ChapterPDF Available

Areas of Worklife: A Structured Approach to Organizational Predictors of Job Burnout

Authors:

Abstract and Figures

This chapter evaluates a model of the organizational context of burnout with direct reference to a new measure, the Areas of Worklife Scale (AWS). The model proposes a structured framework for considering six areas of worklife – workload, control, reward, community, fairness, and values – that have resonated through the literature on burnout over the previous two decades. The chapter presents extensive data on the AWS, testing a model of the six areas’ interrelationships as well as their overall relationship to the three aspects of burnout. The results of these analyses are discussed in reference to the psychometric qualities of the measure and the implications of a structured approach to work environments for future development of research on burnout. Implications for developing workplace interventions are also considered.
Content may be subject to copyright.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
AREAS OF WORKLIFE:
A STRUCTURED APPROACH TO
ORGANIZATIONAL PREDICTORS
OF JOB BURNOUT
Michael P. Leiter and Christina Maslach
ABSTRACT
This chapter evaluates a model of the organizational context of burnout with
direct reference to a new measure, the Areas of Worklife Scale (AWS). The
model proposes a structured framework for considering six areas of worklife
workload, control, reward, community, fairness, and values that have res-
onated through the literature on burnout over the previous two decades. The
chapter presents extensive data on the AWS, testing a model of the six areas’
interrelationships as well as their overall relationship to the three aspects
of burnout. The results of these analyses are discussed in reference to the
psychometric qualities of the measure and the implications of a structured ap-
proach to work environments for future development of research on burnout.
Implications for developing workplace interventions are also considered.
INTRODUCTION
For several decades, the term “burnout” has been used to describe a fundamental
disconnect between the worker and the workplace. The basic story goes like this:
Emotional and Physiological Processes and Positive Intervention Strategies
Research in Occupational Stress and Well Being, Volume 3, 91–134
Copyright © 2003 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1479-3555/doi:10.1016/S1479-3555(03)03003-8
91
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
92 MICHAEL P. LEITER AND CHRISTINA MASLACH
the worker entered a job with positive expectations, enthusiasm, and the goal to
be successful in the job. Over time, things changed and now the worker has an
overwhelming exhaustion; feelings of frustration, anger and cynicism; and a sense
of ineffectiveness and failure. The initial flame has burned out. The experience
impairs both personal and social functioning on the job, and thus carries some
real costs for the individual worker, the people affected by him or her, and for
the organization as a whole. While some people may quit the job as a result of
burnout, others will stay on but will only do the bare minimum rather than their
very best.
Burnout was recognized as an important social problem by practitioners long
before it became a focus of systematic study by researchers. Thus, it was more of
a “grass-roots” phenomenon, grounded in the realities of people’s experiences in
the workplace, rather than a topic derived from a scholarly theory and empirical
studies. This pragmatic conceptual framework of a social problem that needed
to be solved shaped the trajectory of the research on burnout. The early studies
followed a “bottom-up” approach of describing and defining the phenomenon,
and developing hypotheses about its causes and its effects. Later, this initial work
was linked to a wide variety of theoretical perspectives and research literatures in
social, clinical, and industrial/organizational psychology (see reviews by Maslach
& Schaufeli, 1993; Schaufeli & Enzmann, 1998).
An underlying theme of this pragmatic framework has been to discover
solutions to the problem of burnout. From the beginning, the growing research
literature was matched (or even outstripped) by a parallel literature of workshop
and self-help materials. As burnout became more clearly identified as a form
of job stress, it received increasing attention from administrators and policy
makers in the workplace. It is thus fair to say that the field of job burnout
has always had a primary thrust toward application, in addition to scholarly
contributions.
Our recent work has been explicitly designed to bridge the gap between basic
and applied research on burnout. Our goal has been to design tools that can be used
by both researchers and practitioners the former to study hypotheses within the
context of field studies, and the latter to assess the workplace within the context
of organizational interventions. Toward that end, we have developed a new model
that draws on the extant research literature on job stress and proposes that six
areas of job-person mismatch are the critical sources of burnout (Maslach &
Leiter, 1997). We have now developed a new tool to assess these six areas, which
can be used as part of a program of organizational assessment and intervention
(Leiter & Maslach, 2000). This chapter will provide a comprehensive analysis of
our model and measures, and will demonstrate how we are using this approach
both for empirical tests and applied interventions.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 93
BURNOUT AND ENGAGEMENT
Burnout is defined as a psychological syndrome of exhaustion, cynicism, and
inefficacy, whichisexperiencedin response to chronic job stressors. This definition
is a broader statement of the multidimensional model that has been predominant in
the burnout field (Maslach, 1993, 1998; Maslach & Jackson, 1981). The original
model emerged from research with workers in human service and educational
occupations, and thus was framed in terms of the interpersonal relationships that
characterize such jobs. However, more recent work has established that the basic
model can be broadened to apply to any kind of occupation (Leiter & Schaufeli,
1996; Maslach et al., 1996).
Of the three dimensions of burnout, the exhaustion component represents the
basic individual stress experience. It refers to feelings of being overextended and
depleted of one’s emotional and physical resources. The cynicism component
represents the interpersonal context dimension of burnout. It refers to a negative,
callous, or excessively detached response to various aspects of the job. It usually
develops in response to the overload of exhaustion, and is self-protective at first
an emotional buffer of “detached concern. But the risk is that the detachment can
result in the loss of idealism and the dehumanization of others. This detachment, or
distancing, is such an immediate reaction to exhaustion that a strong relationship
from exhaustion to cynicism is found consistently in burnout research, across a
wide range of organizational and occupational settings (Maslach et al., 1996). The
third component of inefficacy represents the self-evaluation dimension of burnout.
It refers to feelings of incompetence and a lack of achievement and productivity
in work. In some instances, it appears to be a function, to some degree, of either
exhaustion or cynicism, or a combination of the two (Byrne, 1993; Lee & Ashforth,
1996). A work situation with chronic, overwhelming demands that contribute to
exhaustion or cynicism is likely to erode one’s sense of effectiveness. However,
in other job contexts, inefficacy appears to develop in parallel with the other two
burnout aspects, rather than sequentially (Leiter, 1993). Here the lack of efficacy
seems to arise more clearly from a lack of relevant resources, while exhaustion
and cynicism emerge from the presence of work overload and social conflict.
Unlike acute stress reactions, which develop in response to specific critical
incidents, burnout is a cumulative reaction to ongoing occupational stressors.
With burnout, the emphasis has been more on the process of psychological
erosion, and the psychological and social outcomes of this chronic exposure,
rather than just the physical ones. Because burnout is a prolonged response to
chronic interpersonal stressors on the job, it tends to be fairly stable over time.
Burnout is one end of a continuum in the relationship people establish with
their jobs. As a syndrome of exhaustion, cynicism, and inefficacy, it stands in
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
94 MICHAEL P. LEITER AND CHRISTINA MASLACH
contrast to the energetic, involved, and effective state of engagement with work.
Recently, the multidimensional model of burnout has been expanded to this other
end of the continuum (Leiter & Maslach, 1998). Engagement is defined in terms
of the same three dimensions as burnout, but the positive end of those dimensions
rather than the negative. Thus, engagement consists of a state of high energy
(rather than exhaustion), strong involvement (rather than cynicism), and a sense
of efficacy (rather than inefficacy).
Engagement is distinct from established constructs in organizational psychol-
ogy such as organizational commitment, job satisfaction, or job involvement.
Organizational commitment refers to an employee’s allegiance to the organization
that provides employment. The focus is on the organization, whereas engagement
focuses on the work itself. Job satisfaction is the extent to which work is a source of
need fulfillment and contentment, or a means of freeing employees from hassles or
dissatisfiers; it does not encompass the person’s relationship with the work itself.
Job involvement is similar to the involvement aspect of engagement with work,
but does not include the energy and effectiveness dimensions. Thus, engagement
provides a more complex and thorough perspective on an individual’s relationship
with work.
In terms of application, the concept of engagement may be more functional
than burnout. A worksetting that is designed to support the positive development
of the three core qualities of energy, involvement, and effectiveness should be
successful in promoting the well-being and productivity of its employees. Thus,
we have found that a focus on what would promote engagement in the workplace
is a better framework for developing effective interventions than a focus simply
on what would reduce stress. Moreover, the former is more likely to change the
job context, while the latter leads to strategies of changing the person.
The Organizational Context for Burnout and Engagement
Job stress has been recognized as a significant occupational hazard, which can
impair both health and work performance (e.g. Sauter & Murphy, 1995). The
worker’s internal experience of stress is assumed to play a mediating role between
the impact of external job demands (stressors) and work-related outcomes (such
as absenteeism or illness). This basic mediation model should be especially
true of the stress phenomenon of burnout, which involves a prolonged response
to chronic interpersonal job stressors. Thus, organizational conditions should
influence a worker’s experience of burnout or engagement, which in turn will
determine outcomes of importance to both the worker and the organization. For
example, assessments of employees’ level of experienced burnout or engagement
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 95
have predicted clients’ evaluation of service quality (Leiter et al., 1998) and
employees’ evaluation of organizational change (Leiter & Harvie, 1998).
Two decades of research on burnout have identified a plethora of organizational
risk factors across many occupations in various countries, as well as some
work-related outcomes (see Maslach et al., 2001; Schaufeli & Enzmann, 1998).
However, there has not been much research that directly tested the mediation
model by including measures of all three model components: organizational fac-
tors, experienced burnout, and work-related outcomes. One of our recent studies
was designed as a first approximation of such a test (Leiter & Maslach, 2003).
In this chapter, we will present not only this initial study on the mediation
model, but the psychometric research that led to the development of a key measure
of organizational factors, the Areas of Worklife Scale. Our goal has been to
develop research tools that are also appropriate for use in applied settings, and this
requires measures that are relatively brief and easily accessible to a wide range
of employees. The standard measure of burnout, the Maslach Burnout Inventory
(Maslach et al., 1996), already meets those criteria. However, there was not a
comparable tool that assesses the multiple job stressors that contribute to burnout,
so our challenge was to devise a measure of these organizational factors.
Six Areas of Worklife
In reviewing the proliferation of organizational correlates in many studies of
burnout and job stress, we had identified six key domains: workload, control, re-
ward, community, fairness, and values (Leiter & Maslach, 1999; Maslach & Leiter,
1997, 1999). The first two areas are reflected in the Demand-Control model of job
stress (Karasek & Theorell, 1990), and reward refers to the power of reinforce-
ments to shape behavior. Community captures all of the work on social support
and interpersonal conflict, while fairness emerges from the literature on equity and
social justice. Finally, the area of values picks up the cognitive-emotional power
of job goals and expectations.
Workload
The most obvious, and most commonly discussed area of worklife is overload: job
demands exceeding human limits. People have to do too much in too little time
with too few resources. Increasing workload has a consistent relationship with
burnout, especially with the exhaustion dimension (Cordes & Dougherty, 1993;
Maslach et al., 2001; Schaufeli & Enzmann, 1998). Structural models of burnout
have shown that exhaustion then mediates the relationship of workload with the
other two dimensions of burnout (Lee & Ashforth, 1996; Leiter & Harvie, 1998).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
96 MICHAEL P. LEITER AND CHRISTINA MASLACH
This association reflects the relationship of work demands with occupational stress
in the stress and coping literature (Cox et al., 1993).
Both qualitative and quantitative work overload contribute to exhaustion by
depleting the capacity of people to meet the demands of the job. The critical point
occurs when people are unable to recover from work demands. That is, acute
fatigue resulting from an especially demanding event at work meeting a deadline
or addressing a crisis need not lead to burnout if people have an opportunity to
recover during restful periods at work or at home (Shinn et al., 1984). When this
kind of overload is a chronic job condition, not an occasional emergency, there is
little opportunity to rest, recover, and restore balance. Such exhaustion can lead to
a deterioration in the quality of the work and a disruption of collegial relationships.
A sustainable workload, in contrast, provides opportunities to use and refine
existing skills as well as to become effective in new areas of activity (Landsbergis,
1988). It builds involvement by opening new opportunities, and by removing
concern about work overwhelming personal capacity. A sustainable workload
stops the cycle of exhaustion that is a driving force in the experience of burnout
for many people.
Control
The Demand-Control theory of job stress (Karasek & Theorell, 1990) has made
the case for the enabling role of control. This area includes employees’ perceived
capacity to influence decisions that affect their work, to exercise professional
autonomy, and to gain access to the resources necessary to do an effective job. As
human beings, people have the ability to think and solve problems, and want to
have the opportunity to make choices and decisions. In other words, they want to
have some input into the process of achieving the outcomes for which they will
be held accountable. Control problems occur when workers have insufficient
authority over their work or are unable to shape the work environment to be
consistent with their values. A sense of efficacy is unlikely to occur when workers
are feeling buffeted by circumstances or powerful people within the organization.
A major control problem occurs when people experience role conflict. Many
burnout studies have found that greater role conflict is strongly and positively
associated with greater exhaustion (Cordes & Dougherty, 1993; Maslach et al.,
1996). Role conflict arises from multiple authorities with conflicting demands
or incongruent values, and people in this situation cannot exercise effective
control in their job. Contradictory demands interfere with their capacity to set
priorities or to commit themselves fully to their work. Role conflict is not simply
an indicator of additional work demands, but is emotionally exhausting in itself
(e.g. Siefert et al., 1991; Starnaman & Miller, 1992). Moreover, role conflict is,
almost by definition, a direct signal of an authority problem at work. It means that
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 97
a worker’s preferred role is out of sync with important qualities of the job, such
as supervisors’ expectations, client demands, or ethical constraints. The critical
issue is not the amount or even the type of work demands, but the consistency of
those demands with the capacity to determine the job.
Studies that examine role conflict usually also consider role ambiguity the
absence of direction in work. Generally, role ambiguity is associated with greater
burnout, but the relationship is not nearly as consistent as that of role conflict
(Cordes & Dougherty, 1993; Maslach et al., 1996). Ambiguity may enhance some
work contexts by providing the freedom to pursue one’s values, while conflict
directly inhibits a course of action.
When people have more control in their work, their actions are more freely
chosen and this can lead to greater satisfaction with the job, and more commit-
ment to it. The process of making a decision has an enduring impact on employees’
experience of participating in organizational life and the responsibility they take
for its outcomes. Participative decision making is a cornerstone of job enrichment
strategies (Hackman, 1986) as much because of its power to engender commitment
as for its capacity to make good use of knowledge and experience within a group
of colleagues. Active participation in organizational decision making has been
consistently found to be associated with higher levels of efficacy and lower levels
of exhaustion (Cherniss, 1980; Lee & Ashforth, 1993; Leiter, 1992).
Reward
The reward area of worklife addresses the extent to which rewards monetary,
social, and intrinsic are consistent with expectations. Lack of recognition from
service recipients, colleagues, managers, and external stakeholders devalues both
the work and the workers, and is closely associated with feelings of inefficacy
(Cordes & Dougherty, 1993; Maslach et al., 1996). When people feel neglected
by the material and social reward system of an organization, they feel out of sync
with its values.
In contrast, consistency in the reward dimension between the person and the
job means that there are both material rewards and opportunities for intrinsic
satisfaction (Richardsen et al., 1992). Intrinsic rewards (such as pride in doing
something of importance and doing it well) can be just as critical as extrinsic
rewards, if not more so. What keeps work involving for most people is the
pleasure and satisfaction they experience with the day-to-day flow of work that
is going well (Leiter, 1992). An enjoyable workflow supports both psychological
well being and physical health, and is also the source of recognition from others.
The results of various studies have shown that insufficient reward (whether
financial, institutional, or social) increases people’s vulnerability to burnout (e.g.
Chappell & Novak, 1992; Glicken, 1983; Maslanka, 1996; Siefert et al., 1991).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
98 MICHAEL P. LEITER AND CHRISTINA MASLACH
Community
Community is the overall quality of social interaction at work, including issues
of conflict, mutual support, closeness, and the capacity to work as a team. People
thrive in community and function best when they share praise, comfort, happiness,
and humor with people they like and respect. In addition to emotional exchange
and instrumental assistance, this kind of social support reaffirms a person’s mem-
bership in a group with a shared sense of values. Unfortunately, some jobs isolate
people from each other, or make social contact impersonal. However, what is most
destructive of community is chronic and unresolved conflict with others on the
job. Such conflict produces constant negative feelings of frustration and hostility,
and reduces the likelihood of social support.
Burnout research has focused primarily on social support from supervisors,
coworkers, and family members (Cordes & Dougherty, 1993; Greenglass et al.,
1994; Greenglass et al., 1988; Maslach et al., 1996). Distinct patterns have been
found for informal coworker support and supervisor support (Jackson et al.,
1986; Leiter & Maslach, 1988). Supervisor support has been more consistently
associated with exhaustion, reflecting the supervisors’ impact on staff members’
workload. Coworker support is more closely related to accomplishment or
efficacy, reflecting the value staff members put on the expert evaluation by their
peers. A sense of community has been found to buffer the impact of feelings of
inequity at work (Truchot & Deregard, 2001). Regardless of its specific form,
social support has been found to be associated with greater engagement (Leiter &
Maslach, 1988; Schnorpfeil et al., 2002).
Research on the social context of burnout has also attended to the broader issues
associated with a sense of community in an organization (Drory & Shamir, 1988;
Farber, 1984; Royal & Rossi, 1996). Research on community orientation (Buunk
& Schaufeli, 1993) provides a distinct but consistent perspective. Both of these
approaches consider ways in which the overall quality of personal interactions
among people in an organization have an impact on the relationships people
have with their work. The consistent finding through this research is that a lively,
attentive, responsive community is incompatible with burnout. People’s subjective
appraisal of their social context their sense of community with colleagues or
their communal orientation towards service recipients reflects the extent to
which the organizational community is consistent with their expectations.
Fairness
Fairness is the extent to which decisions at work are perceived as being fair and
people are treated with respect. Fairness communicates respect and confirms peo-
ple’s self-worth. Mutual respect between people is central to a shared sense of
community. Unfairness can occur when there is inequity of workload or pay, or
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 99
when there is cheating, or when evaluations and promotions are handled inappro-
priately. If procedures for grievance or dispute resolution do not allow for both
parties to have voice, then those will be judged as unfair.
Relevant research on procedural justice (e.g. Lawler, 1968; Tyler, 1990) has
shown that people are more concerned with the fairness of the process than with
the favorableness of the outcome. People use the quality of the procedures, and
their own treatment during the decision-making process, as an index of their
place in the community. They will feel alienated from that community if they
are subject to unfair, cursory, or disrespectful decision-making. In contrast, a fair
decision is one in which people have an opportunity to present their arguments
and in which they feel treated with respect and politeness. Thus, fairness shares
some qualities with community, as well as with reward.
Fairness is also central to equity theory (Walster et al., 1973), which posits
that perceptions of equity or inequity are based on people’s determination of
the balance between their inputs (i.e. time, effort, and expertise) and outputs
(i.e. rewards and recognition). This core notion of inequity is also reflected in
the effort-reward imbalance model (Siegrist, 2002). Research based on these
theoretical frameworks has found that a lack of reciprocity, or imbalanced social
exchange processes, is predictive of burnout (e.g. Bakker et al., 2000; Schaufeli
et al., 1996).
Fairness has also emerged as a critical factor in administrative leadership
(e.g. White, 1987). Employees who perceive their supervisors as being both fair
and supportive are less susceptible to burnout, and are more accepting of major
organizational change (Leiter & Harvie, 1997, 1998). It appears that employees
value fairness in itself and consider it to be indicative of a genuine concern for the
long-term good of the organization’s staff, especially during difficult times. When
employees are experiencing stress, they look to management not only for problem
solving, but for optimism, fairness, and high expectations for organizational and
personal performance. They expect that management will give due consideration
to people’s contributions and will allocate resources and opportunities equitably
(and not to the personal advantage of privileged individuals or cliques).
Values
The values area is at the heart of people’s relationship with their work. It
encompasses the ideals and motivations that originally attracted them to the
job. It is the motivating connection between the worker and the workplace
that goes beyond the utilitarian exchange of time for money or advancement.
Contributing to a meaningful personal goal is a powerful incentive for individuals.
When this work contributes as well to the organizational mission, people
may be rewarded with additional opportunities for meaningful work. As such,
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
100 MICHAEL P. LEITER AND CHRISTINA MASLACH
mutually compatible values produce a self-perpetuating dynamic that supports
engagement.
However, when there is a values conflict on the job, it can undermine people’s
engagement with work. The greater the gap between individual and organizational
values, the more often staff members find themselves making a trade-off between
work they want to do and work they have to do. In some cases, people might
feel constrained by the job to do things that are unethical and not in accord with
their own values. For example, in order to make a sale or to obtain a necessary
authorization, they might have to tell a lie or be otherwise deceptive or not
forthcoming with the truth. People can also be caught between conflicting values
of the organization, as when there is a discrepancy between the lofty mission
statement and actual practice, or when the values are in conflict (e.g. high quality
service and cost containment do not always co-exist). In other instances, there
may be a conflict between their personal aspirations for their career and the values
of the organization, as when people realize that they entered an occupation with
mistaken expectations.
One resolution of the tension resulting from value conflicts is to bring personal
expectations in line with those of the organization (Stevens & O’Neill, 1983);
another is to leave the organization in search of more fulfilling career opportunities
(Pick & Leiter, 1991). The distress associated with value conflicts and the lengths
to which people go to reduce the associated tension are indicative of their central
role in the burnout and engagement process. Research has found that a conflict in
values is related to all three dimensions of burnout (Leiter & Harvie, 1997).
MISMATCH BETWEEN PERSON AND THE JOB
A consistent theme throughout this research literature is the problematic relation-
ship between the person and the environment, which is often described in terms
of imbalance or misalignment or misfit. For example, the demands of the job
exceed the capacity of the individual to cope effectively, or the person’s efforts
are not reciprocated with equitable rewards. There is a long history within psy-
chology of trying to explain behavior in terms of the interaction of person and
environment, and this is particularly evident within the fields of personality and of
vocational psychology (e.g. see Chartrand et al., 1995; Walsh et al., 1992). Many
of these interactional models view person and environment as independent entities,
but characterize them along commensurate dimensions so that the degree of fit,
or congruence, between person and environment can be assessed. This approach
is evident in some of the earliest models of job-person fit (French et al., 1974,
1982), in which better fit was assumed to predict better adjustment and less stress.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 101
Subsequent theorizing continued to highlight the importance of both individual
and contextual factors (see Kahn & Byosiere, 1992), and recent research con-
tinues to utilize this person-environment approach (e.g. Finnegan, 2000; Lauver
& Kristof-Brown, 2001; O’Reilly et al., 1999).
Thus, a model of job-person fit would seem to be an appropriate framework for
understanding burnout. However, prior conceptualizations of job-person fit are
limited in terms of their direct application to this phenomenon. For example, the
“person” is usually framed in terms of personality or an accurate understanding
of the job, rather than in terms of emotions or motivations or stress responses.
Similarly, the “job” is often defined in terms of specific tasks, and not the larger
situation or organizational context. The notion of “fit” is often presumed to predict
such outcomes as choice of job/occupation or of organization (entry issues), or
adjustment to the job (newcomer issues). In contrast, burnout involves a later point
in the process, when the person has been working for a while and is experiencing
a more chronic misfit between self and the job. Thus, the challenge is to extend
the job-person paradigm to a broader and more complex conceptualization of the
person situated in the job context.
We have begun to address this challenge by formulating a model that focuses on
the degree of experienced congruence between the person and the six domains of
his or her job environment (Maslach & Leiter, 1997). We propose that the greater
the perceived gap between the person and the job, the greater the likelihood of
burnout; conversely, the greater the consistency, the greater the likelihood of en-
gagementwith work. One newaspect of this approach is thefocus is onthe enduring
working relationship that people have with their job. This relationship is similar to
the notion of a psychological contract (Rousseau, 1995). Problems arise when the
process of establishing a psychological contract leaves critical issues unresolved,
or when the working relationship changes to something that a worker finds
unacceptable.
A second new aspect of this model is that it specifies not one, but six areas
in which this mismatch can take place. In each area, the nature of the job is not
in harmony with the nature of people, and the result is the increased exhaustion,
cynicism, and inefficacy of burnout. On the other hand, when better compatibility
exists in these six areas, then engagement with work is the likely outcome.
The Areas of Worklife Scale
Our goal was to develop a measure that would apply the concept of job-person fit
to the assessment of the six key areas of worklife, in a generic format that could
be utilized easily by a wide range of employees. We chose to focus on the fit
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
102 MICHAEL P. LEITER AND CHRISTINA MASLACH
itself, rather than on the two component parts of person and of job, and thus asked
respondents to rate their level of experienced congruence with the job within these
six domains. This new measure, the Areas of Worklife Scale, has the potential to
provide useful diagnostic information to organizations interested in interventions
to deal with burnout (Leiter & Maslach, 2000).
Description of the Measure
The Areas of Worklife Scale (AWS) is comprised of 29 items that produce distinct
scores for each of the six areas of worklife: workload (6), control (3), reward (4),
community (5), fairness (6), and values (5). Each scale includes positively worded
items, e.g. “I have enough time to do what’s important in my job” (workload)
and negatively worded items, e.g. “Working here forces me to compromise my
values” (values). Respondents indicate their degree of agreement with these state-
ments on a 5-point Likert-type scale ranging from 1 (strongly disagree), through
3 (hard to decide), to 5 (strongly agree). The scoring for the negatively worded
items is reversed. For each of the six subscales, the AWS defines a job-person
fit or match as a high score (greater then 3.00), indicating a higher degree of
congruence between the workplace and the respondent’s preferences; it defines a
mismatch as a low score (less than 3.00), indicating more incongruence between
the worker and the workplace. The AWS items were developed from a series of
staff surveys conducted by the Centre for Organizational Research and Develop-
ment (Leiter & Harvie, 1998; Maslach & Leiter, 1997) as a means of assessing the
constructs underlying our analysis of the six areas of worklife. The developmental
research found that the new scale had a consistent factor structure across these
initial samples and showed consistently high correlations with the three burnout
dimensions measured by the Maslach Burnout Inventory-General Scale (MBI-GS;
which is the general version of the MBI that can be used with all occupations).
The AWS is available through Leiter and Maslach (2000) or through Leiter and
Maslach (2002).
Samples
The normative sample for the AWS was drawn from a variety of worksettings in the
United States (English), Canada (English), Italy (Italian), and Finland (Finnish);
the number of participants is noted in Table 1. For those for whom demographic
information is available, there were 2,515 males and 5,139 females. In terms of
employment status, 6,343 were full time, 1,005 part time, and 112 casual. In terms
of age, participants were 18–29 years (650), 30–39 years (1,072), 40–49 years
(1,375), 50–59 years (1,061), and 60 and over (139). In terms of supervisory status,
the sample included non-supervisory employees (1,151), supervisors (1,545), and
management (810).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 103
Table 1. Research Settings.
Setting Source Nof Participants
USA: University Library Maslach & Leiter 388
USA: University Student Services Maslach & Leiter 738
Canada: Teachers Leiter 380
USA: University Library Maslach & Leiter 285
Finland: Post Office Aro, K¨
arn¨
a, Salmela-Aro 756
Italy: Hospital Maslach, Leiter, & Aroasio 390
Canada: Public Service Employees Leiter 17
Canada: Hospital Leiter 2,633
Finland: Hospital Aro, K¨
arn¨
a, Salmela-Aro 468
USA: University Administrative Employees Maslach & Leiter 1005
USA: Retail Maslach & Leiter 385
Italy: Hospital Maslach, Leiter, & Aroasio 295
USA: Teachers 39
Canada: Nurses 80
Finland: University Employees Aro, K¨
arn¨
a, Salmela-Aro 230
Finland: Postal Workers Aro, K¨
arn¨
a, Salmela-Aro 57
Finland: Telecommunications Aro, K¨
arn¨
a, Salmela-Aro 193
Total 8,339
Scale Properties
Table 2 displays the means, standard deviations, Cronbach alphas, and correlations
among the six areas of worklife and the three subscales of the MBI-GS. The alpha
values for all scales meet the 0.70 criterion. All of the correlations among the
subscales are significant. (The MBI-GS was not administered with every AWS
sample, thus producing only 6,815 cases for the combined sample in contrast
with the 7,574 cases for the AWS alone.) The highest correlation of the AWS and
the MBI-GS was between Workload and Exhaustion (0.54) while the lowest was
between Workload and Efficacy (0.04). The average of the 18 correlations of the
AWS with the MBI-GS was |0.26|; the average of the 15 correlations among the
AWS subscales was |0.20|.
Principal Components Factor Analyses
A principal components analysis of the normative sample provided evidence
supporting a six-factor structure for the AWS. The scree plot determined that
eigen values began leveling after six factors: 7.64, 2.53, 1.83, 1.60, 1.33, 1.24. The
six factor structure (see Table 3) assigned all 29 items to the appropriate factor.
Two items had loadings that were less than |0.50|: Workload6 loaded on Workload
0.46 with a second highest loading of 0.17 on control. Values5 loaded on Values
at 0.44 with a second highest loading of 0.22 on Fairness. As the second
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
104 MICHAEL P. LEITER AND CHRISTINA MASLACH
Table 2. Means, Standard Deviations, Cronbach Alphas, and Correlations: Normative Sample.
Mean S.D. Alpha Cynicism Efficacy Workload Control Rewards Community Fairness Values
Exhaustion 2.53 1.49 0.90 0.55 0.15 0.53 0.32 0.24 0.29 0.33 0.21
Cynicism 1.81 1.35 0.80 0.31 0.20 0.30 0.28 0.31 0.34 0.35
Efficacy 4.52 1.01 0.74 0.04 0.23 0.17 0.17 0.13 0.20
Workload 2.83 0.84 0.70 0.28 0.19 0.21 0.26 0.13
Control 3.35 0.89 0.70 0.38 0.40 0.47 0.33
Rewards 3.06 0.56 0.82 0.38 0.42 0.26
Community 3.46 0.84 0.82 0.48 0.34
Fairness 2.83 0.83 0.82 0.47
Values 3.43 0.74 0.73
Note: N =6,815, all correlations, p<0.001.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 105
Table 3. Principal Components Factor Analysis.
Fairness Community Workload Reward Values Control
Fair5 0.70 0.16 0.05 0.26 0.11 0.06
Fair4 0.70 0.15 0.09 0.18 0.22 0.19
Fair6 0.67 0.17 0.09 0.28 0.13 0.06
Fair3 0.65 0.14 0.09 0.16 0.15 0.24
Fair1 0.64 0.19 0.14 0.10 0.21 0.22
Fair2 0.54 0.07 0.04 0.04 0.08 0.13
Community3 0.14 0.84 0.07 0.12 0.10 0.10
Community4 0.18 0.80 0.09 0.11 0.08 0.12
Community2 0.13 0.74 0.01 0.18 0.17 0.14
Community1 0.26 0.61 0.10 0.02 0.12 0.19
Community5 0.08 0.57 0.06 0.28 0.07 0.04
Workload4 0.09 0.10 0.78 0.17 0.04 0.02
Workload1 0.08 0.01 0.76 0.01 0.07 0.04
Workload3 0.14 0.13 0.70 0.21 0.02 0.03
Workload5 0.07 0.04 0.63 0.02 0.13 0.27
Workload2 0.11 0.02 0.62 0.01 0.06 0.10
Workload6 0.06 0.09 0.46 0.02 0.07 0.17
Reward3 0.18 0.14 0.11 0.78 0.10 0.08
Reward4 0.18 0.10 0.17 0.73 0.05 0.05
Reward1 0.13 0.19 0.00 0.72 0.11 0.23
Reward2 0.19 0.21 0.03 0.69 0.15 0.30
Values1 0.16 0.11 0.02 0.05 0.78 0.07
Values3 0.11 0.08 0.04 0.10 0.77 0.07
Values4 0.25 0.08 0.11 0.05 0.66 0.11
Values2 0.05 0.07 0.06 0.05 0.60 0.03
Values5 0.22 0.16 0.19 0.17 0.44 0.06
Control3 0.14 0.14 0.04 0.15 0.12 0.73
Control1 0.12 0.12 0.17 0.16 0.05 0.72
Control2 0.30 0.15 0.11 0.20 0.13 0.58
loadings for both items were considerably lower than the loading on the proper
factor, the overall structure is acceptable.
In contrast, a five-factor solution provided a worse fit with a structure that
collapsed fairness and control into one factor. Both Workload6 and Values5
continued to have loadings less than 0.50. In addition all three control items loaded
less than |0.50| on the combined Fairness/Control factor. A four-factor solution
had a factor structure that scrambled items from control, reward, fairness, and
values with 8 items having loadings less than |0.50| and one item loading at 0.34.
To examine whether further differentiation of the scale was appropriate, a seven
factor solution defined a new factor from the two negatively worded fairness items
(Fair5 and Fair6) with the negatively worded Values5. Values5 continued to have a
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
106 MICHAEL P. LEITER AND CHRISTINA MASLACH
relatively low factor loading (0.45) and Workload6 had a loading on Workload of
0.48. The seven-factor solution failed to improve the fit over a six-factor solution
because: (1) it did not improve the factor loadings of the two weakest items;
and (2) it introduced a factor that was contrary to the objective of combining
negatively and positively worded items within most subscales of the AWS.
Confirmatory Factor Analysis
An EQS confirmatory factor analysis considered the factor structure of the AWS.
This analysis freed the four most highly correlated errors between items within
the subscales: Fairness5/Fairness6, Reward3/Reward4, Reward2/Reward1, Work-
load5/Workload1. Also, all co-variances among the factors were freed. The six fac-
tor solution was found to be an excellent fit to the data (x2
(358) =5,138.98, CFI =
0.939, RMSEA =0.042) with all factors loading significantly on the appropriate
item. In contrast, a one-factor solution showed a very poor fit (x2
(373) =25,514.14,
CFI =0.679, RMSEA =0.094). A two-factor solution (assigning the workload,
control, and community items to Factor 1 and the remaining items to Factor 2)
also showed a very poor fit (x2
(372) =23,364.14, CFI =0.706, RMSEA =0.090).
The assignment of items to the appropriate subscales in the six-factor solution
is displayed in Fig. 1.
Fig. 1. Confirmatory Factor Analysis.
Pl. provide the
missing matter
denoted by “?”.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 107
Table 4. Comparisons of Scores: Areas of Worklife Scale, Normative Sample.
Pair Mean Dif. S.D. tSig.
Pair1 Workload Control 0.52 1.04 40.95 0.01
Pair2 Workload Rewards 0.23 0.92 20.75 0.01
Pair3 Workload Community 0.63 1.05 49.44 0.01
Pair4 Workload Fairness 0.00 1.01 0.20 n.s.
Pair5 Workload Values 0.60 1.05 47.05 0.01
Pair6 Control Rewards 0.29 0.85 27.68 0.01
Pair7 Control Community 0.11 0.95 9.82 0.01
Pair8 Control Fairness 0.52 0.89 48.00 0.01
Pair9 Control Values 0.08 0.95 6.92 0.01
Pair10 Rewards Community 0.40 0.82 40.44 0.01
Pair11 Rewards Fairness 0.23 0.78 24.48 0.01
Pair12 Rewards Values 0.37 0.80 37.56 0.01
Pair13 Community Fairness 0.63 0.85 61.06 0.01
Pair14 Community Values 0.03 0.91 3.03 0.01
Pair15 Fairness Values 0.60 0.82 60.65 0.01
Note: N =6,815; df =6,814.
Comparisons Among the Areas of Worklife
Table 4 displays the contrasts among means of the six areas of worklife displayed
in Table 2. All contrasts were significantly different except for the contrast of
workload with fairness; these two areas of worklife were lower than the other four
areas. Community had the highest rating overall (M=3.46) followed closely by
values (M=3.43).
Comparison Among Demographic Groups
Gender differences were examined by a series of t-tests (see Table 5). Men rated
workload, control, and fairness more positively than did women; in contrast,
Table 5. Gender Differences.
Variable Male Female tdf Sig. (2-Tailed)
Exhaustion 2.31 2.62 7.80 6,289 0.01
Cynicism 1.87 1.79 2.28 6,289 0.05
Efficacy 4.58 4.50 2.99 6,289 0.01
Workload 3.02 2.79 10.11 6,289 0.01
Control 3.45 3.30 6.20 6,289 0.01
Rewards 3.09 3.05 2.31 6,289 0.05
Community 3.49 3.45 1.44 6,289 n.s.
Fairness 2.90 2.79 5.09 6,289 0.01
Values 3.40 3.46 2.65 6,289 0.01
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
108 MICHAEL P. LEITER AND CHRISTINA MASLACH
women scored more positively on values. The differences for community and
reward did not attain the p<0.01 level required for repeated t-tests.
Differences among employment status (full time, part time, casual) were
examined by a one-way analysis of variance with a Least Squares Difference
(LSD) test for post-hoc comparisons for the 6,147 respondents for which this
information was available (see Table 6). These groups did not differ on exhaustion
or cynicism, but full time employees scored higher on efficacy. On workload,
each group differed from the other two, with casual staff least positive about
workload and part time staff most positive. On control, each group differed from
the other two with full time staff most positive and casual staff least positive. The
groups did not differ on reward or community. On fairness and values, full time
staff reported less congruence than part time staff.
Contrasts among supervisory level (no supervision, supervisor, management)
were examined with a one-way analysis of variance with a LSD test for
post-hoc comparisons for the 3,417 respondents for which this information was
available (see Table 7). The comparisons indicate that front-line supervisors
were more exhausted and experienced less efficacy than either management or
non-supervisory employees. Non-supervisory employees were the most cynical,
while management reported the least cynicism. Management employees were
least positive about their workload and most positive about their sense of control.
Supervisors experienced the least congruence with rewards and reported the
strongest sense of community, while non-supervisory employees reported the
least sense of community. Non-supervisory employees also reported the lowest
rating of fairness and the least congruence of personal and organizational values.
In contrast, management employees reported the strongest congruence in values.
These patterns are summarized in the graph in Fig. 2.
Contrasts among age groups were examined with a one-way analysis of variance
with a LSD test for post-hoc comparisons for the 3,438 respondents for which this
information was available (see Table 8). For all six areas of worklife there was a
significant F, but the pattern of differences varied (see Fig. 3). For reward, control,
and values, there was a steady increase of positive ratings with age. In contrast,
fairness and community started high, dropped in the middle range, and increased
for older age groups. A sense of workload congruence steadily decreased with age.
Validity
Evidence for the validity of the items was providing by examining the corres-
pondence of scores on the Areas of Worklife measure with written comments
provided by participants in a hospital study (Leiter & Maslach, 2003). The over-
whelming proportion of comments from the 1,443 participants who commented
contained complaints. A qualitative analysis of the comments assigned comments
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 109
Table 6. Employment Status Differences.
Variable Group NMean S.D. F(2,6144) Sig. Group Differed From
Exhaustion Full time 5,345 2.55 1.49
Part time 711 2.46 1.48
Casual 91 2.76 1.61
Total 6,147 2.54 1.49 2.05 n.s.
Cynicism Full time 5,345 1.82 1.35
Part time 711 1.80 1.30
Casual 91 1.75 1.33
Total 6,147 1.82 1.34 0.17 n.s.
Efficacy Full time 5,345 4.54 1.01
Part time 711 4.38 1.00 Full time
Casual 91 4.25 1.00 Full time
Total 6,147 4.52 1.01 11.30 0.01
Workload Full time 5,345 2.84 0.82 Part time, casual
Part time 711 2.94 0.84 Full time, casual
Casual 91 2.63 0.82 Full time, part time
Total 6,147 2.85 0.83 8.00 0.01
Control Full time 5,345 3.37 0.90 Part time, casual
Part time 711 3.26 0.84 Full time, casual
Casual 91 2.86 0.87 Full time, part time
Total 6,147 3.35 0.89 18.57 0.01
Rewards Full time 5,345 3.07 0.56
Part time 711 3.07 0.57
Casual 91 2.98 0.57
Total 6,147 3.06 0.56 1.09 n.s.
Community Full time 5,345 3.46 0.84
Part time 711 3.39 0.83
Casual 91 3.41 0.80
Total 6,147 3.45 0.84 1.81 n.s.
Fairness Full time 5,345 2.80 0.84 Part time
Part time 711 2.89 0.77
Casual 91 2.74 0.80
Total 6,147 2.81 0.83 4.07 0.05
Values Full time 5,345 3.42 0.75 Part time
Part time 711 3.49 0.67
Casual 91 3.56 0.68
Total 6,147 3.43 0.74 3.98 0.05
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
110 MICHAEL P. LEITER AND CHRISTINA MASLACH
Table 7. Supervisory Level Differences.
Variable Group NMean S.D. F(2,2844) Sig. LSD Test: Group
Differed From
Exhaustion Non-supervisory 803 2.50 1.46 Supervisor
Supervisor 1,341 2.21 1.35 Both
Management 703 2.50 1.47 Supervisor
Total 2,847 2.36 1.42 14.85 0.01
Cynicism Non-supervisory 803 2.11 1.41 Both
Supervisor 1,341 1.81 1.32 Both
Management 703 1.55 1.25 Both
Total 2,847 1.83 1.34 33.25 0.01
Efficacy Non-supervisory 803 4.66 1.02 Supervisor
Supervisor 1,341 4.37 1.05 Both
Management 703 4.62 0.95 Supervisor
Total 2,847 4.52 1.03 24.94 0.01
Workload Non-supervisory 803 3.00 0.81 Management
Supervisor 1,341 3.06 0.83 Management
Management 703 2.60 0.85 Both
Total 2,847 2.93 0.85 76.05 0.01
Control Non-supervisory 803 3.45 0.86 Management
Supervisor 1,341 3.49 0.91 Management
Management 703 3.71 0.83 Both
Total 2,847 3.53 0.88 20.97 0.01
Rewards Non-supervisory 803 3.17 0.60 Supervisor
Supervisor 1,341 3.02 0.56 Both
Management 703 3.14 0.54 Supervisor
Total 2,847 3.10 0.57 20.63 0.01
Community Non-supervisory 803 3.38 0.87 Both
Supervisor 1,341 3.70 0.79 Both
Management 703 3.60 0.81 Both
Total 2,847 3.58 0.83 39.30 0.01
Fairness Non-supervisory 803 2.60 0.87 Both
Supervisor 1,341 3.03 0.75 Non-supervisory
Management 703 2.99 0.86 Non-supervisory
Total 2,847 2.90 0.83 78.15 0.01
Values Non-supervisory 803 3.15 0.71 Both
Supervisor 1,341 3.43 0.76 Both
Management 703 3.68 0.73 Both
Total 2,847 3.41 0.76 97.49 0.01
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 111
Fig. 2. Areas of Worklife as a Function of Supervisory Position.
from individuals to nodes, many of which were relevant to the six areas of worklife.
Table 9 displays correlations of scores on the six areas of worklife with a bi-
nary indicator of whether an individual wrote a complaint within the various
categories listed in the first column of Table 9. The second column of
Table 9 indicates the area of worklife most directly relevant to each node. The
pattern of correlations in Table 9 indicates that complaints were most strongly
correlated with scores on the area of worklife to which it was most directly
relevant.
Non-English Translations of the AWS
A confirmatory factor analysis assessed the extent to which the factor struc-
ture of the English version of the scale transferred to a Finnish translation
(Aro K¨
arn¨
a et al., 2001). The four Finnish samples noted in Table 1 included a
range of occupational groups: health care, university education, postal workers,
and telecommunications. This range of occupations requires a robust measure to
span the diverse occupational issues faced in these occupations in all six areas
of worklife. Table 10 displays the factor loadings of the six-factor solution.
All 29 items loaded on the appropriate scale. As with the overall CFA, all
correlations among the factors were freed as were the errors between four
pairs of items within subscales: Workload1/Workload5, Reward3/Reward4,
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
112 MICHAEL P. LEITER AND CHRISTINA MASLACH
Table 8. Age Group Differences.
Variable Group NMean S.D. F(2,2844) Sig.
Exhaustion 18–29 488 2.14 1.26 3.43 0.01
30–39 891 2.28 1.39
40–49 1,122 2.35 1.43
50–59 843 2.42 1.47
60+94 2.39 1.51
Total 3,438 2.32 1.41
Cynicism 18–29 488 1.93 1.30 0.82 n.s.
30–39 891 1.87 1.35
40–49 1,122 1.82 1.39
50–59 843 1.90 1.38
60+94 1.97 1.52
Total 3,438 1.87 1.37
Efficacy 18–29 488 4.38 0.99 4.51 0.01
30–39 891 4.56 0.99
40–49 1,122 4.62 1.01
50–59 843 4.54 1.05
60+94 4.53 1.05
Total 3,438 4.55 1.02
Workload 18–29 488 3.28 0.81 19.00 0.01
30–39 891 3.05 0.82
40–49 1,122 2.94 0.82
50–59 843 2.94 0.80
60+94 2.87 0.81
Total 3,438 3.01 0.82
Control 18–29 488 3.41 0.87 2.71 0.05
30–39 891 3.47 0.86
40–49 1,122 3.46 0.92
50–59 843 3.51 0.91
60+94 3.72 0.79
Total 3,438 3.47 0.89
Rewards 18–29 488 3.08 0.57 5.01 0.01
30–39 891 3.06 0.57
40–49 1,122 3.11 0.58
50–59 843 3.11 0.60
60+94 3.32 0.58
Total 3,438 3.10 0.58
Community 18–29 488 3.66 0.86 4.47 0.01
30–39 891 3.63 0.87
40–49 1,122 3.52 0.84
50–59 843 3.53 0.82
60+94 3.69 0.79
Total 3,438 3.58 0.85
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 113
Table 8. (Continued )
Variable Group NMean S.D. F(2,2844) Sig.
Fairness 18–29 488 2.96 0.77 6.81 0.01
30–39 891 2.91 0.81
40–49 1,122 2.79 0.88
50–59 843 2.87 0.86
60+94 3.14 0.84
Total 3,438 2.87 0.84
Values 18–29 488 3.28 0.72 4.49 0.01
30–39 891 3.32 0.74
40–49 1,122 3.32 0.80
50–59 843 3.40 0.76
60+94 3.56 0.84
Total 3,438 3.34 0.77
Community3/Community4, and Values1/Values2. The overall fit of the model
was good (x2
(358) =897.00, CFI =0.914, RMSEA =0.047).
A confirmatory factor analysis assessed the extent to which the factor structure
of the English version of the scale transferred to an Italian translation (Leiter
et al., 2002). The Italian hospital samples noted in Table 1 included a range of
Fig. 3. Areas of Worklife as a Function of Age.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
114 MICHAEL P. LEITER AND CHRISTINA MASLACH
Table 9. Correspondence With Qualitative Analysis.
Category Area of Worklife Workload Control Rewards Community Fairness Values
Workload on wards Workload 0.14 0.02 0.05 0.00 0.02 0.02
Workload (administrative) Workload 0.13 0.06 0.02 0.05 0.03 0.01
Patient care concerns Workload 0.09 0.05 0.06 0.07 0.04 0.04
Number of staff Workload 0.06 0.04 0.04 0.03 0.06 0.01
Clerical support Workload 0.08 0.07 0.01 0.02 0.02 0.04
Distribution of workload Workload 0.06 0.04 0.06 0.06 0.04 0.04
Timing of amalgamation Workload 0.10 0.00 0.00 0.01 0.03 0.02
Excessive/unproductive meetings Control 0.04 0.08 0.06 0.00 0.05 0.05
Professional autonomy/control Control 0.06 0.08 0.05 0.03 0.07 0.00
Flexible work times and place Control 0.02 0.06 0.01 0.01 0.03 0.00
Positive feedback and appreciation Reward 0.03 0.02 0.07 0.01 0.03 0.01
Appreciation Reward 0.05 0.08 0.11 0.03 0.02 0.07
Accountability for work Community 0.07 0.02 0.04 0.10 0.05 0.01
Fairness Fairness 0.00 0.05 0.12 0.07 0.14 0.07
Educational opportunities Fairness 0.04 0.03 0.06 0.01 0.08 0.06
Trust Fairness 0.01 0.06 0.06 0.04 0.11 0.03
Working relationships Fairness 0.07 0.06 0.07 0.05 0.10 0.03
Respect Fairness 0.05 0.04 0.11 0.07 0.07 0.03
Fair distribution of rewards Fairness 0.03 0.07 0.08 0.06 0.08 0.02
Staff involvement/input Values 0.03 0.01 0.03 0.04 0.09 0.09
Social get-togethers/functions Values 0.02 0.01 0.05 0.03 0.02 0.08
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 115
Table 10. CFA: Finnish Translation. Pl. provide
missing matter
denoted by “?”.
Item ?r2
Workload1 0.520 0.854 0.270
Workload2 0.158 0.987 0.025
Workload3 0.664 0.748 0.440
Workload4 0.850 0.527 0.722
Workload5 0.493 0.870 0.243
Workload6 0.362 0.932 0.131
Control1 0.675 0.738 0.455
Control2 0.751 0.661 0.563
Control3 0.539 0.842 0.291
Reward1 0.724 0.690 0.524
Reward2 0.884 0.468 0.781
Reward3 0.481 0.877 0.231
Reward4 0.377 0.926 0.142
Community1 0.478 0.879 0.228
Community2 0.750 0.661 0.563
Community3 0.689 0.725 0.474
Community4 0.619 0.785 0.384
Community5 0.437 0.899 0.191
Fair1 0.688 0.726 0.473
Fair2 0.096 0.995 0.009
Fair3 0.694 0.720 0.482
Fair4 0.736 0.677 0.542
Fair5 0.575 0.818 0.331
Fair6 0.626 0.780 0.391
Value1 0.712 0.702 0.507
Value2 0.499 0.866 0.249
Value3 0.783 0.622 0.614
Value4 0.598 0.801 0.358
Value5 0.548 0.836 0.300
health care workers, including direct service providers as well as administrative
and support staff. Table 11 displays the factor loadings of the six-factor solution.
All 29 items loaded on the appropriate scale. As with the overall CFA, all
correlations among the factors were freed as were the errors between four
pairs of items within subscales: Workload3/Workload4, Reward3/Reward4,
Community1/Community2, and Fairness5/Fairness6. The overall fit of the model
was good (x2
(358) =802.37, CFI =0.921, RMSEA =0.046).
Tests of the Mediation Model
With the establishment of the AWS, we have met the first challenge of devising a
measure of the organizational conditions variable in the basic mediation model of
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
116 MICHAEL P. LEITER AND CHRISTINA MASLACH
Table 11. CFA: Italian Translation. Pl. provide
missing matter
denoted by “?”.
Item ?r2
Workload1 0.705 0.709 0.498
Workload2 0.484 0.875 0.234
Workload3 0.624 0.782 0.389
Workload4 0.630 0.776 0.397
Workload5 0.582 0.814 0.338
Workload6 0.172 0.985 0.030
Control1 0.423 0.906 0.179
Control2 0.707 0.707 0.500
Control3 0.621 0.784 0.385
Reward1 0.801 0.599 0.641
Reward2 0.840 0.543 0.705
Reward3 0.590 0.808 0.348
Reward4 0.531 0.847 0.282
Community1 0.553 0.833 0.306
Community2 0.783 0.622 0.613
Community3 0.895 0.446 0.801
Community4 0.778 0.629 0.605
Community5 0.463 0.886 0.215
Fair1 0.655 0.755 0.429
Fair2 0.642 0.767 0.412
Fair3 0.679 0.734 0.462
Fair4 0.689 0.725 0.474
Fair5 0.511 0.859 0.261
Fair6 0.547 0.837 0.300
Value1 0.660 0.751 0.436
Value2 0.057 0.998 0.003
Value3 0.573 0.819 0.329
Value4 0.744 0.669 0.553
Value5 0.484 0.875 0.234
burnout. However, to test the model across multiple samples, we faced a second
measurement challenge. We needed to identify a generic work-related outcome
that would be relevant and significant for all respondents. We decided to measure
employees’evaluationofthe general change within theorganization whetherthey
saw things getting better or worse within the workplace. A positive perception of
change is a central outcome in post-industrialized organizations that emphasize
quality of service and must continually adapt to volatile environments.
Organizational change is best viewed as a continuous process shaped by
strategic decisions, in contrast with a model of rigidity disrupted on occasion by
change agents (Weick & Quinn, 1999). The relevant question from the perspective
of continuous change is the extent to which employees perceive the organization
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 117
as changing for the better or worse, not whether they perceive any change at
all. Especially important to employees’ capacity to function in a productive and
fulfilling fashion are high performance management practices that pertain to job
security, decision making, training, hiring, compensation, communication, and
reduced status distinctions (Pfeffer, 1998). Employees’ evaluation of the direction
of continuous change in these practices is an informative indicator of their overall
relationship with their work, and, as such, a central outcome measure.
Hypotheses
The basic hypothesis in our mediation model is that the greater the misfit between
the person and the job in the six areas of worklife, the greater the likelihood
of burnout; conversely, the greater the fit or match, the greater the likelihood
of engagement with work. This would suggest a simple additive model, in which
mismatches in each of the six areas would contribute separately to greater burnout.
However, the research literature reviewed earlier suggests the possibility of more
complex interrelationships between the six areas.
Because control is so central to employees’ ability to influence the people and
processes that determine the quality of worklife, we propose that it serves as the
starting point in our mediation model and will influence the extent to which people
can attain a match in the other areas, especially workload, reward, fairness, and
community. The area of values plays an integrating role in the model, reflecting
the overall consistency in the other areas of worklife. As such, it mediates the
relationship of the other areas with the psychological experience of burnout or
engagement. A match in values indicates that the organization’s central values are
consistent with those of the employee. In a values match, individuals embrace the
organization’s mission as a personal mission whose fulfillment is consistent with
personal aspirations. In a significant mismatch on values, employees perceive the
organization’s mission to be incompatible with their own well-being and that of
the larger community.
This conceptual analysis leads to the following hypotheses. The first set
proposes the replication of the standard pathways among the three dimensions of
burnout: exhaustion predicts cynicism, which in turn negatively predicts efficacy.
Second, all three dimensions are proposed to predict the outcome of evaluation
of change. The third set of hypotheses concerns the relationship of the six areas
of worklife to the three dimensions of burnout. As discussed earlier, workload is
predicted to have a direct path to exhaustion. Values is predicted to mediate the
relationship of all areas (except workload) with the three dimensions of burnout.
Control is predicted to be related to the other areas of workload, reward, fairness,
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
118 MICHAEL P. LEITER AND CHRISTINA MASLACH
Fig. 4. Hypothesized Model.
and community. The combined set of hypotheses forms the Mediation Model
depicted in Fig. 4.
Measures
Three measures assessed the primary elements of the mediation model: the six
areas of worklife, the three dimensions of burnout, and people’s evaluation of
organizational change. The Areas of Worklife Scale (AWS) is the measure we
developed to assess the six domains of organizational contributors to burnout; the
details of the measure have been described earlier in this chapter.
The Maslach Burnout Inventory-General Scale (MBI-GS; Schaufeli et al.,
1996) was used to measure the three dimensions of the burnout-engagement
continuum: exhaustion-energy, cynicism-involvement, and inefficacy-efficacy.
The items are framed as statements of job-related feelings (e.g. “I feel burned out
from my work, “I feel confident that I am effective at getting things done”), and
are rated on a 6-point frequency scale (ranging from “never” to “daily”). Burnout
is reflected in higher scores on exhaustion and cynicism, and lower scores on
efficacy, while the opposite pattern reflects greater engagement. Developed from
the original MBI (Maslach & Jackson, 1981), which was designed for human
service occupations, the MBI-GS is a 16-item measure that evaluates burnout
among people in all occupations. Thus, the MBI-GS was appropriate for all
employees within the participating organizations, providing comparative data
among units and occupational groups.
Evaluation of change was assessed by 11 items, of which the first three were
used in the model testing. This measure has served as an outcome measure in
previous research (Leiter & Harvie, 1998). Participants rated items on a five-point
Likert-type scale from 1 (much worse) through 3 (no change) to 5 (much better), in
response to an introductory sentence: “How do you perceive changes over the past
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 119
six months in the following”: The three items used in the model testing referred to
“services you provide, “your involvement in decisions that affect your work, and
“your job security. All three issues services, decision-making, and job security
have been the focus of concern in burnout research (Schaufeli & Enzmann, 1998)
and were identified as critical challenges in the participating organizations. Pfeffer
(1998) identified job security and decentralized decision making as basic
conditions for employees’ positive evaluation of organizational change. The
timeframe of six months was consistent with that of the MBI-GS, as well as with
a reasonable span for employees to consider when evaluating their worksettings.
Further, it provided a consistent metric across all the samples. The overall
variable of evaluation of change is computed as the average rating across all of
the change items.
Model Testing: Cross Sectional
An EQS analysis assessed a cross-sectional model, using data from the same nor-
mative sample that was reported earlier for the psychometric research on the AWS.
The model comprised sets of pathways representing the three sets of hypotheses
in the Mediation Model (Fig. 4). In this analysis only three indicators were used as
indicators of each of the latent variables: the three dimensions of burnout, the six
areas of worklife, and perception of change. Limiting the number of indicators to
three focuses the analysis primarily on the structural equation model, which is the
primary focus of this study. This approach is in contrast to one in which scale relia-
bility is considered solely in reference to overall inter-item consistency among the
items. Structural equation analysis considers, in addition to high inter-correlations
among the items within a latent construct, the consistency in the pattern of each
indicator within that latent construct with the indicators within the model’s other
latent construct (Bentler & Chou, 1987; Hayduk, 1987; Jaccard & Wan, 1996).
Whereas each indicator added to the causal model makes a distinct demand on the
predictive power of the model, a limit of three indicators for each construct yields
the most parsimonious perspective on the structural model. A limit of three indi-
cators also brings a rigor to measurement construction in that it requires that every
item maintain a strong level of inter-item correlation with the other two items, and
that all three items maintain a consistent pattern relative to the other constructs in
the model. Selecting the first three indicators of each scale rather than searching
for the most auspicious items emphasizes their strong inter-item consistency.
The EQS analysis (maximum likelihood, robust) confirmed a good fit of the
model to the data (x2
(304) =3,618.32, CFI =0.952, RMSEA =0.039), with all
paths at the 0.05 level of significance on the LM test (see Fig. 5). For this model
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
120 MICHAEL P. LEITER AND CHRISTINA MASLACH
Fig. 5. Causal Model Coefficients: Normative Sample.
correlations of eight errors within scales were freed; such correlations occur
frequently with measures as a function of the common response format in the
scale (Byrne, 1994).
Model Testing: Longitudinal
A second set of EQS analyses was used to test a longitudinal model, utilizing
data from an organization that had adapted our organizational assessment program
(Leiter & Maslach, 2000).
Background
Data were collected from the administrative and support staff of a large North
American university. The survey was conducted in an effort to systematically
assess current organizational strengths and weaknesses from the point of view of
the staff; to establish a baseline of data upon which to measure future gains; and
to inspire improvement. The survey was executed three times: Time 1 in 2000,
Time 2 in 2001, and Time 3 in 2002.
Participation in the study was voluntary, anonymous and confidential. The
survey was fully supported by top administration who, in their survey introduction,
pledged that survey responses and comments would be considered in future
discussions about how to improve the organization and operations. In addition,
a Balanced Scorecard Strategic Planning Group, who steered the survey process,
stated that they would be held personally accountable for ensuring that this
occurred.
During Time 1, responses were received from 1,005 of the possible 1,119
participants (90% response rate) who received the survey. At Time 2, a total
of 992 responses of the possible 1,140 participants (87% response rate) were
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 121
collected. At Time 3, a total of 812 responses of the potential 1,128 participants
(72% response rate) were received. The present study compared data of Time 1
with Time 2 and of Time 2 with Time 3.
Demographics: Time 1 to Time 2
Data from Time 1 and Time 2 were linked through an employee-generated code
that permitted the researchers to link the data without knowing the identity of
the person from whom the data were generated. Due to procedural changes,
some elements of the code were lost between Time 1 and Time 2, resulting in
linking only 207 of 800 participants’ data. Of the 207 participants for whom
data were linked, there were 70 females and 134 males with three not iden-
tified. The age ranges were 18–29 (11), 30–39 (32), 40–49 (80), 50–59 (69),
and 60 or older (11). The units for time of employment were less than six
months (9), six months to 1 year (9), 1–2 years (28), 2–5 years (34), 5–10 years
(10), 10–15 years (44), 15–20 years (27), 20–25 years (20), and more than 25
years (24). The positions included front line staff (164), front line supervisors
(19), and management (18). They included 184 career employees and 16 casual
employees.
Demographics: Time 2 to Time 3
Unfortunately, the two sets of linked data draw from different groups in Time 2
because of challenges in the code for linking data between periods. Although the
system of participant-generated codes provided a high degree of confidentiality, it
also led to a large number of incomplete links as individuals failed to provide a
consistent code across the various surveys. Of the 206 participants for whom data
were linked, there were 105 females and 98 males with three not identified. The
age ranges were 18–29 (30), 30–39 (41), 40–49 (63), 50–59 (57), and 60 or older
(8). The units for time of employment were less than six months (34), six months
to 1 year (27), 1–2 years (27), 2–5 years (34), 5–10 years (24), 10–15 years (23),
15–20 years (14), 20–25 years (10), and more than 25 years (7). The positions
included front line staff (156), front line supervisors (28), and management (19).
They included 178 career employees and 25 casual employees.
Changes in MBI-GS and AWS Over Time
Table 12 displays the correlations of the Time 1 MBI-GS and AWS subscales with
their counterparts at Time 2, along with their means and standard deviations. A
series of t-tests assess the changes over time. All correlations were significant.
T-tests identified two changes over the assessment interval: fairness increased
from Time 1 (M=2.64)toTime2(M=3.05) while values increased from Time
1(M=3.21) to Time 2 (M=3.40).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
122 MICHAEL P. LEITER AND CHRISTINA MASLACH
Table 12. Changes Over Time: Administrative Services: Time 1 to Time 2.
Variable rM S.D. t(203) Sig.
Time 1 Time 2 Time 1 Time 2
Exhaustion 0.47 2.19 2.17 1.33 1.42 0.15 n.s.
Cynicism 0.36 1.75 1.96 1.27 1.33 2.38 n.s.
Efficacy 0.38 4.88 4.64 0.93 1.06 3.40 n.s.
Workload 0.48 2.97 3.11 0.82 0.78 1.25 n.s.
Control 0.33 3.40 3.52 0.94 0.90 1.99 n.s.
Rewards 0.17 3.20 3.26 0.52 0.51 1.05 n.s.
Community 0.36 3.32 3.45 0.85 0.88 2.64 n.s.
Fairness 0.36 2.64 3.05 0.87 0.81 2.89 0.01
Values 0.26 3.21 3.40 0.73 0.71 2.45 0.01
Table 13 displays the parallel information for the participants match from
Time 2 to Time 3. Again, all correlations were significant between the measures
from one assessment to the other. In this interval, exhaustion increased (Time
2, M=2.10; Time 3, M=2.38) as did cynicism (Time 2, M=1.67; Time 3,
M=1.98). Community decreased (Time 2, M=3.61; Time 3, M=3.46), as
did fairness (Time 2, M=3.17; Time 3, M=3.01). Table 14 displays the alpha
levels for the two samples, indicating that the measures maintained an acceptable
level of internal consistency at both assessments.
The organization implemented a large number of interventions over each of
the assessment intervals, which were intended to enhance the quality of the work
environment. The increases from Time 1 to Time 2 suggest initial success, but
the subsequent changes in the negative direction from Time 2 to Time 3 suggest
that the organization encountered difficulty in the later stages.
Table 13. Changes Over Time: Administrative Services Time 2 to Time 3.
Variable rM S.D. t(203) Sig.
Time 2 Time 3 Time 2 Time 3
Exhaustion 0.61 2.10 2.38 1.50 1.54 2.99 0.01
Cynicism 0.52 1.67 1.98 1.33 1.37 3.38 0.01
Efficacy 0.49 4.65 4.72 0.95 1.06 1.03 n.s.
Workload 0.57 2.83 2.88 0.58 0.62 1.25 n.s.
Control 0.56 3.63 3.52 0.83 0.90 1.99 n.s.
Rewards 0.41 3.06 3.01 0.58 0.66 1.05 n.s.
Community 0.52 3.61 3.46 0.85 0.85 2.64 0.01
Fairness 0.60 3.17 3.01 0.82 0.91 2.89 0.01
Values 0.51 3.51 3.38 0.72 0.77 2.45 n.s.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 123
Table 14. Time 1 to Time 2: Cronbach’s Alpha.
Measure Time 1 Time 2
Workload 0.76 0.73
Control 0.76 0.70
Rewards 0.82 0.80
Community 0.84 0.85
Fairness 0.83 0.80
Values 0.70 0.71
Model Testing: Time 1 to Time 2
The EQS Model testing assessed the Mediation Model against data from this
sample in two steps. In the first step, the model included all of the pathways and
factor loadings in the Mediation Model with the variation that Time 1 Change was
dropped and Time 2 Exhaustion, Cynicism, and Efficacy were added. At this step
the only paths from Time 1 to Time 2 were those predicting a steady state: Time
1 Exhaustion to Time 2 Exhaustion, Time 1 Cynicism to Time 2 Cynicism, and
Time 1 Efficacy to Time 2 Efficacy (see Fig. 6). The EQS analysis (maximum
likelihood, robust) confirmed a good fit of the model to the data (x2
(565) =887.47,
CFI =0.920, RMSEA =0.041), with all paths at the 0.05 level of significance on
the Lagrange Multiplier (LM) test.
In the second step, this model was contrasted with one a-priori model (Burnout
Lag Model) and one exploratory model (Areas of Worklife Lag Model). The
Burnout Lag Model added to the Mediation model two paths: Time 1 Exhaustion
to Time 2 Cynicism, and Time 1 Cynicism to Time 2 Efficacy. These paths
Fig. 6. Longitudinal: Time 1 to Time 2.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
124 MICHAEL P. LEITER AND CHRISTINA MASLACH
parallel the relationships among Time 1 constructs. Although the overall model
retained the characteristics of a good fit data (x2
(563) =885.27, CFI =0.920,
RMSEA =0.041), neither of the new paths were significant, and they did not
result in a significant improvement of Chi Square (2
(2) =2.20, n.s.).
The Areas of Worklife Lag Model added to the Mediation Model one a-priori
path (Time 1 Workload to Time 2 Exhaustion) and two pathways identified on
the basis of Modification Indices: Time 1 Fairness to Time 2 Cynicism and Time
1 Values to Time 2 Efficacy. The resulting model maintained a good fit data
(x2
(562) =868.32, CFI =0.926, RMSEA =0.040), with all of the new paths
significant, and together resulting in a significant improvement of Chi Square
(2
(3) =19.15, p<0.01). Pl. check symbol
of “Chi square” is
OK?
Model Testing: Time 2 to Time 3
With the data from Time 2 to Time 3, the EQS Model testing followed the sequence
establishedfor assessing Time 1 to Time2: it assessed the Mediation Model against
data from the organization in two steps. In the first step, the model included the
pathways and factor loadings in the Mediation Model with Time 2 exhaustion,
cynicism, and efficacy. At this step the only paths from Time 1 to Time 2 were
those predicting a steady state: Time 1 Exhaustion to Time 2 Exhaustion, Time 1
Cynicism to Time 2 Cynicism, and Time 1 Efficacy to Time 2 Efficacy (see Fig. 7).
The EQS analysis (maximum likelihood, robust) confirmed a good fit of the model
to the data (x2
(565) =984.46, CFI =0.919, RMSEA =0.044), with all paths at the
0.05 level of significance on the LM test.
In the second step, this model was contrasted with the Burnout Lag Model and
the Areas of Worklife Lag Model. The Burnout Lag Model added to the Mediation
model two paths: Time 1 Exhaustion to Time 2 Cynicism, and Time 1 Cynicism
to Time 2 Efficacy. Although the overall model retained the characteristics of a
Fig. 7. Longitudinal: Time 2 to Time 3.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 125
good fit data (x2
(563) =981.18, CFI =0.920, RMSEA =0.044), neither of the
new paths were significant, and they did not result in a significant improvement
of Chi Square (2
(2) =3.28, n.s.).
Areas of Worklife Lag Model added to the Mediation Model the paths from
the previous analysis: Time 1 Workload to Time 2 Exhaustion, Time 1 Fairness
to Time 2 Cynicism, and Time 1 Values to Time 2 Efficacy. The resulting model
maintained a good fit data (x2
(562) =978.29, CFI =0.920, RMSEA =0.044),
but only the path from workload to exhaustion was significant, and the overall
improvement in fit was not significant (2
(3) =6.17, n.s.). An exploratory analysis
indicated that a Worklife Lag Model with three paths from Time 1 Workload to
Time 2 Exhaustion, Time 2 Cynicism, and Time 2 Efficacy provided a good fit
(x2
(562) =969.00, CFI =0.920, RMSEA =0.044) that was a significant increase
in fit (2
(2) =15.48, p<0.01).
DISCUSSION
Taken together, all of this research provides considerable support for our mediation
model. The experience of burnout or engagement is the mediating link between the
organizational context and work-related outcomes. It is not simply that burnout
is an important psychological outcome in its own right, but that it is related to
people’s commitment to their job and their evaluation of organizational change.
The clear implication is that the burnout experience should be predictive of other
job-related outcomes, such as work behaviors, and this should be the focus of
future research.
These analyses provide strong evidence for the utility of the Areas of Worklife
Scale (AWS) as a means of assessing organizational life, which is a key factor
in the mediation model. The scale produces a consistent factor structure, defining
six areas of worklife of specific relevance to the continuum from burnout to
engagement as assessed by the MBI-GS. The remarkable consistency of the
psychometric data across a variety of occupations, organizational settings,
national contexts and languages attests to the robust nature of the measure.
Implications for Research
The pattern of empirical results has important implications for both the validity
of the AWS measure, and the tests (both cross-sectional and longitudinal) of our
conceptual model.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
126 MICHAEL P. LEITER AND CHRISTINA MASLACH
Properties of the AWS
The factor structure identified in both the principal components analysis and the
confirmatory factor analysis support the six-factor solution including all 29 items.
Despite a strong level of correlation among the six subscales, the individual items
maintain their distinct status. The two items with loadings just below the ideal level
of |0.50| did not show cross loadings on other factors but were clearly associated
with the indicated factor. As they both were negatively worded items on their
respective factors, it is important to retain them in the overall scale, as they help to
avoid a strong unidirectional response set. It is a further confirmation of the factor
structure that a good level of fit was established using only three items for each
latent variable.
The contrasts among the various subscales confirm that there are distinct
normative levels for the various areas of worklife. The only two that did not differ
were workload and fairness, the two areas that consistently received the lowest
ratings across the various samples. To some extent the 3.00 level of the five-point
scale provides a clear demarcation between the range of congruence (from 3.01
to 5.00) and the range of incongruence (1.00–2.99). The normative levels on
these scales indicate that across a wide range of work settings, workload and
fairness are incongruent for most people. In contrast, control, community, and
values are generally congruent for most people, and rewards are at a neutral level.
Further research may explore the extent to which these scores reflect fundamental
qualities of worklife in post-industrial societies or whether on a more modest
scale they are qualities of the AWS measure. At this point they indicate an
important reference point in assessing the extent to which a work setting is
confronting distinct challenges or whether it is contending with the general nature
of work.
Differences among the demographic levels provide another reference point.
Many of these differences are quite small in absolute terms, becoming evident
only in a large sample such as the combined normative sample used in this
analysis. Overall, there was a high degree of consistency among the patterns of
scores on the AWS and the MBI-GS. However, a gender pattern did emerge, with
women reporting a more negative pattern than did men, with higher exhaustion,
lower efficacy, and less congruence on workload, control, and fairness. The one
element on which women were more positive than men was greater congruence on
values. Another pattern emerged with regard to age, with older people reporting
a greater congruence in most of the areas of worklife. This increasing congruence
may reflect either a greater capacity to shape the workplace with experience
or authority, or it may reflect a tendency to accommodate more readily to the
qualities of a workplace over time. The one exception was workload, which
showed a declining congruence for older workers. This pattern may reflect
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 127
increasing demands for people with greater experience or authority, or a more
constrained capacity of aging employees to manage work demands.
In regard to employment status, the higher scores of full-time staff on efficacy
and on control are consistent with their increased time and commitment to
their positions. The pattern of scores for casual staff is consistent with their
somewhat marginal status with the organizations: they have a minimal authority
and are generally called upon when workload is heavy. The more positive score
on workload for part-time employees is consistent with the lower demands
associated with a reduced time commitment to the organizations. The more
negative score on fairness for full-time employees relative to part-time employees
is surprising and could suggest systematic problems with organizational justice
or resource allocations associated with the factors that cause fairness to have a
lower normative level of congruence than the other areas.
Contrasts among supervisory level suggest widespread difficulties for front
line supervisory positions, as indicated by higher scores on exhaustion and
lower scores on efficacy. Supervisors reported the least congruence on rewards,
despite a relatively strong sense of community, suggesting that there may be a
benefit in reconsidering reward structures for front-line supervisors. The pattern
for managers suggests that their positional gains in control come at the price
of increased work demands beyond their expectations for the position. Their
relatively high scores on values are consistent with their greater capacity to shape
the organizational agenda and the symbolic role of management in representing
the organizational mission. Overall, these patterns indicate that people develop
distinct perspectives on the six areas of worklife, which are related to their own
positions in the organization and their personal characteristics.
Tests of the Mediation Model
The cross sectional analyses confirmed the central concepts of the mediation
model. All three components of burnout mediated the relationship between the
work setting (as assessed by the AWS) and the work outcome (evaluation of
change). A striking aspect of the findings is the complex way in which the six areas
of worklife predicted burnout. Workload and control each played critical roles
(thus replicating conceptually the Demand-Control model) but were not sufficient.
Reward, community, and fairness added further power to predict values, which
in turn was the critical predictor of the three dimensions of burnout. The results
revealed that this was not a simple additive model, but a more complicated
mediation model in its own right. The strong pathways from control to reward,
community, and fairness acknowledge the role of autonomy and participative
decision making to empower people to shape other key areas of their work
experience. The subsequent pathways from these three areas to values are
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
128 MICHAEL P. LEITER AND CHRISTINA MASLACH
consistent with people integrating their job experiences on various fronts into a
coherent perspective on their working life.
The results of the longitudinal analysis provide additional support for the
importance of considering areas of worklife and raise important points for
consideration in further research. First, the interrelationships among the three
dimensions of burnout were confirmed only within a single time period, and did
not show a lagged pattern across time. For example, the well-established path
from exhaustion to cynicism was seen at Time 1 and again at Times 2 and 3, but
exhaustion at one time period did not predict cynicism at a later point. These find-
ings suggest that the processes through which one component of burnout becomes
aligned with the others occur relatively quickly. These relationships are already in
place at the initial assessment and maintain a steady pattern of interrelationships
through the subsequent assessment, as indicated by the horizontal paths from
each burnout component to its subsequent state in Figs 6 and 7.
Second, and even more intriguing, are the longitudinal results that show
that some of the worklife areas (workload, fairness, and values) at Time 1 are
predictive of burnout at Time 2. In a partial longitudinal replication, workload at
Time 2 predicted burnout at Time 3. These lagged relationships clearly suggest
that the connection between organizational factors and burnout has a longer time
frame. Workload evidenced a consistent relationship with exhaustion across the
one-year interval for both steps of the analysis. This path is in addition to that
within Time 1 from workload to exhaustion and the subsequent step from Time
1 exhaustion to Time 2 exhaustion. This pattern implies that workload that is
incongruent with a worker’s expectations may have both long term and short term
implications. The other lagged relationships of areas of worklife with burnout
suggest that these areas may operate on a similar time frame. The path from
Time 1 fairness to Time 2 cynicism and from Time 1 values to Time 2 efficacy
supplement existing paths within Time 1. The failure to replicate these lagged
paths in the subsequent interval may indicate that these relationships are specific
to transitory conditions within the work setting. In contrast, the confirmation of
the predicted path from workload to exhaustion provides strong evidence for its
enduring nature.
Further progress in unraveling the key elements underlying burnout require
a well thought out concept of organizational life. The AWS strikes a balance
between becoming lost in the myriad elements of the organizational context on the
one hand, and limiting the focus to one or two simple elements on the other. Initial
research suggests the presence of consistent patterns of relationship among the
six areas of worklife, but there are also indications of situation-specific patterns,
as demonstrated by changes over the two steps of the longitudinal analysis.
Greater clarification of the unique and enduring patterns among aspects of the
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 129
organizational environment will provide a firmer basis for developing theory and
for implementing interventions to enhance the quality of worklife.
Implications for Intervention
By positing a complex framework through which people make sense of their work
settings, the mediation model constitutes a major advance over simple listings of
organizational characteristics. The set of organizational correlates of burnout has
grown so diverse that its further adumbration does not appear constructive in it-
self. The mediation model identifies six distinct dimensions of work settings that
encompass a large scope of burnout’s organizational correlates while remaining
sufficiently focused to be manageable. The structure of their relationships in the
mediation model including the pivotal role of control, the relative independence
of workload, and the pervasive influence of value congruence define a psycho-
logical environment in which people perceive and experience the world of work.
Underlying the model and the measures is the concept of people’s fit or match with
their job environment. Rather than proposing an ideal job or the ideal employee,
the model accepts a wide range of functional job environments and a wide range
of personal aspirations and inclinations shaping the way people work. The focus
of both our basic and applied research is examining the interactions of these two
dynamic and complex entities.
The mediation model opens avenues for interventions that will enhance the
quality of people’s job experience. It defines leverage points for changing the key
elements of burnout: the level of energy people bring to their jobs, the extent to
which they are involved in the work, and their sense of efficacy in their work.
These are deep-seated qualities of personal experience, not directly available
to the influence of management or an immediate supervisor. It is difficult to
conceive of how a manager would directly bestow additional energy, involvement,
or efficacy upon an employee. The mediation model points towards the proper
domain for organizational interventions: workplace policies and practices that
will shape the six key areas of worklife. These areas will affect the cognitive and
emotional experiences of employees (energy, involvement, efficacy), which in turn
will affect their attitudes and behavior at work. The challenge for organizational
interventions is being able to identify which areas need change, and this is where
the AWS becomes an especially useful diagnostic measure.
The AWS is a key element in our Organizational Checkup Survey (Leiter &
Maslach, 2000), which has proved to be a powerful tool for mobilizing organi-
zational self-reflection and change. The process of the organizational checkup
is designed to inspire the full participation of all the employees, as indicated by
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
130 MICHAEL P. LEITER AND CHRISTINA MASLACH
the strikingly high response rates in the organizations in which we have worked
(e.g. Leiter & Maslach, 2003). The main intent of the survey is to generate a
comprehensive profile of the organization’s workforce, which can be used to
inform decisions about intervention. However, the participative nature of the
checkup process can be viewed as an intervention in itself, which engages all
employees in an organizational dialogue and prepares them to get involved in
future change.
The results of our work with several organizations (which are included in
our normative sample for the AWS) propose two basic points of contact for
intervention: workload and values. An organization can enhance the energy levels
of employees by managing workload to be compatible with their expectations
and capacity. This apparently simple advice holds huge implications for organi-
zations operating in a fiercely competitive global market, or for public services
attempting to address growing demand with shrinking resources. The challenge of
managing workload is enormous. But the persistent relationship of unmanageable
workload with exhaustion, of exhaustion with cynicism, and of both with per-
formance problems, underscores the necessity to address this area of job-person
mismatch.
To address the second point of contact, values, organizations face the challenge
of building a shared vision of the organization. Although most organizations
in the post-industrial world have completed the initial steps of articulating a
mission statement and core objectives, few have succeeded in making that vision
permeate their policies and practices to the point that they affect everyone from
senior management to front line employees. It has been argued that the capacity to
imbue the organization with the core mission is a critical factor separating highly
effective companies from those with more modest levels of accomplishment
(Pfeffer & Sutton, 2000).
The mediation model suggests that a fundamental issue for managing em-
ployees’ experiences is their capacity to shape their worklife and to participate
in decisions. The position of control at the foundation of the model implies
that significant gains or losses in autonomy or authority can have widespread
implications for employees’ views of their job, their position on the continuum
from burnout to engagement, and their performance or attitudes about their work.
In addition to the overall themes of the model, the AWS provides the capacity
to assess specific work settings in regard to the six areas of worklife. Although
control or workload are key issues in general, a specific work setting may have
greater difficulties with their procedures for recognizing excellent performance
(reward), their internal processes for promotion (fairness), or the level of conflict
within the workplace (community). The AWS not only directs the organization’s
efforts to where they may have the greatest impact, it also provides a metric to
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Areas of Worklife 131
assess the extent to which interventions had their intended impact on specific
qualities of the work environment.
The model and the measure make a major contribution to making burnout a
problem that can be solved in better ways than having employees either endure the
chronic stress or quit their jobs. For the individual employees, the organizations
for which they work, and the clients whom they serve, the preferred solution is to
build a work environment that supports the ideals to which people wish to devote
their efforts. This is a formidable challenge, but one that becomes more possible
with the development of effective measures and a conceptual framework to guide
intervention.
REFERENCES
Aro, A., K¨
arn¨
a, P., Salmela-Aro, K., Leiter, M. P., & Maslach, C. (2001). Development of the Finnish
version of Areas of Worklife Survey. Unpublished data. University of Helsinki.
Bakker, A. B., Schaufeli, W. B., Sixma, H. J., Bosveld, W., & van Dierendonck, D. (2000). Patient
demands, lack of reciprocity and burnout: A five-year longitudinal study among general prac-
titioners. Journal of Organizational Behavior,21, 425–441.
Bentler, P. M., & Chou, C. P. (1987). Practical issues in structural modeling. Sociological Methods and
Research,16, 78–117.
Buunk, B. P., & Schaufeli, W. B. (1993). Professional Burnout: A perspective from social compari-
son theory. In: W. B. Schaufeli, C. Maslach & T. Marek (Eds), Professional Burnout: Recent
Developments in Theory and Research (pp. 53–69). Washington, DC: Taylor & Francis.
Byrne,B.M.(1993).TheMaslachBurnoutInventory: Testingforfactorialvalidityandinvariance across
elementary, intermediate and secondary teachers. Journal of Occupational and Organizational
Psychology,66, 197–212.
Byrne, B. M. (1994). Structural equ