Workplace Affective Commitment, Emotional Labor and Burnout: A Multiple Mediator Model

The objective of this study was to investigate the mediating effects of emotional labor strategies on the relationships between multiple targets of workplace affective commitment and burnout components. A total of 370 service employees from Canada completed a questionnaire. Hierarchical regression analyses were performed and the significance of indirect effects was computed using bias-corrected bootstrap 95% confidence intervals. Results revealed that: (a) indirect effects were almost absent in the prediction of emotional exhaustion, (b) www.ccsenet.org/ijbm International Journal of Business and Management Vol. 7, No. 1; January 2012 ISSN 1833-3850 E-ISSN 1833-8119 4 surface acting significantly mediated the relations between affective commitment and cynicism, and (c) expression of naturally felt emotions significantly mediated the relation between affective commitment and feelings of professional inefficacy. These findings suggest that specific emotional labor strategies may represent psychological mechanisms intervening in the relationship between affective commitment and burnout development or alleviation. Further research is needed to fully understand how each emotional labor strategy impact on relevant individual and organizational outcomes.

Considering that occupational stress is now commonly depicted as an epidemiological phenomenon (Quick, Quick, Nelson, & Hurrell, 1997), the identification of internal resources that may help to foster employees' well-being and reduce risks of developing burnout would be particularly helpful. Among potential internal resources, affective commitment in the workplace (Meyer, 2009;Meyer & Herscovitch, 2001), an intrinsic motivational force known to predict favourable outcomes for individuals and organizations, appears particularly promising.
More precisely, commitment in the workplace is a force that binds an individual to a course of action that is of relevance to one or more targets, accompanied by a specific mind-set (affective in this case) reflecting the emotional attachment and identification to the target, which plays a role in shaping and motivating workplace behaviors (Meyer & Allen, 1991;Meyer & Herscovitch, 2001;Meyer Becker, & Vandenberghe, 2004). As such, workplace affective commitment represents an indicator of person-environment fit in the workplace. Moreover, focusing on multiple targets of commitment, in addition to the organization itself, is a particularly relevant approach, considering the changing nature of work and work contracts and the increasing complexity of organizational structures (Cohen, 2003;Klein, Molloy, & Cooper, 2009;Meyer, 2009;Morin, Madore, Morizot, Boudrias, & Tremblay, 2009;Morin, Morizot, Boudrias, & Madore, 2011;Morrow, 1993;Stinglhamber, Bentein, & Vandenberghe, 2002). Among all the possible workplace-related targets, organization, supervisor, colleagues, customers, work itself, career, and job can be considered as "universal targets" because they apply to a vast majority of workers (Cohen, 2003;Morin et al., 2009Morin et al., , 2011Morrow, 1993).
In addition to these targets, commitment towards the organization's emotional display rules may also be particularly important for service employees since these rules represent a central aspect of their daily work. Thus, commitment to these rules might be intimately related to their reliance on particular emotional labor strategies, and may reflect the ease with which employees will adopt these rules (Ashforth & Humphrey, 1993;Gosserand & Diefendorff, 2005;Hochschild, 1983). Indeed, previous research showed that the relationships between attitudes (i.e. commitment) and behaviors (i.e. emotional labor displays) tend to be stronger and more significant when they refer to similar contexts, actions, timelines, and targets (e.g. Fishbein & Ajzen, 1975;Lavelle, Rupp, & Brockner, 2007). Moreover, service organizations, which tend to be emotionally demanding, have a propensity to employ workers that are commonly under-committed and overstressed, as suggested by their higher rates of turnover and burnout (Hartline & Ferrell, 1993;Henkoff & Sample, 1994;Milbank, 1993).

The present study
It is thus apparent that workplace affective commitment, emotional labor, and burnout, three major issues for service organizations, are potentially closely interrelated. Indeed, the role of workplace affective commitment as an internal resource related to lower risks of burnout has been previously well documented (e.g. Lee & Ashforth, 1996;Maslach et al., 2001). This relationship can be partly explained by the energizing nature of affective commitment that may provide employees with the extra edge necessary to overcome daily difficulties (Meyer et al., 2004) and by the fact that affective commitment toward specific social targets generally occurs in the context of positive social bonds, which also represent an important resource against daily adversities (Becker, 2009).
In addition, emotional displays are central to service employees' daily tasks and will undoubtedly be influenced by work attitudes as central as affective commitment (Grandey, 2000;Morris & Feldman, 1996). Affective commitment directed at specific targets is a force that is expected to bind employees to behaviors that should benefit these specific targets (Becker & Billings, 1993;Meyer & Hescovitch, 2001). Hence, it is probable that, for service employees, affective commitment toward relevant targets, such as display rules or customers, will also predict an increased reliance on the emotional labor strategies that are most beneficial to these targets. For instance, an employee committed towards customers might be more prone to genuinely express the proper emotions during service encounters. In turn, the authenticity of these displays will increase customers' impressions that the employee is truly willing to help, which is clearly a valuable outcome to them (Groth et al., 2009).
Moreover, previous authors (Panaccio & Vandenberghe, 2009), building on the conservation of resources theory (Hobföll, 1989) argued that commitments shape the resources necessary for employees to carry out their work-related responsibilities. Affective commitment, as it entails emotional attachment and identification to a target, provides valuable internal resources to employees by providing them with a sense of direction and purpose, as well as with related feelings of confidence and self-esteem (Hobföll, 2002;Panaccio & Vandenberghe, 2009). These positive emotions also mean that highly committed employees will need to rely less on surface acting and deep acting strategies in order to modify their emotional displays, which will already be more appropriate. These added internal resources are then available for employees when they need to cope with work related stressors and to help them in the achievement of work-related goals, in turn making it easier for them to regulate their emotional displays (Grandey, 2000;Hobföll, 1989;Panaccio & Vandenberghe, 2009). The reliance on specific emotional labor strategies, each underlying different degrees of resource depletion, could therefore partially depend on affective commitment. For instance, if an employee is weakly committed towards his organization and his job, he might lack of resources to cope with the emotional demands of his daily tasks, in addition to feeling increased levels of negative emotions reflecting his lack of commitment, and thus have to rely on the more exhausting surface acting strategy.
The proposed theoretical linkage between affective commitment and emotional labor presented here differs from previous studies (Seery & Corrigall, 2009;Yang & Chang, 2008) that positioned commitment as a general attitude (Mowday, Porter, & Steers, 1979) or an affective state (Meyer, Allen, & Smith, 1993) resulting from emotional labor (also see Klein et al., 2009). However, this rationale is more consistent with recent integrative representations of commitment as an intrinsic driving force for behaviors (Becker, 2009;Meyer et al., 2004;Meyer & Hescovitch, 2001).
If workplace affective commitment towards different targets is truly related to emotional labor strategies, these strategies may then differently affect employees' internal resources therefore leading to more or less important levels of burnout, a known consequence of demands-resources imbalances (Baumeister et al., 1998;Goldberg & Grandey, 2007;Hobföll, 1989;Muraven & Baumeister, 2000;Richard & Gross, 1999;Wilk & Moynihan, 2005). Thus, it can be hypothesized that the effects of workplace affective commitment towards various targets on burnout components might be at least partially mediated by its effects on specific emotional labor strategies, as shown in Figure 1.
Notwithstanding this generic proposition, the present study remains mostly exploratory given the newness of the questions studied here and the lack of previous empirical research linking these three constructs while taking their multidimensional nature into account. This is especially true of the multi-target perspective on workplace affective commitment, which has yet to be integrated into mainstream commitment research (Cohen, 2003;Morin et al., 2009Morin et al., , 2011. Still, the inclusion of emotional labor strategies as intermediate variables in this relation represents an attempt to capture how organizationally-driven emotional processes may explain the relation between individuals' internal resources on important outcomes like burnout. This investigation is particularly relevant for emotionally demanding workplaces like service organizations, where emotional displays are often extrinsically monitored and have a huge impact on service quality (Ashkanasy et al., 2002;Morris & Feldman, 1996). Therefore, this study will investigate the role of multiple targets of workplace affective commitment as potential predictors of efficient emotional labor by employees and by examining how these different strategies predict burnout components.

Participants
The study was conducted in 14 service organizations located in the province of Quebec, Canada. Data were collected in winter 2009 from voluntary French-speaking service workers interacting directly with customers on a regular basis. All participants received the questionnaire from their managers and returned it once completed to the research team in a sealed envelope within a three weeks period. Of the 1164 questionnaires distributed, 370 were returned completed (response rate of 31.8%). Of these participants, 44% are from organizations offering specialized services (e.g. optometry services), 34.9% are from grocery stores, 10% are from retailers, 8.1% are from call centers, and 3% are from the hotel industry. Overall, 76.5% of the participants were females, 66.8% were full-time employees, 72.8% were frontline employees and 27.2% held a managerial position. In the sample, 13.7 % of participants were less than 20 years old, 30.7% were between 20 and 29 years old, 21.8% were between 30 and 39 years old, 19.0% were between 40 and 49 years old, and 14.8% were over 50 years old. In addition, 20.3% of the employees had been working for their organization for less than 1 year, 17.5% between 1 year and 2 years, 24.5% between 2 and 5 years, 29.3% between 5 and 15 years, and 8.5% had been working for their organization for more than 15 years.

Controls
Many variables were assessed as potential controls to ensure obtaining results that could be generalized beyond the specific characteristics of the present sample: gender (coded 0 for males and 1 for females), age (coded on a 1 to 6 scale), marital status (coded 0 for participants not in a relationship, and 1 for participants in a relationship), number of children (categories ranging from 0 to 4 or more), socio-economic level (including income and education level; coded on 0 a to 8 scale), job type (coded 0 for part-time and 1 for full-time), job title (coded 0 for frontline employees and 1 for employees holding a managerial position), and tenure (coded on a 1 to 7 scale). Two additional control variables that could potentially bias the observed relationships (Brotheridge & Grandey, 2002;Diefendroff et al., 2005;Grandey, Fisk, & Steiner, 2005), were also considered. First, social desirability was evaluated with the short version of Paulhus ' (1984) inventory, validated in French by Sabourin et al. (1989) and shortened by Frenette, Valois, Lussier, Sabourin etVilleneuve (2000;α=.71; 18 items, e.g. "I never regret my decisions"). Second, general affectivity was measured by a back-translated French version of the Positive Affectivity-Negative Affectivity Schedule (Watson, Clark, & Tellegen, 1988; α=.78 for both 10-item scales; 10 items, e.g. "Interested" or "Irritable"). Given the large number of controls, preliminary analyses will be used to select an appropriate, reduced set of controls for each analysis (results available upon request).

Emotional labor
Emotional labor was measured by a back-translated French version of the Emotional Labor Strategy Inventory (Diefendorff et al., 2005) including: surface acting (α=.88; 7 items, e.g. "I put on an act in order to deal with customers in an appropriate way"), deep acting (α=.75; 4 items, e.g. "I work hard to feel the emotions that I need to show to customers") and expression of naturally felt emotions (α=.72; 3 items, e.g. "The emotions I express to customers are genuine").

Workplace affective commitment
The short version of the Workplace Affective Commitment Multidimensional Questionnaire (WACMQ; Morin, Courcy, Madore, Loiselle, & Desclos, 2006;Morin et al., 2009Morin et al., , 2011) and a back-translated version of the Commitment to Display Rules questionnaire (Gosserand & Diefendorff, 2005) were used to measure workplace affective commitment. Eight targets of commitment were assessed: organization (α=.88; 3 items, e.g. "I am proud to say that I work for this organization"), supervisor (α=.85; 3 items, e.g. "I like the values my immediate supervisor conveys"), colleagues (α=.87; 3 items, e.g. "I'm happy to work with my coworkers"), work itself (α=.80; 3 items, e.g. "Work is a priority in my life"), job (α=.92; 3 items, e.g. "I find the tasks I perform in my current position stimulating"), career (α=.70; 3 items, e.g. "It is important for me to move up through the ranks or obtain promotions"), customers (α=.87; 3 items, e.g. "Delivering quality products and/or services to customers is a major source of satisfaction for me"), and display rules (α=.78; 5 items, e.g. "I think displaying the organizationally desired emotions on the job is a good goal to shoot for").

Analytical Strategy
Four multivariate outliers were excluded prior to the analyses. Preliminary exploratory factor analyses (maximum likelihood, oblique rotation), confirmed the factorial structure of each variable studied. For each burnout dimension, simple regressions were first conducted with each predictor or mediator. Then, predictors and mediators were entered together in series of multivariate regressions. Similar sets of regressions were used to predict emotional labor strategies. Inspection of the results showed that multicollinearity was not an issue. To verify the mediation hypothesis, the three mediators were simultaneously entered, resulting in 24 specific indirect effects tested for each burnout dimension (i.e. eight affective commitment targets mediated the three emotional labor strategies). To estimate whether mediated/indirect effects (i.e. the product of the a and b relations showed in Figure 1) were significant, bias-corrected bootstrap 95% confidence intervals were computed from 5000 bootstrap samples (MacKinnon, Lockwood, & Williams, 2004;Preacher & Hayes, 2008). Confidence intervals indicate significant mediation when they exclude zero. Table 2)

Emotional exhaustion
Preliminary analyses resulted in the selection of three control variables: age, positive affectivity, and negative affectivity. When considered independently, all workplace affective commitment targets, besides career, represent significant negative predictors of emotional exhaustion. Yet, when considered simultaneously, even with control variables and emotional labor strategies entered in the equation, workplace affective commitment towards the supervisor and the job are the only targets that remain significantly related to emotional exhaustion. Similarly, when considered independently, all emotional labor strategies are significantly related to emotional exhaustion. Yet, when all entered in the same model, only surface acting and naturally felt emotions remain significant predictors of emotional exhaustion, although only surface acting remains significant after adding control variables. Interestingly, when workplace affective commitment targets are added to the equation, surface acting becomes non-significant and deep acting becomes significant again. This final model explains 23% of emotional exhaustion's variance (R 2 aj ).

Cynicism
Preliminary analyses resulted in the selection of six control variables: gender, age, marital status, job title, social desirability, and positive affectivity. Each workplace affective commitment target independently and negatively predicts cynicism. When considered simultaneously, even with control variables and emotional labor strategies entered in the equation, workplace affective commitment towards the organization, the job and the display rules are the only targets still predicting cynicism. The three emotional labor strategies, when considered independently, are significantly related to cynicism. Nevertheless, when they are considered conjointly, the only remaining significant predictors of cynicism are surface acting and naturally felt emotions. However, only the surface acting strategy is still significant once control variables are included, and remains significant after the addition of workplace affective commitment targets to the equation. Deep acting also becomes significant again in this final model. The resulting model explains 33% of cynicism's variance (R 2 aj ).

Feelings of professional inefficacy
Preliminary analyses resulted in the selection of four control variables: age, job type, social desirability, and positive affectivity. Considered independently, each target of workplace affective commitment negatively predicts feelings of professional inefficacy. Yet, when entered simultaneously, workplace affective commitment towards the job and the display rules are the two targets that still predict significantly feelings of professional inefficacy. Moreover, when controls and emotional labor strategies are included, colleagues and career become significant once again. Results from the simple regressions indicate that surface acting and naturally felt emotions are significantly related to feelings of professional inefficacy. These strategies remain significant when the emotional labor strategies are included. However, only naturally felt emotions remains significant beyond the addition of control variables and workplace affective commitment targets. The resulting model explains 36% of feelings of professional inefficacy's variance (R 2 aj ). Table 3)

Surface acting
Preliminary analyses resulted in the selection of two control variables: gender and social desirability. Taken separately, the majority of the workplace affective commitment targets (except for work and career) negatively predict surface acting. When these targets are simultaneously considered, workplace affective commitment towards the career becomes significant, whereas only two (customers and display rules) of the previously significant targets remain significant. These variables remain significant when the control variables are included, and workplace affective commitment towards colleagues also shows up as a significant predictor of surface acting. The resulting model explains 20% of surface acting's variance (R 2 aj ).

Deep acting
Preliminary analyses resulted in the selection of social desirability as the single control. No target of workplace affective commitment separately predicts deep acting. Yet, when they are considered simultaneously, workplace affective commitment towards the supervisor becomes significant and remains so with controls. The final model explains 3% of deep acting's variance (R 2 aj ).

Naturally felt emotions
Preliminary analyses resulted in the selection of four control variables: age, job title, social desirability, and positive affectivity. Taken one by one, the majority of workplace affective commitment targets, besides career, positively predict naturally felt emotions. When these targets are considered simultaneously, only workplace affective commitment towards the customers remains significant. Adding the controls to the equation does not change this result and reveal a significant relationship between naturally felt emotions and workplace affective commitment towards the organization. The resulting model explains 24% of naturally felt emotions' variance (R 2 aj ). Table 4)

Mediation Analyses (see
On the basis of the previous analyses, it appears that few workplace affective commitment targets actually predict both a specific burnout component (dependent variable) and at least one emotional labor strategy, and thus appears to represent potential mediators. In fact, these preliminary regressions suggest that the indirect effects of workplace affective commitment towards the supervisor on emotional exhaustion, as mediated by emotional labor strategies, should be significant. Preliminary results also suggest that the indirect effects of workplace affective commitment towards the organization and display rules on cynicism, as mediated by emotional labor strategies, should reach significance. Moreover, preliminary analyses suggest that the indirect effects of workplace affective commitment towards colleagues and display rules on feelings of professional inefficacy, as mediated by emotional labor strategies, should be significant. However, these expectations could be misleading since it is also possible that non significant direct relation between the predictor and the outcome become significant in the context of the full mediation models, and mediation can occur whether or not there is a statistically significant effect of the independent variable on the dependent variable (MacKinnon, 2008). This phenomenon have been referred to in many ways in the scientific literature (e.g. suppressor effect) and generally occur when the effect of the predictor on the outcome and on the mediator are of opposite signs. Considering these observations and the number of mediating effects tested, only the significant indirect effects are presented in Table 4.

Emotional exhaustion
Regarding emotional exhaustion, a single indirect effect appeared significant: the one involving workplace affective commitment towards colleagues and surface acting. The indirect effect involving commitment to the supervisor proved to be non-significant.

Cynicism
Six out of eight workplace affective commitment targets (organization, supervisor, colleagues, customers, job, and display rules) showed a significant indirect effects on cynicism, as transmitted through surface acting, whereas no other emotional labor strategies mediated the effects of workplace affective commitment on cynicism. Among these targets, organization and display rules' indirect effects on cynicism confirmed prior expectations. The remaining effects, because they turned non-significant in preliminary analyses (see Table 2), disappeared when controlling for other targets, as confirmed by additional analyses (not reported here). However, these effects support the fact that affective commitment generally tends to be mediated by surface acting in the prediction of cynicism.

Feelings of professional inefficacy
In line with the results observed for cynicism, most indirect effects of workplace affective commitment targets on feelings of professional inefficacy were mediated by naturally felt emotions, while the mediating effects of other emotional labor strategies did not reach significance. Indeed, as expected, workplace affective commitment towards colleagues and display rules were indirectly related to professional inefficacy through naturally felt emotions. Furthermore, workplace affective commitment towards supervisor, organization, work, and job followed the same mediating pattern. These additional indirect effects were unexpected, since they came out as non-significant in the preliminary analyses (see Table 2), and could be due to the fact that additional targets of workplace affective commitment were not controlled in the present analyses. Additional analyses not reported here corroborate this interpretation. Still, these findings corroborate the fact that the naturally felt emotions strategy mediates the relations between workplace affective commitment and feelings of professional inefficacy.

Discussion
The main objective of the present study was to investigate the mediating effects of emotional labor strategies on the relationships between multiple targets of workplace affective commitment and burnout dimensions. The main findings were that indirect effects involving emotional exhaustion were almost absent, that surface acting was the only significant mediator of the relations between workplace affective commitment and cynicism and that naturally felt emotions was the only mediator of the relations between workplace affective commitment and feelings of professional inefficacy.
Preliminary analyses revealed that workplace affective commitment targets, particularly the supervisor and the job, tend to be negatively related to emotional exhaustion, which confirms previous findings (e.g. Boyas & Wind, 2010;Lee & Ashforth, 1996;Turmel & Morin, 2008). According to Lewin's (1943) force field theory, the forces that are more proximal to individuals should exert a greater influence on their behaviors. Similarly, social support theory (Maslach et al., 2001;van Dierendonck, Schaufeli, & Buunk, 1998) argues that employees who can rely on social and personal resources, such as the support from their supervisor, can deal more easily with the environment's emotional demands, which would reduce emotional exhaustion (van Dierendonck et al., 1998). In accordance, empirical studies have found positive relationships between such supportive relationships between employees and supervisors and workplace affective commitment (Eisenberger et al., 2010;Stinglhamber & Vandenberghe, 2003). In line with these assertions, the supervisor is usually an important proximal entity in service organizations, as well as an important source of social support. In service organizations, supervisors might therefore play a key role in emotional exhaustion's development among subordinates. Hence, the supervisor, as a target of workplace affective commitment, is most likely to substantially influence employees' emotional displays.
In fact, the supervisor was the only target of workplace affective commitment predicting both emotional exhaustion and an emotional labor strategy (i.e. deep acting) in preliminary analyses. Yet, the indirect/mediated effect of workplace affective commitment towards the supervisor on emotional exhaustion was not significant. Therefore, it might be argued that the explicative role of emotional labor is not relevant or that it is offset by the stronger direct relationship between workplace affective commitment towards the supervisor and emotional exhaustion. Additionally, the only emotional labor strategy predicted by workplace affective commitment towards the supervisor was deep acting, which is known to show little or no relationships with emotional exhaustion (e.g. Brotheridge & Grandey, 2002;Brotheridge & Lee, 2003;Goldberg & Grandey, 2007;Judge et al., 2009;Martinez-Iñigo et al., 2007;Näring et al., 2006;Totterdell & Holman, 2003). In accordance, the prediction of emotional exhaustion by deep acting in the present study appeared positive, but weak.
Besides affective commitment to the supervisor, employees that are affectively committed to their job are more likely to appreciate their tasks, to perform well and to experience positive feelings as a result of doing their job (Turmel & Morin, 2008;Vandenberghe, 2009). Thereby, affective commitment may prevent burnout and, more specifically, emotional exhaustion, which is presumed to be the first component to appear in burnout development (Maslach et al., 2001). This finding suggests that employees who identify with their jobs may gain from it by enhanced positive emotional states.
Subsequently, mediation analyses revealed the existence of a single indirect/mediated effect of workplace affective commitment towards colleagues on emotional exhaustion, as mediated by surface acting. Interestingly, colleagues represent another proximal target of commitment (Lewin, 1943). This finding was unexpected, since workplace affective commitment towards colleagues was not directly related to emotional exhaustion. More precisely, employees who are affectively committed towards their colleagues appeared to rely less on surface acting in order to cope with the emotional demands of their work, a strategy that has been shown to increase emotional exhaustion. Surface acting theoretically increases emotional dissonance by creating a gap between felt and expressed emotions, inducing feelings of self-alienation (Erickson & Wharton, 1997;Parkinson, 1991;Sutton, 1991). This finding suggests that the management of emotional demands could be facilitated by support from colleagues and social integration at work (just as it was previously suggested for the supervisor; see Luchman & Gonzalez- Moráles, 2010;van Dierendonck et al., 1998), which would reduce the use of surface acting among service employees and in turn its deleterious effects on emotional exhaustion.
Results from the present study also revealed that workplace affective commitment targets were almost all directly related to cynicism. Specifically, the observed negative relations between workplace affective commitment and cynicism converge with the results from previous studies (e.g. Abraham, 2000;Boyas & Wind, 2010;Lee & Ashforth, 1996;Turmel & Morin, 2008) and corroborate the conceptual antagonism between these constructs. Indeed, workplace affective commitment is defined as a "(…) force that binds an individual to a course of action of relevance to one or more targets" (Meyer & Herscovitch, 2001, p. 299), while cynicism is defined as the emergence of a cognitive distance between the employee and their workplace-related experiences with these same targets (Maslach et al., 2001). Since employees cannot be simultaneously bound and detached from their job and its different aspects, the negative relations between those elements were predictable.
However, this obvious finding is further refined by mediation analyses showing that surface acting plays an important role in these relationships. These results show that the more employees are affectively committed towards a variety of targets, the less they tend to use the surface acting strategy, a strategy related to increased levels of cynicism. As mentioned previously, the use of the surface acting strategy as a mean to manage emotions at work is generally viewed as detrimental for the employee, but also for the employee's social interactions, since the resulting state of emotional dissonance can be unfavourably perceived by others (Erickson & Wharton, 1997;Grandey, 2000;Parkinson, 1991;Sutton, 1991;Totterdell & Holman, 2003). Indeed, as a result of surface acting, others may respond adversely to employees' inauthentic display of emotions, thus creating a feedback loop that may in turn increase the employees' cognitive detachment, or cynicism (Côté, 2005). In itself, surface acting may therefore contribute to the cognitive distance inherent in cynicism by distorting the fundamental signal function of emotional reactions, which intrinsically help individuals to make sense of situations and to stay connected to others (Adelmann, 1995;Ashforth & Humphrey, 1993;Thoits, 1989). In this context, it can be argued that the energizing, motivating, and binding force inherent to workplace affective commitment (Meyer & Allen, 1991;Meyer et al., 2004;Meyer & Herscovitch, 2001) limits the expression of behaviors that are likely to be prejudicial to the targets that employees are emotionally attached to and to encourage employees sense of connectedness at work. This finding implies that, in order to promote warm and effective service encounters, managers should try to support and sustain affective commitment in the workplace.
In line with the results observed for cynicism, workplace affective commitment targets were almost all directly related to employees' feelings of professional inefficacy, supporting the results from previous studies (e.g. Boyas & Wind, 2010;Lee & Ashforth, 1996;Turmel & Morin, 2008). Indeed, when employees are affectively committed in their workplace, they will tend to invest more energy at work (Meyer & Allen, 1991;Meyer et al., 2004;Meyer & Herscovitch, 2001). This investment may ultimately lead to better performances. In fact, positive relations between several commitment targets and performance have already been found in other studies (Bentein, Stinglhamber, & Vandenberghe, 2000;Meyer et al., 2002;Vandenberghe, 2003). Since feelings of professional inefficacy depends on the evaluation that one makes of his own achievements at work (Maslach, 2001;Maslach et al., 2001), this increased performance may lead employees to legitimately consider themselves as more effective.
Moreover, partial mediation effects transmitted through naturally felt emotions were observed for six out of eight workplace affective commitment targets in their relation to employees' feelings of professional inefficacy. These results indicate that workplace affective commitment decreases employees' feelings of professional inefficacy, partly by increasing their tendency to express their naturally felt emotions as a way to meet organizational goals. Contrary to surface acting, the expression of naturally felt emotions is expected to be beneficial as it preserves emotional consonance (Diefendroff et al., 2005;Zapf, 2002). As discussed before, some authors argue that the negative consequences of emotional labor mostly emerge from employees' experiences of emotional dissonance (Erickson & Wharton, 1997;Hochschild, 1983;Parkinson, 1991;Sutton, 1991). In this way, service workers who genuinely experience and express expected emotions are likely to be emotionally consonant. From an interpersonal point of view, these authentic emotional displays are likely to appear more sincere to others, thus creating another reinforcing feedback loop, sustaining employees' feelings of professional efficacy (Côté, 2005). The expression of naturally felt emotions may thus represent a self-enhancing process (Pugliesi, 1999;also see Sheppard, Malone, & Sweeny, 2008), but may also improve the actual efficacy of the employee and, as a result, contribute to organizational efficiency. Still, the fact that service employees cannot always truly experience positive emotions underneath the requested positive emotional displays of their job suggests that there might be a limit to naturally felt emotions' positive outcomes. Future studies should attempt to disentangle these issues by considering the valence of the naturally expressed emotion as a potential moderator.
Finally, it is worth noting that no emotional labor strategies other than surface acting mediated the relationships between commitments and cynicism and that only naturally felt emotions mediated the relationships between commitments and felt professional inefficacy. Thus, commitments may be related to employees' well-being through two complementary mechanisms: First by increasing authentic emotional displays and felt professional efficacy, and second by lowering inauthentic emotional displays and cynicism. Clearly, these results confirm that emotional labor strategies have a differentiated impact on burnout components (Ashkanasy et al., 2002;Goldberg & Grandey, 2007). Besides, the multiple mediator model used in this study implies that each specific indirect/mediated effect has to be significant beyond other mediators' effects to reach significance (Preacher & Hayes, 2008). Thus, it typically leads to a smaller number of significant mediators, especially when the mediators are correlated, as it is the case in the present study (MacKinnon et al., 2004). In this context, if a potential mediator does not mediate a relation but another does, it increases the likelihood of it being the real or most important mediator (MacKinnon et al., 2004), reinforcing the observed results.
The cross-sectional design of the present study seriously limits the interpretation of the results by precluding any kind of causal inference regarding the directionality of the observed direct and indirect effects. Indeed, there may be several equivalent models to this multiple mediator model that could explain the observed relations equally well (MacKinnon et al., 2004). Longitudinal analyses would represent an interesting follow up to the present study. In addition, participation in the present study was voluntary, and resulted in a low participation rate. This might have biased the results, since the employees who completed the questionnaire may have initially been more committed towards their organization than the others. Moreover, the sample was obtained from service organisations and presented specific characteristics; it was indeed mostly composed of young females with few years of tenure, characteristics which may have influenced the results. Finally, given the exploratory nature of the present study, multiple relations were investigated, thus inflating the risk of committing type 1 errors and capitalizing on chance against which replication is the only safeguard. Therefore, reproducing this study with more diversified samples would be important to ensure that the results can be generalized and do not reflect capitalization on chance.
In addition to taking these limitations into account, further research should also take two additional issues in considerations. First, despite its potential key role, studies that examined naturally felt emotions are currently scarce (Diefendroff et al., 2005). In the present study, this strategy brought up nuances to the explanation of the relationship between workplace affective commitment and burnout. Still, more studies would be needed to fully understand its role and to establish if there are actually limits to its beneficial effects. Secondly, the balance between psychological congruence (emotional consonance) and relational rewards (positive feedback) as a mean to explain the indirect effects of emotional labor strategies on the relationships between workplace affective commitment targets and burnout components seems to offer interesting avenues of explanations deserving additional consideration in the study of emotions in organizations (Martínez-Iñigo et al., 2007).  Behavior, 14, 177-190. http://dx.doi.org/10.1002/job.4030140207 Hartline, M., & Ferrell, O.C. (1993. Service quality implementation: The effects of organizational socialization and managerial actions on customer-contact employee behaviors . Cambridge, MA: Marketing Science Institute. Henkoff, R., & Sample, A. (1994). Finding, training, and keeping the best service workers. Fortune, October 3, pp.110-116. Hobföll, S.E. (1989). Conservation of resources: A new attempt at conceptualizing stress. American Psychologist, 44, 513-524. http://dx.doi.org/10.1037/0003-066X.44.3.513 Hobföll, S.E. (2002. Social and psychological resources and adaptation. Journal of General Psychology, 6, 307-324. http://dx.doi.org/10.1037/1089-2680.6.4.307 Hochschild, A.R. (1983. The managed heart: Commercialization of human feeling. Los Angeles, CA: University of California Press.