Burnout is defined as a state of physical, mental, and emotional exhaustion caused by a combination of very high expectation and persistent situational stress (Freudenberger, 1974). Maslach et al. (1996) conceptualized burnout as a three-dimensional syndrome: (a) emotional exhaustion, (b) cynicism (also known as depersonalization), and (c) diminished professional efficacy (also known as lack of personal accomplishment). Emotional exhaustion represents a depletion of emotional energy and resources. Employees who are emotionally exhausted typically experience physical and cognitive fatigue. Cynicism describes where employees take cold, indifferent attitudes toward their job, coworkers, and organization. Diminished professional efficacy reflects feelings of reduced ability on the job. When employees feel a sense a decline in personal job competence, they feel a growing sense of inadequacy.

With regard to the sequence of the three burnout components, Leiter (1993) proposed a mixed sequential and parallel development model, indicating that burnout starts with emotional exhaustion, which in turn contributes to increased cynicism. This suggests that emotional exhaustion arises from work stressors, that is, from being continuously challenged with high job demands (e.g., work pressure, high emotional demands). Subsequently, employees can develop a cynical attitude toward the job as a coping strategy to distance themselves from the stress (Naus et al., 2007; Taris, LeBlanc, Schaufeli, & Schreurs, 2005). Thus, cynicism may mediate the relation between work stressors and behavioral outcomes. Meanwhile, the diminished professional efficacy triggered by the work environment develops in parallel with emotional exhaustion.

Job performance is one of the major organizational outcomes associated with employee burnout (Brandes et al., 2008; Wright & Cropanzano, 1998). Borman and Motowidlo (1993) identified two broad dimensions of job performance: (a) task performance (also known as in-role performance) and (b) contextual performance (also known as extra-role performance). Task performance includes behaviors that directly contribute to the technical core processes and maintenance activities in an organization, such as producing products, distributing finished products, acquiring inventory, managing subordinates, or delivering services. Contextual performance involves behaviors that contribute to the culture and climate of the organization which influences others to carry out organizationally valuable work. Examples would include defusing hostility and conflict and encouraging interpersonal trust.

Although the burnout literature suggests it may be associated with poorer job performance, this relationship has received inconsistent empirical support. For example, some studies supported a negative association between burnout and performance (Bhagat, Allie, & Ford, 1995; Wright & Cropanzano, 1998), while others have found a positive one (Keijsers, Schaufeli, Le Blanc, Zwerts, & Reis-Miranda, 1995; Randall & Scott, 1988) or no relationship at all (Hakanen & Koivumäki, 2014; Parker & Kulik, 1995; Wright & Bonett, 1997). Demerouti et al. (2005) suggested these conflicting results may reveal a relationship between burnout configurations, rather than burnout dimensions and performance. Nevertheless, because the weight of the literature suggests a negative burnout-performance relationship, we expect that each of the burnout components will be negatively associated with both forms of job performance.

Prosocial behavior refers to positive acts, such as helping, sharing, donating, cooperating, and volunteering. Prosocial behavior could overlap with a number of constructs, including organizational citizenship behavior (OCB) and contextual performance. However, prosocial behavior differs from OCB in that it can be either role-prescribed or extra-role; OCB by definition can only be extra-role (Wilkerson et al., 2008). Further, prosocial behavior can be positive toward a coworker, but negative toward an organization. An example of this might be helping a coworker solve a personal issue at the expense of the timely task completion for an important client (Borman & Motowidlo, 1997). Thus, by engaging in the prosocial behavioral act, the individual rather than the organization benefits.

Prosocial behavior, albeit conceptually similar to contextual performance, also differs meaningfully from the construct. Borman and Motowidlo (1997) posited a five-part taxonomy that allows for a more fine-grained understanding of contextual performance that includes: (a) persisting enthusiastically and with extra effort to successfully complete one's own tasks (e.g., perseverance, conscientiousness), (b) volunteering to take on tasks not necessarily related to one's job (e.g., taking initiative), (c) helping and cooperating (e.g., helping coworkers, altruism), (d) following organizational rules and procedures (e.g., complying with organizational values and policies, and (e) supporting and embracing organizational objectives (e.g., organizational loyalty). The taxonomy demonstrates that engaging in prosocial behavior is part of contextual performance (most closely aligned with a-c above), yet because it is primarily motivated by egoistic and self-serving concerns (Dovidio, Piliavin, Gaertner, Schroeder, & Clark, 1991), unlike contextual performance, it does not necessarily have to contribute to the overall success of the organization (Borman & Motowidlo, 1997). Research suggests that emotional exhaustion, cynicism, and diminished professional efficacy, as a result of overwhelming demands and lack of physical and psychological resources, have a negative influence on prosocial behavior (DeWall, Baumeister, Gailliot, & Maner, 2008; Xu et al., 2012). For example, a challenging, stressful work environment would make a high burnout employee less likely to be prosocial or willing to help and cooperate with others.

The JD-R model proposes that working conditions can be categorized as job demands and job resources, each leading to two psychologically different processes. In the first process, demanding aspects of work (e.g., work overload, unreasonable customers) lead to constant overtaxing and in the long run to exhaustion (e.g., Leiter, 1993; Wright & Cropanzano, 1998). In the second process, a lack of job resources complicates the meeting of job demands and impedes actual goal accomplishment, which leads to failure and frustration (Bakker, Demerouti, DeBoer, & Schaufeli, 2003). Regardless of the type of occupation, burnout develops when job demands are high and job resources are limited because such negative working conditions lead to energy depletion and undermine employees’ motivation (Demerouti et al., 2001). Research has provided empirical evidence that job demands were likely the most important antecedents of emotional exhaustion, which, in turn, would predict task performance (Bakker et al., 2004; Hockey, 1993). Conversely, job resources (e.g., autonomy, performance feedback) tend to be the most powerful predictors of contextual performance through their relationship with cynicism (Bakker et al., 2004).

This study incorporated both Leiter's (1993) process model and the JD-R model as they applied to burnout, job performance, and prosocial behavior. The negative influence of emotional exhaustion on job performance can be explained because it not only directly influences job performance, but also because it can lead to cynicism, which in turn influences contextual performance in particular, but also task performance to a significant, but lesser extent. This indicates that both the job demands-exhaustion-task performance sequence and the job demands-exhaustion-cynicism-contextual performance sequence may coexist. Furthermore, the combined model involves not only the indirect influence of job demands on cynicism through emotional exhaustion, but also the direct influence of job resources on cynicism. Given those possible sequences started from either emotional exhaustion or cynicism separately or jointly, the present study suggests that cynicism is associated with inadequate job resources and/or is developed through exhaustion resulting from excessive job demands, and thus will have a negative association with both task performance and contextual performance.

As for prosocial behavior, this study posits that stressful working conditions with excessive demands or lack of resources may wear down employees, leading them to experience burnout, which progressively worsens the employees’ desire to initiate prosocial behavior. Emmerick, Jawahar, and Stone (2005) noted that employees’ sense of emotional and mental resource depletion averted them from putting extra effort or spending personal time in activities that exceed job requirements. If an individual's resources are depleted, it becomes difficult for the individual to reflect on their behavior, reassess the decision process, reach factual or counterfactual conclusions, and store appraisal and evaluation information in memory (Xu et al., 2012). Studies have demonstrated that resource depletion decreases the likelihood of prosocial behavior (DeWall et al., 2008; Xu et al., 2012). Relating to the three burnout components, emotional exhaustion and cynicism react to a work situation with chronic, overwhelming demands which foster feelings of depleted resources and lack of energy (Bakker et al., 2004; Leiter, 1993). Diminished professional efficacy reflects being frazzled and exhausted as a result of an excessive demand on energy, power, and resources (Freudenberger, 1974). Accordingly, this study proposes that all three burnout components are negatively associated with prosocial behavior.

Based on the aforementioned literature, we hypothesize:

Hypothesis 1: Emotional exhaustion is negatively related to task performance, contextual performance, and prosocial behavior.

Hypothesis 2: Cynicism is negatively related to task performance, contextual performance, and prosocial behavior.

Leiter's (1993) mixed sequential and parallel development model argued that the feelings of diminished professional efficacy triggered by the work environment develops in parallel with the other two burnout components, rather than sequentially. Contrary to the other components, personal accomplishment reflects an individual difference characteristic similar to self-efficacy (Cordes & Dougherty, 1993). Research suggested that employees with low levels of professional efficacy tend to have a deflated view of their progress, being disappointed with tasks, projects, and relationships (Shoss, Jiang & Probst, 2016). That is, diminished professional efficacy, which occurs with feelings of inefficacy in one's job and poor professional self-esteem, is associated with low task and contextual performance (Demerouti et al., 2005).

Hypothesis 3: Diminished professional efficacy is negatively related to task performance, contextual performance, and prosocial behavior.

A large body of research has proposed that cynicism is negatively associated with both job performance and prosocial behavior (e.g., Bakker et al., 2004; Demerouti et al., 2005). In addition to the relationship between cynicism and job-related outcomes, this study investigates the possible positive role of cynicism at work where an individual suffers from the various accompanying symptoms of burnout. Mirvis and Kanter (1989, 1991) explained that cynicism is different from skepticism in that it is a strategy individuals can employ to cope with an unfriendly, unstable, and insecure world, which is in essence a healthy response to work and life. The authors argued that employees who experienced, for example, a lack of job security under unstable economic conditions tend to build a self-protective shell around them. Thus, cynicism can be understood as a defensive, cognitive method of creating a protective distance in stressful situations (Cartwright & Holmes, 2006). Individuals may hold a distant attitude (i.e., cynicism), which prevents one themselves from letting their job performance suffer when dealing with the physical, emotional exhaustion, and feelings of ineffectiveness caused by excessive and prolonged stress.

Supporting the notion of cynicism as a coping strategy, Taris et al.’s (2005) study provided longitudinal evidence from data derived from two Dutch samples of 1,185 oncology care providers and teachers. Their findings revealed that high levels of exhaustion triggered high levels of depersonalization, which is consistent with previous studies (i.e., Lee & Ashforth, 1993; Leiter & Maslach, 1988). Research (e.g., Golembiewski, Munzenrider, & Stevenson, 1986; Leiter & Maslach, 1988) has proposed the causal relationships among Maslach et al.’s (1996) three burnout components. However, the relationship among the other components and organizational outcomes has not been sufficiently investigated, especially with those who are simultaneously experiencing all three symptoms of burnout. Brandes and Das (2006) proposed a non-linear relationship between the different aspects of cynical behavior and performance, emphasizing that moderate levels of cynicism can improve performance because it serves as an employee coping mechanism (yet, low and high levels of cynicism are detrimental to performance). Thus, despite cynical employees’ feelings of disillusionment and frustration, they do not necessarily convert their displeasure into action, which can directly influence not only job performance, but also organizational performance (Johnson & O’Leary-Kelly, 2003).

Moreover, cynical employees may have an experienced, critical eye, which could be a positive force for change (Brandes & Das, 2006). Due to the work insights gained from the critical analysis of prior performances, cynical employees can work better (Cutler, 2000). Brandes and Das (2006) labeled cynical employees as “closet idealists” who desire to reform and improve, but become disillusioned with their work. Empirical research supports the claim that cynical employees may positively affect organizations. Cynical employees appear to have less intention to comply with requests to engage in unethical behavior, for instance (Andersson, 1996). Reichers, Wanous, and Austin (1997) also found that more than half of the employees falling within the high cynicism rating indicated their willingness to make efforts to change. Further Cutler (2000) argued that cynical employees may accept personal responsibility for their own actions and ideas. “Towing the corporate line” without taking risks can be rather safe although it can be dull and uninspiring (Cutler, 2000; Turnbull, 1999). Demerouti et al.’s (2005) study demonstrated that the employees who depersonalized toward their customers showed that they were successful in selling products and services (i.e., satisfactory task performance). The employees also showed sufficient organizational citizenship behaviors (i.e., satisfactory contextual performance). These employees tended to take the perspective of their customers, yet they still had concern for their own problems and/or demands, necessitating the need for maintaining emotional detachment (Demerouti et al., 2005). Therefore, while emotional exhaustion and diminished professional efficacy can negatively influence employees’ behavioral outcomes, it may be incorrect to propose that cynical employees will necessarily have low job performance and low prosocial behavior. The present study proposes that employees’ cynicism can play a positive role in partially mediating the relationships between their emotional exhaustion and diminished professional efficacy and their job performance and prosocial behavior while those high in emotional exhaustion and diminished professional efficacy still exhibit lower levels of job performance and prosocial behavior. Given the above discussion, we formulated the following hypothesis (see also Figure 1).

Figure 1.

Conceptual Model of Hypothesized Associations of Burnout with Behavioral Outcome Variables.

(0.19MB).

Hypothesis 4: Cynicism positively and partially mediates the relationships between the two burnout components (emotional exhaustion and diminished professional efficacy) and the three outcome variables (task performance, contextual performance, and prosocial behavior), while emotional exhaustion and diminished professional efficacy have direct negative relationships with the three outcomes.

Participants were 262 working adults (a response rate of 78.8%), 44.4% men, 55.6% women; average age = 25.54 years, SD = 7.32, attending a large southeastern university in the United States. There are arguments for and against the use of student-recruited samples, indicating whether results obtained from student subjects of a research study are generalizable to non-student populations (Peterson & Merunka, 2014). However, numerous researchers (Demerouti & Rispens, 2014; Hochwarter, 2014; Wheeler, Shanine, Leon, & Whitman, 2014) have suggested that student-recruited sampling is a useful data collection strategy. The Wheeler et al.’s (2014) meta-analysis revealed that no substantive, demographic difference existed between student-recruited samples and non-student-recruited samples. They found few differences in the observed correlations (e.g., effect sizes), which would not lead to different practical conclusions when inferring relationships between constructs irrespective of sampling method. In addition, student-recruited samples have several advantages of increasing the heterogeneity of the participants, having students work on faculty research as an effective educational practice, reducing research budgets, and conducting studies with more elaborate research designs (Demerouti & Rispens, 2014).

The researchers administered each of the survey batteries in class over two separate times; completion time for each survey battery was roughly 15minutes. Because this was a study of worker attitudes and beliefs as they were linked to important organizational outcomes, the independent variables were collected first, followed a month later by the dependent variables. Of the participants, 54.5% were working part-time while 45.5% were employed full-time. The majority of participants (over 80%) was working as customer and professional service providers (e.g., retail sales, cashier, hostess, recreation/sport service providers, teachers). Data showed that Hispanics made up 37.7% of the participants, 30.6% were African American, 23.8% were Caucasian, and 7.9% were “Other.” The majority of the participants were single (77.5%) while 15.3% were married, and 7.2% were partnered.

Use of each of the measures employed in this research was strongly supported by the reliability and validity evidence presented by the respective instruments’ authors, as well as the bulk of the research literature. The items for the measures are presented in Table 1.

Maslach Burnout Inventory. This study measured burnout using the Maslach Burnout Inventory (MBI; Maslach et al., 1996). The scale includes the three burnout dimensions: emotional exhaustion (5 items), cynicism (5 items), and professional efficacy (6 items). The burnout items were scored on a 6-point Likert-type scale from 1 (never) to 6 (a few times a week). As a high score on the professional efficacy scale demonstrated low burnout, which is opposite to the other two burnout scales where a high score demonstrated high burnout; the professional efficacy scores were reflected during the analyses to be consistent with the other two scale scores (Kline, 1998).

Job performance. The job performance scale developed by Motowidlo and Van Scotter (1994) was used to measure the participants’ task performance (5 items) and contextual performance (6 items). The responses to the job performance items were assessed on a 5-point Likert-type scale from 1 (not at all likely) to 5 (extreme likely).

Prosocial behavior. Prosocial behavior was measured using a 10-item scale developed by Caprara and Pastorelli (1993). The prosocial scale assessed their degree of helpfulness, sharing, kindness, and cooperation on a 5-point Likert response format 1 (never) to 5 (always).

Demographic characteristics. The questionnaire also included four demographic questions: sex, age, ethnicity, and marital status, to address demographic characteristics of the participants.

In this study, we did not include any control variables. Methodologists recommend that researchers provide theoretical justification for the need and relevance of control variables (Atinc, Simmering, & Kroll, 2012; Schjoedt & Bird, 2014; Spector & Brannick, 2011). They noted that control variables can profoundly change the relationships between the independent variables and the dependent variables and controlling for the confounding influence of the extraneous variables on the predictor variables does not control for the confounding effect of the control variables on the criterion variable and vice versa. This indicates that using control variables can be problematic and may invalidate the research findings. Thus, no control variables were included in this study.

Because the data were self-reported, where information was collected from the same sources, the Harman single-factor test was employed as a diagnostic tool to assess the presence of common method variance bias (CMV; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Confirmatory factor analysis (CFA) was used to verify how well the variables defined their respective constructs (Guarino, 2004) using the LISREL 8.72 program. To examine the reliability of the data, the internal consistency estimates were calculated with Cronbach's alpha for the items selected to represent the variables (Brown, 2002). Structural equation modeling (SEM) was then performed to test the research hypotheses, assessing how well the data fit the proposed model (see Figure 1).

According to Malhotra, Kim, and Patil (2006) and Podsakoff et al. (2003), two criteria can lead to the presence of CMV, if (a) a single strong factor emerges from an exploratory factor analysis or (b) a first factor accounts for the majority of the variance in the variables. However, the Harman single-factor results showed there were nine factors with eigenvalues greater than 1 and thus no single factor accounted for the majority of the variance as the factors ranged from 2.87% to 21.70%. Consequently, this indicates that CMV was unlikely to be a major source of the variations in the observed items of the present study.

The measurement model included 37 items comprising the three burnout components of exhaustion, cynicism, and professional efficacy, the two constructs of task performance and contextual performance, and prosocial behavior. The CFA results indicated that the items for each construct significantly loaded on their corresponding construct with their factor loadings above .4 for the respective manifest variables (see Table 1). The chi-square test was significant (χ2 = 1482.88, df = 614, p < .01), which is regarded as unacceptable. However, the chi-square test of absolute model fit is affected by large sample sizes and non-normality in the underlying distribution of the input variables (Kline, 1998). Other measures of fit were thus used to examine the overall fit of the model to the data. The alternative fit indices indicated that the measurement model had a reasonable fit to the data: RMSEA = .071, CFI = .92, NNFI = .91, IFI = .92 and RMR = .069. The RMSEA met the recommended criteria of .08 or less (Steiger, 1998) for a reasonable fit. The CFI, the NNFI, and the IFI all exceeded the minimum cut-off of .90 (Kelloway, 1998; Kline, 1998). The RMR value was also acceptable with the recommended criteria of .08 or less (Browne & Cudeck, 1993). Therefore, the results indicated that the measurement model achieved an adequate fit for the proposed factor dimensions, providing evidence of factorial validity and convergent validity of the study's measurement (Bagozzi & Yi, 1988).

The reliability measurements were found to be above an acceptable level. Cronbach's alpha ranged from .74 to .92, exceeding the minimum level (.70) recommended by Nunnally and Bernstein (1994). The Cronbach's coefficients for the three burnout components of exhaustion, cynicism, and professional efficacy were .92, .84, and .78, respectively. The Cronbach's alpha values for task performance and contextual performance, and prosocial behavior were .74, .82, and .86, respectively. The internal consistency estimates of the variables are shown in Table 1.

Table 2 presents the intercorrelations among the burnout and job performance variables along with their means and standard deviations. Most of the correlations were significant and in the direction suggested by theory and research; even the highest correlation of .58 between task performance and contextual performance resulted in a shared variance of only 33.6 percent. The results therefore suggested preliminarily that the three burnout components and the two job performance constructs were sufficiently distinct from each other and that they represented unique constructs.

Hypothesis 1. The SEM results showed that the model addressing the negative relationships of emotional exhaustion with the three outcome variables had a reasonable fit to the data. Although the chi-square value (χ2 = 811.95, df = 293, p < .01) was significant, other fit indices indicated a good fit of the model (RMSEA = .080, CFI = .93, NNFI = .92, IFI = .93, RMR = .069). The direct negative effect of exhaustion with contextual performance was accepted (β = -.18; p < .01), while the relationships of exhaustion with task performance and prosocial behavior were not significant. Thus, the first hypothesis was partially supported.

Hypothesis 2. The results revealed that the model had a marginally acceptable fit to the data (RMSEA = .082, CFI = .91, NNFI = .90, IFI = .91, RMR = .074), (χ2 = 837.20, df = 293, p < .01). Although the RMSEA (.082) slightly exceeded the recommended criteria of .08 (Steiger, 1998) for an acceptable fit, these types of fit indices may degrade when models enclose large numbers of variables as we tested in this research (Kenney & McCoach, 2003). In such situations, Kenny and McCoach (2003) suggest that as long as a number of other fit indices meet the criterion for being acceptable, the achieved structural model represents an acceptable fit to the data. Cynicism was found to have a significant, direct negative effect on task performance (β = -.21, p < .01), yet the negative relationships between cynicism and contextual performance and between cynicism and prosocial behavior were not statistically significant. Overall, the hypothesis was partially supported.

Hypothesis 3. The model with the relationships of professional inefficacy with the three outcome variables had an acceptable fit to the data as the fit indices appeared satisfactory (χ2 = 879.97, df = 318, p < .01, RMSEA = .080, CFI = .92, NNFI = .91, IFI = .92, RMR = .068). Professional inefficacy had a significant and powerful direct negative effect on task performance (β = -.51, p < .001), contextual performance (β = -.41, p < .001), and prosocial behavior (β = -.36, p < .001). Therefore, the hypothesis was fully supported.

Hypothesis 4. According to the SEM results, the proposed model had an acceptable fit to the data (χ2 = 1528.28, df = 617, p < .01, RMSEA = .073, CFI = .91, NNFI = .91, IFI = .91, RMR = .079). Supporting the fourth hypothesis, all the relationships were found to be significant. This corroborates the notion that although emotional exhaustion and professional inefficacy have direct negative effects on the three outcome variables, cynicism plays a partial mediating role in the positive relationships between the other two burnout components and the outcome variables (see Figure 2).

Figure 2.

Results from SEM Analysis for the Proposed Model.

* p < .05, ** p < .01, *** p < .001.

(0.35MB).