Service Climate and Display of Employees' Positive Emotions: the Mediating Role of Burnout and Engagement in Services

1 This article aims to test a model linking service climate to the frequency of expression of positive emotions by frontline employees. We propose that burnout and engagement at work mediate the relationship between service climate and the expression of positive emotions. Service climate impacts negatively on burnout and positively on engagement; in turn, burnout and engagement are significantly related to the frequency of expression of positive emotions. This model was tested both at the individual and work-unit levels. In addition, it was compared with a direct model that proposes an additional direct link from service climate to frontline employees' positive emotions. Models were tested through structural equation modeling. The sample consisted of 508 frontline employees working in 151 work-groups. At the individual level, results confirmed the mediating role of engagement in the relationship between service climate and the expression of frontline employees' positive emotions. At the work-unit level, both burnout and engagement mediated the relationship between service climate and positive emotions. Engagement was the most important direct predictor of the display of positive emotions, both at the individual and work-unit levels. During the last few decades, scholars and practitioners have shown a growing interest in quality. It is assumed that quality and customer satisfaction assure organizations'

This effect is especially important in services, given the social interaction that often exists between customers and employees.In addition to the increasing workload in the service sector (Peeters & Le Blanc, 2001), frontline employees are required to express socially desirable emotions as a role demand (Hochschild 1979(Hochschild , 1983)).A critical point of this emotional labor, directed toward producing customer satisfaction, is the display of positive emotions by frontline employees in their relationship with customers (e.g., Grandey, 2003).The expression of positive emotions has been shown to be closely associated with customers' perceptions of service quality, customer satisfaction, and loyalty (Pugh, 2001;Rogelberg, Barnes-Farrell, & Creamer, 1999;Tsai & Huang, 2002).Taking this observation into account, there is a need to study the factors that explain the frequency of expression of positive emotions among frontline employees, directed toward their customers.This facet of service quality is more difficult to control and predict than others of a more technical nature.However, we suggest that organizations are able to increase the display of employees' positive emotions by considering service climate, burnout, and engagement in service work.
Service climate refers to employees' shared perceptions of the practices, procedures, and behaviors concerning quality customer service (Schneider, 1990).Traditionally, researchers have focused on the relationship between service climate and customer evaluations of services.The most important finding from the so-called "linkage research" (Wiley, 1996) points out that climate experiences reported by employees were validated by customer experiences.Many researchers have observed significant relationships between service climate and positive customer perceptions of service (Burke, Rupinski, Dunlap, & Davison, 1996;Dietz, Pugh, & Wiley, 2004;Martínez-Tur et al., in press;Paradise-Tornow, 1991;Salanova, Agut, & Peiró, 2005;Schneider, Salvaggio, & Subirats,2002;Schneider & Bowen, 1992;Schneider, White, & Paul, 1998;Tornow & Wiley, 1991).Nevertheless, as Schneider and colleagues (Schneider, Ehrhart, Mayer, Saltz, & Niles-Jolly, 2005) have noted, little is known about the process through which service climate impacts employee reactions (e.g.expression of positive emotions) responsible for positive customer evaluations.
With this in mind, this investigation proposes a mediated model where service climate impacts on well-being at work, and well-being, in turn, influences the frequency of expression of positive emotions among frontline service employees in their relationships with customers.Service climate is considered as a precursor of well-being at work because it informs frontline employees that the organization is serving the customers well, and it supports efforts related to attending to customers (Martin, 2008).In addition, well-being at work facilitates employee investment in efforts related to positive emotions (Lam, Huang & Janssen, 2010).
The present study contributes to the literature on frontline service employees' display of positive emotions in three ways.First, we offer an integrative approach that synthesizes concepts from three different literatures (service climate, well-being at work, and emotional labor) in understanding the process through which employees display positive emotions.This integrative strategy describes an additional step in the investigation of the process linking service climate to outcomes (see Schneider et al., 2005).Previous research efforts observed direct links from service climate to display of positive emotions (e.g., Lam et al., 2010), but the potential role of intermediate constructs was not tested.In the current study, we examine the role of well-being as a mediator.This approach provides the opportunity to better understand the process.Second, we consider the positive and negative sides of well-being at work simultaneously: burnout and engagement.Traditionally, the literature about positive emotions has focused the attention on burnout, indicating that employees with burnout tend to reduce emotional labor efforts as a way to protect their resources (Lam et al., 2010).The consideration of engagement allows a complementary approach where the direct existence of well-being at work (and not its absence) is also measured.In other words, it has been well-established that lack of well-being (burnout) reduces the display of positive emotions among frontline employees (Lam et al., 2010).However, there is a need to test whether the positive promotion of well-being (engagement) is able to increase employees' emotional labor efforts.Finally, we test the model at both the individual and work-unit levels.
Service climate and its effects have been examined at the individual (e.g., Martin, 2008;Yoon, Beatty & Suh, 2001) and work-unit (Salanova et al., 2005;Schneider et al., 2005) levels.Psychological service climate (individual level) refers to individual differences in employees' perceptions of support in the attention to customers.In contrast, organizational service climate (collective level) focuses on a construct level beyond the individual.The joint consideration of individual and work-unit levels allows us to examine differences and similarities in the process linking service climate to positive emotions in individuals and work-units.The tendency to organize work using teams and work-units makes it necessary to focus the attention not only on the individual but also at the work-unit level.The consideration of different levels permits a better understanding of organizations, and it is an indicator of maturity in research (Kozlowski & Klein, 2000).

Concepts
As indicated above, our model involves three different constructs: service climate, well-being at work (burnout and engagement), and display of positive emotions.
In this section, we present a short definition of these concepts.
Service climate.Service climate has been defined as employees' shared perceptions regarding the practices, procedures, and kinds of behaviors that get rewarded and supported in a particular organization (Schneider, 1990).Because multiple climates often exist simultaneously within a single organization, climate is best regarded as a specific construct that has a referent -a climate must be a climate for something, such as service, support, innovation, or safety (Moliner, Martínez-Tur, Peiró, Ramos, & Cropanzano, 2005;Schneider, Gunnarson, & Niles-Jolly, 1994).
When a facet of the organization is important or critical, a specific climate is created.In the service sector, service climate is defined as employees' perceptions of the practices, procedures, and behaviors that get rewarded, supported, and are expected by the organization in the areas of customer service and service quality.
Service climate reflects perceptions that customer service is important to the organization and that management supports service quality efforts (Schneider et al., 1998).Some authors have focused the attention on psychological service climate, measuring individual differences in perceptions (Martin, 2008;Yoon et al., 2001).From this perspective, the focus is on individual interpretations of stimuli related to the importance the organization attributes to service quality efforts and customer satisfaction.It is measured in terms of perceptions of service climate that are meaningful to individuals (see Yoon et al., 2001, p. 502).Thus, the level of construct is the individual frontline employee.Nevertheless, and taking into account that constructs in organizations often behave at different levels at the same time (Bliese, Chan, & Ployhart, 2007), other scholars have opted for a definition of service climate with a level beyond the individual one (e.g., work-unit, branch): organizational service climate.It is assumed that employees pertaining to the same work-unit (e.g., Hui, Chiu, Yu, Cheng & Tse, 2007) or branch (e.g., Schneider et al, 1998) are able to develop consensual views about the importance organizations attribute to service quality and the degree to which efforts related to customer service are supported.Shared perceptions are possible because employees are subjected to identical processes and structures, and they interact and share experiences in their daily work activities (Martínez-Tur, Tordera, Peiró, & Potocnik, 2011).
Burnout and Engagement.Affective well-being is conceptualized by Warr (1987) as one of five components of mental health.Within the well-being component, he differentiated between context-free and job-related well-being.The latter includes positive experiences, such as job satisfaction, as well as negative ones, such as job related tension, job-related depression and job-related burnout.In this study we focus on job-related well-being, taking into account its negative and positive sides.Burnout has been considered a critical indicator of job-related well-being (Maslach, Schaufeli, & Leiter, 2001;Moliner et al., 2005;Tummers, Landeweerd, & Merode, 2002), describing a negative experience related to work.
Burnout was defined originally as a three dimensional syndrome including exhaustion (e.g., the draining of emotional resources because of demanding interpersonal contacts with others), cynicism (e.g.negative, callous, and cynical attitude toward the recipients of one's care or services), and lack of professional efficacy (e.g. the tendency to evaluate one's work with recipients negatively) (Schaufeli, Salanova, González-Romá & Bakker,2002).Whereas some authors have found support for this three-component burnout model (Maslach & Jackson, 1981, 1986), other scholars have obtained support for a two-component burnout model (Green, Walkey, & Taylor, 1991;Schaufeli et al., 2002) that includes only the dimensions of exhaustion and cynicism as the so-called "core of burnout".Lack of personal efficacy is considered as a personality trait rather than a specific dimension of burnout (Shirom, 2003).Initially, burnout was measured with the Maslach-Burnout Inventory (Maslach & Jackson, 1981), but when the burnout concept broadened to all kinds of employees, its measurement changed through an MBI adaptation called the MBI-General Survey (Schaufeli, Leiter, Maslash, & Jackson, 1996), which measures the same factors without referring to other people as recipients of work.
Developments in burnout research have also focused on its positive side, job engagement.Burnout is rephrased as an erosion of engagement with the job (Schaufeli, et al., 2002), which is congruent with the Diener (1984) proposition that research should include positive measures to overcome a limited view of well-being that assumes it stems mostly from the absence of negative symptoms.
The differentiation between burnout and engagement reflects an emerging trend toward a positive psychology that focuses on human strengths and optimal functioning rather than on weaknesses and malfunctioning (Seligman & Csikszentmihalyi, 2000).This approach is consistent with the definition of health by the World Health Organization (1946): "Health is a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity".This definition opens the door to the elaboration of concepts where the focus is on the positive side of well-being (e.g., engagement) and on the promotion of health beyond simply reducing problems.
Research studies include engagement as a relevant factor because interventions may be more effective if they are framed in terms of building engagement rather than just reducing burnout (Maslach, 2003).Using Role Theory, Khan and Byosiere (1992) define "psychological presence" or "to be fully there" as an experiential state that accompanies "personally engaging behaviors" which involve the channeling of personal energies into physical, cognitive and emotional labor.This approach previews the Maslach and Leiter (1997) engagement concept characterized by energy, involvement and efficacy.Engaged employees have a sense of an energetic and effective connection with their work activities, and they see themselves as able to deal completely with job demands.Engagement refers to a persistent and pervasive affective-cognitive state that is not focused on any particular object, event, individual, or behavior (Schaufeli et al., 2002).Similar to the "core of burnout" (with exhaustion and cynicism), a "core of engagement" has been proposed, with the antithetical dimensions of vigor (characterized by high levels of energy and mental resilience while working, and by the willingness and ability to invest effort in one's work) and dedication (characterized by a sense of significance, enthusiasm, inspiration, pride and challenge) (González-Romá, Schaufeli, Bakker, & Lloret, 2006;Schaufeli et al., 2002).Engagement was initially measured by the opposite pattern of scores on the dimensions of burnout (Maslach & Leiter, 1997).
However, assessing the presence of something is more definite than assessing the absence of its opposite.In other words, the absence of burnout does not necessarily lead to engagement.Therefore, although there is a debate about the nature of the relationship between burnout and engagement (e.g., González-Romá et al., 2006), several scholars measure burnout and engagement separately with different instruments (González-Romá et al., 2006;Schaufeli et al., 2002).
Traditionally, burnout has been defined at the individual level of construct and analysis, with some exceptions (see Maslach et al., 2001).Accordingly, the focus is on individual differences in burnout experiences.However, burnout is also an emergent phenomenon beyond the individual.Bakker, Schaufeli, Sixma, and Bosveld (2001) demonstrated that burnout contagion exists among general practitioners in a hospital.In addition, Westman and Etzion (1995) observed burnout contagion in private personal relationships.The contagion of burnout offers a richer portrait of burnout because this mechanism is able to explain how people who interact within work-units develop shared experiences of burnout (Moliner et al., 2005).
As Edelwich and Brodsky (1980) pointed out: "If burnout only affected individuals in isolation, it would be far less important and far less devastating in its impact than it is.Burnout in human services agencies is like an infection in hospitals: it gets around.It spreads from clients to staff, from one staff member to another, and from staff back to clients.Perhaps it ought to be called staff infection".(p.25).Additional information about burnout contagion may be found in Bakker and Schaufeli (2000), Bakker, et al. (2001) and Bakker, Demerouti, and Schaufeli (2005).

Display of Employees' Positive Emotions. Display of employees' positive emotions
is defined as the extent to which employees show positive emotions in order to deliver quality experiences in service encounters (Lam et al., 2010, p. 369).In the service sector, it is expected that frontline employees will express cheerful and friendly emotions in order to increase customer satisfaction.Services tend to be produced and consumed simultaneously, making heavy interpersonal demands on employees who interact with a number of customers a day, each one with specific desires and behaviors.Co-production places employees and customers in proximity, both physically and psychologically (Ryan & Ployhart, 2003).Given this situation, frontline employees are required to display positive emotions, following the assumption that this emotional labor will impact positively on customer evaluations of service.In fact, previous empirical efforts demonstrated that the display of employees' positive emotions is significantly connected to customer satisfaction, quality perceptions and customer loyalty (Pugh, 2001;Rogelberg et al., 1999;Tsai & Huang, 2002).
The display of positive emotions is regulated by individuals.In other words, each frontline employee responds to this emotional labor requirement when interacting with customers.However, this regulation of positive emotions also has a collective nature.In service organizations, employees often work in teams or work-units, and in each work-unit employees tend to perform similar tasks, they have the same supervisor, and they interact with each other during their daily work (Moliner et al., 2005).Although members of work-units and teams differ in their experiences and interpretations of their work-environment and organizational life, a primary effect of groups is to generate consensus about the social reality (Bliese & Halverson, 1998).Job experiences and emotional states are influenced by other actors, generating convergence.Mechanisms for convergence in the display of emotions are both subconscious and conscious.Subconscious or involuntary processes that should be mentioned are: a) the automatic imitation of facial expressions, postures, movements, etc. leading to emotional convergence (Hatfield, Cacioppo, & Rapson, 1994); b) similarity in affective reactions associated with the performance of similar tasks (George, 1996); and c) through vicarious learning, employees may be experiencing emotions very similar to those of other actors or models (e.g., colleagues in the work-unit) (Kelly & Barsade, 2001).
In addition to these involuntary or subconscious processes, employees and organizations are able to generate emotional convergence through conscious mechanisms.We outline three main mechanisms.First, selection processes tend to accentuate uniformity of personality traits and emotional convergence (Watson, 2000), with the participation of leaders who select employees with certain defined personality traits and employees who are attracted to an organization because their personality traits are consistent with the working environment ("attractionselection-attrition", Schneider, 1987;Schneider, Goldstein & Smith, 1995).Second, there are important interpersonal processes such as social comparison and socialization.The appropriateness of their own emotions is evaluated by comparing them to the emotions of others (Festinger, 1954), thus increasing uniformity in the display of emotions (Sullins, 1991).Socialization also offers appropriate models for displaying emotions at work (Fisher, 1986).Third, the social structure establishes and reinforces rules about the desired emotions and the manner in which they should be displayed, pressing toward uniformity (Hochschild, 1990).
In sum, our concepts (service climate, burnout-engagement, and display of employees' positive emotions) exist at different construct levels in service organizations.Individuals differ in their interpretations of their work environment and experiences, but members of the same group, team, or work-unit are also able to develop consensual views and similar reactions.Accordingly, to explore the process from service climate to display of positive emotions, we consider both the individual and the work-unit levels of construct and analysis.

The mediated model
As we indicated above, we propose a model where well-being at work (burnout and engagement) mediates the link from service climate to display of employees' positive emotions directed toward customers (see Figure 1).The theoretical approach underlying the hypotheses of this model is the conservation of resources (COR) theory (Hobfoll, 1989(Hobfoll, , 1990)).

Figure 1. Mediated model
According to COR theory, employees are motivated to obtain resources to compensate for the depletion of resources associated with work demands.In the service sector, frontline employees are required to positively interact with customers in an effort to offer an excellent service.In other words, frontline employees take on an important emotional labor and invest resources in order to offer good service.Service climate informs frontline employees that service quality is important to the organization, and employee service efforts are supported and rewarded (Schneider et al., 1998;2002).These compensatory resources are clearly connected to burnout, in consonance with the COR theory.In fact, Wright and Cropanzano (1998) indicated the COR theory is especially valuable in understanding burnout.These authors remarked that burnout is more likely to occur when "there is an actual resource loss, a perceived threat of resource loss, a situation in which one's resources are inadequate to meet work demands, or when the anticipated returns are not obtained on an investment of resources"

Display of employees' positive emotions
Service Climate Engagement Burnout (Wright & Cropanzano, 1998, p. 487).Service climate is an important source of resources for frontline employees who meet work demands related to interacting with customers (Lam et al., 2010).It is reasonable to expect that frontline employees will feel engaged if they perceive that required efforts devoted to customer service are truly recognized, rewarded and supported.In contrast, burnout is more likely to occur if support associated with service climate is not present.Consistent with these ideas, Martin (2008) observed at the individual level of analysis that service climate acts as a significant resource for university staff, reducing job-induced tension and increasing job satisfaction.In our current integrative research study, we consider both individual and work-unit levels of construct and analysis, and burnout-engagement as outcomes of service climate.
Hence, we predict the following: Hypothesis 1: Psychological service climate is negatively related to burnout and positively related to engagement.
Hypothesis 2: Work-unit service climate is negatively related to burnout and positively related to engagement.
According to the mediated model, well-being at work is significantly related to the display of employees' positive emotions (see Figure 1).COR theory helps to understand the connection between burnout-engagement and display of employees' positive emotions.Burned-out employees do not have sufficient resources for emotional labor, and COR theory predicts that, under this situation of lack of resources, employees will develop strategies to protect or replenish their emotional resources (e.g., Halbesleben, 2006;Lee & Ashforth, 1996).Leiter (1991;1993) indicated that burned-out employees emphasize withdrawal coping mechanisms.Because one of the typical withdrawal strategies is performance reduction, Lam et al. (2010) proposed that burnout in frontline employees tends to cause them to reduce emotional efforts and, therefore, make less effort in displaying positive emotions.In contrast, it is reasonable to expect that engaged employees, who are characterized by energy and abundant emotional resources, will devote their efforts to displaying the required positive emotions directed toward customers.Therefore, we test the links from burnout and engagement to display of employees' positive emotions, both at the individual and workunit levels.
Hypothesis 3: At the individual level, employee burnout is negatively related to display of positive emotions, while employee engagement is positively related to display of positive emotions.
Hypothesis 4: At the work-unit level, burnout in work-units is negatively related to display of positive emotions, while engagement in work-units is positively related to display of positive emotions.
In summary, our mediated model proposes that support for service climate increases well-being at work (reducing burnout and increasing engagement).
Well-being at work (low burnout and high engagement), in turn, predicts an increase in the display of employees' positive emotions directed toward customers.Because these constructs exist at different levels in organizations, the model is proposed for both individual employees and work-teams in organizations.

Mediated versus direct model
In previous sections we have described arguments suggesting the mediating role of burnout and engagement in the relationship between service climate and the display of employees' positive emotions.However, there are also reasons to expect direct relationships between service climate and employees' positive emotions.In fact, the links between service climate and customer evaluations are based on the assumption that service climate increases the effort (e.g., in terms of expression of positive emotions) devoted to the attention of customers (Schneider et al., 1998) Burnout.To measure burnout, we adapted the Spanish Version (Schaufeli et al., 2002) of the Maslach-Burnout Inventory-General Survey (Schaufeli et al., 1996) to hotel employees.The instrument consisted of 10 items assessing exhaustion (5 items) and cynicism (5 items), tapping the core aspects of burnout.High scores on exhaustion and cynicism, ranging from 0 (never) to 6 (every day), are indicative of work burnout.Engagement.To measure engagement, we also adapted the Spanish version of the Utrecht Work Engagement Scale (UWES) assessing vigor (6 items) and dedication (5 items) (Schaufeli et al., 2002), in order to capture the core aspects of the construct.High scores on the scale, ranging from 0 (never) to 6 (every day), represent work engagement.
Employees' positive emotions.Six researchers from three universities in Spain were involved again in the elaboration of this measure for our projects.Based on the Emotion Work Scale by Zapf, Vogt, Seifert, Mertini, and Isic (1999), they analyzed the best options to represent positive emotions displayed by frontline employees in hotels.Three positive emotions were finally selected for the current study: We measured the frequency of expression of these positive emotions directed toward customers (3 items).More specifically, frontline employees indicated the degree to which they express these positive emotions to customers in their daily work.
Measured items ranged from 1 (Never) to 5 (Several times a day).
Sample items and Alpha Coefficients of each scale are shown in Table 1.In general terms, the magnitude of Alpha coefficients was satisfactory for all measures involved in this research study.

Data aggregation
We statistically justified aggregation of our measures because we focused our attention not only on the individual level of construct, but also on the work-unit level.Accordingly, we explored within-work unit agreement and reliability and between-work unit differences.The median values on the interrater agreement index r wg(J) (James, Demaree, & Wolf, 1984) ranged from .68 (burnout) to .80 (engagement).Because r wg(J) is above or near the .70cutoff value, agreement at the work-unit level was satisfactory (Dunlap, Burke, & Smith-Crowe, 2003).In addition, the intra-class correlations ICC (1) and ICC (2) were calculated (James, 1982).

Analysis
Both hypothesized mediated models and the alternative direct models were analyzed through Structural Equation Modeling (SEM) at the individual-level and team-level of analysis.Maximum likelihood method of estimation was used, and inputs for each analysis were the correlation matrix and asymptotic covariance matrix of the items.As no statistical test or critical values are available to deter-mine the adequacy of absolute indices, researchers also recommend the computation of relative goodness-of-fit (Bentler, 1990;Kelloway, 1998).Thus, absolute and relative goodness-of-fit indices were calculated.Absolute goodness-of-fit indices computed were the c² goodness-of-fit statistic, the Root Mean Square Error of Approximation (RMSEA), the goodness-of-fit index (GFI), and the adjusted goodness-of-fit index (AGFI).The relative goodness of fit indices calculated were: the Normed Fit Index (NFI), the Non-Normed Fit Index (NNFI), and the Comparative Fit Index (CFI).In addition, mediated and direct models were compared, examining differences between c² values and the Non-Normed Fit Index (NNFI) (Marsh, Balla, & McDonald, 1988).

Preliminary results
Mean ratings, standard deviations, and Pearson correlations among the variables used in this study are presented in Table 2. Service climate correlated significantly and negatively with burnout and positively with engagement, as expected.The negative correlation between burnout and engagement was significant.The correlation between service climate and display of positive emotions was positive and significant, but with less magnitude than other correlations.Finally, the positive correlation between engagement and display of positive emotions was significant, while the correlation between burnout and display of positive emotions was not significant.In sum, these preliminary results indicate that service climate has a direct relationship with the display of positive emotions.However, the structural equation modeling will allow us to test whether this relation remains significant when the different study variables are considered simultaneously.

Mediated vs. direct model at the individual level
Mediated and direct models are represented in Figure 3.With regard to service climate, H1 was confirmed: service climate was negatively related to burnout and positively related to engagement.In contrast, H3 was only partially supported.
Engagement was positively related to display of employees' positive emotions, as expected.However, the connection between burnout and display of positive emotions was not significant.In addition, the correlation between service climate and display of positive emotions was not significant.To test the proposed mediated model more accurately, we compared it with the direct model.Goodness-of-fit indices for both models were satisfactory and very similar (see Table 3), and they did not differ in statistical terms ∆χ 2 (1) = 1.29, p > .05).In fact, the difference between the two models in their NNFI values (0) was below .01,indicating that the difference between them is irrelevant (Marsh et al., 1988).All these results, in addition to the non-significant path from service climate to display of positive emotions, revealed that the mediated model is more parsimonious than the direct model, and it is able to describe the process more adequately.

Mediated vs. direct model at the work-unit level
Mediated and direct models at the work-unit level are represented in Figure 4.
With regard to service climate, H2 was confirmed: service climate was negatively related to burnout and positively related to engagement.In contrast, H4 was only partially supported.Engagement was positively related to display of employees' positive emotions, as expected.However, the connection between burnout and display of positive emotions presented the opposite pattern of that described in H4.Finally, the correlation between service climate and display of emotions was not significant.When the two models were compared, findings were very similar to those observed at the individual level.Goodness-of-fit indices for both models at the work-unit level were satisfactory and very similar (see Table 4), and they did not differ in statistical terms ∆χ 2 (1) = 3.65, p > .05).In fact, the difference between the two models in their NNFI values (0.004) was below 0.01, indicating that the difference between them is irrelevant (Marsh et al., 1988).All these results, in addition to the non-significant path from service climate to display of positive emotions, revealed that the mediated model is more parsimonious than the direct model, and it is able to describe the process more adequately.

Discussion
The Nevertheless, our model suggests that organizations can increase the display of employees' positive emotions by creating an adequate service climate and improving well-being at work.In consonance with the COR theory (Hobfoll, 1989;1990;Wright & Cropanzano, 1998), our results confirmed that service climate is an important source of resources for frontline employees in services, informing employees that their efforts in terms of service quality are rewarded and supported.Accordingly, and consistent with previous research studies (Martin, 2008), service climate reduces burnout and increases engagement.
Engagement, in turn, reflects an abundance of resources that employees, according to COR theory, are able to invest by displaying positive emotions directed toward customers.The role of burnout is secondary in understanding the display of positive emotions.Different implications of the mediated model are discussed below.
Previous research efforts pointed out that service climate is a direct precursor of the display of employees' positive emotions (Lam et al., 2010;Tsai & Huang, 2002).However, the mediating role of well-being at work was not examined in these previous efforts.In general, climate perceptions are not automatically translated into employee behavior; therefore, we considered that there was a need to unfold the process through which service climate influences employees' Our results also supported the differentiation between burnout and engagement as two independent but related concepts.Consistent with previous empirical evidence indicating that the two constructs do not have the same relationships with criteria (e.g., Moliner et al., 2008), the power to predict display of employees' positive emotions is different for burnout and engagement.More specifically, our data showed that the relationship between engagement and display of employees' positive emotions is strong and statistically significant, both at the individual and work-unit levels.This finding reinforces the trend towards positive psychology, demonstrating the utility of the incorporation of positive measures beyond the negative aspects massively reported by scholars (Seligman & Csikszentmihalyi, 2000).In other words, measuring the existence of well-being at work is more definitive than examining lack of burnout in an attempt to stimulate display of employees' positive emotions directed toward customers.
In fact, the relationship between burnout and display of positive emotions was not significant at the individual level, and this relation was counter-intuitive at the work-unit level, with a small, but positive and significant, link from burnout to display of positive emotions.A tentative explanation for this link could be related to the idea that burnout is required for the display of emotional labor (Goldberg & Grandey, 2007).Nevertheless, future research is needed to examine this alternative hypothesis.
Another contribution of our research study is the examination of the models both at the individual and work-unit levels.Given the complexity of organizations and the tendency to organize the work into teams and work-units, many constructs behave at different levels at the same time (Bliese et al., 2007).Thus, the consideration of different levels in organizations is considered an indicator of maturity in theory and research (Kozlowski & Klein, 2000).In our case, we confirmed that, in addition to individual differences, service climate, burnout, engagement and display of positive emotions can also be conceptualized at the work-unit level.
Similarities are important when individual and work-unit levels are compared.
For both individuals and work-units, service climate was confirmed as a relevant precursor of well-being at work (reducing burnout and increasing engagement).
In addition, engagement maintained a strong relationship with display of positive emotions by employees and work-units.
Our findings revealed practical implications.Specifically in service work, the existence of a service climate appears to positively impact employees' emotional behavior directed toward customers.This major finding, in addition to the positive impact of engagement on the display of positive emotions, suggests that enhancing service climate is a mechanism that can improve service encounters between employees and customers.Also, managers of services should be aware that critical indicators of performance (in terms of display of positive emotions in services) and engagement are not only compatible, but it is likely that engaged workers will display positive emotions directed toward customers.
The present results have several implications for future research, some of which are related to the limitations of this study.First, the cross-sectional nature of the data limits the study.Because all the data were collected at the same research time, alternative models cannot be specifically ruled out by these data (James, Mulaik, & Brett, 2006), and causal inferences should be interpreted with caution.For example, the counter-intuitive positive relation between burnout and display of positive emotions at the work-unit level could be better explored with longitudinal approaches.Although longitudinal studies reinforce the sequence from burnout to performance (Wright & Cropanzano, 1998), other alternatives indicating how positive emotional display provokes burnout are possible (e.g., Goldberg & Grandey, 2007).Thus, there is a need to conduct longitudinal research in order to better test causality in models integrating service climate, burnout, engagement and emotional labor indicators.Second, the sample in this research study only included hotel employee data, and, therefore, generalization of the results is somewhat limited.Testing the present model in other samples, including different types of actors (employees, supervisors), companies, and sectors, will provide good opportunities to explore the generalizability of the proposed models.Nevertheless, hotels have shared characteristics with other service organizations (e.g., co-production, direct contact between customers and employees), and it is possible to anticipate similar results in other types of services.Finally, in this study all data were collected through self-report surveys.The self-report bias warns us about the possibility of inflated relationships due to common method variance (Podsakoff et al., 2003;Podsakoff & Organ, 1986).The use of alternative measures of our constructs is recommended.For example, quality efforts could be evaluated by objective conditions (e.g., amount of money invested by the organization to improve quality).

Figure
Figure 4. Mediated and direct model at the work-unit level * p < .05;** p < .01;*** p < .001 main objective of this study was to propose and test a causal model integrating service climate, well-being at work and display of employees' positive emotions.This model proposes that well-being at work, measured as burnout and engagement, mediates the relationship between service climate and display of employees' positive emotions directed toward customers.Our results confirmed this mediated model both at the individual and work-unit levels.The interaction between employees and customers is the common place where service work happens and where customers evaluate service quality.The display of positive emotions by frontline employees is considered a critical indicator of performance in services.As Pugh (2001) demonstrated, positive emotions displayed by employees are closely related to customer positive affect and service quality, following a contagion process.Because frontline employees are ultimately responsible for the interaction with customers, the aforementioned facet of service quality (employees' positive emotions) is under their control.
Despite these limitations and the need for additional research, the present study contributes to the initial investigation of an important process, integrating the service climate, burnout, engagement and emotional labor literatures.Although frontline employees are ultimately responsible for the display of positive emotions toward customers, service organizations are able to increase the frequency of these O Clima de Serviço e a Demonstração das Emoções Positivas dos Empregados: O Papel Mediador de Burnout e Engagement nos Serviços Este artigo teve como objectivo testar um modelo que relaciona o clima de serviço com a frequência de expressão de emoções positivas por parte dos trabalhadores de frontline.Propõe-se que o burnout e o engagement no trabalho medeiam a relação entre o clima de serviço e a expressão de emoções positivas.O clima de serviço tem um impacto negativo sobre o burnout e positivo sobre o engagement; por sua vez, burnout e engagement estão significativamente relacionados com a frequência da expressão de emoções positivas.Este modelo foi testado quer ao nível individual quer ao nível das unidades de trabalho, tendo sido comparado com um modelo que propõe uma associação directa adicional entre o clima de serviço e as emoções positivas dos trabalhadores de frontline.Os modelos foram testados através de modelos de equações estruturais (SEM).A amostra foi composta por 508 trabalhadores de frontline, pertencentes a 151 grupos de trabalho.No nível individual, os resultados confirmaram o papel mediador do engagement na relação entre o clima de serviço e a expressão de emoções positivas por parte dos trabalhadores de frontline.Ao nível da unidade de trabalho, quer o burnout quer o engagement medeiam a relação entre o clima de serviço e as emoções positivas.O engagement revelou-se o mais importante preditor directo da expressão de emoções positivas, tanto no nível individual como na unidade de trabalho.PALAVRAS-CHAVE: burnout, emoções, engagement, trabalhadores de frontline, clima de serviço