You are on page 1of 11

Journal of Occupational Health Psychology Copyright 2006 by the American Psychological Association

2006, Vol. 11, No. 1, 52– 62 1076-8998/06/$12.00 DOI: 10.1037/1076-8998.11.1.52

A Cluster Analysis Investigation of Workaholism as a Syndrome


Shahnaz Aziz Michael J. Zickar
East Carolina University Bowling Green State University

Workaholism has been conceptualized as a syndrome although there have been few tests that
explicitly consider its syndrome status. The authors analyzed a three-dimensional scale of
workaholism developed by Spence and Robbins (1992) using cluster analysis. The authors
identified three clusters of individuals, one of which corresponded to Spence and Robbins’s
profile of the workaholic (high work involvement, high drive to work, low work enjoyment).
Consistent with previously conjectured relations with workaholism, individuals in the workaholic
cluster were more likely to label themselves as workaholics, more likely to have acquaintances
label them as workaholics, and more likely to have lower life satisfaction and higher work–life
imbalance. The importance of considering workaholism as a syndrome and the implications for
effective interventions are discussed.

Keywords: workaholism; syndrome; cluster analysis; work–life balance; life satisfaction.

Workaholism is a popular term used to describe holism as a multifaceted syndrome and examines its
individuals who are captivated by work. Ever since correlates; by doing so, we hope to provide more
Oates (1971) first coined the term workaholic to refer coherence to this important research area.
to an individual whose increased need to work hin-
ders one or more life functions, it has become a Definitional and Measurement-Related Issues
colloquial term used increasingly in the popular
press, on Web sites, and in the scientific literature Although there are other instruments designed to
(e.g., Burke, 2001a; Bonebright, Clay, & Anken- measure workaholism (e.g., Mudrack & Naughton,
mann, 2000; Porter, 2001; Schor, 1991). The philos- 2001; Robinson, 1998), Spence and Robbins’s (1992)
ophy of squeezing more of everything into a single instrument is the most frequently used self-report
24-hr day has become an accepted way of life. In measure of workaholism (Burke, 2001b; Porter,
fact, working excessive hours is often seen as a 1996). Their scale consists of three factors that con-
prerequisite for success. As a result, it appears that stitute workaholism: excessive work involvement,
some individuals may find it extremely problematic drive to work, and lack of work enjoyment. Work
to release themselves from work, even when they are involvement refers to the extent to which individuals
given the opportunity to do so (Porter, 2001). Even constructively use their time (both on and off the job)
though the term workaholism has become common- and dedicate themselves to working productively.
place, there has been little empirical research exam- Drive to work reflects the individual’s internal moti-
ining what it means when someone is referred to as a vation to work. Work enjoyment is the degree to
workaholic. The extant research, though limited, sup- which individuals derive pleasure from work. Spence
ports the notion that workaholism is an important and Robbins’s facets of workaholism are similar to
construct that warrants further empirical research. other conceptualizations of workaholism (e.g., Por-
However, most of that research considers facets of ter, 1996; Scott, Moore, & Miceli, 1997; Snir &
workaholism in isolation. This study tests worka- Zohar, 2000). In addition to the facets of worka-
holism, Spence and Robbins identified six work-
related profiles (see Table 1). Although these profiles
have been proposed, there has been little systematic
Shahnaz Aziz, Department of Psychology, East Carolina
University; and Michael J. Zickar, Department of Psychol- attempt to identify these groups empirically in work-
ogy, Bowling Green State University. ing samples.
Portions of this manuscript were conducted as part of the
first author’s dissertation at Bowling Green State University.
Correspondence concerning this article should be ad- Workaholism as a Syndrome
dressed to Shahnaz Aziz, Department of Psychology, East
Carolina University, 104 Rawl Building, Greenville, NC In the medical field, a syndrome is defined as a
27858-4353. E-mail: azizs@ecu.edu “term applied to a group of symptoms occurring

52
WORKAHOLISM AS A SYNDROME 53

Table 1
Spence and Robbins’s (1992) Classification of Worker Types
Work Drive Work
Worker type involvement to work enjoyment
Work enthusiast High Low High
Workaholic High High Low
Enthusiastic workaholic High High High
Unengaged worker Low Low Low
Relaxed worker Low Low High
Disenchanted worker Low High Low

regularly and thus constituting a disease to which lowed by subsequent validation of the purported
some particular name is given” (Macpherson, 2004, group of workaholics by comparing them to other
p. 602). In addition, syndrome is used instead of classes of individuals in terms of correlates of work-
disease when “the full picture of a condition as a true aholism. Relations among workaholics and total
entity has not been yet defined” (italics added, weekly work hours, job involvement, work stress, job
Haubrich, 1984, p. 236). The concept of a syndrome dissatisfaction, work–life imbalance, and life dissat-
has been used in psychological research to character- isfaction were examined; these measures are likely to
ize burnout (Maslach, 1982), effects of physical be correlates of workaholism. A more narrow focus
abuse in women (e.g., battered-women syndrome; on these variables helps verify past findings and
Kaser-Boyd, 2004), and posttraumatic stress (e.g., clarifies the relations among the current study con-
Burke & Mayer, 1985), among many other areas. In structs. Theory and empirical findings were used to
burnout research, for example, Maslach (1982) char- explain propositions regarding correlates that have
acterized burnout as a “syndrome of emotional ex- been suggested in the popular press (e.g., work stress)
haustion, depersonalization, and reduced personal ac- and to develop new hypotheses regarding correlates
complishment” (p. 3). Hence, people who experience that lie outside the workplace (e.g., life satisfaction).
burnout feel these characteristics in tandem. Simi-
larly, Spence and Robbins (1992) present worka- Multirater Approach
holism as a three-dimensional scale that implicitly
suggests that it is a syndrome. If workaholism is truly With self-report research, several studies have rec-
a syndrome, each of the three components (i.e., high ommended collecting information from acquain-
work involvement, high work drive, and low work tances and others who can verify it (e.g., Spector,
enjoyment) would be necessary (though not suffi- 1994). Collecting data from other sources to supple-
cient) conditions for someone to be classified as a ment the primary respondent may be especially im-
workaholic. Previous research has examined each of portant in workaholism research. Porter (1996) spec-
the constituent elements of workaholism in isolation. ulated that workaholics are often unaware that they
For example, studies have shown relationships be- are completely immersed in their work. As a result,
tween elements of the workaholism scales with other they might be oblivious to the detrimental effects that
variables (e.g., work enjoyment is negatively related working long hours has on both their physical and
to intentions to quit; Burke, 2001c). These univariate mental well-being. It is likely those people with
findings, however, fail to capture the syndrome na- whom the workaholic spends most of his or her time
ture of workaholism. For instance, low work enjoy- (i.e., family, friends, coworkers) who observe the
ment, by itself, does not signify that someone is a endless hours the workaholic devotes to work and the
workaholic. If workaholism is really a syndrome, it is harmful effects that occur as a result. Given that
important to investigate the dimensions concurrently. workaholics may be in denial of their condition, a
multirater approach was applied to alleviate the pre-
Current Study cept-percept bias that might be present in organiza-
tional behavior literature (Spector, 1994).
The primary goal of our study was to test if work- Going beyond self-reports and considering the
aholism was truly a syndrome with the three dimen- viewpoints of others may provide for a more reliable
sions proposed by Spence and Robbins (1992), fol- indicator of workaholism. In fact, Harpaz and Snir
54 AZIZ AND ZICKAR

(2003) suggested the inclusion of reports from fam- Correlates of Workaholism


ily, friends, and coworkers in future workaholism
research. Thus, in addition to self-reports, data from Job Involvement. Kanungo (1982) defined job
family, friends, and coworkers were also gathered. involvement as “a belief descriptive of the present
The group of family, friends, and coworkers is here- job that tends to be a function of how much the job
after referred to as the “acquaintance” sample; par- can satisfy one’s present needs” (p. 342). Job-in-
ticipants were asked to select only one acquaintance volved individuals actively engage in their job, con-
of their choice. sider their job as a central life interest, consider job
performance as a key to their self-esteem, and view
Workaholism Syndrome job performance as consistent with their self-concept
(Pinder, 1998). Individuals who are high on job in-
Though the literature cited has focused on inde- volvement are intrinsically motivated, more satisfied,
pendent main effects of individual aspects of worka- happier with their organization, and less absent from
holism, some researchers (e.g., Flowers & Robinson, work (Pinder, 1998). Other researchers have found
2002; McMillan, Brady, O’Driscoll, & Marsh, 2002) that highly job-involved individuals are disinterested
have found support for the idea that workaholism is in nonwork activities and are more likely to engage in
likely a multidimensional construct. To test our hy- organizational citizenship behaviors (Mudrack &
pothesis of whether workaholism is really a syn- Naughton, 2001).
drome with the three dimensions proposed by Spence Of note is the commonly held perspective that
and Robbins (1992), we conducted a cluster analysis workaholism is simply an extreme form of job in-
(Aldenderfer & Blashfield, 1984) to investigate volvement (Scott et al., 1997). Although the two
whether there was a cluster of individuals who fit the constructs have been considered synonymous in the
purported pattern of workaholism as posited by practitioner literature, job involvement is clearly dis-
Spence and Robbins (1992; i.e., high work involve- tinct from workaholism in several ways. Scott et al.
ment, high work drive, low work enjoyment). The (1997) indicate that job involvement has an attitudi-
second phase of data analysis was to show that peo- nal component regarding work, whereas worka-
ple in that cluster were more likely to experience the holism refers to behavioral patterns and an overall
correlates of workaholism (e.g., higher hours, re- outlook on work. Moreover, high job involvement
duced job satisfaction, more work–life imbalance). does not necessarily relate to workaholism in that
people might be highly engaged in their jobs and
Hypothesis 1: We expected workaholism to be a
“syndrome” in that there will be a cluster that consider work as a key element in their lives but may
fits the configuration of workaholics as proposed not be workaholics (e.g., they can leave at the end of
by Spence and Robbins’s (1992) three dimen- an 8-hr day and not think about work until returning
sions (i.e., high work involvement, high drive to the next day; Harpaz & Snir, 2003). Hence, worka-
work, and low work enjoyment). holism is not merely an extreme case of job involve-
ment in that other constituent components such as
Moreover, to further validate that the cluster of high drive to work and low work enjoyment might
individuals high on work involvement, high on drive also be present in the workaholic.
to work, and low on work enjoyment is indeed com-
posed of workaholics, we compared the workaholism Hypothesis 2a: Workaholics will be associated
cluster to the other clusters on constructs and vari- with higher levels of job involvement.
ables that have been proposed as correlates of work-
aholism (e.g., hours worked, self- and acquaintance- Work Stress. Stanton, Balzer, Smith, Parra, and
labels of workaholism, work–life imbalance). We Ironson (2001) applied the Lazarus and Folkman
expected significant mean differences between each (1984) model as a framework to better understand
of these outcomes with the workaholic cluster com- previous attempts to measure work stress. Lazarus
pared to other clusters identified in the analysis. Spe- and Folkman defined stress as “a particular relation-
cifically, we expected that workaholics would be ship between the person and the environment that is
higher on self-labels of workaholism, hours worked, appraised by the person as taxing or exceeding his or
job involvement, work stress, and work–life imbal- her resources and endangering his or her well-being”
ance and lower on job satisfaction and life satisfac- (p. 19). Given that we used Stanton et al.’s (2001)
tion compared to the other clusters. Stress in General (SIG) scale to measure overall work
WORKAHOLISM AS A SYNDROME 55

stress, our study follows their conceptualization work/nonwork conflict by developing a broader con-
of stress. struct referred to as work–life balance. Though
Excessive time devoted to work is related to stress work–life balance is related to work–family conflict,
(Porter, 2001). Compared to their nonworkaholic it is a broader, all-encompassing concept.
counterparts, workaholics experience greater stress Fisher (2001) defined work–life balance as an oc-
(e.g., Burke, 2000a; Burke & MacDermid, 1999). cupational stressor based on lost resources of time
Additionally, Spence and Robbins (1992) found pos- (e.g., amount of time spent at work relative to time
itive correlations between their work involvement spent in nonwork activities), energy (e.g., having
and drive to work facets and a measure of work energy available to pursue nonwork activities after a
stress. One should note, however, that correlations full day’s work), lack of goal accomplishment, and
solely between hours worked and psychological role strain between work and personal life demands.
strain are quite modest (Sparks, Cooper, Fried, & Employees are being required to do more work as
Shirom, 1997). Perhaps workaholics create stress and organizations downsize, merge, or undergo restruc-
pressure for themselves and tend to self-select them- turing. A trend of working longer hours emerges as a
selves into more stressful and demanding occupa- result of such organizational changes (Danna & Grif-
tions. fin, 1999; Bonebright et al., 2000), which in turn
hinders work–life balance. Seybold and Salomone
Hypothesis 2b: Workaholics will be associated (1994) have also suggested that workaholics over-
with greater levels of work stress. commit to work as a means to escape their personal
lives, whereas hard workers neglect their personal
Job Satisfaction. Job satisfaction has been de- lives in an attempt to succeed at work. For the pur-
scribed as “an affective reaction” to the job or job poses of the current study, work–life balance is
situation (Cranny, Smith, & Stone, 1992). Research equated with the absence of work interference with
on the relation between workaholism and job satis- personal life (WIPL).
faction has been virtually nonexistent. Job dissatis-
faction is possible depending on the degree to which Hypothesis 2d. Workaholics will be associated
workaholics can achieve work-related goals. If work- with greater levels of WIPL.
aholics are unable to fulfill their work aspirations,
they may feel less satisfied with the job (Burke, Life Satisfaction. Life satisfaction is the degree
to which individuals judge the overall quality of their
2000b; Scott et al., 1997). This perspective is also
lives favorably (Diener, Emmons, Larsen, & Griffin,
supported by Schaef and Fassel (1988). They char-
1985). It is reasonable to assume that workaholism
acterize workaholism as an addiction or compulsion
influences employee satisfaction that is external to
that leads to reduced satisfaction; this may lie in the
the job itself (i.e., life satisfaction). Stress, burnout,
fact that workaholics experience greater stress, which
and health complaints that result from workaholism
in turn lowers satisfaction. Furthermore, Burke
would not only have an effect on work-related out-
(2000b) found a negative relation between worka-
comes (e.g., job performance) but also have adverse
holism and extrawork satisfactions (e.g., family,
effects on things that lie outside the boundaries of
friends, community). It appears that most of the re-
work, namely, life satisfaction.
search that has been conducted regarding the relation
between workaholism and job satisfaction has re- Hypothesis 2e: Workaholics will be associated
vealed a negative relation. with higher levels of life dissatisfaction.
Hypothesis 2c: Workaholics will be associated
with higher levels of job dissatisfaction.
Methods

Work–Life Balance. Past research in the area of


Participants
work/nonwork conflict has mainly focused on “fam- Because white-collar professionals are typically associ-
ily” as the prominent aspect of life in the nonwork ated with workaholism, professional employees (N ⫽ 174)
domain. However, by focusing solely on family, we from various white-collar organizations (e.g., consulting
fail to consider aspects of nonwork life (e.g., house- firms, oil companies, government agencies) located in Can-
ada (78%) and the United States (22%) served as the current
hold chores, personal hobbies) in addition to family study sample. The Wilcoxon’s Matched Pairs test indicated
(Fisher, 2001). Therefore, Fisher sought to enhance that Canadians and Americans were not significantly differ-
our understanding of the stress that results from ent in terms of the study variables and demographics (e.g.,
56 AZIZ AND ZICKAR

age, gender, ethnicity); therefore, the Canadian and U.S. 1.5 for the positively worded items and “Y” ⫽ 0, “N” ⫽ 3,
samples were combined to conduct analyses. “?” ⫽ 1.5 for the negatively worded items. Higher scores
The complete sample consisted of both males (56%) and indicate higher levels of overall stress. The present study
females (44%). Employees ranged in age from 23 to 70 obtained an overall alpha value of 0.86.
years, with a mean age of 43. The majority (68%) of the Job Satisfaction. The 18-item Job in General scale
employees in this sample were Caucasian. Organizational (JIG; Ironson, Smith, Brannick, Gibson, & Paul, 1989) was
tenure ranged from less than one year to 15 years or more; used to measure overall job satisfaction. The response scale
26% of the survey respondents had been with their current for the JIG scale is similar to the SIG: “Y” for “yes,” “N”
organization for more than 15 years. Furthermore, approx- for “no,” and “?” for “cannot decide.” Scoring for the JIG is
imately a quarter (28%) of those surveyed had served be- as follows: “Y” ⫽ 3, “N” ⫽ 0, “?” ⫽ 1 for the positively
tween 3 and 4 years in their current position. worded items and “Y” ⫽ 0, “N” ⫽ 3, “?” ⫽ 1 for the
negatively worded items. Higher scores indicate higher lev-
els of overall job satisfaction. The current study found a
Procedure reliability coefficient of 0.91 for the employee sample.
Similarly, an alpha of 0.85 was obtained for the acquain-
Convenience sampling was used to recruit participants: tance sample.
alumni, friends, and acquaintances who were affiliated with Work–Life Balance. Fisher’s (2001) 13-item WIPL
professional organizations were contacted as sources for scale was used to measure work–life balance. Following
data collection. In this sense, the white-collar organizations Fisher’s methods, respondents were asked to indicate the
that were selected for inclusion in the current study were frequency that they felt a particular way during the previous
those in which familiar others served as contact people. In 3 months using a six-point scale. The current study obtained
general, the acquaintance rated the participant on such vari- a reliability coefficient of 0.89 for the WIPL scale. Simi-
ables as job satisfaction and work–life balance, thereby larly, an alpha of 0.87 was obtained for the acquaintance
providing a “knowledgeable other” check of the participant. sample.
For brevity purposes, not all scales were rated by acquain- Life Satisfaction. Diener et al.’s (1985) five-item Satis-
tances. When answering the questions, acquaintances were faction with Life scale was used to assess overall life
asked to think about the individual who provided them with satisfaction. The Diener et al. scale measures items on a
the survey. This individual was referred to as the ratee. We seven-point Likert format ranging from “strongly disagree”
told acquaintances that we were interested in how they to “strongly agree.” The current study obtained an alpha
believe this individual would respond to each question and value of 0.89.
that we would link their survey responses to a survey that In addition to the previously indicated measures, respon-
was administered to the ratee. Acquaintances were assured dents indicated the total number of hours per week they
that an identification number would be used to anonymously worked on the job (including outside of work). Finally, our
track both sets of survey responses. Out of the 385 paired study also employed a self-label measure of the employees’
surveys (i.e., surveys sent to employees and their acquain- ratings of themselves as workaholics. Participants were
tances) that were mailed out, 174 completed paired surveys asked, “On a scale of 1 to 10, with 10 being high, how much
were returned and used in the study, yielding a response rate do you consider yourself to be a workaholic?” Acquain-
of 45%. tances were asked a similar question about the study
participant.

Measures
Results
Workaholism. Spence and Robbins’s (1992) measure
was used to assess work involvement, drive to work, and Table 2 shows total employee and acquaintance
work enjoyment. Each facet consists of its own set of items.
The scale measures items on a five-point Likert format sample means and standard deviations for all vari-
ranging from “very true of me” to “very untrue of me.” For ables. Overall, a comparison of the means of the
the employee sample, the following reliability coefficients employee and the acquaintance reveals very similar
were obtained: 0.74 for work involvement, 0.81 for feeling patterns. For example, the average workaholism rat-
driven to work, and 0.90 for work enjoyment. Similarly, for
the acquaintance sample, the reliability coefficients were as
ing was 6 on a 10-point scale for both samples. In
follows: 0.73 for work involvement, 0.78 for feeling driven addition, both employees and acquaintances indi-
to work, and 0.88 for work enjoyment. cated the average total weekly work hours to be 50
Job Involvement. Reeve and Smith’s (2001) 20-item hours per week (including outside of work).
scale was used to measure job involvement. The response Pearson correlations and Cronbach alphas for all
format consists of four-point bipolar scales anchored by
“strongly agree” to “strongly disagree.” The current study employee and acquaintance variables appear in Ta-
employee sample obtained an alpha of 0.80. Similarly, an bles 3 and 4, respectively. Alpha values ranged from
alpha of 0.81 was obtained for the acquaintance sample. .73 to .91 and are consistent with those cited in
Work Stress. The 15-item SIG scale (Stanton et al., previous research. Although several significant cor-
2001) was used to measure overall work stress. Responses
to each of the 16 items on the SIG measure were recoded as
relations are observed in the tables, to be succinct, we
“Y” for “yes,” “N” for “no,” and “?” for “cannot decide.” highlight the fact that the pattern of correlations is
Scoring for the SIG is as follows: “Y” ⫽ 3, “N” ⫽ 0, “?” ⫽ similar to bivariate relations that have been found in
WORKAHOLISM AS A SYNDROME 57

Table 2
Means and Standard Deviations of All Variables for Employees and Acquaintances (N ⫽ 174)
Possible range Range of scores
of scores obtained M SD
Employee
Workaholism rating 1–10 1–10 5.89 2.21
Hours per week — 30–105 49.88 11.24
Job involvement 9–36 10–35 22.46 5.26
Work stress 0–48 0–48 24.12 12.32
Job satisfaction 0–54 0–54 43.54 10.32
Work interference with personal life 1–6 1.31–4.62 2.75 0.71
Life satisfaction 5–35 9–35 24.11 6.39
Work involvement 8–40 13–38 25.69 5.61
Drive to work 7–35 9–35 23.56 5.53
Work enjoyment 10–50 15–50 35.08 7.52
Acquaintance
Workaholism rating 1–10 1–10 5.90 2.18
Hours per week — 30–100 49.70 10.45
Job involvement 9–36 10–34 22.02 5.19
Job satisfaction 0–54 8–54 43.18 8.90
Work interference with personal life 1–6 1.23–4.38 2.81 0.70
Work involvement 8–40 13–39 25.58 5.72
Drive to work 7–35 8–34 23.70 5.27
Work enjoyment 10–50 11–49 33.14 7.21

other samples. Therefore, the data appear to be con- by the employees. Wilcoxon’s Matched Pairs tests
sistent with previous research. identified no mean differences between acquaintance
To assess interrater agreement, intraclass correla- and employee respondents. Therefore, employees did
tions between the employee and the acquaintance not tend to deny their behavior in comparison to
were calculated on all study variables (e.g., total acquaintances and seemed to have a fairly accurate
weekly work hours, job involvement, WIPL) (see view of themselves.
Table 5). Results indicated significant (p ⬍ .01)
intraclass correlations between the employee and the Hypothesis 1: Workaholism Syndrome
acquaintance on all study variables (e.g., total weekly
work hours; r ⫽ .69, p ⬍ .01). In essence, the To test the syndrome hypothesis, scores for each of
acquaintances substantiated the responses provided Spence and Robbins’s (1992) three dimensions of

Table 3
Correlations Among All Employee Variables (N ⫽ 174)
1 2 3 4 5 6 7 8 9 10
1. Employee’s rating —
2. Hours per week 0.47** —
3. Job involvement 0.57** 0.36** 0.80
4. Work stress 0.38** 0.30** 0.24** 0.86
5. Job satisfaction 0.07 0.01 0.33** ⫺0.35** 0.91
6. WIPL 0.41** 0.38** 0.34** 0.67** ⫺0.26** 0.89
7. Life satisfaction ⫺0.29** ⫺0.09 ⫺0.31** ⫺0.38** 0.22* ⫺0.39** 0.90
8. Work involvement 0.49** 0.34** 0.62** 0.21* 0.04 0.38** ⫺0.23* 0.74
9. Drive to work 0.43** 0.25** 0.43** 0.47** ⫺0.22* 0.48** ⫺0.27** 0.65** 0.81
10. Work enjoyment 0.29** 0.20** 0.66** ⫺0.08 0.59** ⫺0.01 0.16 0.39** 0.18 0.90
11. Acquaintance’s rating — 0.48** 0.46** 0.28** 0.03 0.32** ⫺0.23** 0.36** 0.31** 0.24**
Note. Internal consistency reliabilities on diagonal are Cronbach’s alpha.
* p ⬍ 0.05. ** p ⬍ 0.01.
58 AZIZ AND ZICKAR

Table 4
Correlations Among All Acquaintance Variables (N ⫽ 174)
1 2 3 4 5 6 7 8
1. Acquaintance’s rating —
2. Hours per week 0.53** —
3. Job involvement 0.57** 0.44** 0.81
4. Job satisfaction 0.11 0.06 0.49** 0.85
5. WIPL 0.45** 0.33** 0.49** ⫺0.03 0.87
6. Work involvement 0.44** 0.21** 0.53** 0.11 0.43** 0.73
7. Drive to work 0.47** 0.19* 0.50** ⫺0.02 0.42** 0.42** 0.78
8. Work enjoyment 0.37** 0.33** 0.65** 0.61** 0.18* 0.44** 0.21* 0.88
Note. Internal consistency reliabilities on diagonal are Cronbach’s alpha.
* p ⬍ 0.05. ** p ⬍ 0.01.

workaholism (i.e., work involvement, drive to work, definition of workaholism, individuals falling into
work enjoyment) were first transformed (or standard- cluster 1 are referred to instead as positively engaged
ized) into z scores. Next, a K-means cluster analysis workers. In contrast, individuals who scored below
(Aldenderfer & Blashfield, 1984) was performed to the mean on all three dimensions were classified as
show there is a cluster that fits the configuration of unengaged workers (cluster 3). Cluster 2 fit the con-
workaholics as proposed by Spence and Robbins figuration of standard workaholics proposed by
(i.e., high work involvement, high drive to work, and Spence and Robbins, thereby supporting hypothesis
low work enjoyment). We analyzed two-cluster, 1. Table 6 shows the number of employees per cluster
three-cluster, and four-cluster solutions. The three- and the resulting means of the three workaholism
cluster solution was more interpretable, sensible, and dimensions. In sum, 40% (n ⫽ 69) of the sample
consistent with Spence and Robbins’s worker types. belonged to cluster 1, 23% (n ⫽ 39) to cluster 2, and,
For interpretability’s sake, scores above the mean finally, 37% (n ⫽ 64) to cluster 3.
(positive z scores) were classified as “high” scores
and those below the mean (negative z scores) as
“low” scores. Applying Spence and Robbins’s de- Hypothesis 2: Correlates of Workaholism
fined patterns of high and low scores on their three
facets (see Table 1), participants in each cluster were To further validate that the cluster indeed consists
classified into a particular worker type. For example, of workaholics, we compared that cluster with the
an individual who scored above the mean on all three other clusters to see if they were higher (or lower) on
dimensions was classified as an enthusiastic worka- variables that have been proposed to be associated
holic (cluster 1). Given that the term “enthusiastic with workaholism using one-way analyses of vari-
workaholic” is inconsistent with the current study ance. Both employee and acquaintance variables
were used in these analyses. Mean scores on all
measures for positively engaged workers, worka-
holics, and unengaged workers are shown in Table 6.
Table 5 Significant differences were found across all three
Intraclass Correlations (ICC) Between Employee clusters for each employee and acquaintance corre-
Variables with Acquaintance Variables (N ⫽ 174) late (see F tests in Table 7). We employed Tukey
follow-up tests for pairwise comparisons that cor-
Variable ICC rected for experiment-wise error rates. For succinct-
1. Workaholism rating 0.54** ness, only those comparisons that include the pur-
2. Hours per week 0.69** ported workaholic group are emphasized. We found
3. Job involvement 0.62** workaholics (cluster 2) to have significantly higher
4. Job satisfaction 0.40**
5. WIPL 0.60** self-labels of workaholism (M ⫽ 7.05), hours worked
6. Work involvement 0.51** per week (M ⫽ 52.90), job involvement (M ⫽ 23.92),
7. Drive to work 0.48** work stress (M ⫽ 29.38), work–life imbalance (M ⫽
8. Work enjoyment 0.49** 2.95), and life dissatisfaction (M ⫽ 20.87) than un-
** p ⬍ 0.01. engaged workers (cluster 3) and significantly less job
WORKAHOLISM AS A SYNDROME 59

Table 6
Number of Employees per Cluster/Type, Percentage of the Sample, and Mean z Scores for Spence and
Robbins’s (1992) Dimensions
Work Drive Work
Cluster n (%) involvement to work enjoyment
1 (Positively engaged worker) 69 (40) 0.3183 0.0506 0.8381
2 (Workaholic) 39 (23) 0.8089 1.1068 ⫺0.3091
3 (Unengaged worker) 64 (37) ⫺0.8361 ⫺0.7290 ⫺0.7152

satisfaction (M ⫽ 38.69), work–life balance, and life professionals are typically associated with worka-
satisfaction than positively engaged workers (cluster holism, the current study was composed of employ-
1). Likewise, acquaintance reports found workaholics ees from various professions in white-collar organi-
to have significantly higher acquaintance labels of zations; few studies have investigated workaholism
workaholism (M ⫽ 6.79), job involvement (M ⫽ in organizational samples. Furthermore, in addition
22.95), and work–life imbalance (M ⫽ 2.92) than to self-reports, our study also used reports from fam-
unengaged workers. Overall, these results provide ily, friends, and coworkers. There was agreement on
support for hypotheses 2a through 2e. the relative standing of individuals with respect to
workaholism, in that the acquaintance’s results sub-
Discussion stantiated the employee’s findings. This agreement
alleviates some of the standard concerns that have
The current study extended previous research in been made about self-report.
several important ways. First, workaholism was con-
ceptualized as a syndrome in that a set of key com- Workaholism Relations
ponents was considered to characterize the worka-
holic. Employees who experienced these characteristics With a few exceptions, substantial support was
in conjunction with each other were more likely to be found for the individual relations among Spence and
workaholics than those who just experienced a subset Robbins’s (1992) triad of workaholism (i.e., work
of the symptoms. In addition, because white-collar involvement, drive to work, work enjoyment) with

Table 7
Means (and Standard Deviations) of Correlates for All Three Clusters of Worker Types
Cluster 1 Cluster 2 Cluster 3
Correlate M SD M SD M SD F(2, 171)
Employee
Workaholism rating 6.28 (2.24)a 7.05 (1.69)a 4.80 (1.96)b 17.06***
Hours per week 51.51 (12.90)a 52.90 (10.70)a 46.33 (7.93)b 5.79**
Job involvement 24.46 (4.44)a 23.92 (5.03)a 19.22 (4.93)b 22.76***
Work stress 23.96 (12.27)a,b 29.38 (11.27)a 20.58 (11.93)b 6.61**
Job satisfaction 47.44 (6.08)a 38.69 (15.05)b 42.78 (8.74)b 10.57***
Work-life imbalance 2.58 (0.67)a 2.95 (0.69)b 2.37 (0.63)a,c 9.23***
Life satisfaction 24.93 (6.26)a 20.87 (7.26)b 25.41 (5.40)a,c 7.30***
Acquaintance
Workaholism rating 6.24 (2.05)a 6.79 (1.94)a 4.94 (2.16)b 11.35***
Hours per week 51.97 (11.22)a 50.92 (10.00)a,b 46.85 (9.00)b 4.37*
Job involvement 24.21 (4.98)a 22.95 (4.66)a 19.76 (5.07)b 13.55***
Job satisfaction 45.52 (8.06)a 41.79 (10.20)a,b 41.74 (8.74)b 3.73*
Work-life imbalance 2.70 (0.66)a 2.92 (0.72)a,b 2.42 (0.55)c 7.90***
Note. Means in the same row that have the same subscript are not significantly different at p ⬍ .05 in the Tukey honestly
significant difference comparison.
* p ⬍ 0.05. ** p ⬍ 0.01. *** p ⬍ 0.001.
60 AZIZ AND ZICKAR

most of the correlates of workaholism (e.g., total light (i.e., social desirability). This is not improbable
weekly work hours, workaholism ratings, job in- given that the positive connotation associated with
volvement). For the most part, the pattern of corre- being a workaholic has become fairly prevalent in
lations among the different variables reflects the find- our society. Second, individuals might be more aware
ings of other researchers. of and have better insights into their own behavior
The following results are based on replication of (Spector, 1994).
past studies. In their study on workaholism, Burke
(2001c) found average total weekly hours to be 51 Workaholism as a Syndrome
hours per week; in our study, both employees and
acquaintances indicated the mean total weekly work The current study helped to clarify some of the
hours to be 50 hours per week (including outside of confusion in the realm of workaholism by determin-
work). In addition, given the positive relations found ing that the workaholic is portrayed by a set of
in the literature between job involvement and job distinct characteristics. In this sense, workaholism
satisfaction (e.g., Furnham, Brewin, & O’Kelly, can be conceptualized as a multifaceted syndrome.
1994; Morrow & Crum, 1998), Mudrack and Naugh- Considerable support was found for the notion of
ton (2001) expected (and found) positive relations workaholism as a syndrome in that each of Spence
between work involvement and work enjoyment. Our and Robbins’s (1992) components (i.e., high work
findings also showed significantly positive correla- involvement, high drive to work, low work enjoy-
tions between work involvement and work enjoy- ment) were necessary (though not sufficient) for an
ment. Furthermore, Bonebright et al. (2000) found individual to be classified as a workaholic. In fact, a
that workaholics were likely to work more hours per class of employees was found to match the pattern of
week than their nonworkaholic counterparts; our workaholics proposed by Spence and Robbins,
study revealed a positive correlation between total thereby substantiating that workaholism is truly a
weekly work hours and workaholism ratings. syndrome.
Bivariate correlations between job satisfaction and Results provided strong evidence that the class of
work involvement were not statistically significant employees was in fact composed of workaholics in that
for both self and acquaintance reports. The absence the employees were more likely to experience the
of a significant correlation between these two vari- correlates of workaholism (e.g., more acquaintance-
ables is not surprising given the mixed findings from labeling, greater job involvement, reduced job satis-
previous literature; the verdict with regard to the faction). Our findings support those of past research-
workaholism–job satisfaction relation remains unre- ers, who, for example, found positive relations
solved. In fact, Mudrack and Naughton (2001) did between workaholism and job involvement (e.g.,
not expect a strong directional relation between work- Mudrack & Naughton, 2001), and workaholism and
aholism and job satisfaction. Additionally, the fact work stress (e.g., Burke, 2000a; Burke & MacDer-
that we did not find a clear relation between worka- mid, 1999). Results from our study also indicate that
holism and job satisfaction mirrors the results of workaholics experience more work–life imbalance
Bonebright et al. (2000). Perhaps this is due to the and less life satisfaction than nonworkaholics. To the
nature of the items that were used to assess overall authors’ knowledge, this has not been previously
job satisfaction. The items that constitute the JIG explored in empirical studies on workaholism. One
scale (e.g., ideal, better than most, makes me content) should also note that hours worked alone do not
are personal in nature, thereby creating a strong po- indicate a workaholic. Many external reasons, such
tential for individual differences in overall job as the need for money or organizational culture, may
satisfaction. account for the long work hours. Therefore, perhaps
Though not surprising, another observation is the workaholics are (in part) those individuals who are
slightly higher magnitude of correlation coefficients intrinsically motivated to work long hours because of
for the employee’s self-label of workaholism com- an inability to disengage. In this way, workaholics
pared to the acquaintance rating. For example, the may be somewhat similar to Type A personality types
correlation between self-label and job involvement (see Seybold & Salomone, 1994, for more discussion
was 0.57 versus 0.46 for the acquaintance’s rating of on the connection between Type A and workaholism).
the employee as a workaholic. Two possible expla- To create effective intervention and treatment pro-
nations are offered. First, perhaps employees rated grams, it is imperative that the notion of workaholism
themselves higher on workaholism because they as a syndrome be taken into account by mental health
wanted to present themselves to others in a better professionals and career counselors. A dimensional
WORKAHOLISM AS A SYNDROME 61

focus enables practitioners and their clients to exam- post-traumatic stress syndrome in Vietnam era veterans.
ine specific symptoms of workaholism instead of the Journal of Clinical Psychology, 41, 152–156.
Burke, R. J. (2000a). Workaholism in organizations: Psy-
global construct (Burke, 2001c). For example, in
chological and physical well-being consequences. Stress
terms of work–life balance, Burke (2001d) notes that Medicine, 16, 11–16.
it is imperative that “workplace values encourage Burke, R. J. (2000b). Workaholism and extra-work satis-
more balanced priorities and healthier lifestyles to factions. The International Journal of Organizational
support those workaholics [who] want to change their Analysis, 7(4), 352–364.
behaviors” (p. 643). Burke, R. J. (2001a). Predictors of workaholism compo-
nents and behaviors. International Journal of Stress
Management, 8(2), 113–127.
Burke, R. J. (2001b). Spence and Robbins’ measures of
Limitations and Future Research workaholism components: Test–retest stability. Psycho-
logical Reports, 88, 882– 888.
One of the main limitations of the study was that Burke, R. J. (2001c). Workaholism components, job satis-
current marital/relationship status, number of chil- faction, and career progress. Journal of Applied Social
dren, and management level were unknown. Perhaps Psychology, 31(11), 2339 –2356.
stronger relations are attained between workaholism Burke, R. J. (2001d). Workaholism in organizations: The
role of organizational values. Personnel Review, 30(6),
and such variables as work stress and work–life im- 637– 645.
balance for managers who are married with children. Burke, R. J., & MacDermid, G. (1999). Are workaholics job
Based on several grounds, McMillan et al. (2002) satisfied and successful in their careers? Career Devel-
suggest that Spence and Robbins’s (1992) scale war- opment International, 4(5), 277–282.
rants further analysis. Once we acquire a thorough Cranny, C. J., Smith, P. C., & Stone, E. F. (Eds.). (1992).
Job satisfaction. New York: Lexington.
understanding of what constitutes the workaholic, Dana, K., & Griffin, R. W. (1999). Health and well-being in
future studies must focus on developing a valid, the workplace. Journal of Management, 25, 357–384.
reliable scale that identifies workaholics. In addition, Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S.
attempts should be made to further examine types of (1985). The satisfaction with life scale. Journal of Per-
workaholics in different occupations and the effects sonality Assessment, 49, 71–75.
Fisher, G. G. (2001). Work/personal life balance: A con-
various types have on organizations. Such research struct development study. Unpublished doctoral disserta-
may shed light on the conflicting conclusions that tion, Bowling Green State University.
exist in both the clinical and practitioner literature Flowers, C. P., & Robinson, B. (2002). A structural and
pertaining to workaholism (e.g., Burke, 2001d). Fur- discriminant analysis of the work addiction risk test.
thermore, the ability to generalize between cultures Educational and Psychological Measurement, 62(3),
517–526.
and national origins was limited given that most
Furnham, A., Brewin, C. R., & O’Kelly, H. (1994). Cogni-
respondents were of Caucasian American decent. tive style and attitudes to work. Human Relations, 47,
Therefore, future studies should sample from differ- 1509 –1521.
ent populations to explore whether workaholism is a Harpaz, I., & Snir, R. (2003). Workaholism: Its definition
universal concept. Finally, empirical research on and nature. Human Relations, 56(3), 291–319.
Haubrich, W. S. (1984). Medical meanings: A glossary of
workaholism and gender has been scant; therefore,
word origins. San Diego, CA: Harcourt Brace Jovanovich.
future studies must investigate the importance of Ironson, G. H., Smith, P. C., Brannick, M. T., Gibson,
gender roles in shaping workaholic behaviors. The W. M., & Paul, K. B. (1989). Construction of a job in
ability to distinguish gender differences will allow general scale: A comparison of global, composite and
for better predictions of workaholism in the work- specific measures. Journal of Applied Psychology, 74,
place; this will assist organizations in recognizing 1– 8.
Kanungo, R. N. (1982). Measurement of job and work
workaholism and taking an appropriate plan of involvement. Journal of Applied Psychology, 67,
action. 341–349.
Kaser-Boyd, N. (2004). Battered woman syndrome: Clini-
cal features, evaluation, and expert testimony. In B. J.
References Cling (Ed.), Sexualized violence against women and chil-
dren: A psychology and law perspective (pp. 41–70).
Aldenderfer, M. S., & Blashfield, R. K. (1984). Cluster New York: Guilford Press.
analysis. Thousand Oaks, CA: Sage. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and
Bonebright, C. A., Clay, D. L., & Ankenmann, R. D. coping. New York: Springer.
(2000). The relationship of workaholism with work–life Macpherson, G. (2004). Black’s medical dictionary. Lan-
conflict, life satisfaction, and purpose in life. Journal of ham, MD: Scarecrow Press.
Counseling Psychology, 47(4), 469 – 477. Maslach, C. (1982). Understanding burnout: Definitional
Burke, H. R., & Mayer, S. (1985). The MMPI and the issues in analyzing a complex phenomenon. In W. S.
62 AZIZ AND ZICKAR

Paine (Ed.), Job stress and burnout (pp. 29 – 40). Beverly Schaef, A. W., & Fassel, D. (1988). The addictive organi-
Hills, CA: Sage. zation. San Francisco: Harper Row.
McMillan, L. H. W., Brady, E. C., O’Driscoll, M. P., & Schor, J. B. (1991). The overworked American. New York:
Marsh, N. V. (2002). A multifaceted validation study of Basic Books.
Spence and Robbins’ (1992) workaholism battery. Jour- Scott, K. S., Moore, K. S., & Miceli, M. P. (1997). An
nal of Occupational and Organizational Psychology, 75, exploration of the meaning and consequences of worka-
357–368. holism. Human Relations, 50(3), 287–314.
Morrow, P. C., & Crum, M. R. (1998). The effects of Seybold, K. C., & Salomone, P. R. (1994). Understanding
perceived and objective safety risk on employee out- workaholism: A review of causes and counseling ap-
comes. Journal of Vocational Behavior, 53, 300 –313. proaches. Journal of Counseling and Development, 73,
Mudrack, P. E., & Naughton, T. J. (2001). The assessment 410 – 418.
of workaholism as behavioral tendencies: Scale develop- Snir, R., & Zohar, D. (2000). Workaholism: Work-addiction
ment and preliminary empirical testing. International or workphilia? Paper presented at the June Internal Con-
Journal of Stress Management, 8(2), 93–111. ference on Psychology: Psychology after the year 2000,
Oates, W. (1971). Confession of a workaholic. New York: University of Haifa, Israel.
World. Sparks, K., Cooper, C., Fried, Y., & Shirom, A. (1997). The
Pinder, C. C. (1998). Work motivation in organizational effects of hours of work on health: A meta-analytic
behavior. Saddle River, NJ: Prentice Hall. review. Journal of Occupational and Organizational
Porter, G. (1996). Organizational impact of workaholism: Psychology, 70, 391– 408.
Suggestions for researching the negative outcomes of Spector, P. E. (1994). Using self-report questionnaires in
excessive work. Journal of Occupational Health Psy- OB research: A comment on the use of a controversial
chology, 1(1), 70 – 84. method. Journal of Organizational Behavior, 15, 385–392.
Porter, G. (2001). Workaholic tendencies and the high po- Spence, J. T., & Robbins, A. S. (1992). Workaholism:
tential for stress among co- workers. International Jour- Definition, measurement, and preliminary results. Jour-
nal of Stress Management, 8(2), 147–164. nal of Personality Assessment, 58(1), 160 –178.
Reeve, C. L., & Smith, C. S. (2001). Refining Lodahl and Stanton, J. M., Balzer, W. K., Smith, P. C., Parra, L. F., &
Kejner’s job involvement scale with a convergent evi- Ironson, G. (2001). A general measure of work stress:
dence approach: Applying multiple methods to multiple The stress in general (SIG) scale. Educational and Psy-
samples. Organizational Research Methods, 4(2), chological Measurement, 61(5), 866 – 888.
91–111.
Robinson, B. E. (1998). Chained to the desk: A guidebook
for workaholics, their partners and children, and the
Received December 16, 2004
clinicians who treat them. New York: New York Uni- Revision received May 5, 2005
versity Press. Accepted June 16, 2005 y

You might also like