**4. Discussion**

The aim of the present study was to revisit the happy productive worker model, extending it to consider not just the synergies between happiness and productivity, but also the antagonistic relations between these two constructs. Moreover, we aimed to clarify the implications of different operationalizations of relevant theoretically-based constructs for the model. Finally, we aimed to identify demographic antecedents for each cluster solution. In this way, this work has addressed important limitations of the happy-productive worker model by incorporating both the hedonic and eudaimonic components of wellbeing, considering different aspects of job performance as well as their different sources of evaluation, and focusing not just on the synergies between the two constructs (happiness and productivity), but also on the antagonistic relations, an issue that has hardly been considered in the research based on the model.

The results support a different way to specify and expand the happy-productive worker model. Indeed, by analyzing the relationships between different constructs, we are not taking a positive relationship that leads to being a "happy-productive" or "unhappy-unproductive" worker for granted. The present research has also contemplated a negative relationship between constructs that would appear on a daily basis and that would lead to being "happy-unproductive" or "unhappy-productive" at work. In this study, we provide an affirmative response to Research Question 1, which asks whether "employees show different patterns considering the antagonist relation beyond the traditional synergetic relation between performance and wellbeing (i.e., happy-productive)". In fact, we have found antagonist patterns of wellbeing and performance (i.e., happy-unproductive and, in some cases, unhappy-productive) that are well represented in our sample. We found these alternative patterns by taking into account different operationalizations of wellbeing (i.e., hedonic, eudaimonic) and performance (i.e., self-rated, evaluated by the supervisor). In fact, the results indicate that, on average, over 50% of the respondents belong to the unhappy-productive/happy-unproductive clusters, which suggests that it is important to consider the antagonistic patterns of wellbeing and performance when re-defining the happy-productive worker thesis. Thus, we contribute to filling the gap identified by Warr and Nielsen [51], who pointed out that it is important to learn more about individuals who are outside the happy-productive pattern by considering additional measures of performance and wellbeing.

In fact, Research Question 2 asks whether the same employees belong to the same patterns of wellbeing and performance in their different operationalizations. The results show that a large number of employees do not belong to analogous clusters in different operationalizations of wellbeing and performance, which means that some employees are classified as unhappy in a hedonic way and, simultaneously, happy in an eudaimonic way (and vice-versa). This result draws our attention to the complexity of the phenomenon of wellbeing and the importance of considering both the hedonic

and eudaimonic dimensions in studies on wellbeing. It clearly shows that merely considering the hedonic aspect of wellbeing provides only half the picture. We believe future research should more thoroughly investigate the antecedents and outcomes for "hedonically-happy" and "eudaimonically unhappy" employees.

In addition, the results suggest that employees' self-rated performance is often not reflected in their supervisor's evaluation of their performance. This draws our attention to the importance of considering more than one source of evaluation of work performance in order to obtain valid information about the employees' task performance, extra-role performance, and creativity. It is possible that the disparity in the evaluation of the employees' performance level is due to the fact that employees might be more lenient when self-rating their general performance [38]. It is also possible that, when assessing their own performance, employees' responses reflect not only their past behavior, but also their expectations of current and future behavior [58]. We think it would be interesting to investigate more in depth the reasons for the differences between employees' ratings of their own performance and the ratings given by their direct supervisors.

Finally, the results provide an affirmative response to Research Question 3 about whether there are any demographic variables that play a role as antecedents of the clusters in different operationalizations of the "happy-productive" worker. The existence of differences in the demographic variables between clusters provides yet another way to validate the clusters and the different operationalizations of wellbeing and performance. This means that it is reasonable to expand the study of employees and their different outcomes at work to different patterns of wellbeing and performance, and include alternative configurations of "happy-unproductive" and "unhappy-productive" clusters.

Following the recommendations of Warr and Nielsen [51], we identified a number of situational and personal features associated with membership in each profile when additional measures of wellbeing and performance are considered. Our study examines whether personal features, such as gender, age, and educational level, and situational features, such as sector, type of contract, occupational category, and seniority in the position, play a predictor role in the different profiles obtained, based on the operationalizations of wellbeing (hedonic-eudaimonic) and performance (self- or supervisor- evaluated) considered. The exploratory results provide relevant information showing that occupational category is the only variable with a predictor role in the four models studied. Moreover, another situational variable (sector) and a personal variable (gender) significantly predict the profiles in three of the four models studied. Interestingly, the type of contract is a significant antecedent in the two models in which the supervisors' performance assessment is considered, whereas the educational level is a significant antecedent in the two models where self-assessed performance is considered. More specifically, women, workers in the production sector, and management or highly qualified professionals are more likely to be included in the happy-productive profile, whereas men, workers in the services sector, employees with a low education level, and technicians/administrative workers are more likely to be included in the unhappy-unproductive cluster.

We also identified the main features of employees included in the happy-unproductive profiles. These features differ across the four models studied. The "high hedonic/low performance (self-rated)" pattern is populated more by employees from the services sector with professional training and technician-administrative jobs. In the case of the "high eudaimonic/low performance (self-rated)" pattern, it is mostly composed of women and employees in unqualified or technician/administrative jobs. It is interesting to note that, when we look at the two similar profiles generated using supervisor ratings of performance, the employees with a higher probability of belonging to these patterns (both hedonic and eudaimonic) have permanent contracts and are employed in unqualified or manual jobs. Finally, it is interesting to identify the features that more often characterize employees included in the unhappy/productive profiles. The employees included in the "low hedonic/high performance (self-rated)" profile work in the production sector, have seniority (>5 years) and professional education, and work in technician-administrative jobs. The employees included in the "low eudaimonic/high performance (self-rated)" profile are mostly men over 50 years old working in unqualified-manual

or technician-administrative jobs. Considering this complex picture of personal and situational characteristics associated with the different profiles obtained with different types of wellbeing and performance, we can conclude that the different models are not redundant, and different types of wellbeing and different sources of performance need to be considered to better understand the happy-productive worker model. Further research is needed to confirm the predictive power of the variables studied and extend the study by including other personal and situational variables, in order to better describe the employees in each profile.

In sum, the present study addresses a number of limitations of the happy productive worker thesis, and it sheds light on a number of issues that may clarify the previous inconsistencies of the model. First, this study included both the hedonic and eudaimonic aspects of wellbeing, coinciding with recent conceptualizations of wellbeing as having both pleasurable and meaningful components [3–5]. The identification of the hedonic "happy-productive" and "unhappy-unproductive" patterns coincides with studies indicating that there is a positive relationship between hedonic wellbeing and performance [13–18]. The identification of the "unhappy-productive" pattern agrees with research that shows a negative relationship between positive affect and the dimensions of performance [48]. Simultaneously, the identification of the eudaimonic "happy-productive" pattern supports research that suggests a synergetic relationship between eudaimonic wellbeing and performance [31]. These patterns support previous research showing that daily increases in perceived meaning at work were related to employees' increased focus on tasks and greater exploratory behavior [31]. Second, this study considers different dimensions and sources of evaluation of employees' performance. On the one hand, we operationalize job performance as consisting of different facets or dimensions (i.e., in-role performance, extra-role performance, creative performance) that can help to capture its manifestations. On the other hand, we consider two sources of information about performance: self-rated performance and performance rated by the direct supervisor. Third, the present research analyzes alternative configurations that have not been considered in the happy-productive worker thesis. It shows the importance of these alternative configurations, reflected by the number of employees who belong to the "happy-unproductive" and "unhappy-productive" clusters (over 50% on average), suggesting that the work reality is built on these antagonistic patterns, as well as on the synergetic ones. Thus, antagonistic patterns should not be neglected in future research. Finally, this study has identified a number of individual and situational features that significantly distinguish the different profiles in each of the operationalizations of the happy-productive worker model.
