Emotional labour profiles: Associations with key predictors and outcomes
Work & Stress
The present study examines how three emotional labor strategies (hiding feelings, faking emotions, and deep acting) combine within different profiles of workers among two samples characterized by different types and intensity of customer contact. In addition, this research investigates the role of perceived workload as well as perceived organizational support, supervisor support, and colleagues support in the prediction of profile membership. Finally, this research also documents the relation
... tween emotional labor profiles and adaptive and maladaptive work outcomes (job satisfaction, work performance, emotional exhaustion, sleeping problems, psychological detachment, and counterproductive work behaviors). Latent profile analysis revealed three similar emotional labor profiles in both samples. Results also showed the most desirable levels on all outcomes to be associated with Profile 3 (Low Emotional Labor/Low Surface Acting and Moderate Deep Acting), followed by Profile 2 (Moderate Emotional Labor/Moderate Surface Acting and High Deep Acting) and Profile 1 (High Emotional Labor), with most comparisons being statistically significant in both samples. In contrast, a more diversified pattern of findings was observed in the prediction of profile membership. For instance, perceived colleagues support did not predict membership into any of the profiles, while supervisor support predicted an increased likelihood of membership into Profile 3 relative to Profiles 1 and 2. Gabriel et al. (2015) Profile 1 deep actors Very low levels of deep and surface acting Display rules in the US sample: 1 = 4 = 5 > 2; 4 > 3 Display rules in the Singapore sample: 4 > 1 > 2 = 3; 4 > 5 > 2 Positive affect in the US sample: 1 > 2 = 3 = 4 > 5 Positive affect in the Singapore sample: 1 = 3 > 4 Negative affect in the US sample: 4 = 5 > 1 = 2 = 3 Negative affect in the Singapore sample: 4 > 3 Customer orientation in the Singapore sample: 1 = 2 = 3 = 4 > 5 Emotion demands-abilities fit in the Singapore sample: 2 > 1 = 3 = 4 = 5 Emotional exhaustion in the US sample: 5 > 4 > 1 = 2 = 3 Emotional exhaustion in the Singapore sample: 4 = 5 > 1 = 3 > 2 Job satisfaction in the US sample: 1 > 2 > 3 > 4 > 5 Job satisfaction in the Singapore sample: 3 > 1 = 2 > 4 > 5 Felt inauthenticity in the Singapore sample: 5 > 4 > 1 = 3 > 2 Profile 2 non-actors Low to moderate levels of deep and surface acting Profile 3 low actors Low levels of deep acting and high levels of surface acting Profile 4 regulators High levels of deep acting and low levels of surface acting Profile 5 surface actors High levels of deep and surface acting Supplements for Emotional Labor Profiles S1 Online Supplemental Materials for: Emotional Labor Profiles: Associations with Key Predictors and Outcomes Preliminary Measurement Models Preliminary measurement models were estimated using Mplus 8 . Due to the complexity of the multi-sample measurement models underlying all constructs assessed in the present study, these preliminary analyses were conducted separately for the emotional labor variables, the predictors, and the outcomes. These models were estimated as multiple group models, allowing for the estimation of similar models across both samples, and for the progressive integration of invariance constraints to the models. The emotional labor models included, in each sample, three factors for hiding feelings, faking emotions, and deep acting. The predictor model included, in each sample, four factors related to perceived organizational support, perceived supervisor support, perceived colleagues support, and workload. Finally the outcome model included four factors related to emotional exhaustion, sleeping problems, psychological detachment, and counterproductive work behaviors. The emotional regulation measurement models were estimated using exploratory structural equation modeling (ESEM; , forcing cross-loadings (even as small as .100) present in the population model to be exactly zero according to typical confirmatory factor analytic (CFA) specification forces these cross loadings to be expressed through the inflation of the factor correlations. In contrast, these same studies showed that the free estimation of cross-loadings, even when none are present in the population model, still provides unbiased estimates of the factor correlations. These ESEM factors were specified in a confirmatory manner, using an oblique target rotation (Asparouhov & Muthén, 2009; Browne, 2001) , allowing for the pre-specification of target loadings in a confirmatory manner, while cross-loadings are targeted to be as close to zero as possible, yet still freely estimated. However, because the factors included in the predictors and outcomes models are taken from distinct measurement instruments, these factors were estimated using classical CFA representations. In addition, in the predictors models, five orthogonal method factors were integrated to control for the methodological artefact associated with the parallel wording of the four items used to assess respondents' perceptions of organizational, supervisor, and colleagues support, as well as the negative wording of 6 items (Marsh et al.