Psychological Reports, 1990, 66,451-457.

Q Psychological Reports 1990

GOLEMBIEWSKI'S PHASE MODEL O F PSYCHOLOGICAL BURNOUT: SOME ISSUES ' JACOB WOLPIN, RONALD J. BURKE, AND ESTHER R. GREENGLASS York University Summary.-The present investigation examined three issues relevant to Golembiewski's phase model of psychological burnout. These were use of the mean versus median in creating high and low subgroups on Maslach Burnout Inventory subscales, use of different item structures on Maslach subscales, and different sequences of Maslach subscales proposed by Golembiewski and Maslach in the development of psychological burnout. Use of mean versus median, or different item structures on Maslach subscales, made relatively little difference. Golembiewski's ordering of the subscales, compared to Maslach's, produced a more linear progression over the 8 phases, as well as on three antecedents and consequences, but both sequences were related to these three variables in an almost identical fashion.

Researchers have shown increasing interest in psychological burnout in organizations during the past few years (Shirom, 1989; Journal of Health and Human Resources Adminis&ation, 1984, 1986, 1989). Burnout develops over time, hence is a process. Cherniss (1980) has explicitly proposed a process model of burnout. Maslach (1982) implicitly writes about burnout as a developmental phenomenon. One way in which burnout might be examined as a process involves the notion of phases within the concept of burnout. Golernbiewski and his colleagues (1986) have proposed and pursued this line of research. They assigned different priorities or prepotencies to the three subscales of the Maslach Burnout Inventory (Maslach & Jackson, 1981) on theoretical grounds. Depersonalization was considered the least important contributor to burnout. Lack of Personal Accomplishment was rated a more important contribution to burnout. Finally, s motional Exhaustion was considered to be the most important contribution to burnout. Dichotomizing the distribution of scores on each Maslach subscale at the median as high and low generated eight phases of burnout (see Table 1). Research using the progressive phase notion has had two emphases. The first, and the largest body of work, has attempted to validate the notion of progressive phases itself. This stream of research has compared individuals at various phases on a wide array of potential antecedents and consequences of

'Preparation of this manuscript was supported in part by the Faculty of Administrative Scud~es, the Department of Psycholo , York University, and by Imperial Oil Limited. We acknowledge the cooperation of the B O of Education ~ i n collecting the data. Request reprints from R J Burke, Organizational Behavior/Industrial Relations, Faculty of Administrative Studies, Yorh University, 4700 Keele Street, North York, Ontario, Canada M3J lP3.

452

J. WOLPIN, ETAL.

psychological burnout (Burke & Deszca, 1986; Burke & Greenglass, 1989; Burke, Shearer, & Deszca, 1984; Golembiewsh, et al., 1986). This research has provided considerable support for the validity of the underlying notion of progressive phases of burnout. That is, individuals in more advanced . phases almost always report more negative experiences (e.g., greater stress, less job autonomy) and more negative outcomes (e.g., less job satisfaction, more psychosomatic symptoms) than do individuals in less advanced phases. TABLE 1 GOLEMBIEWSKI'S EIGHT-PHASE MODELOF BURNOUT Maslach Subscales Depersonalization Personal Accomplishment Emotional Exhaustion

Phase

IV

v

VI

VII

VIII

Low Low

High

Low

High

High High

Low

Low

Low

High

High

High High

High High High

I

I1

Low Low Low

High

Low

Low Low

111

A second but considerably smaller stream of research has examined the advantages of three and four phases of psychological burnout over the eight-phase model initially proposed by Golembiewski (Cahoon & Rowney, 1984; Rowntree, 1984). Cahoon and Rowney collapsed Golembiewski's eight phases into three: low (Phases I, 11, and 111), moderate (Phases IV and V), and high (Phases VI, VII, and VIII). Using fewer phases is proposed as one way of dealing with entry-transition questions inherent with the full model (Rountree, 1984). The steady accumulation of research findings using Golembiewski's phase model (Golembiewski & Munzenrider, 1988; special issues of the Journal of Health and Human Resources Administration in 1984, 1986, and 1989) and the opportunity to contrast two models of burnout (Golembiewski, 1989; Leiter, 1989) suggested that the time was right to take stock of what we knew, highlight potential concerns, and identify needed research directions. Group and Organization Studies (1989) devoted part of one issue to Golembiewski's phase model. In this collection Burke (1989) raised some conceptual and methodological concerns. The former included issues of definition, translating from advanced phases in Golembiewski's model to levels of distress, ways in which individuals get to a particular phase or move between phases and the appropriateness of the three Maslach scales for all jobs. The latter included the appropriate sequence of the three Maslach scales in the development of burnout, the high intercorrelation among the three Maslach subscales, Golembiewski's modification of the original subscales, use of the median versus mean in defining high and low subgroups, and the use of a normative sample to determine absolute benchmarks to per-

ISSUES IN PSYCHOLOGICAL BURNOUT

453

rnit comparisons across studies. Burke (1989) raised some issues for future research on Golembiewski's phase model to build on and hopefully improve our understanding of processes of psychological burnout at work. The present study addresses three of these issues: (1) the sequence of the three Maslach scales in developing the eight phases, (2) the use of the median versus mean in creating high and low subgroups, and (3) the item structure of the Maslach subscales. It appears that Golernbiewski and Maslach propose a different developmentaIsequence involving the three Maslach scales. Is one sequence superior to another in terms of the eight phases themselves and their external validity? Do they differ? And does it make a difference whether the mean versus the median is used to create the eight phases? Finally, do the changes Golembiewski made to the item structure of the three Maslach subscales make a dfference? METHOD This study involves a secondary analysis of data collected in a longitudinal study of psychological burnout among school-based educators. Respondents This study involved 746 educators (teachers, department heads, vice principals, principals) from a single Board of Education in a large Canadian city. Questionnaires were sent to over 2000 potential respondents. The response rate obtained was about 35%. Respondents were of both sexes (50% women, 50% men), mostly married (71%) and almost evenly split across three levels of students (elementary, junior, secondary). Ages ranged from 25 to 65 yr., with a mean of 41.1 yr. Eighty-seven percent had been in education for more than 10 yr., and 33% for over 20 yr. Measures Psychological burnout.-This was measured by the Maslach Burnout Inventory (Maslach & Jackson, 1981) since this instrument is the basis for Golembiewski's phase model. This measure has three subscales (see Table 1). Only the frequency response format was used in the research. Depersonalization had 5 items for Maslach and 8 items for Golembiewski, of which we included 7. Lack of Personal Accomplishment had the same 8 items for both Golembiewski and Maslach. Emotional Exhaustion had 9 items in Maslach's version and 7 items in Golembiewslu's. The external validity of the phase model was examined using three measures previously used in research on psychological burnout. Work-setting characteristics.-Cherniss (1980) identified eight worksetting characteristics as antecedents of burnout. These were Adequacy of orientation, Workload, Scope of Client Contacts, Agreement with Institutional Goals, Autonomy, Leadership and Supervision, Lack of Stimulation,

454

J. WOLPIN, ET AL.

and Social Isolation. Each was measured by multiple items, then combined into a composite score. Job satisfaction.-This was assessed by a five-item measure of global job satisfaction created by Quinn and Shepard (1974). Marital satisfaction.-This was measured by a scale developed by Orden and Bradburn (1968). Respondents indicated whether they and their partners had experienced particular positive events in the preceding two weeks and whether they and their partner had experienced particular negative events in the preceding two weeks. The latter were then subtracted from the former.

Mean Versus Median The mean and the median were determined for each of the three Maslach subscales using both the original Maslach item structure and the revised Golembiewski structure. Although the mean was always higher than the median, the differences were small and never statistically significant; see Table 2. These findings resulted from the fact that the distribution of individual scores on the three Maslach subscales formed a normal distribution. TABLE 2

COMPARISON OF EIGHT PHASES F*

Phase

I

I1

I11

IV

v

VI

VII

VIII

Golembiewsh Sequence (Masl.ach Composite Measure)

n M

Mdn Range Minimum Maximum

184 38.3 38.7

45 50.0 50.7

90 52.1 51.8

54 61.4 61.5

60 53.9 53.2

82 68.0 66.7

33 66.7 66.0

198 83.6 81.8

22 51

36 60

35 79

46 76

40 70

48 110

55 92

61 122

93 52.0 52.8

54 68.7 68.0

54 62.1 62.6

177 81.8 83.4

35 60

55 92

46 79

61 122

Maslach Sequence (Maslach Composite Measure) n 186 69 49 67 M 38.6 56.6 49.0 69.0 Mdn 38.8 55.0 48.9 67.4 Range M u m 22 47 36 53 54 81 60 110 Maximum

370.3 339.4

370.3 339.4

* p = ,0001.

Two questions can be raised at this point. Does it make any difference whether one uses the mean or median in creating high and low subgroups on the Maslach subscales? Which is better? The answer to the first question is that it does not make much difference whether one uses the mean or median

ISSUES IN PSYCHOLOGICAL BURNOUT

455

for a particular sample. I t does make a difference, however, if one is comparing the same sample at two points in time, or two different samples. I n this case creating groups based on the mean would be better, using the means on the first administration as norms for a repeated measures design (a longitudinal study), or the means for one group as a benchmark to compute the high and low subgroups for a second occupational group. The issue of a normative group to anchor group comparisons must still be confronted.

Item Structure of Maskzch Subscales Does it matter whether one uses the item structure proposed by Maslach or the revised item structure offered by Golembiewski? Our analyses indicated that use of both produced very similar results. Obviously the specific values of the mean, median, mode and range of scores differed as a function of the different number of items in the subscales based on recommended analyses of Maslach and GoIembiewski. Otherwise no differences were apparent. The means were higher than the median in both approaches. The distribution of people across the eight phases was very similar. The progression of mean values for the eight phases was similar. Both were predictive of work-setting characteristics, job satisfaction and marital satisfaction to the same extent. Our data indicated it did not make much difference whether one uses the Maslach item structure or the Golembiewski item structure of the Maslach subscales in the creation of the eight phases. We prefer to use the Maslach structure since we, and others, have found factor structures identical to hers.

Sequencing the Maslach Subscales Golembiewski and his colleagues order the three Maslach scales in a particular sequence of virulence (see Table 1) in the development of psychological burnout. Our reading of Maslach (1982) implied a different order or sequence. We compared the relative merits of these different sequences in two ways: (1) the progression of mean values across the eight phases, and (2) the relationship of the eight burnout phases to other variables (work setting characteristics, job satisfaction, marital satisfaction). Let us first examine characteristics of the eight phases themselves. Table 2 presents the mean, median and range of scores for each of the eight phases of Golembiewski's phase model using Golembiewski's sequence (top half) and Maslach's sequence (bottom half). We see a more linear progression across the eight phases when the Golembiewski sequence is used. Let us now examine the validity of the eight phase models, one based on the Golembiewski sequence, the other on the Maslach sequence. These data are shown in Table 3 . Both sequences were related in the same direction and with approximately the same magnitude (see F ratios) to the three ante-

-

456

J, WOLPIN, ETAL

cedents and consequences. Once again, a smoother (more linear) progression appeared on these three variables when the Golembiewslu sequence was used to create the eight-phase model. TABLE 3 COMPARINGSUBSCALE SEQUENCES FOR GOLEMBIEWSKI'S AND MASLACH'S BURNOUTINVENTORIES

Ft

Phase* Golembiewski's Sequence Work Setting Characteristics Job Satisfaction Marital Satisfaction Work Setting Characteristics Job Satisfaction Marital Satisfaction

I

I1

111

IV

76.5'"'" 22. 7'b"" 7.3'

79.9bh' 22.2'h' 7.1

85.7"" 20.3"' 6.6

90.9~" 18.8~" 6.1

v

VI

92.1" 17.8" 6.9

96.3" 17.8~'" 6.1

VII 99.4*' 17.0'~'"

39.4 52.2 6.4

VIII 109.3'~~" 13, 8ih~m"o

6.1

5.5'

I11

IV

Phase*

I Maslach's Sequence Work Setting Characteristics Job Satisfaction Marital Satisfaction

76.7'1dm 22. 7'"d" 7.4'

v Work Setting Characteristics Job Satisfaction Marital Satisfaction

85.1dh' 20.2"' 6.7

I1 91.7~' 19.2" 6.7

VI 97.ahh 16.4~~' 6.3

81.TUi 98.5" ~ 2 . 2 ~ ~ ' 17.7bgm 6.7 6.1

VII

VIII

91.2'" 19.0'" 6.3

107.3'~"

39.2 49.0 7.4

14,

5.3'

'Phase means with same superscripts are significantly different ( p = .05). tp = .0001.

Implications Our own program of research has examined Golembiewski's phase model of burnout in two samples of police officers and a sample of schoolbased educators. We were greatly attracted to the model when we first read about it. There was a strong appeal in its abdity to place individuals in different phases of burnout. But as we became more familiar with it and its use, we identified some potential concerns (Burke, 1989). The present study addressed some of these concerns and raised others. These concerns must be examined if the phase model is to be a useful approach to burnout research and intervention. REFERENCES BURKE,R. J. (1989) Toward a phase model of burnout: some conceptual and methodological concerns. Group and Organization Studies, 14, 23-32. BURKE,R. J., & DESZCA, E. (1986) Correlates of psychological burnout phases among police officers. Human Relations, 39, 487-502.

ISSUES IN PSYCHOLOGICAL BURNOUT

457

& GREENGLASS, E. R. (1989) Correlates of psychological burnout phases among teaches. Journal of Health and Human Resources Administration, in press. BURKE, R. J., SHEARER,J., & DESZCA,E. (1984) Correlates of burnout phases among police officers. Group &r Organization Studies, 9, 451-466. CAHOON, A. R., & ROWNEY, J. I. A. (1984) Managerial burnout: a comparison by sex and level of responsibility. Journal of Health and Human Resources Administration, 7, 240-263. CHERNISS,C. (1980) Professional burnout in human service organizations. New York: Praeger. GOLEMBIEWSKI, R. T. (1989) A note on Leiter's study: highlighting two models of burnout. Group and Organization Studies, 14(1), S-13. GOLEMBIEWSKI, R. T., & MUNZENRIDER, R. F. (1988) Phases of burnout. New York: Praeger. GOLEMBIEWSKI, R. T., MUNZENRIDER, R. F., & STEVENSON, G . (1986) Stress in organizations. New York: Praeger. Group and Organization Studies. (1989) Journal, 14, March. Journal of Health and Human Resources Administration (1984) Journal, 7, Fall. Journal of Health and Human Resources Adminisfration (1986) Journal, 9, Summer. Journal of Health and Human Resources Administration ( 1989) Journal, 11, Spring. LEITER, M. (1989) Conceptional implications of two models of burnout: a response to Golembiewzki. Group and Organization Studies, 14(1), 15-22. MASLACH, C. (1982) Burnout: the cost ofcaring. Englewood Cliffs, NJ: Prentice-Hall. MASLACH,C., & JACKSON, S. E. (1981) Measurement of experienced burnout. Journal of Occupational Behavior, 2, 99- 113. ORDEN,S., & BRADBURN, N. M. (1968) Dimensions of marriage happiness. The American Journal of Sociology, 73, 715-73 1. QUINN,R. P., & SHEPARD,L. J. (1974) The 1972 quality of employment study. Ann Arbor, M I : Survey Research Center, Institute of Social Research. ROWTREE, B. (1984) Psychological burnout in task groups. Journal of HeaW and Human Resources Administration, 7, 235-248. SHIROM,A. (1989) Burnout in work organizations. In C. L. Cooper & I. Robertson (Eds.), International review of industrial and organizational psychology. New York: Wiley. Pp. 25-48.

BURKE,R. J.,

Accepted February 10, 1990.

Golembiewski's phase model of psychological burnout: some issues.

The present investigation examined three issues relevant to Golembiewski's phase model of psychological burnout. These were use of the mean versus med...
247KB Sizes 0 Downloads 0 Views