Running Head: Workaholism Profiles
Workaholism Profiles: Associations with Determinants, Correlates, and Outcomes
Nicolas Gillet
1*
, Alexandre J. S. Morin
2
, Baptiste Cougot
1
, & Marylène Gagné
3
1
Université François-Rabelais de Tours, France
2
Concordia University, Canada
3
University of Western Australia, Australia
* Corresponding author
Nicolas Gillet,
Université François-Rabelais de Tours,
UFR Arts et Sciences Humaines,
Département de psychologie,
3 rue des Tanneurs, 37041 Tours Cedex 1, France.
E-mail: nicolas.gillet@univ-tours.fr
This is the prepublication version of the following manuscript:
Gillet, N., Morin, A. J. S., Cougot, B., & Gagné, M. (in press). Workaholism profiles: Associations
with determinants, correlates, and outcomes. Journal of Occupational and Organizational Psychology.
Early view. doi: 10.1111/joop.12185.
© 2017. This paper is not the copy of record and may not exactly replicate the authoritative document
published in Journal of Occupational and Organizational Psychology.
Abstract
The present series of studies examines how the two dimensions of workaholism (working excessively
and compulsively) combine within different profiles of workers. This research also documents the
relations between these workaholism profiles and a series of correlates (psychological need thwarting)
and adaptive and maladaptive work outcomes. In addition, this research investigates the role of
emotional dissonance sychosocial safety climate
(Study 1, n = 465), as well as job demands, resources, and perfectionism (Study 2, n = 780) in the
prediction of profile membership. Latent profile analysis revealed four identical workaholism profiles
in both studies. In Study 1, emotional dissonance predicted a higher likelihood of membership in the
Very High, Moderately High, and Moderately Low profiles relative to the Very Low profile. In
contrast, Study 2 revealed a more diversified pattern of predictions. In both studies, levels of need
thwarting were the highest in the Very High and Moderately High profiles, followed by the
Moderately Low profile, and finally by the Very Low profile. Finally, in both studies, the most
desirable outcomes levels (e.g., lower levels of work-family conflict and emotional exhaustion, and
higher levels of perceived health) were associated with the Very Low profile, followed by the
Moderately Low profile, then by the Moderately High profile, and finally by the Very High profile.
Keywords: Workaholism profiles; Working excessively and compulsively; Job demands and
resources; Health; Psychological needs
Workaholism Profiles 1
added that workaholics tend to allocate as much time as
possible to work. Indeed, scholars (e.g., Schaufeli, Bakker, van der Heijden, & Prins, 2009; Schaufeli,
Shimazu, & Taris, 2009) generally propose to differentiate the behavioral (i.e., being hardworking,
spending a great deal of time in work activities, neglecting other spheres of life) and cognitive (i.e.,
being obsessed with work, thinking compulsively about work) facets of workaholism. Recently,
research has started to examine how these two facets combine within specific individuals (Kravina,
Falco, Girardi, & De Carlo, 2010; Schaufeli, Bakker et al., 2009). Variable-centered analyses,
designed to test how specific variables relate to other variables are able to tests for interactions among
predictors (i.e., if the effect of a predictor differs as a function of another variable). However, through
their focus on the identification of subgroups characterized by distinct configurations, or profiles, on a
set of variables, person-centered analyses are more naturally suited to the consideration of the joint
effect of variable combinations. The present research extends prior studies of workaholism profiles
(e.g., Buelens & Poelmans, 2004) by (1) simultaneously and exclusively considering the two
behavioral (working excessively) and cognitive (working compulsively) facets of workaholism, rather
than relying on a mixture of indicators conflating workaholism facets with other variables; (2)
assessing the construct validity of the workaholism profiles through the consideration of correlates,
predictors, and a wide range of attitudinal and health outcomes; and (3) relying on state-of-the art
latent profile analyses (LPA) rather than cluster analyses which have been criticized (see Meyer &
Morin, 2016), particularly for research involving covariates. Meyer and Morin (2016) emphasize the
importance of clear a priori specifications of which covariates can be assumed to predict profile
membership (predictors), to be predicted by it (outcomes), or to relate to the profiles with no
assumption of directionality (correlates). However, although our treatment of covariates as correlates,
determinants, and outcomes is theoretically anchored (Clark, Michel, Zhdanova, Pui, & Baltes, 2016;
Schaufeli, Bakker et al., 2009) and necessary for methodological reasons, our cross-sectional design
precludes interpretations regarding the directionality of the associations.
Workaholism
Workaholism can be seen as an addiction to work (e.g., Schaufeli, Shimazu, & Taris, 2009;
Spence & Robbins, 1992), leading to preoccupations and compulsions regarding work, loss of self-
control, and continued work engagement despite negative outcomes (Ng, Sorensen, & Feldman,
2007). Workaholic behaviors thus involve an excessive involvement in work that goes well beyond
normal job requirements. Workaholics are also constantly obsessed with work, even when they are not
working (Schaufeli, Bakker et al., 2009). These two behavioral and cognitive facets of workaholism
(working excessively and compulsively) are not mutually exclusive, but rather seen as complementary
and co-existing to various degrees within individuals (Clark et al., 2016). It thus follows that
workaholism cannot be reduced to either of these two components. However, many studies have
shown that the two dimensions of workaholism tend to be positively and moderately to strongly
related (e.g., Huyghebaert et al., 2016), leaving as an open research question whether these two forms
of workaholism really represent distinct components.
So far, the predictive validity of working excessively and compulsively has been documented in
relation to a variety of work outcomes in the context of variable-centered studies (for a meta-analysis,
see Clark et al., 2016). For instance, working compulsively and excessively both share positive
relations with s of emotional exhaustion, presenteeism, and work-family conflict, as
well as negative relations with happiness and performance (Huyghebaert et al., 2016; Schaufeli,
Bakker et al., 2009). However, limited research has looked at the combined effects of these two
dimensions of workaholism on these important work-related outcomes. Interestingly, emerging
person-centered research suggests that employees characterized by a high level on both dimensions
tend to experience fewer sleeping hours and poorer sleep quality on weekdays and weekends, relative
to those scoring high on only one dimension (Salanova et al., 2016).
Workaholism Profiles
Variable-centered approaches examine relations occurring between variables, on the average, in a
specific sample. In contrast, person-centered approaches identify homogeneous subgroups (or profiles)
of workers sharing similar configurations of workaholism components. Therefore, the person-centered
approach provides a complementaryyet uniquely informativeperspective on the same questions,
focusing on individual profiles rather than on specific relations among variables (Marsh, Lüdtke,
Workaholism Profiles 2
Trautwein, & Morin, 2009; Morin & Wang, 2016). In particular, person-centered analyses are
naturally suited to the verification of how the two types of workaholism will be combined among
different profiles of employees, and the relative consequences of membership into these various
profiles. However, little person-centered research has been conducted on workaholism. Among the
few available studies, Salanova, Del Libano, Llorens, and Schaufeli (2014) examined different profiles
of wellbeing at work and identified a workaholic profile corresponding to employees characterized by
moderate to high levels of energy, challenge, skills and identification, and by low levels of pleasure.
Other investigations relied on a mixture of workaholism dimensions and additional constructs
(Buelens & Poelmans, 2004; Spence & Robbins, 1992), making it impossible to identify workaholism
configurations occurring independently from these additional dimensions.
Among the few relevant investigations, and despite some variations, four workaholism profiles
have typically been identified (Kravina et al., 2010; Salanova et al., 2016): high levels of working
compulsively and excessively (HC-HE), high levels of working compulsively and low levels of
working excessively (HC-LE), low levels of working compulsively and high levels of working
excessively (LC-HE), and low levels of working compulsively and excessively (LC-LE). For instance,
Schaufeli, Bakker et al. (2009) identified these four workaholism profiles and showed the HC-HE
profile to be associated with the most unfavorable outcomes in terms of mental health (i.e., burnout,
happiness, and recovery) and organizational behaviors (i.e., presenteeism and performance) (also see
Kravina et al., 2010). The reliance on cluster analyses is a key limitation of these studies. Indeed,
cluster analyses have been previously criticized as showing a greater level of reactivity to the retained
clustering algorithm, relying on rigid statistical assumptions, forcing the exact assignment of
participants into a single profile (rather than taking into account participant likelihood of
membership in all profiles based on their prototypical similarity), and making it impossible to directly
incorporate covariates into the model as predictors, correlates, or outcomes (for details, see Meyer &
Morin, 2016; Morin, Morizot, Boudrias, & Madore, 2011).
The first purpose of the present research was thus to identify workaholism profiles using LPA,
while simultaneously and exclusively considering the two facets of workaholism proposed by
Schaufeli, Shimazu, and Taris (2009). To the best of our knowledge, no research has yet relied on
LPA to identify workaholism profiles. Still, in line with past cluster analytic studies, it was expected
that a relatively small number of profiles (i.e., between four and five) corresponding to the four
previously identified configurations (1. HC-HE; 2. LC-LE; 3. HC-LE; 4. LC-HE) would be identified.
Determinants of Workaholism Profiles
Little research has investigated the structural determinants of workaholism profiles (Caesens,
Stinglhamber, & Luypaert, 2014). The job demands-resources model (Bakker & Demerouti, 2007;
Bakker, Demerouti, & Euwema, 2005) classifies job characteristics in two general categories, job
demands and job resources, providing an overarching model applicable to any work contexts. Job
demands refer to those aspects of a job that require sustained physical and/or psychological effort and
are assumed to be associated with a variety of physiological and/or psychological costs. In contrast,
job resources help employees to achieve workrelated goals, thus helping to balance the costs
associated with job demands and to stimulating personal development. Based on the job demands-
resources model, Schaufeli, Bakker et al. (2009) tested the relations between job demands (work
overload, mental demands, emotional demands) and resources (social support from colleagues,
supervisory coaching, opportunities to learn), and workaholism profiles. Their results showed that
higher levels of job demands and lower levels of job resources predicted a higher likelihood of
membership into the HC-HE profile. Similarly, Kravina et al. (2010) showed that higher levels of time
pressure were associated with a higher likelihood of membership into the HC-HE profile. Also based
on the job demands-resources model, Molino, Bakker, and Ghislieri (2016) recently examined the
determinants of workaholism. Results revealed that job demands (i.e., workload, cognitive demands,
emotional demands, and customer-related social stressors) were positively related to workaholism. In
addition, job resources (job security and opportunities for development) buffered the relations between
job demands and workaholism. In sum, past studies showed that job demands and resources were
significant determinants of workaholism. In the present research, we also examine the role of various
job demands and resources in the prediction of the likelihood of membership into workaholism
profiles, and extend these prior investigations by considering a more extensive set of indicators of job
demands (emotional dissonance in Study 1, role ambiguity in Study 2) and resources (psychosocial
Workaholism Profiles 3
safety climate in Study 1, independence in Study 2).
Scott, Moore, and Miceli (1997) also suggested that some traits might be involved in the
emergence of workaholism. Many others have similarly considered that workaholism may be
influenced by personal characteristics (for a meta-analysis, see Clark et al., 2016) such as self-esteem
(Ng et al., 2007) and perfectionism (Clark, Lelchook, & Taylor, 2010). More generally, dispositional
traits are known to play a major role in the emergence of addictions (e.g., Eysenck, 1997). Still,
irrespective of the fact that workaholism represents a form of addiction to work (Schaufeli, Shimazu,
& Taris, 2009; Spence & Robbins, 1992), very little attention has been paid to the effects of
perfectionism on working compulsively and excessively. Thus, to increase our understanding of the
role of individual characteristics, we also examine the links between self-oriented and socially
prescribed perfectionism and the likelihood of membership into the various profiles in Study 2.
Study 1: Emotional Dissonance and Perceptions of the Psychosocial Safety Climate
Emotional dissonance reflects a discrepancy between the emotions one feels and the emotions one
is required to display (Holman, Chissick, & Totterdell, 2002). Emotional dissonance is experienced as
a role conflict, leading to an unpleasant state of tension due to the inability to display authentic
feelings (Hülsheger & Schewe, 2011). Emotional dissonance is linked to feelings that they
have not functioned optimally or in accordance with their values, and presents a known association
ruminate about their work and with their levels of working
compulsively (Sonnentag & Bayer, 2005). Emotional dissonance is also an important form of job
demand (Zapf, 2002), because it requires effortful regulatory processes (Baumeister, Bratslavsky,
Muraven, & Tice, 1998) likely to eir tasks, and increase their
feelings of work overload. Emotional dissonance may thus directly increase the time spent at work as
employees tend to catch up on what they perceive to be an unreasonable workload, thus leading to
working excessively (Zohar, Tzischinski, & Epstein, 2003). Although no research has yet analyzed the
association between emotional dissonance and workaholism, Molino et al. (2016) showed that
emotional demands were positively linked to workaholism. In line with these results, we hypothesized
that emotional dissonance would predict a greater likelihood of membership in the HC-HE profile.
Psychosocial safety climate is defined as tection of
worker psychological Dollard & Bakker, 2010, p. 580). Psychosocial safety
climate stems emerges when organizations support stress prevention through involvement and
commitment, and clearly communicate that employee psychological health and safety is as important
as productivity (Hall, Dollard, & Coward, 2010). Recent research has shown that psychosocial safety
climate was negatively correlated with job demands, such as work pressure (Bailey, Dollard,
McLinton, & Richards, 2015; Law, Dollard, Tuckey, & Dormann, 2011). This result suggest that
organizations with high psychosocial safety climate might implement more efficient built-in workload
management procedures, thus possibly helping to reduce workaholism (Schaufeli, Bakker et al., 2009).
We thus hypothesized that perceptions of the psychosocial safety climate would be associated with a
higher likelihood of membership in the LC-LE profile.
Study 2: Job Demands and Resources, and Perfectionism
In Study 1, we considered the role of one type of job demand (emotional dissonance) and resource
(psychosocial safety climate) in the prediction of workaholism profiles. In Study 2, we extend this
investigation by considering a more extensive set of indicators of job demands (mental and emotional
load, role ambiguity) and resources (support from colleagues, hierarchical support, independence). In
line with aforementioned results (e.g., Kravina et al., 2010; Molino et al., 2016; Schaufeli, Bakker et
al., 2009), we hypothesized that job demands would predict a higher likelihood of membership in the
HC-HE profile. Arguably, the more important job demands are, the more workers may be tempted to
invest efforts and energy to meet these demands, possibly leading them to work excessively
(Schaufeli, Taris, & van Rhenen, 2008). Important job demands may also generate anxiety regarding
ability to meet them, leading employees to spend more time ruminating about work, possibly
leading them to work compulsively (Huyghebaert et al., 2016).
According to the conservation of resources theory, support from colleagues, hierarchical support,
and independence are powerful resources to help maintain worker (Hobfoll, 1989) and
their ability to manage job demands effectively (Spurk, Hirschi, & Kauffeld, 2016). Employees who
feel supported by their supervisor and colleagues may not come to rely on destructive forms of work
overinvestment compared to those who feel more isolated at work (Spurk et al., 2016). Moreover,
Workaholism Profiles 4
supervisor support have been found to be more frequently associated with a work environment where
employees are not pushed to work extra hours, possibly leading to a reduced risk of workaholism
(Mazzetti, Schaufeli, Guglielmi, & Depolo, 2016). Finally, workers who feel sufficiently independent
at work may dispose of a greater level of latitude to deal with their job demands within regular work
schedules without feeling compelled to go overboard (Molino et al., 2016). Independence provides
workers with opportunities to use their strengths without feeling that their personal resources are
challenged or drained, or that they need to spend a great deal of time at work or to obsess about it to
avoid losing these resources. We thus hypothesized that job resources would predict a higher
likelihood of membership into the LC-LE profile (Schaufeli, Bakker et al., 2009).
Study 2 also focuses on the relations between self-oriented and socially prescribed perfectionism
and the likelihood of membership into the various profiles. Self-oriented perfectionism is an internal
drive to uphold exceedingly high personal standards and to criticize oneself harshly. Socially
prescribed perfectionism comprises beliefs that others have high standards for oneself that must be met
to achieve social acceptance (Hewitt & Flett, 1991). Because perfectionists are driven by strong
strivings for perfection, it would be logical to assume that self-oriented and socially prescribed
perfectionism would foster these two workaholism components. This link is supported by evidence
showing that global perfectionism was associated with higher levels of workaholism (Clark et al.,
2016). Taris, van Beek, and Schaufeli (2010) further showed that the effect of socially prescribed
perfectionism on global workaholism was stronger than that of self-oriented perfectionism. These
results suggest that self-oriented and socially prescribed perfectionism should be important in the
prediction of the likelihood of membership into the HC-HE profile. However, in line with Taris et al.
(2010) and because socially prescribed perfectionism appears more detrimental than self-oriented
perfectionism (Flett, Hewitt, & Heisel, 2014), we leave as an open research question whether the two
forms of perfectionism would differentially relate to the workaholism profiles.
Outcomes of Workaholism Profiles
To support a substantive interpretation of latent profiles as meaningful and relevant, it is critical to
demonstrate that they relate to key outcomes and that they can be reliably replicated across samples
(Marsh et al., 2009; Meyer & Morin, 2016; Morin & Wang, 2016). The research was specifically
designed to address this issue, allowing for a direct test of whether the profiles, as well as their
relations with outcomes and correlates, would replicate across samples. We now turn our attention to
the outcomes, which were selected to be both complementary and similar across studies. Specifically,
attitudinal and health outcomes were assessed in the present series of studies (i.e., work-family
conflict, emotional exhaustion, perceived stress, turnover intentions, psychological detachment, job
satisfaction, and perceived health in Study 1, as well as work-family conflict, emotional exhaustion,
perceived health, and life satisfaction in Study 2). First, we studied the effects of workaholism profiles
on various work outcomes previously documented to be associated with workaholism (e.g., emotional
exhaustion) across a variety of cultural samples (e.g., Dutch: Schaufeli, Bakker et al., 2009; Italian:
Kravina et al., 2010; Spanish: Salanova et al., 2016). Second, we also considered outcomes already
found to be related to workaholism, but only in the context of past variable-centered research in order
to see whether these results would generalize to but person-centered studies (e.g., work-family
conflict). Third and finally, to complement prior research, we considered three potential outcomes of
workaholism profiles not assessed in past studies (i.e., turnover intentions, psychological detachment,
and life satisfaction).
Prior research has documented associations between workaholism profiles and work outcomes.
Schaufeli, Bakker et al. (2009) showed that the HC-HE profile reported the highest levels of burnout
and presenteeism, and the lowest levels of recovery, happiness, and performance. In contrast, the LC-
LE profile reported the lowest levels of burnout and presenteeism, and the highest levels of recovery,
happiness, and performance. In addition, their results also showed that the HC-LE and LC-HE profiles
did not differ in terms of recovery, happiness, presenteeism, and performance. Kravina et al. (2010)
found that the HC-HE profile presented the highest levels of psychological strain and emotional
exhaustion, while the HC-LE and LC-HE profiles did not differ on work satisfaction, emotional
instability, and compliance. Salanova et al. (2016) showed that the HC-HE profile had lower sleep
quantity and quality, as well as greater levels of alcohol use, and risk of cardiovascular difficulties.
These relations may be explained by the conservation of resources theory (Hobfoll, 1989).
Workaholics spend excessive amounts of time and energy on their work, leading to a state of extreme