Daily deep acting toward coworkers: An examination of day- specific antecedents and consequences

Correspondence Hadar Nesher Shoshan, Department of Psychology, University of Mannheim, A5, 6 C-103 D-68159, Mannheim, Germany. Email: hneshers@mail.uni-mannheim.de Summary Emotional labor in coworker interactions is a prevalent phenomenon in everyday work. Yet, it is largely unknown whether it is also a relevant phenomenon, that is, whether emotional labor toward coworkers matters for employee daily work life. Addressing this question, we investigate day-specific antecedents and consequences of coworker-directed emotional labor, especially deep acting. We hypothesized that deep acting toward coworkers will be rewarded by coworkers providing emotional and task support and that this coworker support, in turn, will predict enhanced positive affect at the end of work. Further, we suggest that high morning positive affect enables employees to deep act toward their coworkers in the first place. During a 10-workday diary study, 102 employees answered surveys on 618 days. Multilevel path analysis showed that morning positive affect predicted daily deep acting toward coworkers, which was positively related to emotional (but not task) support from coworkers. Emotional (but not task) support predicted higher end-of-work positive affect and mediated the relationship between deep acting and end-of-work positive affect. Findings highlight the importance of studying deep acting toward coworkers as part of a positive dynamic process that employees can experience at work. Our results bring along vital theoretical and practical implications.

Such knowledge is crucial, however, because prevalence does not equal relevance. Yet if emotional labor toward coworkers is beneficial, advancing its enactment becomes meaningful. Accordingly, we examine predictors and outcomes of daily emotional labor toward coworkers with the present diary study.
The classic emotional labor literature, which focuses on employee-customer interactions (Grandey & Gabriel, 2015), suggests two main types of emotional labor strategies (Grandey, 2000): Deep acting (i.e., modifying inner feelings toward the desired ones) and surface acting (i.e., faking emotions without changing the felt ones).
Recently, Gabriel et al. (2020) confirmed that employees in principle also engage in emotional labor when interacting with their coworkers.
Specifically, the authors showed that different types of emotional labor actors exist in coworker interactions (e.g., deep actors and nonactors), which have different antecedents and consequences. Accordingly, emotional labor matters not only in service interactions but in any social interaction at work (Becker et al., 2018). Yet, emotional labor in coworker interactions has received insufficient research attention (Diefendorff et al., 2020). This limits a thorough understanding of emotional labor overall. In this regard, two points are especially worth noting.
First, Gabriel et al. (2020) showed that deep acting is more common in coworker interactions than surface acting. However, previous studies on emotional labor toward coworkers often focused on surface acting (e.g., Hu & Shi, 2015;Ozcelik, 2013). Hence, especially the knowledge about deep acting toward coworkers is limited. We aim to help gain such knowledge. Accordingly, we focus our study on predictors and outcomes of deep acting toward coworkers. Nevertheless, we also consider surface acting to draw a comprehensive picture of emotional labor between coworkers. Studying emotional labor in coworker interactions, especially deep acting as a generally positive emotional labor strategy (Becker et al., 2018), broadens the understanding both of how good coworker relationships are maintained and of the phenomenon of emotional labor as a whole (Diefendorff et al., 2020). Thus, examining emotional labor toward coworkers with a focus on deep acting is a meaningful undertaking from theoretical and practical viewpoints.
Second, because emotions are dynamic (Grandey & Melloy, 2017), emotional labor in service interactions is commonly assessed as "within-person variations across days" (Grandey et al., 2020, p. 152). Yet, no existing study on coworker-directed emotional labor incorporated daily dynamics. Doing so is necessary, however, to adequately capture emotional labor in coworker interactions (cf. Diefendorff et al., 2020, p. 318). This need is further substantiated given that coworker interactions in fact differ from day to day (Tschan et al., 2005), with some days being "good" and others "bad" (Halbesleben & Wheeler, 2011). Accordingly, we use a diary approach and examine within-person processes related to daily emotional labor toward coworkers.
Altogether, we offer a fresh look into emotional labor research by investigating day-specific predictors and outcomes of emotional labor, especially deep acting, toward coworkers. Precisely, we examine state positive affect as the starting point (Hur et al., 2020) and the ultimate outcome of daily deep acting toward coworkers (see Figure 1). Employee state positive affect is of high value for organizations, as its benefits go "beyond simply feeling good in a given moment" (Vacharkulksemsuk & Fredrickson, 2013, p. 47): State positive affect enhances critical organizational and personal outcomes, such as job performance (Shockley et al., 2012), self-efficacy (Xanthopoulou et al., 2012), and subjective well-being (Diener et al., 2020).
Taken together, our study provides crucial contributions to the emotional labor literature, especially to that on emotional labor between coworkers (Diefendorff et al., 2020). With our focus on deep acting, we contribute to theory and research on the "bright side of emotional labor" (Humphrey et al., 2015). We suggest that deep acting is generally a positive, adaptive strategy to manage workplace relationships (Huang et al., 2015;Humphrey et al., 2015). However, we also acknowledge that deep acting is emotional "labor" (i.e., a resource-demanding behavior; Kammeyer-Mueller et al., 2013). In this regard, we extend theoretical models of emotional labor by considering that employees not only need reasons to engage in deep acting (e.g., display rules; see Grandey et al., 2020), but must also be able to actually deep act. Drawing on broaden-and-build theory (Fredrickson, 2001(Fredrickson, , 2004, we advance the dynamic model of emotional labor (Diefendorff et al., 2020) by including positive affect as a predictor (Hur et al., 2020). Specifically, we suggest that morning positive affect (Binnewies & Wörnlein, 2011), an "energized-to" state F I G U R E 1 Path model of processes related to day-specific emotional labor toward coworkers. Note: All depicted paths were modeled at both the day level (i.e., within person) and the person level (i.e., between person). Solid lines relate to hypothesized paths and core study variables. Dotted lines relate to additionally modeled paths and variables (Lian et al., 2017), provides an employee with the resources needed to engage in deep acting toward coworkers on a given day.
We further broaden theoretical models on emotional labor by considering rewards for deep acting toward coworkers, suggesting that deep acting is worth the resources invested in it. Synthesizing broaden-and-build theory (Cohn & Fredrickson, 2006;Fredrickson, 2004) and the social interaction model of emotion regulation (Côté, 2005), our study bears the potential to uncover how dynamic emotional labor relates to changes in state positive affect.
Specifically, we propose that day-specific deep acting toward coworkers will be socially rewarded in the form of coworker support (Gabriel et al., 2020), which in turn will enhance positive affect. As noted, employee state positive affect is of high value for organizations (Shockley et al., 2012) and employees themselves (Diener et al., 2020) and is, therefore, an important outcome to examine.
Lastly, with our diary study approach, we meet calls to use within-person methods to capture the transient nature of emotional labor and its correlates (Grandey et al., 2020). Our methodological approach allows us to adequately test within-person processes (McCormick et al., 2020) as they are theoretically proposed in the dynamic model of emotional labor (Diefendorff et al., 2020). In particular, our approach enables us to examine how deep acting produces changes in positive affect. It is important to actually examine dynamic, within-person emotional labor because effects found at the betweenperson level (e.g., Gabriel et al., 2020) do not necessarily reflect effects within persons (Gabriel et al., 2019). Our study captures the dynamic nature of daily coworker interactions, advancing knowledge on how positive coworker relationships and thus their positive outcomes (e.g., Becker et al., 2018) might be fostered in everyday work.

| EMOTIONAL LABOR TOWARD COWORKERS
Emotional labor is defined as the regulation of emotions as part of one's work role (Grandey et al., 2020;Hochschild, 1983). More precisely, it describes the regulatory efforts that employees engage in when attempting to display appropriate emotions in social interactions at work (Grandey & Melloy, 2017). Emotional labor has been studied primarily as emotion regulation enacted by service employees when interacting with customers (Grandey et al., 2020). Researchers, however, have started to acknowledge that emotional labor indeed plays a role in any interpersonal interaction at work, including leader-follower interactions (Humphrey et al., 2008), team interactions (Becker & Cropanzano, 2015), and dyadic coworker interactions (Gabriel et al., 2020).
Emotional labor can take the form of two main emotion regulation strategies, deep acting and surface acting. When deep acting, employees attempt to change their inner emotions toward those required. This can be done by different means, for instance, by reappraising the situation (Humphrey et al., 2015). In surface acting, employees do not try to modify their inner emotions. Instead, they suppress or hide their true emotions and fake the emotional expression (Brotheridge & Grandey, 2002;Grandey, 2003). Surface acting is known to have well-being costs for service employees whereas deep acting is unrelated to service employees' well-being overall (Hülsheger & Schewe, 2011; see also Grandey et al., 2020 (Shanock et al., 2013;Thomas et al., 2018). To summarize, until recently, research regarding emotional labor toward coworkers focused on surface acting as a negative workplace phenomenon (cf. Gabriel et al., 2020).
We see this narrow focus as somehow surprising, as it does not consider that coworker relationships meaningfully differ from employee-customer relationships. We believe that especially in closer, long-term relationships, such as coworker relationships, employees will choose more adaptive strategies of emotional labor, that is, deep acting (Humphrey et al., 2015). Indeed, in their study on different types of emotional labor actors, Gabriel et al. (2020) found that although employees do engage in surface acting toward coworkers, deep acting is the more common emotional labor strategy in coworker interactions. Further, they reported initial evidence that deep acting toward coworkers is rewarding, being related to receiving coworker support. Moreover, Gabriel et al. (2020) found that employees with high prosocial motives were more likely to use deep acting toward coworkers. Using a within-person approach, we offer an additional perspective to these between-person findings: We suggest that deep acting toward coworkers will also be rewarding on a day-to-day basis and that state positive affect in the morning before work predicts day-specific deep acting toward coworkers. Such a within-person investigation is important to thoroughly understand emotional labor between coworkers given that emotional labor strategies as well as coworker interactions are highly dynamic, fluctuating from day to day (Diefendorff et al., 2020).

| Morning positive affect as a predictor of daily deep acting toward coworkers
Positive affect describes the psychological experience of feeling active, enthusiastic, and positively energized (Watson et al., 1988).
According to broaden-and-build theory (Fredrickson, 2001), positive affect is a crucial personal resource that promotes a repertoire of beneficial thoughts and actions, including investment in social relationships (Cohn & Fredrickson, 2006). Accordingly, the beneficial effects of positive affect are not restricted to the person experiencing it but transfer to interpersonal behaviors in social interactions (Fredrickson, 2013). Based on this notion, we suggest that positive affect predicts deep acting in coworker interactions.
Deep acting is taxing (Kammeyer-Mueller et al., 2013), because "trying to feel what should be felt" (Liu et al., 2008(Liu et al., , p. 2417) requires the enactment of active emotion regulation techniques (e.g., reappraisal and attention deployment; Alabak et al., 2020;Ashforth & Humphrey, 1993;Brotheridge & Lee, 2003;Humphrey et al., 2015). Hence, deep acting indeed is emotional labor. In principle, employees must be both motivated and able to invest resources to show taxing behaviors such as deep acting. With regard to general motivation, Gabriel et al. (2020) found that employees high on prosocial motives are more likely to use deep acting in coworker interactions. On a given day, however, being principally motivated might not suffice to engage in deep acting when interacting with coworkers on that specific day. Rather, an employee needs to have enough resources available to be able to effectively engage in deep acting on a given day. Accordingly, day-specific resources need to be considered to understand daily emotional labor in coworker interactions. In line with broaden-and-build theory (Fredrickson, 2001), we suggest that state positive affect is such a crucial day-specific resource that allows "employees to expend their energies in the form of more mentally taxing emotion regulation strategies such as deep acting"  et al., 2020). Accordingly, state positive affect, for example, in the morning before work, has been shown to predict a range of dayspecific positive work behaviors such as job performance and work engagement (e.g., Binnewies & Wörnlein, 2011;Ouweneel et al., 2012) as well as customer-directed deep acting (Hur et al., 2020). In line with these findings, we suggest that starting the day with high positive affect will enable employees to engage in deep acting when interacting with their coworkers on that day.
Hypothesis 1. On a given day at work, state positive affect in the morning will be positively related to deep acting toward coworkers.

| Positive consequences of deep acting toward coworkers
According to broaden-and-build theory, behaviors triggered by positive affect help in building social resources (Fredrickson, 2004).
Specifically, the person investing in other people (e.g., by engaging in deep acting) might "receive advice or emotional support" as a response (Cohn & Fredrickson, 2006, p. 41; see also Fredrickson, 2004). Relating to emotional labor, these assumptions are mirrored and further explicated in the social interaction model of emotion regulation (Côté, 2005). The social interaction model of emotion regulation suggests that recipients of emotional labor (i.e., the interaction partners) perceive the emotions expressed in deep acting as more genuine as compared to those expressed in surface acting.
People who perceive genuine emotions in others, in turn, tend to respond more favorably (Becker et al., 2018;Humphrey et al., 2015).
In more detail, we propose that deep acting will be positively related to both emotional and task support from coworkers, which are complementary types of social support (Ducharme & Martin, 2000). Emotional support includes coworkers' expression of personal interest in oneself such as by listening to personal problems. Task support includes coworkers' provision of tangible help to perform work-related tasks, such as assisting with job problems (Settoon & Mossholder, 2002). Looking at differences between types of emotional labor actors (i.e., between-person differences), Gabriel et al. (2020) found that deep actors reported receipt of both higher emotional and higher task-focused support from their coworkers as compared to non-and low actors (i.e., employees who generally show none or low deep acting in coworker interactions).
Similarly, Becker et al. (2018) found that individual deep acting positively predicted perceived team support. We draw on these between-person findings and examine the relationship between deep acting toward coworkers and coworker support on the withinperson level.
As outlined before, deep acting likely fluctuates within person on a daily basis. We suppose that coworkers' reactions to deep acting shown toward them fluctuate accordingly. The dynamic model of emotional labor supports this notion by suggesting that service employees' dynamic emotional labor may impact the corresponding customers' immediate behavior, for example, how they treat the employee (Diefendorff et al., 2020). Translating this view to coworker-directed emotional labor, we expect that an employee's day-specific deep acting in coworker interactions will be answered with higher support provision from their coworkers on the same day.
Empirical evidence for such day-specific interpersonal reciprocity comes from diary studies that show positive behaviors toward coworkers to indeed come along with reciprocal positive behavior on the same day in turn (e.g., Halbesleben & Wheeler, 2011). To summarize, based on broaden-and-build theory (Fredrickson, 2001), the social interaction model of emotion regulation (Côté, 2005), and empirical evidence from within-person studies that showed daily fluctuations in emotional labor and coworker support exchange (e.g., Uy et al., 2017), we hypothesize: Hypothesis 2. On a given day at work, deep acting toward coworkers will be positively related to (a) coworker emotional support and (b) coworker task support.
Broaden-and-build theory suggests that building resources, such as social support, enhances well-being (Fredrickson, 2004;Fredrickson & Joiner, 2002), especially positive affect (Fredrickson, 2013). In fact, broaden-and-build theory as "a naturally social theory" (Cohn & Fredrickson, 2006, p. 42) suggests that positive affect has a strong social part. In other words, positive interpersonal exchange is particularly important to the experience of positive affect (Vacharkulksemsuk & Fredrickson, 2013). This is presumably because positive social exchange satisfies basic psychological needs and strengthens employees' sense of self-worth (Bowling et al., 2004;Xanthopoulou et al., 2008), experiences that are related to positive affect. Evidence from diary studies supports these theoretical notions with regard to day-specific processes: On days when employees receive more support from their coworkers, they experience higher well-being and positive affect (see Sonnentag, 2015;Venz et al., 2020). Accordingly, we suggest that on days when employees receive support from their coworkers, their state positive affect at the end of the workday will be higher.
Hypothesis 3. On a given day at work, (a) coworker emotional support and (b) coworker task support will be positively related to increased state positive affect at the end of work.
Taken together, we hypothesize that daily deep acting will be fostered by morning positive affect and will predict coworker social support, which in turn will predict increased positive affect. Accordingly, coworker support might act as a mediator between deep acting facilitated by positive affect and enhanced end-of-work positive affect.
Hypothesizing an indirect effect of deep acting is in line with the social interaction model of emotion regulation (Côté, 2005), which suggests that deep acting may start an interpersonal process that counteracts the resource consumption inherent to the original effort involved in it. We detail this theoretical notion and suggest that positive affect will not only enable deep acting in the first place but that receiving social support from coworkers in response to one's deep acting might ultimately booster positive affect. This assumption is reflected in broaden-and-build theory (Fredrickson, 2004), which suggests a self-reinforcing process of positive affect. In this process, positive affect first enables a person to invest resources in positive, but effortful, behaviors, which then promote (social) resources, which in turn make the actor experience elevated positive affect later on (Fredrickson, 2013;Vacharkulksemsuk & Fredrickson, 2013). Translated into our research model (see Figure 1), we suggest, first, that day-specific deep acting toward coworkers will be more likely on days when people experience higher morning positive affect and, second, that this affect-facilitated deep acting will result in coworkers providing social support, which in turn will produce an increase in positive affect. Thus, we hypothesize: Hypothesis 4. On a given day at work, there will be a positive indirect effect between deep acting toward coworkers and increased state positive affect at the end of work via (a) coworker emotional support and (b) coworker task support.

| Procedure and participants
We collected data as part of a larger project on interpersonal workplace relationships (Venz & Nesher Shoshan, 2021). Our study began with a general survey, followed by the diary part that covered 2 weeks (i.e., 10 workdays) with two daily measurement points, one before work and one after work. Supervised by the authors, four students recruited study participants within their empirical bachelor theses 1 by posting in social media (e.g., Facebook) and by approaching their personal networks. As an incentive, participants had a chance to win one out of 10 Amazon vouchers worth 10 Euro each when they responded to at least 50% of the diary surveys. To ensure that participants were eligible to take part in this study, we asked them in the registration survey whether they had daily contact with their coworkers and whether they worked at least 20 h per week. We sent eligible participants the general survey via email 1 week before the diary part began. During the diary period, we emailed the morning survey at 5:00 AM (participation was possible until 11:00 AM) and the end-of-work survey at 3:00 PM (participation was possible until midnight

| State positive affect
We assessed state positive affect twice a day (i.e., in the morning survey and the end-of-work survey) with six items of the German version (Krohne et al., 1996) of the Positive and Negative Affect Schedule (Watson et al., 1988) as used in previous diary studies (e.g., Sonnentag et al., 2008). We instructed the participants to report how they feel "at the moment". A sample item is "active."

| Emotional labor toward coworkers
We assessed deep acting toward coworkers in the end-of-work survey with the original three-item deep acting subscale of the emotional labor scale (ELS; Brotheridge & Lee, 2003), adjusted to assess dayspecific deep acting toward coworkers. A sample item is "Today, I really tried to feel the emotions that I should show my coworkers." To draw a full picture of daily emotional labor toward coworkers, we also included surface acting in our analysis. Surface acting is theoretically and empirically related to deep acting (i.e., both are the main types of emotional labor; Grandey, 2000).
We considered surface acting in parallel to deep acting (see Figure 1) to rule out the possibility that emotional labor in general, instead of specifically deep acting, predicts support receipt and endof-work positive affect. In addition, we wanted to make sure that morning positive affect predicts specifically deep acting and not any type of emotional labor (see Hur et al., 2020

| Discriminant validity
We conducted a set of multilevel confirmatory factor analyses using Mplus 7.11 (Muthén & Muthén, 1998 to test the study variables' discriminant validity. We compared a six-factor model (morning positive affect, daily deep acting, daily surface acting, daily coworker emotional support, daily coworker task support, end-of-work positive affect), which we specified on the within-person level and on the between-person level simultaneously, with alternative models. 2 We took a conservative approach in which error terms of the same items measuring morning and end-of-work positive affect were not allowed to correlate (Hermida, 2015). The six-factor model, χ 2 = 1153.920,  .24* .20* Note: Standard deviations at the within-person level (SD w ) and at the between-person level (SD b ) are presented. ICC = intraclass correlation coefficient. Mean Cronbach's α as well as level-specific omega at the within-person level (ω w ) and at the between-person level (ω b ) are presented. Correlations below the diagonal are between-person level correlations (N = 102). Correlations above the diagonal are within-person level correlations (N = 618 days). The correlations were calculated with Mplus 7.11 to account for the nested data structure. *p < .05. **p < .01.

| Data analysis strategy
Considering the multilevel nature of the data (i.e., days nested within persons), we tested our hypotheses with a multilevel path analysis We specified one single path model, that is, we tested all hypotheses within one model. To examine change in positive affect, we used a lagged dependent-variable approach (Gabriel et al., 2019; see also Diefendorff et al., 2020): We controlled for morning positive affect when predicting end-of-work positive affect. We allowed correlations between deep acting and surface acting as well as between emotional and task support. All paths were specified as fixed slopes. We report unstandardized estimates. To make sure that multilevel analysis is appropriate, we tested whether the study variables fluctuate on a daily basis by computing their ICCs (see Table 1). ICCs ranged from 0.46 (end-of-work positive affect) to 0.63 (deep acting toward coworkers). The ICCs support the multilevel analysis approach. In Hypothesis 2, we suggested that deep acting toward coworkers will be positively related to (

| Supplementary analyses
To investigate the option that social support might have triggered deep acting rather than vice versa, we tested for reversed day-specific relationships. 3 We changed the path model (see Figure 1)

| DISCUSSION
We found that on days when employees had higher state positive affect in the morning, they engaged in more deep acting (but not more or less surface acting) toward their coworkers. In addition, on days when employees deep acted more toward their coworkers, they reported receiving more emotional (but not more task) support from their coworkers. Receiving emotional support from coworkers was positively related to increased state positive affect at the end of work.
Deep acting toward coworkers was indirectly related to enhanced positive affect at the end of work via emotional support from coworkers.

| Theoretical and empirical implications
We integrated core notions from broaden-and-build theory (Fredrickson, 2001(Fredrickson, , 2004 into theoretical models on emotional labor (Côté, 2005;Diefendorff et al., 2020) to examine positive processes related to day-specific deep acting in coworker interactions. Focusing on potential benefits of deep acting toward coworkers, we contribute to research that adopts a "bright" view on emotional labor (Humphrey et al., 2015). We advance this view to research on emotional labor in coworker interactions (e.g., Gabriel et al., 2020). Hence, our study has implications for research and theory on both emotional labor and coworker relationships, as we detail in the following.
Previous studies on the predictors of emotional labor toward coworkers focused on aspects of the general work situation (e.g., team politics; Ozcelik, 2013) and rather stable personal motives (Gabriel et al., 2020). Considering that deep acting and its correlates substantially fluctuate from day to day (e.g., Hülsheger et al., 2015;Xanthopoulou et al., 2018), our diary approach provides an essential extension to these studies (cf. Diefendorff et al., 2020). Also, given that deep acting is taxing yet beneficial, it is important to understand what enables employees to deep act in their daily work. With that in mind, we applied broaden-and-build theory and added morning positive affect as an internal, energized-to predictor to the prevalent emotional labor research focus on external, reason-to antecedents (e.g., customer mistreatment; Grandey & Melloy, 2017). We found that morning positive affect indeed predicts at-work deep acting toward coworkers. This finding corroborates theoretical assumptions that positive affect energizes people and equips them with the resources needed to successfully invest in their interpersonal relationships (Fredrickson, 2004). Also, this finding adds to recent evidence regarding a positive relation between state positive affect and deep acting in customer interactions (Hur et al., 2020).
Further, our results indicate that no matter how taxing deep acting may be, it may also be beneficial. In particular, on days when employees engaged in more deep acting toward their coworkers, they received more coworker emotional support and ultimately ended their workday higher on positive affect. These results provide support for the "build" hypothesis in broaden-and-build theory (Vacharkulksemsuk & Fredrickson, 2013) and strengthen the notion that these building processes can appear on a daily basis (e.g., Xanthopoulou et al., 2012). Our study highlights the usefulness of extending the investigation of emotional labor processes to coworker interactions (Diefendorff et al., 2020;Gabriel et al., 2020) as it showed that daily emotional labor can enhance positive affect (and not only impair well-being; Diestel et al., 2015). This finding is important for employees and organizations given that state positive affect benefits several favorable personal and organizational outcomes (e.g., Diener et al., 2020;Shockley et al., 2012).
In this conjunction, our results regarding indirect effects help to explain why deep acting relates to enhanced positive affect and thereby to develop theoretical ideas on self-reinforcing processes as they are proposed in the emotional labor literature: The social interaction model (Côté, 2005) suggests a dynamic process in which an employee's emotional labor triggers receivers' reactions, which in turn shape the employee's strain outcomes. Based on broaden-and-build theory (Fredrickson, 2004), we extended this model and empirically showed that this dynamic emotional labor process can end up in improved well-being, precisely higher state positive affect, via building social resources.
Interestingly, deep acting only predicted emotional support, but not task support, from coworkers. Also, only emotional support, but not task support, predicted enhanced positive affect. We may speculate that this result mirrors a better "match" (Daniels & de Jonge, 2010) of emotional support with the variables under study. On a daily basis, coworkers may more likely respond with emotional behaviors than with task-related behaviors to the emotional labor strategy of deep acting. Similarly, emotional rather than task-related behaviors might matter for positive affect.
It is worth noting that surface acting toward coworkers did not predict any form of coworker support. Our findings indicate that coworkers might be able to identify genuine emotional labor efforts (i.e., deep acting) and to distinguish them from non-genuine ones (i.e., surface acting). In light of these results, we suggest that previous studies' sole focus on surface acting in coworker interactions (e.g., Hu & Shi, 2015;Ozcelik, 2013) might have covered only one part of the emotional labor phenomenon. Precisely, deep acting seems not only to be used more often (Gabriel et al., 2020), but might in fact play a more important role in coworker interactions than surface acting-at least when it comes to positive outcomes. As such, our study highlights the importance of deep acting for building and maintaining good workplace relationships and for promoting well-being.
The results of our additional analysis yet show that surface acting should not be disregarded when studying emotional labor between coworkers either. Indeed, we found that surface acting predicts decreased end-of-work positive affect. This result advances previous findings regarding the detrimental psychological role of surface acting toward coworkers (e.g., Hu & Shi, 2015;Ozcelik, 2013). What is more, receiving emotional support from coworkers negatively predicted surface acting toward them, which indicates that emotional labor and support between coworkers might indeed be parts of a reciprocal social exchange process. Interestingly, whereas positive affect predicted deep acting which then predicted emotional support, the supplementary analysis revealed that morning positive affect did not predict coworker support directly. This finding, although not at the heart of our study, informs the current debate about the causal connectedness between affect, interpersonal work behaviors (e.g., emotional labor), and interpersonal work events (e.g., support receipt; see Venz et al., 2020).
To sum up, our study has important implications for research and theory on emotional labor and coworker relationships. First, the inclusion of a more general psychological theory, namely broaden-andbuild theory (Fredrickson, 2001), allowed us to consider predictors (i.e., state positive affect) and outcomes (i.e., support receipt and state positive affect) that are so far underrepresented in theoretical models on emotional labor (e.g., Diefendorff et al., 2020;Grandey & Melloy, 2017). Second, our diary approach helped to better understand the positive processes related to deep acting by considering its inherently dynamic nature (e.g., Alabak et al., 2020). Third, our findings indicate that the different nature of coworker relationships has to be considered when applying ideas from research on emotional labor in service interactions to research on coworker-directed emotional labor. In particular, further research on emotional labor in coworker interactions should include deep acting and consider positive processes related to it.

| Limitations and future directions
Despite its important implications, our study is not free of limitations.
First, the self-report nature of our data raises concerns regarding common method bias (Podsakoff et al., 2012). We tried to minimize these effects by separating the measurement of morning positive affect as the predictor and deep acting as an outcome. In addition, we directed the measure of positive affect as an outcome to a different point in time: When measuring deep acting toward coworkers and coworker support, we asked about processes that happened during the workday. When assessing end-of-work positive affect, we asked about feelings at the exact moment. Finally, we controlled for morning positive affect when predicting end-of-work positive affect, thus predicting changes in positive affect (Gabriel et al., 2019). To take further steps in reducing possible biases, future studies may survey meaningful others. For example, coworkers might rate emotional labor efforts directed toward them. However, one must keep in mind that deep acting might not always be accurately identified by others (Groth et al., 2009). Therefore, targets of deep acting might perceive it as genuine behavior, neglecting the effort that was made to achieve it. Hence, asking employees to report the amount of social support they provided to their coworkers might be more fruitful. Yet, it is essential to also include the individual perception of receiving support, because it is the perception that is responsible for positive outcomes (Binnewies et al., 2009).
Second, our study design does not allow to infer causality, in particular regarding the relationship between emotional labor toward coworkers and coworker support provision. We based our assumptions on Gabriel et al.'s (2020) supposition that deep acting predicts social support because emotional labor has "exchange value" (Hochschild, 1983, p (Parker et al., 2010).
Accordingly, future studies could integrate different types of predictors of day-specific emotional labor toward coworkers (e.g., positive affect, self-efficacy, and social motives) as well as investigate their interactive effects (e.g., Schmitt et al., 2016). For example, future studies may investigate if being highly socially motivated on a given day facilitates the translation of morning positive affect into deep acting toward coworkers.
Another interesting question regards the role of specific aspects of the coworker interactions themselves. For example, one could imagine that in some interactions with coworkers it will be harder (or easier) to deep act (Humphrey et al., 2015), depending on the emotion expected or appropriate in the specific situation (Scott et al., 2020). It is important to note that deep acting in coworker interactions might not always involve positive emotions. This is different from most service interactions, which usually do (i.e., service with a smile; Rafaeli & Sutton, 1987). For example, coworkers might be expected to show sympathy as an empathic concern (e.g., after a coworker got harsh supervisor feedback). Accordingly, future studies will gain from examining specifics of the coworker interactions in which emotional labor is enacted. Particular resources that employees need in order to show specific emotions toward their coworkers are also worth studying (Gabriel et al., 2020;Scott et al., 2020).
Lastly, future studies might examine whether processes that promote positive affect in the morning (e.g., recovery; Sonnentag, Venz, & Casper, 2017) translate into deep acting toward coworkers at work (see Hur et al., 2020). Such research could provide important knowledge on how employees' emotional labor processes in coworker interactions and their potential positive consequences may be positively influenced and fostered.

| Practical implications
Our study supports the notion that emotional labor in coworker interactions is an important process in organizational life (Gabriel et al., 2020;Tschan et al., 2005). Our results suggest that day-specific deep acting toward coworkers comes along with emotional support receipt followed by higher positive affect, but surface acting does not.
Thus, one implication of our study is to foster deep acting in coworker interactions. To this end, organizations might actively seek to encourage deep acting among their employees (Becker & Cropanzano, 2015).
In this regard, employees may gain from training emotion-regulation skills (e.g., Buruck et al., 2016;Denny & Ochsner, 2014). Specifically, it might be worthwhile to instruct employees on how to use techniques such as positive reappraisal (e.g., Pogrebtsova et al., 2018) to promote deep acting.
In addition, our results suggest that any means that promote positive affect in the morning might translate into the use of deep acting in coworker interactions. Accordingly, another practical implication is to boost morning positive affect. Evening recovery experiences (e.g., Hur et al., 2020), good sleep, and doing sports are all predictors of high positive affect and energy in the morning (Sonnentag, Venz, & Casper, 2017). Therefore, we advise employees to take part in these well-being promoting activities. To this end, recovery training (e.g., Hahn et al., 2011) might be a fruitful approach.

| CONCLUSION
Acknowledging the unique nature of coworker relationships, this study addresses the question if emotional labor toward coworkers has relevance for employee daily life. Adopting a dynamic approach, this study found that morning positive affect enables employees to invest in their coworkers by engaging in deep acting-indicating that deep acting indeed is "labor" and thus requires resource investment.
Revealing the "bright side" of emotional labor, however, we showed that deep acting toward coworkers is worth the effort, being rewarded in terms of coworker emotional support provision, which in turn enhances positive affect. The study highlights the importance of investigating dynamic processes related to coworkerinteractions, especially emotional labor between coworkers.

ACKNOWLEDGEMENTS
We thank Martin Ganz, Marina Kamburova, Andrea Laura Macavei, and Tabea Anna Schiele for their invaluable help in collecting the data.

ENDNOTES
1 We asked each of the four students to recruit 30 participants. This is in line with recommendations for data collection done by students, which emphasize that the goal for the number of acquired participants should be feasible (Demerouti & Rispens, 2014).
2 For all of these models, the comparative fit index (CFI) and the Tucker-Lewis index (TLI) are not informative, because the RMSEA value of the baseline model (χ 2 = 5781.154, df = 552, p < .001) was 0.124 and CFI and TLI values will be too small when the RMSEA of the baseline model is smaller than 0.158 (Kenny, 2020). Therefore, we do not report CFI and TLI values.