European Journal of Social Sciences Volume 11, Number 4 (2009)

Emotional Labor and Psychological Distress: Testing the Mediatory Role of Work-Family Conflict
Jahanvash Karim CERGAM Institute dAdministration des Entreprises dAix-en-Provence Clos Guiot Puyricard BP 30063 13089 Aix en Provence cedex 2, France E-mail: j_vash@hotmail.com Tel: 33-643055276 Abstract The purpose of the current study was to examine the relationship between emotional labor, work-family conflict (WFC) and psychological distress among a sample of employees working in three public sector organizations in Pakistan. Data were collected on a structured questionnaire containing standard scales of deep acting, surface acting, workfamily conflict, and psychological distress. After establishing the psychometric properties of the scales, hypotheses were tested using Partial Least Squares structural equation modeling. Results indicated that both surface acting and deep acting were linked positively to work-family conflict. Surface acting emerged as a most important variable contributing to more than seventy five percent of variance of work-family conflict. Finally, work-family conflict mediated the relationship between emotional labor strategies (i.e. surface acting/deep acting) and psychological distress. Keywords: Emotional Labor, Work-Family Conflict, Mediation

Introduction
Although study of emotions have long been a topic of interest to psychologists and sociologists (e.g., Clark, 1992; Thoits, 1990) recently organizational behavior researchers have started increasingly recognizing the important role of emotions within the work setting (e.g., Hartel, Zerbe, & Ashkanasy, 2005). One area of emotion research that has received considerable attention within work setting is emotional labor (EL) (e.g., Ashforth & Tomiuk, 2000; Grandey, 2000; Hochschild, 1983). EL is the regulation of emotions for a wage (Hochschild, 1983). EL consists of personal interactions (mainly apart from job descriptions) among employees and between employees and customers that facilitate the effective and smooth operation of the organization (Meier, Mastracci, & Wilson, 2006). Since the conceptualization of EL concept by Hochschild (1983), considerable amount of theoretical and empirical research has been done regarding its conceptualization, antecedents and consequences (e.g., Ashforth & Tomiuk, 2000; Grandey, 2000; Grandey & Brauburger, 2002; Rafaeli & Sutton, 1987; Schaubroeck & Jones, 2000; Zapf, 2002). There is accumulating evidence that EL influences psychological distress within work setting (e.g., Kruml & Geddes, 2000; Totterdell & Holman, 2003). Although this stream of research has contributed substantially to explicating the role of EL in relation to psychological distress, it is not without limitations. For instance, while the role of EL in predicting psychological distress is well established, the mechanism through which EL predicts psychological distress is little researched. One potential construct that can serve as a linking 584

European Journal of Social Sciences Volume 11, Number 4 (2009) mechanism between EL and psychological health is work-family conflict (WFC) (Montgomery, Panagopolou, Wildt, & Meenks, 2006). There is increasingly scholarly interest in the relationship between work and family life (Eby, Casper, Lockwood, Bordeaux, & Brinely, 2005; Michel, Mitchelson, Kotrba, LeBreton, & Baltes, 2009). Studying the links between work and family are important because conflict between family and work life has been linked to many undesirable outcomes such as, burnout, stress, depression and psychological distress (e.g., Frone & Rice, 1987; Frone, 2003; Frone, Russell, & Cooper, 1992). Recently, researchers have started giving attention to the relationship between EL and WFC (e.g., Kinman, 2009; Montgomery, Panagopolou, & Benos, 2005; Montgomery et al., 2006; Seery, Corrigall, & Harpel, 2008; Yanchus, Eby, Lance, & Drollinger, in press). Despite this initial stream of research, there are still several gaps in the existing literature. For instance, various models have been presented within the work-family literature explicating the mediatory role of work-family construct between work and family domain antecedents, and certain life outcomes (e.g., Carlson & Kacmar, 2000; Carlson & Perrew, 1999; Frone et al.,1992; Michel et al., 2009), but none of the reviewed research specifically examined the mediatory role of WFC in the relationship between EL and psychological distress. Moreover, results concerning the relationship between EL and WFC are somewhat inconsistent. For instance, in two different studies conducted on nurses, Seery et al. (2008) found a positive and significant relationship between surface acting (SA) (a form of EL) and WFC, whereas Montgomery et al. (2005) found no significant relationship between SA and WFC. Thus, more research is clearly required that examines relationship between aspects of EL and WFC in other occupational and cultural settings. This study sought to test a model of EL that includes WFC in the EL-Psychological distress relationship in a sample of employees working in public sector organizations. Psychological distress is a serious problem faced by many employees (Tsaousis & Nikolaou, 2005). An understanding of the relationship between EL, WFC, and psychological distress will help managers to take care of problem of distress in employees. Furthermore, most of the previous studies relating EL to WFC and psychological distress have been conducted in the West raising the questions about the extent to which these findings are generalizable to the East. This study adds to the literature by testing the proposed model in the South Asia, thus providing some empirical cross-cultural validity of EL-psychological distress relationship.

Conceptual Framework
Emotional Labor In many situations in our daily lives, we often display particular emotions (which may not correspond to the emotions we are actually experiencing) in response to social expectations of appropriate behavior. For example, there are times we suppress anger when being cut off by someone in waiting line. The process in which people regulate (and display) their emotions to comply with social norms is referred to as emotion work (Hochschild, 1990; p.118). We also regulate our emotions (displaying particular ones and suppressing others) in response to job-related expectations of appropriate emotional behavior or for a wage. Hochschild (1983) was among the first to examine such phenomenon in the modern work setting and coined the term emotional labor. She defined EL as the management of feeling to create a publicly observable facial and bodily display to keep up with job requirements. EL is sold for a wage and therefore has exchange value (p. 7). With the increasing trend of service oriented organizations (e.g., Bowen, Siehl, & Schneider, 1989), where internal clients and coworkers are equally treated as important customer-like constituencies (Witt, 1999), the role of EL has gone beyond the traditional focus on only service provider-client interaction to universally applied phenomenon in the work place (Liu, Perrew, Hochwarter, & Kacmar, 2004). Hence, employees as emotional beings play the key roles and must be instructed by the organization to present themselves according to certain emotional rules. 585

European Journal of Social Sciences Volume 11, Number 4 (2009) Researchers proposed that employees perform EL through two types of acting mechanism: surface acting (SA) and deep acting (DA) (e.g., Grandey, 2003; Hochschild, 1983). In DA, employee attempts to deeply modify internal feelings to match the required organizational display rules. DA involves changing inner feelings by altering something more than outward appearance. Employees engaged in DA make an effort to understand people, be empathetic to their circumstances, and internalize their feelings. Rafaeli and Sutton (1987) referred to this act as faking in good faith because employees intent is to seem authentic to the audience. For example, a cashier tries to look concerned by feeling what a demanding customer is experiencing (from within the frame of reference of that particular customer). In SA, employee modifies outward displays to be consistent with display rules without shaping inner feelings. In other words, employee hide felt emotions or fake unfelt emotions. SA increases emotional dissonance (a gap between felt and expressed emotions) (Grandey, 2003). Furthermore, SA is also detrimental to organization in a way that employees conform to organizational display rules in order to keep their jobs, but not to help customers or organization (Grandey, 2003). Hence, SA is termed as faking in bad faith (Rafaeli & Sutton, 1987). For example, a cashier maintains a smile and social demeanor even though internally he might be feeling sad. The expression of positive emotion is fitting for the customer-service context, but negative emotion is being experienced. EL and Work-Family Conflict Greenhaus and Beutell (1985) define WFC as, a form of inter-role conflict in which the role pressures from the work and family domains are mutually incompatible in some respect (p. 77). According to Greenhaus and Beutell (1985), WFC can be time-based, strained based, or behavior based. In timebased conflict, the time demands of one role make it difficult to participate fully in another role. Strained based conflict occurs, when the strain experienced in one role domain intrude or spill over into the other role, making it difficult to fulfill the responsibilities of that role. Finally, behavior-based conflict occurs when the behaviors that are expected or appropriate in one role (family/work) are viewed as inappropriate or dysfunctional when used in other role (family/work). In other words, behavior-based conflict is described as conflict stemming from incompatible behaviors demanded by competing roles. The relationship between EL and WFC can be traced into behavior-based and strained based conflict. According to Hochschild (1983), performance of EL duties may engender a fusion of self and work roles. Employees performing DA are often too identified (preoccupied) with their work roles that they find it difficult to depersonalize and detach themselves from their work roles. They find it difficult to recover their true feelings (even after the role performance) and hence lose track of when they are acting or not (Hochschild, 1983). This preoccupation may interfere with their efforts to fulfill the demands of a competing role (Greenhaus & Beutell, 1985). Hence, due to this fusion of self and work role, a person is unable to adjust behavior to comply with the expectations of different roles (e.g., son, father, brother, husband), and is likely to experience conflict between the roles (work role and family role). For example, behavioral styles that employees exhibit at work (e.g., power, authority) may be incompatible with behaviors desired by members within the family domain. Thus, high involvement in a role due to DA is predicted to influence WFC. Hypothesis 1: DA is positively related to WFC. Furthermore, Hochschild (1983) assert that, employees performing EL duties may become estranged from their true feelings. This estrangement between work roles and oneself leads to emotional dissonance (a gap between felt and expressed emotions) and is associated with many negative psychological outcomes (e.g., Ashforth & Humphrey, 1993; Zapf, 2002). The emotional dissonance has important implication for strained based conflict (when the strain experienced in one role domain intrude or spill over into the other role). According to Wharton (1999), the major reason for emotional dissonance is that the organizational display rules prevent employees from interacting with customers based on spontaneous intuition, which compel employees to replace and suppress their 586

European Journal of Social Sciences Volume 11, Number 4 (2009) own emotional response by an organizationally sanctioned response. Since SA leads to inauthentic/fake emotional displays, it has been argued that emotional dissonance is the direct outcome of SA. SA exerts its negative effects through emotional dissonance on other negative psychological outcomes. Prior research has found a strong relationship between SA and emotional exhaustion (e.g., Brotheridge & Lee, 2003; Brotheridge & Grandey, 2002; Grandey, 2003; Grandey et al., 2005; Johnson, 2007). Emotional exhaustion refers to the depletion of emotional resources and physical energy to such a degree that adequate resources are no longer available to effectively manage the emotions (Maslach & Jackson, 1981). Research suggests that certain stressful events at work, burnout, or fatigue at work make it difficult to pursue satisfying nonworking life (Bartolome & Evans, 1980), have a debilitating effect on quality of family life (Jackson & Maslach, 1982) and may cause withdrawal from personal contact at home (Kanter, 1977). This leads to following prediction: Hypothesis 2. SA is positively related to WFC. There is ample evidence for the negative effects of WFC on psychological factors. For example, WFC has been found associated with Burnout (Netemeyer, Boles, & McMurrian, 1996), high levels of depersonalization (Kinnunen & Mauno, 1998), psychological distress (Kafetsios, 2007; ODriscoll, Ilgen, & Hildreth, 1992), and psychological well being (Parasuraman, Purohit, Godshalk, & Beutell, 1996). This leads to following prediction: Hypothesis 3. WFC is positively related to psychological distress. A given variable functions as a mediator when its inclusion in analysis results in a significant reduction in the relationship between the independent and outcome variable (Baron & Kenny, 1986).Various models have been presented within the work-family literature explicating the mediatory role of work-family construct between work and family domain antecedents, and certain outcomes (e.g., Frone et al., 1992; Carlson & Kacmar, 2000; Carlson & Perrew, 1999; Michel et al., 2009). Montgomery et al. (2006) hypothesized that work family interference (WFI) mediates the relationship between EL (DA and SA) and burnout, psychosomatic complaints, and cynicism. WFI was found to partially mediate between SA and both cynicism and psychosomatic complaints but no support was found for the mediatory role of WFI between DA and cynicism and psychosomatic complaints. In line with Montgomery et al. (2006) hypothesis, this study expects that WFC will mediate the relationship between EL and psychological distress. This leads to following predictions: Hypothesis 4a. WFC mediates the relationship between SA and psychological distress. Hypothesis 4b. WFC mediates the relationship between DA and psychological distress.

Method
Participants The sample for this study consisted of 200 employees from three public sector organizations situated in province of Balochistan, Pakistan. 92 participants of the total sample (46 percent) were males and 108 (54 percent) were females. The mean age for this sample was 31.48 years (SD = 8.10). All participants were treated in accordance with the Ethical principles of Psychologists and Code of Conduct (American Psychological Association, 2002). Administration of the questionnaires was carried out by post graduate students who acted as research assistants and no monetary incentive was provided. Measures Psychological distress. Psychological distress was measured by Chans (2005) twenty items scale. This scale measures psychological distress in terms of current non-psychotic symptoms in the five symptom areas represented by scales of health concerns, sleep problems, anxiety, dysphoria, and suicidal ideas. Respondents were requested to rate each symptom statement on a 5-point scale (not at all to extremely) by comparing themselves during the past 2 weeks with their usual selves. Coefficients alphas for the 587

European Journal of Social Sciences Volume 11, Number 4 (2009) five dimensions were: health concerns: .75; sleep problems: .68; anxiety: .60; dysphoria: .86; and suicidal ideas: .78. Emotional Labor. SA was measured by five items adopted from Grandeys (2003) EL scale. The sample items include, I just pretend to have the emotions I need to display for my job. DA was measured by three items adopted from Brotheridge and Lee (1998) EL scale. The sample items include, I make an effort to actually feel the emotions that I need to display to others. The response scale has been seven point Likert-type scale ranging from one (strongly disagree) to seven (strongly agree). In this study, the SA and DA scales had good internal consistencies (Alphas .72 and .86 respectively). Work-to-family conflict. WFC was measured by five item scale that was developed by Netemeyer et al., (1996). The sample items include, things I want to do at home do not get done because of the demands my job puts on me. The response scale has been seven point Likert-type scale ranging from one (strongly disagree) to seven (strongly agree). In the present study, the coefficient of internal consistency (Cronbachs alpha) for this scale was .89.

Design and Analysis

Inspection of skewness and kurtosis statistics revealed non normality for most of the items. Furthermore, Mardias (1970) coefficient of multivariate kurtosis provided by AMOS (Arbuckle, 2006) indicated that the assumption of multivariate normality was not tenable (Mardias coefficient = 48.57, CR = 11.5, p < .05). Based on these results, I resorted to Partial Least Squares (PLS) path modeling algorithm. Like covariance based structural equation modeling (CBSEM), PLS is a latent variable modeling technique that incorporates multiple dependent constructs and explicitly recognizes measurement error. However, PLS is far less restrictive in its distributional assumptions and sample size restrictions as compared to covariance-based structural equation modeling. Furthermore, maximum likelihood models are based on assumptions of a specific joint multivariate distribution and independence of the observations (independently and identically distributed, i.e, iid), PLS does not impose such requirements on data. PLS applies to situations where knowledge about distribution of the latent variables is limited and requires the estimates to be more closely tied to the data compared to covariance structure analysis (Fornell & Cha, 1994). Moreover, the application of PLS requires a minimum sample size that is (1) ten times the number of items comprising the most formative constructs, or (2) ten times the largest number of structural paths directed at a particular construct in the inner path model (Barclay, Higgins, & Thompson, 1995). With a sample size of 200 in this study, these requirements were well met. Specifically, I employed SmartPLS (Ringle, Wende, and Will, 2005), which allows for estimating both measurement model and structural model simultaneously. According to Henseler, Ringle, and Sinkovics (2009) recommendations PLS model was analyzed and interpreted in two stages: the measurement model and the structural model. The measurement model relates to the relations between manifest variables (observed items) and latent variables. The measurement model is tested by assessing the validity and reliability of the items and constructs in the model. This ensures that only reliable and valid constructs measures are used before assessing the nature of relationships in the overall model. In PLS, individual item reliability is assessed by examining the loadings of respective items on their respective latent construct (Hulland, 1999). The higher loadings imply that there is more shared variance between the construct and its measures than error variance. Whereas, low loadings add very little to the explanatory power of the model while attenuating the estimates of the parameters linking constructs (Hulland, 1999).Composite reliability (c) (Werts, Linn, & Joreskog, 1974) and Cronbachs alpha (1951) were used to assess the reliability of scales. Composite reliability is preferred over Cronbachs alpha because it offers a better estimate of variance shared by the respected indicators and because it uses the item loadings obtained within the nomological network (Hair, Anderson, Tatham, & Black, 2006). For assessing the convergent validity1 of constructs, I employed Fornell and Larckers (1981) average variance extracted (AVE) criterion. An AVE value greater than 0.50 indicates that a latent variable is able to explain more than half of the 588

European Journal of Social Sciences Volume 11, Number 4 (2009) variance of its indicators on average (Henseler et al., 2009). Discriminant validity2 of measurement model was tested through Fornell and Larckers (1981) AVE test and cross loadings criterion (Chin, 1998).. Evidence of discriminant validity occurs when square root of the variance extracted estimation exceed the correlations between the factors making each pair (Fornell and Larcker, 1981). Each latent variable shares more variance with its own block of indicators than with another latent variable representing a different block of indicators. According to cross loading criterion (Chin, 1998), the loading of each indicator is expected to be greater than all of its cross-loadings. Structural model specifies relations between latent constructs. The structural model is tested by estimating the paths between the constructs, which are an indicator of the models predictive ability. The nonparametric bootstrapping procedure (Chin, 1998; Davison & Hinkley, 2003) using 1000 subsamples was performed to evaluate the statistical significance of each path coefficient and to provide confidence intervals for all parameter estimates. Contrary to CBSEM (covariance based structural equation modeling) PLS path modeling does not report any kind of fit indices like TFI, RMSEA or CFI (since PLS makes no distributional assumptions for parameter estimation). The evaluation of PLS model is therefore based on prediction oriented measures that are non-parametric (Chin, 1998). The PLS structural model is mainly evaluated by R2 of endogenous latent variable Chin (1998), effect size f2 (Cohen, 1988), Goodness of Fit index (GoF) (Tenenhaus, Esposito Vinzi, Chatelin, & Lauro, 2005), and by using the Stone-Geiser Q-square test for predictive relevance (Stone, 1974; Geiser, 1975). According to Chin (1998), R2 values of 0.67, 0.33, and 0.19 for endogenous latent variables are described as substantial, moderate and weak respectively. In the present study, effect sizes were determined by a method identified by Cohen (1988, p.410) and adopted by Schroer and Herterl (2009) in PLS path models. In this method, effect sizes of single predictors are obtained by comparing the explained amount of variance when a predictor is either included or not included in the model, that is, f2 = (R2incl R2excl)/ (1 R2includ). According to Cohen (1988), f2 values of 0.02, 0.15, and 0.35 signify small, medium, and large effects, respectively. Goodness-of-fit (GoF) (Tenanhaus et al., 2005) was employed to judge the overall fit of the model. GoF, which is the geometric mean of the average communality (outer measurement model) and the average R2 of endogenous latent variables, represents an index for validating the PLS model globally, as looking for a compromise between the performance of the measurement and the structural model, respectively. GoF is normed between 0 and 1, where a higher value represents better path model estimations. The Q-squares statistics measure the predictive relevance of the model by reproducing the observed values by the model itself and its parameter estimates. A Q-square greater than 0 means that the model has predictive relevance; whereas Q-square less than 0 mean that the model lacks predictive relevance (Fornell & Cha, 1994). In PLS, two kinds of Q-squares statistics are estimated, that is, crossvalidated communality (H2j) and cross-validated redundancy (F2j). Both statistics are obtained through blindfolding procedure in PLS. Blindfolding procedure (while estimating Q-squares) ignores a part of the data for a particular block during parameter estimation (a block of indicators is the set of measures for a construct). The ignored data part is than estimated using the estimated parameters, and the procedure is repeated until every data point has been ignored and estimated. Omission and estimation of data point for the blindfolded construct depend on the chosen omission distance G (Chin, 1998). A cross-validated communality H2j is obtained if prediction of the omitted data points in the manifest variables block is made by underlying latent variable (Chin, 1998). In other words, the cvcommunality H2j measures the capacity of the path model to predict the manifest variables (MVs) directly from their own latent variable (LV) by cross-validation. It uses only the measurement model. The prediction of a MV of an endogenous block is carried out using only the MVs of this block (Tenanhaus et al., 2005). On the other hand, a cross-validated redundancy predicts the omitted data points by constructs that are predictors of the blindfolded construct in the PLS model (Chin, 1998). In other words, the cv-redundancy F2j measures the capacity of the path model to predict the endogenous MVs indirectly from a prediction of their own LV using the related structural relation, by crossvalidation. The prediction of an MV of an endogenous block j (e.g., WFC) is carried out using all the 589

European Journal of Social Sciences Volume 11, Number 4 (2009) MVs of the blocks j* (e.g., SA and DA) related with the explanatory LVs of the dependent LV j (e.g., WFC) (Tenanhaus et al., 2005). This index is used for measuring the quality of the path model. In accordance to effect size (f2), the relative impact of the structural model on the observed measures for latent dependent variable is evaluated by means of q2 (Henseler et al., 2009). The q2 values of 0.02, 0.15, and 0.35 signify small, medium, and large predictive relevance of certain latent variable, thus explaining the endogenous latent variable under evaluation. Finally, in order to test the mediation effect of WFC, I employed product of coefficients strategy (Sobel, 1982; Preacher & Hayes, 2004; 2008). Product of coefficients strategy is preferred over Baron and Kennys (1986) casual step approach because of two main reasons. First, causal step approach does not consider the estimate of the indirect effect, nor a standard error for this effect that might permit direct investigation of statistical significance. That is, it ignores the central question: Is the indirect effect different from zero? (Preacher & Hayes, 2008). Second, testing the null hypothesis that indirect effect = 0 requires one fewer hypothesis test, and thus type II error in the testing of mediation would be less likely (Preacher & Hayes, 2004). Mediation analysis was conducted with latent variable scores obtained in PLS analysis and then using these latent variable scores as an input for the SPSS macro provided by Preacher and Hayes (2004).

Results
Measurement Model The factor loadings from the final PLS measurement models are reported in Figure 1. All items loaded significantly (> .50) on their respective factors which was an indication of indicator reliability. Composite reliability (c) (Werts, Linn, & Joreskog, 1974) and Cronbachs alpha (1951) values for all scales exceeded the minimum threshold level of .70 (Nunnally & Bernstein, 1994), thus indicating the reliability of all scales used in this study (Table 1). Results revealed that the variance extracted for all factors exceeded the minimum threshold value of .50 which was an indication of convergent validity of all scales (Table 1). Fornell and Larckers (1982) test for discriminant validity revealed relatively high variances extracted for each factor compared to the inter-scale correlations, which was an indication of discriminant validity of four constructs (i.e., WFC, psychological distress, SA, and DA) (Table 1).
Table1: Reliability, Convergent and Discriminant Validity
Correltations 2 3 CRb c AVEd

1 2 3 4 Note:

1 3 WFC 0.83a 0.92 0.89 0.70 PD 0.50 0.75 0.87 0.81 0.57 SA 0.41 0.34 0.70 0.83 0.76 0.51 DA 0.29 0.26 0.42 0.88 0.91 0.86 0.78 WFC = Work-family conflict; PD = Psychological distress; SA = Surface acting; DA = Deep acting. a square root of AVE. b Composite reliability (c) = ( i)2 / [( i)2 + Var (i)], where i is the outer factor loading, and Var (i) = 1 - i. is the measurement error or the error variance associated with the individual indicator variable(s) for that given factor (Fornell & Larcker, 1981). c = Cronbach Alpha (Cronbach, 1951). d Average variance extracted (AVE) = ( 2i) / [( 2i) + Var (i)], where where i is the outer factor loading, and Var (i) = 1 - i, is the measurement error or the error variance associated with the individual indicator variable(s) for that given factor (Fornell & Larcker, 1981).

590

European Journal of Social Sciences Volume 11, Number 4 (2009)

Figure 1: Structural Model. SA = Surface acting; DA = Deep acting; WFC = Work-family conflict; PD = Psychological Distress; PD1 = Health concerns; PD2 = sleep problems; PD3 = anxiety; PD4 = dysphoria; PD5 =suicidal ideas.

Structural Model The results shown in Table 2 support hypothesis 1and 2: the coefficient of the path from SA to WFC was significant with low effect size and low predictive relevance ( = .33, t = 4.74, p < .01, 95% CI: (.21) (.47), f2 = .11, q2 = .07), as was the coefficient of the path from DA to WFC, with small effect size and predictive relevance ( = .15, t = 2.04, p < .05, 95% CI: (.09) (.29), f2 = .02, q2 = .03). Furthermore, the WFC was significantly related to psychological distress with medium effect size and low predictive relevance ( = .42, t = 8.10, p < .001, 95% CI: (.32) (.52), f2 = .17, q2 = .07) (hypothesis 3). Finally, the coefficient of the path from SA to psychological distress was insignificant (( = .13, t = 1.88, p > .05, 95% CI: (-.0001) (.26), f2 = .01, q2 = .01), as was the coefficient of the path from DA to psychological distress (( = .09, t = 1.63, p > .05, 95% CI: (-.001) (.19), f2 = .01, q2 = .01).
Table 2: Path Coefficients
q2 0.09 0.07 0.02 0.03 0.01

Path t LL 95% CI UL 95%CI f2 WFC -> PD 0.42 8.10 0.32 0.52 0.17 SA -> WFC 0.33 4.74 0.21 0.47 0.11 SA -> PD 0.13 1.88 -0.009 0.26 0.01 DA -> WFC 0.15 2.04 0.09 0.29 0.02 DA -> PD 0.09 1.63 -0.001 0.19 0.01 Note. t values and CIs are calculated through bootstrapping routine with 200 cases and 1000 samples. f2 = (R2incl R2excl)/ (1 R2includ). q2 = (F2j incl F2j excl)/ (1 F2j includ).

Exogenous variables in the model explained low amounts of variance of WFC (R2 = .18) and psychological distress (R2 = .28). The value of R2 may be decomposed in terms of the multiple regression coefficients and correlations between the dependent variable and the explanatory ones (Tenanhaus et al., 2005). This decomposition allows understanding the contribution of each 591

European Journal of Social Sciences Volume 11, Number 4 (2009) explanatory variable to the prediction of the dependent one. For this model, SA was the most important variable in the prediction of WFC, contributing to 75.16 % of the R2. On the contrary, DA contribution was only 24.12 % (far less than SA). Further, for the endogenous variable psychological distress, WFC emerged as a most important variable contributing to 75 % of the R2 (Table 3).
Table 3: The Explanation of Psychological Distress
(R2 = .17) a r Contribution to R2 (%) j .26 .35 52.25 -.19 -.14 15.64 .21 .26 32.11 , where j is the path coefficient, y is the dependent latent variable and, xj is the independent latent variable.

SA EI DA Note: a

The blindfolding results (G = 30 blocks) are presented in Table 4. We may notice that for this model all blocks had high values for cv-communality index H2. These values were well above the threshold level of zero (Fornell and Cha, 1994). However, regarding cv-redundancy index F2, WFC and psychological distress had low values. Furthermore, the 0.37 value of GoF index was quite acceptable. In sum, results indicated that model had an acceptable predictive relevance.
Table 4: Communality and Redundancy
R2 0.18 0.28
a

Block

WFC PD SA DA Average 0.23 0.14 c GoF 0.37 Note: a Communality j = 1/p cor2 (Xjh, Yj), where p = is the total number of MVs in the block; Xjh = hth MV in jth block; Yj = latent variable. The communality index measures the quality of the measurement model for each block (Tenanhaus et al., 2005). b 2 H = CV-communality index c Redundancy j = communality j X R2. The redundancy index measures the quality of the structural model for each endogenous block, taking into account the measurement model (Tenanhaus et al., 2005). d CV-redundancy (F2j) = 1 D = (D SSED) / (D SSOD), where D is the omission distance, SSE is the sum of squares of prediction errors, and SSO is the sum of squares of observations (Henseler et al., 2009). It measures the capacity of the path model to predict the endogenous MVs indirectly from a prediction of their own LV using the related structural relation, by cross-validation (Tenanhaus et al., 2005). e GoF = (Tenanhaus et al., 2005).

Communality 0.70 0.57 0.50 0.78 0.63

H2 .52 .35 .32 .52

Redundancy 0.12 0.16

F2 .10 .13

To test Hypothesis 4a and 4b proposing that WFC mediates the effect of EL strategies on psychological distress, two separate mediation analysis were conducted (one for each EL strategy) with case values of composite latent variables obtained in the PLS analysis. The bootstrapped confidence interval and the product of coefficient approach with second order SE estimate (Preacher and Hayes, 2004) were used to test the significance of indirect effects in both models. As can be seen in the Table 5, WFC mediated the relationship between SA and psychological distress (indirect effect = 0.17, Sobel z = 6.17, p < .001, 95% CI: 0.10 - 0.27) and between DA and psychological distress (indirect effect = 0.13, Sobel z = 3.70, p < .001, 95% CI: 0.08 - 0.21). Furthermore, the signs of the indirect effects were consistent with the interpretation that SA and DA increase WFC, which in turn influences psychological distress.

592

European Journal of Social Sciences Volume 11, Number 4 (2009)

Table 5: Bootstrap Results for Indirect Effects
UL 95 CI 0.27 0.21

Model Path Indirect effect S.E Sobel z LL 95 CI 1 SA -> WFC -> PD 0.17 0.042 6.17** 0.10 2 DA -> WFC -> PD 0.13 0.039 3.70** 0.08 Note: Values are calculated through a bootstrapping routine with 200 cases and 1000 samples. ** p < .001

Discussion
This study examined the associations amongst EL, WFC, and psychological distress in a sample of public sector organizations in Pakistan. Proposed hypothesis received considerable support. This is the first study (to my knowledge) in any South Asian countrys context to assess the relationship of EL strategies with other variables. The findings of this study suggest that the degree of EL undertaken by employees may have negative implications for their psychological well-being that extend beyond their work domain. In terms of first hypothesis, DA was significantly related to WFC. This suggests that employees are more often preoccupied with their roles at work (due to DA) that they lose track of when they are acting (at work) and when they are off the job (at home). Hence, this preoccupation with job role makes it difficult for the employees to comply with their other roles at family. In other words, selling feelings for a wage results in alienation from ones real self (Hochschild, 1983). Moreover, the depletion of cognitive and energy resources (while modifying internal states in DA) suggests that emotional demands at work produce strain (Grandey, 2003) that makes it difficult for employees to fulfill family duties and responsibilities. In accordance with previous research findings (e.g., Kinman, 2009; Montgomery et al., 2005; Montgomery et al., 2006; Seery et al., 2008; Yanchus et al., in press), evidence has been provided that employees who commonly engage in SA (faking and suppressing emotions) have more WFC conflict. SA increases emotional dissonance (a gap between felt and expressed emotions), which in turn leads to emotional exhaustion (a key component of burnout) (Grandey, 2003). Although not directly tested in this study, it is assumed that SA through emotional exhaustion intrude or spill over in the family domain and has debilitating effects on family life. The results suggest that the direct effect of SA is stronger for WFC as compared to DAs effect. Among two EL strategies SA emerged as the most important predictor of WFC contributing to substantial proportion of the variance in the WFC (75.16%). This is an important finding supporting accumulating evidence for the primacy of SA over DA as an important predictor of WFC (e.g., Montgomery et al., 2006; Yanchus et al., in press). These results were consistent with Grandeys (2003) assertion that, DA minimizes emotional dissonance by bringing feelings in line with expressions, so DAs relationship with emotional exhaustion should be weaker than the relationship between SA and emotional exhaustion (p. 89). In other words, both SA and DA impact WFC through emotional exhaustion but SA has more profound impact on WFC than DAs impact. The results of this study confirmed expectation for an inverse relationship between WFC and psychological distress. This finding is in line with previous research that documented the adverse effects of WFC on psychological distress (Frone et al., 1997; Kafetsios, 2007). Moreover, WFC emerged as a most important variable contributing to 75% of the variance in psychological distress. This is an important finding delineating the important role of WFC in predicting psychological distress. The results from the mediation analysis further point to the importance of WFC when examining SA and DA influence on psychological distress, as stated in hypothesis 4a and 4b. The mediation results are consistent with the literature denoting the important role of WFC as a mediator (e.g., Montgomery et al., 2006). The mediating role of WFC found for both SA and DA provides a new perspective regarding the way that the work can spill-over into non work life and impact psychological distress. 593

European Journal of Social Sciences Volume 11, Number 4 (2009)

Implications
Effort-Recovery Model (E-R Model: Meijman & Mulder, 1998) suggests that employees build up negative effects on the job. This does not necessarily give rise to negative consequences for employees well-being as long as employees are give adequate time to recover from these effects. Jobs that require substantial amounts of EL (emotional demands) should take into consideration the adequate amount of private time (e.g., schedule breaks) that can be sufficient to recover from (emotional) load effects built up at work resulting from high level of emotional work. This can help employees from carrying these negative effects from their work to non work life. Training programs focusing on emotional regulation skills and DA techniques to cope with emotional demands of work and family can help in reducing the deleterious effects of SA (Yanchus et al., in press). Besides training, employers can help employees to internalize their roles rather to fake the emotions (Ashforth & Humphrey, 1983) by providing them adequate resources needed to meet the demands of the job. The negative impacts of emotional labor can be minimized by selecting employees with the aim of achieving the best person-job fit. For instance, applicants who demonstrate a high level of positive affectivity and emotional regulation skills would be considered good job fit for jobs requiring high level of EL strategies. Positive affectivity and emotional regulations skills are shown to be negatively related to WFC and psychological distress (Kafetsios, 2007; Yanchus et al., in press). Based on Fredricksons (2001) broaden-and-build model of positive emotions, it is suggested that intervention strategies should be introduced within work setting that cultivate positive emotions among employees. Research indicates that SA impacts WFC via negative affectivity (Yanchus et al., in press). According to Fredricksons (2000, p. 1), positive emotions (such as, joy and contentment) broaden an individuals momentary thought action repertoire, and help in eradicating the hold of negative emotions on an individuals mind and body. Enhancement of positive emotions help in preventing and treating problems such a psychological distress, deeply rooted in negative emotions. Fredricksons (2000), suggested many intervention strategies that may help in preventing and treating psychological health related problems, as well as, help in building personal strengths, resilience and wellness of people. These intervention strategies include, (a) Relaxation therapies (e.g., imagery exercises, muscle exercises, mediation exercises), (b) decreasing the intensity of unpleasant events and increasing the rates of engagement in pleasant activities, (c) cognitive therapies, (d) training employees in finding positive meaning in daily life, (e) building empathy between people and groups.

Limitations and Recommendations

The findings of this study are subject to several limitations which are common in this type of research. First, the results are specific to organizations in one geographical area and may or may not be generalizable to other areas. Second, the cross-sectional data precludes any inference of causality. The direction of causality (in cross-sectional studies) cannot be established and will have to be examined using longitudinal data. Third, respondents in this study were full-time employees and these findings may not be applicable to part-time employees. Fifth, this study only explored the relationship of WFC with EL strategies. Future studies should also explore the impact of EL strategies on family-to-work conflict (FWC). Five, since all measures were self report based measures we cannot avoid the social desirability bias. Six, the relationships among these variables might differ depending on the type of job. Future research can examine these relationships for jobs that differ in terms of intellectual and interpersonal demands (e.g., sales vs. Engineering).

594

European Journal of Social Sciences Volume 11, Number 4 (2009)

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] American Psychological Association, 2002. Ethical Principles of Psychologists and Codes of Conduct, APA, Washington, DC Arbuckle, J. 2006. Amos 7.0 Users Guide, Amos Development Corporation: Spring House. Ashforth, B. E., & Tomiuk, M, A. 2000. Emotional labor and authenticity: Views from service agents. In S. Fineman (Ed.), Emotion in organizations (2nd ed.) (pp. 184-203). Sage, Thousand Oaks, GA. Ashforth, B.E., & Humphrey, R.H. 1993. Emotional labor in service roles: The influence of identity, Academy of Management Review, 18(1), pp. 88-115. Barclay, D. W., Higgins, C., & Thompson, R. 1995. The partial least square approach to causal modeling: Personal computer adoption use as illustration, Technology Studies, 2(2), pp. 285-309. Baron, R. M, & Kenny, D. A. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations, Journal of Personality and Social Psychology, 51, pp.1173-1182. Bartolome, F., & Evans, P. A. L. 1980. Must success cost so much?, Harvard Business Review, 58(2), pp. 137-148. Bowen, D. E., Siehl, C., & Schneider, B. 1989. A framework for analyzing customer service orientations in manufacturing, Academy of Management Review, 14, pp. 75-95. Brotheridge, C.M., & Grandey, A. A. 2002. Emotional labor and burnout: Comparing two perspectives of people work, Journal of Vocational Behavior, 60, pp. 7-39. Brotheridge, C.M., & Lee, R.T. 2003. Development and validation of the Emotional Labour Scale, Journal of Occupational and Organizational Psychology, 76(3), pp. 365-379. Brotheridgen, C. M., & Lee, R. E. 1998. On the dimensionality of emotional labor: Development and validation of the emotional labor scale, Paper presented at the first conference on emotions in organizational life, San Diego. Carlson, D. S., & Kacmar, K. M. 2000. Work-family conflict in the organization: Do life role values make a difference?, Journal of Management, 26, pp. 10311054. Carlson, D. S., & Perrew, P. L. 1999. The role of social support in the stressor-strain relationship: An examination of work-family conflict, Journal of Management, 25, pp. 513 540. Chan, D. W. 2005. Emotional Intelligence, social coping, and psychological distress among Chinese gifted students in Hong Kong, High Ability Studies, 16(2), pp. 163-178. Chin, W.W. 1998. The partial least squares approach to structural equation modeling. In G. A. Marcoulides(Ed.), Modern Methods for Business Research (pp. 295358). Lawrence Erlbaum Associates: Mahwah, NJ. Clark, M. S. (Ed.). 1992. Review of personality and social psychology: Emotion and social behavior, vol. 14, Sage, Newbury Park, CA. Cohen, J. 1988. Statistical power analysis for the behavioral sciences (2nd ed.),Hillsdale, Lawrence Erlbaum Associates, NJ. Cronbach, L. J. 1951. Coefficient Alpha and the internal structure of tests, Psychometrika, 16(3), 297-333. Davison, A.C., & Hinkley, D.V. 2003. Bootstrap methods and their application (2nd ed.), Cambridge University Press, New York, NY. Eby, L. T., Casper, W. J., Lockwood, A., Bordeaux, C., & Brinely, A. 2005. Work and family research in IO/OB: Content analysis and review of the literature (19802002), Journal of Vocational Behavior, 66, pp. 124197. Fornell, C., & Cha, J. 1994. Partial least squares. In Bagozzi, R.P. (Ed.), Advanced methods of marketing research (pp. 52-78). Basil Blackwell: Cambridge, MA. 595

European Journal of Social Sciences Volume 11, Number 4 (2009) [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] Fornell, C., & Larcker, D. F. 1981. Evaluating structural equation models with unobservable variables and measurement error, Journal of Marketing Research, 18(1), pp. 39-50. Fredrickson, B. L. 2001. The role of positive emotions in positive psychology: The broadenand-build theory of positive emotions, American Psychologist, 56, pp. 218-226. Fredrickson, B. L. 2000. Cultivating positive emotions to optimize health and well-being. Target article in Prevention and Treatment, 3, Available on the World Wide Web: http://journals.apa.org/prevention. Frone, M. R., Russell, M., & Cooper, M. L. 1992. Antecedents and outcomes of work- family conflict: Testing a model of the work-family interface, Journal of Applied Psychology, 77, pp. 6578. Frone, M.R., & Rice, R.W. 1987. Work-family conflict: The effect of job on family involvement, Journal of Occupational Behavior, 8, pp. 45-53. Frone, M. R. 2003. Work-family balance. In J. C. Quick & L. E. Tetrick (Eds.), Handbook of occupational health psychology (pp. 143162), American Psychological Association., Washington, DC. Geissser, S. 1975. The predictive samples reuse method with applications, Journal of the American Statistical Association, 70, pp. 320-328. Grandey, A. 2000. Emotion regulation in the workplace: A new way to conceptualize emotional labor, Journal of Occupational Health Psychology, 5, pp. 95-110. Grandey, A. A. 2003. When the show must go on: Surface acting and deep acting as determinants of emotional exhaustion and peer-rated service delivery, Academy of Management Journal, 46(1), pp. 86-96. Grandey, A. A., & Brauburger, A. L. 2002. The emotion regulation behind the customer service smile, In R. G. Lord, R. J. Klimoski, & R. Kanfer (Eds.), Emotions in the workplace: Understanding the structure and role of emotions in organizational behavior (pp. 260-294), Jossey-Bass, San Francisco, CA. Grandey, A. A., Fisk, G. M., Mattila, A. S., Jansen, K. J., & Sideman, L. A. 2005. Is 'service with a smile' enough? Authenticity of positive displays during service encounters, Organizational Behavior and Human Decision Processes, 96(1), pp. 38-55. Greenhaus, J. H., & Beutell, N. J. 1985. Sources of conflict between work and family roles, Academy of Management Review, 10, pp. 76-88. Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. 2006. Multivariate data analysis, Prentice-Hall, Upper Saddle River, NJ. Hartel, C. E. J., Zerbe, W. J., & Ashkanasy, N. M. (Ed.). 2005. Emotions in organizational Behavior, Lawrence Erlbaum, London. Henseler, J., Ringle, C. M., & Sinkovics, R. R. 2009. The use of partial least squares path modeling in international marketing. In Sinkovics, R. R. & Ghauri, P. N. (eds.), Advances in International Marketing (pp. 277-320), Emerald, Bingley. Hochschild, A. 1983. The managed heart: Commercialization of human feeling, University of California Press, Berkeley. Hochschild, A. 1990. Ideology and emotion management: A perspective and path for future research. In T. D. Kemper (Ed.), Research agendas in the sociology of emotions (pp.117-142), State University of New York Press, Albany, N. Y. Hulland, J. 1999. Use of partial least squares (PLS) in strategic management research: a review of four recent studies, Strategic Management Journal, 20, pp. 195204. Jackson, S. E., & Maslach, C. 1982. After-effects of job-related stress: Families as victims, Journal of Occupational Behavior, 3, pp. 63-77. Johnson, H. M. 2007. Service with a smile: Antecedents and consequences of emotional labor strategies, Unpublished doctoral dissertation, University of South Florida. 596

[32] [33] [34] [35] [36] [37] [38] [39] [40] [41]

European Journal of Social Sciences Volume 11, Number 4 (2009) [42] [43] [44] [45] [46] Kafetsios, K. 2007. Work-family conflict and its relationship with job satisfaction and psychological distress: The role of affect at work and gender, Hellenic Journal of Psychology, 4, pp. 15-35. Kanter, R. M. 1977. Work and family in the United States: A critical review and agenda for research and policy, Russell Sage Foundation, New York. Kinman, G. 2009. Emotional labour and strain in front-line service employees. Does mode of delivery matter?, Journal of Managerial Psychology, 24 (2), pp. 118-135. Kinnunen, U., & Mauno, S. 1998. Antecedents and outcomes of work-family conflict among employed women and men in Finland, Human Relations, 51, pp. 157-177. Kruml, S., & Geddes, D. 2000. Catching fire without burning out: Is there an ideal way to perform emotional labor? In N. M. Ashkanasy, C. E. J. Haertel, & W. J. Zerbe. (Eds.) Emotions in the workplace: Research, theory and practice (pp. 177-188), Quorum Books, Westport, CT. Liu, Y., Perrew, P. L., Hochwarter, W. A., & Kacmar, C. J. 2004. Dispositional antecedents and consequences of emotional labor at work, Journal of Leadership and Organizational Studies, 10 (4), pp. 12-25. Mardia, K. 1970. Measures of multivariate skewness and kurtosis with applications, Biometrika, 57, pp. 519-530. Maslach, C., & Jackson, S. E. 1981. The measurement of experienced burnout, Journal of Occupational Behavior, 2, pp. 99-113. Meier, K. J., Mastracci, S. H., & Wilson, K. 2006. Gender and emotional labor in public organizations: An empirical examination o f the link to performance, Public Administration Review, Nov/Dec, pp. 899-909. Meijman, T. F., & Mulder, G. 1998. Psychological aspects of workload. In P. J. D. Drenth & H. T. Thierry (Eds.), Handbook of work and organizational psychology: Vol. 2. Work psychology (pp. 5-33), Psychology Press, Hove, England. Michel, J. S., Mitchelson, J. K., Kotrba, L. M., LeBreton, J. M., & Baltes, B. B. 2009. A comparative test of work-family conflict models and critical examination of work-family linkages, Journal of Vocational Behavior, 74, pp. 199-218. Montgomery, A. J., Panagopolou, E., & Benos, A. 2005. Emotional labour at work and at home among Greek health-care professionals. Journal of Health and Organizational Management, 19, pp. 395408. Montgomery, A. J., Panagopolou, E., de Wildt, M., & Meenks, E. 2006. Work-family interference, emotional labor and burnout. Journal of Managerial Psychology, 21, pp. 3651. Netemeyer, R. G., Boles, J. S., & McMurrian R. 1996. Development and validation of workfamily conflict and family-work conflict scales. Journal of Applied Psychology, 81, pp. 400410. Nunnally, J.C., & Bernstein, I. H. 1994. Psychometric theory (3rd ed.). McGraw-Hill, NewYork, NY. ODriscoll, M. P., Ilgen, D. R., & Hildreth, K. 1992. Time devoted to job and off-job activities, interrole conflict, and affective experiences, Journal of Applied Psychology, 77, pp. 272-279. Parasuraman, S., Purohit, Y. S., Godshalk, V. M., & Beutell, N. J. 1996. Work and family variables, entrepreneurial career success, and psychological well-being. Journal of Vocational Behavior, 48, pp. 275-300. Preacher, K. J., & Hayes, A. F. 2004. SPSS and SAS Procedures for Estimating Indirect Effects in Simple Mediation Models, Behavior Research Methods, Instruments, & Computers, 36, pp. 717-731.

[47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59]

597

European Journal of Social Sciences Volume 11, Number 4 (2009) [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] Preacher, K. J., & Hayes, A. F. 2008. Asymptotic and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models, Behavior Research Methods, 40, pp. 879-891. Rafaeli, A., & Sutton, R. L. 1987. Expression of emotion as part of the work role. Academy of Management Review, 12, pp. 23-37. Ringle, C. M, Wende, S., & Will. 2005. SmartPLS Version 2.0 M2, www.smartpls.de. Schaubroeck, J., & Jones, J. R. 2000. Antecedents of work place emotional labor dimensions and moderators of their effects on physical symptoms, Journal of Organizational Behavior, 21, pp. 163-183. Schroer, J. & Hertel, G. 2009. Voluntary engagement in an open web-based encyclopedia: Wikipedians, and why they do it, Media Psychology, 12(1), pp. 96-120. Seery, B. L., Corrigall, E. A., & Harpel, T. 2008. Job-related emotional labor and its relationship to work-family conflict and facilitation. Journal of Family and Economic Issues, 29, pp. 461-477. Sobel, M. E. 1982. Asymptotic confidence intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociological methodology (pp. 290-312). Jossey-Bass, San Francisco. Stone, M. 1974. Cross-validiation choice and assessment of statistical predictions, Journal of the Royal Statistical Society,36(2), pp. 111-133. Tenenhaus, M., Esposito Vinzi, V., Chatelin,Y.-M., & Lauro, C. 2005. PLS path modeling, Computational Statistics & Data Analysis, 48(1), pp. 159205. Thoits, P. A. 1990. Emotional deviance: Research agendas. In T. Kemper (Ed.), Research agendas in the sociology of emotions (pp. 180-203), State University of New York Press, Albany. Totterdell, P. & Holman, D. 2003. Emotion regulation in customer service roles: Testing a model of emotional labor, Journal of Occupational Health Psychology, 8(1), pp. 55-73. Tsaousis, I., & Nikolaou, I. 2005. Exploring the relationship of emotional intelligence with physical and psychological health functioning, Stress and Health, 21, pp. 7786. Werts, C. E., Linn, R. L., & Joreskog, K.G. 1974. Intra class reliability estimates: Testing structural assumptions, Educational and Psychological Measurement, 34(1), pp. 2533. Wharton, A. 1999. The psychosocial consequences of emotional labor. In R. J. Steinberg, & D. M. Figart (Eds.), Emotional labor in service economy (pp. 38-45), 561. The Annals of the American Academy of Political and Social Science. Witt, L. A. 1999. This job is too much: Emotional labor on the job, Paper presented at the Annual Meeting of Society for Industrial and Organizational Psychology, Atlanta, GA. Yanchus, N. J., Eby, L. T., Lance, C. L., & Drollinger, S. (in press). The impact of emotional labor on workfamily outcomes. Journal of Vocational Behavior (2009), doi:10.1016/j.jvb.2009.05.001. Zapf, D. 2002. Emotion work and psychological wellbeing: A review of the literature and some conceptual considerations, Human Resource Management Review, 12, pp. 237-268.

598