How Other - and Self-Compassion Reduce Burnout Through Resource Replenishment

r Academy of Management Journal
2022, Vol. 65, No. 2, 453–478.

https://doi.org/10.5465/amj.2019.0493
HOW OTHER- AND SELF-COMPASSION REDUCE

BURNOUT THROUGH RESOURCE REPLENISHMENT
KIRA SCHABRAM
YU TSE HENG
University of Washington
The average employee feels burnt out, a multidimensional state of depletion likely to per-
sist without intervention. In this paper, we consider compassion as an agentic action by
which employees may replenish their own depleted resources and thereby recover. We
draw on conservation of resources theory to examine the resource-generating power of two
distinct expressions of compassion (self- and other-directed) on three dimensions of burn-
out (exhaustion, cynicism, inefficacy). Utilizing two complementary designs—a longitudi-
nal field survey of 130 social service providers and an experiential sampling methodology
with 100 business students across 10 days—we find a complex pattern of results indicating
that both compassion expressions have the potential to generate salutogenic resources
(self-control, belonging, self-esteem) that replenish different dimensions of burnout. Specif-
ically, self-compassion remedies exhaustion and other-compassion remedies cynicism—
directly or indirectly through resources—while the effects of self- and other-compassion on
inefficacy vary. Our key takeaway is that compassion can indeed contribute to human sus-
tainability in organizations, but only when the type of compassion provided generates
resources that fit the idiosyncratic experience of burnout.
“Whenever the topic of job burnout gets raised, the 2017) including diabetes, heart disease, and pre-
key question is ‘What can we do about it?’” (Maslach, mature death (Alarcon, 2011).
2017: 143) These deleterious consequences make understand-
The World Health Organization (2019) recently ing potential recovery mechanisms for addressing
added burnout to its international classification of dis- burnout of vital importance. To date, remedies have
eases that significantly impair health. In the United largely focused on passive activities that treat those
States, more than half of employees report such burn- experiencing burnout as the object of, rather than an
out (Gallup, 2018), described as a multidimensional active participant in, their own recovery. Burnout
state of depletion negatively related to performance scholars have advocated for organizational interven-
and commitment (for meta-analyses, see Lee & tions that lessen work demands or increase social
Ashforth, 1996; Swider & Zimmerman, 2010). Left support from others (Maslach, 2017). Meanwhile,
the organizational recovery literature has prioritized
unchecked, burnout persists over time (Maslach,
reprieve from work through psychological detach-
2017), resulting in an estimated $109 billion in
ment, sleep, vacations, or breaks (Sonnentag, Venz, &
annual U.S. health-related consequences (Garton,
Casper, 2017; Trougakos, Beal, Green, & Weiss, 2008).
It is somewhat paradoxical, however, that, although
a persistent assumption about burnout is that “the
problem lies within the person” (Maslach, 2003:
We would like to thank associate editor Jessica Rodell 191), scholars also suggest that “chronically burned-
and the three anonymous reviewers for their exceptional out employees or those at risk for burnout need
guidance and patience. This manuscript also benefited help from others in order to change” (Bakker &
from much generous feedback. We are grateful to Emily Costa, 2014: 117).
Cox-Pahnke, Ashley Hardin, Michael Johnson, Klodiana
Taking our inspiration from conceptual work that
Lanaj, Keith Leavitt, Chris Myers, Scott Reynolds, Sandra
argues individuals can play an agentic role in their
Robinson, Danielle van Jaarsveld, and Amy Wrzesniewski
as well as to seminar participants of the Harvard Business own recovery, and that acts of care may be one such
School, Oregon State University, the University of Massa- pathway (Lilius, 2012), our guiding research question
chusetts Amherst, and the Wharton Organizational Behav- is: When can compassionate action reduce an actor’s
ior Conference. Funding was provided in part by the Social own burnout? Building on conservation of resources
Sciences and Humanities Research Council of Canada. theory principles (COR; Hobfoll, 1989) that employees
453
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder's express
written permission. Users may print, download, or email articles for individual use only.
454 Academy of Management Journal April
must combat resource loss by engaging in replenishing have focused on recovery away from work or directed
activities, we propose that compassion is most restor- by others, we identify agentic behaviors by which
ative when its particular expression generates the very individuals may reduce their own burnout while at
resources depleted by burnout. We consider this work—and conversely also indicate the limits of such
compassion–burnout alignment by attending to the actions. Moreover, by highlighting the distinct recov-
multidimensional nature of the burnout experience— ery patterns for each burnout dimension, we engage
comprising exhaustion, cynicism, and inefficacy—and with the multidimensional nature of burnout and
the variety of resource categories that compassionate underscore concerns about collapsing the construct
acts might generate (energies, conditions, attributes). into exhaustion-only models (Leiter & Maslach, 2016).
Specifically, we hypothesize that self-compassion can Indeed, such an approach would have missed our
remedy exhaustion by increasing self-control (a type of robust results for the cynicism dimension, which is
energy resource), other-compassion can remedy cyni- particularly relevant to the study of employees (Dean,
cism by fostering a sense of belonging (a type of condi- Brandes, & Dharwadkar, 1998). Our work also answers
tion resource), and both can remedy inefficacy by
calls to explain whether and why compassion matters
increasing self-esteem (a type of personal attribute
at work (Dutton, Workman, & Hardin, 2014; Rynes,
resource). We test our model (see Figure 1, below)
Bartunek, Dutton, & Margolis, 2012), extending our
across two complementary samples (full-time social
understanding in two complementary ways. First, we
service providers vs. business school students) and
shift empirical focus from compassion recipients to
two experiences of burnout (chronic vs. acute). We
also utilize two distinct methodologies (a longitudinal those offering it, highlighting the counter-intuitive
field survey and an experiential sampling methodol- insight that caring for others must not come at a cost to
ogy) in order to provide a robust exploration of this the self, a trade-off remarked upon for other prosocial
important phenomenon. Our results highlight a com- work gestures (see Bolino & Grant, 2016; Lanaj, John-
plex pattern of relationships between compassion, son, & Wang, 2016; Uy, Lin, & Ilies, 2017). Second, we
resources, and burnout such that self-compassion introduce the concept of self-compassion from psy-
reduces exhaustion and other-compassion reduces chology and map its unique process and impact when
cynicism—directly in Study 1 and indirectly in Study compared to other-compassion, thereby stressing the
2—while the effects of self- and other-compassion on necessity of distinguishing compassion targets. We
inefficacy are more varied. also hope that these insights may offer guidance to
Our work offers a number of important insights for practitioners eager to embrace compassion as a way to
research on burnout. Whereas previous interventions manage for human sustainability.
FIGURE 1
Proposed Theoretical Model
Initial Burnout Compassion Resources Later Burnout
–
Exhaustion Self-Control Exhaustion
Self-Compassion +
–
Inefficacy Self-Esteem Inefficacy
Other-Compassion +
–
Cynicism Belonging Cynicism
Study 2
Study 1
2022 Schabram and Heng 455
LITERATURE REVIEW AND MODEL dimensions into one or by considering only exhaustion
DEVELOPMENT (for recent examples in the Academy of Management
Journal, see Baer, Dhensa-Kahlon, Colquitt, Rodell,
The starting point of our model—actor burnout—
Outlaw, & Long, 2015; Grant, Berg, & Cable, 2014; Lin,
represents a state of resource depletion comprising
Scott, & Matta, 2019). Indeed, exhaustion-only mod-
three dimensions (Hobfoll, Halbesleben, Neveu,
els have received by far the most research attention
& Westman, 2018; Lee & Ashforth, 1996). The
(Schaufeli, Leiter, & Maslach, 2009). Leading burn-
exhaustion dimension refers to feelings of being
out scholars, however, are critical of this develop-
overextended or depleted in physiological resources.
Exhaustion correlates with work demands (Maslach, ment, arguing against efforts to “rename exhaustion
2003) and relates closely to prior conceptualization as burnout” (Maslach & Leiter, 2016b: 108). Exhaus-
of strain (Gaines & Jermier, 1983) and ego depletion tion alone can be indicative of a host of ailments
(Baumeister, Faber, & Wallace, 1999). The cynicism1 other than burnout (Leiter & Maslach, 2016) and fails
dimension refers to a callous and diminished connec- to capture the fact that burnt out people do not simply
tion to various aspects of one’s work, including one’s feel tired but also alienated and dejected (Pines, 2017).
clients, coworkers, or the work itself. Cynicism repre- Following this guidance, we consider the influence of
sents a state of social depletion characterized by feel- compassion on all three burnout dimensions.
ings of rejection and alienation. Finally, inefficacy As noted by our introductory quote, a driving push
captures the self-evaluative dimension of burnout and in the burnout literature is to identify solutions. To
refers to a depleted sense of one’s self or accomplish- date, efforts to remedy burnout have focused on what
ments. Inefficacy is closely intertwined with perfor- others can do to help the person, including re-engaging
mance, both resulting from evaluations of reduced employees, providing social support, improving their
performance (Maslach & Leiter, 2016a) as well as work conditions, or teaching coping techniques (for
being predictive of subpar performance (Swider & a review, see Maslach, 2017). Such other-directed
Zimmerman, 2010). approaches have been subject to the criticism that
The interplay of these three dimensions is not yet they are patronizing at best or stigmatizing at worse,
well understood. Early research sought to map the leaving employees reluctant to share their burnout
dimensions sequentially, with exhaustion commonly, with others. Moreover, many such proposed solu-
but not always, being the first indicator of burnout (for tions have not undergone empirical scrutiny (for a
a review of early research, see Maslach, Schaufeli, & review, see Maslach & Goldberg, 1998). Our empha-
Leiter, 2001). In contrast, cross-sectional studies have sis on agency and theory testing seeks to assuage
generally suggested that exhaustion and cynicism are such concerns.
correlated with each other, while inefficacy is not (Lee Before proceeding, we note one additional charac-
& Ashforth, 1996; Maslach, 2003). Most recently, teristic of burnout important to our model: burnout
empirical work has pursued latent profile analysis to is rarely an isolated phenomenon. It develops pri-
compare “full burnout” involving all three dimen- marily in social contexts (Buunk & Schaufeli, 1993),
sions to various “partial burnout” profiles (e.g., high both because of the shared experience of contextual
exhaustion and cynicism only; Leiter & Maslach, demands and because burnout can be contagious
2016). Consistent across these research approaches are (Alarcon, 2011; Gonzalez-Morales, Peir o, Rodrıguez,
insights that the three burnout dimensions are distinct & Bliese, 2012; Hobfoll et al., 2018). We stress this
but interrelated (Leiter & Maslach, 2016), and that any because it means that those experiencing burnout
given person can experience various patterns that are are commonly surrounded by others who are as
likely to change over time. well. This lived reality makes compassion a poten-
Despite this multidimensional richness, most tially important agentic remedy because it can be
scholars have moved toward treating burnout as a offered to oneself or others.
singular construct, either by collapsing the three
Compassion as Other- and Self-Care
1 Organizational scholars conceptualize compassion
Early studies of burnout were conducted in social ser-
vice occupations and this detachment dimension (then
as a dynamic, interpersonal process that begins with
termed “depersonalization”) referred only to distance from noticing and empathizing with another’s pain and
service recipients. Depersonalization was reconceptual- culminates in a response to alleviate it (Kanov, Mai-
ized as “cynicism” to be occupation neutral and capture tlis, Worline, Dutton, Frost, & Lilius, 2004). This
broader alienation from all aspects of work. emphasis on the “activation and mobilization of
action triggered by human pain” (Dutton, Worline, up for a yoga class or social activity, or writing a per-
Frost, & Lilius, 2006: 62) represents a unique schol- sonal reflection acknowledging difficulty. To date,
arly focus and is an essential aspect of our upcoming self-compassion has rarely been studied at work (for
theoretical model. In contrast, psychologists primarily exceptions, see Kreemers, van Hooft, & van Vianen,
study the affective nature of compassion and its asso- 2018; Shepherd & Cardon, 2009). Further, few studies
ciated emotions (e.g., empathy, sympathy, warmth, have considered both compassion expressions, which
pity, etc.)—what organizational scholars consider could advance our understanding of the complexity of
the second step of the compassion process—while compassion and its impact. We are, however, encour-
acknowledging that such emotions ultimately pre- aged to pursue this comparative approach by a select
dict action (i.e., the third step; see Goetz, Keltner, & number of surveys, albeit cross-sectional, that suggest
Simon-Thomas, 2010). that the two expressions are distinct but positively cor-
In organizations, compassionate action can take a related (Crocker & Canevello, 2008; Neff & Beretvas,
range of forms, from passively listening to the suf- 2013) at least in specific samples (for an exception,
ferer to actively investing abstract or tangible resour- see L opez, Sanderman, Ranchor, & Schroevers,
ces into their care (Dutton et al., 2006). For example, 2018): older adults and those practicing meditation
one could invite a suffering coworker to a yoga class, report greater self- and other-focused concern (Pom-
organize a social activity, or leave a kind note on mier, 2010), while both are lower in students high in
their desk acknowledging their difficulty. Among self-judgment (Beaumont, Durkin, Martin, & Carson,
the family of organizational prosocial actions, com- 2016). Moreover, both can be cultivated by the same
passion thus distinguishes itself not in form (i.e., lis- practices (Gilbert & Irons, 2005), suggesting that
tening to a coworker is not unique to compassion), employees have the capacity to engage in either or
but in intention (i.e., the reason for listening is). both compassion expressions.
Whereas other prosocial actions like citizenship
behaviors are ultimately intended to promote the The Restorative Potential of Compassion
effective functioning of organizations (Organ, 1988), for Burnout
compassion is directly targeted at human needs
(Tsui, 2013).2 We link the depleting experience of burnout and
To date, compassion scholars have primarily stud- restorative potential of compassion through the salu-
ied its impact on recipients. Promising insights sug- togenic lens of COR theory and its two key principles
gest that receiving compassion speeds recovery from for how resources are allocated (Halbesleben, Neveu,
pain, reduces anxiety, and makes people feel valued Paustian-Underdahl, & Westman, 2014).
(for a review, see Dutton et al., 2014). Some research The primacy of resource loss principle indicates that
also highlights potential benefits for the actor; namely, resource loss is more salient than resource gain and
that those who offer compassion are viewed in a more disproportionate in terms of its degree and speed (Hob-
positive light (Melwani, Mueller, & Overbeck, 2012). foll et al., 2018). Burnout functions according to this
Tangential research on prosocial gestures (Bolino & principle because early experiences deplete resources,
Grant, 2016; Lanaj et al., 2016; Uy et al., 2017), how- making it more difficult to contend with subsequent
ever, warns that such benefits likely come with cav- demands (Bakker & Costa, 2014; Maslach & Leiter,
eats related to the exhaustion dimension of burnout: 2016b). Burnout “represents erosion in values, dignity,
helping others takes time and effort away from the self spirit, and will—an erosion of the human soul. It is a
and is therefore costly and depleting. malady that spreads gradually and continually over
Intriguingly, psychologists have expanded the con- time” (Maslach & Leiter, 2008: 17). Indeed, we see this
ceptualization of compassion to that which is directed principle play out for each of the burnout dimensions.
at and intended to benefit the self (for a review, see For instance, early exhaustion diminishes self-control
Barnard & Curry, 2011). “Self-compassion” mirrors (Baumeister et al., 1999) making it harder for the
other-compassion, but entails offering kindness to one- depleted individual to protect themselves from further
self (Neff, 2003a, 2003b). As with other-compassion, work demands (Bakker & Costa, 2014), while early
acts of self-care can take various forms, such as signing cynicism means individuals can’t suppress behaviors
such as complaining that will further alienate them
2
We stress, however, that these behaviors are not mutu- from others (Gonzalez-Morales et al., 2012), and early
ally exclusive and may share downstream consequences. inefficacy shuts individuals out of decision-making
Attending to an employee’s human needs can lead to (Schaufeli et al., 2009; ten Brummelhuis, ter Hoeven,
improved downstream performance and vice versa. Bakker, & Peper, 2011) putting them in a position
of increasingly diminished self-esteem. Thus, a One quintessential COR energy resource that past
foundational prediction of our paper is that people research has identified as helping against exhaustion
who are experiencing burnout will continue to do so is self-control (Lanaj, Johnson, & Barnes, 2014),
in the absence of an appropriate intervention. which matters because a reserve helps individuals
withstand exertion (Baumeister, 2002). Self-control
Hypothesis 1a. Past exhaustion is positively related to
future exhaustion. is particularly suited as an agentic resource to restore
exhaustion as it is often referred to as a metaphorical
Hypothesis 1b. Past cynicism is positively related to “muscle” that the focal individual can strengthen
future cynicism. (Baumeister et al., 1999). Self-control can be gener-
ated via rest activities in the form of breaks, formal
Hypothesis 1c. Past inefficacy is positively related to time off, or improved sleep (Sonnentag et al., 2017;
future inefficacy. Trougakos, Hideg, Cheng, & Beal, 2014), or via more
active forms of self-care such as exercising, improv-
To stop and reverse such loss, the resource invest-
ing nutrition (Leiter & Maslach, 2016), or practicing
ment principle suggests that employees must invest in
mindfulness (H€ ulsheger, Lang, Depenbrock, Fehr-
resource-generating activities. The unit of analysis in
mann, Zijlstra, & Alberts, 2014).
COR is the resources that can be generated (Halbesle-
Self-compassion is likely to manifest in these kinds
ben et al., 2014). Resources fall into three categories
of self-control-restorative behaviors because it brings
(Hobfoll, 1989), all of which have been previously
into focus failures or depletion and motivates personal
operationalized to link depletion and recovery3 (Baer
initiative to tackle them (Neff, Kirkpatrick, & Rude,
et al., 2015; ten Brummelhuis & Bakker, 2012). Ener-
2007). Self-compassion is predictive of self-regulatory
gies refer to emotional and physical resources that
and health-related behaviors including seeking and
have little value in their own right (e.g., sense of
sticking to treatment (Terry & Leary, 2011; Terry,
self-control, time), but can be spent to accomplish
Leary, Mehta, & Henderson, 2013). Indeed, contrary
goals or acquire other resources; conditions relate to
to the notion that self-compassion promotes self-
one’s positive state and social circumstances (e.g.,
indulgence or passivity, it has been shown to predict
sense of belonging, status); and attributes are personal
greater responsibility for recovery (Leary, Tate, Adams,
characteristics, attitudes, or skills that aid the individ- Batts Allen, & Hancock, 2007). In short, we predict that
ual, particularly in the realm of stress resistance (e.g., individuals suffering from exhaustion can recover
sense of self-esteem, health). Below, we outline our when they engage in self-compassionate activities that
predictions for how self- and other- compassionate generate self-control.
actions can generate resources with the potential to
reduce burnout. Specifically, we propose that each Hypothesis 2. (a) Self-compassion is negatively
resource category is relevant for each particular burn- related to the exhaustion dimension of burnout,
which is (b) mediated by self-control.
out dimension, with self-control (an energy resource)
being negatively related to exhaustion, belonging (a In contrast, we do not predict other-compassion
condition resource) to cynicism, and self-esteem (an will reduce exhaustion by generating self-control.
attribute resource) to inefficacy. On the one hand, other-compassion likely saps the
Exhaustion. As previously noted, the exhaustion self-control reserve because of its inherent self-
dimension refers to feelings of being overextended, sacrificial impulse. Research on organizational citi-
strained (Gaines & Jermier, 1983) or ego depleted zenship (Bergeron, 2007; Bergeron, Shipp, Rosen, &
(Baumeister et al., 1999). To date, burnout scholars Furst, 2011; Lanaj et al., 2016) and COR (Hobfoll,
have proposed that organizations or supervisors 1989) has consistently warned that supporting others
reduce workload to reduce exhaustion (Maslach & can come at a personal cost. For instance, other-
Leiter, 2016a). In accordance with COR’s resource compassion might require sacrificing a much-needed
investment principle, however, we are interested in break to care for a suffering coworker. This persis-
identifying resource-generative activities employees tent focus on others’ needs runs the risk of giving
could initiate themselves to remedy exhaustion. up control not only over one’s time and energy but
also general sense of self (Lilius, 2012) as one is
3
We exclude objects (e.g., tools), the fourth original increasingly out of touch with one’s own needs. At
resource category, from our theorizing because acquiring the extreme, this has been documented by practi-
physical goods has not been examined as a relevant rem- tioners examining “compassion fatigue,” a condi-
edy for burnout and is unlikely to function as such. tion wherein individuals “lose control” (Figley,
2013: 154) by over-emphasizing others’ problems. suited to connecting with others. Self-compassion
Thus, other-compassion has the potential to sap self- deactivates the threat system associated with feel-
control. ings of insecurity (Gilbert & Irons, 2005) and thus
On the other hand, this relationship may not be makes individuals’ positive self-attitudes less con-
entirely negative. In contrast to traditional citizen- tingent on others (Neff, 2003a). The focal actor prac-
ship behaviors that are more bounded by work ticing self-compassion therefore has fewer motives
requirements or leadership demands (Zhao, Peng, & to engage with others at work. Furthermore, self-
Chen, 2014), other-compassion’s focus on human compassion activates the self-soothing system (Gil-
needs likely gives the individual more leeway and bert & Irons, 2005), making the employee less reliant
thus has the potential to enhance self-control. For on others for support. Finally, some empirical evi-
instance, they may care for another by proposing dence suggests that focusing one one’s own suffering
activities they consider personally restorative, such makes one less caring toward others going through
as taking a yoga class together. Thus, we propose the same experiences (Ruttan, McDonnell, & Nordg-
that only self-compassion generates self-control to ren, 2015).
reduce exhaustion, while other-compassion neither That having been said, we do acknowledge that
reduces nor contributes to exhaustion. self-compassion traces its origins to Buddhist notions
Cynicism. Cynicism occurs when an employee of social connectedness, and Neff (2003a) has stressed
experiences alienation from the various aspects of that a defining feature is the recognition of “common
their work such as the work itself or other people in humanity.” Empirical evidence also indicates that
it (i.e., clients, coworkers). Burnout scholars have self-compassion correlates positively with some
suggested that cynicism might be remedied by self-ratings of social connectedness (Barnard & Curry,
addressing the latter; that is, reaffirming a person’s 2011; Neff, 2003b; Neff et al., 2007). Thus, taken
membership and sense of community (Maslach & together, these impulses may cancel each other out,
Leiter, 2008). Interventions whereby others support since individuals acting self-compassionately may dis-
or reintegrate the burnt out employee, however, tance themselves from specific others at work while
have not always been successful (Halbesleben, also embracing a broader sense of common humanity.
2006). Building again on COR’s resource investment Therefore, self-compassion should have no noticeable
principle, we propose that reducing cynicism might effect on belonging or subsequent cynicism.
require the individual to reaffirm their own social Inefficacy. Inefficacy is associated with a lack of
connections and thereby generate feelings of belong- accomplishment and feelings of ineffectiveness
ing (Lambert, Stillman, Hicks, Kamble, Baumeister, (Maslach, 2003). We propose that recovery could be
& Fincham, 2013). facilitated by enhancing self-esteem, defined as
We propose that other-compassion, as an affilia- one’s evaluation of and attitude toward oneself
tive, self-transcendent action (Halbesleben & Whee- (Rosenberg, 1965) and identified as a prototypical
ler, 2015), may be well suited to the task. We base example of COR’s personal attributes category of
this on past findings that giving compassion results resources (Hobfoll, 1989). When employees view
in greater affiliative feelings of commitment (Grant, themselves in a more positive light, this not only
Dutton, & Rosso, 2008), relatedness (Lilius, 2012), improves their subjective evaluation of their own
and trust (Crocker & Canevello, 2008). Individuals accomplishments (Farh & Dobbins, 1989), but also
who give other-compassion also report perceiving motivates efforts to pursue further accomplishments
more compassion from others (Lemay & Clark, 2008) (Sommer & Baumeister, 2002). We propose that both
thereby indicating mutual attachment (Grant et al., self- and other-compassionate actions can increase
2008). Thus, offering other-compassion should self-esteem and thereby replenish inefficacy, though
increase the perception (Lemay & Clark, 2008) and for different reasons.
actualization (Crocker & Canevello, 2008) of a sense Other-compassion is likely to enhance self-esteem
of belonging, which in turn should reduce cynicism. by allowing the individual to see their own value
vis-a-vis their impact on others. Indeed, other-
Hypothesis 3. (a) Other-compassion is negatively compassion has already been identified as one oppor-
related to the cynicism dimension of burnout, which
tunity to increase self-esteem by benefitting others
is (b) mediated by belonging.
(Canevello & Crocker, 2011; Mongrain, Chin, & Sha-
In contrast, we do not predict self-compassion to pira, 2011). Moreover, we know that positive recogni-
reduce cynicism by generating feelings of belonging, tion from others can be self-enhancing (Cordes &
because self-compassionate activities are not as Dougherty, 1993). Thus, we reason that the focal
individual will evaluate themselves more positively do so, we recruited employees from a West Coast
when they care for a suffering coworker, especially social service provider, because this is a context in
when their kindness is appreciated. which our phenomenon is likely transparently
Self-compassion, as a private act, does not afford observable (Eisenhardt, 1989). Burnout has been
the same opportunities for recognition. We suggest, well documented in social service employees (Mas-
however, that it can replenish self-esteem, not by dem- lach & Leiter, 2008) and the profession’s mission
onstrating worth to others, but by directly prompting attracts those prone to compassionate care (Schab-
re-evaluation of one’s self-worth. We know that self- ram & Maitlis, 2017). We collected data in three
compassion is associated with “more careful, thorough waves, each one year apart, to capture the experience
processing of unflattering self-relevant information” of burnout as it unfolds over time and to mitigate
(Neff, 2003a: 93) and a reduction in feelings of insecu- against potential common method bias from tran-
rity (Barnard & Curry, 2011). Self-compassion pro- sient sources (Podsakoff, MacKenzie, Lee, & Podsak-
motes an unconditional acceptance of oneself off, 2003). Every year, all employees were invited to
(Thompson & Waltz, 2008) after a broad spectrum complete an online survey, given three weeks and
of negative experiences, including failure (Breines two reminders to participate, and incentivized with
& Chen, 2012), difficult emotions (Sbarra, Smith, & the chance to win one of three $200 gift cards. To test
Mehl, 2012), and chronic pain (Sirois, Molnar, & our predictions, we collected measures of burnout at
Hirsch, 2015). Through honest self-acceptance, Times 1 and 3 and compassion at Time 2.
self-compassion also fuels growth (Zhang & Chen,
2016) and resilience capabilities (Neff & McGehee, Sample
2010). In short, self-compassion lends itself to
thoughts and processes that enhance self-esteem, We received 284 responses at Time 1 (83% organi-
which in turn should reduce inefficacy. zational response rate) and obtained 179 responses
from prior participants at Time 2 (63% Time 1
Hypothesis 4. (a) Self-compassion and (b) other-
response rate) and 130 at Time 3 (73% Time 2
compassion are negatively related to the inefficacy
dimension of burnout, which is (c) mediated by self- response rate). After listwise deletion of those who
esteem. had not participated at all three points in time, we
retained the sample of 130 respondents. Partici-
pants’ average age was 40.75 (SD 5 11.17), average
OVERVIEW OF STUDIES tenure was 10.02 years (SD 5 7.17); 86% were
Our research follows a generalization and exten- female, and 32% were unionized. We compared all
sion approach (Tsang & Kwan, 1999) as we test our demographics to non-respondents at Time 1 and
model (see Figure 1) across two distinct samples found that respondents were less likely to be union-
(social service providers vs. business school stu- ized than non-respondents (32% vs. 45% union-
dents), experiences of burnout (chronic vs. acute), ized), x2(1, N 5 442) 5 6.75, p , .01.
and methodologies. In Study 1, we test our temporal
predictions—that each burnout dimension will per- Measures
sist over time, but can be reduced through acts of
A list of all measures and items can be found in
self- and other-compassion—via a three-year field
Appendix A.
survey. For Study 2, we design a within-person ran-
Burnout. We collected the Maslach burnout inven-
domized compassion intervention and apply an
tory general survey (MBI-GS; Schaufeli, Leiter, Mas-
experiential sampling technique to test the mediat-
lach, & Jackson, 1996), which includes five items for
ing effects of self-control, belonging, and self-esteem
exhaustion, five items for cynicism, and six reverse-
on the relationship between compassion and subse-
coded items for inefficacy. Participants rated items on
quent burnout. In accordance with the multidimen-
a 5-point Likert-type scale (1 5 strongly disagree, 5 5
sional nature of burnout (Leiter & Maslach, 2016),
strongly agree). All measures exhibited strong internal
we include all three dimensions in both our studies’
consistency: exhaustion (Time 1, a 5 .91; Time 3, a 5
path analyses.
.86), cynicism (Time 1, a 5 .85; Time 3, a 5 .91), and
inefficacy (Time 1, a 5 .77; Time 3, a 5 .74).
STUDY 1
Enacted other-compassion. We included three of
In Study 1, we sought to examine our main effects the four kindness subscale items from Pommier’s
hypotheses (Hypotheses 1a–c, 2a, 3a, 4a, and 4b). To (2010) compassion inventory, asking employees to
indicate their answer to the question “To what the burnout dimensions (exhaustion, cynicism, ineffi-
extent did you try to engage in the following behav- cacy) at Time 1 and Time 3 and self- and other-
iors?” with reference to a 5-point scale (1 5 never, compassion at Time 2. Given the large number of
5 5 all the time). Given our emphasis on action, we items, we utilized random assignment (Little, Cun-
dropped one item from the original measure (“My ningham, Shahar, & Widaman, 2002) to create two to
heart goes out to employees who are unhappy”); this three item parcels per construct (Williams & O’Boyle,
did not alter the significance of our results, a 5 .84. 2008) in order to reduce the sample-size-to-parameter
Enacted self-compassion. We captured self- ratio and yield more stable latent estimates. Our
compassion using the 5-item kindness subscale from hypothesized eight-factor model showed acceptable
Neff’s (2003b) self-compassion inventory, using fit to the data, x2 5 147.22 (comparative fit index
the same prompt and scale as for other-compassion [CFI] 5 .97, root-mean-square error of approximation
(a 5 .84). [RMSEA] 5 .05, standardized root-mean-square resid-
Controls. We controlled for participant gender ual [SRMR] 5 .05), and all factor loadings were signif-
(“0” 5 female, “1” 5 male), age, tenure, and union icant, p , .01. The hypothesized model fit the data
status (“0” 5 non-unionized, “1” 5 unionized), as significantly better than more parsimonious seven-,
each has been shown to influence the likelihood of six-, five-, and four-factor models (all Dx2 , .01).
burnout and compassionate action (Davis, 2011; Yar- Given our use of same-source data, we also followed
nell, Stafford, Neff, Reilly, Knox, & Mullarkey, 2015). Podsakoff and colleagues’ (2003) precedent and
conducted Harman’s single-factor test to evaluate
Analysis Strategy whether substantial variance in the data would be
accounted for by a single factor. Fit indices suggested
We tested our hypotheses in Mplus 8.2 (Muthen & that a single-factor solution fit the data poorly,
Muth en, 2017), grand-mean centering all our predictor x2 5 852.08 (CFI 5 .37, RMSEA 5 .20, SRMR 5 .14),
variables (Hofmann & Gavin, 1998) because employ- and significantly worse than our hypothesized
ees were nested in 45 units. We utilized the “Type 5 eight-factor solution (Dx2 , .01). We therefore pro-
Complex” syntax to account for data nonindepend- ceeded with our proposed model.
ence (Schaubroeck, Shen, & Chong, 2017) by adjusting
for parameter estimate standard errors that were a
Results
result of the sampling design (Wu & Kwok, 2012).
Prior to hypothesis testing, we conducted a con- Study 1’s descriptive statistics and correlations
firmatory factor analysis (CFA) to assess the fit of appear in Table 1. Multilevel path analysis results
our eight-factor measurement model, including are given in Table 2.
TABLE 1
Study 1: Means, Standard Deviations, and Correlations of the Focal Variables
Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12
1 Exhaustion (T1) 2.85 0.88 (.91)

2 Cynicism (T1) 2.00 0.82 .60 (.85)
3 Inefficacy (T1) 1.71 0.42 .17† .37 (.77)
4 Other-compassion (T2) 3.98 0.68 2.18 2.23 2.25 (.84)
5 Self-compassion (T2) 3.07 0.68 2.02 .01 2.19 .18 (.84)
6 Exhaustion (T3) 3.12 0.83 .49 .37 .21 2.16† 2.21 (.86)
7 Cynicism (T3) 2.25 0.98 .26 .52 .34 2.28 2.06 .54 (.91)
8 Inefficacy (T3) 1.82 0.49 .17† .33 .53 2.21 2.18 .37 .48 (.74)
9 Age 40.75 11.17 .06 .08 2.05 .11 .10 2.18 2.06 2.12 (–)
10 Gender 0.14 0.35 2.05 2.01 2.08 .03 .02 2.03 .04 2.20 .25 (–)
11 Tenure 10.02 7.17 .15† .19 2.07 .04 .10 2.12 .04 2.09 .53 .24 (–)
12 Union 0.32 0.47 .37 .36 .01 2.03 .07 .18 .17 .07 .19 2.03 .34 (–)
Notes: N 5 130. Gender: 0, female, 1, male. Union: 0, non-unionized; 1, unionized. “T” denotes time (i.e., Time 1, Time 2, Time 3).
Coefficient alphas are reported in parentheses along the diagonal.
†
p , .10
p , .05
p , .01
TABLE 2 inefficacy at Time 3, b 5 2.06, SE 5 0.05, ns; nor was

Study 1: Path Model Results other-compassion at Time 2 related to inefficacy at
Exhaustion Cynicism Inefficacy Time 3, b 5 2.05, SE 5 0.06, ns. As such, neither
(T3) (T3) (T3) Hypotheses 4a nor 4b were supported.
Variables b SE b SE b SE
Discussion
Exhaustion (T1) .47 0.07
Cynicism (T1) .48 0.07 Study 1 provided support for our baseline predic-
Inefficacy (T1) .50 0.09 tion that past burnout will be positively related to
Other-compassion (T2) 2.02 0.09 2.24 0.11 2.05 0.06 future burnout for each of the three dimensions;
Self-compassion (T2) 2.21 0.09 2.03 0.07 2.06 0.05
Age 2.01 0.01 2.01 0.01 2.003 0.004
according to COR (Hobfoll et al., 2018), this highlights
Gender .17 0.17 .24 0.19 2.20 0.11 the necessity for resource-generating interventions. In
Tenure 2.02 0.01 2.002 0.02 .001 0.01 line with the resource investment principle, we also
Union .16 0.14 .10 0.13 .08 0.05 found evidence that giving compassion constituted
such an effective intervention for two of the three
Notes: N 5 130. SE, standard error. Unstandardized coefficients
are reported. All predictors were grand-mean centered. Gender: 0, burnout dimensions. As expected, we found that self-
female, 1, male. Union: 0, non-unionized; 1, unionized. “T” but not other-compassion remedied exhaustion, while
represents the data wave. other- but not self-compassion remedied cynicism.
p , .05
p , .01
Surprisingly, neither form of compassion remedied
inefficacy. These results support our prediction that,
for compassion to matter, it must fit the idiosyncratic
experience of burnout—at least when it comes to two
To test Hypotheses 1a–c that past burnout is posi- of the three burnout dimensions.
tively related to future burnout, we regressed each Though these findings are promising, our study
burnout dimension at Time 3 on Time 1. We found a design was subject to a number of limitations. First,
positive relationship between Time 1 and Time 3 we did not collect data on our mediating mecha-
burnout across dimensions (see Table 2): exhaus- nisms (Hypotheses 2b, 3b, 4c). Second, the use of
tion, b 5 .47, SE 5 0.07, p , .001; cynicism, b 5 .48, surveys prevented us from pinpointing whether
SE 5 0.07, p , .001; inefficacy, b 5 .50, SE 5 0.09, p compassionate acts, as opposed to underlying stable
, .001. Thus, Hypotheses 1a–c were supported. traits that motivate such behavior, drove our effects.
Hypothesis 2a predicted that self-compassion Third, we sampled employees particularly inclined
would be negatively related to exhaustion. Control- toward compassion and burnout, which limits our
ling for past exhaustion (i.e., at Time 1), we found generalizability. Finally, Study 1 took the more con-
that self-compassion at Time 2 was negatively ventional approach of studying chronic burnout (Mas-
related to exhaustion at Time 3, b 5 2.21, SE 5 0.09, lach et al., 2001; Schaufeli et al., 2009). More recently,
p , .05, thus supporting Hypothesis 2a. Though not however, scholars have called for examinations of
hypothesized, we also examined and found support daily variances in acute burnout (Xanthopoulou &
for our assumption that other-compassion at Time 2 Meier, 2014) and short-term recovery (Sonnentag et al.,
was not related to exhaustion at Time 3, b 5 2.02, 2017), neither of which our results can speak to. Given
SE 5 0.09, ns. these considerations, we designed Study 2 to extend
Hypothesis 3a predicted that other-compassion our insights.
would be negatively related to cynicism. Controlling
for cynicism at Time 1, we found that other-
STUDY 2
compassion at Time 2 was negatively related to cyni-
cism at Time 3, b 5 2.24, SE 5 0.11, p , .05, thus In Study 2, we sought to test our complete model
supporting Hypothesis 3a. We also examined and using a new sample, context, and methodology—
found support for our non-hypothesized assumption experiential sampling methodology (ESM)—called for
that self-compassion at Time 2 was not related to by both compassion (Lilius, 2012) and burnout (Xan-
cynicism at Time 3, b 5 2.03, SE 5 0.07, ns. thopoulou & Meier, 2014) scholars. We recruited
Hypotheses 4a and 4b predicted that self- and undergraduate business students, because they tend
other-compassion are both negatively related to inef- to offer less self- or other-compassion (Beaumont et al.,
ficacy. Controlling for Time 1 inefficacy, we found 2016). Further, we conducted the study during their
that self-compassion at Time 2 was not related to midterm period, in order to maximize the likelihood
of acute burnout (Cushman & West, 2006). Participant On average, students completed 9.48 days of sur-
consent, contact information, and demographics were veys (range, 3–10), constituting 948 full day-level data
collected via a baseline survey two weeks prior to the points (95% response rate). Since our hypothesized
study. Over 10 consecutive business days (Monday model involved the analysis of two consecutive days
through Friday), students were emailed survey links (i.e., Day t morning, Day t evening, Day t11 morning),
twice daily. The average start times for surveys were removing responses for which more than one day
8:47 a.m. and 8:16 p.m., respectively. To increase par- passed between responses (Mondays or when partici-
ticipation, students had to complete the baseline sur- pants skipped days), left 757 data points. We also
vey and at least 80% of daily surveys to obtain removed 55 responses because participants indicated
research credit, were eligible to win gift cards for that they did not follow that day’s instructions (e.g.,
100% completion, and were texted personalized because they did not go to school) or provided careless
reminders at 11:30 a.m./9:30 p.m. if they had not yet response as indicated by two independent raters. This
participated. resulted in a final sample of 702.
To increase variance in our independent variable,
every day, participants were directed to engage in a Measures
specific action adapted from past compassion inter-
ventions (see Appendix B for instructions; Breines & We adapted all Study 1 measures to a diary study
Chen, 2012; Leary et al., 2007). On self-compassion design (Gabriel et al., 2019) by including five or fewer
days, participants were asked to notice one episode items per scale, referencing the context of school in
during which they would experience difficulty and items, changing phrasing to present tense, and includ-
to treat themselves with kindness. On other-compas- ing “at this moment” in question prompts to empha-
sion days, this language pertained to others. Finally, size the immediate time frame (Fisher & To, 2012).
we also included neutral (i.e., no directed compas- A list of all measures and items can be found in
sion) days in which participants were asked to report Appendix A.
on what they ate (see Foulk, Lanaj, Tu, Erez, & Arch- Burnout. We measured burnout every morning
ambeau, 2018). Every morning, participants were using the three highest-loading items from each
assigned one of these three instructions using a con- dimension of the MBI-GS (Schaufeli et al., 1996).
strained random matrix (Foulk et al., 2018), such Measures were rated on a 5-point scale (1 5 strongly
that the order was random both within and across disagree, 5 5 strongly agree) and had strong internal
participants; we ensured that participants were consistency: aexhaustion 5 .88 (Day t) and .87 (Day
given each direction three times, and that a third of t11); acynicism 5 .79 (Day t) and .79 (Day t11); and
participants were assigned each on any given day. ainefficacy 5 .74 (Day t) and .75 (Day t11).
We manipulated daily instructions for the purpose Directed compassion. To account for our daily
of creating greater variance in compassion (our inde- instructions, we created two dummy variables:
pendent variable) than we might have seen had we directed other-compassion (coded “1” for other-
passively sampled participants’ experiences (for compassion, “0” for self-compassion, and “0” for
another example of this approach, see Quinn, Myers, neutral) and directed self-compassion (“1” 5 self-
Kopelman, & Simmons, 2021). compassion, “0” 5 other-compassion, “0” 5 neutral).
Enacted compassion. In line with our emphasis
on action, to measure whether participants had actu-
Sample
ally acted compassionately, in the evening survey,
We recruited 100 undergraduate students from a we asked participants to complete the three highest-
business school in the Pacific Northwest. Participants’ loading items from Pommier’s (2010) other-kindness
average age was 21.4 (SD 5 3.22); 69% identified as and Neff’s (2003b) self-kindness subscales. Items
female; 48% identified as White, 5% identified as His- were rated on a 5-point Likert-type scale (1 5 strongly
panic, and 39% identified as Asian/Pacific Islander. disagree, 5 5 strongly agree); aother 5 .84, aself 5 .83.
Only 1% of the sample worked full-time, while 56% As an additional check, participants also described
worked part-time and 39% were unemployed, con- their action via an open-ended question (see Appen-
firming that being a student was our participants’ pri- dix B for sample responses).
mary occupation and therefore a relevant context in Self-control. Self-control was measured using
which to consider burnout. As all students who Lanaj and colleagues’ (2014) 5-item scale on a 5-point
signed up for the study provided at least three full scale (1 5 strongly disagree, 5 5 strongly agree;
days of data, we retained the entire sample. a 5 .90). Self-control was collected in the morning
survey because it is related to sleep quality (Lanaj Shi, 2013): we specified direct pathways from directed
et al., 2014). compassion (Day t morning) to enacted compassion
Belonging. Belonging was measured using three (Day t evening) to the three resource mediators (Day t
items from Zadro, Williams, and Richardson’s (2004) evening for belonging and self-esteem; Day t11 morn-
measure each evening with a 5-point scale (1 5 very ing for self-control) to the three burnout dimensions
slightly or not at all, 5 5 extremely; a 5 .77). (Day t11 morning). For our hypotheses tests, we also
Self-esteem. Self-esteem was measured using the included direct and indirect pathways from directed
four highest-loading items from the Rosenberg (1965) and enacted compassion to the three burnout dimen-
self-esteem scale each evening with a 5-point scale sions. We included controls from Day t (day and burn-
(1 5 strongly disagree, 5 5 strongly agree; a 5 .83). out) and Day t11 (sleep quality). Our within-level
Controls. As is common practice in ESM studies, variables (compassion predictors, resources media-
we controlled for day to account for time trends tors, burnout outcomes) were modeled at level 1 using
(Lanaj et al., 2016; Sonnentag et al., 2017). To isolate random slopes. The random effects of proposed medi-
daily burnout fluctuations, we controlled for prior ations were allowed to covary (Bauer, Preacher, & Gil,
day (Day t) burnout. To account for participants’ 2006). Our control variables were modeled as fixed
degree of rest that day, we controlled for sleep quality slopes, with day, prior-day burnout, and sleep quality
(Day t11) using Scott and Judge’s (2006) four-item modeled at level 1 (Wang et al., 2013). We group-mean
measure on a 5-point scale (1 5 very slightly or not at centered level 1 predictors (Hofmann, Griffin, & Gavin,
all, 5 5 very much; a 5 .78). 2000). We followed Preacher, Zyphur, and Zhang’s
(2010) recommendations for testing mediation in mul-
Analysis Strategy tilevel models to account for the asymmetric sam-
pling distribution of indirect effects. This procedure
Since daily experiences are nested within the per- involves using a parametric bootstrap to estimate
son in ESM studies, we first partitioned the variance and evaluate the significance of our indirect effects
of each focal variable into within- and between- (Selig & Preacher, 2008) and a Monte Carlo simula-
person components using Mplus 8.2 (Muthen & tion with 20,000 replications to build confidence
Muth en, 2017). We noted substantial within-person intervals around these estimated indirect effects.
variance (exhaustion 5 18.2%, cynicism 5 23%, To assess the fit of our measurement model, we con-
inefficacy 5 24.6%, enacted other-compassion 5 ducted a multilevel CFA with the eight latent varia-
62.9%, enacted self-compassion 5 68%, self- bles at the within-person level. As in Study 1, we
control 5 17.4%, belonging 5 24.7%, self-esteem 5 utilized random assignment to create item parcels (Lit-
28.6%). Thus, we chose to proceed with multilevel tle et al., 2002; Williams & O’Boyle, 2008), but only for
path analysis. self-control and self-esteem, as these variables had
Next, given our emphasis on action, we sought to more than three items each. Our eight-factor model
confirm that self-ratings of enacted compassion showed acceptable fit, x2 5 379.09 (CFI 5 .96, RMSEA
indeed captured behavior (as opposed to just inten- 5 .04, SRMR 5 .05); all factor loadings were signifi-
tion). To do so, two independent coders, blind to the cant, p , .01, and this fit the data significantly better
study, independently rated 400 of the written descrip- than more parsimonious seven-, six-, and five-factor
tions (interrater reliability: mean rself-compassion 5 .90, models (all Dx2 , .01). Given our same-source data,
p , .01; mean rother-compassion 5 .85, p , .01). These we again conducted Harman’s single-factor test, with
coder-ratings were correlated with self-ratings for both fit indices suggesting that a single-factor solution
forms of compassion (rself-compassion 5 .19, p , .001; fit the data poorly, x2 5 2665.52 (CFI 5 .47,
rother-compassion 5 .23, p , .001), but not correlated for RMSEA 5 .13, SRMR 5 .14), and significantly
opposing compassion ratings (rcoder self-compassion/self worse than our hypothesized eight-factor solution
other-compassion 5 .06, p . .05; rcoder other-compassion/self self- (Dx2 , .01).
compassion 5 2.05, ns).
We proceeded with a single model4 and lagged anal-
Results
ysis (see Wang, Liu, Liao, Gong, Kammeyer-Mueller, &
Study 2’s descriptive statistics and correlations
4
For completeness, this model includes directed and appear in Table 3, multilevel path analysis results in
enacted compassion. Alternate models with only directed Table 4, and indirect effects for mediation in Table 5.
compassion and only enacted compassion can be found in We first examined Hypotheses 1a–c, predicting
an online supplement. All raw data is also made available. that each burnout dimension would be positively
TABLE 3
Study 2: Means, Standard Deviations, and Correlations of the Focal Variables
Variables M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 Directed OC (t) 0.31 0.46 (—)

2 Directed SC (t) 0.32 0.47 2.47 (—)
3 Enacted OC (t) 3.66 0.68 .27 2.12 (.84)
4 Enacted SC (t) 3.54 0.72 2.01 .09 .37 (.83)
5 Exhaustion (t11) 3.21 1.00 .00 .02 .00 2.25 (.87)
6 Cynicism (t11) 2.74 0.87 .01 2.00 2.06 2.16 .56 (.79)
7 Inefficacy (t11) 2.54 0.74 .00 .00 2.25 2.32 .45 .57 (.75)
8 Self-control (t11) 3.06 0.89 2.02 .01 .01 .22 2.82 2.63 2.49 (.90)
9 Belonging (t) 3.45 0.84 .02 2.04 .33 .28 2.37 2.33 2.50 .34 (.77)
10 Self-esteem (t) 3.77 0.73 2.03 .03 .20 .47 2.48 2.47 2.64 .51 .47 (.83)
Controls
11 Exhaustion (t) 3.21 1.00 .01 .01 .00 2.21 .83 .51 .40 2.76 2.36 2.45 (.88)
12 Cynicism (t) 2.75 0.87 .06 2.04 2.06 2.17 .52 .80 .52 2.61 2.31 2.47 .56 (.79)
13 Inefficacy (t) 2.56 0.73 .01 .02 2.21 2.27 .45 .54 .77 2.49 2.49 2.64 .46 .58 (.74)
14 Day (t) 4.84 2.73 2.03 .00 .03 .07† .04 .05 2.01 2.05 .02 2.04 .04 .06 .01 (—)
15 Sleep quality 4.03 0.90 2.01 .05 2.06 .11 2.37 2.25 2.18 .42 .06 .26 2.32 2.26 2.16 2.02 (.78)
(t11)
Notes: N(level 1) 5 702; N(level 2) 5 100. OC, other compassion; SC, self-compassion. Directed OC: 1, other-compassion; 0,
self-compassion, 0, neutral. Directed SC: 1, self-compassion; 0, other-compassion; 0, neutral. Correlations, means, and standard deviations
for level 1 variables represent relationships among the daily variables at the within-individual level of analysis. Coefficient alphas
calculated at level 1 are reported in parentheses along the diagonal.
†
p , .10
p , .05
p , .01
related to future burnout, by analyzing the effect of support for Hypothesis 2b and partial support for
prior day on next day burnout on neutral days only Hypothesis 2a, since self-compassion was indirectly
(i.e., those without directed compassion). We found related to exhaustion through self-control, but not
that exhaustion was positively related to next day directly. As supplemental analysis, we tested and
exhaustion, b 5 .22, SE 5 0.08, p , .01, and cynicism found no significant direct or indirect effects of
to next day cynicism, b 5 .23, SE 5 0.09, p , .05, but other-compassion on exhaustion, as reported in
inefficacy was not positively related to next day inef- Tables 4 and 5, respectively.
ficacy, b 5 .09, SE 5 0.11, ns. This offered support Hypothesis 3 predicted that (a) other-compassion
for Hypotheses 1a and 1b but not 1c. would be negatively related to cynicism through (b)
Hypothesis 2 predicted that (a) self-compassion belonging. We found no direct effect of directed
would be negatively related to exhaustion through other-compassion, b 5 2.02, SE 5 0.03, ns, or enacted
(b) self-control. We found no direct effects of directed other-compassion, b 5 .03, SE 5 0.04, ns, on next
self-compassion, b 5 .05, SE 5 0.04, ns, or enacted day cynicism. We did, however, find direct effects
self-compassion, b 5 2.02, SE 5 0.04, ns, on next day of directed other-compassion on enacted other-
exhaustion. However, we did find a direct effect of compassion, b 5 .39, SE 5 0.05, p , .001, of enacted
directed self-compassion on enacted self-compassion, other-compassion on belonging, b 5 .09 SE 5 0.05,
b 5 .15, SE 5 0.05, p , .01, of enacted self-compassion p , .05, and of belonging on cynicism, b 5 2.09,
on self-control, b 5 .05, SE 5 0.03, , .05, and of self- SE 5 0.04, p , .05. Further, the indirect effect of
control on exhaustion, b 5 2.62, SE 5 0.07, p , .001. enacted other-compassion (indirect effect 5 2.01,
Further, the indirect effect of enacted self-compassion 95% CI [2.03, 2.00]) and serial indirect effect of
on exhaustion excluded zero (enacted self-compassion directed other-compassion (indirect effect 5 2.003,
! self-control ! exhaustion 5 2.03, 95% CI [2.07, 95% CI [2.01, 2.00]) excluded zero. Therefore, we
2.00]) as did the serial indirect effect of directed found complete support for Hypothesis 3b and partial
self-compassion (directed self-compassion ! enacted support for Hypothesis 3a, since other-compassion
self-compassion ! self-control ! exhaustion 5 2.01, was indirectly related to cynicism through belonging,
95% CI [2.01, 2.00]). These results provide complete but not directly. As supplemental analysis, we tested
TABLE 4
Study 2: Path Model Results
Other-compassion Self-compassion Self-control Belonging Self-esteem Exhaustion Cynicism Inefficacy
(t) (t) (t11) (t) (t) (t11) (t11) (t11)
Variables b SE b SE b SE b SE b SE b SE b SE b SE
Directed OC (t) .39 0.05 2.01 0.06 2.02 0.03 2.00 0.03
Directed SC (t) .15 0.05 .05 0.04 .02 0.04 .03 0.03
Enacted OC (t) 2.03 0.04 .09 0.05 .08 0.04 .03 0.11 .03 0.04 2.04 0.04
Enacted SC (t) .05 0.03 .05 0.04 .15 0.04 2.02 0.04 .01 0.03 2.00 0.03
Self-control (t11) 2.62 0.07 2.39 0.07 2.20 0.05
Belonging (t) 2.04 0.04 2.09 0.04 2.10 0.04
Self-esteem (t) 2.11 0.06 2.16 0.06 2.19 0.05
Exhaustion (t) 2.00 0.06
Cynicism (t) .02 0.05
Inefficacy (t) 2.06 0.07
Day (t) .00 0.01 .01 0.01 2.01 0.01
Sleep quality (t11) 2.03 0.02 .01 0.03 2.03 0.03
Notes: N(level 1) 5 702; N(level 2) 5 100. SE, standard error. OC, other compassion; SC, self-compassion. Directed OC: 1,
other-compassion; 0, self-compassion, 0, neutral. Directed SC: 1, self-compassion; 0, other-compassion; 0, neutral. Unstandardized
coefficients are reported. Predictors were group-mean centered.
p , .05
p , .01
and found no significant direct or indirect effects of complete support for Hypothesis 4c and partial sup-
self-compassion on cynicism, as reported in Tables 4 port for Hypotheses 4a and 4b, since self-compassion
and 5, respectively. and other-compassion were indirectly related to inef-
Hypothesis 4 predicted that (a) self-compassion ficacy through self-esteem, but not directly.
and (b) other-compassion are negatively related to
inefficacy through (c) self-esteem. We found no direct Supplemental Analysis
effect of directed self-compassion, b 5 .03, SE 5 0.03, Given the consistent support for our hypothesized
ns, or directed other-compassion, b 5 2.00, SE 5 0.03, indirect effects through the three resources, we con-
ns, nor of enacted self-compassion, b 5 2.00, SE 5 ducted a post hoc examination of potential spillover
0.03, ns, or enacted other-compassion, b 5 2.04, SE 5 paths. This would clarify whether self-control,
0.04, ns, on next day inefficacy. However, we did find belonging, and self-esteem can reduce only one
direct effects of directed self-compassion on burnout dimension each, as hypothesized, or if their
enacted self-compassion, b 5 .15, SE 5 0.05, p , benefits extend further. Including these spillover
.01, and of directed other-compassion on enacted effects did not change the significance of any of our
other-compassion, b 5 .39, SE 5 0.05, p , .001. hypothesized paths, and, for this reason, we present
Further, the relationships between enacted self- the complete and robust model in Table 4. This sup-
compassion and self-esteem, b 5 .15, SE 5 0.04, p , plemental analysis identified a number of significant
.001, as well as enacted other-compassion and additional mediation paths, which we report in
self-esteem, b 5 .08, SE 5 0.04, p , .05, were both Table 5. Self-esteem mediated the relationships
significant, as was the relationship between self- between self- and other-compassion and exhaustion,
esteem and inefficacy, b 5 2.19, SE 5 0.05, p , as well as between self- and other-compassion and
.001. Finally, both the indirect effects of enacted cynicism. Moreover, self-control mediated the rela-
compassion (self-compassion: indirect effect 5 2.03, tionship between self-compassion and inefficacy,
95% CI [2.05, 2.00]; other-compassion: indirect while belonging mediated the relationship between
effect 5 2.02, 95% CI [2.04, 2.00]) and the serial other-compassion and inefficacy.
indirect effects of directed compassion (self-compas-
Discussion
sion: indirect effect 5 2.004, 95% CI [2.01, 2.00];
other-compassion: indirect effect 5 2.01, 95% CI To extend our Study 1 insights from the field, in
[2.01, 2.00]) excluded zero. Therefore, we find Study 2, we adopted a new methodology, a new time
TABLE 5
Study 2: Indirect Effects for Mediation
95% CI
Indirect effect [LL, UL]
Criterion: Exhaustion
Enacted SC ! self-control ! exhaustion 2.03 [2.07, 2.00]
Directed SC ! Enacted SC ! self-control ! exhaustion 2.01 [2.01, 2.00]
Supplemental analyses of null predictions (OC paths)
Enacted OC ! self-control ! exhaustion .02 [2.01, .05]
Directed OC ! Enacted OC ! self-control ! exhaustion .01 [2.01, .03]
Criterion: Cynicism
Enacted OC ! belonging ! cynicism 2.01 [2.03, 2.00]
Directed OC ! Enacted OC ! belonging ! cynicism 2.003 [2.01, 2.00]
Supplemental analyses of null predictions (SC paths)
Enacted SC ! belonging ! cynicism 2.004 [2.02, .00]
Directed SC ! Enacted SC ! belonging ! cynicism 2.001 [2.00, .00]
Criterion: Inefficacy
Enacted OC ! self-esteem ! inefficacy 2.02 [2.04, 2.00]
Directed OC ! Enacted OC ! self-esteem ! inefficacy 2.01 [2.01, 2.00]
Enacted SC ! self-esteem! inefficacy 2.03 [2.05, 2.00]
Directed SC ! Enacted SC ! self-esteem ! inefficacy 2.004 [2.01, 2.00]
Supplemental cross-path analyses
Enacted SC ! self-esteem ! exhaustion 2.02 [2.04, 2.00]
Directed SC ! Enacted SC ! self-esteem ! exhaustion 2.002 [2.01, 2.00]
Enacted OC ! self-esteem ! exhaustion 2.01 [2.03, 2.00]
Directed OC ! Enacted OC ! self-esteem ! exhaustion 2.004 [2.01, 2.00]
Enacted OC ! self-esteem ! cynicism 2.01 [2.03, 2.00]
Directed OC ! Enacted OC ! self-esteem ! cynicism 2.01 [2.01, 2.00]
Enacted SC ! self-esteem ! cynicism 2.02 [2.05, 2.01]
Directed SC ! Enacted SC ! self-esteem ! cynicism 2.003 [2.01, 2.00]
Enacted SC ! self-control ! inefficacy 2.01 [2.02, 2.00]
Directed SC ! Enacted SC ! self-control ! inefficacy 2.001 [2.00, 2.00]
Enacted OC ! belonging ! inefficacy 2.01 [2.03, 2.00]
Directed OC ! Enacted OC ! belonging ! inefficacy 2.004 [2.01, 2.00]
Notes: Supplemental analyses are denoted in italics. OC, other compassion; SC, self-compassion. Directed OC: 1, other-compassion; 0,
self-compassion, 0, neutral. Directed SC: 1, self-compassion; 0, other-compassion; 0, neutral. LL, lower limit; UL, upper limit.
frame (acute, daily burnout over two weeks), and a either undercutting more stable inclinations cap-
sample less inclined toward either form of compas- tured in the Study 1 sample or creating discomfort
sion (Beaumont et al., 2016). Two initial distinctions in a Study 2 sample disinclined toward compas-
from Study 1 stand out. First, we found that, in this sionate gestures.
design, in the absence of an intervention, cynicism Study 2 does provide consistent evidence for the
and exhaustion persisted, but inefficacy fluctuated application of COR’s resource investment principle
over the 10 days. This suggests that the depleting as all our mediating hypotheses were supported: acts
experiences of cynicism and exhaustion may be more of self-compassion generated self-control and thereby
predictable across time frames, but also suggests that reduced exhaustion, acts of other-compassion gener-
individuals must not find themselves in a persistent ated belonging and thereby reduced cynicism, and
state of burnout to reap the benefits of compassion. both forms of compassion raised self-esteem and
Second, we found no direct effects for any form of through it reduced inefficacy. Moreover, in probing
compassion on next day burnout. This could be for spillover effects, we identified additional surpris-
attributable to the shorter time frame, though this ing restorative paths: other- and self-compassion
would be counterintuitive as it should have made also reduced exhaustion and cynicism via self-
detecting direct effects more likely. Instead, we esteem, while other-compassion reduced inefficacy
speculate that this might be the case because we via belonging and self-compassion reduced ineffi-
directed compassion via our instructions, thus cacy via self-control.
GENERAL DISCUSSION directed via instruction). This research strategy

afforded us the confidence to generalize from consis-
This research was inspired by two contemporary,
tent results across our “imprecise replication” (Tsang
organizational phenomena: the recognition of
& Kwan, 1999: 768), while remaining cautiously aware
employee burnout (Gallup, 2018; World Health
of the importance of context, and seeding future
Organization, 2019) and growing academic (Rynes
research directions. Our findings principally advance
et al., 2012; Tsui, 2013) and practitioner (Tan, 2012)
the study of burnout and compassion, but our design
interest in bringing compassion into the workplace.
also provides potentially important insights to the the-
Adopting the multidimensional view of burnout, we
oretical COR lens.
applied COR’s resource allocation principles to pre-
dict how actors could act compassionately toward
themselves or others and thereby generate resources Contributions
to remedy their own burnout. Burnout. Though scholars have long emphasized
Across two studies, we found that the exhaustion the need to study burnout solutions (Maslach, 2017;
and cynicism dimensions of burnout persisted over Maslach & Leiter, 2008), researchers remain focused
time, and that actors experiencing each could inter- primarily on antecedents (Schaufeli et al., 2009). In
vene via self- and other-compassion respectively, part, this may be attributed to a move toward reduc-
thereby impacting them directly in our first study tive models of burnout (Baer et al., 2015; Grant et al.,
and indirectly in our second. We also noted that, 2014; Lin et al., 2019), which make it difficult to
under the conditions of our second study, both forms identify holistic remedies, since “effective interven-
of compassion were negatively and indirectly tions to deal with burnout should be framed in terms
related to subsequent inefficacy, even when ratings of these three dimensions” (Maslach, 2003: 190).
of inefficacy fluctuated by day rather than persisting. In our work, we consistently probed for all three
Taken together, these results indicate that acting dimensions of burnout and thereby identified impor-
compassionately can be a way to pull oneself out of tant effects that would have been missed by an
burnout, but only when the type of compassion exhaustion-only approach. We uncovered persistent
offered fits the idiosyncratic experience of burnout. cynicism among members of two distinct populations,
We also found support for all predictions about the thereby suggesting that cynicism remains as relevant
role of resources in this recovery: self-control medi- as when it was first identified as a widespread organi-
ated the relationship between self-compassion and zational phenomenon (Dean et al., 1998). More prom-
exhaustion, belonging between other-compassion isingly, we found that such cynicism could be readily
and cynicism, and self-esteem between both forms remedied, including via fairly minimal directions.
of compassion and inefficacy. Finally, our analysis Cynicism was once considered a promising area of
also revealed a number of spillover effects. In Study organizational research and described as crucial to
1, we noted that the correlations among burnout understanding how employees’ relationships to their
dimensions were stronger at Time 3 than Time 1, organization sour (Dean et al., 1998; Naus, van Iterson,
suggesting that, over time, either wear and tear has a & Roe, 2007). We hope that our insights encourage
greater impact on all three dimensions, or depletion more work on this empirically neglected construct
in one makes one more vulnerable in another. In (Leiter & Maslach, 2016).
Study 2, supplemental analysis identified spillover Furthermore, inclusion of all three burnout dimen-
pathways such that self-control and self-esteem, but sions served as a means to consider employees’
not belonging, could replenish all three dimensions. agency in addressing their own burnout. To date,
We note that these results emerged from a general- research into remedies has primarily considered how
ization and extension approach (Tsang & Kwan, others can aid the burnt out employee (Maslach,
1999) wherein we tested our predictions across two 2017). While some such external (and passive) solu-
complementary research procedures: samples tions to an internal problem have shown promise, not
drawn from two populations (older, social service all have been successful (Halbesleben, 2006). Our
employees inclined to embrace compassion; Bun- results indicate that employees can have important
derson & Thompson, 2009; Schabram & Maitlis, agency in their own recovery (Lilius, 2012). This sug-
2017; and younger, business students disinclined to; gests two potential explanations for the relationship
Beaumont et al., 2016); two time frames (a three-year between passive and agentic burnout remedies. On
field survey and 10-day diary study); as well as two the one hand, it may be that the two strategies are com-
manifestations of compassion (unprompted and plementary and depend on the depleted dimension(s).
While past research has shown how others can help depleted, giving support to others should sap them
recovery from exhaustion and inefficacy, we found further; such personal costs have also been identi-
that the actor can reduce their own cynicism. The fied in previously cited research on prosocial ges-
notion that we might be better suited than others to tures (Bolino & Grant, 2016; Lanaj et al., 2016; Uy
help ourselves recover from depletion in this most et al., 2017). Why, then, did other-compassion serve
interpersonal of burnout dimensions (Leiter & Mas- a singularly restorative function? As we noted in our
lach, 2017) is somewhat counterintuitive. It is, how- literature review, compassion is distinguished
ever, consistent with self-determination theory, a among the family of prosocial behaviors by its prin-
motivational theory related to COR, which has posited cipal attendance to human needs (Tsui, 2013) rather
that autonomy and relatedness needs interact (Halbe- than organizational effectiveness, and this may offer
sleben et al., 2014). It may be that, to reduce one’s an explanation. Perhaps, there is something funda-
sense of alienation, one must feel that one played a mentally more beneficial for actors about engaging
role in one’s own reintegration. Alternatively, there in acts of kindness and care (e.g., taking someone
may be interaction effects or a temporal order between who is having a hard time out for coffee) than in pro-
external and internal remedies not yet uncovered. Our viding instrumental support (e.g., exerting oneself to
results indicate that acting compassionately benefits provide a friendly review). We further note that our
the actor, but do not speak to whether the actor must study also did not find any evidence of “compassion
initiate or even intend such action. Indeed, our second fatigue” (Figley, 2013), identified frequently by prac-
study highlights that compassion worked when it was titioners among the social service employees that
directed via prompt. Thus, others may serve to create comprised our first sample. In line with the
the conditions needed for actors to help themselves or “desperation corollary” of COR (Hobfoll et al., 2018),
to prompt them to do so. We are reminded of the old which suggests that individuals can reach a state of
adage “give a man a fish and you feed him for a day. extreme depletion characterized by maladaptive cop-
Teach him how to fish and you feed him for a life- ing, it may be that there exists a tipping point after
time.” To sum, our insights lay the foundation for which compassion ceases to offer benefits. If there is,
incorporating agentic with external efforts to more however, it must be quite high to not have registered
fully understand how recovery unfolds. in either the longitudinal or the diary designs.
Compassion. Our findings should also speak to A second contribution emerged from our decision
scholars embracing compassion as a potential anti- to import the construct of self-compassion from psy-
dote against a lack of humanity in the modern work- chology, in line with directives to draw on compas-
place (Rynes et al., 2012; Tsui, 2013). To date, sion advancements from the wider social sciences
compassion research remains “dominated by theo- (Rynes et al., 2012). As in past psychological studies,
retical and theory-building studies, opening wide we found that both compassion expressions were
possibilities for theory-testing studies that address positively correlated (Crocker & Canevello, 2008;
both the process and the outcomes of compassion at Neff & Beretvas, 2013) and could be primed by the
work” (Dutton et al., 2014: 293). When scholars have same type of intervention (Gilbert & Irons, 2005).
tested the consequences of compassion, they have Despite such similarities, however, we note that
focused primarily on the benefits for those who their consequences are distinct: each form of com-
receive it. What little we know about benefits to passion had the potential to generate two of three
actors themselves has been framed vis- a-vis their resource categories and diminish two of three burn-
standing in the eyes of such recipients. For instance, out dimensions. This makes the point that we cannot
a series of studies by Melwani et al. (2012) demon- simply expect all forms of compassion to remedy all
strated that compassionate actors are judged as more depleting experiences. Instead, scholars must spec-
intelligent and better leaders. ify, and ideally incorporate into their theorizing and
Our work suggests a more immediate benefit; methodologies, the different compassion expres-
namely, that giving compassion can serve an impor- sions—an advancement that is not uncommon as
tant resource generative function for the self. Indeed, areas of study move out of their infancy (e.g., see
in neither of our studies did we find either compas- Fulmer & Gelfand, 2012, for a review of trust refer-
sion expression to ever have a deleterious effect. ents). Indeed, the broader study of prosocial gestures
While this is in line with the broader literature on may benefit from such an approach, as research has
self-compassion (Neff, 2011), it is somewhat surpris- contrasted giving and receiving (Grodal, Nelson, &
ing when it comes to other-compassion. Hobfoll Siino, 2015) but rarely considered the impact of giv-
(1989) speculated that, when people find themselves ing to different targets including the self.
Conservation of resources. Though we drew on substitution (i.e., turning one’s attention away from
COR principally to explain why compassion impacts one’s own pain and to caring for others). Thus, our
burnout, our research also touches upon important results would counter predictions for the primacy of
nuances in this theoretical lens worthy of consider- replacement strategies, as self- and other-compassion
ation. Just as we have emphasized the limitations of generated the same number of resources (two out of
studying burnout as an exhaustion-only phenome- three categories) and comparable effect sizes. More-
non, reviews of COR research (Halbesleben et al., over, they also contradict the notion that replacement
2014; Hobfoll et al., 2018) have noted that COR’s mul- is more likely when individuals are on a loss trajectory
tiple and dynamic principles are too frequently exam- (Hobfoll, 1989). In our second study, self-compassion
ined via static or overly minimalist approaches. To (replacement) and other-compassion (substitution)
date, studies have prioritized resource consumption were effective in remedying exhaustion and cynicism
over generation (Hobfoll et al., 2018), cross-sectional respectively, and both reduced inefficacy. While we
designs (Halbesleben et al., 2014), and the examination stress again that ours is not a direct test of replacement
of only one resource at a time (Parker, Johnson, Col- versus substitution, these results warrant further
lins, & Nguyen, 2013). By testing a more comprehen- inquiry. It may be that an individual’s idiosyncratic
sive model that included three resource categories, we state of depletion matters for strategy choice and
can offer proof of concept of various under-examined success.
COR theory dynamics (Hobfoll et al., 2018). Specifi-
cally, by demonstrating how actions can generate Limitations and Future Directions
self-esteem concurrently to self-control and self-
efficacy, we provide support for the recently intro- Our results should be considered in light of a num-
duced concept of “resource caravans” (Halbesleben ber of limitations. Foremost, our generalization and
et al., 2014; Hobfoll et al., 2018), defined as patterns of extension approach (Tsang & Kwan, 1999) makes it
resources that arise from the same developmental con- difficult to know whether our inconsistent findings
ditions. Moreover, our supplemental analysis revealed are attributable to the instability of our results or our
that self-control and self-esteem could replenish all design choices. For instance, as previously noted,
three burnout dimensions to varying degrees; this pro- we cannot say whether our inconsistent direct
vided evidence for the concepts of “resource equi- effects results should be attributed to our two differ-
finality,” defined as when multiple resources can ent time frames or voluntary versus directed com-
achieve the same goal, and “multifinality,” defined as passion. Future research should both seek to extend
when the same resource can achieve multiple goals. our consistent results and replicate our inconsistent
Taken together, this suggests the value in developing a ones, ideally by varying only one rather than multi-
taxonomy of previously identified recovery strategies ple elements (sample, timeframe, design).
(Sonnentag et al., 2017) based on whether they pro- We particularly encourage extension to addi-
duce complementary or substitutable resources. tional populations: both our studies largely comprised
In addition, our comparison of self- and other- women, who tend to engage in more other-
compassion also allows us to probe a foundational compassion but less self-compassion (L opez et al.,
question regarding the nature of resource-generating 2018; Neff, 2003b), and we are curious whether and
strategies. In his introductory piece on COR, Hobfoll how compassion might matter in male-dominated
(1989) suggested that individuals can cope with professions. Both samples also comprised partici-
resource loss either via direct replacement (e.g., pants of equal status (i.e., coworkers or fellow stu-
seeking ways to improve on a failed task) or substitu- dents). We know, however, that elevated power and
tion (e.g., shifting focus away from this failure to status tend to inhibit other-compassion (van Kleef,
other areas of competence), alternatively referred to Oveis, van der L€ owe, LuoKogan, Goetz, & Keltner,
as “direct replacement” versus “indirect invest- 2008) while low status can inhibit self-compassion
ment” (Hobfoll et al., 2018). Replacement was pos- (Vigna, Poehlmann-Tynan, & Koenig, 2018), making
ited as both a more common and more effective it difficult to know whether our findings would hold
strategy (Hobfoll, 1989), and, perhaps not surpris- in more hierarchical organizations or for leader–
ingly, has dominated research (Halbesleben et al., subordinate interactions.
2014; Hobfoll et al., 2018). Though not a direct test, Both studies also relied on self-reports, including,
we propose that self-compassion constitutes a in Study 2, where we found only indirect effects of
replacement strategy (i.e., self-care to alleviate burn- directed compassion through enacted compassion.
out), while other-compassion functions as a form of We particularly encourage replication via controlled
experiments and using objective or multisource fundamental assumption that the necessary starting
measures of compassion and burnout, though we point of compassion is suffering. Further, it would be
acknowledge that this will be more difficult for self- helpful to understand the unfolding effects of com-
than other-compassion. passion based on recipient requests or reactions (Dut-
Incorporating compassion recipients would also ton et al., 2014). Finally, as we noted, there may exist
shed light on the interpersonal dynamics of compas- a tipping point after which offering compassion may
sion and recovery. Our focus was very much on the become deleterious, which could be identified in
focal actor offering compassion, who happened to be another ESM study.
one and the same when it comes to self-compassion. Beyond such limitations, we would encourage
We took this targeted approach both because the more research into some of the complex elements
compassion literature has prioritized the impact on incorporated into this study but ultimately beyond
recipients and because we were specifically inter- the scope of our paper to fully consider. As we noted
ested in agentic efforts to recover. There remains a in our literature review, the interplay of the three
great deal of work to be done, however, to integrate burnout dimensions is not yet well understood. We
the various parties in a compassion episode. Do com- showed that compassion, and the resources it gener-
passionate acts need to be appreciated to have an ates, can matter for all three burnout dimensions,
effect, as we have speculated? Moreover, in line with but more work is needed on potential order effects
COR’s emphasis on dyadic resource crossover (Hob- (e.g., if exhaustion is often an early marker of burn-
foll et al., 2018), does other-compassion generate out, is early self-compassion more advantageous
resources for recipients such that both actor and recip- than other-compassion?) or considerations of com-
ient benefit? We are particularly interested in the role passion when employees are experiencing only
of leaders in promoting compassion and shaping the “partial burnout” (e.g., does other-compassion trump
compassion context as well as whether attachment self-compassion for a cynicism-only profile?). The
styles (Mikulincer & Shaver, 2005) and relational type of latent profile analysis employed to develop
schemas between the various parties influence how burnout profiles (Leiter & Maslach, 2016) might also
compassion is given or received. With others already lend itself to further explore the resource caravans
considering collective other-compassion (Dutton et al., identified via our supplemental analysis in Study 2
2006), it would be interesting to also study collective (e.g., what combination of resources would lend itself
self-compassion and the influence of such collective most to improving employee well-being?).
effects on recovery (Sonnentag et al., 2017). Consider- We conclude by highlighting that recent reviews
ing organizational compassionate culture and practi- have indicated that “the area of greatest need for test-
ces (Dutton et al., 2014) also opens the door to ing COR theory and for the health of organizations is
examine whether such compassion can function not the application of resource theory to interventions
just as a remedy but also as a prophylactic, as well as and clinical trials, both randomized and more natu-
what makes those already experiencing burnout ralistic” (Hobfoll et al., 2018: 120–121). We encour-
choose to act compassionately in the first place. age future field experiments as a form of co-creation
We took pains to consider time in our research of knowledge with participating organizations.
approach, seeking to generalize our findings across a
longitudinal and a shortitudinal design—and noting
Practical Implications
the surprise of direct effects only in the former. In
Study 2, however, the availability of a morning sur- Practitioners are increasingly interested in
vey led to our decision to measure self-control right humanizing the workplace through compassion ini-
after sleep, inadvertently time separating it from the tiatives (Women at Work, 2019). As Google’s
measures of belonging and self-control at night as Chade-Meng Tan (2012) proclaimed, “Compassion
well as testing Hypothesis 2 with the same time and is something that creates a vibrant, energetic com-
source mediator and dependent variable. Future munity. Compassion is good for business.” Our work
work should aim to avoid such time separation offers guidance to such employers.
unless it is theoretically justified. Careful temporal On the one hand, we highlight that compassion
designs could also serve to answer a number of ques- can function as one means to address the current
tions raised by this study, such as whether compas- burnout epidemic. Participants in both our studies
sion can function as a preventative strategy for those benefited from both compassion offered to others
not yet experiencing burnout. Exploring this possibil- and invested into the self. However, we also found
ity would shed light on the compassion literature’s that the calculus is not as straightforward, as all
compassion combats all burnout. Our results different recuperative functions, depending on the
depended on the right fit between the two compas- idiosyncratic experience of burnout. We hope that
sion expressions and the three burnout dimensions. this emphasis on fit and multifinality will inspire fur-
Moreover, when employing compassion instruc- ther research on care and recovery at work.
tions, we found only indirect effects. This suggests
that reaping the benefits of compassion in organiza- REFERENCES
tions (Tsui, 2013) is not an easy enterprise and likely
Alarcon, G. M. 2011. A meta-analysis of burnout with
requires ongoing investment of resources, particu- job demands, resources, and attitudes. Journal of
larly in assessing the specific needs of burnt out Vocational Behavior, 79: 549–562.
employees and matching appropriate recovery strate-
Baer, M. D., Dhensa-Kahlon, R. K., Colquitt, J. A., Rodell,
gies. We do hope that our work can offer some useful
J. B., Outlaw, R., & Long, D. M. 2015. Uneasy lies the
ideas toward those ends. Answering the calls to iden- head that bears the trust: The effects of feeling trusted
tify measurements of compassion (Dutton et al., on emotional exhaustion. Academy of Management
2014), we utilized validated, quantitative measures of Journal, 58: 1637–1657.
self- and other-compassion from psychology (Neff, Bakker, A. B., & Costa, P. L. 2014. Chronic job burnout and
2003b; Pommier, 2010), which employers could read- daily functioning: A theoretical analysis. Burnout
ily use to measure fluctuations in compassion. In Research, 1: 112–119.
addition, we adapted our compassion instructions Barnard, L. K., & Curry, J. F. 2011. Self-compassion: Con-
from prior research (Breines & Chen, 2012; Leary et al., ceptualizations, correlates, & interventions. Review of
2007) and our data (and lack of direct results) can General Psychology, 15: 289–303.
help practitioners further discern whether interven-
Bauer, D. J., Preacher, K. J., & Gil, K. M. 2006. Conceptual-
tions are a viable strategy. As noted in our future izing and testing random indirect effects and moder-
directions section, we do encourage future field ated mediation in multilevel models: New procedures
experiments through which managers and academics and recommendations. Psychological Methods, 11:
could co-create knowledge on burnout compassion. 142–163.
Finally, while our paper considers compassion as Baumeister, R. F. 2002. Ego depletion and self-control fail-
a remedy for burnout, we raise the question of ure: An energy model of the self’s executive function.
whether it could also function as a prophylactic. Self and Identity, 1: 129–136.
Here, we echo burnout scholars’ warnings that pre- Baumeister, R. F., Faber, J. E., & Wallace, H. M. 1999. Cop-
vention is better than waiting until burnout becomes ing and ego depletion. In C. R. Snyder (Ed.), Coping:
a problem; the best treatment is not just about The psychology of what works: 50–69. Oxford, U.K.:
fixing the person but also fixing the job that led to Oxford University Press.
such burnout (Maslach, 2017). To leave on an opti- Beaumont, E., Durkin, M., Martin, C. J. H., & Carson, J.
mistic note, however, we believe that fostering 2016. Compassion for others, self-compassion, quality
compassionate cultures may facilitate both (see of life and mental well-being measures and their asso-
Lilius, Worline, Dutton, Kanov, & Maitlis, 2011). ciation with compassion fatigue and burnout in stu-
Such cultures are likely particularly adept at identi- dent midwives: A quantitative survey. Midwifery, 34:
fying or eliminating the situational and structural 239–244.
drivers of burnout as well as promoting compassion Bergeron, D. M. 2007. The potential paradox of organiza-
to combat burnout. In short, while we positioned tional citizenship behavior: Good citizens at what cost?
compassion as an agentic remedy in this paper, we Academy of Management Review, 32: 1078–1095.
believe it will be most effective in a culture of com- Bergeron, D. M., Shipp, A. J., Rosen, B., & Furst, S. A. 2011.
passion that supports the individual rather than Organizational citizenship behavior and career out-
putting the onus on them to take better care of them- comes the cost of being a good citizen. Journal of
selves and others. Management, 39: 958–984.
Bolino, M. C., & Grant, A. M. 2016. The bright side of being
Conclusion prosocial at work, and the dark side, too: A review
and agenda for research on other-oriented motives,
We examined compassion offered to oneself or to behavior, and impact in organizations. Academy of
others as an agentic action by which employees might Management Annals, 10: 599–670.
remedy their own burnout. From our results emerged Breines, J. G., & Chen, S. 2012. Self-compassion increases
a complex view of the restorative potential of compas- self-improvement motivation. Personality and Social
sion, suggesting self- and other-compassion play Psychology Bulletin, 38: 1133–1143.
Bunderson, S. J., & Thompson, J. A. 2009. The call of the actor-centric approach to daily psychological power,
wild: Zookeepers, callings, and the double-edged abusive leader behavior, and perceived incivility.
sword of deeply meaningful work. Administrative Academy of Management Journal, 61: 661–684.
Science Quarterly, 54: 32–57. Fulmer, C. A., & Gelfand, M. J. 2012. At what level (and in
Buunk, B. P., & Schaufeli, W. B. 1993. Burnout: A perspec- whom) we trust: Trust across multiple organizational
tive from social comparison theory. In W. B. Schau- levels. Journal of Management, 38: 1167–1230.
feli, C. Maslach, & T. Marek (Eds.), Professional Gabriel, A. S., Podsakoff, N. P., Beal, D. J., Scott, B. A., Son-
burnout: Recent developments in theory and nentag, S., Trougakos, J. P., & Butts, M. M. 2019. Expe-
research: 53–69. Philadelphia, PA: Taylor & Francis. rience sampling methods: A discussion of critical
Canevello, A., & Crocker, J. 2011. Interpersonal goals, trends and considerations for scholarly advancement.
others’ regard for the self, and self-esteem: The para- Organizational Research Methods, 22: 969–1006.
doxical consequences of self-image and compassion-
Gaines, J., & Jermier, J. M. 1983. Emotional exhaustion in a
ate goals. European Journal of Social Psychology,
high stress organization. Academy of Management
41: 422–434.
Journal, 26: 567–586.
Cordes, C. L., & Dougherty, T. W. 1993. A review and an
Gallup. 2018. Employee burnout. Retrieved from https://
integration of research on job burnout. Academy of
www.gallup.com/workplace/237059/employee-
Management Review, 18: 621–656.
burnout-part-main-causes.aspx
Crocker, J., & Canevello, A. 2008. Creating and undermin-
Garton, E. 2017, April 6. Employee burnout is a problem
ing social support in communal relationships: The
with the company, not the person. Retrieved from
role of compassionate and self-image goals. Journal of
https://hbr.org/2017/04/employee-burnout-is-a-
Personality and Social Psychology, 95: 555–575.
problem-with-the-company-not-the-person
Cushman, S., & West, R. 2006. Precursors to college student
Gilbert, P., & Irons, C. 2005. Focused therapies and com-
burnout: Developing a typology of understanding.
passionate mind training for shame and self-attacking.
Qualitative Research Reports in Communication, 7:
In P. Gilbert (Ed.), Compassion: Conceptualizations,
23–31.
research, and use in psychotherapy: 263–325. New
Davis, R. S. 2011. Blue-collar public servants: How union York, NY: Routledge.
membership influences public service motivation.
Goetz, J. L., Keltner, D., & Simon-Thomas, E. 2010. Com-
American Review of Public Administration, 41:
705–723. passion: An evolutionary analysis and empirical
review. Psychological Bulletin, 136: 351–374.
Dean, J. W., Brandes, P., & Dharwadkar, R. 1998. Organiza-
tional cynicism. Academy of Management Review, Gonz o, J. M., Rodrıguez, I., & Bli-
alez-Morales, M. G., Peir
23: 341–352. ese, P. D. 2012. Perceived collective burnout: A multi-
level explanation of burnout. Anxiety, Stress, and
Dutton, J. E., Workman, K. M., & Hardin, A. E. 2014. Coping, 25: 43–61.
Compassion at work. Annual Review of Organi-
zational Psychology and Organizational Behav- Grant, A. M., Berg, J. M., & Cable, D. M. 2014. Job titles as
ior, 1: 277–304. identity badges: How self-reflective titles can reduce
emotional exhaustion. Academy of Management
Dutton, J. E., Worline, M. C., Frost, P. J., & Lilius, J. 2006. Journal, 57: 1201–1225.
Explaining compassion organizing. Administrative
Science Quarterly, 51: 59–96. Grant, A. M., Dutton, J. E., & Rosso, B. D. 2008. Giving com-
mitment: Employee support programs and the proso-
Eisenhardt, K. M. 1989. Building theories from case study cial sensemaking process. Academy of Management
research. Academy of Management Review, 14: Journal, 51: 898–918.
532–550.
Grodal, S., Nelson, A. J., & Siino, R. M. 2015. Help-seeking
Farh, J.-L., & Dobbins, G. H. 1989. Effects of self-esteem on and help-giving as an organizational routine: Contin-
leniency bias in self-reports of performance: A struc- ual engagement in innovative work. Academy of
tural equation model analysis. Personnel Psychology, Management Journal, 58: 136–168.
42: 835–850.
Halbesleben, J. R. 2006. Sources of social support and
Figley, C. R. 2013. Treating compassion fatigue. New burnout: A meta-analytic test of the conservation of
York, NY: Routledge. resources model. Journal of Applied Psychology, 91:
Fisher, C. D., & To, M. L. 2012. Using experience sampling 1134–1145.
methodology in organizational behavior. Journal of Halbesleben, J. R., Neveu, J.-P., Paustian-Underdahl, S. C.,
Organizational Behavior, 33: 865–877. & Westman, M. 2014. Getting to the “COR” under-
Foulk, T., Lanaj, K., Tu, M.-H., Erez, A., & Archambeau, L. standing the role of resources in conservation of
2018. Heavy is the head that wears the crown: An resources theory. Journal of Management, 40: 1–31.
Halbesleben, J. R., & Wheeler, A. R. 2015. To invest or not? treating oneself kindly. Journal of Personality and
The role of coworker support and trust in daily recip- Social Psychology, 92: 887–904.
rocal gain spirals of helping behavior. Journal of Man- Lee, R. T., & Ashforth, B. E. 1996. A meta-analytic exam-
agement, 41: 1628–1650. ination of the correlates of the three dimensions of
Hobfoll, S. E. 1989. Conservation of resources: A new job burnout. Journal of Applied Psychology, 81:
attempt at conceptualizing stress. American Psychol- 123–133.
ogist, 44: 513–524.
Leiter, M. P., & Maslach, C. 2016. Latent burnout profiles:
Hobfoll, S. E., Halbesleben, J., Neveu, J.-P., & Westman, M. A new approach to understanding the burnout experi-
2018. Conservation of resources in the organizational ence. Burnout Research, 3: 89–100.
context: The reality of resources and their consequen-
Leiter, M. P., & Maslach, C. 2017. Burnout and engage-
ces. Annual Review of Organizational Psychology
ment: Contributions to a new vision. Burnout
and Organizational Behavior, 5: 103–128.
Research, 5: 55–57.
Hofmann, D. A., & Gavin, M. B. 1998. Centering decisions
Lemay, E., & Clark, M. 2008. How the head liberates the
in hierarchical linear models: Implications for
heart. Journal of Personality and Social Psychology,
research in organizations. Journal of Management,
94: 647–671.
24: 623–641.
Lilius, J. M. 2012. Recovery at work: Understanding the
Hofmann, D. A., Griffin, M. A., & Gavin, M. B. 2000. The
restorative side of “depleting” client interactions.
application of hierarchical linear modeling to organi-
Academy of Management Review, 37: 569–588.
zational research. In K. J. Klein & S. W. J. Kozlowski
(Eds.), Multilevel theory, research, and methods in Lilius, J. M., Worline, M. C., Dutton, J. E., Kanov, J. M., &
organizations: Foundations, extensions, and new Maitlis, S. 2011. Understanding compassion capabil-
directions: 467–511. San Francisco, CA: Jossey-Bass. ity. Human Relations, 64: 873–899.
H€
ulsheger, U. R., Lang, J. W. B., Depenbrock, F., Fehr- Lin, S.-H. (J.), Scott, B. A., & Matta, F. K. 2019. The dark
mann, C., Zijlstra, F. R. H., & Alberts, H. J. E. M. 2014. side of transformational leader behaviors for leaders
The power of presence: The role of mindfulness at themselves: A conservation of resources perspective.
work for daily levels and change trajectories of psy- Academy of Management Journal, 62: 1556–1582.
chological detachment and sleep quality. Journal of Little, T. D., Cunningham, W. A., Shahar, G., & Widaman,
Applied Psychology, 99: 1113–1128. K. F. 2002. To parcel or not to parcel: Exploring the
Kanov, J. M., Maitlis, S., Worline, M. C., Dutton, J. E., question, weighing the merits. Structural Equation
Frost, P. J., & Lilius, J. M. 2004. Compassion in organi- Modeling, 9: 151–173.
zational life. American Behavioral Scientist, 47: L
opez, A., Sanderman, R., Ranchor, A. V., & Schroevers,
808–827.
M. J. 2018. Compassion for others and self-compas-
Kreemers, L. M., van Hooft, E. A. J., & van Vianen, A. E. M. sion: Levels, correlates, and relationship with psycho-
2018. Dealing with negative job search experiences: logical well-being. Mindfulness, 9: 325–331.
The beneficial role of self-compassion for job seekers’
Maslach, C. 2003. Job burnout: New directions in research
affective responses. Journal of Vocational Behavior,
and intervention. Current Directions in Psychologi-
106: 165–179.
cal Science, 12: 189–192.
Lambert, N. M., Stillman, T. F., Hicks, J. A., Kamble, S.,
Maslach, C. 2017. Finding solutions to the problem of
Baumeister, R. F., & Fincham, F. D. 2013. To belong is
burnout. Consulting Psychology Journal, 69:
to matter: Sense of belonging enhances meaning in
143–152.
life. Personality and Social Psychology Bulletin, 39:
1418–1427. Maslach, C., & Goldberg, J. 1998. Prevention of burnout:
New perspectives. Applied & Preventive Psychology,
Lanaj, K., Johnson, R. E., & Barnes, C. M. 2014. Begin-
7: 63–74.
ning the workday yet already depleted? Consequen-
ces of late-night smartphone use and sleep. Maslach, C., & Leiter, M. P. 2008. The truth about burn-
Organizational Behavior and Human Decision out: How organizations cause personal stress and
Processes, 124: 11–23. what to do about it. New York, NY: John Wiley &
Sons.
Lanaj, K., Johnson, R. E., & Wang, M. 2016. When lending
a hand depletes the will: The daily costs and benefits Maslach, C., & Leiter, M. P. 2016a. Burnout. In G. Fink
of helping. Journal of Applied Psychology, 101: (Ed.), Stress: Concepts, cognition, emotion, and
1097–1110. behavior: 351–357. San Diego, CA: Academic Press.
Leary, M. R., Tate, E. B., Adams, C. E., Batts Allen, A., & Maslach, C., & Leiter, M. P. 2016b. Understanding the
Hancock, J. 2007. Self-compassion and reactions to burnout experience: Recent research and its implica-
unpleasant self-relevant events: The implications of tions for psychiatry. World Psychiatry, 15: 103–111.
Maslach, C., Schaufeli, W. B., & Leiter, M. P. 2001. Job recommended remedies. Journal of Applied Psychol-
burnout. Annual Review of Psychology, 52: 397–422. ogy, 88: 879–903.
Melwani, S., Mueller, J. S., & Overbeck, J. R. 2012. Look- Pommier, E. 2010. The compassion scale [Doctoral disser-
ing down: The influence of contempt and compas- tation, University of Texas at Austin]. Texas Scholar-
sion on emergent leadership categorizations. Works Repository. Retrieved from https://repositories.
Journal of Applied Psychology, 97: 1171–1185. lib.utexas.edu/handle/2152/ETD-UT-2010-12-2213
Mikulincer, M., & Shaver, P. R. 2005. Attachment theory Preacher, K. J., Zyphur, M. J., & Zhang, Z. 2010. A general
and emotions in close relationships: Exploring the multilevel SEM framework for assessing multilevel
attachment-related dynamics of emotional reac- mediation. Psychological Methods, 15: 209–233.
tions to relational events. Personal Relationships, Quinn, R. W., Myers, C., Kopelman, S., & Simmons, S. A.
12: 149–168. 2021. How did you do that? Exploring the motivation
Mongrain, M., Chin, J. M., & Shapira, L. B. 2011. Practicing to learn from others’ exceptional success. Academy of
compassion increases happiness and self-esteem. Management Discoveries, 7. doi: 10.5465/amd.2018.
Journal of Happiness Studies, 12: 963–981. 0217
Muthen, L. K., & Muthen, B. O. 2017. Mplus user’s guide Rosenberg, M. 1965. The measurement of self-esteem. In
(8th ed.). Los Angeles, CA: Muthen & Muthen. M. Rosenberg (Ed.), Society and the adolescent self
image: 297–307. New York, NY: Princeton University
Naus, F., van Iterson, A., & Roe, R. 2007. Organizational cyn-
Press.
icism: Extending the exit, voice, loyalty, and neglect
model of employees’ responses to adverse conditions in Ruttan, R. L., McDonnell, M.-H., & Nordgren, L. F. 2015.
the workplace. Human Relations, 60: 683–718. Having “been there” doesn’t mean I care: When prior
experience reduces compassion for emotional dis-
Neff, K. 2003a. Self-compassion: An alternative conceptu-
tress. Journal of Personality and Social Psychology,
alization of a healthy attitude toward oneself. Self
108: 610–622.
and Identity, 2: 85–101.
Rynes, S. L., Bartunek, J. M., Dutton, J. E., & Margolis, J. D.
Neff, K. 2003b. The development and validation of a scale
2012. Care and compassion through an organizational
to measure self-compassion. Self and Identity, 2:
lens: Opening up new possibilities. Academy of Man-
223–250.
agement Review, 37: 503–523.
Neff, K. 2011. Self-compassion: The proven power of
Sbarra, D. A., Smith, H. L., & Mehl, M. R. 2012. When leav-
being kind to yourself. New York, NY: Harper
ing your ex, love yourself: Observational ratings of
Collins.
self-compassion predict the course of emotional
Neff, K., & Beretvas, S. N. 2013. The role of recovery following marital separation. Psychological
self-compassion in romantic relationships. Self and Science, 23: 261–269.
Identity, 12: 78–98.
Schabram, K., & Maitlis, S. 2017. Negotiating the chal-
Neff, K., Kirkpatrick, K. L., & Rude, S. S. 2007. Self-com- lenges of a calling: Emotion and enacted sensemaking
passion and adaptive psychological functioning. Jour- in animal shelter work. Academy of Management
nal of Research in Personality, 41: 139–154. Journal, 60: 584–609.
Neff, K., & McGehee, P. 2010. Self-compassion and psy- Schaubroeck, J. M., Shen, Y., & Chong, S. 2017. A
chological resilience among adolescents and young dual-stage moderated mediation model linking
adults. Self and Identity, 9: 225–240. authoritarian leadership to follower outcomes. Jour-
Organ, D. W. 1988. Organizational citizenship behavior: nal of Applied Psychology, 102: 203–214.
The good soldier syndrome. Lexington, MA: Lexing- Schaufeli, W. B., Leiter, M. P., & Maslach, C. 2009. Burn-
ton Books. out: 35 years of research and practice. Career Devel-
Parker, S. K., Johnson, A., Collins, C., & Nguyen, H. 2013. opment International, 14: 204–220.
Making the most of structural support: Moderating Schaufeli, W. B., Leiter, M. P., Maslach, C., & Jackson, S. E.
influence of employees’ clarity and negative affect. 1996. Maslach burnout inventory—general survey
Academy of Management Journal, 56: 867–892. (MBI-GS). In C. Maslach, S. E. Jackson, & M. P. Leiter
Pines, A. M. 2017. Burnout: An existential perspective. In (Eds.), MBI manual (3rd ed.): 19–26. Mountain View,
W. M. Schaufeli, C. Maslach, & T. Marek (Eds.), Pro- CA: CPP, Inc.
fessional burnout: Recent developments in theory Scott, B. A., & Judge, T. A. 2006. Insomnia, emotions, and
and research: 33–51. New York, NY: Routledge. job satisfaction: A multilevel study. Journal of Man-
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, agement, 32: 622–645.
N. P. 2003. Common method biases in behavioral Selig, J. P., & Preacher, K. J. 2008. Monte Carlo method for
research: A critical review of the literature and assessing mediation: An interactive tool for creating
confidence intervals for indirect effects [Computer moderator of recovery during lunch. Academy of
software]. Retrieved from http://quantpsy.org Management Journal, 57: 405–421.
Shepherd, D. A., & Cardon, M. S. 2009. Negative emotional Tsang, E. W., & Kwan, K.-M. 1999. Replication and theory
reactions to project failure and the self-compassion to development in organizational science: A critical real-
learn from the experience. Journal of Management ist perspective. Academy of Management Review,
Studies, 46: 923–949. 24: 759–780.
Sirois, F. M., Molnar, D. S., & Hirsch, J. K. 2015. Self-com- Tsui, A. S. 2013. 2012 Presidential address—On compas-
passion, stress, and coping in the context of chronic sion in scholarship: Why should we care? Academy
illness. Self and Identity, 14: 334–347. of Management Review, 38: 167–180.
Sommer, K. L., & Baumeister, R. F. 2002. Self-evalua- Uy, M. A., Lin, J. K., & Ilies, R. 2017. Is it better to give or
tion, persistence, and performance following receive? The role of help in buffering the depleting
implicit rejection: The role of trait self-esteem. Per- effects of surface acting. Academy of Management
sonality and Social Psychology Bulletin, 28: 926– Journal, 60: 1442–1461.
938.
van Kleef, G. A., Oveis, C., van der L€
owe, I., LuoKogan,
Sonnentag, S., Venz, L., & Casper, A. 2017. Advances in A., Goetz, J., & Keltner, D. 2008. Power, distress,
recovery research: What have we learned? What and compassion: Turning a blind eye to the suffer-
should be done next? Journal of Occupational Health ing of others. Psychological Science, 19: 1315–
Psychology, 22: 365–380. 1322.
Swider, B. W., & Zimmerman, R. D. 2010. Born to burnout: Vigna, A. J., Poehlmann-Tynan, J., & Koenig, B. W. 2018.
A meta-analytic path model of personality, job burn- Does self-compassion facilitate resilience to stigma? A
out, and work outcomes. Journal of Vocational school-based study of sexual and gender minority
Behavior, 76: 487–506. youth. Mindfulness, 9: 914–924.
Tan, C.-M. 2012. Everyday compassion at Google. Wang, M., Liu, S., Liao, H., Gong, Y., Kammeyer-Mueller,
Retrieved from https://www.ted.com/talks/chade_ J., & Shi, J. 2013. Can’t get it out of my mind: Employee
meng_tan_everyday_compassion_at_google rumination after customer mistreatment and negative
ten Brummelhuis, L. L., & Bakker, A. B. 2012. A resource mood in the next morning. Journal of Applied Psy-
perspective on the work–home interface: The chology, 98: 989–1004.
work–home resources model. American Psycholo- Williams, L. J., & O’Boyle, E. H. 2008. Measurement mod-
gist, 67: 545–556. els for linking latent variables and indicators: A
ten Brummelhuis, L. L., ter Hoeven, C. L., Bakker, review of human resource management research using
A. B., & Peper, B. 2011. Breaking through the loss parcels. Human Resource Management Review, 18:
cycle of burnout: The role of motivation. Journal of 233–242.
Occupational and Organizational Psychology, 84: Women at Work. 2019, November 11. How we take care of
268–287. ourselves. Retrieved from https://hbr.org/podcast/
Terry, M. L., & Leary, M. R. 2011. Self-compassion, 2019/11/how-we-take-care-of-ourselves
self-regulation, and health. Self and Identity, 10: World Health Organization. 2019, May 29. Burn-out—an
352–362. “occupational phenomenon.” Retrieved from http://
Terry, M. L., Leary, M. R., Mehta, S., & Henderson, K. www.who.int/mental_health/evidence/burn-out/en
2013. Self-compassionate reactions to health threats. Wu, J.-Y., & Kwok, O. 2012. Using SEM to analyze com-
Personality and Social Psychology Bulletin, 39: 911– plex survey data: A comparison between design-
926. based single-level and model-based multilevel
Thompson, B. L., & Waltz, J. A. 2008. Mindfulness, approaches. Structural Equation Modeling, 19:
self-esteem, and unconditional self-acceptance. Jour- 16–35.
nal of Rational-Emotive & Cognitive-Behavior Ther- Xanthopoulou, D., & Meier, L. L. 2014. Daily burnout
apy, 26: 119–126. experiences: Critical events and measurement
Trougakos, J. P., Beal, D. J., Green, S. G., & Weiss, H. M. challenges. In M. P. Leiter, A. B. Bakker, & C. Mas-
2008. Making the break count: An episodic examina- lach (Eds.), Burnout at work: 88–109. London,
tion of recovery activities, emotional experiences, and U.K.: Psychology Press.
positive affective displays. Academy of Management Yarnell, L. M., Stafford, R. E., Neff, K. D., Reilly, E. D.,
Journal, 51: 131–146. Knox, M. C., & Mullarkey, M. 2015. Meta-analysis of
Trougakos, J. P., Hideg, I., Cheng, B. H., & Beal, D. J. 2014. gender differences in self-compassion. Self and Iden-
Lunch breaks unpacked: The role of autonomy as a tity, 14: 499–520.
Zadro, L., Williams, K. D., & Richardson, R. 2004. How

low can you go? Ostracism by a computer is sufficient
to lower self-reported levels of belonging, control,
self-esteem, and meaningful existence. Journal of Kira Schabram ([email protected]) is an assistant
Experimental Social Psychology, 40: 560–567. professor at the University of Washington’s Foster School
of Business. Her research focuses on employees who seek
Zhang, J. W., & Chen, S. 2016. Self-compassion promotes
to make the world a better place through prosocial
personal improvement from regret experiences via
gestures and meaningful work. She earned her doctorate
acceptance. Personality and Social Psychology Bul-
from the University of British Columbia’s Sauder School
letin, 42: 244–258.
of Business.
Zhao, H., Peng, Z., & Chen, H.-K. 2014. Compulsory citi-
zenship behavior and organizational citizenship Yu Tse Heng ([email protected]) is a doctoral candidate at the
behavior: The role of organizational identification and University of Washington Foster School of Business. Her
perceived interactional justice. Journal of Psychol- research uncovers ways to humanize workplaces, with a
ogy, 148: 177–196. focus on compassion, grief, and suffering in organizations.
APPENDIX A: STUDIES 1 AND 2—MEASURES

The same measures were sourced for both studies. Please see the methods sections for descriptions of how
items were selected and adapted to each study’s requirements.
TABLE A1
Studies 1 and 2—Measures
Study 1 Study 2
Burnout Exhaustion: MBI-GS items 1–4, 6 Exhaustion: MBI-GS items 2, 3, 6

Schaufeli et al. (1996) Cynicism: MBI-GS items 8, 9, 13–15 Cynicism: MBI-GS items 8, 9, 14
Inefficacy: MBI-GS items 5, 7, 10–12, 16 Inefficacy: MBI-GS items 7, 10, 12
Enacted other-compassion 1. If I see someone at work going through a 1. Today, when I saw another at school going
Pommier (2010) difficult time, I try to act caring toward that through a difficult time, I tried to act caring
person. toward them.
2. I like to be there for other employees in times 2. Today, I liked to be there for others in times
of difficulty. of difficulty.
3. When other employees feel sadness, I always 3. Today, when others felt sadness, I tried to
try to comfort them. comfort them.
Enacted self-compassion 1. I try to be understanding and patient toward 1. Today, I was kind to myself when I was
Neff (2003b) those aspects of my personality I don’t like. experiencing suffering.
2. I’m kind to myself when I’m experiencing 2. Today, when I was going through a hard
suffering. time, I gave myself the caring and tenderness
3. When I’m going through a very hard time, I I needed.
give myself the caring and tenderness I need. 3. Today, I was tolerant of my own flaws and
4. I’m tolerant of my own flaws and inadequacies.
inadequacies.
5. I try to be loving toward myself when I’m
feeling emotional pain.
Self-control — 1. I feel drained from school.
Lanaj et al. (2014) 2. My mind feels unfocused at school.
3. At school, it takes a lot of effort for me to
concentrate on something.
4. At school, I cannot absorb any additional
information.
5. At school, I feel like my willpower has gone.
Belonging — 1. I feel accepted by others at [school name].
Zadro et al. (2004) 2. I feel as though I have a “connection” or
bond with others at [school name].
3. I feel like an outsider at [school name].
TABLE A1
(Continued)
Study 1 Study 2
Self-esteem — 1. I feel that I am a person of worth, at least on

Rosenberg (1965) equal plane with others.
2. I am inclined to feel that I am a failure
[reverse-coded].
3. I have much to be proud of.
4. I take a positive attitude toward myself.
Sleep quality — 1. I had trouble falling asleep.
Scott and Judge (2006) 2. I had trouble staying asleep (including
waking up too early).
3. I woke up several times during the night.
4. I woke up after my usual amount of sleep
feeling tired and worn out.
Note: MBI-GS, Maslach burnout inventory general survey.
APPENDIX B: STUDY 2—INTERVENTION
TABLE B1
Study 2—Intervention
Condition Prompt Sample Responses
Self- At some point today, you will find yourself “Today, at school, I was working on a group project in a team I
compassion facing a difficulty related to school (e.g., don’t get along with. One member in particular I have a long
problems with a team, struggling with an history with and it just makes me uncomfortable to be with him. I
assignment or deadline, etc.). We will ask tried to be really kind to myself by listening to uplifting music
you to describe this situation in the evening beforehand and used a breathing meditation app afterwards to keep
survey. Now, here comes your task: It is easy my mood at bay.”
to criticize or blame ourselves for perceived “Today, I was very stressed because I have a group project coming
shortcomings. But, remember that all up and none of my group members are really pulling their weight. I
students at [school name] face challenges and am worried we won’t be able to finish in time, but I also don’t
this is completely normal! Instead of being want to do all the work. I was kind to myself by allowing myself to
critical, come up with one specific action to take a two-hour nap after my classes, and I really enjoyed it
be compassionate toward yourself: say because I need some extra sleep.”
something nice to make yourself feel better, “I struggled today with an essay I am writing for one of my courses.
get yourself a treat, engage in an act of I felt like I should know this material but am just slower at
self-care, etc. understanding the material. To be kind to myself, I told myself it is
okay to not understand the first time around and I will eventually
learn. Additionally, I treated myself to a meal [near campus] to fuel
my endeavors and I felt a little better.”
Other- At some point today, you will witness another “Today, when I was with my friend at lunch, he told me about how
compassion person at [school name] facing a difficulty he did poorly on his practice dental school exam. In his time of
(e.g., problems with a team, struggling with sadness, I gave him words of encouragement in order to try and
an assignment or deadline, etc.). We will ask give him a positive outlook. Doing this made me feel as though I
you to describe this situation in the evening was making a difference.”
survey. Now, here comes your task: It is easy “My friend had a hard time in her lab today because a classmate
to criticize or blame others for perceived was being mean to her. He was so rude, it made her cry. We had a
shortcomings. But, remember that all nice long talk in my room until she was giggly and happy. It made
students at [school name] face challenges and me feel happy to help my friend and knowing she has more
this is completely normal! Instead of being confidence to stand up to him the next time it happens.”
critical, come up with one specific action to “A difficult situation faced by someone at school today was in my
be compassionate toward them: say management class in the afternoon with one of my group members.
something nice to them to make them feel They were struggling with little sleep and weren’t able to do the
better (email/text if you’re not on campus), readings that were assigned for the day so they were lost with what
get them a treat, engage in an act of care, etc. we were talking about. It made me feel sad for them so I helped
them by explaining the material.”
TABLE B1
(Continued)
Condition Prompt Sample Responses
Control Your task today is to keep careful track of what “For dinner, I ate: rice, shrimp, cooked asparagus, broccoli and
you are eating. Please try to be mindful of all carrots, and a side salad. I also had a peanut butter cookie. The
the ingredients in your food, especially in meal was at my sorority with my friends at about 5:20 and made
your bigger meals. If possible, please write me feel happy.”
down everything you eat today. “I ate a sandwich—two slices of toasted whole-wheat bread with
oven-roasted chicken breast. I also had some veggie chips. I ate at
my apartment around noon with my sister. It made me feel full.”
“I had a southwestern quinoa salad (vegan—corn, black beans,
tomatoes, jalape~ no) with avocado toast and eggs for lunch at 11:30.
It felt great because that’s my favorite meal but it didn’t fill me up
for very long, which is weird because that meal usually keeps me
full for at least a few hours.”
Copyright of Academy of Management Journal is the property of Academy of Management
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.

How Other - and Self-Compassion Reduce Burnout Through Resource Replenishment

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

How Other - and Self-Compassion Reduce Burnout Through Resource Replenishment

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

How Other - and Self-Compassion Reduce Burnout Through Resource Replenishment

Uploaded by

Copyright:

Available Formats

r Academy of Management Journal

2022, Vol. 65, No. 2, 453–478.

HOW OTHER- AND SELF-COMPASSION REDUCE

Initial Burnout Compassion Resources Later Burnout

1 Exhaustion (T1) 2.85 0.88 (.91)

TABLE 2 inefficacy at Time 3, b 5 2.06, SE 5 0.05, ns; nor was

1 Directed OC (t) 0.31 0.46 (—)

GENERAL DISCUSSION directed via instruction). This research strategy

Zadro, L., Williams, K. D., & Richardson, R. 2004. How

APPENDIX A: STUDIES 1 AND 2—MEASURES

Burnout Exhaustion: MBI-GS items 1–4, 6 Exhaustion: MBI-GS items 2, 3, 6

Self-esteem — 1. I feel that I am a person of worth, at least on

Note: MBI-GS, Maslach burnout inventory general survey.

APPENDIX B: STUDY 2—INTERVENTION

You might also like