ArticlePDF Available

Work engagement and burnout: real, redundant, or both? A further examination using a bifactor modelling approach


Abstract and Figures

Researchers have often debated whether burnout and work engagement are truly different concepts, or whether they are opposite poles of the same construct and therfore redundant. Recent perspectives postulate that they are both real and redundant. In this paper we examine these three competing views using a bifactor modelling approach. A sample of 1787 Argentine employees completed the Utrecht Work Engagement Scale (UWES), the Maslach Burnout Inventory–General Survey (MBI-GS), and other work-related measures. Bifactor analysis showed that at the construct level, work engagement and burnout are not adequately represented by a single well-being dimension, indicating that they are distinct constructs. At the dimension level, vigour and exhaustion could not be accounted for by a general energy factor, meaning that these constructs are distinct rather than direct opposites of one energy continuum. In contrast, dedication and cynicism were substantially explained by a single identification factor, suggesting that they represent each other’s opposite along an identification continuum. However, dedication and cynicism displayed a unique pattern of relationships with different external variables, implying that they are also real constructs. Collectively, the findings indicate that each of the competing views on the work engagement-burnout relationship has its merits. Implications for conceptualization and measurement are discussed.
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
European Journal of Work and Organizational Psychology
ISSN: (Print) (Online) Journal homepage:
Work engagement and burnout: real, redundant,
or both? A further examination using a bifactor
modelling approach
Mario A. Trógolo , Luis P. Morera , Estanislao Castellano , Carlos Spontón &
Leonardo A. Medrano
To cite this article: Mario A. Trógolo , Luis P. Morera , Estanislao Castellano , Carlos Spontón &
Leonardo A. Medrano (2020): Work engagement and burnout: real, redundant, or both? A further
examination using a bifactor modelling approach, European Journal of Work and Organizational
Psychology, DOI: 10.1080/1359432X.2020.1801642
To link to this article:
Published online: 02 Aug 2020.
Submit your article to this journal
View related articles
View Crossmark data
Work engagement and burnout: real, redundant, or both? A further examination
using a bifactor modelling approach
Mario A. Trógolo, Luis P. Morera, Estanislao Castellano, Carlos Spontón and Leonardo A. Medrano
Healthy Organizations Institute, Universidad Siglo 21, Córdoba, Argentina
Researchers have often debated whether burnout and work engagement are truly dierent concepts, or
whether they are opposite poles of the same construct and therfore redundant. Recent perspectives
postulate that they are both real and redundant. In this paper we examine these three competing views
using a bifactor modelling approach. A sample of 1787 Argentine employees completed the Utrecht
Work Engagement Scale (UWES), the Maslach Burnout Inventory–General Survey (MBI-GS), and other
work-related measures. Bifactor analysis showed that at the construct level, work engagement and
burnout are not adequately represented by a single well-being dimension, indicating that they are
distinct constructs. At the dimension level, vigour and exhaustion could not be accounted for by a
general energy factor, meaning that these constructs are distinct rather than direct opposites of one
energy continuum. In contrast, dedication and cynicism were substantially explained by a single identi-
cation factor, suggesting that they represent each other’s opposite along an identication continuum.
However, dedication and cynicism displayed a unique pattern of relationships with dierent external
variables, implying that they are also real constructs. Collectively, the ndings indicate that each of the
competing views on the work engagement-burnout relationship has its merits. Implications for con-
ceptualization and measurement are discussed.
Received 4 May 2020
Accepted 22 July 2020
Work engagement; burnout;
employee well-being;
bifactor modelling; structural
equation modelling
Since its emergence at the beginning of the century, the con-
cept of work engagement has attracted burgeoning interest
among academics and professionals. The amount of scientic
research on this subject has increased rapidly, with thousands
of publications produced each year (Schaufeli & Salanova,
2011). However, alongside the mounting popularity of the
concept, the conceptual and empirical distinctiveness of work
engagement and other long-standing organizational con-
structs, particularly burnout, has often been questioned (Cole
et al., 2012; Newman & Harrison, 2008). At the construct level,
the debate is about whether work engagement represents a
distinct construct from burnout and is thus real, or whether it is
merely the antipode (i.e., absence) of burnout and, thus, redun-
dant (Schaufeli & De Witte, 2017a). At a more specic level,
disagreement persists regarding whether the core elements of
work engagement (vigour and dedication) and burnout
(exhaustion and cynicism) represent opposite ends of the
same continuum (i.e., energy -vigour and exhaustion and iden-
tication -dedication and cynicism) or whether they are in fact
distinct (Nimon & Shuck, 2019). On the basis of the dialectical
perspective on the work engagement-burnout relationship
(Leon et al., 2015), more recent approaches have proposed
that the two constructs are both real and redundant (Leiter &
Maslach, 2017; Schaufeli & De Witte, 2017b).
Using a bifactor modelling approach (Reise, 2012), the goal
of this paper is to examine the extent to which work engage-
ment and burnout, both at the construct and dimension level,
can be considered as real (i.e., distinct constructs), redundant (i.
e., polar opposites of the same construct), or both. In doing so,
the current study builds on previous research (e.g., Demerouti
et al., 2010; González-Romá et al., 2006; Taris et al., 2017) and
contributes to the literature in at least two ways. Firstly, pre-
vious variable-centred studies addressing the opposite versus
distinct issue through a conrmatory factor-analytic approach
have been limited to the assessment of higher-order factor
models. As we discuss in the next section, there are statistical
and substantive reasons that may seriously limit the usefulness
of higher-order models for testing competing views on the
conceptualization of work engagement and burnout. The bifac-
tor model overcomes these limitations and allows for a more
complete and direct examination of the matter. Secondly, past
research examining the work engagement-burnout relation-
ship has focused on the opposite versus distinct – i.e., real
versus redundant – view. To the best of our knowledge, the
“real and redundant” view has so far not been assessed directly.
Thus, for the rst time we address all theoretical perspectives
simultaneously. The bifactor model has been recommended as
a useful technique for examining the relative inuence of gen-
eral and several orthogonal specic factors on item responses
(Flores Kanter et al., 2018). As such, bifactor model allows
researchers to examine the extent to which items pertaining
to a scale or dierent subscales (e.g., exhaustion and vigour
items in the present study) are independently inuenced by a
general factor (e.g., energy) and by specic factors (exhaustion
and vigour), thereby providing valuable information to judge
whether items are essentially unidimensional and reect the
same construct (i.e., a strong general factor and weak specic
factors), whether they tap distinct constructs (i.e., a weak
CONTACT Luis P. Morera
© 2020 Informa UK Limited, trading as Taylor & Francis
general factor and strong specic factors), or whether they
simultaneously reect both (Bonifay et al., 2017). Accordingly,
the bifactor model represents a particularly useful framework
for testing competing perspectives and brings clarity into the
debate about the relationship between work engagement and
Burnout was dened by Maslach and Jackson (1986) as a work-
related stress syndrome characterized by emotional exhaus-
tion, depersonalization and reduced personal accomplishment,
potentially occurring in any individual working in some capa-
city with people. Emotional exhaustion refers to feelings of
being overextended and depleted of one’s emotional resources
due to interpersonal demands; depersonalization involves a
distant negative attitude towards others, particularly the reci-
pients of one’s services; and reduced personal accomplishment
refers to the tendency to negatively evaluate one’s compe-
tences and achievements in the work. Based on this denition,
the authors developed the Maslach Burnout Inventory (MBI;
Maslach & Jackson, 1981) for assessing burnout in human
services occupations.
Over time, however, it has become clear that burnout
can occur in all occupations, not just in those involving
work with people (Maslach & Leiter, 1997). Consequently,
the concept of burnout and its dimensions were later
amended to include non-human services occupations.
Exhaustion was accordingly redened as feelings of fatigue
and being overextended without reference to other people
as the source of the fatigue. The term depersonalization was
replaced by cynicism, which reects a distant attitude
towards work in general and not necessarily towards the
people with whom one works. Finally, reduced personal
accomplishment was relabelled as reduced professional e-
cacy, referring to the perceived inability to perform work
ecaciously, including both social and non-social aspects of
work (Maslach et al., 2001). In accordance with this broader
denition, Schaufeli et al. (1996) proposed a new version of
the MBI, namely the MBI-General Survey (MBI-GS), for asses-
sing burnout symptoms irrespective of occupation.
Despite the three-dimensional concept of burnout, var-
ious studies have shown that exhaustion and cynicism are
its core dimensions (Schaufeli & Taris, 2005; Halbesleben &
Demerouti, 2005; Schaufeli et al., 2017) while professional
self-ecacy has been considered more as a personality trait
(Coders & Dougerthy, 1993), an antecedent (Ventura et al.,
2015), a consequence (Schaufeli & De Witte, 2017a), or a
moderator of the inuence of job stressors on burnout (Gil-
Monte et al., 2008). Schaufeli and Salanova (2007) argued
that professional “inecacy” rather than (reduced) profes-
sional “ecacy” constitutes the third component of burnout;
however, recent ndings did not nd support for the pro-
fessional inecacy scale (Spontón et al., 2019). In view of
the divergent role of the professional ecacy, we focused
our study on the core burnout dimensions.
Work engagement
Despite the lack of agreement on the conceptualization of work
engagement (see Schaufeli & Bakker, 2010), the most cited
denition refers to a positive, fulling, work-related aective-
cognitive state of mind characterized by vigour, dedication and
absorption (Schaufeli et al., 2002). Vigour implies high levels of
energy while working, the willingness to invest eort in one’s
work and perseverance in the face of diculties. Dedication
involves a sense of enthusiasm, inspiration, pride and being
challenged by one’s work. Lastly, absorption is characterized by
being so deeply immersed in one’s work that time ies and one
has diculty disconnecting. On this basis Schaufeli and Bakker
(2003) developed the Utrecht Work Engagement Scale (UWES),
the most widely used work engagement measure among
researchers (Schaufeli & Salanova, 2011).
Notwithstanding this three-dimensional concept of work
engagement, several authors agree that vigour and dedication
represent its core dimensions (Schaufeli & Bakker, 2004; Llorens
et al., 2007) while the role of the third dimension – absorption –
is less clear. In particular, it has been suggested that absorption
represents a more independent construct (Innanen et al., 2014).
Others authors pointed out that absorption may represent a
consequence of engagement (Salanova et al., 2003) or a non-
exclusive feature of work engagement (Schaufeli et al., 2008). In
light of the existing doubts concerning this dimension, we have
excluded it from the present study.
Distinct states versus opposite poles perspectives:
theoretical and measurement issues
Initially, Maslach and Leiter (1997) dened work engagement
as the opposite of burnout, each representing opposite ends
of occupational well-being. As stated by the authors, indivi-
duals typically feel bursting with high energy (vigour) and
dedication at the beginning of a new job. Under stressful
conditions, however, this state may erode and convert into
low energy (exhaustion) and low dedication (cynicism).
Accordingly, exhaustion-vigour and dedication-cynicism are
direct opposites that reect the level of energy and identica-
tion with one’s work, respectively (González-Romá et al.,
2006). Work engagement is thus a positive well-being state
at work characterized by a high level of energy and strong
identication with one’s work, whereas burnout is the nega-
tive pole of employees’ well-being characterized by a low
level of energy and poor identication with one’s work.
Assuming that work engagement and burnout are each
other’s opposite implies that they are mutually exclusive
experiences (i.e., one can feel either engaged or burned-out,
but not both at the same time) and the same instrument can
therefore be used to capture both experiences (Maslach &
Leiter, 1997). Consequently, according to these authors, low
scores on MBI subscales represent burnout and high scores
are indicative of work engagement.
Schaufeli and colleagues (Schaufeli & Bakker, 2004, 2010;
Schaufeli et al., 2002) take a dierent perspective by consider-
ing work engagement as a separate construct from burnout,
rather than its direct opposite. In fact, even though the authors
agree that in conceptual terms work engagement and burnout
are “each other’s opposites” (Schaufeli & Bakker, 2004, p. 295),
they advocate that burnout and work engagement are in fact
distinct states. Hence, the absence of burnout does not neces-
sarily imply the presence of work engagement and vice versa.
By implication, work engagement cannot be adequately
assessed by opposite (i.e., low) proles in MBI scores, but rather
by using a dierent measure.
Distinct states versus opposite poles perspectives: review
of empirical evidence
Various studies have been carried out to investigate the extent
to which work engagement is distinct or opposite from burn-
out. Based on conrmatory factor analysis (CFA), Schaufeli et al.
(2002) found that a second-order model assuming that all
burnout and work engagement subscales reect an underlying
general well-being dimension did not t the data well (CFI =.83,
GFI = .87, RMSEA = .16), whereas two correlated second-order
factors assuming work engagement and burnout as dierent
but related constructs did (CFI = .92, GFI = .93, RMSEA = .10).
These factors correlated at – .47 and – .62 in dierent samples.
Similarly, Schaufeli and Bakker (2004) found that a second-
order well-being factor with work engagement and burnout
subscales as rst-order factors did not t to the data well
(CFI = .87, GFI = .90, NFI = .86 RMSEA = .10). An alternative
two second-order factor model consisting of “burnout” and
“engagement” showed an acceptable t to the data across
four dierent samples of workers (CFI = .91, GFI = .97,
NFI = .91, RMSEA = .08). In these samples, the correlation
between burnout and work engagement ranged from – .38 to
.51. Furthermore, these constructs exhibited a dierent pat-
tern of relationships with antecedents (e.g., job demands and
resources) and outcomes (e.g., health outcomes and turnover
intention), patterns which were remarkably similar across the
samples. In contrast, Taris et al. (2017) found that burnout and
work engagement could be distinguished empirically based on
second-order CFA ndings (CFI = .97, RMSEA = .05, SRMR = .06);
however, the correlation between latent burnout and work
engagement constructs (– .75) as well as between their core
dimensions (ranging from – .60 to .87) suggests that they over-
lap substantially (any correlation of |r ≥ 0.70| or |r ≥ 0.80| is
considered redundant; see Byrne et al., 2016; Carlson &
Herdman, 2012). In addition, correlations with job demands
and job resources were highly similar for burnout and work
engagement, making it hard to dierentiate between the two
constructs based on their relationship with other constructs. In
sum, evidence concerning whether work engagement and
burnout are distinct constructs or opposite ends of the same
dimension is conictive.
Other studies have focused their interest on the core dimen-
sions of work engagement and burnout. For instance,
González-Romá et al. (2006) using a scaling method based on
Item Response Theory showed that items of cynicism and
dedication as gauged by MBI-GS and UWES – were scalable
on a single underlying identication dimension in dierent
samples (Mokken’s H scalability coecients above .50), thus
representing opposite ends of one continuum. Exhaustion
and vigour items on the other hand did not form a strong
bipolar energy dimension (H coecients ranging from .36 to
.44) implying that they represent distinct constructs. Further
studies from a variable- (Demerouti et al., 2010) and person-
centred perspective (Mäkikangas et al., 2012, 2017) and,
interestingly, using a variety of burnout and work engagement
scales – yielded similar conclusions, reinforcing the idea that
dedication and cynicism are each other’s opposite, whereas
vigour and exhaustion are distinct. Finally, a meta-analysis of
50 samples conducted by Cole et al. (2012) showed that the
estimated mean true correlation between dedication-cynicism
and vigour-exhaustion was – .69, and – .43, respectively, indi-
cating that dedication and cynicism practically overlap,
whereas vigour and exhaustion do not. A subsequent analysis
revealed that the overall association of dedication and cynicism
with a common set of antecedents and outcomes was nearly
identical (mean vr = – .96) and the same held true for exhaus-
tion and vigour (mean vr = – .93), indicating that these two
pairs of constructs cannot be dierentiated based on their
nomological networks, i.e. they are redundant.
Dialectical view on work engagement and burnout
Based on the literature review outlined above, it seems clear
that the relationship between work engagement and burnout
is complex and that a denitive conclusion remains elusive, as
both perspectives distinct and opposite have received
empirical support. Recently, Leon et al. (2015) argued that
these apparently contradictory perspectives may in fact be
complementary. On the basis of dialectical theory, the authors
suggested that a plausible way to explain the inconsistent
empirical ndings regarding the relationship between work
engagement and burnout is to consider that they are both
opposite and distinct states. According to this view, while
work engagement and burnout are opposing forces they do
not negate each other (i.e., the presence of work engagement
does not necessarily imply the absence of burnout and vice
versa). For example, a very highly engaged employee may work
so hard to burned-out, whereas another employee might feel
mentally exhausted but still continue to be fully engaged with
his or her work. Accordingly, work engagement and burnout
are mutual states (i.e., paired opposites) that can coexist within
an individual (i.e. they are distinct) and result in unique out-
comes. In line with tenets of dialectical thinking,
Leiter and Maslach (2017) further proposed that work engage-
ment and burnout would be better conceptualized as both
“real and redundant” (p. 56). They are strongly negatively
and are thus opposites (redundant), but they also have unique
relationships with other constructs, meaning they are dierent
Methodological issues: CFA versus ESEM, higher-order
versus bifactor model
Along with theoretical issues surrounding the work engage-
ment-burnout relationship, there are some methodological
concerns that warrant consideration. Firstly, the estimated cor-
relation at the construct- and dimension-level reported in pre-
vious studies has been based on latent correlations derived
from CFA models. In such models, each item loads on the factor
it is designed to measure, and no cross-loadings are allowed. By
constraining nontarget factor loadings to zero, CFA relies on
the implicit assumption that items are perfectly pure indicators
of the constructs they are intended to measure, which is often
unrealistic (Marsh et al., 2014). Rather, they tend to present
some degree of association with others constructs (typically
expressed through cross-loadings), which may be even more
pronounced when the instruments assess related constructs
(Gillet et al., 2019) such as work engagement and burnout.
Importantly, when these additional sources of true score var-
iance are ignored (i.e., by forcing items’ cross-loading to zero),
the unmodeled complexity tend to be absorbed by parts of the
model resulting in biased estimation of the model parameters
such as inated correlations (Schmitt & Sass, 2011; Tóth-Király
et al., 2017; Xiao et al., 2019), even with negligible cross-load-
ings as small as .10 (Asparouhov et al., 2015). These results
suggest that the magnitude of correlations reported in pre-
vious work engagement-burnout studies could be biased and,
given that the meaning of a construct depends on the way it is
related to other constructs, it is important to take into account
potential cross-loadings, especially when assessing theoreti-
cally related constructs (Morin et al., 2017).
Exploratory structural equation modelling (ESEM;
Asparouhov & Muthén, 2009) allows for the estimation of
cross-loadings between items and conceptually related con-
structs, thus oering a more exible approach than traditional
CFA. Indeed, ESEM shares many features with CFA (e.g., calcula-
tion of goodness-of-t indices, relationship between latent
constructs adjusted for measurement error, correlated unique-
nesses) but unlike CFA, secondary loadings (cross-loadings) are
estimated and not forced to zero (Marsh et al., 2014). Thus,
ESEM integrates the best features of both CFA and EFA and can
be used in a conrmatory manner to test a-priori measurement
models using target rotation (Asparouhov & Muthén, 2009;
Browne, 2001), which allows for the pre-specication of target
loadings while specifying cross-loadings as close to zero as
possible. Several studies have shown that ESEM results in a
more accurate estimation of factor correlations than CFA (e.g.,
Asparouhov & Muthén, 2009; Gomes et al., 2017), while allow-
ing for cross-loadings when the true population model corre-
sponds to CFA assumptions does not result in biased
estimation of factor correlations (Asparouhov et al., 2015).
A second methodological issue relates to the use of higher-
order models to test the opposite versus distinct views (e.g.,
Demerouti et al., 2010; Schaufeli et al., 2002, 2008; Taris et al.,
2017). For example, Demerouti et al. (2010) examined whether
dedication and cynicism on the one hand and vigour and
exhaustion on the other can be considered extreme opposites
of two continuums – identication and energy, respectively – or
as distinct constructs based on the logic of higher-order CFA
models (i.e., a higher-order factor supports the existence of the
continuum between the constructs). On the basis of the higher-
order CFA results, the authors concluded that “while the atti-
tude constructs [dedication and cynicism] represent opposite
ends of one continuum, the energy constructs [vigour and
exhaustion] do not” (p. 259). However, it should be noted that
in these models (see Figure 1(a)) the higher-order factor (e.g.,
identication) is directly estimated from the lower-order factors
(e.g., dedication and cynicism). The lower-order factors are thus
a component of the higher-order factor, rather than being
separate from it (Morin et al., 2017). As the higher-order factor
operates only through (i.e., fully mediated by) the lower-orders
factors (Yung et al., 1999), it is impossible to disentangle the
unique variance attributable to the higher-order factor from
the variance associated with the lower-order factors (Canivez,
2016), which is crucial for clarifying whether the constructs are
distinct or not (Chen et al., 2013). A further limitation lies in the
stringent proportionality constraints underlying the higher-
order model (Gignac, 2016). Specically, the eect of the
higher-order factor on the items is estimated as the product
of (a) the item’s lower-order factor loading by (b) the loading of
this lower-order factor on the higher-order factor. This second
term (b) is thus a constant for all items associated with a specic
lower-order factor. These constraints imply that the ratio of the
higher to lower-order factor loadings for all items associated
with a specic lower-order factor will be exactly the same (i.e.,
proportional), which is unlikely to be true in real-world settings
(Reise, 2012; Yung et al., 1999). Finally, it is important to note
that the higher-order factor is an abstraction of lower-order
factors, which in turn are an abstraction of the items. The
higher-order factor is thus an abstraction from an abstraction
(Thompson, 2004), making it dicult to determine what the
higher-order factor really is or means (Gorsuch, 2003).
Bifactor models have been proposed as a more exible alter-
native to higher-order models that do not suer from
Figure 1. Graphical representation of confirmatory higher-order and bifactor models. Note. X1–X3, Y1–Y3, and Z1–Z3 = items; F1–F3 = lower-order factors; HF = higher-
order factor; SF1–SF3 = specific factors in a bifactor model; GF = general factor in a bifactor model. Ovals represent latent factors and squares represent observed
indicators. Full unidirectional arrows linking ovals and squares represent factor loadings.
proportionality constraints (Brunner et al., 2012; Chen et al.,
2006). In these models (Figure 1(b)) all items are inuenced by
a general factor (analogous to the higher-order factor in the
higher-order model) and dierent subset of items by specic
factors (analogous to the lower-order factors). However, both
the general and the specic factors are directly estimated from
the items, thereby providing an easy and straightforward inter-
pretation. Importantly, the general and specic factors are all
assumed to be orthogonal (Reise, 2012) which means that the
specic factors are conceptually independent of the general
factor. Statistically, this implies that item variance can be parti-
tioned into dierent sources, allowing one to examine the rela-
tive inuence of the general versus the specic factors and,
consequently, to determine whether items are primarily unidi-
mensional, multidimensional, or both (Bonifay et al., 2017; Reise,
2012). For this reason, bifactor model represents a useful tool in
assisting researchers to answer critical questions regarding the
conceptualization of psychological constructs (Brunner et al.,
2012; Cai et al., 2011). For example, this method can be used to
test whether variance from the exhaustion and vigour items is
substantially explained by a general energy factor (which means
that they reect the same construct and are thus redundant);
whether the general energy factor is weak and the variance is
mainly accounted for by the specic exhaustion and vigour
factors (which suggest that they are distinct or real constructs);
or alternatively, whether there is a strong general energy factor
underlying all items but the specic exhaustion and vigour
factors explain unique variance over and beyond the general
factor (which means that these constructs are real and
The present study
In this study we investigated competing perspectives on the
conceptualization of work engagement and burnout using
bifactor modelling approach. In particular, we examined
whether work engagement and burnout are better represented
as opposite ends of a single global well-being construct (in
which case the distinction is untenable, i.e. they are redundant);
as conceptually related but distinct constructs; or whether a
single global well-being construct coexists with more specic
work engagement and burnout constructs (in which case the
constructs are real and redundant; see Figure 2(c)). Similarly, we
tested whether core dedication-cynicism and vigour-exhaus-
tion dimensions are opposite ends of two bipolar constructs
(identication and energy, respectively), are dierent con-
structs, or whether they reect the same global construct but
also retain some specicity over and beyond the global con-
struct (see Figure 2(d)). When data were consistent with a
bifactor structure (i.e., a well-dened general factor and at
least some well-dened specic factors), we then examined
the pattern of relationships between the specic factors and
other variables and the contribution of these factors over and
beyond the general factor in predicting external variables,
including measures of job resources, professional self-ecacy,
positive aect and negative aect. These variables have been
shown to be related with burnout and work engagement in
previous studies (e.g., Bakker et al., 2014; Castellano et al., 2013;
Nahrgang et al., 2011; Salanova et al., 2011) and can therefore
be usefully applied to address the theoretical debate surround-
ing the work engagement-burnout relationship. As indicated
by Le et al. (2010), if two constructs (i.e., specic factors) are
similarly related with other constructs and do not contribute
uniquely to the prediction of external variables over and
beyond the general factor, this would mean that these con-
structs are empirically redundant. On the contrary, if the two
constructs are dierentially related to other variables and
explain additional variance over and beyond the general factor,
this would mean that despite being opposites they result in
unique outcomes, which would support the real and redundant
view. On the other hand, when data do not support a bifactor
structure and only the specic factors are meaningful, then it
means that the constructs (i.e., specic factors) are empirically
Figure 2. Measurement models tested in the present study. Note. VI = vigour; DE = dedication; EX = exhaustion; CY = cynicism; BUR = burnout; ENG = work
engagement; WB = worker well-being; EN = energy; ID = identification.
distinct and therefore it makes no sense to pursue further
analyses to explore whether or not they are truly dierent
Participants and procedure
A cross-sectional study was conducted using a convenience
sample of 1787 employees from dierent companies in
Córdoba, Argentina. The companies were contacted and
invited to participate as part of a Healthy Organizations
Research Project. The response rate was 78%. After obtaining
permission from executive management, the managers and
Human Resources department were informed of the study
during management meetings. Thereafter, all employees
received paper-and-pencil questionnaires which they were
asked to complete. A letter explaining the purpose of the
study accompanied the questionnaires, emphasizing the volun-
tary nature of participation as well as the condentiality and
anonymity of the responses. All questionnaires were completed
and handed over to the researchers on the same day. After
completing the questionnaires, each company received feed-
back outlining the main ndings and providing some recom-
mendations. Data were collected between March and
November 2018. This study was approved by the Ethics
Committee of the Faculty of Medical Sciences, National
University of Córdoba.
The participants were predominantly female (51.5%). Ages
ranged from 18 to 65 (M = 33.18; SD = 8.50). A total of 21.5%
held a university degree. The majority worked in the private
sector (65.3%), in small companies (41.4%) and in the service
sector (66.8%). A detailed description of sample characteristics
is provided in Table 1.
Burnout and work engagement
Core burnout dimensions were assessed with the Argentinean
version (Spontón et al., 2019) of the Maslach Burnout Inventory-
General Survey (MBI-GS; Schaufeli et al., 1996). The exhaustion
and cynicism scales each comprised four items (e.g., “I feel
emotionally drained from my work” and “I have become less
enthusiastic about my work”, respectively). Core work engage-
ment dimensions were assessed with the Argentinean version
(Spontón et al., 2012) of the Utrecht Work Engagement Scale
(UWES; Schaufeli et al., 2002). The vigour and dedication scales
included six items (e.g., “at my work, I feel bursting with energy”
and “I am enthusiastic about my work”, respectively). All items
tapping core burnout and work engagement dimensions were
rated on a 7-point frequency scale, ranging from 0 (never) to 6
(daily). In order to avoid response bias, we randomly merged all
burnout and work engagement items into one questionnaire.
Job resources
Job resources were measured with a 9-item self-report ques-
tionnaire based on job characteristics proposed by Warr (1990):
role clarity, job autonomy, social support, skill utilization, skill
variety, task feedback, salary, safety and task signicance. All
items were rated on a 5-point scale ranging from 1 (very little) to
5 (very much). Previous research (Medrano & Trógolo, 2018)
supported the unidimensional structure of the scale and accep-
table level of internal consistency for the scale items
(Cronbach’s α = .79). In this study, the ordinal alpha coecient
was .87.
Professional self-efficacy
We used a 10-item self-ecacy scale developed by Salanova et
al. (2011). All items measure the perceived self-ecacy in per-
forming tasks at work (e.g., “I feel condent to achieve work
goals”). Responses were rated on a 7-point response scale
ranging from 0 (never) to 6 (always). Prior studies have shown
good reliability and validity of the scale in Argentina (Maei et
al., 2012). In this study, the ordinal alpha coecient was .89.
Positive and negative affect
Positive aect and negative aect were measured with the
Argentinean adaptation (Moriondo et al., 2012) of the Positive
Aect and Negative Aect Schedule (PANAS; Watson et al.,
1988). The PANAS consists of 20 adjectives that measure two
dimensions of aective experience: positive aect (10 items;
“interested”, “proud”) and negative aect (10 items; “guilty”,
“irritable”). Items are rated on a 5-point scale, ranging from 1
(very slightly or not at all) to 5 (extremely). The PANAS can be
administered with various instructional sets reecting dierent
time frames. In the present study, employees were asked to
complete the questionnaire on the basis of how they felt at
Table 1. Sociodemographic characteristics of the study’s sample.
N (%) N (%)
Gender Occupational sector
Male 867
Public 543
Female 920
Private 1167
Age Non-profit 52 (2.9)
18–30 390
Do not know/do not
25 (1.4)
31–40 481
41–50 459
Industry 213
51–65 457
Commerce 357 (20)
Educational level Service 1194
Incomplete primary school 18 (1) Agriculture 14 (0.8)
Complete primary school 73 (4.1) Do not know/do not
9 (0.5)
Incomplete secundary
113 (6.3) Type of company
Complete secundary school 393 (22) Micro (< 10 employees) 402
Incomplete tertiary level 125 (7) Small (10–50 employees) 740
Complete tertiary level 255
Medium (50–250
Incomplete university
Large (> 250 employees) 363
Complete university studies 384
Do not know/do not
41 (2.3)
Incomplete postgraduate
11 (0.6)
Complete postgraduate
108 (6)
Do not know/do not answer 12 (0.7)
work during recent weeks. In this sample, the ordinal alpha
coecients were .90 for positive aect and .88 for negative
Data analyses
All models were estimated using Mplus 7.11 (Muthén &
Muthén, 1998–2013) robust weighted least square (WLSMV)
estimator based on the polychoric correlation matrix, which
explicitly treats the variables as ordinal (Finney & DiStefano,
2006; Holgado-Tello et al., 2010; Lubke & Muthén, 2004) and it is
thus theoretically more appropriate for the Likert-type data
used in the present study. Previous research has indeed
shown that WLSMV based on the polychoric correlation matrix
results in a more accurate estimation of key model parameters
than popular ML or other methods (Beauducel & Herzberg,
2006; Li, 2016). Full information maximum likelihood (FIML)
was used to treat missing data, although the percentage of
missing data was insignicant (0.4%).
Measurement models
Following recommendations by Morin et al. (2015), Morin et al.
(2017)), a correlated four-factor model representing dedication,
cynicism, exhaustion and vigour was rst estimated using CFA
(Model 1) and ESEM (Model 2). In the latter case, items loading
on the factor they are intended to measure were freely esti-
mated and small cross-loadings (≈.10) on nontarget factors
were allowed. The pre-specied factor structure was then
tested using an oblique target rotation (Asparouhov &
Muthén, 2009). Based on the fact that ESEM yields more accu-
rate estimates of factor correlations when cross-loadings are
present in the population model but remain unbiased other-
wise (Asparouhov et al., 2015), observation of a dierent pat-
tern of factor correlations in ESEM versus CFA argues in favour
of the ESEM measurement model. Otherwise, the CFA model is
retained because of its greater parsimony. We then compared
the four-factor correlated model with bifactor solutions, relying
on bifactor-CFA or bifactor-ESEM (see Morin et al., 2015, 2017),
depending on the results of the rst step. In particular, we
tested two theoretically-driven
bifactor models: The rst
(Model 3) consisted of one general well-being factor and two
specic factors represented by work engagement and burnout.
The second bifactor model (Model 4) consisted of two corre-
lated general factors (energy and identication) and four spe-
cic factors (exhaustion, vigour, dedication and cynicism). Here,
in addition to better t of the data, the bifactor model is
supported if a well-dened general factor and at least some
well-dened specic factors are present.
Model fit evaluation
Given the well-known oversensitivity of the chi-square test to
sample size and minor model misspecications (e.g., Marsh et al.,
2005), model t was assessed through the comparative t index
(CFI), the Tucker-Lewis index (TLI), the root mean square error of
approximation (RMSEA) with its 90% condence interval (CI), and
the weighted root-mean-square residual (WRMR). Previous
research has shown that these indices perform reasonably well
under the WLSMV estimator (Beauducel & Herzberg, 2006; Yu,
2002). According to common guidelines (e.g., Browne & Cudeck,
1993; Marsh et al., 2004; Hu & Bentler, 1998), values ≥ .90 and
.95 for the CFI and TLI, respectively, are considered to be indica-
tive of adequate and excellent t to the data, while RMSEA values
≤ .08 and ≤ .05, respectively, support good and excellent model
t. Values ≤ 1.00 for WRMR are taken to be indicative of satisfac-
tory t to the data (Yu, 2002). In the comparison between nested
models, typical guidelines suggest that ΔCFI ≥ – .01 and ΔRMSEA
.015 between the more restrictive and the less restrictive
model are indicative of an equivalent level of t (Chen, 2007;
Cheung & Rensvold, 2002). These criteria have also been shown
to perform well for comparing nested models with ordinal data
using WLSMV, particularly with large samples (Sass et al., 2014),
as in the case of the present study.
Complementary bifactor model indices
When t indices supported the bifactor model, we then pro-
ceeded to investigate the relative strength of the general and
specic factors by examining the magnitude of the standar-
dized factor loadings on the general versus the specic factors,
the explained common variance (ECV), the percentage of
uncontaminated correlations (PUC), the construct replicability
(H) and the hierarchical omega reliability coecients for the
total scale (ω
) and for the subscales (ω
).The ECV is the
proportion of all common variance explained by the general
factor. This index can also be calculated for the specic factors
(ECV_S), computing the strength of a specic factor relative to
all explained variance only of the item loadings on that specic
factor (Stucky & Edelen, 2014). According to Reise (2012), ECV
.70 or .80 indicates that most of the common variance is attri-
butable to the general factor and that items are essentially
unidimensional. The PUC represents the degree to which the
factor strength will be biased when forcing multidimensional
data into a unidimensional model, and supports or moderates
the interpretation of ECV (Dominguez-Lara & Rodriguez, 2017).
High PUC (≥ .70) and ECV (≥ .70) indicate that tting data into a
unidimensional model would result in little bias in the model
parameters (factor loadings) and, therefore, common variance
can be regarded as unidimensional (Rodriguez et al., 2016).
Even if the ECV is modest (.50), relative bias is still small when
the PUC is high (Reise et al., 2013). The construct replicability (H)
informs how well a latent variable is represented by a set of
items (Hancock & Mueller, 2001). High H values (≥ .70) suggest a
well-dened latent variable, whereas low H values suggest that
the latent variable is poorly dened. Finally, omega hierarchical
) is a model-based estimate of internal reliability and reects
the percentage of reliable variance in total scores that can be
attributable to the general factor. The omega hierarchical coef-
cient can also be extended to the specic factors in the form
of the omega hierarchical subscale (ω
), which reects the
percentage of reliable variance in a subscale score after partial-
ling out the variance attributable to the general factor. High
values (≥ .80) indicate that the majority of the variance in
subscale scores reects the general factor and not the specic
factors they are intended to measure and, consequently, the
specic factors are meaningless. However, if the ω
values are
.30, then specic factors explain substantial unique variance
apart from the general factor (Smits et al., 2015), which must be
construed as measuring unique constructs above and beyond
the general factor (Bonifay et al., 2017). Other researchers
(Urbán et al., 2016) have proposed a more lenient cut-o (≥ .20)
as indicative of the uniqueness of the specic factors over and
beyond the general factor.
Association with external criteria
The relationship between factors of the bifactor model and exter-
nal variables was examined using structural equation modelling
(SEM). Thus, this analysis included both the bifactor measurement
model and structural model in which relations were tested. The
standardized regression coecients (β) and the percentage of
explained variance (R
) associated with the general and the spe-
cic factors were examined. According to Ellis (2010), R
values of
.02, .13 and .26 represent respectively small, moderate and large
contributions to the explained variance of a criterion.
CFA versus ESEM
Table 2 presents the goodness-of-t indices of the various
models. The four-factor CFA model (Model 1) did not t well
to the data according to all t indices; CFI = .89, TLI = .87,
RMSEA = .13, WRMR = 3.92. In contrast, the four-factor ESEM
model (Model 2) provided an acceptable (RMSEA = .07) to
excellent (CFI = .97, TLI = .95, WRMR = 0.89) degree of t to
the data. The estimated factor correlations from the ESEM
model were substantially lower (׀r׀=.24 to .58, M = .42) than
those estimated from the CFA model (׀r׀=.43 to .91, M = .68),
resulting in a clearer dierentiation of the factors (Table 3).
Since various studies have consistently shown that ESEM pro-
vides a better representation of the true factor correlations
when cross-loadings are present in the population model (e.
g., Asparouhov & Muthén, 2009; Gomes et al., 2017), the ESEM
model was retained for subsequent analyses. An examination
of parameter estimates from this model (Table 4) reveals well-
dened factors, with all items loading strongly on their
respective factors (ranging from |λ| = .39 to .89, M = .59) and
cross-loadings systematically lower than the main loadings
(ranging from |λ| = .00 to .41; M = .14). Out of 60 cross-loadings,
17 are non-signicant, 14 are between |.05| and |.10|, 10 are
between |.10| and |.20|, 13 are between |.20| and |.30|, and 6 are
over |.30|. Interestingly, dedication and cynicism are strongly
negatively related to one another (r = – .54) and share substan-
tial nontarget cross-loadings (ranging from |λ| = .06 to .41,
M = .22). In contrast, vigour and exhaustion are weakly nega-
tively correlated (r = – .24) and share small nontarget cross-
loadings (ranging from |λ| = .01 to .26, M = .09), most of which
are under the boundaries of what is typically considered neg-
ligible in ESEM applications (Marsh et al., 2013). Thus, it appears
from ESEM ndings that while dedication and cynicism have
much in common with one another, exhaustion and vigour do
not and, indeed, represent largely independent constructs (any
correlation below .33 is typically considered as indicative of
independency between two constructs; Tabachnick & Fidell,
2007). Complementary analysis showed that the average var-
iance extracted (AVE) for the latent vigour and exhaustion
variables were larger (.37 and .48) than their shared variance
(ϕ = .06), providing additional support to the distinctiveness
between the two constructs. In the case of dedication and
cynicism, the AVEs were .27 and .45 and their shared variance
was .29, thus suggesting that they are distinct constructs
despite being strongly negatively related (i.e., for any two con-
structs, A and B, the AVE for A and the AVE for B both need to
be larger than the shared variance between A and B to con-
clude for discriminant validity; see Farrell, 2010).
ESEM versus bifactor-ESEM
Since the ESEM model provided a better representation of the
data than the CFA model, we kept within the ESEM framework
Table 2. Summary of goodness-of-fit indices for CFA, ESEM and bifactor-ESEM
Model Description χ
RMSEA (90%
CFA model
4824.86*** 164 .89 .87 .126
ESEM model
1335.85*** 116 .97 .95 .077
bifactor ESEM:
well-being (SF:
2065.77*** 133 .95 .93 .090
bifactor ESEM:
366.08*** 85 .99 .98 .043
Note: χ
= chi-square; df = degree of freedom; CFI = comparative fit index;
TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation;
CI = confidence interval; WRMR = weighted root-mean-square residual;
CFA = confirmatory factor analysis; ESEM = exploratory structural equation
modelling; GF = general factor; SF = specific factors.
*** p <.001
Table 3. Intercorrelations among variables obtained from ESEM (above the
diagonal) and CFA (below the diagonal) solutions.
1. Dedication .58*** −.31*** −.54***
2. Vigour .91*** −.24*** −.59***
3. Exhaustion −.43*** −.51*** .29***
4. Cynicism −.86*** −.77*** .58***
*** p <.001 (two-tailed)
Table 4. Standardized factor loadings for the four-factor ESEM model.
Items Dedication Vigour Exhaustion Cynicism
Dedication 1 .51*** −.01 .25*** −.18***
Dedication 2 .60*** .30*** −.14*** −.07***
Dedication 3 .53*** .15*** .09*** −.19***
Dedication 4 .40*** .29*** .03 −.31***
Dedication 5 .59*** .21*** −.03 −.17***
Dedication 6 .44*** .33*** .02 −.28***
Vigour 1 .28*** .39*** −.21*** −.02
Vigour 2 −.01 .71*** .15*** −.03
Vigour 3 .25*** .43*** −.09*** .03
Vigour 4 .31*** .39*** −.26*** .04
Vigour 5 −.24*** .76*** .09*** .06*
Vigour 6 −.03 .82*** −.01 .05*
Exhaustion 1 .28*** −.01 .69*** .08**
Exhaustion 2 .08*** .04 .80*** .00
Exhaustion 3 −.28*** −.06* .47*** .17***
Exhaustion 4 .24*** −.07* .76*** .06*
Cynicism 1 −.33*** −.10*** .15*** .42***
Cynicism 2 −.41*** .01 .29*** .47***
Cynicism 3 −.06* .10*** .00 .89***
Cynicism 4 .11*** .02 .09*** .79***
Note: Target factor loadings in the ESEM solution are in bold.
to test bifactor models using an orthogonal bifactor rotation
(Reise et al., 2011). In bifactor-ESEM, all items are allowed to
load on the general factor and on their corresponding specic
factors, and cross-loadings across the nontarget specic factors
are also estimated. All specic factors are specied as orthogo-
nal in line with bifactor assumptions. As shown in Table 2, the
bifactor-ESEM model assuming a general well-being factor and
two specic factors representing work engagement and burn-
out (Model 3) had an acceptable degree of t to the data
according to the CFI (.95) and TLI (.93), but not the RMSEA
(.09) and WRMR (1.64). Thus, this model did not account well
for the data suggesting that burnout and work engagement do
not form a common well-being factor. The bifactor-ESEM
model consisting of two general energy and identication
factors and four specic factors representing vigour, exhaus-
tion, dedication and cynicism (Model 4) provided an excellent t
to the data (CFI = .99, TLI = .98, RMSEA = .04, WRMR = 0.57) and
an apparently better representation of the data than the four-
factor ESEM model (ΔRMSEA = – .03; ΔCFI = .02). However, as
mentioned before, in addition to the model t indices it is also
necessary to take into account model parameters and specic
bifactor indices to be able to more accurately judge the appro-
priateness of a bifactor solution.
Tables 5 and 6 provide separately the standardized factor
loadings and bifactor indices of the bifactor-ESEM identication
and the bifactor-ESEM energy parts of Model 4. In terms of
identication, results show that the general identication factor
is well-dened by the presence of strong and signicant target
loadings of all items (|λ| = .36 to .87, M = .69). Over and above
the general factor, the specic dedication factor is also well-
dened through substantial target loadings (|λ| = .34 to .53,
M = .43), whereas the specic cynicism factor appears to be
poorly dened (|λ| = – .05 to .46, M = .17). In support of this,
ESEM factor loadings of cynicism items (|λ| = .42 to .79, M = .64)
were considerably lower (|λ| = .05 to .46, M = .17) and factor
loadings of some items became nonsignicant or even nega-
tive (items had been reverse scored) once the general factor
was taken into account. These results suggest that cynicism
may not exist as a specic factor over and above the general
factor, and that the variance related to cynicism is explained by
the general factor. In contrast, ESEM factor loadings of dedica-
tion items (|λ| = .40 to .60, M = .51) were somewhat smaller (|
λ| = .34 to .53, M = .43) but remained signicant once the
general factor was taken into account, implying that the ded-
ication factor does indeed tap into relevant specicity and add
information to the general identication factor. Further analysis
revealed that the general factor accounted for 76% of the
common variance (ECV = .76) suggesting a strong general
factor. In addition, the H value for the general factor was high
(.92) meaning that the items dened the latent identication
construct well; reliability estimates indicate that the majority of
reliable variance of total scores is attributable to the general
factor (ω
H =
.82). Thus, ECV, H and ω
support the idea that
items are essentially unidimensional and reect a global iden-
tication construct. However, it should be noted that PUC was
moderate (.53) suggesting the presence of some multidimen-
sionality. In line with this, the hierarchical omega subscale of
the dedication factor (ω
= .27) indicates that unique reliable
variance of the subscale scores is explained by this factor after
controlling for the general factor. In sum, dedication and cyni-
cism items collapse into a general identication factor and can
thus be considered as opposite indicators of the same con-
struct (redundant), but dedication also retains some specicity
over and beyond the general identication factor and can thus
be considered as real.
With regard to the bifactor energy part of the model (Table
6), results show that the general energy factor is relatively well-
dened by the presence of weak-to-moderate and signicant
target loadings from most of the items (|λ| = .13 to .86, M = .43).
Only three out of ten items had strong factor loadings (> .50) on
the general factor. The specic vigour and exhaustion factors
were better dened by the presence of stronger and signicant
factor loadings (|λ|
= .24 to .69, M = .48; |λ|
= .23 to
.71, M = .58). Comparison of factor loadings from ESEM and
bifactor-ESEM solutions reveals that factor loadings pertaining
to vigour items are slightly lower but remain signicant and
Table 5. Standardized factor loadings and specific bifactor indices for the bifac-
tor-identification ESEM solution.
GF-identification SF-dedication SF-cynicism
Dedication 1 .36*** .44*** .05***
Dedication 2 .82*** .34*** .19***
Dedication 3 .57*** .53*** .00
Dedication 4 .72*** .38*** .16***
Dedication 5 .73*** .46*** .05***
Dedication 6 .71*** .41*** .08***
Cynicism 1 .75*** .01 −.05
Cynicism 2 .87*** .03 −.20**
Cynicism 3 .73*** .07* .45***
Cynicism 4 .61*** .02 .46***
.69 .43 .16
Hierarchical omega .82 .27 .04
H.92 .58 .36
ECV .76 .30 .17
PUC .53
Note: GF = general factor; SF = specific factor; λ
= mean factor loadings;
H = construct replicability; ECV = explained common variance;
PUC = percentage of uncontaminated correlations. Items from the cynicism
subscale are reverse scored.
** p <.01. *** p <.001
Table 6. Standardized factor loadings and specific bifactor indices for the bifac-
tor-energy ESEM solution.
GF-energy SF-vigour SF-exhaustion
Vigour 1 .64*** .39*** .09
Vigour 2 .32*** .60*** .15
Vigour 3 .47*** .37*** .02
Vigour 4 .86*** .24*** −.02
Vigour 5 .13*** .58*** −.04
Vigour 6 .38*** .69*** .01
Exhaustion 1 .16*** −.02 .68***
Exhaustion 2 .36*** −.11 .68***
Exhaustion 3 .72*** .06 .23***
Exhaustion 4 .29*** .00 .71***
.43 .48 .58
Hierarchical omega .51 .44 .56
H.84 .71 .74
ECV .44 .48 .66
PUC .53
Note: GF = general factor; SF = specific factor; λ
= mean factor loadings;
H = construct replicability; ECV = explained common variance;
PUC = percentage of uncontaminated correlations. Items from the exhaustion
subscale are reverse scored.
*** p <.001
generally strong once the general energy factor is taken into
account in the model, and the same holds true for exhaustion
items (|λ|
= .39 to .82, M = .58; |λ|
= .47 to
.80, M = .68). The examination of the bifactor indices show that
vigour and exhaustion explains more common variance than
the general factor on their respective items (ECV_S
= .48,
= .66, ECV = .44). In addition, the omega hier-
archical coecient was inacceptable for the general factor
= .51) but not for the specic factors (ω
= .44 and .56),
indicating that reliable variance of subscale scores mostly
reects the specic factors rather than the general factor.
Finally, H values were .71 and .74 for the specic factors and,
therefore, items represent well-dened latent variables in these
factors. The H coecient for the general factor was also high
(.84), suggesting that the items may be a good representation
of the latent energy construct. However, it should be kept in
mind that this coecient is quite sensitive to high factor load-
ings and that just a few items with high factor loadings can
articially elevate its value (Rodriguez et al., 2016). A closer look
into factor loadings on the general factor show two items with
very high loadings relative to the rest: “When I wake up in the
morning I feel like going to work” (λ = .86) and “It’s more and
more of an eort every day to get up and go to work” (λ = .72).
As these items possess high content similarity, it is feasible that
the high H value is a mere artefact due to the fact that items tap
essentially the same question and likely elicit the same
response, giving rise to an illusory index (Rodriguez et al.,
2015). Taken together, these ndings support the idea that
rather than being opposites of one energy construct and thus
redundant, exhaustion and vigour represent distinct constructs
and are thus real, which is also consistent with the weak corre-
lation observed between them (see Table 3).
Relations with other constructs
To provide complementary evidence of the identication part
of the bifactor model, relationships with other constructs
including professional self-ecacy, job resources, positive
aect and negative aect were examined. Results from SEM
indicate that the model t the data well, χ
(1147) = 3697.48,
CFI = .93, TLI = .92, RMSEA = .052, 90% CI [.050, .054]. Only the
WRMR value (1.32) was slightly above the acceptable range. As
seen in Table 7, professional self-ecacy and job resources
were more strongly related to dedication than to cynicism.
Positive aect and negative aect were related to dedication
and unrelated to cynicism. Together, the two constructs con-
tributed uniquely to the prediction of the external variables
over and beyond the general identication construct, explain-
ing strong variance across all the variables (R
ranging from
24% to 50%). Thus, instead of a similar pattern (equally strong)
of relationships, dedication and cynicism were dierentially
related to professional self-ecacy, job resources, positive
aect and negative aect, and had unique eects across all
these variables over and beyond the general identication
construct, which support the view that despite being each
other’s opposite of an identication continuum and thus
redundant, dedication and cynicism result in unique outcomes
and are thus real.
Work engagement and burnout have been linked to employee
health and organizational performance (e.g., Christian et al.,
2011; Hakanen et al., 2008; Rich et al., 2010) and therefore
constitute relevant topics for researchers and practitioners.
However, because work engagement and burnout are gener-
ally high correlated (Halbesleben, 2010), there has been recent
debate in the literature concerning the nature of the relation-
ship between these constructs. As mentioned, while some
scholars have argued that work engagement is a distinct con-
struct from burnout (Schaufeli & Bakker, 2004; Schaufeli et al.,
2002), others doubt whether work engagement is a novel and
useful concept, suggesting it to be simply the opposite of
burnout and thus redundant (Cole et al., 2012). Finally, more
recent perspectives move beyond the distinct/real versus
opposite/redundant dilemma and claim work engagement
and burnout to be both real and redundant concepts (Leiter
& Maslach, 2017). Bifactor analysis provides a viable means for
testing these competing views, because it simultaneously
examines the variance shared by two constructs as reected
in the general factor, and the variance unique to each construct
as reected in the specic factors of the bifactor model. In the
current study, we addressed for the rst time the three con-
ceptual perspectives through the lens of bifactor model.
The ndings show that a bifactor model consisting of a
general well-being factor and two specic factors represented
by work engagement and burnout did not account well for
data. This nding suggests that at the construct level, burnout
and work engagement do not represent opposite sides of a
single employee well-being construct, but rather dierent con-
structs. Thus, as Schaufeli and Salanova (2011) colourfully put
forward, it appears that “rather than two sides of the same
[well-being] coin, work engagement and burnout are two dif-
ferent coins” (p. 45). On the other hand, the bifactor model
consisting of two general factors representing energy and
identication and four specic factors representing exhaustion,
vigour, dedication and cynicism, provided the best t to the
data. An inspection of bifactor indices leads to dierent con-
clusions regarding these two pairs of constructs. On the one
hand, exhaustion and vigour dimensions did not form a strong
Table 7. Relationship between general identification factor, specific dedication
and cynicism factors to external variables.
Standardized regression coefficients
cynicism R
Professional .36*** .31*** .17*** .13
.77*** .44*** −.05 .59
−.48*** −.29*** −.03 .23
Job resources .67*** .39*** .12*** .45
(27%) .62
Note: GF = general factor; SF = specific factors. Items from the cynicism subscale
are reverse scored. Percentage of the explained variance is included in the
*** p <.001
general energy factor, and the majority of reliable variance was
explained by the specic factors. Accordingly, although in a
theoretical sense exhaustion and vigour refer to low and high
energy towards work, respectively, they are in fact independent
constructs. Indeed, the correlation between vigour and exhaus-
tion was weak (r = – .24), sharing only 6% of their variance,
which further substantiates the argument that they represent
independent constructs. On the other hand, dedication and
cynicism were strongly negatively related (r = – .54) and the
bifactor-based indices consistently supported the presence of a
strong general identication factor that captured the over-
whelming variance of dedication and cynicism items, suggest-
ing that these construct are opposite ends of a bipolar
identication construct and thus redundant. However, dedica-
tion and cynicism show a dierent pattern of relationships with
professional self-ecacy, job resources, positive aect and
negative aect, and have unique inuences on these variables
over and beyond the global identication construct. Hence,
while dedication and cynicism act as direct opposites along
an identication continuum they also result in unique out-
comes, which, according to Leiter and Maslach (2017), implies
that they are real and redundant constructs. Collectively, these
ndings are largely consistent with those of psychometric
(Demerouti et al., 2010; González-Romá et al., 2006) and long-
itudinal (Mäkikangas et al., 2012, 2017) studies suggesting that
cynicism and dedication are opposite poles of a single identi-
cation construct, whereas exhaustion and vigour represent
independent constructs. However, our ndings also revealed
that dedication and cynicism have unique, non-shared charac-
teristics and result in unique eects, meaning that they also
represent unique constructs.
Theoretical implications
The ndings of the present study are in disagreement with the
assertion of Cole et al. (2012) that work engagement is a
redundant concept tapping an old construct under a new
label. Rather, it appears to be a real construct that should be
understood in its own right. When the core dimensions of work
engagement and burnout are considered, the conclusions are
not unequivocal: vigour and exhaustion appear to be empiri-
cally distinguishable constructs and thus real. In practical terms,
this implies that feeling vigorous at work does not presuppose
lack of fatigue, and vice versa. Indeed, employees may feel
simultaneously vigorous and exhausted, as evidenced by daily
diary studies (Mäkikangas et al., 2014). Dedication and cynicism
on the other hand were strongly explained by a global identi-
cation construct and thus represent redundant concepts indi-
cating high and low identication, respectively. Accordingly,
employees can feel either cynical or enthusiastic about their
jobs, but it seems unlikely they can feel both simultaneously
(Demerouti et al., 2010), since dedication and cynicism are each
other’s opposite. However, dedication and cynicism also
showed unique eects on dierent external variables, substan-
tiating the view that they are also real constructs. As suggested
by dialectical theory (Leon et al., 2015), employees may move
from one state to another across time, and this dynamic inter-
play may produce dierent outcomes. Overall, these ndings
suggest that each theoretical perspective on the work engage-
ment-burnout relationship has its own merit.
Measurement implications
From a measurement perspective, considering work engage-
ment to be a distinct construct from burnout indicates that it
may not be accurately assessed by the opposite pattern scores
(i.e., low) in the MBI subscales (Maslach & Leiter, 1997). Rather,
we agree with Schaufeli and Bakker (2010) that work engage-
ment would be more appropriately assessed using a stand-
alone measure. If the primary interest lies in the core compo-
nents of work engagement and burnout, then our ndings
clearly indicate that exhaustion and vigour should be measured
separately. In the case of dedication and cynicism, as they
largely represent opposite poles along an identication conti-
nuum, one could argue that a single measure would suce,
with a high score on identication indicating a high level of
dedication and a low score indicating a high level of cynicism.
However, the correlation between dedication and cynicism was
.54, implying that although strongly negatively related they are
not perfect opposites and can be thus regarded as complemen-
tary rather than mutually exclusive states. Consequently, a
more accurate measurement of dedication and cynicism
would require dierent scales rather than considering only
MBI-cynicism or UWES-dedication items.
Limitations and suggestions for future research
Certain limitations of the present study should be men-
tioned. Firstly, although we used a large sample of employ-
ees, they were mostly from private sector services
companies, which limit the generalization of the current
results. Thus, further studies are needed to test the general-
izability of the ndings across a wider range of organiza-
tional and occupational settings. Secondly, all data were
gathered through self-reports, meaning that the results
may be biased due to common method variance
(Podsako et al., 2003). However, given that we evaluate
intrapsychic factors such as beliefs, aects and feelings
about work, collecting data on these phenomena by other
means would be challenging. In addition, although we can-
not completely discard the possibility that common method
variance bias plays a role, several studies (e.g., Brannick et
al., 2010; Semmer et al., 1996) have indicated that this bias
is not as troublesome as one might expect. Nevertheless,
future research could potentially benet from including
more objective measures such as co-worker or supervisor
ratings. Thirdly, we used the most popular measures of
work engagement and burnout, i.e., the UWES and the
MBI-GS. One might therefore argue that the results do not
necessarily make the construct redundant (real, or both),
but rather the measures. However, it should be noted that
previous studies using alternative measures of work
engagement and burnout yielded similar conclusions to
those of the present study (see Demerouti et al., 2010;
Mäkikangas et al., 2012). Beyond this, it would be fruitful
to replicate the present study using dierent scales.
Fourthly, we have only contrasted two dimensions of burn-
out and engagement: exhaustion with vigour and cynicism
with dedication. As mentioned, the third component of
burnout (i.e., low professional ecacy) and engagement (i.
e., absorption) were excluded because the evidence in sup-
port of these components remains controversial and, theo-
retically speaking, they are distinct rather than each other’s
opposite. However, it would be important for future
research to include all six dimensions captured by the burn-
out and engagement concepts. Fifthly, we used a variable-
centred approach to investigate the underlying dimension-
ality of the work engagement and burnout constructs. It
would be valuable in future investigation to complement
this approach with a person-centred perspective to provide
a more complete investigation of this matter. As Morin et al.
(2017) noted, variable- and person-centred analyses are dif-
ferent ways of looking at the same phenomena and, hence,
complementary person-centred perspective may also be
useful for exploring the dimensionality of psychological
constructs (see also Marsh et al., 2009, for a synergic view
between the two approaches). For example, using this
approach researchers would be able to examine how the
dierent components of burnout and work engagement
combine at the within-person level, and whether they act
as direct opposites or not. Moreover, diary studies using a
person-centred approach could be useful for examining
intra-individual uctuations of burnout and work engage-
ment components and how they interact throughout the
workday (see for example, Sonnentag, 2017). Finally, as in
the case of most empirical research surrounding the work
engagement-burnout debate, the present study relied
exclusively on psychological measures. It would be interest-
ing to examine the physiological and neurobiological corre-
lates of work engagement and burnout. In this respect,
research has shown that burnout leads to a hypothalamus-
pituitary-adrenal (HPA) axis dysregulation (Chow et al., 2018)
which stimulates the adrenal cortex to release glucocorti-
coids such as cortisol into the blood (Sapolsky et al., 2000).
However, other studies found no evidence of disruption in
HPA-axis functioning in burned-out employees (Langelaan
et al., 2006). Less is known about the physiological and
neurobiological mechanisms involved in the case of work
engagement. Langelaan et al. (2006) showed that engaged
employees did not dier from burned-out employees with
regard to HPA-axis functioning. On the other hand, recent
studies have suggested that the locus coeruleus–norepi-
nephrine (LC-NE) system may be associated with the regula-
tion of task engagement, particularly in its phasic mode, in
which NE from the LC is rapidly released in stimuli-evoked
responses (Hopstaken et al., 2015). This output mode of the
LC–NE system contributes to maintain high task engage-
ment (Minzenberg et al., 2008). In summary, ndings on
physiological and neurobiological correlates are inconsistent
and limited, particularly with respect to work engagement.
Further exploration of the neural bases of burnout and work
engagement could enhance our understanding of these
work-related states. For example, are neural bases of work
engagement and burnout similar or dissimilar? How and to
what extent are they interconnected? Addressing these
questions will provide additional evidence to those of
cross-sectional, longitudinal and psychometric studies and
may ultimately have implications for the conceptualization
of burnout and work engagement.
In this study, we sought to provide further evidence in relation
to the ongoing debate on the conceptualization of work
engagement and burnout, using a data analytic method that
remains underutilized in organizational research. The ndings
reveal that each of the three theoretical perspectives has some
explanatory value, thereby providing an insightful look into the
work engagement-burnout relationship. A proper conceptuali-
zation should take into account the level of analysis (construct-
or dimension-level) as well as the specic components of burn-
out and work engagement.
Cronbach’s alpha coecient is by far the most used reliability
index for estimating internal consistency (Sijtsma, 2009).
Calculation of the alpha coecient involves the correlation
matrix among all items of a scale, or among a subset of items
in the case of multidimensional measures. In mainstream sta-
tistical software such SPSS and SAS, Cronbach’s alpha is calcu-
lated using the Pearson correlation matrix as the default. An
important assumption for the use of the Pearson correlation
matrix is that the variables (items) are continuous. However, the
most common scenario in social sciences is that data comes
from measures using ordinal response scales (e.g., Likert-type
response formats). In such cases, the use of Cronbach’s alpha
tends to result in a biased estimation (attenuation) of the
reliability (see e.g., Gelin et al., 2003; Maydeu-Olivares et al.,
2007). Zumbo et al. (2007) proposed the ordinal alpha, a version
of Cronbach’s alpha coecient for ordinal data, which has been
shown to provide a more accurate estimation of reliability in
ordinally-scaled data.
Ordinal alpha is conceptually equivalent to Cronbach’s
alpha. The main dierence between the two is that ordinal
alpha is estimated using the polychoric correlation matrix
rather than the Pearson correlation matrix (Gadermann et al.,
2012). Thus, instead of focusing on the reliability of the
observed scores by treating the observed item responses as if
they were continuous, the ordinal alpha coecient focuses on
the reliability of the unobserved continuous variables under-
lying the observed item responses. Gadermann et al. (2012)
provides a step-by-step example of how to calculate the ordinal
alpha coecient from the polychoric correlation matrix using
statistical software package R. There is also available a user-
friendly tool in MS Excel format for the calculation of ordinal
alpha coecient (Dominguez-Lara, 2012).
Although it is reasonable to argue that bifactor analysis
should only be conducted when results from the rst step
suggest that data may be consistent with a bifactor structure
and is thus somewhat data-driven such analysis should be
based on substantive theory and expectations as the driving
forces (Morin et al., 2017). Since work engagement and burnout
constructs as well as their core dimensions represent
conceptually-related constructs that might reect the presence
of an overarching global construct, it is theoretically justied to
explore a bifactor representation of the data.
Discriminant validity refers to the extent to which a con-
struct actually diers from others empirically and can be initi-
ally evaluated by examining the correlations between two
constructs, or by examining cross-loading of the observed
variables onto factors they are not expected to measure (see
Zaiţ & Bertea, 2011, for a review of dierent methods for testing
discriminant validity). A more rigorous approach for assessing
the discriminant issue was proposed by Fornell and Larcker
(1981), who compare the average variance extracted (AVE) for
each construct with the shared variance (ϕ) between those
constructs. The AVE estimate is the average amount of variation
that a latent construct is able to explain in the observed vari-
ables (items) to which it is theoretically related. It can be
calculated by averaging the squared correlations (factor load-
ings) between the latent variable and the associated items. The
shared variance is the amount of variance that a construct is
able to explain in another construct. It is represented by the
square of the correlation between two constructs. According to
Fornell and Larcker (1981), if the AVE for each construct is
greater than its shared variance with any other construct,
then discriminant validity is supported.
Disclosure statement
No potential conict of interest was reported by the authors.
Asparouhov, T., Muthén, B., & Morin, A. J. S. (2015). Bayesian structural
equation modeling with cross-loadings and residual covariances.
Journal of Management, 41(6), 1561–1577.
Asparouhov, T., & Muthén, B. O. (2009). Exploratory structural equation
modeling. Structural Equation Modeling, 16(3), 397–438.
Bakker, A., Demerouti, E., & Sanz-Vergel, A. (2014). Burnout and work
engagement: The JD–R approach. Annual Review of Organizational
Psychology and Organizational Behavior, 1(1), 389–411.
Beauducel, A., & Herzberg, P. Y. (2006). On the performance of maximum
likelihood versus means and variance adjusted weighted least squares
estimation in CFA. Structural Equation Modeling, 13(2), 186–203. https://
Bonifay, W., Lane, S. P., & Reise, S. P. (2017). Three concerns with applying a
bifactor model as a structure of psychopathology. Clinical Psychological
Science, 5(1), 184–186.
Brannick, M. T., Chan, D., Conway, J. M., Lance, C. E., & Spector, P. E. (2010).
What is method variance and how can we cope with it? A panel discus-
sion. Organizational Research Methods, 13(3), 407–420.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model t.
In K. Bollen & J. Long (Eds.), Testing structural equation models (pp. 136–
162). Sage.
Browne, M. W. (2001). An overview of analytic rotation in exploratory factor
analysis. Multivariate Behavioral Research, 36(1), 111–150. https://doi.
Brunner, M., Nagy, G., & Wilhelm, O. (2012). A tutorial on hierarchically
structured constructs. Journal of Personality, 80(4), 796–846. https://doi.
Byrne, Z. S., Peters, J. M., & Weston, J. W. (2016). The struggle with employee
engagement: Measures and construct clarication using ve samples.
Journal of Applied Psychology, 101(9), 1201–1227.
Cai, L., Yang, J. S., & Hansen, M. (2011). Generalized full-information item
bifactor analysis. Psychological Methods, 16(3), 221–248.
Canivez, G. L. (2016). Bifactor modeling in construct validation of multi-
factored tests: Implications for multidimensionality and test interpreta-
tion. In K. Schweizer & C. DiStefano (Eds.), Principles and methods of test
construction: Standards and recent advancements (pp. 247–271). Hogrefe.
Carlson, K. D., & Herdman, A. O. (2012). Understanding the impact of
convergent validity on research results. Organizational Research
Methods, 15(1), 17–32.
Castellano, E., Cifre, E., Spontón, C., Medrano, L. A., & Maei, L. (2013).
Emociones positivas y negativas en la predicción del burnout y engage-
ment en el trabajo [Positive and negative emotions on engagement and
burnout prediction]. Revista de Peruana de Psicología y Trabajo Social, 2
(1), 75–88.
Chen, F. F. (2007). Sensitivity of goodness of t indices to lack of measure-
ment invariance. Structural Equation Modeling, 14(3), 464–504. https://
Chen, F. F., Jing, Y., Hayes, A., & Lee, J. M. (2013). Two concepts or two
approaches? A bifactor analysis of psychological and subjective well-
being. Journal of Happiness Studies, 14(3), 1033–1068.
Chen, F. F., West, S. G., & Sousa, K. H. (2006). A comparison of bifactor and
second-order models of quality of life. Multivariate Behavioral Research,
41(2), 189–225.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-t indexes
for testing measurement invariance. Structural Equation Modeling, 9(2),
Chow, Y., Masiak, J., Mikołajewska, E., Mikołajewski, D., Wójcik, G. M.,
Wallace, B., Eugene, A., & Olajossy, M. (2018). Limbic brain structures
and burnout—A systematic review. Advances in Medical Sciences, 63(1),
Christian, M. S., Garza, A. S., & Slaughter, J. E. (2011). Work engagement: A
quantitative review and test of its relations with task and contextual
performance. Personnel Psychology, 64(1), 89–136.
Coders, C. L., & Dougerthy, T. W. (1993). A review and integration of research
on job burnout. Academy of Management Review, 18(4), 621–656. https://
Cole, M. S., Walter, F., Bedeian, A. G., & O’Boyle, E. H. (2012). Job burnout and
employee engagement: A meta-analytic examination of construct pro-
liferation. Journal of Management, 38(5), 155–1581.
Demerouti, E., Mostert, K., & Bakker, A. B. (2010). Burnout and work engage-
ment: A thorough investigation of the independency of both constructs.
Journal of Occupational Health Psychology, 15(3), 209–222. https://doi.
Dominguez-Lara, S. (2012). Propuesta para el cálculo del alfa ordinal y theta
de armor [Proposal for the calculation of the ordinal alpha and armor
theta]. Revista De Investigación En Psicología, 15(1), 213–217. https://doi.
Dominguez-Lara, S. A., & Rodriguez, A. (2017). Índices estadísticos de mod-
elos bifactor [Statistical indices from bifactor models]. Interacciones, 3(2),
Ellis, P. (2010). The essential guide to eect sizes: Statistical power, meta-analysis,
and the interpretation of research results. Cambridge University Press.
Farrell, A. M. (2010). Insucient discriminant validity: A comment on Bove,
Pervan, Beatty, and Shiu (2009). Journal of Business Research, 63(3), 324–
Finney, S. J., & DiStefano, C. (2006). Non-normal and categorical data in
structural equation modeling. In G. R. Hancock & R. O. Mueller (Eds.),
Structural equation modeling: A second course (pp. 269–314). Information
Age Publishing.
Flores Kanter, P. E., Dominguez-Lara, S., Trógolo, M. A., & Medrano, L. A.
(2018). Best practices in the use of bifactor models: Conceptual grounds,
t indices and complementary indicators. Evaluar, 18(3), 44–48. https://
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models
with unobservable variables and measurement error. Journal of
Marketing Research, 18(1), 39–50.
Gadermann, A. M., Guhn, M., & Zumbo, B. D. (2012). Estimating ordinal
reliability for Likert-type and ordinal item response data: A conceptual,
empirical, and practical guide. Practical Assessment, Research, and
Evaluation, 17(3), 3.
Gelin, M. N., Beasley, T. M., & Zumbo, B. D. (2003, April). What is the impact
on scale reliability and exploratory factor analysis of a Pearson correla-
tion matrix when some respondents are not able to follow the rating
scale? [Paper presented] The annual meeting of the American educational
research association (AERA), Chicago, Il.
Gignac, G. E. (2016). The higher-order model imposes a proportionality
constraint: That is why the bifactor model tends to t better.
Intelligence, 55, 57–68.
Gillet, N., Caesens, G., Morin, A. J., & Stinglhamber, F. (2019).
Complementary variable-and person-centred approaches to the dimen-
sionality of work engagement: A longitudinal investigation. European
Journal of Work and Organizational Psychology, 28(2), 239–258. https://
Gil-Monte, P. R., García-Juesas, J. A., & Hernández, M. (2008). Inuencia de la
sobrecarga laboral y la autoecacia sobre el síndrome de quemarse por
el trabajo (burnout) en profesionales de enfermería [The inuence of
overload and self-ecacy on Burnout: A study in nursing professionals].
Interamerican Journal of Psychology, 42(1), 113–118. http://pepsic.bvsa-
Gomes, C. M. A., Almeida, L. S., & Núñez, J. C. (2017). Rationale and applic-
ability of exploratory structural equation modeling (ESEM) in psychoe-
ducational contexts. Psicothema, 29(3), 396–401.
González-Romá, V., Schaufeli, W., Bakker, A. B., & Lloret, S. (2006). Burnout
and work engagement: Independent factors or opposite poles? Journal
of Vocational Behavior, 68(1), 165–174.
Gorsuch, R. L. (2003). Factor analysis. In J. A. Schinka & W. F. Velicer (Eds.),
Handbook of psychology: Vol. 2. Research methods in psychology (pp. 143–
164). John Wiley.
Hakanen, J. J., Schaufeli, W. B., & Ahola, K. (2008). The Job Demands-
Resources model: A three-year cross-lagged study of burnout, depres-
sion, commitment, and work engagement. Work and Stress, 22(3), 224–
Halbesleben, J. R. B. (2010). A meta-analysis of work engagement:
Relationships with burnout, demands, resources and consequences. In
A. Bakker & M. Leiter (Eds.), Work engagement: A handbook of essential
theory and research (pp. 102–117). Psychology Press.
Halbesleben, J. R. B., & Demerouti, E. (2005). The construct validity of an
alternative measure of burnout: Investigating the English translation of
the Oldenburg Burnout inventory. Work and Stress, 19(3), 208–220.
Hancock, G. R., & Mueller, R. O. (2001). Rethinking construct reliability within
latent systems. In R. Cudeck, S. Du Toit, & D. Sörbom (Eds.), Structural
equation modeling: Present and future e A festschrift in honor of Karl
Jöreskog (pp. 195–216). Scientic Software International.
Holgado-Tello, F. P., Chacón-Moscoso, S., Barbero-García, M. I., & Vila-Abad,
E. (2010). Polychoric versus pearson correlations in exploratory and
conrmatory factor analysis of ordinal variables. Quality & Quantity, 44
(1), 153–166.
Hopstaken, J. F., van der Linden, D., Bakker, A. B., & Kompier, M. A. (2015).
The window of my eyes: Task disengagement and mental fatigue covary
with pupil dynamics. Biological Psychology, 110, 100–106. https://doi.
Hu, L.-T., & Bentler, P. M. (1998). Fit indices in covariance structure model-
ing: Sensitivity to underparameterized model misspecication.
Psychological Methods, 3(4), 424–453.
Innanen, H., Tolvanen, A., & Salmela-Aro, K. (2014). Burnout, work engage-
ment and workaholism among highly educated employees: Proles,
antecedents and outcomes. Burnout Research, 1(1), 38–49. https://doi.
Langelaan, S., Bakker, A. B., Schaufeli, W. B., van Rhenen, W., & van Doornen,
L. J. (2006). Do burned-out and work-engaged employees dier in the
functioning of the hypothalamic-pituitary-adrenal axis? Scandinavian
Journal of Work, Environment & Health, 32(5), 339–348.
Le, H., Schmidt, F. L., Harter, J. K., & Lauver, K. J. (2010). The problem of
empirical redundancy of constructs in organizational research: An
empirical investigation. Organizational Behavior and Human Decision
Processes, 112(2), 112–125.
Leiter, M. P., & Maslach, C. (2017). Burnout and engagement: Contributions
to a new vision. Burnout Research, 5, 55–57.
Leon, M. R., Halbesleben, J. R. B., & Paustian-Underdahl, S. C. (2015). A
dialectical perspective on burnout and engagement. Burnout Research,
2(2–3), 87–96.
Li, C.-H. (2016). Conrmatory factor analysis with ordinal data: Comparing robust
maximum likelihood and diagonally weighted least squares. Behavior Research
Methods, 48(3), 936–949.
Llorens, S., Schaufeli, W. B., Bakker, A. B., & Salanova, M. (2007). Does a
positive gain spiral of resources, ecacy beliefs and engagement exist?
Computers in Human Behavior, 23(1), 825–841.
Lubke, G. H., & Muthén, B. O. (2004). Applying multigroup conrmatory
factor models for continuous outcomes to Likert scale data complicates
meaningful group comparisons. Structural Equation Modeling, 11(4),
Maei, L., Spontón, C., Spontón, M., Castellano, E., & Medrano, L. A. (2012).
Adaptación del Cuestionario de Autoecacia Profesional (AU-10) a la
población de trabajadores cordobeses [Adaptation of the professional
self-ecacy questionnaire (AU-10) to the working population of
Cordoba]. Pensamiento Psicológico, 10(1), 51–62. https://dialnet.unirioja.
Mäkikangas, A., Feldt, T., Kinnunen, U., & Tolvanen, A. (2012). Do low
burnout and high work engagement always go hand in hand?
Investigation of the energy and identication dimensions in longitudinal
data. Anxiety, Stress, and Coping, 25(1), 93–116.
Mäkikangas, A., Hyvönen, K., & Feldt, T. (2017). The energy and identica-
tion continua of burnout and work engagement: Developmental proles
over eight years. Burnout Research, 5, 44–54.
Mäkikangas, A., Kinnunen, S., Rantanen, J., Mauno, S., Tolvanen, A., & Bakker,
A. B. (2014). Association between vigor and exhaustion during the work-
week: A person-centered approach to daily assessments. Anxiety, Stress,
& Coping, 27(5), 555–575.
Marsh, H. W., Hau, K. T., & Grayson, D. (2005). Goodness of t evaluation in
structural equation modeling. In A. Maydeu-Olivares & J. McArdle (Eds.),
Contemporary psychometrics. A Festschrift for Roderick P. McDonald (pp.
275–340). Mahwah, NJ: Erlbaum.
Marsh, H. W., Hau, K. T., & Wen, Z. (2004). In search of golden rules:
Comment on hypothesis-testing approaches to cuto values for t
indexes and dangers in overgeneralizing Hu & Bentler’s (1999).
Structural Equation Modeling, 11(3), 320–341.
Marsh, H. W., Lüdtke, O., Nagengast, B., Morin, A. J. S., & Von Davier, M.
(2013). Why item parcels are (almost) never appropriate: Two wrongs do
not make a right—Camouaging misspecication with item parcels in
CFA models. Psychological Methods, 18(3), 257–284.
Marsh, H. W., Lüdtke, O., Trautwein, U., & Morin, A. J. S. (2009). Classical
latent prole analysis of academic self-concept dimensions: Synergy of
person-and variable-centered approaches to theoretical models of self-
concept. Structural Equation Modeling: A Multidisciplinary Journal, 16(2),
Marsh, H. W., Morin, A. J. S., Parker, P. D., & Kaur, G. (2014). Exploratory
structural equation modeling: An integration of the best features of
exploratory and conrmatory factor analysis. Annual Review of Clinical
Psychology, 10(1), 85–110.
Maslach, C., & Jackson, S. E. (1981). The measurement of experienced
burnout. Journal of Organizational Behavior, 2(2), 99–113. https://doi.
Maslach, C., & Jackson, S. E. (1986). Maslach burnout inventory (2nd ed.).
Consulting Psychologists Press.
Maslach, C., & Leiter, M. P. (1997). The truth about burnout: How organiza-
tions cause personal stress and what to do about it. Jossey-Bass.
Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual
Review of Psychology, 52(1), 397–422.
Maydeu-Olivares, A., Coman, D. L., & Hartmann, W. M. (2007).
Asymptotically distribution free (ADF) interval estimation of coecient
alpha. Psychological Methods, 12(2), 157–176.
Medrano, L. A., & Trógolo, M. (2018). Employee well-being and life satisfac-
tion in Argentina: The contribution of psychological detachment from
work. Journal of Work and Organizational Psychology, 34(2), 69–81.
Minzenberg, M. J., Watrous, A. J., Yoon, J. H., Ursu, S., & Carter, C. S. (2008).
Modanil shifts human locus coeruleus to low-tonic, high-phasic activity
during functional MRI. Science, 322(5908), 1700–1702.
Morin, A. J. S., Arens, A. K., & Marsh, H. W. (2015). A bifactor exploratory
structural equation modeling framework for the identication of distinct
sources of construct-relevant psychometric multidimensionality.
Structural Equation Modeling, 23(1), 116–139.
Morin, A. J. S., Boudrias, J. S., Marsh, H. W., McInerney, D. M., Dagenais-Desmarais,
V., Madore, I., & Litalien, D. (2017). Complementary variable-and person-
centered approaches to the dimensionality of psychometric constructs:
Application to psychological wellbeing at work. Journal of Business and
Psychology, 32(4), 395–419.
Moriondo, M., De Palma, P., Medrano, L. A., & Murillo, P. (2012). Adaptación
de la Escala de Afectividad Positiva y Negativa (PANAS) a la población de
adultos de la ciudad de Córdoba: Análisis psicométricos preliminares
[Adaptation of Positive and Negative Aectivity Scale (PANAS) to adults
in Cordoba city: Preliminary psychometric analysis]. Universitas
Psychologica, 11(1), 187–196.
Muthén, L. K., & Muthén, B. O. (1998–2013). Mplus user’s guide (7th ed.).
Muthén & Muthén.
Nahrgang, J. D., Morgeson, F. P., & Hofmann, D. A. (2011). Safety at work: A
meta-analytic investigation of the link between job demands, job
resources, burnout, engagement, and safety outcomes. Journal of
Applied Psychology, 96(1), 71–94.
Newman, D. A., & Harrison, D. A. (2008). Been there, bottled that: Are state
and behavioral work engagement new and useful construct “wines”?
Industrial and Organizational Psychology, 1(1), 31–35.
Nimon, K., & Shuck, B. (2019). Work engagement and burnout: Testing the
theoretical continuums of identication and energy. Human Resource
Development Quarterly, 1–18.
Podsako, P. M., MacKenzie, S. B., Lee, J. Y., & Podsako, N. P. (2003).
Common method biases in behavioral research: A critical review of the
literature and recommended remedies. Journal of Applied Psychology, 88
(5), 879–903.
Reise, S. P. (2012). The rediscovery of bifactor measurement models.
Multivariate Behavioral Research, 47(5), 667–696.
Reise, S. P., Moore, T. M., & Maydeu-Olivares, A. (2011). Targeted bifactor
rotations and assessing the impact of model violations on the para-
meters of unidimensional and bifactor models. Educational and
Psychological Measurement, 71(4), 684–711.
Reise, S. P., Scheines, R., Widaman, K. F., & Haviland, M. G. (2013).
Multidimensionality and structural coecient bias in structural equation
modeling: A bifactor perspective. Educational and Psychological
Measurement, 73(1), 5–26.
Rich, B. L., Lepine, J. A., & Crawford, E. R. (2010). Job engagement:
Antecedents and eects on job performance. Academy of Management
Journal, 53(3), 617–635.
Rodriguez, A., Reise, S. P., & Haviland, M. G. (2015). Evaluating bifactor
models: Calculating and interpreting statistical indices. Psychological
Methods, 21(2), 137–150.
Rodriguez, A., Reise, S. P., & Haviland, M. G. (2016). Applying bifactor
statistical indices in the evaluation of psychological measures. Journal
of Personality Assessment, 98(3), 223–237.
Salanova, M., Llorens, S., Cifre, E., Martínez, I., & Schaufeli, W. B. (2003).
Perceived collective ecacy, subjective well-being and task perfor-
mance among electronic work groups: An experimental study. Small
Group Research, 34(1), 43–73.
Salanova, M., Llorens, S., & Schaufeli, W. (2011). “Yes, I can, I feel good, and I
just do it!” On gain cycles and spirals of ecacy beliefs, aect, and
engagement. Applied Psychology, 60(2), 255–285.
Sapolsky, R. M., Romero, L. M., & Munck, A. U. (2000). How do glucocorti-
coids inuence stress responses? Integrating permissive, suppressive,
stimulatory, and preparative actions. Endocrine Reviews, 21(1), 55–89.
Sass, D., Schmitt, T., & Marsh, H. (2014). Evaluating model t with ordered
categorical data within a measurement invariance framework: A com-
parison of estimators. Structural Equation Modeling: A Multidisciplinary
Journal, 21(2), 167–180.
Schaufeli, W., & De Witte, H. (2017a). Work engagement in contrast to
burnout: Real or redundant? Burnout Research, 5, 1–2.
Schaufeli, W., & De Witte, H. (2017b). Outlook work engagement. Real and
Redundant! Burnout Research, 5, 8–60.
Schaufeli, W., Maslach, C., & Marek, T. (Eds.). (2017). Professional burnout:
Recent developments in theory and research (2nd ed.). Taylor & Francis.
Schaufeli, W. B., Leiter, M. P., Maslach, C., & Jackson, S. E. (1996). Maslach
burnout inventory General survey. In C. Maslach, S. E. Jackson, & M. P.
Leiter (Eds.), The maslach burnout inventory–Test manual (3º ed., pp. 22–
26). Palo Alto, CA: Consulting Psychologists Press.
Schaufeli, W. B., & Bakker, A. B. (2010). Dening and measuring work
engagement: Bringing clarity to the concept. In A. B. Bakker & M. P.
Leiter (Eds.), Work engagement: A handbook of essential theory and
research (pp. 10–24). Psychology Press.
Schaufeli, W. B., & Bakker, A. B. (2003). Test manual for the utrecht work
engagement scale. [Unpublished manuscript]. Utrecht University. http://
Schaufeli, W. B., & Bakker, A. B. (2004). Job demands, job resources, and their
relationship with burnout and engagement: A multi-sample study.
Journal of Organizational Behavior, 25(3), 293–315.
Schaufeli, W. B., & Salanova, M. (2007). Ecacy or inecacy, that’s the
question: Burnout and work engagement, and their relationships with
ecacy beliefs. Anxiety, Stress, and Coping, 20(2), 177–196. https://doi.
Schaufeli, W. B., & Salanova, M. (2011). Work engagement: On how to better
catch a slippery concept. European Journal of Work and Organizational
Psychology, 20(1), 39–46.
Schaufeli, W. B., Salanova, M., González-Romá, V., & Bakker, A. B. (2002). The
measurement of burnout and engagement: A conrmatory factor ana-
lytic approach. Journal of Happiness Studies, 3(1), 71–92.
Schaufeli, W. B., & Taris, T. W. (2005). The conceptualization and measure-
ment of burnout: Common ground and worlds apart. Work and Stress, 19
(3), 256–262.
Schaufeli, W. B., Taris, T. W., & van Rhenen, W. (2008). Workaholism, burnout,
and work engagement: Three of a kind or three dierent kinds of
employee well-being? Applied Psychology: An International Review, 57
(2), 173–203.
Schmitt, T. A., & Sass, D. A. (2011). Rotation criteria and hypothesis testing
for exploratory factor analysis: Implications for factor pattern loadings
and interfactor correlations. Educational and Psychological Measurement,
71(1), 95–113.
Semmer, N., Zapf, D., & Greif, S. (1996). ‘Shared job strain’: A new approach
for assessing the validity of job stress measurements. Journal of
Occupational and Organizational Psychology, 69(3), 293–310. https://
Sijtsma, K. (2009). On the use, the misuse, and the very limited usefulness of
Cronbach’s Alpha. Psychometrika, 74(1), 107–120.
Smits, I. A. M., Timmerman, M. E., Barelds, D. P. H., & Meijer, R. R. (2015). The
Dutch symptom checklist-90-revised: Is the use of the subscales justi-
ed? European Journal of Psychological Assessment, 31(4), 263–271.
Sonnentag, S. (2017). A task-level perspective on work engagement: A new
approach that helps to dierentiate the concepts of engagement and
burnout. Burnout Research, 5, 12–20.
Spontón, C., Medrano, L. A., Maei, L., Sponton, M., & Castellano, E. (2012).
Validación del cuestionario de Engagement UWES a la población de
trabajadores de Córdoba, Argentina [Validation of the engagement
Questionnaire UWES for the population of workers of Córdoba,
Argentina]. Liberabit, 18(2), 147–154.
Spontón, C., Trógolo, M., Castellano, E., & Medrano, L. A. (2019). Medición
del burnout: Estructura factorial, validez y conabilidad en trabajadores
argentinos [Measurement of burnout: Factor structure, validity and relia-
bility in Argentinean]. Interdisciplinaria, 36(1), 87–103. http://www.
Stucky, B. D., & Edelen, M. O. (2014). Using hierarchical IRT models to create
unidimensional measures from multidimensional data. In S. P. Reise & D.
A. Revicki (Eds.), Handbook of item response theory modeling: Applications
to typical performance assessment (pp. 183–206). Routledge/Taylor &
Francis Group.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.
ed.). Pearson Education.
Taris, T. W., Ybema, J. F., & van Beek, I. (2017). Burnout and engagement:
Identical twins or just close relatives? Burnout Research, 5, 3–11. https://
Thompson, B. (2004). Exploratory and conrmatory factor analysis:
Understanding concepts and applications. American Psychological
Tóth-Király, I., Bõthe, B., Rigó, A., & Orosz, G. (2017). An illustration of the
exploratory structural equation modeling (ESEM) framework on the
passion scale. Frontiers in Psychology, 8, 1968.
Urbán, R., Arrindell, W. A., Demetrovics, Z., Unoka, Z., & Timman, R. (2016).
Cross-cultural conrmation of bi-factor models of a symptom distress
measure: Symptom checklist-90-revised in clinical samples. Psychiatry
Research, 239, 265–274.
Ventura, M., Salanova, M., & Llorens, S. (2015). Professional self-ecacy as a
predictor of burnout and engagement: The role of challenge and hin-
drance demands. The Journal of Psychology, 149(3), 277–302. https://doi.
Warr, P. (1990). The measurement of well-being and other aspects of
mental health. Journal of Occupational Psychology, 63(3), 193–210.
Watson, D., Clark, L., & Tellegen, A. (1988). Development and validation of
brief measures of positive and negative aect: The PANAS scales. Journal
of Personality and Social Psychology, 54(6), 1063–1070.
Xiao, Y., Liu, H., & Hau, K. T. (2019). A comparison of CFA, ESEM, and BSEM in
test structure analysis. Structural Equation Modeling: A Multidisciplinary
Journal, 26(5), 665–677.
Yu, C. Y. (2002). Evaluating cuto criteria of model t indices for latent variable
models with binary and continuous outcomes. University of California.
Yung, Y.-F., Thissen, D., & McLeod, L. D. (1999). On the relationship between
the higher-order factor model and the hierarchical factor model.
Psychometrika, 64(2), 113–128.
Zaiţ, A., & Bertea, P. E. (2011). Methods for testing discriminant validity.
Management & Marketing Journal, 9(2), 217–224.
Zumbo, B. D., Gadermann, A. M., & Zeisser, C. (2007). Ordinal versions of
coecients Alpha and Theta for likert rating scales. Journal of Modern
Applied Statistical Methods, 6(1), 21–29.
... In line with this Research about the impact of transformational leadership behavior on leaders' general health and work engagement is missing so far, and prior empirical studies have exclusively focused on health impairment indicators. Nevertheless, persisting states of emotional exhaustion will deteriorate general health and are related to work engagement (as emotional exhaustion and work engagement are negatively correlated and yet distinct constructs; Trógolo et al., 2020). ...
Full-text available
While profound evidence indicates certain kinds of leadership behavior promote followers' well‐being, little is known about the role of leaders' own well‐being for leading in a beneficial way. This study examined the link between leaders' well‐being, which is reflected by perceived general health and work engagement (dedication and vigor), their self‐appraised transformational leadership behavior, and team performance in a sample of 276 leaders (55.8% women; mean age = 45 years) across three waves. Building on Hobfoll's Conservation of Resources theory, we assumed that leaders' well‐being functions as a resource and, therefore, facilitates transformational leadership behavior, resulting in higher team performance. Moreover, we explored if this ultimately means a resource gain or loss for the leader. The results of cross‐lagged path‐analyses indicated a significant positive effect of leaders' well‐being on their transformational leadership behavior over time and their general health on team performance. However, our results further indicate that through transformational leaders' investment in higher team performance, the leaders' well‐being over time might be diminished, indicating a resource loss. The present findings provide support for the assumption that leaders' well‐being is a requirement for their ability to execute transformational leadership behavior, which in turn fosters team performance and subsequently depletes leaders' resources. Practical implications concerning leaders' well‐being and the benefits for organizations as well as implications for future research focused on Conservation of Resources theory are highlighted and discussed.
... 10 In fact, work engagement as a positive phenomenon among employees can overcome destructive phenomena such as burnout, mistakes, sabotage and work weaknesses. 11 In nursing, work engagement is associated with an increase in nurses' efficiency, improvement in patient survival and safety, financial productivity and other positive organizational results. 12 Given the importance and effectiveness of the work engagement and the existence of job problems in the operating room, hospital managers must attempt to strengthen work engagement among surgical technologists in order to reduce their job problems. ...
Background : Work engagement can be a savior phenomenon for surgical technologists who will experience a high level of job stress, workload, Occupational Injury, and burnout. Therefore, the present study aimed to provide a model of work engagement in surgical technologists. Methods : This qualitative study was performed based on the grounded theory method at medical sciences University of Hamadan in 2020. Sampling was purposeful and continued until data saturation. Sixteen surgical technologists, 2 operating room head nurses, and 2 surgeons were interviewed using Semi-structured interviews . This study was carried out based on four criteria of Lincoln and Guba, and the data were analyzed using the Strauss-Corbin method. Results : The financial appreciation and the professional competence as causal conditions, interest in work and the work cognition as context conditions, the organizational support and the organizational justice as intervening conditions can have influence on the work engagement of surgical technologists. It was also shown that the Workplace attraction and Communicative skills as strategies can enhance the work engagement of surgical technologists, leading to responsibility, achieving organizational goals and job satisfaction. Conclusion : Results of this study reveal that the job motivation of surgical technologists can be affected by various factors, so hospital managers must use effective strategies to promote this phenomenon.
Full-text available
The worldwide spread of work‐related mental unhealth suggests that this is a major problem affecting organizations and employees on a global scale. In this paper, we therefore provide a thematic review of the literatures that address this issue in management and organization studies (MOS) and related fields. While these literatures examine how employee mental health is affected by organizational and occupational structures and managed by organizations and employees, they have paid relatively little attention to the capitalist labour relations which underpin the unhealthy conditions of contemporary working life. They have paid even less attention to how these conditions may be resisted. To help future scholarship in MOS challenge this state of affairs, we draw on some of the most basic but central notions of exploitation, alienation and resistance in classic and current critiques of capitalism, optimistic that this may help strengthen the field's capacity to confront mental unhealth in settings of work and organization.
Full-text available
Using meta‐analytic correlations from Cole et al. (2012), we conducted secondary data analysis to further explore the empirical overlap between measures of work engagement (Utretch Work Engagement Scale) and burnout (Maslach Burnout Inventory). We found that the dimensions of work engagement and burnout did not align with previously positioned theoretical continuums. In addition to finding a negative association between the aspects of dedication and inefficacy, we found that aspects of vigor and absorption were positively related to cynicism and exhaustion, respectively. We also found that very little of the shared variance between the measures of work engagement and burnout were the result of semantic equivalence between item sets.
Full-text available
El burnout representa uno de los daños de carácter laboral y psicosocial más importantes en el mundo actual. En el presente trabajo se analiza la estructura factorial, la validez y la confiabilidad de una medida de burnout en Argentina, compuesta por las escalas del Maslach Burnout Inventory-General survey (MBI-GS)y la subescala de despersonalzación correspondiente al Maslach Burnout Inventory-Human services. Se utilizó una muestra de 1903 trabajadores del sector público, privado y organizaciones sin fines de lucro. Para evaluar la estructura interna se pusieron a prueba un total de seis modelos. Los resultados obtenidos por medio de análisis factorial confirmatorio, indican que un modelo de dos factores correlacionados, compuesto por las dimensiones corazón del burnout (agotamiento y cinismo), es el que posee un mejor ajuste a los datos. Por su parte, los análisis de confiabilidad evidenciaron una consistencia interna aceptable, como así también una adecuada fiabilidad de constructo para los factores. Finalmente, se obtuvieron correlaciones significativas y en la dirección esperada entre los factores del burnout y medidas de engagement y afecto, proporcionando así evidencia externa de validez concurrente para los puntajes de la escala. Globalmente, los resultados son satisfactorios y avalan el uso del MBI-GS en el contexto argentino, aunque se requiere de nuevos estudios que examinen otras propiedades psicométricas relevantes. Se discuten las implicaciones de este trabajo para la evaluación y la investigación sobre burnout en Argentina
Full-text available
El afecto o emoción como tema de investigación científica ha despertado un gran interés en los últimos años (Isen, 2007). Aunque las emociones experimentadas por las personas constituyen un tópico de gran relevancia actual y con múltiples y variadas repercusiones en el ejercicio profesional del psicólogo, en nuestro medio no se cuenta con instrumentos psicométricos adaptados que permitan medir de manera válida y confiable dicha variable. Tomando esto en consideración, el presente trabajo tuvo por objetivo realizar una adaptación psicométrica de la Escala de Afectividad Positiva y Negativa (PANAS; Watson & Clark, 1988) a la población de adultos de la ciudad de Córdoba. Con este fin, luego de realizar un estudio piloto, se administró el PANAS a una muestra de 205 adultos con edades comprendidas entre los 25 y 65 años. Los estudios de estructura interna realizados mediante el Análisis Factorial Exploratorio sugieren la existencia de dos dimensiones que explican un 39% de la variabilidad de la prueba. Se calculó además la correlación ítem total y el índice de discriminación de cada ítem y se examinó la consistencia interna de la escala utilizando el coeficiente Alfa de Cronbach (0.73; 0.82). Finalmente, se obtuvieron evidencias externas contrastando grupos según la edad, género y nivel de estudio de los participantes. Los resultados son similares a los observados en estudios antecedentes (Robles y Páez, 2003; Sandin et al, 1999). Aunque se requiere de investigaciones adicionales, los análisis preliminares efectuados sostienen el uso de la escala PANAS en la población de adultos de nuestro medio.
Full-text available
This study illustrates complementary variable- and person-centred approaches to the investigation of the underlying dimensionality of the work engagement construct. A sample of 730 participants completed a questionnaire twice across a four-month period. The results showed that employees’ ratings of their work engagement simultaneously reflected a global overarching work engagement construct, which co-existed with three specific dimensions (vigour, dedication, and absorption). Relying on factor scores from this initial measurement model, the present study examined latent profiles of employees defined based on their global (work engagement) and specific (vigour, dedication, and absorption) levels of work engagement. The results revealed five distinct work engagement profiles, which proved to be fully identical, and highly stable, across the two time points. These profiles characterized disengaged-vigorous, normative, totally disengaged, vigorously absorbed, and engaged yet distanced employees. These profiles were also showed to be meaningfully related to employees’ levels of stress, intentions to leave the organization, health, and job satisfaction.
Full-text available
Bifactor models have gained increasing popularity in the literature concerned with personality, psychopathology and assessment. Empirical studies using bifactor analysis generally judge the estimated model using SEM model fit indices, which may lead to erroneous interpretations and conclusions. To address this problem, several researchers have proposed multiple criteria to assess bifactor models, such as a) conceptual grounds, b) overall model fit indices, and c) specific bifactor model indicators. In this article, we provide a brief summary of these criteria. Using data from a recent publication on factorial structure of the Positive and Negative Affect Schedule (Seib-Pfeifer, Pugnaghi, Beauducel, & Leue, 2017), we illustrate errors and misinterpretation that can result from exclusively relying on SEM model fit indices without taking into consideration certain theoretical basis and more specific bifactor model indices. We want to show how taking into account all criteria may prevent researchers from drawing wrong conclusions.
Full-text available
Previous research has demonstrated the impact of various life domains on employee well-being. However, these domains have been commonly examined separately. In addition, most existing studies on this topic stem from North America and Western European countries, particularly Spain and Netherlands. Comparatively, little research has been conducted in Latin American countries. The aim of this research was to develop and test a model of employee well-being in Argentina. One thousand and sixty employees from a national representative sample completed measures of leisure, psychological detachment from work, job resources, work-family conflict, work-related well-being (engagement and burnout), and subjective well-being (life satisfaction). Results from structural equation modeling indicated that the model fit the data well. We discuss practical implications of the findings for employee well-being and suggest future research building upon study limitations that may contribute to a more refined understanding of the results outlined in this study.
Full-text available
More profound understanding of the relationship between the burnout and the limbic system function can provide better insight into brain structures associated with the burnout syndrome. The objective of this review is to explore all evidence of limbic brain structures associated with the burnout syndrome. In total, 13 studies were selected. Four of them applied the neuroimaging technology to investigate the sizes/volumes of the limbic brain structures of burnout patients. Six other studies were to investigate the hypothalamus-pituitary-adrenal (HPA) axis of burnout patients. Based on the results of the studies on the HPA-axis and neuroimaging of the limbic brain structures, one can see great impact of the chronic occupational stress on the limbic structures in terms of HPA dysregulation, a decrease of BDNF, impaired neurogenesis and limbic structures atrophy. It can be concluded that chronic stress inhibits the feedback control pathway in the HPA axis, causes the decrease of brain-derived neurotrophic factor (BDNF), then impaired neurogenesis and eventually neuron atrophy.
Full-text available
While exploratory factor analysis (EFA) provides a more realistic presentation of the data with the allowance of item cross-loadings, confirmatory factor analysis (CFA) includes many methodological advances that the former does not. To create a synergy of the two, exploratory structural equation modeling (ESEM) was proposed as an alternative solution, incorporating the advantages of EFA and CFA. The present investigation is thus an illustrative demonstration of the applicability and flexibility of ESEM. To achieve this goal, we compared CFA and ESEM models, then thoroughly tested measurement invariance and differential item functioning through multiple-indicators-multiple-causes (MIMIC) models on the Passion Scale, the only measure of the Dualistic Model of Passion (DMP) which differentiates between harmonious and obsessive forms of passion. Moreover, a hybrid model was also created to overcome the drawbacks of the two methods. Analyses of the first large community sample (N = 7,466; 67.7% females; Mage = 26.01) revealed the superiority of the ESEM model relative to CFA in terms of improved goodness-of-fit and less correlated factors, while at the same time retaining the high definition of the factors. However, this fit was only achieved with the inclusion of three correlated uniquenesses, two of which appeared in previous studies and one of which was specific to the current investigation. These findings were replicated on a second, comprehensive sample (N = 504; 51.8% females; Mage = 39.59). After combining the two samples, complete measurement invariance (factor loadings, item intercepts, item uniquenesses, factor variances-covariances, and latent means) was achieved across gender and partial invariance across age groups and their combination. Only one item intercept was non-invariant across both multigroup and MIMIC approaches, an observation that was further corroborated by the hybrid model. While obsessive passion showed a slight decline in the hybrid model, harmonious passion did not. Overall, the ESEM framework is a viable alternative of CFA that could be used and even extended to address substantially important questions and researchers should systematically compare these two approaches to identify the most suitable one.
Minor cross-loadings on non-targeted factors are often found in psychological or other instruments. Forcing them to zero in confirmatory factor analyses (CFA) leads to biased estimates and distorted structures. Alternatively, exploratory structural equation modeling (ESEM) and Bayesian structural equation modeling (BSEM) have been proposed. In this research, we compared the performance of the traditional independent-clusters-confirmatory-factor-analysis (ICM-CFA), the nonstandard CFA, ESEM with the Geomin- or Target-rotations, and BSEMs with different cross-loading priors (correct; small- or large-variance priors with zero mean) using simulated data with cross-loadings. Four factors were considered: the number of factors, the size of factor correlations, the cross-loading mean, and the loading variance. Results indicated that ICM-CFA performed the worst. ESEMs were generally superior to CFAs but inferior to BSEM with correct priors that provided the precise estimation. BSEM with large- or small-variance priors performed similarly while the prior mean for cross-loadings was more important than the prior variance.