Psychological Reports, 1991, 68, 963-977.
@ Psychological Reports 1991
RELATIONSHIP BETWEEN BURNOUT AND PERFORMANCE ' ANNA-MARIA GARDEN GDS Consultants Summary.-This paper seeks to clarify the nature of the relationship between burnout and both perceived and actual performance. A sample was chosen for whom a clear and objective measure of actual performance was available. This was a group of 95 mid-career MBA students, for whom examination scores were obtained. A questionnaire containing items assessing burnout and perceived performance was administered. There were significant positive correlations between the measure of burnout and the items assessing perceived performance, particularly those reflecting self-esteem. There was no significant association between the measure of burnout and actual performance. It is suggested that (i) a clear distinction between actual and perceived performance be made in the burnout Literature; (ii) it is premature to assume burnout will lead to an actual decline in ~erformance;(iii) the notion of deteriorating ~erceivedperformance in relation to burnout may have more to do with general feelings of self-esteem than performance or accomplishment per se.
Burnout is a phenomenon that has gained attention and empirical focus as a form of psychological strain. I t may be understood as a type of psychological distress arising from over-extension of the self that manifests as a severe loss of energy that cannot be renewed. A number of conceptualisations of the phenomenon have been made (Freudenberger, 1980; Cherniss, 1980; Perlrnan & Hartman, 1982; Maslach, 1982a; Meier, 1983; Shirom, 1989). The framework adopted here follows that of Shirom (1989) who, in the most recent and comprehensive review of the literature on burnout, stated that "the major conclusion which may be drawn from past validation efforts is that the unique content of burnout has to d o with the depletion of an individual's energetic resources" (p. 331). Like Shirom, I consider there to be sufficient shared meaning unique to the concept which provides it with legitimacy. Maslach (1982b, p. 40) states that "most conceptualisations of burnout have hypothesised a direct link between experienced burnout and a deterioration in the quality of service or care provided . . . this is a very serious consequence of burnout and is probably the bottom line for most institutions." Snibbe, Radcliffe, Weisberger, Richards, and Kelly (1989) also stated in a recent article that one of the consequences of burnout is "productivity problems" (p. 775). However, the exact relationship between burnout and a decline in performance remains ambiguous and tentative. Shirom (1989, p. 36) notes "the sparse research on the relationships of burnout to job performance. I n spite of this, items measuring performance-related behaviour are sometimes incorporated into measures of burnout, such as in Jones's 'Request reprints from G D S Consultants, 40 Roe Lane, London NW7 9BD, England
964
A-M. GARDEN
(1981) Staff Burnout Scale for Health Professionals, or in the personal accomplishment scale of the [Maslach Burnout Inventory]." This paper examines a number of propositions concerning burnout and performance in an attempt to clarify the relationship between the two. -
-
BACKGROUND There are several sources of confusion which require clarification. (a) The first arises from the lack of precision about whether it is perceived or actual deterioration in performance to which reference is made. Indeed, the two meanings are used interchangeably. Some authors refer to both perceived and actual performance as being implicated in burnout. Cherniss (1980), for example, states that burnout "reduces motivation and effectiveness" (p. 9) and includes as a symptom "a sense of failure" (p. 17). "Feelings of reduced personal accomplishment" is specified as the third dimension of burnout by Maslach (1982a) and is measured on the third scale of the Maslach Burnout Inventory (Maslach &Jackson, 1981). I t is assumed to follow sequentially from emotional exhaustion and depersonalisation and is described in terms of "a growing sense of inadequacy about their ability to relate to recipients, and this may result in a self-imposed verdict of failure" (Maslach, 1982a, p. 5 ) . However, for other authors it is actual performance, not perceived performance, that is implicated in burnout. Farber (1983, p. 3) states that, for burned-out professionals, "performance at work deteriorates markedly." Nagi (1985) also asserts that burnout results in "loss of productivity" (p. 195). Carroll and White (1982, p. 41) define burnout as "a construct used to explain observable decrements in the typical quality and quantity of work performed by a person on the job." They state (p. 43) that, of the typical signs of burnout, "first and foremost [is] a significant decrement in the quality of services provided to clients or customers." This definition was derived from observation but has not been empirically supported. Perlman and Hartrnan (1982), in the first review article on burnout, proposed a definition from a content analysis of existing definitions and included "lowered work productivity" as a key component (p. 293). I n contrast to Maslach, they state that research did not support the inclusion of a "negative self-concept" into burnout's definition; for them it is actual performance and not negative feelings about one's performance which is appropriate to include. I n this study, separate measures of both actual and perceived performance were obtained to see if they were differentially related to burnout. (b) The second source of confusion arises when considering the relationship between feelings of reduced personal accomplishment and burnout. For Maslach (1982a), as stated above, the latter is part of the definition and measure of burnout. Other authors have used the Maslach inventory and retained the personal accomplishment scale (notably Golembiewski & Mun-
BURNOUT AND PERFORMANCE
965
zemider, 1983). However, Jayartre and Chess (1983, p. 130), in relating burnout to job strain and satisfaction, use only the emotional exhaustion and depersonalisation subscales of the Maslach Burnout Inventory because these are the "factors that fall strictly within the domain of burnout." Similarly, Gaines and Jermier (1983) ignore the personal accomplishment subscale (as well as the depersonalisation subscale), arguing that it is only emotional exhaustion which is the core aspect of burnout. Brookings, Bolton, Brown, and McEvoy (1985), in an item factor analysis of the Maslach Burnout Inventory items, concluded that "a perceived lack of personal accomplishment in one's work is not a critical feature of the burnout phenomenon" (p. 148). Snibbe, et al. (1989) studied burnout in a Health Maintenance Organisation using Maslach's inventory as the measure of burnout. They found-both emotional exhaustion and depersonalisation in the "high" or "moderate" range and concluded that "moderate burnout exists" (p. 778). Yet, alongside this, their subjects retained high not low levels of feelings of personal accomplishment; "these physicians are exhausted and depersonahse their patients but mysteriously retain a strong sense of professional identification" (p. 778, my emphasis). Shirom (1989), in his review of validation studies, concluded that, in relation to the Maslach Burnout Inventory, only the emotional exhaustion subscale, but neither the depersonalisation nor personal accomplishment subscales, had sufficient validity (pp. 32-34) and that the latter could not be included as part of the core definition of burnout. Construct validation efforts have, on balance, concentrated more on the other two components and, as Maslach and Jackson (1981) state, validation of the personal accomplishment component is weaker than for the other two components (see pp. 7-9). As Jackson, Turner, and Brief (1987) state, this third component "has received little direct attention" (p. 340). Clarification of the relationship between feelings of deterioration in performance and burnout would, then, be useful. I n this study, the association between the measure of burnout and that of perceived performance was examined. (c) The thud main source of confusion arises from the fact that empirical vahdation of the link between burnout and actual job performance is ambiguous. As Lazaro, Shinn, and Robinson (1985) comment, "most theories . . . hypothesise that burnout leads to poor performance, absenteeism and turn-over, but there is little empirical evidence for or against this claim" (p. 213), noting that job performance is rarely measured. As Maslach (1982b, p. 40) states, "In spite of its importance, the quality of staff members' work with clients is almost never assessed. This is largely because of the difficulty in determining just what "good quality" and "poor qualityMare. The problems involved in assessing the quality of job performance are ones that people have struggled with for years." Golembiewslu, Hilles, and Daly
966
A-M. GARDEN
(1987, p. 296) cite a number of "dreary effects" which characterise advanced phases of burnout. Whilst references are cited for most of the effects specified, none is available for the two of interest, viz., "lower performance appraisals" and "generally lower productivity." Studies by Rafferty, Lemkau, Purdy, and Rudisill (1986) and C. Ezrahi (1987) provided limited or no support for the notion that burnout was accompanied by an actual lack of personal accomplishment. However, other studies have provided some support for this notion. One study looked at burnout and performance in college students using measures of memory and learning style. Using two measures of burnout, they found both to be "correlated significantly to the measure of memory performance" (p. 67). However, the positive relationship may have been obtained through confounding occurring between the measures of performance and the measures of burnout. One of the measures of burnout correlated with the measure of performance was the Maslach Burnout Inventory, which includes as one component, as indicated above, feelings about one's performance. Similarly, the second measure of burnout, the Meier Burnout Assessment (Meier, 1983), included items such as "I can d o all the tasks required to be a successful student" and "I have considerable talent for school work" (Meier & Schmeck, 1985, pp. 64-65.). Again, the danger is one of potential confounding of measures. One study used two different measures of performance. Roelens (1983), using a nursing sample, examined the relationship of burnout to job performance and found a significant negative relationship with self-ratings of performance, i.e., feelings about one's performance, but did not find a consistent, significant negative relationship when actual performance was rated by the person's supervisor. Lazaro, Shinn, and Robinson (1985) also attempted to gauge the relationship between burnout and job performance, using both self and supervisor's ratings. They found significant negative correlations for each type of rating, with the correlation for the self-ratings approximately double that for the supervisors' ratings. Regression analysis of job performance on the burnout measures, controlling for individual differences, showed the selfrated performance to be significant, but not the supervisor-rated performance. [These two studies also confirm the importance of using different measures of performance as argued in point (a) above.] Shirom (1989), in his review article, concluded that reduced levels of personal accomplishment and job performance are "additional components that several authors sought to add to this core negative emotional experience," thereby confounding burnout with some of its sequels. I n his view, reduced actual accomplishment (as well as self-perceived accomplishment) should not be treated as part of burnout. - -
BURNOUT AND PERFORMANCE
967
Some authors suggest that feelings of accomplishment might deteriorate but actual accomplishment remains the same. Freudenberger (1980) describes burnout as a process within which the person starts to work harder and for longer hours as performance deteriorates, thereby enabling particular standards actually to be maintained. Pelletier (1984) also describes the syndrome as one of a "lengthening workday to compensate for diminished productivity" (p. 67). I n other words, these authors suggest that individual action may circumvent an actual decline in performance per se, through allocating more time to work and a decline in leisure hours. Over-all, the different findings and perspectives above are difficult to integrate and ambiguous to interpret. This may partly be because research on the issue is scarce. One of the recurring problems is that of the measurement of performance as discussed earlier by Maslach. The task addressed in this paper is to clarify these three areas of confusion, by examining the relationship between burnout and both perceived and actual performance. Since objective measures of actual performance are needed, a sample was chosen where a measure of performance was clear, precise, and objective. This is critical, since "subjective performance ratings are known to suffer from rater biases" (Lazaro, et al., 1985). Data on feelings about one's performance were also obtained.
Sample The population chosen was a group of 136 full-time MBA students, of whom 95 (69%) voluntarily completed a "stress questionnaire." (There was no attrition during the study.) The MBA programme is an intensive 12-month one of four terms. I n the first two terms, all participants take identical courses. I n the third and fourth terms, some variation in courses occurs due to some choice in elective courses, in addition to core courses which were also taken in these terms. The stress questionnaire was administered nine months into the programme, prior to the third-term exams. The questionnaire was distributed through the normal pigeon-hole mailing system and returned directly to this author. Over 200 items were included. The measure of burnout as well as questions concerning their feelings and reactions on the MBA, including feelings about their performance, were contained in this questionnaire. Most were answered on a five-point frequency scale rated from "Never" to "Always.'' Participants were predominantly mid-career managers and engineers. The ages of the sample ranged from 25 to 46 years, with an average of 31. Most of the sample were single (55%), a third were married, the remainder being either divorced or separated. Most (65%) had no children. The major-
968
A-M. GARDEN
ity (85%) were men. They worked an average of 64 hours per week. This sample was chosen because there is a clear, unambiguous, and anonymous rating of performance. Students have been used as subjects in other studies of burnout (Pines, Arenson, & Kafry, 1981; Garden, 1987; Powers & Gose, 1986; Meier & Schmeck, 1985). Whilst they are not part of the human services, it is acknowledged that burnout may be experienced in a wide range of situations. To restrict the study of burnout arbitrarily to only those situations in which the work requires working with other people cannot be justified. Claims for such a restriction seem to have arisen solely because the human services were the site where the phenomenon was first formally differentiated. As has been established elsewhere (Garden, 1985, 1987, 1989), if the initial research site had been a nonhuman services one, many of the implicit assumptions which have become wedded to the conventional understanding of the phenomenon would not hold. Shirom (1989) concluded from reviewing research from the early 70s to the late 80s that studies of burnout "need not and should not be restricted to individuals whose work requires large amounts of contact with people in need of aid . . . Therefore, a theory of burnout must not allow itself to be exclusively concerned with the people occupations" (p. 33, my emphasis).
Measures The measure of burnout used represents the core symptom of burnout, i.e., energy depletion. The dimension of burnout for which there is most definitional agreement is exhaustion or energy depletion (Maslach, 1982b). Most definitions refer to exhaustion or depletion as a key component of burnout or as the sole defining component (Freudenberger, 1980, 1983; Pines, et a / . , 1981; Jackson, Schwab, & Schuler, 1986). As indicated earlier, one study took as the measure of burnout the emotional exhaustion subscale only of the Maslach Burnout Inventory, without using the other two subscales. They refer to "general consensus in the literature that emotional exhaustion is the central or core dimension of burnout," and that it "may be conceptualised as the first stage of burnout" (Gaines & Jerrnier, 1983). Shirom (1989, p. 33) states that "the unique content of burnout had to do with the depletion of an individual's energetic resources . . . This is the dimension of the burnout experience for which there is most support in the validation efforts. This is the dimension identified in longitudinal research on burnout and by most scale constructors, as the predominant component of this concept" (p. 33). He notes that this depletion "denoted now by the term burnout" is distinguishable from other established concepts and, moreover, that "not surprisingly, several researchers already used only this primary dimension of burnout to operationalise it in field research" (p. 33). The second and third components of the Maslach inventory are seen respectively as antecedent coping behaviour or a sequel of the syndrome, both of which have been
969
BURNOUT AND PERFORMANCE
confounded with the phenomenon. Note that the construct or measure of energy depletion used in this study is comparable to the definition of burnout used in other research. The majority of burnout research has both measured and defined burnout in terms of exhaustion or energy depletion. This is typically defined as preceding other manifestations or dimensions of burnout, such as deterioration in performance. In addition to the above arguments, the Maslach Burnout Inventory is not used as a measure due to the difficulties in generalising this instrument to fields outside the human services and its seeming applicability to only certain types of personality and not others, irrespective of occupational setting (Garden, 1987, 1988, 1989). Other studies have also suggested difficulties with this measure (Shirom, 1989; Firth, McIntee, & McKeown, 1985; Meier, 1984; Rafferty, Lemkau, Purdy, & Rudisill, 1986; C. Ezrahi, 1987). The measure of burnout, called the Energy Depletion Index, was similar to that used in previous studies (Garden, 1985, 1987, 1988, 1989). It was obtained from a factor analytic ~rocedureon seven items, each of which related to energy level or exhaustion as it had been described in the burnout literature. Note that the experience of exhaustion specific to burnout would appear to be the difficulty renewing one's energy not simply exhaustion per se. The items reflected thls characteristic. They were rated on a scale from 1 to 5 (Never to Always), assessing the frequency of their occurrence over the previous 9-mo, ~ e r i o d ,thereby attempting to gauge long-term or chronic energy depletion, not acute tiredness. Results of the analysis are set out in Table 1. The criterion for selection of an item is that it load above .5 on one factor and not above .3 on any other factor. Two factors emerged with an TABLE 1 FACTOR LOADINGS FORENERGY DEPLETION FACTOR ( N = 95) Item
Factor 1
Factor 2
Feeling Exhausted Upon Waking .53 -.lo Severe Energy Fluctuations Nonrenewal of Energy* At Limit t Feeling Energetic Enthusiasm (-.27) Boredom *Nomenewd comprises questions gauging whether sleeping, resting or vacation relieved sensations of tiredness. t A composite item consisting of three items reflecting different exhaustion themes. It refers to being unable to perform a range of specified activities, e.g., reading, talking to others.
eigenvalue above 1.0. The first factor accounted for 40% of total variance, and this was taken to represent an energy depletion factor. The second factor contained only one clear item and was, therefore, ignored. The four items in
970
A-M. GARDEN
Factor 1 with significant and clear loadings were added, with equal weighting, to form the Energy Depletion Index. Cronbach alpha was 0.82. The mean of the index was 2.4, SD = .8. The items were also loaded into a separate factor analysis containing, in addition to the above, the items in the Pines, et al. measure which has been criticised for confounding burnout and depression (Y. Ezrahi, 1986; Shirorn, 1989). These additional items included, for example, "feeling depressed," and "feeling trapped." Factor analysis yielded two separate factors. The first contained the items used in the Energy Depletion Index in this study and the second factor contained high-loading items representing the concept of Depression as per the Pines measure (Garden, 1985). Measures of actual performance consisted of an average examination score for each participant for each term. Since burnout was measured towards the end of Term 3, scores for that term, as well as for Term 4, are relevant. Mean scores and standard deviations are in Table 2. Examination scores for each of the four terms of the programme were made available, after agreement with student-representatives. TABLE 2 MEANSAND STANDARD DEWATIONS OF ITEMS( N = 9 5 )
Item M SD Worry~ngAbout One's Performance 2.3 0.7 Forgcr What You Have Been Taught 2.6 1.0 Bran Giving Out on You 2.0 0.9 Not as Quick as Before 2.0 1.0 Making Silly Mistakes 2.8 1.0 Working Harder but Accomplishing Less 2.1 0.8 Finding Fault With Yourself 2.8 0.9 Feeling Helpless 1.6 0.9 Exam Score, Tlud Term 4.6 60.8 Exam Score, Fourth Term 61.1 5.0 Average Weekly Hours Worked 63.6 12.1 10.1 2.3 Average Weekly Leisure Hours *Scale for all items except exam scores and hours is from 1 (Never) to 5 (Always).
Measures of perceived performance were obtained from the stress questionnaire. Means and standard deviations are illustrated in Table 2. Items were worded to reflect aspects of performance appropriate to this context (cf. Golembiewski, et al., 1987). The specific wording of the items was obtained through prior interviews with three of the MBA faculty, as well as three MBA graduates from previous courses. These interviews were designed solely for the purpose of identifying which aspects of experience, including performance, might be relevant to this particular group. Examples are "worrying about one's performance," and "forget what you have been taught." I n corn-
971
BURNOUT AND PERFORMANCE
mon with the Maslach inventory (Maslach &Jackson, 1981) items measuring "finding fault with yourself" and "feeling helpless" are included. I n addition, the average number of hours worked per week for each term was measured, to test whether there is any relationship between the measures of burnout, performance and hours worked. As part of this assessment of time allocation, an assessment of hours spent on leisure activities was also obtained.
(a) Perceived Performance The Pearson correlations between the various items reflecting feelings about one's performance and the measure of burnout are illustrated in Table 3 . All of the items show significant positive correlations. In other words, higher energy depletion was accompanied by more negative feelings about one's own performance. I n particular, "finding fault with yourself" and "feeling helpless" had high positive correlations with the Energy Depletion Index. I t was necessary to control for the effect of other potential explanatory variables. Age, sex, marital status, number of children, education, and allocation of hours did not affect the relationships by more than .02. A question asking respondents to rate the similarity of skills required of the MBA to those they already possessed was also used as a control variable, but, again, there was no effect on the relationships depicted in Table 3. TABLE 3 CORRELATIONS BETWEENENERGY DEPLETION INDEXAND FEELMGSABOUTPERFORMANCE ( N = 95) Worrying About One's Performance Forget What You Have Been Taught Brain Giving Out on You Not as Quick as Before Making Sdly Mistakes Working Harder but Accomplishing Less Finding Fault With Yourseu Feeling Helpless *p
@ Psychological Reports 1991
RELATIONSHIP BETWEEN BURNOUT AND PERFORMANCE ' ANNA-MARIA GARDEN GDS Consultants Summary.-This paper seeks to clarify the nature of the relationship between burnout and both perceived and actual performance. A sample was chosen for whom a clear and objective measure of actual performance was available. This was a group of 95 mid-career MBA students, for whom examination scores were obtained. A questionnaire containing items assessing burnout and perceived performance was administered. There were significant positive correlations between the measure of burnout and the items assessing perceived performance, particularly those reflecting self-esteem. There was no significant association between the measure of burnout and actual performance. It is suggested that (i) a clear distinction between actual and perceived performance be made in the burnout Literature; (ii) it is premature to assume burnout will lead to an actual decline in ~erformance;(iii) the notion of deteriorating ~erceivedperformance in relation to burnout may have more to do with general feelings of self-esteem than performance or accomplishment per se.
Burnout is a phenomenon that has gained attention and empirical focus as a form of psychological strain. I t may be understood as a type of psychological distress arising from over-extension of the self that manifests as a severe loss of energy that cannot be renewed. A number of conceptualisations of the phenomenon have been made (Freudenberger, 1980; Cherniss, 1980; Perlrnan & Hartman, 1982; Maslach, 1982a; Meier, 1983; Shirom, 1989). The framework adopted here follows that of Shirom (1989) who, in the most recent and comprehensive review of the literature on burnout, stated that "the major conclusion which may be drawn from past validation efforts is that the unique content of burnout has to d o with the depletion of an individual's energetic resources" (p. 331). Like Shirom, I consider there to be sufficient shared meaning unique to the concept which provides it with legitimacy. Maslach (1982b, p. 40) states that "most conceptualisations of burnout have hypothesised a direct link between experienced burnout and a deterioration in the quality of service or care provided . . . this is a very serious consequence of burnout and is probably the bottom line for most institutions." Snibbe, Radcliffe, Weisberger, Richards, and Kelly (1989) also stated in a recent article that one of the consequences of burnout is "productivity problems" (p. 775). However, the exact relationship between burnout and a decline in performance remains ambiguous and tentative. Shirom (1989, p. 36) notes "the sparse research on the relationships of burnout to job performance. I n spite of this, items measuring performance-related behaviour are sometimes incorporated into measures of burnout, such as in Jones's 'Request reprints from G D S Consultants, 40 Roe Lane, London NW7 9BD, England
964
A-M. GARDEN
(1981) Staff Burnout Scale for Health Professionals, or in the personal accomplishment scale of the [Maslach Burnout Inventory]." This paper examines a number of propositions concerning burnout and performance in an attempt to clarify the relationship between the two. -
-
BACKGROUND There are several sources of confusion which require clarification. (a) The first arises from the lack of precision about whether it is perceived or actual deterioration in performance to which reference is made. Indeed, the two meanings are used interchangeably. Some authors refer to both perceived and actual performance as being implicated in burnout. Cherniss (1980), for example, states that burnout "reduces motivation and effectiveness" (p. 9) and includes as a symptom "a sense of failure" (p. 17). "Feelings of reduced personal accomplishment" is specified as the third dimension of burnout by Maslach (1982a) and is measured on the third scale of the Maslach Burnout Inventory (Maslach &Jackson, 1981). I t is assumed to follow sequentially from emotional exhaustion and depersonalisation and is described in terms of "a growing sense of inadequacy about their ability to relate to recipients, and this may result in a self-imposed verdict of failure" (Maslach, 1982a, p. 5 ) . However, for other authors it is actual performance, not perceived performance, that is implicated in burnout. Farber (1983, p. 3) states that, for burned-out professionals, "performance at work deteriorates markedly." Nagi (1985) also asserts that burnout results in "loss of productivity" (p. 195). Carroll and White (1982, p. 41) define burnout as "a construct used to explain observable decrements in the typical quality and quantity of work performed by a person on the job." They state (p. 43) that, of the typical signs of burnout, "first and foremost [is] a significant decrement in the quality of services provided to clients or customers." This definition was derived from observation but has not been empirically supported. Perlman and Hartrnan (1982), in the first review article on burnout, proposed a definition from a content analysis of existing definitions and included "lowered work productivity" as a key component (p. 293). I n contrast to Maslach, they state that research did not support the inclusion of a "negative self-concept" into burnout's definition; for them it is actual performance and not negative feelings about one's performance which is appropriate to include. I n this study, separate measures of both actual and perceived performance were obtained to see if they were differentially related to burnout. (b) The second source of confusion arises when considering the relationship between feelings of reduced personal accomplishment and burnout. For Maslach (1982a), as stated above, the latter is part of the definition and measure of burnout. Other authors have used the Maslach inventory and retained the personal accomplishment scale (notably Golembiewski & Mun-
BURNOUT AND PERFORMANCE
965
zemider, 1983). However, Jayartre and Chess (1983, p. 130), in relating burnout to job strain and satisfaction, use only the emotional exhaustion and depersonalisation subscales of the Maslach Burnout Inventory because these are the "factors that fall strictly within the domain of burnout." Similarly, Gaines and Jermier (1983) ignore the personal accomplishment subscale (as well as the depersonalisation subscale), arguing that it is only emotional exhaustion which is the core aspect of burnout. Brookings, Bolton, Brown, and McEvoy (1985), in an item factor analysis of the Maslach Burnout Inventory items, concluded that "a perceived lack of personal accomplishment in one's work is not a critical feature of the burnout phenomenon" (p. 148). Snibbe, et al. (1989) studied burnout in a Health Maintenance Organisation using Maslach's inventory as the measure of burnout. They found-both emotional exhaustion and depersonalisation in the "high" or "moderate" range and concluded that "moderate burnout exists" (p. 778). Yet, alongside this, their subjects retained high not low levels of feelings of personal accomplishment; "these physicians are exhausted and depersonahse their patients but mysteriously retain a strong sense of professional identification" (p. 778, my emphasis). Shirom (1989), in his review of validation studies, concluded that, in relation to the Maslach Burnout Inventory, only the emotional exhaustion subscale, but neither the depersonalisation nor personal accomplishment subscales, had sufficient validity (pp. 32-34) and that the latter could not be included as part of the core definition of burnout. Construct validation efforts have, on balance, concentrated more on the other two components and, as Maslach and Jackson (1981) state, validation of the personal accomplishment component is weaker than for the other two components (see pp. 7-9). As Jackson, Turner, and Brief (1987) state, this third component "has received little direct attention" (p. 340). Clarification of the relationship between feelings of deterioration in performance and burnout would, then, be useful. I n this study, the association between the measure of burnout and that of perceived performance was examined. (c) The thud main source of confusion arises from the fact that empirical vahdation of the link between burnout and actual job performance is ambiguous. As Lazaro, Shinn, and Robinson (1985) comment, "most theories . . . hypothesise that burnout leads to poor performance, absenteeism and turn-over, but there is little empirical evidence for or against this claim" (p. 213), noting that job performance is rarely measured. As Maslach (1982b, p. 40) states, "In spite of its importance, the quality of staff members' work with clients is almost never assessed. This is largely because of the difficulty in determining just what "good quality" and "poor qualityMare. The problems involved in assessing the quality of job performance are ones that people have struggled with for years." Golembiewslu, Hilles, and Daly
966
A-M. GARDEN
(1987, p. 296) cite a number of "dreary effects" which characterise advanced phases of burnout. Whilst references are cited for most of the effects specified, none is available for the two of interest, viz., "lower performance appraisals" and "generally lower productivity." Studies by Rafferty, Lemkau, Purdy, and Rudisill (1986) and C. Ezrahi (1987) provided limited or no support for the notion that burnout was accompanied by an actual lack of personal accomplishment. However, other studies have provided some support for this notion. One study looked at burnout and performance in college students using measures of memory and learning style. Using two measures of burnout, they found both to be "correlated significantly to the measure of memory performance" (p. 67). However, the positive relationship may have been obtained through confounding occurring between the measures of performance and the measures of burnout. One of the measures of burnout correlated with the measure of performance was the Maslach Burnout Inventory, which includes as one component, as indicated above, feelings about one's performance. Similarly, the second measure of burnout, the Meier Burnout Assessment (Meier, 1983), included items such as "I can d o all the tasks required to be a successful student" and "I have considerable talent for school work" (Meier & Schmeck, 1985, pp. 64-65.). Again, the danger is one of potential confounding of measures. One study used two different measures of performance. Roelens (1983), using a nursing sample, examined the relationship of burnout to job performance and found a significant negative relationship with self-ratings of performance, i.e., feelings about one's performance, but did not find a consistent, significant negative relationship when actual performance was rated by the person's supervisor. Lazaro, Shinn, and Robinson (1985) also attempted to gauge the relationship between burnout and job performance, using both self and supervisor's ratings. They found significant negative correlations for each type of rating, with the correlation for the self-ratings approximately double that for the supervisors' ratings. Regression analysis of job performance on the burnout measures, controlling for individual differences, showed the selfrated performance to be significant, but not the supervisor-rated performance. [These two studies also confirm the importance of using different measures of performance as argued in point (a) above.] Shirom (1989), in his review article, concluded that reduced levels of personal accomplishment and job performance are "additional components that several authors sought to add to this core negative emotional experience," thereby confounding burnout with some of its sequels. I n his view, reduced actual accomplishment (as well as self-perceived accomplishment) should not be treated as part of burnout. - -
BURNOUT AND PERFORMANCE
967
Some authors suggest that feelings of accomplishment might deteriorate but actual accomplishment remains the same. Freudenberger (1980) describes burnout as a process within which the person starts to work harder and for longer hours as performance deteriorates, thereby enabling particular standards actually to be maintained. Pelletier (1984) also describes the syndrome as one of a "lengthening workday to compensate for diminished productivity" (p. 67). I n other words, these authors suggest that individual action may circumvent an actual decline in performance per se, through allocating more time to work and a decline in leisure hours. Over-all, the different findings and perspectives above are difficult to integrate and ambiguous to interpret. This may partly be because research on the issue is scarce. One of the recurring problems is that of the measurement of performance as discussed earlier by Maslach. The task addressed in this paper is to clarify these three areas of confusion, by examining the relationship between burnout and both perceived and actual performance. Since objective measures of actual performance are needed, a sample was chosen where a measure of performance was clear, precise, and objective. This is critical, since "subjective performance ratings are known to suffer from rater biases" (Lazaro, et al., 1985). Data on feelings about one's performance were also obtained.
Sample The population chosen was a group of 136 full-time MBA students, of whom 95 (69%) voluntarily completed a "stress questionnaire." (There was no attrition during the study.) The MBA programme is an intensive 12-month one of four terms. I n the first two terms, all participants take identical courses. I n the third and fourth terms, some variation in courses occurs due to some choice in elective courses, in addition to core courses which were also taken in these terms. The stress questionnaire was administered nine months into the programme, prior to the third-term exams. The questionnaire was distributed through the normal pigeon-hole mailing system and returned directly to this author. Over 200 items were included. The measure of burnout as well as questions concerning their feelings and reactions on the MBA, including feelings about their performance, were contained in this questionnaire. Most were answered on a five-point frequency scale rated from "Never" to "Always.'' Participants were predominantly mid-career managers and engineers. The ages of the sample ranged from 25 to 46 years, with an average of 31. Most of the sample were single (55%), a third were married, the remainder being either divorced or separated. Most (65%) had no children. The major-
968
A-M. GARDEN
ity (85%) were men. They worked an average of 64 hours per week. This sample was chosen because there is a clear, unambiguous, and anonymous rating of performance. Students have been used as subjects in other studies of burnout (Pines, Arenson, & Kafry, 1981; Garden, 1987; Powers & Gose, 1986; Meier & Schmeck, 1985). Whilst they are not part of the human services, it is acknowledged that burnout may be experienced in a wide range of situations. To restrict the study of burnout arbitrarily to only those situations in which the work requires working with other people cannot be justified. Claims for such a restriction seem to have arisen solely because the human services were the site where the phenomenon was first formally differentiated. As has been established elsewhere (Garden, 1985, 1987, 1989), if the initial research site had been a nonhuman services one, many of the implicit assumptions which have become wedded to the conventional understanding of the phenomenon would not hold. Shirom (1989) concluded from reviewing research from the early 70s to the late 80s that studies of burnout "need not and should not be restricted to individuals whose work requires large amounts of contact with people in need of aid . . . Therefore, a theory of burnout must not allow itself to be exclusively concerned with the people occupations" (p. 33, my emphasis).
Measures The measure of burnout used represents the core symptom of burnout, i.e., energy depletion. The dimension of burnout for which there is most definitional agreement is exhaustion or energy depletion (Maslach, 1982b). Most definitions refer to exhaustion or depletion as a key component of burnout or as the sole defining component (Freudenberger, 1980, 1983; Pines, et a / . , 1981; Jackson, Schwab, & Schuler, 1986). As indicated earlier, one study took as the measure of burnout the emotional exhaustion subscale only of the Maslach Burnout Inventory, without using the other two subscales. They refer to "general consensus in the literature that emotional exhaustion is the central or core dimension of burnout," and that it "may be conceptualised as the first stage of burnout" (Gaines & Jerrnier, 1983). Shirom (1989, p. 33) states that "the unique content of burnout had to do with the depletion of an individual's energetic resources . . . This is the dimension of the burnout experience for which there is most support in the validation efforts. This is the dimension identified in longitudinal research on burnout and by most scale constructors, as the predominant component of this concept" (p. 33). He notes that this depletion "denoted now by the term burnout" is distinguishable from other established concepts and, moreover, that "not surprisingly, several researchers already used only this primary dimension of burnout to operationalise it in field research" (p. 33). The second and third components of the Maslach inventory are seen respectively as antecedent coping behaviour or a sequel of the syndrome, both of which have been
969
BURNOUT AND PERFORMANCE
confounded with the phenomenon. Note that the construct or measure of energy depletion used in this study is comparable to the definition of burnout used in other research. The majority of burnout research has both measured and defined burnout in terms of exhaustion or energy depletion. This is typically defined as preceding other manifestations or dimensions of burnout, such as deterioration in performance. In addition to the above arguments, the Maslach Burnout Inventory is not used as a measure due to the difficulties in generalising this instrument to fields outside the human services and its seeming applicability to only certain types of personality and not others, irrespective of occupational setting (Garden, 1987, 1988, 1989). Other studies have also suggested difficulties with this measure (Shirom, 1989; Firth, McIntee, & McKeown, 1985; Meier, 1984; Rafferty, Lemkau, Purdy, & Rudisill, 1986; C. Ezrahi, 1987). The measure of burnout, called the Energy Depletion Index, was similar to that used in previous studies (Garden, 1985, 1987, 1988, 1989). It was obtained from a factor analytic ~rocedureon seven items, each of which related to energy level or exhaustion as it had been described in the burnout literature. Note that the experience of exhaustion specific to burnout would appear to be the difficulty renewing one's energy not simply exhaustion per se. The items reflected thls characteristic. They were rated on a scale from 1 to 5 (Never to Always), assessing the frequency of their occurrence over the previous 9-mo, ~ e r i o d ,thereby attempting to gauge long-term or chronic energy depletion, not acute tiredness. Results of the analysis are set out in Table 1. The criterion for selection of an item is that it load above .5 on one factor and not above .3 on any other factor. Two factors emerged with an TABLE 1 FACTOR LOADINGS FORENERGY DEPLETION FACTOR ( N = 95) Item
Factor 1
Factor 2
Feeling Exhausted Upon Waking .53 -.lo Severe Energy Fluctuations Nonrenewal of Energy* At Limit t Feeling Energetic Enthusiasm (-.27) Boredom *Nomenewd comprises questions gauging whether sleeping, resting or vacation relieved sensations of tiredness. t A composite item consisting of three items reflecting different exhaustion themes. It refers to being unable to perform a range of specified activities, e.g., reading, talking to others.
eigenvalue above 1.0. The first factor accounted for 40% of total variance, and this was taken to represent an energy depletion factor. The second factor contained only one clear item and was, therefore, ignored. The four items in
970
A-M. GARDEN
Factor 1 with significant and clear loadings were added, with equal weighting, to form the Energy Depletion Index. Cronbach alpha was 0.82. The mean of the index was 2.4, SD = .8. The items were also loaded into a separate factor analysis containing, in addition to the above, the items in the Pines, et al. measure which has been criticised for confounding burnout and depression (Y. Ezrahi, 1986; Shirorn, 1989). These additional items included, for example, "feeling depressed," and "feeling trapped." Factor analysis yielded two separate factors. The first contained the items used in the Energy Depletion Index in this study and the second factor contained high-loading items representing the concept of Depression as per the Pines measure (Garden, 1985). Measures of actual performance consisted of an average examination score for each participant for each term. Since burnout was measured towards the end of Term 3, scores for that term, as well as for Term 4, are relevant. Mean scores and standard deviations are in Table 2. Examination scores for each of the four terms of the programme were made available, after agreement with student-representatives. TABLE 2 MEANSAND STANDARD DEWATIONS OF ITEMS( N = 9 5 )
Item M SD Worry~ngAbout One's Performance 2.3 0.7 Forgcr What You Have Been Taught 2.6 1.0 Bran Giving Out on You 2.0 0.9 Not as Quick as Before 2.0 1.0 Making Silly Mistakes 2.8 1.0 Working Harder but Accomplishing Less 2.1 0.8 Finding Fault With Yourself 2.8 0.9 Feeling Helpless 1.6 0.9 Exam Score, Tlud Term 4.6 60.8 Exam Score, Fourth Term 61.1 5.0 Average Weekly Hours Worked 63.6 12.1 10.1 2.3 Average Weekly Leisure Hours *Scale for all items except exam scores and hours is from 1 (Never) to 5 (Always).
Measures of perceived performance were obtained from the stress questionnaire. Means and standard deviations are illustrated in Table 2. Items were worded to reflect aspects of performance appropriate to this context (cf. Golembiewski, et al., 1987). The specific wording of the items was obtained through prior interviews with three of the MBA faculty, as well as three MBA graduates from previous courses. These interviews were designed solely for the purpose of identifying which aspects of experience, including performance, might be relevant to this particular group. Examples are "worrying about one's performance," and "forget what you have been taught." I n corn-
971
BURNOUT AND PERFORMANCE
mon with the Maslach inventory (Maslach &Jackson, 1981) items measuring "finding fault with yourself" and "feeling helpless" are included. I n addition, the average number of hours worked per week for each term was measured, to test whether there is any relationship between the measures of burnout, performance and hours worked. As part of this assessment of time allocation, an assessment of hours spent on leisure activities was also obtained.
(a) Perceived Performance The Pearson correlations between the various items reflecting feelings about one's performance and the measure of burnout are illustrated in Table 3 . All of the items show significant positive correlations. In other words, higher energy depletion was accompanied by more negative feelings about one's own performance. I n particular, "finding fault with yourself" and "feeling helpless" had high positive correlations with the Energy Depletion Index. I t was necessary to control for the effect of other potential explanatory variables. Age, sex, marital status, number of children, education, and allocation of hours did not affect the relationships by more than .02. A question asking respondents to rate the similarity of skills required of the MBA to those they already possessed was also used as a control variable, but, again, there was no effect on the relationships depicted in Table 3. TABLE 3 CORRELATIONS BETWEENENERGY DEPLETION INDEXAND FEELMGSABOUTPERFORMANCE ( N = 95) Worrying About One's Performance Forget What You Have Been Taught Brain Giving Out on You Not as Quick as Before Making Sdly Mistakes Working Harder but Accomplishing Less Finding Fault With Yourseu Feeling Helpless *p
