout was viewed as a change in volunteers' attitudes and behaviors over time. In measuring burnout, it was therefore necessary to look at differences in attitudes and behaviors between two time periods: one at the start of the volunteer's experience, the other at the end. Thus change—scores of each of the two Semantic Differential factors were used to measure burnout. Prior to calculating change—scores, each factor score was standard- ized at both time periods, across subjects. This was done to put both scores in the same metric and to equate their distributions. Difference scores were then calculated for both factors by subtracting the time—two standard score from the time—one standard score. Factor scores were coded so 38 that a positive difference indicated burnout. A negative difference indicated change in attitudes in the direction opposite to burnout. Because standardization at times one and two masks over— all groupchanges over time, raw-score means for each factor were examined prior to any further analysis. Raw score means and standard deviations for each factor are listed in Table 2. Neither factor shows significant change in raw-score mean over time. The steps taken in developing the Semantic Differential for use as a burnout measure are summarized in Table 3. Peer Rating. The Peer Rating measure was administered before training (pre) and at the end of Spring term (post). The. measure consisted of 23 attributes, each followed by a five—point Likert scale. Each volunteer was rated on each attribute by all the members of his or her supervision/ training group. Only data from peersrwne analyzed (that is, data from self and supervisors were not analyzed). Examples of attributes include ”ability to get along with others,” ”perseverance," ”empathy,” and ”effectiveness in using skills.” The 23 original items are listed in Appendix B. 0f the 23 original items, ten were chosen that relatedto some aspect of burnout. For example, ”empathy” was chosen because according to the theory of burnout, a decrease in a volunteer's level of empathy is one indicator of burnout. ”Intelligence" and ”well-liked by supervisor" were not chosen Raw Score Means and Standard Deviations for Burnout Measure Items or Factors 39 Table 2 Item or Within- Time 1 Time 2 Measure Factor Mean SD Mean SD Semantic Differentiala Self 4.11 7.17 5.41 9.97 Youth 3.32 7.31 5.33 9.06 Peer Ratingb Effectiveness 3.13 .46 3.56 .60 Planfulness 3.71 .57 3.82 .59 Perseverance 3.76 .54 3.91 .56 Resourcefulness 3.37 .59 3.71 .52 Motivation 3.81 .57 3.96 .52 Empathy 3.60 .54 3.77 .53 Harmth 3.72 .57 3.99 .57 Input 3.53 .70 3.61 .77 Interest 3.61 .54 3.73 .62 Helpfulness 3.44 .54 3.55 .61 a . . Hithln measure factors are used. bOriginal items are used. 40 Table 2 (Cont.d.) Item or Within- Time 1 Time 2 Measure Factor Mean SD Mean Supervisor Evaluationb Helpful Consistent Concerned Knowledgeable Resourceful Involved Supportive Comfortable Available Understanding Conscientious "Expert" Generates Discussion Familiar with Resources Got Along . a Course Evaluation Overall Satisfaction Talkative Outside Discussion Training 4.11 2.81 4.34 3.68 .06 .02 .88 .73 .86 .86 .11 .OO .79 .80 .87 .07 .79 .90 .79 .83 .80 .58 .71 3.16 3.15 4.44 4.32 4.22 3.11 4.04 3.03 3.07 3.10 3.14 3.18 3.93 4.01 3.12 4.21 3.51 .81 .66 .67 .70 .70 .69 .14 .78 .77 .60 .58 .05 .81 .70 .69 .82 .80 .64 .78 a . . Within-measure factors are used. 0 . . . Orlglnal items are used. 41 Table 3 Steps in Development of Semantic Differential Original Measure 22 concepts, each rated by vol Unteers on six dimensions. Steps in Development 1. 2. Principle—components analysis of original items, producing 23 factors. Superordinate factor analysis of 22 of these factors, producing three super~ ordinate factors. Two of these factors chosen for further analysis. Standardization of factor-scores at two time periods. Calculation of two differences over time (i.e. ”change scores"). Within—Measure Factors 1. 2. Change in attitude toward self. Change in attitude toward client. Final Burnout Scores 1. Both within-measure factor scores were used as burnout scores. 42 because changes in these attributes do not necessarily indicate burnout. The ten Peer Rating items that were further analyzed are listed in Appendix B. Each item was standardized at times one and two, and ten change scores were calculated. Table 2 again shows that raw-score means for each item did not change significantly over time. The ten change scores were then factor—analyzed. Factor analysis was used to help reduce the number of items to be studied. This also provided more reliable ”items” (actually factor scores). It helped in interpretation of item content as well. The ten change scores were factor—analyzed using principle-components analysis with varimax rotation. Kaiser's criterion and the scree method (Cattell, 1965) were used to first estimate the number of factors. Of the possible solutions, the one that made the most sense on theoretical grounds was chosen. Using these procedures, the three-factor solution was chosen. Factor loadings for this solution are listed in Table 4. The three Peer Rating factors were called Perse— verance at Task, Involvement in Group and Effectiveness/ Empathy. Items are listed under factor headings in Appendix B. The steps taken in developing the Peer Rating for use as a burnout measure are summarized in Table 5. Peer-Ranking. The Peer Ranking measure was also ad- ministered before training and at the end of Spring term. For this measure, each volunteer ranked all the members of 43 Table 4 Factor Analysis of Peer Rating Measure Eigenvalues and Proportion of Variance Factor Eigenvalue % of Variance Cummulative % 1 4.912 49.1 49.1 2 1.145 11.4 60.6 3 .837 8.4 68.9 4 .699 7.0 75.9 5 .561 5.6 . 81.5 6 .533 5.3 86.9 7 .452 4.5 91.4 8 .378 3.8 95.2 9 .294 2.9 98.1 10 .190 1.9 100.0 Rotated Factor Matrixa Factor 1 2 3 Item (Perseverance) (Involvement) (Effectiveness/ Effectiveness .434 -.015 .684 Planfulness .655 .392 .175 Perseverance .834 .135 .202 Resourcefulness .650 .133 .310 Motivation .678 .251 .172 Empathy .216 .315 .768 Warmth .202 .401 .766 Input .180 .892 .081 Interest .175 .699 .361 Helpfulness .319 .783 .239 a . . . . . . Varlmax relation with Kalser normalizat1on was used. 44 Table 5 Steps in Development of Peer Rating Original Measure 23 attributes, rated on a 5—point scale, by fellow supervision-group members. Steps in Development 1. Ten items related to burnout model chosen. 2. Standardization of each item at two time periods. 3. Calculation of ten change scores. 4. Factor analysis of change scores. Within-Measure Factors 1. Perseverance at task 2. Involvement in group. 3. Effectiveness/Empathy. Final Burnout Score. 1. Average of three within—measure factor scores. 45 his or her group on ten attributes. The attributes are listed in Appendix C. The procedure here was the same as that used for Peer Ratings, except that, because only ten items were available, all were used. All ten items were standardized at both time periods, and ten change scores were calculated. Raw-score means are not listed in Table 2 because, due to the nature of ranking data, all means were virtually the same. The ten change scores were factor—analyzed, using the same method as for Peer Ratings. This produced a five—factor solution. Factor loadings are listed in Table 6. The five Peer Ranking factors were labeled Talkative, Gets Along With Others, Motivation, Empathy and Perseverance. Items are listed under factor headings in Appendix C. Table 7 lists the steps in development of the Peer Ranking measure. Supervisor Evaluation. The Supervisor Evaluation measure consisted of 15 items. Each item was answered on a five—point Likert scale. Each volunteer rated each of his or her supervisors on the 15 items. The measure was administered at the end of training and at the end of Spring term. The Supervisor Evaluation items are listed in Appendix D. The procedure used with this measure followed that used for Ratings and Rankings. The fifteen scores were standard— ized at both time periods, and change scores were calculated. Again Table 2 shows no significant change in raw-score means over time for these items. The fifteen change scores were 46 Table 6 Factor Analysis of Peer Ranking Measure Eigenvalues and Proportion of Variance Factor Eigenvalue % of Variance Cummulative % 1 3.395 42.4 42.4 2 1.169 14.6 57.0 3 1.082 13.5 70.6 4 .733 9.2 79.7 5 .541 6.8 86.5 6 .438 5.5 92.0 7 .378 4.7 96.7 8 .263 3.3 100.0 Rotated Factor Matrixa Factor 1 2 3 4 5 (Talkative) (Gets Along) (Motivation) (Empathy) (Perseverance) Confident .906 .190 -.054 —.004 .186 Talkative in class .768 .108 .450 .005 —.009 Gets along with youth .162 .864 -.100 .188 .123 Perseverance .093 .167 .163 .132 .946 Motivation .211 .096 .849 .255 .243 Well liked by students .166 .824 .331 —.022 .098 Empathy .052 .119 .206 .936 .137 Input .629 .184 .458 .299 -.054 a . . . . . . Var1max rotation with Kaiser norma11zation was used. 47 Table 7 Steps in Development of Peer Ranking Original Measure 10 attributes, on which each volunteer is ranked by fellow supervision—group members. Steps in Development 1. Standardization of ten items, each at two time periods. 2. Calculation of ten change-scores. 3. Factor analysis of change-scores. Within-Measure Factors 1. Talkative. 2. Gets along with others. 3. Motivation. 4. Empathy. 5. Perseverance. Final Burnout Score 1. Average of five within-measure factor scores. 48 factor-analyzed, using the procedure already described, producing four factors. Factor loadings are listed in Table 8. The four Supervisor Evaluation factors were Knowledgeable/ Concerned, Available/Helpful, Quality of Discussion and Involved. Items are listed under factor headings in Appendix D. Steps in developing the Supervisor Evaluation measure are listed in Table 9. Course-Evaluation. The Course Evaluation was also filled out by each volunteer after the training period had ended and at the end of Spring term. The Course Evaluation measure was a 36—item questionnaire measuring volunteers' satisfaction with various aspects of the field work experi— ence. In previous research (Davidson & Redner, 1982), factor analysis of this measure had produced a 12-factor solution. Four of these factors were chosen for further study, based on their relevance to the theory of burnout. The four factors that were kept were Satisfaction with Class Dis- cussion, Satisfaction with Course, Talkative Outside Group and Satisfaction with Training. Some of the factors discarded were Relevance to Career, Satisfaction with Grading Scheme, and Satisfaction with Outside Readings. Items that compose the factors that were kept are listed under factor headings in Appendix E. Each of the four factor scores was standardized at both time periods. Four change scores were then calculated, one for each factor. Table 2 shows that there was no significant 4+9 Table 8 Factor Analysis of Supervisor Evaluation Measure Eigenvalues and Proportion of Variance Factor Eigenvalue % of Variance Cummulative % 1 2.731 22.8 22.8 2 1.576 13.1 35.9 3 1.326 11.0 46.9 4 1.235 10.3 57.2 5 .916 7.6 64.9 6 .818 6.8 71.7 7 .776 6.5 78.2 8 .698 5.8 84.0 9 .684 5.7 89.7 10 .519 4.3 94.0 11 .411 3.4 97.4 12 .309 2.6 100.0 Rotated Factor Matrixa Factor 1 2 3 4 (Knowledgable/ (Available/ (Discussion Item Concerned Helpful) Quality) (Involved) Helpful .116 .682 .103 —.340 Concerned .712 .166 .098 —.162 Knowledgable .695 -.O78 .079 .161 Resourceful .807 .016 .128 —.001 Involved .025 .207 .033 .719 Supportive .545 .176 -.007 .220 Comfortable .104 .690 .023 .231 Available .007 .810 .027 .036 Understanding .107 -.159 .072 .694 ”Expert” .268 .102 -.716 .184 Generated discussion .332 .067 .710 .123 Got along .226 .205 .664 .208 a o o a o n c Var1max rotation with Kaiser normalization was used. 50 Table 9 Steps in Development of Supervisor Evaluation Original Measure 15 items about supervisor, rated on a 5-point scale by volunteers. Steps in Development 1. Standardization of all items at two time periods. 2. Calculation of fifteen change scores. 3. Factor analysis of change-scores. Within-Measure Factors 1. Knowledgeable/Concerned. 2. Available/ Heldul 3. Quality of Discussion. 4. Involved. Final Burnout Score 1. Average of four within-measure factor scores. 51 change over time in factor scores prior to standardization. Steps in development of the Course Evaluation measure are listed in Table 10. An examination of Appendices A through E shows that factors from these five measures seem to tap most of the areas of volunteers' feelings and behavior that were suggested by the model to be important in volunteer burnout. The Semantic Differential factors measure attitudes about self and client. The volunteer's level of activity and in— volvement is tapped by one of the Peer Rating factors (Group Involvement) and one of the Peer Ranking factors (Talkative in Class). Empathy is measured by one of the Peer Rating factors (Empathy/Effectiveness) and one of the Peer Ranking factors (Empathy). Satisfaction with various aspects of the volunteer experience is measured by the Supervisor and Course Evaluation factors. Predicting Burnout Three methods were used to measure volunteer—related variables predicting burnout. Each measure was administered before training. Demographic Questionnaire. The Demographic Question— naire asked for information about work history, interests, etc. The item expected to predict burnout was Experience, defined as total number of hours, as worker or volunteer, during high school or college, in any of the following settings: education, health, juvenile justice, mental health, geriatrics, recreation, etc. 52 Table 10 Steps in Development of Course Evaluation Original Measure 36 items about the course, rated on a 5-point scale by volunteers. Steps in Development 1. 2. 3. 4. Factor analysis of original items, producing a 12-factor solution. Four factors chosen for further periods. Standardization of factor scores at two periods. Calculation of four change—scores. Within-Measure Factors 1. 2. 3. 4. Satisfaction with class discussion. Satisfaction with course. Talkative outside group. Satisfaction with training. Final Burnout Score 1. Average of four within-measure factor scores. 53 The range of scores on this variable was from zero to almost 6000. There were 38 zeros, 45 scores between zero and 1000, six scores between 1000 and 2000, and two scores over 5000. Because of the greater likelihood that higher scores were less reliable, all scores over 1000 were set at 1000. Personality Research—Form. Jackson's PRF, Form A (Jackson, 1974) scale scores were also used. A list of scales that were used, along with defining trait adjectives, may be found in Appendix F. The fourteen PRF scale scores were factor analyzed using the method already described. The five—factor solution was chosen, based on rational and empirical criteria. Table 11 lists factor loadings for this solution. The PRF factors were labeled Impulsive/Playful, Affiliative/Dependent, Dominant/Aggressive, Hardworking and Nurturant. Behavior Measure. The third method was a Behavior Measure being developed at the Diversion Project. Trained observers rated each volunteer's interactions in role— playing situations with a pseudo—client youth, who presented the volunteer with a number of concerns and problems. Attri- butes on which volunteers were rated included optimism, mention of similarities between self and youth and resource- fulness. A complete list of Behavior Measure items indicative of overinvolvement, high expectations and other traits associated with stage one of burnout may be found in Appendix G. 54 Table 11 Factor Analysis of Jackson's PRF Eigenvalues and Proportion of Variance Factor Eigenvalue % of Variance Cummulative % 1 2.949 21.1 21.1 2 2.349 16.8 37.8 3 2.327 16.6 54.5 4 1.364 9.7 64.2 5 .962 6.9 71.1 6 .721 5.1 76.2 7 .643 4.6 80.8 8 .611 4.4 85.2 9 .492 3.5 88.7 10 .396 2.8 91.5 11 .379 2.7 94.2 12 .307 2.2 96.4 13 .289 2.1 98.5 14 .211 1.5 100.0 55 Table 11 (Cont.d.) . a Rotated Factor Matrix Factor 1 2 3 4 5 (Impulsive/ (Affiliative/ (Dominant/ Scale playful) Dependent) Aggressive) (Hardworking) (Nurturant) Achievement —.321 —.206 .237 .693 .022 Affiliation .165 .660 .184 .044 .499 Aggression .115 .154 .624 -.018 —.541 Autonomy .218 —.632 -.089 .239 -.406 Dominance -.O38 .022 .814 .182 .180 Endurance -.076 -.092 .004 .842 .175 Exhibitionism .385 .130 .712 .065 .310 Harm Avoidance —.535 .066 -.142 -.549 .307 Impulsivity .805 -.008 .209 -.100 -.O71 Nurturance .025 .018 .212 .125 .766 Order —.658 .419 .004 .300 .056 Play .759 .326 -.O42 -.046 .199 Social Recognition —.042 .785 .331 -.070 -.140 Understanding —.161 -.695 .237 .207 .143 a . . . . . . Var1max rotation with Kaiser normalization was used. 56 Only items which had numerical response categories (e.g. ”this never happened, this happened once, twice, etc.") were used. For each item, the number of times the response occurred (that is, for example, the number of times the volunteer discouraged the pseudo-client, or pointed out differences between self and pseudo—client) was calculated by averaging across raters within situation, and then summing across the four role-playing situations. Thus each volunteer started with thirteen scores, each of which was the total number of times during the four role playing situations that a certain type of behavior occurred. These scores were factor analyzed, using the method already described. The six factor solution was chosen, using rational and empirical criteria. Table 12 lists factor loadings for this solution. The factors were labeled Involved, Resourceful, Distant, Angry, Different and Histori- cal. Examination of Appendices F and C shows that factors from those measures used as predictors seem to tap areas of volunteer personality and behavior that were described in the model as traits that lead to volunteer burnout. Need or ability to become involved with colleagues and clients is tapped by one PRF factor (Affiliative/Dependent) and several Behavior Measure factors (Involved, Distant). Need to achieve is tapped by one PRF factor (Hardworking) and one Behavior Measure factor (Resourceful). Ability to be nurturant is tapped by several PRF factors (Nurturant, 57 Table 12 Factor Analysis of Behavior Measure Eigenvalues and Proportion of Variance Factor Eigenvalue % of Variance Cummulative % 1 2.507 17.9 17.9 2 1.848 13.2 31.1 3 1.525 10.9 42.0 4 1.392 9.9 52.0 5 1.190 8.5 60.4 6 1.019 7.3 67.7 7 .795 5.7 73.4 8 .732 5.2 78.6 9 .710 5.1 83.7 10 .572 4.1 87.8 11 .519 3.7 91.5 12 .471 3.4 94.9 13 .395 2.8 97.7 14 .324 2.3 100.0 58 Table 12 (Cont.d.) Rotated Factor Matrixa Factor 1 2 3 4 5 6 Item (Involved) (Resourceful) (Distant) (Angry) (Different) (Historical) Illogical .664 -.156 .310 .008 .123 -.181 Optimistic —.041 .531 .326 -.O37 -.54l -.O72 Pessimistic .123 .069 -.117 .813 .105 —.099 Similarities — now .726 .160 .210 -.025 .026 —.159 Dissimilarities - now .094 .088 .270 -.O68 .803 .023 Similarities - then .192 .012 .751 .183 .019 -.062 Dissimilarities - then .068 -.098 .734 .052 .145 —.010 Historical -.130 .023 —.051 .011 .080 .890 Judgemental -.040 .156 .302 .644 -.227 -.154 Resourceful .042 .859 .055 .010 -.053 .030 Specific Planning -.032 .779 —.341 .074 .172 -.026 Empathic .482 —.077 -.069 -.117 —.421 .503 Discounting .067 -.138 .224 .653 -.002 .243 Impatient .763 -.013 -.103 .269 .012 .202 a . . . . . . Var1max rotation with Kaiser normalization was used. 59 Dominant/Aggressive) and one Behavior Measure factor (Angry). Relating-Burnout to Other Personality Changes Changes in volunteers' locus of control associated with burnout was measured using Rotter's 23-item Locus of Control scale (Rotter, 1954), administered before training and at termination. Relating Burnout to Client Behavior In order to examine the relationship between volunteer burnout and client outcome, two measures of client behavior were used. Recidivism Measure. A measure of recidivism was calculated for each youth, based on the number of contacts with police and courts. This information was gathered for three time periods: during the year prior to participation in the project, during the year of the project, and during the year after termination from the project. Self—Reported Delinquency. A questionnaire in which the youth reported how many times in the last six weeks he or she had engaged in a variety of delinquent activities was used. This measure of self—reported delinquency was administered at time of selection for the project, three times during the academic year and one year after contact with the project ended. Items are listed in Appendix H. Predicting Burnout from Volunteer and Situational Variables The measures used in this pasrt of the study consisted of the volunteer variables discussed earlier, with two 60 additional situational variables. One was a dichotomized youth self—reported delinquency (SRD) score, from the pre time period. Youth SRD scores were dichotomized using the median score as cut-off. The other situational variable was intensity of training and supervision. This was also a dichotomized variable. High intensity training/supervision groups were in general smaller and met more often than low intensity groups. In high intensity groups, volunteers were trained in a specific mode of intervention (e.g. focus on behavioral contracting and work with family of youth; focus on advocacy for the youth with relevant groups — family, school, courts, etc.; focus on one—to-one relationship with youth and therapeutic interaction). In low intensity groups, volunteers were expected to use ”natural helping skills.” A list of all the measures that were used, showing the purpose of each measure and the times at which each was administered, is provided in Table 13. 6]. Table 13 Measures Used in this Study 1. For measuring burnout of volunteers: Semantic Differential Peer Ratings (Peers evaluate student) Peer Rankings (Peers evaluate student) Supervisor Evaluation (Student evaluates supervisor) Course Evaluation (Student evaluates course) For predicting burnout (volunteer variables): Jackson's PRF Demographic Questionnaire Behavioral Measure For measuring locus of control of volunteers: Rotter's LOC questionnaire For measuring behavior of youth: Self-Report Delinquency Questionnaire Data gathered at police stations and courts For predicting burnout (situational variables) Intensity of Training/ Supervision (dichotomized) Youth Self-Reported Delinquency - pre (dichotomized) (pre-post difference) (pre-data only) (pre—post difference) (pre—, during, and post) (pre-data only) CHAPTER THREE: RESULTS Measuring Burnout Initially it was not clear how to use the 18 within- measure factor scores derived from the five ”burnout measures.” The burnout literature suggested that there might be five or six dimensions on which burnout in human services workers could be measured (e.g. involvement, hostility, flexibility, expectations). The questions to be answered in this part of the study were: a) how many dimensions underlie the measures of burnout in this sample of human services volunteers?; b) what are the underlying dimensions: are they similar to those described in the literature on human services workers?: c) how many "burnout scores” should be calculated for each volunteer and how should each be calculated? A superordinate factor analysis was performed, using the 18 within-measure factors as "items.” (These within— measure factors will subsequently be referred to as ”sub— factors.”) A principle-components analysis with varimax rotation was used. Kaiser's criterion and the scree method suggested something between four and seven factors. The seven-factor solution seemed best on rational grounds. The most striking thing about this analysis was that in all factor solutions, variance due to method of measure- ment was much more important than variance due to subfactor content in determining the factor loadings. In all of the 62 63 factor solutions using four or more factors, the first four factors were labelled Peer Ratings, Peer Rankings, Super- visor Evaluation and Course Evaluation. Subfactors that seemed to have similar content did not load together when they came from different measures. For example, Ratings: Effectiveness/Empathy loaded with other Ratings subfactors and Rankings: Empathy loaded with other Rankings subfactors, rather than loading together. Another consistency across all factor solutions was that the two subfactors from the Semantic Differential never loaded together. The loadings for these two subfactors fluctuated as the number of factors in the solution was increased. The two Semantic Differential subfactors were also the first subfactors to load individually on independent factors as the number of factors was increased. Factor loadings for the seven—factor solution are listed in Table 14. Factors one through four were labelled Ratings, Rankings, Supervisor Evaluation and Course Evalua— tion. Factor five, consisting mainly of two subfactors from the Ranking measure and one from the Supervisor Evaluation measure, seemed to be a combination of subfactors that did not fit elsewhere. Factors six and seven were labelled Self-evaluation and Youth—evaluation. The single subfactor from the Course Evaluation measure loading on factor six was not stable across factor solutions, rather this sub- factor separated into its own factor in the eight—factor solution. Factor five was also unstable across solutions. 64 Table 14 Superordinate Factor Analysis Eigenvalues and Proportion of Variance Factor Eigenvalue % of Variance Cummulative % 1 3.656 20.3 20.3 2 1.918 10.7 31.0 3 1.499 8.3 39.3 4 1.457 8.1 47.4 5 1.262 7.0 54.4 6 1.193 6.6 61.0 7 1.047 5.8 66.9 8 .871 4.8 71.7 9 .800 4.4 76.1 10 .726 4.0 80.2 11 .675 3.8 83.9 12 .641 3.6 87.5 13 .571 3.2 90.7 14 .506 2.8 93.5 15 .388 2.2 95.6 16 .322 1.8 97.4 17 .246 1.4 98.8 18 .221 1.2 100.0 65 Table 14 (Cont.d.) . a Rotated Factor Matrix Factor 1 2 3 4 5 6 7 (Supervisor (Course Sub Factor (Ratings) (Rankings) Evaluation) Evaluation) (Self) (Youth) Self—evaluation .058 .028 .205 —.092 -.125 .761 .024 Youth-evaluation .130 .103 -.O39 .028 -.030 .006 .932 Rating: Perseverance .823 .068 .069 .061 .283 .003 .041 Rating: Involvement .793 .191 .063 .082 .009 .025 .109 Rating: Effective- ness/Empathy .855 .053 .115 .092 -.009 -.058 .017 Rank: Talkative .081 .743 —.D73 .104 .263 -.213 -.080 Rank: Gets Along .249 .314 -.069 .176 .578 .148 -.D73 Rank: Motivation .113 .748 .173 .075 .232 .004 .068 Rank: Empathy .150 .680 .215 -.l30 —.018 .191 .241 Rank: Perseverance .193 .252 .334 —.D39 .669 .006 -.123 SE: Knowledgable/ Concerned .097 .085 .714 .191 .013 .037 -.099 SE: Available/ Helpful -.071 -.432 .590 .038 .391 -.096 .251 SE: Discussion .148 .243 .720 .016 -.127 .064 .034 SE: Involved .031 —.096 .331 .204 —.563 .153 —.132 CE: Satisfaction .192 .109 .189 .534 .059 -.398 .221 CE: Talk outside —.112 -.028 —.198 .379 .322 .635 .012 CE: Discussion —.098 —.O36 .055 .800 .077 .047 .004 CE: Training .194 .063 .109 .724 -.162 .049 -.052 a . . . . . . Var1max rotation with Kaiser normalization was used. 66 The other six factors, however, did seem stable across factor solutions. On the basis of this analysis it was decided that the 18 subfactors had more than one underlying dimension. Thus it did not make sense to use only a single burnout score for each volunteer. A sum of all 18 subfactor scores would probably be meaningless. Clearly, the methods of measurement were acting in— dependently of each other and thus measuring separate dimensions. With the exception of the two Semantic Differen— tial subfactors, subfactors within measures generally loaded together. Subfactors that didn't consistently load with the others from within their original measure did not load on any factor consistently across factor solutions. The two Semantic Differential subfactors each seemed to form' a separate factor. Given these results it was decided that, except for the two Semantic Differential subfactors, the other 16 sub— factor scores would be combined within original measures, to produce four ”burnout scores.” The two Semantic Differential subfactors would be used individually, as two additional "burnout scores.” Six ”burnout scores” would thus be assigned to each volunteer. The six scores were labelled Peer Ratings, Peer Rankings, Supervisor Evaluation, Course Evaluation, Self—evaluation and Youth—evaluation. The first four of these ”burnout scores" were calculated by averaging the subfactor scores within each of these four 67 measures. For example, the Peer Ratings burnout score was calculated by averaging the scores of the three Peer Ratings subfactors. The last two burnout scores were simply the original subfactor scores on Self— and Youth—Evaluation. The method of calculating burnout scores — the first step in burnout—measure development — is summarized in Section D, of Tables 3, 5, 7, 9 and 10. These six burnout scores were all that remained after extensive data reduction, first within the five original measures and then across measures. For the most part the six scores represent separate methods of measurement. They will subsequently be referred to as ”burnout scores,” ”burn— out measures” and ”burnout variables” interchangeably. It should be noted that these six dimensions of burnout are quite different from those described in the literature on burnout in human services workers. Once the burnout measures had been developed, analyses could be performed to examine relationships between volunteer burnout and other variables. Predicting Burnout From Volunteer-Variables The questions being examined in this part were a) whether burnout could be predicted on the basis of the volunteers' personality alone, and b) whether the burnout model's hypothesis of an initial predisposing state was correct. Six separate discriminant analyses were run, one on 68 each of the six burnout (dependent) variables. Volunteer personality, behavior and experience variables were used as predictors. Volunteer personality variables were the five PRF factor scores, measured at the beginning of the project. Volunteer behavior variables were the six Behavior Measure factor scores, also measured at the beginning of the project. Volunteer experience was the number of hours worked or volunteered in human services settings. As already mentioned, each of the six burnout variables were based on changes in the volunteers between the start and end of the project. Because the six burnout scores (which would be used as six dependent variables in six separate discriminant analyses) were continuous, it was necessary to convert them into nominal variables for this analysis. It was decided that the dependent variables would be trichotomized. This would allow for a slightly more fine-grained analysis of the predictor—burnout relationship than would dichotomized dependent variables. It would allow for the possibility of a two—dimensional arrangement of groups (whereas dichotomized dependent variables would only allow for a one— dimensional arrangement of groups). Trichotomizing the dependent variables would also allow us to notice curvilinear relationships between predictors and burnout variables which would be missed if the dependent variables were dichotomized. In order to provide for a fairly even distribution across the groups of the dependent variable, cut—off scores 69 for categorization were set at one—half standard deviation above and below the mean for each of the six burnout scores. Thus each burnout score was partitioned into three roughly equal—sized groups. The three groups were: change in the direction of burnout, change in the direction opposite to burnout, and little change. For example, on the Self- evaluation burnout score, one group of volunteers demon— strated more negative attitudes and expectations of theme selves over time, one group improved in these attitudes over time, and one group demonstrated little change over time. Clearly, the method of categorizing the continuous burnout scores was chosen arbitrarily, based on the fineness of grain desired in the analysis, on the number of subjects, the number of predictors, and the number of factors believed to underlie the relationship. In the Discussion section, further comments will be made about the use of a continuous dependent variable in discriminant analysis. Discriminant functions of volunteer personality, behavior and experience (independent) variables were able to discriminate between groups for three of the six burnout variables. Discriminant analyses for each of the burnout (dependent) variables will be discussed separately. Self—evaluation Using the Self-evaluation measure (SELF) as the depen- dent variable, a function composed of the variables Distant, Resourceful and Historical accounted for 18.1 percent of 70 the variance in the dependent variable. The first function accounted for virtually all the variance in group centroids. Table 15 summarizes the data on the discriminant functions and shows mean predictor scores within groups for this dependent variable. Examination of within—group mean predictor scores shows that volunteers who improve in self—evaluation are those who initially are least resourceful, most distant and most historical. Those who either burn out or show little change on this measure are initially more resourceful, less distant and less historical. Youth—evaluation On the Youth—evaluation measure (YOUTH), two functions are necessary to explain the variance between groups. The first function accounts for 7.7 percent of the group variance, while the second function accounts for 7.4 percent. The two functions each account for about half the group centroid variance. The predictors Dominant and Hardworking are most important in function one, while Nurturant is most important in function two. Table 16 summarizes the discri- minant function data and shows mean predictor scores within groups for the Youth-evaluation dependent variable. Volunteers who change in a positive direction in youth evaluation are initially most dominant and least hardworking. Volunteers who are initially most nurturant tend to change in either a positive or negative direction on youth evalua- tion. Those who are initially least nurturant change less 71 Table 15 Discriminant Analysis Results: Predicting Self—evaluation Burnout from Volunteer Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 l. Distant .905* -.796 .413 2. Resourceful .839** .626 .727 3. Historical .818** -.377 .602 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .222 99.78 .426 17.47 (6) ** 2. .000 .22 .022 0.04 (2) Mean Predictor Scores Predictors Category Distant Resourceful Historical 1. .85 2.39 .25 2. .36 3.55 .14 3. .45 3.26 .15 Total mean .53 3.12 .18 SD .69 1.86 .45 *p< .05 **p<:.01 72 Table 16 Discriminant Analysis Results: Predicting Youth-Evaluation Burnout From Volunteer Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Nurturant .928* .195 —1.015 2. Dominant .883* .865 .344 3. Hardworking .855* -.665 - .032 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .084 51.11 .278 13.68 (6)* 2. .080 48.89 .272 6.69 (2)* Mean Predictor Scores Predictors Category NurtUrant Dominant Hardworking 1. 16.34 10.59 13.62 2. 15.46 9.38 14.22 3. 16.84 9.20 14.44 Total mean 16.12 9.71 14.09 SD 2.17 2.79 2.55 * p< .05 ** p( .01 73 in their evaluation of youth. Peer Ratings Using the Peer Rating burnout measure, none of the predictors was able to distinguish the burnout groups at a significant level. The function of predictors was also unable to discriminate between groups. Table 17 shows the results of discriminant analysis on this dependent variable. Peer Rankings Again none of the predictors could discriminate between groups, nor could the function. Table 18 summarizes the results of this analysis. Supervisor Evaluation One function significantly discriminated between criterion groups. A function consisting of the predictors Different, Dominant, Historical and Involved accounted for 14.5 percent of the variance in the group variable. The first function accounted for 82.4 percent of group centroid variance. Table 19 presents the data from the discriminant analysis on this dependent variable. Volunteers who later burnout out on supervisor evalua— tions initially scored high on Dominant and Different. Volunteers who improved their evaluations of their super- visors over time initially scored high on Involved. Volunteers who didn't change their supervisor evaluations a great deal initially scored high on Historical and low on Dominant. 74 Table 17 Discriminant Analysis Results: Predicting Peer Ratings from Volunteer Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 l. Affiliative .935 .595 —.823 2. Different .906 —.469 -.560 3. Experience .877 -.617 -.243 4. Nurturant .846 -.567 -.250 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.F.) 1. .125 71.04 .333 14.46 (8) 2. .051 28.96 .220 4.29 (3) Mean Predictor Scores Predictors Category Affiliative Different Experiencea Nurturant 1. .81 .11 1.78 15.81 2. .81 .36 1.89 16.48 3. 2.03 .18 1.78 16.03 Total mean 1.25 .22 1.81 16.12 SD 2.32 .57 .31 2.17 a . . . . Recoded so that range is from 1 to 2, and direction is reversed. *p<.05 **p(f.01 75 Table 18 Discriminant Analysis Results: Predicting Peer Rankings from Volunteer Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Resourceful .947 —.677 .278 2. Affiliative .901 .577 —.497 3. Impulsive .874 .490 .638 4. Dominant .844 —.522 -.166 5. Involved .822 —.516 -.152 6. Distant .795 .469 .199 7. Nurturant .774 .273 .584 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .237 84.69 .437 21.61 (14) 2. .043 15.31 .203 3.56 (6) Mean Predictor Scores Predictors Category Resourceful Affiliative Inpulsive Dominant Involved Distant Nurturant 1. 2.54 1.86 4.21. 9.39 .42 .59 15.96 2. 3.56 .86 3.77 10.09 .67 .47 15.97 3. 3.15 1.14 4.89 9.57 .45 .56 16.46 Toml Mean 3.12 1.25 4.25 9.71 .52 .53 16.12 SD 1.86 2.32 2.90 2.79 .68 .69 2.17 *p<.05 **p<.01 76 Table 19 Discriminant Analysis Results: Predicting Supervisor Evaluation from Volunteer Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Different .936 —.771 .270 2. Dominant .875* -.659 —.067 3. Involved .847* .094 -.929 4. Historical .825 .404 .279 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .170 82.41 .381 16.62 (8)* 2. .036 17.59 .187 3.07 (3) Mean Predictor Scores Predictors Category Different Dominant Involved Historical 1. .24 9.96 .70 .11 2. .10 9.16 .46 .26 3. .48 10.72 .42 .09 Total mean .22 9.71 .52 .18 SD .57 2.79 .68 .45 *p< .05 “p(.01 77 Course Evaluation None of the predictors could discriminate between criterion groups, nor could the function. Table 20 summarizes the results of this discriminant analysis. On the basis of these results at appeared that burnout could be predicted from volunteer personality, behavior and experience variables alone, using specific burnout measures. The hypotheses of the burnout model about personality traits predisposing volunteers to burnout were validated with Self— and Youth—evaluation burnout scores. The Relationship Between Burnout and-Locus of Control One hypothesis about this relationship stated that burnout was related to LOC—change in such a way that volunteers who burned out would shift in the External direction on locus of control. This will be referred to as the "directional hypothesis.” The other hypothesis stated that burnout was related to change in locus of control such that the more the volunteer burned out, the larger the absolute value of his or her LOC-change would be. This will be called the ”non—directional hypothesis.” Correlations were calculated between LOC-change in the External direction and each of the six burnout scores. Another six correlations were calculated between the absolute value of LOC-change and each of the burnout scores. Correlations between the six burnout scores and locus of control change scores are listed in Table 21. Course 78 Table 20 Discriminant Analysis Results: Predicting Course Evaluation from Volunteer Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Angry .948 —.780 .017 2. Resourceful .918 .564 .037 3. Experience .894 .204 .648 4. Involved .865 .135 -.753 5. Different .839 .010 .604 6. Historical .818 .462 -.193 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .119 56.32 .326 17.20 (12) 2. .092 43.68 .291 7.56 (5) Mean Predictor Scores Predictors Category Angry Resourceful Experiencea Involved Different Historical l. 1.85 3.01 1.88 .37 .33 .12 2. 1.97 2.88 1.76 .62 .15 .15 3. 1.24 3.54 1.82 .54 .21 .28 Total mean 1.71 3.12 1.81 .52 .22 .18 SD 1.40 1.86 .31 .68 .57 .45 a . . . . Recorded so that range is from 1 to 2 and direction is reversed. *p< .05 **p< .01 79 Table 20 Correlation Between Burnout Variables and LOC Change Scores Burnout Measures Type of LOC Change Self Youth Ratings Rankings Supervisor Course A. Directionala —.05 —.10 —.05 -.08 —.06 .21* B. Non-directionalb -.19* -.18* .16 .10 .10 .01 Note: For all correlations, g = 91. a . . . . Correlation between burnout score and LOC change in the Internal direction. b . Correlation between burnout and absolute value of the LOC change. * p<.05 80 Evaluation was the only burnout measure that was correlated to a significant degree with a directional shift in locus of control. The fact that the relationship was positive indicates that volunteers whose Course Evaluation score indicated burnout also changed in the Internal direction on locus of control. Volunteers who improved in their evaluations of the course also changed in an External direction on locus of control. This result is contrary to what was predicted under the directional hypothesis. Table 21 shows that Self-evaluation and Youth-evaluation are the only burnout measures related to the absolute value of change in locus of control. Contrary to the prediction of the non—directional hypothesis they are negatively re— lated. This indicates that volunteers who evaluate them— selves and clients more negatively over time (i.e. those who burn out on Self— and Youth-evaluation) show less change in locus of control, while volunteers who evaluate them- selves and clients more positively over time show more change in locus of control. Neither of the original hypotheses seem to be supported by these results, thus the relationship between burnout and locus of control remains unclear. The Relationship Between Burnout-and Client Behavior The effects of the client on the volunteer and the effects of the volunteer on the client were examined. More specifically, the relationships between volunteer burnout 81 and client delinquency, before, during and after the project were studied. Discriminant analysis was used, with the six volunteer burnout scores as independent variables and the client delinquency scores as dependent variables. Two measures of client delinquency were used. The ”recidivism” measure of youth behavior was based on the number of court appearances and contacts with police either during the year prior to selection for the project, the year of the project, or the year after the project. The Self— Reported Delinquency (SRD) measure of youth behavior was based on the number of delinquent acts reported by the youth to have occurred in the past six weeks. Again, the SRD questionnaire was administered at selection, at several times during the project, and at one—year after termination from the project. Client Behavior Prior to Selection It had been assumed that ”tougher” or "more delinquent" youth would create more stress for the volunteer. The hypo- thesis was that these youth would be more likely to have volunteers who burn out than ”less delinquent” youth. Recidivism dependent variable. The recidivism measure showed a distribution too skewed for the measure to be of use in this part of the analysis. Of the 87 youth for whom archival outcome data existed, only seven fell into the less severe category, meaning they were not brought to court or were brought to court only for status offenses in the year prior to selection. Thus, no analysis of this data was done. 82 SRD dependent variable. SRD data for the six weeks prior to selection were analyzed using discriminant analysis with SRD as the dependent variable and the six burnout scores as predictors. SRD was trichotomized. Group one was composed of youth who reported relatively few delinquent acts in the six weeks prior to the project. Group two con- sisted of youth who reported a moderate number of delinquent acts. Group three consisted of youth who reported many delinquent acts. Results of this analysis are listed in Table 22. A strong relationship was found between the number of delinquent acts the youth reported (at selection) and volun- teer burnout. Twenty percent of the variance in the group variable was accounted for by a function composed of Course Evaluation and Supervisor Evaluation. The first function was able to account for almost all the group centroid variance. Volunteers working with youth who reported the greatest number of delinquent acts at selection for the project tended to burn out on both Supervisor and Course Evaluation measures. Volunteers working with youth who reported fewest delinquent acts improved in their evaluations of supervisor and course. These results support the hypothesis that working with ”tougher” clients leads to burnout. Burnout in this case means increasing dissatisfaction with various aspects of the volunteer experience as measured by Supervisor and Course 83 Table 22 Discriminant Analysis Results: Predicting SRD-pre from Burnout Scores Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Course evaluation .864** .740 -.678 2. Supervisor Evaluation .795** .617 .791 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .250 97.37 .447 19.18 (4)** 2. .007 2.63 .082 0.56 (1) Mean Predictor Scores Predictors Category Course Evaluation Supervisor Evaluation 1. —.24 -.22 2. -.14 —.02 3. .36 .23 Total mean -.01 .00 SD .70 .55 *p( .05 **p< .01 84 Evaluation scores. Client Behavior During the Project. It was hypothesized that working with ”more delinquent” youth makes volunteer burnout more likely and also that volunteer burnout makes client failure more likely. It was expected that both of these effects would be seen in a strong relationship between volunteer burnout and client delinquency during the project. Recidivism dependent variable. Discriminant analysis was performed with the six burnout scores as predictors and the recidivism variable as the group (dependent) variable. Once again, the recidivism variable had a skewed distribu— tion. Sixty—five of 87 youth had no police or court con— tacts during the project. The other 22 had at least one contact with police or courts. None of the burnout variables qualified for the discriminant analysis on this variable. Self-Reported Delinquency dependent variable. Discrimi— nant anabmfis was performed with the six burnout scores as predictors and the SRD variable (trichotomized) as the group (dependent) variable. A fairly strong relationship was found between the number of delinquent acts the youth re— ported during the nine months of the project and volunteer burnout. A function composed of Supervisor Evaluation, Course Evaluation, Self-evaluation and Youth-evaluation accounted for 13.9 percent of the variance in the group variable. The first function accounted for 81.7 percent of the group 85 centroid variance. Table 23 presents the results of this analysis. Volunteers whose youth reported the greatest number of delinquent acts during the project burned out on Super- visor, Course and Self-evaluations. Those whose youth reported fewer delinquent acts during the project improved on Supervisor, Course and Self-evaluations. The effects of Youth-evaluation was reversed. Volunteers working with youth who reported fewest delinquent acts during the project burned out most on Youth-evaluation. Those working with youth who reported an average or above—average number of delinquent acts tended to improve in their attitudes about youth. Again, there is evidence of a relationship between client behavior and volunteer burnout. In general burnout is associated with more delinquent behavior on the part of the client. However, when burnout is thought of as an in— crease in negative attitudes about clients, the direction of the relationship between volunteer burnout and client behavior is reversed. Burnout appears to be related to what the youth say they do (regarding illegal activities), but not to their actual number of police and court contacts. Client Behavior After the Project It was expected that the impact of volunteer burnout on the clients would be to make delinquency more likely. Recidivism dependent variable. thirty—four of the 87 youth for whom data existed had at least one contact with 86 Table 23 Discriminant Analysis Results: Predicting SRD-during from Burnout Scores Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Supervisor Evaluation .897** -.733 —.l40 2. Self-evaluation .869* -.356 -.584 3. Youth—evaluation .842* .325 —.715 4. Course Evaluation .819* —.376 .418 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .175 81.71 .386 16.44 (8)* 2. .039 18.29 .194 3.16 (3) Mean Predictor Scores Predictors Category Supervisor Evaluation Self-evaluation Youth—evaluation Course Evaluation 1. —.21 -.08 .31 .23 2. —.06 -.16 -.19 .01 3. .24 .47 —.10 .13 Total mean .00 .05 -.04 —.01 SD .55 1.31 1.23 .70 *p( .05 **p< .01 87 police or courts in the first year after participating in the project. Fifty—three had no police or court contacts. Discriminant analysis was unable to produce functions of burnout scores that could discriminate between these two groups. Table 24 shows the results of this analysis. Self-Reported Delinquency dependent variable. No rela— tionship was found between volunteer burnout and the number of delinquent acts the youth reported after the project. Discriminant functions of burnout scores were unable to differentiate between groups on SRD post—intervention. The results of this analysis are listed in Table 25. These results lead to the rejection of the hypothesis that volunteer burnout increases the likelihood of delinquent behavior after termination from the project. Change in Client Behavior In a further attempt to understand the impact of volunteer burnout on youth outcome, discriminant analysis was used to examine the relationship between the six burnout measures and changes in SRD from pre-intervention to post— intervention and from during—intervention to post—inter— vention. It was expected that volunteers who burned out would have youth who reported continued delinquent activity, while volunteers who improved on the "burnout scores” would have youth who reported a decrease in delinquent activity. Pre—Post SRD difference as dependent variable. When difference scores between pre—SRD and post—SRD were tri- chotomized (i.e. into groups of decrease in reported 88 Table 24 Discriminant Analysis Results: Predicting Recidivism—post from Burnout Scores Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Ratings .984 1.00 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .016 100.00 .126 1.36 (1) Mean Predictor Scores Predictors Category Ratings 1. —.07 2. .12 Total mean .00 SD .74 *p( .05 **p< .01 89 Table 25 Discriminant Analysis Results: Predicting SRD—Post from Burnout Scores Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 1. Self—evaluation .945 .743 .670 2. Rankings .902 -.661 .751 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) 1. .089 83.84 .287 7.45 (4) 2. .017 16.16 .130 1.24 (1) Mean Predictor Scores Predictors Category Self—evaluation Rankings 1. .17 -.02 2. .50 -.45 3. -.34 .01 Total mean .05 ' .00 SD 1.31 .72 *p< .05 **p<.01 90 delinquent activity, increase in reported delinquent acti- vity, and little change), the function of burnout scores did not discriminate between groups to a significant degree. Results of this analysis are presented in Table 26. During-Post SRD difference as dependent variable. When SRD change from during—intervention to post—intervention (trichotomized) was the group (dependent) variable, none of the predictors qualified for inclusion in a discriminant function. In this part of the study, results showed that while youth delinquency (self—reported) before and during the pro- ject were predictable from volunteer burnout scores, youth behavior after the project was not. While youth seemed to have impact on whether or not volunteers burned out, whether or not volunteers burned out did not seem to have any impact on how youth behaved or reported themselves to behave after the project. Furthermore, volunteer burnout was unrelated to changes in youth self-reports. Volunteer burnout was also unrelated to youth recidivism during or after the project. Predicting Burnout from Volunteer and Situational Variables This part of the study examined both the six burnout scores and the various predictors, in an attempt to examine the validity of the burnout scores and the usefulness of the predictors. Specifically, predictors related to the volunteer (i.e. personality, behavior and experience) were 91 Table 26 Discriminant Analysis Results: Predicting PRE-POST SRD Difference from Burnout Scores Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 l. Self-evaluation .947 .723 -.226 2. Rankings .896 .674 —.396 3. Supervisor Evaluation .865 .227 .975 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi-squared (d.f.) l. .119 78.35 .326 9.99 (6) 2. .033 21.65 .178 2.23 (2) Mean Predictor Scores Predictors Category Self—evaluation Rankings Supervisor Evaluation 1. -.06 —.16 .12 2. .39 .13 .06 3. -.36 —.22 -.16 Total mean .00 —.08 .01 SD 1.33 .71 .59 *p( .05 **p('.01 92 compared with predictors related to the situation (i.e. youdi SRD—pre and supervision/training intensity) to see which type of predictor was more strongly related to the dependent variables (i.e. volunteer burnout scores). Attention was focused on the Training/Supervision Intensity variable, in order to understand what effect, if any, it had on volunteer burnout. The relative validity of the burnout scores was judged by observing whether each score could be predicted, which variables predicted each measure, and in what direction the predictors operated. In examining the directions of effect of predictors, it was determined whether each burnout score was related to the predictors in the way described in the burnout model. Thus, the measures most compatible with the model (i.e. the best operational definitions of burnout) could be chosen from among the six burnout scores. By examining the relationships with predictors, hypotheses could be made about what constructs some of the less adequate burn— out scores may have been measuring. Six discriminant analyses were performed, each using one of the six burnout scores as the group variable. In all six cases these scores were trichotomized. Group one consisted of volunteers who changed in a direction opposite that described by the burnout model; group two consisted of volunteers who changed little on the measure; group three consisted of volunteers who changed in the direction 93 described in the burnout model. The predictors were volunteer personality (five PRF factors), behavior (six Behavior Measure factors) and experience (hours worked or volunteered in human services), supervision/training inten— sity (dichotomized), and youth SRD at selection for the pro- ject (also dichotomized). Four of the six discriminant analyses of volunteer and situational variables were successful in differentiating between criterion groups. On Supervisor, Course and Self- evaluation and on Peer Rankings, functions of predictors were able to differentiate between groups. On Youth— evaluation and Peer Ratings the functions were unable to differentiate between groups. The results of the six discriminant analyses are presented individually, for each burnout (dependent) variable. Self—evaluation On the Self-evaluation burnout variable, 21.7 percent of the variance in the group variable was accounted for by a function composed of the predictors Distant, Impulsive, Experience and Resourceful. The function explained 85.6 percent of the group centroid variance. The results of this analysis are presented in Table 27. Volunteers whose Self—evaluations moved in a positive direction over time were initially the most distant from youth, least impulsive and lease resourceful. Those who burned out on self-evaluations were most resourceful, less 94 Table 27 Discriminant Analysis Results: Predicting Self—evaluation from Volunteers and Situational Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 l. Distant .835** .808 -.162 2. Impulsive .802** —.303 -.550 3. Experience .773** -.227 .764 4. Resourceful .748** —.385 —.080 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .277 85.61 .466 20.43 (8)** 2. .047 14.39 .211 3.20 (3) Mean Predictor Scores Predictors Category Distant Impulsive Experience Resourceful 1. .98 3.00 174.0 2.58 2. .31 4.27 266.0 3.45 3. .50 4.81 116.2 3.38 Total mean .58 4.01 194.63 3.15 SD .71 2.91 321.4 1.94 *p< .05 **p<.01 95 distant from youth, more impulsive and least experienced initially. Those who didn't change their self—evaluation were initially the most experienced. Youth-evaluation Functions of predictor scores could not discriminate between groups on this dependent variable. Table 28 summarizes the results of this analysis. Peer Rating Functions of predictor scores could not discriminate between groups on this dependent variable. Table 29 summarizes the results of this analysis. Peer Ranking On the Peer Ranking burnout score, 17.9 percent of the group variance was explained by a function composed of the predictors Historical, Involved, Affiliative, Dominant and Impulsive. The first function explained 65.1 percent of group centroid variance. Table 30 summarizes the results of this analysis. Volunteers who improved in Peer Rankings were initially most affiliative, least dominant and least resourceful. Those who changed little on this measure over time were initially most involved with youth, most resourceful and least impulsive. Those who declined on the Peer Ranking measure were initially the most historical and impulsive. They were also relatively dominant and uninvolved with youth. Supervisor Evaluation On the Supervisor Evaluation group variable, 25 percent Predicting Youth-evaluation from Volunteer and Situational Variables 96 Table 28 Discriminant Analysis Results: Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Dominant .930 -.644 —.182 2. Historical .880 -.592 .589 3. Involved .840 -.571 .171 4. Nurturant .808 .071 —.826 5. Angry .784 -.468 -.285 6. Distant .759 .443 —.106 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .234 77.53 .436 19.21 (12) 2. .068 22.47 .252 4.57 (5) Mean Predictor Scores Predictors Category Dominant Historical Involved Nurturant Angry Distant 1. 11.00 .23 .70 16.52 2.19 .44 2. 9.43 .19 .53 15.47 1.70 .58 3. 9.46 .05 .43 16.21 1.74 .70 Total mean 9.88 .16 .55 16.00 1.85 .58 SD 2.65 .34 .71 2.30 1.45 .71 *p( .05 **p<'.01 97 Table 29 Discriminant Analysis Results: Predicting Peer Ratings from Volunteer and Situational Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Hardworking .944 .738 —.505 2. Affiliative .900 .488 .629 3. Dominant .858 -.733 .214 4. Different .826 -.154 .567 5. Experience .799 .484 .194 6. Impulsive .773 .609 .109 7. Angry .750 .434 .211 8. Resourceful .723 -.443 -.123 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .229 64.38 .431 22.53 (16) 2. .126 35.62 .335 8.15 (7) Mean Predictor Scores Predictors Category Hardworking Affiliative Dominant Different Experience Impulsive Angry Resourceful 1. 14.95 1.00 9.24 .14 261.81 3.90 1.96 2.86 2. 13.81 .89 10.26 .22 108.93 3.85 1.58 3.46 3. 13.52 2.00 10.00 .33 228.07 4.26 2.03 3.08 Total mean 14.03 1.32 9.88 .24 194.63 4.01 1.85 3.15 SD 2.51 2.21 2.65 .61 321.41 2.91 1.45 1.94 *p< .05 **p< .01 98 Table 30 Discriminant Analysis Results: Predicting Peer Rankings from Volunteer and Situational Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Historical .887* .455 .773 2. Involved .842* —.535 .060 3. Affiliative .814* .473 .587 4. Resourceful .786* -.476 .304 5. Dominant .759* -.429 .535 6. Impulsive .735* .436 —.l73 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .218 65.11 .423 21.40 (12)* 2. .117 34.89 .323 7.68 (5) Mean Predictor Scores Predictors Category Historical Involved Affiliative Resourceful Dominant Impulsive l. .10 .44 1.95 2.74 9.43 4.10 2. .07 .75 .94 3.47 10.09 3.59 3. .33 .36 1.27 3.09 10.00 4.55 Total mean .16 .55 1.32 3.15 9.88 4.01 SD .34 .71 2.21 1.94 2.65 2.91 *p< .05 **p<.01 99 of the group variance was explained by a function composed of the predictors Different, Hardworking, SRD, Involved, Historical and Nurturant. The first function accounted for 79.25 percent of group centroid variance. Table 31 summarizes the results of this analysis. Volunteers who burned out on this dependent variable were initially least hardworking, most nurturant, least involved, least historical and they pointed out differences between themselves and youth most often. They were most likely to have youth who reported more delinquency initially. Those who improved on Supervisor Evaluation were initially most involved with youth and most hardworking. Course Evaluation On the Course Evaluation group variable, 12.6 percent of the group variance was explained by a function composed of Angry, SRD, Intensity and Dominant. The first function explained 57.6 percent of group centroid variance. Table 32 summarizes the results of this analysis. On this variable, volunteers who burned out were initially least angry. They tended to be in high—intensity supervision/training groups and worked with youth who re— ported more delinquency. Volunteers whose course ratings increased also tended to be in high intensity supervision/ training groups, but worked with youth who reported less delinquency initially. Those who changed least on this measure were initially more angry and more dominant. They 100 Table 31 Discriminant Analysis Results: Predicting Supervisor Evaluation from Volunteer and Situational Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Different .933 -.802 —.313 2. Dominant .881 -.312 —.576 3. Hardworking .830* .531 -.447 4. SRD .786* -.659 .176 5. Involved .751* .422 -.519 6. Historical .713* .532 .183 7. Nurturant .690* -.333 .351 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .333 79.25 .500 25.59 (14)* 2. .087 20.75 .283 5.77 (6) Mean Predictor Scores Predictors Category Different Dominant Hardworking SRD Involved Historical Nurturant 1. - .27 10.21 14.67 1.42 .71 .13 15.71 2. .10 9.39 13.94 1.44 .48 .21 15.97 3. .51 10.53 13.20 1.67 .45 .08 16.53 Total mean .24 9.88 14.03 1.48 .55 .16 16.00 SD .61 2.65 2.51 .50 .71 .34 2.30 *p<.05 **p<’.01 101 Table 32 Discriminant Analysis Results: Predicting Course Evaluation from Volunteer and Situational Variables Summary of Stepwise Procedure Standardized function coefficients Step Variable Wilk's Lambda Function 1 Function 2 1. Angry .906* .930 -.469 2. SRD .849* -.586 -.282 3. Intensity .816* -.506 -.282 4. Dominant .791* —.049 —.572 Discriminant Functions Function Eigenvalue % of Variance Canonical r Chi—squared (d.f.) 1. .144 57.61 .355 16.56 (8)* 2. .106 42.39 .309 7.09 (3) Mean Predictor Scores Predictors Category Angry SRD Intensitya Dominant 1. 2.10 1.29 1.24 9.33 2. 2.19 1.57 1.47 10.43 3. 1.29 1.54 1.29 9.67 Total mean 1.85 1.48 1.35 9.88 SD 1.45 .50 .48 2.65 *p< .05 **p< .01 a . . . Direction is reversed. 102 tended to be in lowaintensity training/supervision groups and to work with youth who reported more delinquent acts. It seems clear from these results that both personality and situational variables are important in predicting burn- out. The burnout measures differ in how well they can be predicted from initial personality and situational data. Of the burnout scores that are predictable, some do seem to fit the burnout model more closely than others. For instance, on the Self and Course Evaluation measures, pre- dictors tend to operate in the direction described by the model. Specific relationships between predictors and burnout scores will be examined more closely, and conclusions about each of the operationalized burnout measures will be drawn, in the next section. Summary Superordinante factor analysis across the five original methods of measuring volunteer burnout showed that there was more than a single dimension underlying the within— measure factors. Method variance seemed to determine the factor structure. It was decided that six ”burnout scores” r would be calculated for each volunteer. Four of the original methods each produced a single burnout score. The fifth method produced two burnout scores. When burnout scores were trichotomized, group membership for three of the measures could be predicted using volunteer— related variables. For two of these burnout scores, Self— 103 and Youth—evaluation, the volunteer-related predictors operated in the direction described in the burnout model. Volunteers who burned out on Self-evaluation were initially resourceful and not very distant from youth. Those who burned out on Youth—evaluation were initially hardworking and nurturant. An examination of the relationships between burnout scores and changes in locus of control showed that one of the burnout scores (Course Evaluation) was related to directional change in locus of control. Two of the burnout scores (Self— and Youth—evaluation) were related to non- directional change in locus of control. However, both the directional and non-directional hypotheses about the rela— tionship between burnout and locus of control change were rejected, due to the directions of these relationships. The relationships between volunteer burnout and youth behavior before, during and after the project were examined. There was a strong relationship between youth self-reported delinquency prior to selection for the project and later volunteer burnout (as measured by Supervisor and Course Evaluations). The relationship between volunteer burnout and youth self—reported delinquency during the project was also significant. No relationship was found between volun- teer burnout and youth self—reports of delinquency after the project or between burnout and changes in youth self— reported delinquency. No relationships was found between 104 volunteer burnout and the number of contacts the youth had with police or courts either during or after the project. The relative predictive power of ”situational” and ”volunteer-related” predictors of burnout was examined. Neither type of variable was clearly more important as a predictor. Group membership in four of the (trichotomized) burnout scores could be predicted using situational and volunteer— related predictors. For two of these (Self and Course Evaluation), the predictors all operated in the direction described in the burnout model. CHAPTER FOUR: DISCUSSION Measuring Burnout The initial problem involved finding a way of combining within—measure factor scores from five different methods of measurement to produce a burnout measure or measures. Superordinate factor analysis showed that there was more than a single superordinate factor underlying the 18 within measure factors. Method variance predominated. The content of within—measure factors apparently had little impact on the superordinate factor loadings. It was decided that six burnout scores would be used, one for each of the first four methods of measurement and two for the Semantic Differential. The dimensions of volunteer burnout in this sample were quite different from the dimensions of burnout described in the literature on human services workers. This could be a result of real differences in the burnout process be— tween workers and volunteers, or it might be an artifact of the procedures used in studying volunteers. Method variance tended to predominate in the superordinate factor analysis (across burnout measures). In addition, the dimensions of burnout described in workers and volunteers were arrived at in very different ways. For workers, the burnout dimensions were based on clusters of descriptions from many studies. For volunteers, the dimensions were based on superordinate factor analysis of a set of measures from a single sample. More research is needed on both types 105 106 of burnout before firm conclusions about their similarities and differences can be reached. One finding was consistent with the literature on worker burnout. Burnout, whether in workers or volunteers, does not appear to be a: undimensional phenomenon. Changes may take place in a number of different spheres. For example, workers or volunteers may become more or less active and involved over time, or their attitudes about themselves and their clients might change. It is not yet completely clear which of these spheres are most important or why different people change in different spheres. It is clear, however, that using a single burnout measure (or one that taps only one of the spheres) is a research strategy likely to produce oversimplified conclusions. Six burnout measures were used in this study. They measured changes in attitudes about self and client, in satisfaction with various aspects of the volunteer ex— perience, and in peer evaluation of the volunteer. Predicting Burnout from Volunteer Personality, Behavior and Experience Evidence was found that initial volunteer personality traits did predict burnout. It was also found that the specific personality traits described as ”stage one” of the burnout predicted burnout. However, the first finding was true only when three of the six burnout measures were used, while the second was true using two of the burnout measures. 107 On the basis of the burnout model presented earlier, it would be expected that volunteers who burn out would initially have high scores on the Affiliative, Hardworking and Nurturant PRF dimensions. They should initially be more involved with youth, more resourceful (that is, making more plans and attempting to implement them), less distant from the youth and less angry. They should have less experience than volunteers who do not burn out. Table 33 summarizes the directions of the effects of these eight predictor vari— ables on the three successfully predicted burnout scores. The results will be discussed individually for each burnout score. Self-evaluation This was one of the burnout measures successfully pre— dicted by initial volunteer variables. Volunteers who were initially more resourceful and less distant from youth tended to burnout on self—evaluation. Both predictors thus operated in the directions implied by the burnout model. Youth-evaluation This burnout measure could also be predicted from volunteer variables alone. Volunteers who were initially more hardworking and nurturant tended to burn out on Youth- evaluation. Again, the predictors operated in the direction implied by the burnout model. Peer Ratings This burnout score could not be predicted from volunteer variables alone. 108 Table 33 Directions of Effects of Volunteer-Related Predictors on Burnout Scores Burnout Variable Expected Predictor Direction Self Youth Supervisor Affiliative high Hardworking high high Nurturant high high* Involved high low Resourceful high high* Distant low low* Angry low Experience low Historical ? low low Dominant ? low high Different ? high Note: Direction in table is for volunteers who burn out, relative to others. For example, ”High" indicates that volunteers who burn out are expected to score, or actually score, higher than other volunteers on that predictor. * p<'.05 for univariate f. 109 Peer Rankings This burnout score could not be predicted from volun— teer variables alone. SupervisoreEvaluation Although this burnout score was predicted from volun- teer variables, the predictors did not operate in the direction implied by the burnout model. Rather than being more involved, volunteers who burned out on Supervisor Evaluation were initially lggs involved with youth than other volunteers. Volunteers who were initially more involved with youth tended to improve their Supervisor Evaluation scores. It may be that supervisors were concerned, available and involved with their volunteers to the extent that each volunteer was involved in the work with adolescents. This would explain why volunteers who were less involved with their clients found that their supervisors became less in- volved with them. This may mean that when the Supervisor Evaluation score indicates volunteer burnout there is actually a decrease in interest and involvement by the supervisor. Originally it was thought that Supervisor Evaluation scores would de- cline along with volunteer evaluations of other aspects of the volunteer experience, as the volunteer became dis- illusioned and dissatisfied. The results here may mean that a) volunteer burnout is not necessarily related to the quality of interaction between supervisor and volunteer, 110 and b) the Supervisor Evaluation score is more sensitive to the quality of supervisor—volunteer interaction than to volunteer burnout. That would mean that this ”burnout score” is not really measuring burnout. The issue of what is being measured by the burnout scores is taken up in a later part of this chapter. Course Evaluation This burnout score could not be predicted from volunteer variables alone. Summary For three of the six burnout measures, volunteer variables alone were able to predict burnout. It can be concluded that whether or not burnout is predictable from volunteer variables alone depends on how burnout is measured. Using Self, Youth, and Supervisor Evaluation measures, such prediction is possible. Using two of the six burnout scores, Self and Youth- Evaluations, the predictors operated in the directions implied by the burnout model. It can be concluded that the part of the model describing the personality at risk for burnout holds up best when burnout is operationalized as a change toward more negative attitudes and expectations of oneself and one's client. 111 The Relationship-between Burnout and Locus of Control There were two competing hypotheses about what this relationship might be. The ”directional hypothesis” was that volunteers who burned out would move in the External direction on the locus of control measure. The ”non- directional hypothesis” was that to the extent volunteers burned out they would also change on the locus of control measure, in either direction. Only the Course Evaluation score was significantly related to directional change in locus of control. This relationship was such that volunteers who burned out on Course Evaluation changed toward the Internal direction on locus of control. This relationship was fairly weak: only four percent of locus of control change in the Internal direction was explained by concurrent burnout on Course Evaluation. These results refute the directional hypothesis. On the one burnout measure for which there was any relationship be- tween burnout and change in locus of control, the relationship was in a direction opposite to the hypothesized one. However, the non—directional hypothesis would explain these results. It was also found that two of the burnout measures, Self— and Youth—evaluation were related to absolute value of change in locus of control. They were related in such a way that volunteers who burned out on these measures showed the least change in locus of control. Volunteers who changed on these two burnout measures in a direction qxxmifletp Unfimmt 112 changed the most on locus of control. Again the relation— ships were fairly weak, with only four percent of absolute value of locus of control change explained by Self—evaluation burnout and another four percent explained by Youth- evaluation burnout. These results seem to contradict the non—directional hypothesis. The burnout scores that are related to absolute value of locus of control change, are related in the direc— tion opposite to the hypothesized one. This still might be explained, however, by a more general version of the non- directional hypothesis. Perhaps the burnout measures are related to the locus of control score in such a way that volunteers who change, either for better or worse, on the burnout measures, are more likely to change, in either direction, on the locus of control measure. This would be the relationship expected if both the burnout and locus of control-change scores measured related personality changes, but if the directions of the changes were dependent on some initial personality state. The relationship between locus of control and burnout remains unclear. Further analysis of this relationship should go beyond correlations of change—scores. It might prove more fruitful to compare initial and final locus of control scores for volunteers who burn out, volunteers who improve, and volunteers who do not change on each of the burnout measures. A two-way analysis of variance could be used for this purpose. If level of burnout and time were 113 the independent variables and locus of control score the dependent variable, the hypothesis would be that there is a significant interaction effect. Another avenue for future research would be an examina- tion of variability of locus of control scores. The hypo— thesis would be that for the group of volunteers who burn out, initial variability of locus of control scores would be higher than for volunteers who do not burn out, and variability would increase over time. Variability of locus of control scores for volunteers who ”improve" on the burnout measures would initially be lower and would decrease. It is also possible that the dimensions on which one burns out are to some extent determined by initial locus of control scores. It may be that volunteers who are initially more Internal would burn out on Self-evaluation, while those who are initially more External would burn out on a Course or Supervisor Evaluation. Although it is important to continue to try to under— stand the relationship between burnout and other personality changes, we may not get very strong results in this type of study as long as our subjects are volunteers. It may be that the type of burnout seen in volunteers is too mild for significant personality changes to be evident. The Relationship between Burnout and Client Behavior Client Behavior (pre-intervention) It was hypothesized that volunteers working with 114 ”tougher” clients would be more likely to burn out. How ”tough” the client might be was measured by initial Self- Reported Delinquency (SRD) score. Due to an overly skewed distribution, the archical delinquency measure was not used for the pre—intervention time period. It was assumed that clients reporting more delinquent acts prior to selection for the project would be more difficult to engage in a working relationship, slower to change and less likely to provide positive feedback to the volunteer, and thus that working with them would be more stressful for volunteers. There was a strong relationship between SRD (that is, the number of delinquent acts the youth said they had engaged in during the six weeks prior to project selection) and later volunteer burnout. Volunteers working with youth who reported more delinquency tended to burn out on Course and Supervisor Evaluation measures more than other volunteers. It is tempting to conclude from this that working with youth who report more delinquency causes burnout on Course and Supervisor Evaluations. However, because the experi- mental situation was not one in which all variables but youth—SRD were controlled, we cannot infer that the relationship between youth SRD (pre) and volunteer burnout is causal. From two of the volunteer's burnout scores we can tell approximately how many delinquent acts the youth initially reported. That in itself is a strong conclusion, and is consistent with the hypothesis that burnout is at least partially a result of working with ”tougher” clients. 115 Of course the SRD-score does not exactly tell us how ”tough" the client is. It only tells us how "tough” the client says he is. Thus how ”tough” he says he is (or she says she is) is the measure related to burnout. Unfortunate— ly, the measure of how ”tough” the youth actually behaved in the year prior to selection (the archival delinquency measure) could not be used in this part of the analysis. At the pre-intervention time period, the vast majority of youth fell into the ”more severe" category. Thus, discri— minant analysis on this variable was not attempted. Client Behavior (during—intervention) When SRD was measured during the six months that youth and volunteers worked together, the results were similar. Using volunteer burnout scores, it was possible to predict fairly accurately how many delinquent acts the youth were reporting. Course, Supervisor, Self and Youth—evaluations were the burnout scores that were most useful in making this prediction. The Supervisor, Course, and Self-evaluation burnout scores behaved as the model predicted. The Youth-evaluation variable had an effect in the direction opposite to what was predicted. Thus, volunteers working with youth who reported more delinquency during the intervention period burned out on Self, Supervisor and Course evaluations. However, they improved their evaluations of youth. Volunteers working with ”less delinquent” youth improved their Self, Supervisor 116 and Course evaluations. However, those working with the "less delinquent” youth burned out on Youth evaluation. A possible explanation for this unexpected finding is found in the Project's ideological orientation. An important part of the volunteer's training is the philosophy that the youth are not solely responsible for their problems. Instead, these problems are seen as a product of the social milieu in which the youth live. For many of the volunteers (i.e. those trained in family intervention and advocacy), the focus of the intervention is the social network surrounding the youth, rather than in the youth themselves. This ideology may lead volunteers who feel increasingly unsuccessful and incompetent to blame their ”failure” on everyone but their client. Further study of the relation— ship between burnout and treatment ideology might test this hypothesis. In looking at the relationship between volunteer burnout and the number of delinquent acts the youth reports during the intervention period, the combined influence of two effects may be seen. One is the effect of the youth on the volunteer, the other is the effect of the volunteer on the youth. Since the theory is that ”tough” clients produce burned out workers, while burned out workers produce client ' we could expect a fairly strong relationship ”failures,‘ between SRD-during and volunteer burnout. However, analyses indicated canonical correlations which were weaker than those 117 in the analyses examining the relationship between SRD—pre and burnout. Thus, when the effect of volunteers on youth is combined with the effect of youth on volunteers, the relationship is weaker than when the effect of youth on volunteers alone is examined. This may mean that the effect of volunteers on youth is relatively weak. In analyzing the relationship between volunteer burnout and client be- havior (post—intervention), this hypothesis was examined more directly. Client Behavior (post—intervention) It had been assumed that volunteers who burn out are less effective in their work with clients, so that their clients benefit less from working with them. Relationships between volunteer burnout scores and youth recidivism—post- intervention, SRD post-intervention and change in SRD from pre- to post-intervention were examined. It appeared that volunteer burnout was unrelated to youth recidivism (how many court or police contacts the youth had) and youth SRD (how many illegal activities the youth reported). Volunteer burnout was also unrelated to change in SRD from pre- to post-intervention or during- to post- intervention. Thus clients'SRD had more of an impact on whether or not volunteers burned out than volunteer burnout had on client SRD or change in SRD. In order to explain this finding, the structure of the Adolescent Diversion Project must be examined. Each volunteer worked with a single youth, and each youth had 118 a single volunteer. However, for a number of-reasons, that one-to-one relationship may have been more important to the volunteer than to the youth. Although the youth are still in their teens, many of them may also have relationships with parole officers or other court-related caseworkers. If youth do not have such relationships while participating in the Project, they may have had them in the past. Youth may also have had contact with employees of social service programs, school—based remedial programs, recreational programs, etc. All have had contact with the police. Thus youth may have had a good deal of experience with adults who were there to watch them, keep them in line, control them, help them if they behaved, teach them and punish them. For youth, one of the major features of being accepted into the Project is that they do not have to go to court, and thereby avoid further re- strictions on their freedom. For the youth, the investment in the one—to-one relationship with the volunteer may be reduced because a) it is misperceived as another relationship with a punitive, witholding and controlling other, b) it may occur con— currently with other such relationships, c) it is offered without the youth requesting it, and d) it is required if the youth is to gain a reward he or she does want (i.e. freedom from formal court processing). On the other hand, the volunteers are relatively young and inexperienced in intense one-to-one relationships 119 with a client. Initially, they must choose to take a year— long sequence of courses. Then they must go through a selection process involving attending meetings and taking a battery of tests. During this process, they know that they might eventually be rejected. For some, this is the first ”clinical” experience after years of psychology or criminal justice courses. In addition to the usual pressure for grades, career choices are probably being made by most of them. Unlike other courses where they can keep their successes and failures secret, here their work with the client is out in the open for all supervision group members to see. They receive training in: how their youth will act, why he or she will act this way and what they can do about changing their youth's behaviors. Remember that the youth does not receive formal training in how the volunteer will act and how the youth are supposed to respond except to the extent that this relationship is seen as another example of past relationships with adults. In short, the volunteers' college experience and their personalities (i.e. needs to achieve, nurture or be altru— istic) point them in the direction of investing a great deal in the one—to-one relationship with the youth. The youth's experience with society's institutions may lead them to avoid involvement in the relationship. Thus it may be expected that when directional effects between volunteers and youth can be measured, the impact of youth on volunteers will be stronger than the impact of volunteers on youth. 120 Predicting Burnout from Volunteer and Situational-Variables Situational and volunteer predictors were examined to see if either type was clearly more important in predicting volunteer burnout. The burnout measures were also examined to see which were the most valid measures of volunteer burn— out, as defined in the model. The ”less valid” measures were examined so that hypotheses might be made about what these scores actually measured. Youth-SRD—pre (the number of delinquent acts reported by the youth at the start of the project) was the only predictor that both a) was used in a function on more than one burnout variable, and b) always operated in the direction described in the model. Other predictors that always Operated in the direction described in the model were the volunteer—related variables: Nurturant, Distant, Angry and Experience. The volunteer—related predictors Affiliative, Hardworking and Involved did not operate in the direction described by the model. Table 34 summarizes the directions of effects of the predictors on the burnout scores. We can conclude that the number of delinquent acts reported by the youth is the best predictor of volunteer burnout, given these burnout measures. Intensity of training/supervision was used in a single discriminant function (the one predicting group membership in Course Evaluation). It was the third predictor entered in the stepwise procedure on that function. With all the 121 Table 34 Directions of Effects of Volunteer and Situational Predictors on Burnout Scores Burnout Variable Expected Predictor Direction Self Rankings Supervisor Course Affiliative high medium Hardworking high low Nurturant high high Involved high low low Resourceful high high medium Distant low low* Angry low 10w* Experience low low Historical ? high high Impulsive ? high high Dominant ? high high low Different ? high Intensity ? high SRD high high high Note: Direction in table is for volunteers who burn out, relative to those who don't. *p<.05 for univariate F. 122 evidence in the literature for the importance of situational variables in causing job dissatisfasction and turnover, it is surprising to find that the Intensity of training/super- vision had so little impact on burnout. This particular situational variable may be too diffuse or global to be a very useful predictor of volunteer burnout. From these results it does not appear that either volunteer—related or situational variables could be sacrificed without losing a good deal of predictive power. There is no clear evidence that either type of variable is a better predictor. Rather, it seems that both types of variable should be used together to obtain the best pre- dictions. The burnout measures will be examined individually. Self—evaluation Group membership on this burnout score was predicted successfully by a function composed of four volunteer-related predictors (see Table 34). Three of these four predictors (that is Resourceful, Distant and Experience) are clearly related to aspects of stage one of volunteer burnout, as described in the model. All three of these predictors operated in the direction implied by the model. Volunteers who burned out on Self—evaluation were initially very resourceful, not very distant from clients and had little experience. Youth—evaluation Although significant relationship had been found, in 123 an earlier part of the study, between this burnout score and volunteer-related predictors, no relationship was found when the earlier predictors were combined with situational predictors. This lack of statistical significance (in spite of an increase in the canonical correlation over the earlier one, when volunteer—related predictors alone were used) is probably a result of the loss of subjects (from 91 to 75) as situational predictors were added. Peer Rating Group membership on this variable could not be pre— dicted. With the absence of any clear relationship between Peer Rating score and the predictors, specific conclusions about this burnout variable cannot be made. All that can be said is that decline in Peer Rating score is not as good an operational definition of burnout as is decline on some of the other burnout measures. Peer-Ranking Six volunteer—related predictors were used in the discriminant function that successfully predicted group membership on this variable. Of these, three predictors measured personality traits associated with stage one of volunteer burnout as described by the model (i.e. Affili- ative, Involved and Resourceful). However, 9223 of these three predictors operated in the direction implied by the model (see Table 34). Volunteers who burned out on this Peer Ranking measure were initially moderately affiliative, 124 moderately resourceful with youth and not very involved with youth. They were also impulsive, dominant and historical. Although changes in Peer Rankings can be predicted from initial volunteer-related variables, it appears that change for the worse on the Peer—Ranking measure is not a good operational definition of volunteer burnout. While the Peer Ranking score does not appear to be a valid indicator of volunteer burnout, it does appear to be measuring some other construct that is predictable from initial volunteer variables. One hypothesis is that when given the task of ranking their peers, volunteers to some extent disregard item content, and instead give their subjective opinions and feelings about fellow supervision-group members. Thus, Peer Ranking would be a measure of change in popularity, rather than burnout. The fact that volunteers who decline on this measure are initially uninvolved, impulsive, dominant/aggressive and not very affiliative or resourceful tends to support this hypothesis. Supervisor Evaluation Seven predictors wereused in the discriminant function predicting group membership on this burnout variable. Four of those measured volunteer-related traits or situational factors that the burnout model implied were part of stage one of burnout (i.e. Hardworking, Involved, Nurturant, Youth SRD). Of these four, only two operated in the direction implied by the model. Volunteers who burned out on Super- visor Evaluations were initially nurturant, and worked with 125 youth who reported more delinquency. However, they were also not hardworking and not involved with youth. They scored high on the Dominant/Aggressive PRF dimension and tended to point out differences between themselves and youth. It may be that changes in Supervisor Evaluation are to some extent indicative of the type of volunteer burnout described by the theory, while the measure is also sensitive to some other kind of change. What that other kind of change is might be inferred from the variables that predicted it. It was originally thought that volunteers who were affiliative, hardworking and involved would burn out, because they would not be able to get as close as they wanted to clients or colleagues and because their clients would not improve as much as they expected. However, volunteers who produced more negative supervisor evaluations over time were initially 32E hardworking and REE involved with youth. They also tended to be dominant and aggressive. Volunteers who started out with these qualities would be unlikely to burn out, in the sense described by the model. They wppld be likely, however, to come into conflict with their super— visors. As mentioned already, When the relationship between Supervisor Evaluation and volunteer variables was discussed, the Supervisor Evaluation measure seems to be a sensitive gauge of the quality of the relationship between supervisor and volunteer. For this reason it may not be a very good burnout measure. 126 Course-Evaluation Four predictors contributed to the function that successfully predicted group membership on this variable. Two tapped factors discussed in the burnout model as important (i.e. Angry and youth-SRD). Both operated in the direction described in the model: volunteers who burned out on Course Evaluation were initially not angry with youth (i.e. they were not judgemental, pessimistic and dis- couraging), and they worked with youth who reported more delinquency. These volunteers did not rate themselves as dominant and aggressive initially. Thus, it seems that the Course Evaluation measure behaves as would be expected from the model of burnout. It may be considered, along with Self-evaluation, as one of the better burnout measures in the study. The effect of supervision/training intensity on Course Evaluation score was interesting. Both volunteers whose Course Evaluation scores improved and those whose Course Evaluation scores declined tended to be in high—intensity supervision/training groups. Those volunteers who did not change on this burnout score tended to be in low—intensity supervision/training groups. Thus, the increased training and contact with supervisors and peers of high intensity group does produce a reaction in volunteers. That reaction might be either positive or negative. It might prove interesting to further explore this relationship. For 127 example, are there certain personality—situational inter— actions that explain the different directions of this effect? It is also interesting that "situational predictors" figure prominently in the discriminant function predicting Course Evaluation score, but are not used at all in the function on Self-evaluation. It may be that these two burn— out measures are measuring different kinds of burnout. It is tempting to hypothesize that Course Evaluation measures a more ”outer—directed” form of dissatisfaction, while Self— evaluation measures a more ”inner-directed" form. Comments on the Method Discriminant Analysis Discriminant analysis was used repeatedly in this study. Often the dependent variable had to be transformed from a continuous to a categorical variable. Trichotomiza- tion was used, resulting in a fairly even distribution across groups. Any transformation from a continuous to a nominal variable involves a loss of information. Ideally, the categorization should reduce the ”noise” in the data without losing the effects. There was no attempt in this study to find out whether the method of categorization that was used achieved this balance. In future research using discriminant analysis with continuous dependent variables, several methods of categorization of the dependent variable into equal—sized groups might be compared. Canonical correlations and the number of significant functions for each categorization could 128 be examined. Choosing the ”best” of several categorizations, in a process similar to the choice of factor-solution in factor analysis, might be a way of avoiding data—loss, while using as few categories of the dependent variable as possible. Calculation of Burnout Scores Initially, data from five methods of measurement were combined into six burnout scores. Two of the original measures, Course Evaluation and Semantic Differential, had been developed in previous research. Changes over time in factor scores from these measures were used to calculate the burnout scores: Self, Youth and Course Evaluation. On the three other measures (Peer Ratings, Peer Rankings, and Supervisor Evaluation), a slightly different procedure was used to arrive at within—measure factors. Change scores on original items were factor—analyzed within measures to arrive at within-measure factors. These were then averaged within measures to produce three burnout scores. Thus three of the six final burnout scores (Self, Youth and Course Evaluation) were based on within—measure factors that were differences over time of factor scores. The other three final burnout scores (Peer Ratings, Rankings and Supervisor Evaluation), were based on within—measure sub-factors derived through factor analysis within measures of differences over time on items. It was not clear a priori whether this difference in procedure would make any real difference in results. 129 However, the fact is that Self, Youth and Course evaluations seemed to be the better operational definitions of burnout. Peer Ratings, Rankings and Supervisor Evaluation seemed to be measuring other constructs. Future research should try to eliminate the possibility that slight variations in pro— cedure may confound results: possibly by using less complex or "opaque" methods of calculating scores. Comparison of Predictors In the last part of the study, two types of predictors were compared to see whether either type had more predictive power. However, the two types of predictor may not have been given an equal chance. There were twelve "volunteer- related" predictors, each of which was continuous. There were only two "situational” predictors, each of which was dichotomous. Still, the two types of predictor seemed equally important. Had a full range of "situational” predictors been used, they may have proved superior to the volunteer-related predictors. In any case, this type of question would not be easily answered using this type of design with this sample. In examining the question of the relative importance of per— sonality and situational factors as predictors of burnout, it would be important to sample from a broad range of workers, who worked in a broad range of situations. Only in this way could results be generalized beyond the sample. The sample used here was fairly homogeneous in terms of the subjects and the situations in which they worked. It would 130 also be important to somehow equate the variances of the two types of predictor before they were compared. This issue was not addressed in this study. Validity of Burnout Scores Although the validity of the burnout scores is ex- plicitly addressed in the last part of the study, this entire research effort can be seen as an initial attempt to establish the construct validity of a measure (or set of measures) of burnout. The process of establishing the construct validity of a measure is normally drawn out over many years and many studies by many investigators. It has been described as a three—step process (Nunnally, 1978, p. 98): 1) Specification of the domain of observables related to the construct; 2) Determination of the extent to which observables tend to measure the same thing, several things, or many different things, using empirical research and statistical analyses; 3) Determination of the extent to which supposed measures of the construct produce results which are predictable from highly accepted theoretical hypo; theses concerning the construct, by performing studies of individual differences and/or controlled experiments. It is unlikely that any single study can accomplish any one of these steps, much less all three. This research makes contributions in all three areas. The "domain of observables" for the construct of burnout is specified by referring to the literature, which presents a large number of descriptions of symptoms of burnout in 131 various human services workers. This domain is then speci— fied for human services volunteers as well. Based on this domain, face—valid measures of volunteer burnout are derived. The extent to which ”observables” measure the same thing is examined by clustering empirical descriptions for human services workers and by performing factor analysis across volunteer "burnout measures." The relationships between the ”burnout scores” that result from this factor analysis and a variety of other variables are then studied. The burnout scores are predicted from volunteer-related variables. They are related to changes in volunteers' locus of control. They are used as predictors of client behavior. They are predicted from volunteer and situational variables. Following all these tests, we are left with two burn- out scores - Self-evaluation and Course evaluation - that seem to have the most ”construct validity." However, as Nunnally points out (1978, p. 104), the process of ”construct validation” rests on several assumptions. For example, in part five of the study (”Predicting Burnout From Volunteer and Situational Variables"), the assumptions were: 1) The volunteer-related and situational constructs described in the burnout model (e.g. need to achieve, need for affilia— tion, capacities for involvement and resourcefulness, severity of client problems, etc.) are positively related to volunteer burnout; and 2) The volunteer-related and 132 - situational variables used as predictors (e.g. PRF factors Hardworking and Affiliative, Behavior Measure factors In— volved and Resourceful, Youth—SRD, etc.) are measures of the constructs: need to achieve, need for affiliation, etc. If these two assumptions are true, according to Nunnally's logic, it is then possible to say: from the fact (empirically observed) that the predictors are positively related to the burnout scores, we infer that the burnout scores are measures of the construct of volunteer burnout. It is true that there were significant empirical re— lationships between predictors and burnout scores. But before jumping to the final conclusion, we should remind ourselves that at this stage in the process of trying to define and understand the concept of burnout, there is no way assumption #1 above could be called a ”highly accepted theoretical hypothesis.” Perhaps we should follow Nunnally's advice and avoid the language of ”construct validation” altogether. Instead we should consider this research to be an initial step in the process of defining the concept of burnout, which is to say ”the process of making an abstract word explicit in terms of observable variables" (Nunnally, 1978, p. 105). TO. the extent that the construct is more abstract, this process becomes more difficult. The construct of burnout is 2331 abstract. It will take some time and a good deal of work to define it more clearly. However, even at the present vague and ”fuzzy” level of concept definition, we 133 have found relationships, as predicted, between burnout scores and many other variables. This indicates that ”burnout” may be a very important construct and one worthy of further study. Directions for Future Research Various research projects have already been mentioned as follow—ups to analyses done in this study. There are several additional areas of research that could be explored once initial measures of burnout have been derived and that would greatly expand our understanding of burnout. One obvious application would be to study the prevalence of burn- out — just how serious is the problem for various sorts of human services workers? The present study found little impact of volunteer burnout on client outcome. However this result was attri— buted to certain structural aspects of the diversion project. It would be interesting to explore the effects of burnout on workers' creativity, resourcefulness, flexibility and productivity in a wide variety of work settings. The organizational causes of burnout were barely touched on in this study. Researchers studying the effects of situational variables or organizational structures on indivi— duals should consider using a ”burnout measure” as their dependent variable in addition to the traditional measures of satisfaction and turnover. In addition to looking at the effects on the individual, the effects on the work organization need to be studied. 134 It might make sense to think in terms of the burnout of entire work-groups or agencies. Measures of organizational burnout could be developed. This could lead to the study of the life cycles of organizations — somewhat analogous to the burnout cycles of individuals. A good deal of work must be done in order to under— stand the burnout stages. This aspect of the model has relatively little empirical support at present. If a set of common stages could be delineated and measures of these stages derived, the progress of groups of subjects through the stages could be observed. Patterns in the way indivi— duals pass through the stages could then be noted. This might lead into the more general study of how different types of individuals cope with various kinds of stress. The subject of ”coping” seems intimately related to the subject of burnout. In focusing attention on burnout it should not be forgotten that the ultimate goal is not just to understand human failure, but to promote human growth. APPENDICES 135 APPENDIX A Semantic Differential Concepts and Dimensions Concepts Factor 1 College Student, Myself, Volunteers, My Effectiveness (Self-evaluation) In Working with Adolescents, Human Services Career, Average Person, Psychology 370-400—490. Factor 2 Juvenile Justice System, Juvenile Court Staff, Probation (System—evaluation) Services for Juvenile Offenders, Police Officers, School Counselors, School Systems in This Area, School Principals, School Teachers, Lansing Area, Employment Programs for Juveniles, Diversion for Juvenile Offenders, Supervisors. Factor 3 Adolescents, Parents of Adolescents, Juvenile Delinquents. (Client-evaluation) Dimensions Pleasant—Unpleasant Good-Bad Effective-Ineffective Friendly-Unfriendly Helpful—Harmful 1136 APPENDIX 8 Peer Rating Items and Factors Items Factor 1 1. Planfulness (plans and carries out plans.) (Perseverance) (1 = very planful; 5 = not planful at all) 2. Perseverance (l = very perseverant; 5 = not perseverant at all) 3. Resourcefulness (1 = not resourceful at all; 5 = very resourceful) 4. Motivation (1 = very motivated; 5 = not motivated at all) Factor 2 1. Giving input into other people's cases. (Group Involvement) (1 = never gives input; 5 = always gives input) 2. Showing interest in others' cases. (1 = always shows interest; 5 = never shows interest) 3. Helpfulness with others' cases. (1 = never helpful; 5 = always helpful) Factor 3 1. Empathy (understanding and reflections of people's (Empathy/Effectiveness) feelings). (1 = not empathic at all; 5 = very empathic. 2. Warmth. (1 = warm; 5 = cold) 3. Effectivenmm in using skills taught in this course. (1 = very effective; 5 = very ineffective) 1237 APPENDIX C Peer Ranking Items and Factors Factor 1 (Talkative in Class) Factor 2 (Gets Along with Others) Factor 3 (Motivation) Factor 4 (Empathy) Factor 5 (Perseverance) Items not fitting this factor structure Confident when using skills learned in this course. (1 = completely lacks confidence; 5 = is very confi- dent) Talkative in class. (1 = always talks; 5 = never talks) Giving input into other's cases. (1 = never gives input; 5 = always gives input) Getting along with the youth assigned to her/him. (1 = doesn't get along at all; 5 = gets along easily) Well liked by fellow students. (1 = not well-liked at all; 5 = very well—liked) Motivation (1 = very motivated; 5 = not motivated at all) Empathy ( understanding and reflections of people's feelings.) (1 = not empathic at all; 5 = veryempathic. Perseverance (l = very perseverant; 5 = not perseverant at all) Understanding of the model taught in this course. (1 = doesn't understand at all; 5 = understands completely) Success in his/her career. (1 = very likely to succeed; 5 = not likely to succeed at all) 138 APPENDIX 0 Supervisor Evaluation Items and Factors Factor 1 (Resourceful/Concerned) Factor 2 (Available/Helpful) Factor 3 (Discussion Quality) Factor 4 (Involved/Understanding) How concerned about your case was s/he? (l = never concerned; 5 = always concerned) How much knowledge about the field of delinquency does s/he have? (1 = none; 5 = a great deal) How resourceful was s/he about your case? (1 = not resourceful at all; 5 = very resourceful) How supportive was s/he? (1 = very unsupportive; 5 = very supportive) How helpful to you was s/he? (1 = always helpful; 5 = never helpful) If you had a problem, how comfortable would you feel in seeking her/him out about it? (1 = very comfortable; 5 = very uncomfortable) How available was s/he when you wanted him/her? (l = very available; 5 = very unavailable) How often did s/he offer ”expert” answers as solutions rather than helping you generate your own answers? (1 = always; 5 = never) How skilled was s/he at encouraging and generating group discussion? (1 = very unskilled; 5 = very skilled) How well do you feel you got along with her/him? (1 = didn't get along at all; 5 = got along very well) How involved was s/he in your case? (1 = very involved; 5 = very uninvolved) How understanding was s/he about your concerns? (1 = always understanding; 5 = never understanding) 139 APPENDIX D (Cont.d.) Items not fitting 1. How consistent with the approach was s/he? this factor structure ‘ (l = always consistent; 5 = never consistent) 2. How conscientious was s/he? (1 = always conscientious; 5 = never conscientious) 3. How familiar is s/he with community resources? (1 = very familiar; 5 = very unfamiliar) 140 APPENDIX E Course Evaluation Items and Factors Factor 1 1. (Take again) Factor 2 (Talkative outside group) 1. Factor 3 1. (Class discussion) Knowing what you know now about this Course, would you take it again? (1 = definitely not; 5 = definitely yes) Would you recommend this course to a friend? (1 = definitely yes; 5 = definitely not) How completely did this course meet your initial expectations of it? (1 = met all expectations; 5 = met no expectations) How much do you talk with friends not in this class about the training and skills you have received? (1 = never; 5 = very frequently) Outside of this class, how much do you talk with other students in this class about the training and skills you have received? (1 = never; 5 = very frequently) If you have a case already assigned, how frequently 0. 0 you talk with friends not in this class about problems, concerns, and issues dealing with your case? (1 = never; 5 = very frequently) If you have a case assigned, how frequently do you talk outside of class with other students in your class about problems, concerns and issues dealing with your case? (1 = never; 5 = very frequently) If already assigned, how much do you like the class discussion of your case? (1 = like very much; 5 = dislike very much) 1141 APPENDIX E (Cont.d.) Factor 4 (Training) If already assigned, how useful to you is the class discussion of your case? (1 = completely useless; 5 = very useful) How much do you like the class discussion of other students' cases? (1 = dislike very much, 5 = like very much) How useful to you is the class discussion of others' cases? (1 = completely useless, 5 = very useful) How much do you believe you’ve learned from this course? (1 = I learned nothing; 5 = I learned a huge amount) How do you feel about the model of intervention that you were trained in? (1 = liked very much; 5 = extremely disliked) How do you feel about the amount of training you have received? (1 = extremely inappropriate; 5 = extremely appro— priate) 142 APPENDIX F Jackson PRF Scales and Their Definitions Scale Defining Trait Adjectives Factor 1 (Impulsive/Playful) Impulsivity Hasty, rash, uninhibited, spontaneous, reckless, irrepressible. quick-thinking, mercurial, impatient, incautious, hurried, impulsive, foolhardy, excitable, impetuous. Order (—)a Neat, organized, tidy, systematic, well-ordered, dis- ciplined, prompt, consistent, orderly, clean, methodical, scheduled, planful, unvarying, deliberate. Play Playful, jovial, jolly, pleasure—seeking, merry, laughter— 1oving, joking, frivolous, prankish, sportive, mirthful, fun—loving, gleeful, carefree, blithe. Factor 2 (Affiliative/Dependent) Affiliation Neighborly, loyal, warm, amicable, good—natured, friendly, companionable, genial, affable, cooperative, gregarious, hospitable, sociable, affiliative, good—willed. Autonomy (-) Unmanageable, free, self—reliant, independent, autonomous, rebellious, unconstrained, individualistic, ungovernable, self—determined, non—conforming, uncompliant, undominated, resistant, lone—wolf. Social Recognition Approval—seeking, proper, well—behaved, seeks recognition, courteous, makes good impression, seeks respectability, accommodating, socially proper, seeks admiration, obliging, agreeable, socially sensitive, desirous of credit, behaves appropriately. Understanding (-) Inquiring, curious, analytical, exploring, intellectual, reflective, incisive, investigative, probing, logical, scrutinizing, theoretical, astute, rational, inquisitive. 143 APPENDIX F (Cont.d.) Scale Defining Trait Adjectives Aggression Dominance Exhibition Achievement Endurance Nurturance Factor 3 (Dominant/Aggressive) Aggressive, quarrelsome, irritable, argumentative, threatening, attacking, antogonistic, pushy, hot-tempered, easily-angered, hostile, revengeful, belligerent, blunt, retaliative. Governing, controlling, commanding, domineering, in- fluential, persuasive, forceful, ascendant, leading, directing, dominant, assertive, authoritative, powerful, supervising. Colorful, entertaining, unusual, spellbinding, exhibition— istic, conspicuous, noticeable, expressive, ostentatious, immodest, demonstrative, flashy, dramatic, pretentious, showy. Factor 4 (Hardworking) Striving, accomplishing, capable, purposeful, attaining, industrious, achieving, aspiring, enterprising, self— improving, productive, driving, ambitious, resourceful, competitive. Persistent, determined, steadfast, enduring, unfaltering, preservering, unremitting, relentless, tireless, dogged, energetic, has stamina, sturdy, zealous, durable. Factor 5 (Nurturant) Sympathetic, paternal, helpful, benevolent, encouraging, caring, protective, comforting, maternal, supporting, aiding, ministering, consoling, charitable, assisting. Note: From Jackson, 1974, pp. 6—7. aThis sign means the scale is negatively weighted on the factor. léuh APPENDIX 6 Behavior Measure Items and Factors Factor 1 (Involved) Factor 2 (Resourceful) Factor 3 (Distant) Factor 4 (Angry) Factor 5 (Different) Factor 6 (Historical) Items not fitting this factor structure I. Illogical. (Flighty, unreasonable, doesn't make sense, off—the—wall) Voiced similarities - now. (I'm like you in that way) Impatience. (Interrupting youth while s/he is talking; didn't wait for answer to first question before asking another) Resourceful. (Ideas, things to do, courses of action) Specific planning. (Finding out, making steps) Voiced similarities — then. (I was like you in that way when I was your age) Voiced dissimilarities - then. (I wasn't like you in that way when I was your age) Pessimistic. (Discouraged youth) Judgemental. (Voiced opinions, blaming, critical, used ”should,” ”ought to”) Discounting. (Said youth shouldn't feel as s/he does) Voiced dissimilarities - now. (I'm not like you in that way) Historical. (Mentioned past, asked about past of youth) Optimistic. (Encouraged youth) Empathic. (Mentioned feelings of youth) 145 APPENDIX H Self-Report Delinquency Items How often in the last six weeks have you: 1. 2. 10. ll. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. Skipped class when you were in school? Gone onto someone's land when they didn't want you to be in there, or without their permission? Gone into a house or building when you weren't supposed to be there? Threatened to hurt someone? Been told to bring your parents to school for something you did wrong? Damaged or messed up something not belong to you? Hurt someone badly enough for him/her to need bandages or a doctor? Taken some part of a car or some gasoline? Hit a member of your family? (in anger) Has not been allowed to go to school until the superintendent or principal told you that you could go again? (been suspended) Taken something not belonging to you worth less than $2.00? Drunk beer or liquor? (includes sips) Run away from home? Skipped a full day of school? Been sent to the school principal's office for bad behavior in class? Carried a gun or a knife? Taken something not belonging to you worth over $50.00? Set fire to someone else's property? Used or threatened to use a weapon to get something from a person? Taken something from a store without paying for it? (regardless of price) Smoked without your parents knowing about it or without permission? (regular cigs.) Taken a car without the owner's permission? (includes joyriding) Smoked marijuana? Taken something from a person by force? (may or may not use a weapon) Beaten up on somebody or fought someone? (physically) Taken drugs or pills, other than marijuana? Bought or gotten something that was stolen by someone else? Broken into a place and stolen something? Taken things worth less than $50.00? (over $2.00) LI ST OF REFERENCES LIST OF REFERENCES Abdel—Halim, A. Interaction effects of power equalization and subordinate personality on job satisfaction and performance. Human Relations, 1979, 22(6), 489—502. Allesandra, A. Burnout: The symptoms, causes and cures. Registered Representative, April 1981, pp. 25-26. Argyris, C. Personality and organization: The conflict between the system andgthe individual. New York: Harper and Row, 1957. Armstrong, K. An exploratory study of the inter—relation— ships between worker characteristics, organizational structure, management process and worker alienation from clients (How to avoid burnout). (Doctoral dissertation, University of California, Berkeley, 1978). Dissertation Abstracts International, 1978, 29(2), 670B. Bloom, B. and Parad, H. Interdisciplinary training and interdisciplinary functioning: A survey of attitudes and practices in community mental health. American Journal of Orthopsychiatry, 1976, fi6(4), 669—677. Cattell, R.B. Factor analysis: An introduction to essentials. (I) The purpose and underlying models, (II) the role of factor analysis in research. Biometrics, 21, 190—215; 405-435. Cherniss, C. and Egnatios, E. Participation in decision— making by staff in community mental health programs. American Journal of Community Psychology, 1978, 6, 171-190. _ Costello, T. w. and Zalkind, S. S. (Eds.), Psychology in administration: A research orientation. EnglewoodF Cliffs, N.J.: Prentice-Hall,il963. Cowen, E.L. and Zax, M. The mental health fields today: Issues and Problems. In E. L. Cowen, E. A. Gardner and M. Zax (Eds.), Perspectives in community mental health. Chicago: Adeline, 1969. Daley, M. R. Burnout: Smoldering problem in protective services. Social Work, 1979, 2&(5), 375-379. 146 147 Davidson, W. and Redner, R. Assessing differentialimpact of delinquency treatment. Grant proposal submitted to NIMH, Michigan State University, 1982. Denton, R. T. The effects of different leadership behaviors on the job satisfaction and job performance of pro— fessional mental health workers. (Doctoral disserta- tion, Ohio State University, 1976.) Dissertation Abstracts International, 1976, 81(5), 3183A. Durlack, J. A. Comparative effectiveness of paraprofessional and professional helpers. Psychological Bulletin, 1979, 86, 80—92. Durlack, J. A. Evaluating comparative studies of para— professional and professional helpers: A reply to Nietzel and Fisher. Psychological Bulletin, 1981, 89, 566-569. ‘— Edelwich, J. and Brodsky, A. Burnout: Stages of disillusionment in the-helpingprofessions. New York: Human Sciences Press, I980. Effraty, D. Organizational identification: Sources and consequences for person and performance. (Doctoral dissertation, Case Western Reserve University, 1976). Dissertation Abstracts International, 1976, 37(2), 950- 951B. '_— Ellison, K. W. and Genz, J. L. The police officer as burnout samaritan. Law Enforcement Bulletin, 1978, pp. 1-7. Freudenberger, H. J. Staff burnout. Journal of Social Issues, 1974, 30(1), 159—165. Freudenberger, H. J. Burnout: Occupational hazard of the child case worker. Child Care Quarterly, 1977, 6(2), 90-99. - Gilmour—Barrett, K. C. Managerial systems and interpersonal treatment processes in residential centres-for disturbed youth. (Doctoral dissertation, University of Waterloo, 1974). Dissertation Abstracts International, 1974, 22(5), 2428B. Glicken, M. D. A regional study of the job satisfaction of social workers. (Doctoral dissertation, University of Utah, 1977). Dissertation-Abstracts International, 1977, 38(3), 1658A. Gottheil, E. Poor morale in treatment personnel. Alcohol Health and-Research-World; Experimental Issue, l978 (Spring), pp. 20—25. Ci 148 Gudowski, Richard M. An investigation of paraprofessionals in mental health: A descriptive study of attitudinal differences. (Doctoral dissertation, U.S. International University, 1974). Dissertation Abstracts International, 1974, 32(6), 30143—30153., Hall, R., Gardner, E. and Perl, M. The professional burnout syndrome. Psychiatric Opinion, 1979, 16, 12. Harris, Major USA, J. J. Status as a moderator of job satisfaction and role ambiguity: A study of the Army Behavioral Science Specialist. (Doctoral Dissertation, University of Maryland, Baltimore, 1976). Dissertation Abstracts International, 1976, 37(6), 3909A. Hendrickson, B. Teacher burnout: How to recognize it, what to do about it. Learning, 1979, pp. 37—39. Hesse, K. The paraprofessional as a referral link in the mental health delivery system. Community Mental Health Journal, 1976, 12(3), 252—258. Hrebniak, L. G. Effects of job level and participation on employee attitudes and perceptions of influence. Academy of Management Journal, 1974, 11(4), 649-662. Jackson, D. N. Personality research form manual. Goshen, N.Y.: Research Psychologists Press, 1974. Kahn, R. Job burnout: Prevention and remedies. Public Welfare, 1978, 36(2), 61-63. Kamis, E. An epidemiological approach to staff burnout. Paper presented at American Psychological Association convention, Montreal, 1980. Karlsruher, A. E. The nonprofessional as a psychotherapeutic agent. American Journal of Community Psychology, 1974, 2, 61-77. "F Lorber, J. and Satow, R. Creating a company of unequals: Sources of occupational stratification in a ghetto community mental health center. Sociology of Work and Occupations, 1977, 4(3), 281—302. Marshall, R. E. and Kasman, C. Burnout in the neonatal intensive care unit. Pediatrics, 1980, 65(6), 1161— 1165. ‘— Maslach, C. The client role in staff burnout. Journal of Social Issues, 1978, 34(4), 111—124. (a) 149 Maslach, C. Job burnout: How people cope. Public-Welfare,‘ 1978, 33(2), 56-58. (b) Maslach, C. and Jackson, S.E. Lawyer burnout. Barrister, 1978 (Spring), 3, 52-54. Maslach, C. and Pines, A. The burnout syndrome in the day care setting. Child Care Quarterly, 1977, 6(2), 100- 113. — Mervyn, F. V. A comparison of perceptions of decision- making roles and job satisfaction of mental health clinic directors. (Doctoral dissertation, Boston College, 1978). Dissertation-Abstracts-International, 1978, 32(6), 3284A-3285A. Meunerick, L. A. Client typologies: A method of coping with conflict in the service worker-client relationship. Sociology of-Work and Occupations, 1974, 3(4), 396-418. Mitchell, C., Kantrowitz, R., and Davidson, W. Differential attitude change in nonprofessional experience: An experimental comparison. Journal of Counseling Psychology, 1980, 31(6), 625-629. Nietzel, M. T. and Fisher, S. G. Effectiveness of pro— fessional and paraprofessional helpers; A Comment on Durlak. Psychological Bulletin, 1981, 32, 555-565. Nogardi, G. S. An analysis of existing relationships betweai the administration processes and organizational commitment and job involvement of recreational staff members within mental health centers across Canada. (Doctoral dissertation, Unversity of Oregon, 1978). Dissertation Abstracts International, 1978, 38(6), 3367A—3368A. _’ Nunnally, J. C. Psychometric Theory (2nd ed.). New York: McGraw—Hill, 1978. Price, J. T. The study of turnover. Ames, Iowa: Iowa State University Press, 1977. Pines, A. and Maslach, C. Characteristics of staff burnout in mental health settings. Hospital and Community Psychiatry, 1978, 33, 233-237. Quint, G. H. An investigation of the attitudes of rehabilitation workers toward narcotic addictsengaged in a rehabilitation program as they relate to rehabili- tation workers' length of time on the job, type of pro- gram where employed and selected demographic variables. (Doctoral dissertation, New York University, 1974). Dissertation Abstracts International, 1974, 34(8, pt.2), 5316A. ‘— 150 Ray, J. W. A study of paraprofessionals: Factors governing job satisfaction and career objectives. (Doctoral dissertation, New York University, 1975). Dissertation Abstracts International, 1975, 33(6), 3064B—3065B. Roehl, C. T. The Mental Health Associate: The effects of social environmental factors on job satisfaction. (Doctoral dissertation, University of Texas, Austin, 1976). Dissertation Abstracts International, 1976, 31(10), 5280B. Rotter, J. B. Social learning theory and clinical psychology. N.Y.: Prenctice-Hall, 1954. Sarata, B. P. V. Job characteristics, work satisfactions and task involvement as correlates of service delivery strategies. American Journal of Community Psychology, 1977, 3, 99-109. Sarata, B. P. V. and Jeppesen, J. C. Job design and staff satisfaction in human service settings. American Journal of Community Psychology, 1977, 3, 229-236. Schreisheim, C. A. and Murphy, C. I. Relationships between leader behavior and subordinate satisfaction and per- formance: A test of some situational moderators. Journal of Applied Psychology, 1976, 93, 634-641. Shaefor, B. W. The effects of board members on staff of community mental health centers: A study of the relationships of their values to job satisfaction. Journal of Social Welfare, 1976, 3(1), 75-82. .— Shinn, M. Burnout in human services agencies: Patterns of job stress, psychological strain and coping responses. New York University mimeo, 1979. Tinsley, M. Interdisciplinary role perception and job satisfaction: A study of Missouri's nine multi- disciplinary diagnostic clinics. (Doctoral disserta- tion, University of Arkansas, 1974). Dissertation Abstracts International, 1974, 33(6), 3533A-3534A. Veninga, R. L. and Spradley, J. P. The work/stress connection: How to cope with job burnout. Boston: Little, Brown, & Co., 1981. Vroom, V. H. Work and motivation. New York: Wiley, 1964. 151 Weinstein, R. 8., Gibbs, J. T., and Middlestadt, S. E. College students in human services agencies: Per- ceptions of their input on the setting. American Journal of Community Psychology, 1979, 3, 209-221. Zaleznik, A., deVries, M., and Howard, J. Stress reactions in organizations: Syndromes, causes and consequences. Behavioral Science, 1977, 33(3), 151-162. e7 . In , IFF Ill-“L nPflhb (I..._r(_,... . \ 515'. . .l .r. . arthiwaiii .4.¢_.. IE5 N 9 G I "((Iy'yilllllllll