Workplace Emotions: The Role of Supervision and Leadership

Joyce E. Bono, Hannah Jackson Foldes, Gregory Vinson, and John P. Muros University of Minnesota

In this experience sampling study, the authors examined the role of organizational leaders in employees’ emotional experiences. Data were collected from health care workers 4 times a day for 2 weeks. Results indicate supervisors were associated with employee emotions in 3 ways: (a) Employees experienced fewer positive emotions when interacting with their supervisors as compared with interactions with coworkers and customers; (b) employees with supervisors high on transformational leadership experi- enced more positive emotions throughout the workday, including interactions with coworkers and customers; and (c) employees who regulated their emotions experienced decreased job satisfaction and increased stress, but those with supervisors high on transformational leadership were less likely to experience decreased job satisfaction. The results also suggest that the effects of emotional regulation on stress are long lasting (up to 2 hr) and not easily reduced by leadership behaviors.

Keywords: emotional regulation, mood, stress, job satisfaction, leadership

Over the past 2 decades, there has been a growing interest in affective and emotional experiences at work (Brief & Weiss, 2002), including interest in the role of mood and emotions in employee motivation (Erez & Isen, 2002), job performance (Law, Wong, & Song, 2004), creativity (George & Zhou, 2002), and job attitudes (Weiss & Cropanzano, 1996). Emotional regulation at work has also received attention, especially given evidence that regulating emotions is associated with cardiovascular system ac- tivation (Gross & Levenson, 1993, 1997), stress, emotional ex- haustion (Pugliesi, 1999), and physical symptoms such as head- aches (Schaubroeck & Jones, 2000).

There has also been interest in emotions in the leadership domain. Most influential theories of transformational and charis- matic leadership (e.g., Bass, 1985; Conger & Kanungo, 1998; House, 1977; Shamir, House, & Arthur, 1993) posit emotional links between leaders and followers, yet there is little empirical research linking managers and their leadership behaviors to em- ployees’ emotions. A few studies have been conducted, but these studies have not fully elucidated emotional links between leaders and followers because they either manipulated leader emotions (Sy, Coˆte ́, & Saavedra, 2005); used simulated leaders and follow- ers (Bono & Ilies, 2006); or used a single-time, single-source survey to assess follower emotions and leader behavior (McColl- Kennedy & Anderson, 2002). Therefore, the primary purpose of our study was to examine the effects of supervisors and managers on employees’ emotions in a natural work setting. First, we ex-

Joyce E. Bono, Hannah Jackson Foldes, Gregory Vinson, and John P. Muros, Department of Psychology, University of Minnesota.

Hannah Jackson Foldes is now employed at Personnel Decisions Inter- national, and Gregory Vinson is now employed at The Center for Victims of Torture.

This article was supported in part by a Grant-in-Aid of Research and Artistry, University of Minnesota.

Correspondence concerning this article should be addressed to Joyce E. Bono, Department of Psychology, University of Minnesota, 75 East River Road, Minneapolis, MN 55455. E-mail: [email protected]

amined the direct effects of supervisors and managers and their leadership behaviors on employees’ emotional experiences (i.e., experienced emotions, expressed emotions, and emotional regula- tion). Second, we examined the extent to which managers’ lead- ership behaviors can buffer employees from the negative conse- quences of emotional regulation. We used an experience sampling methodology and within-person analyses, which allowed us to focus on within-person covariations and the effects of leaders’ behaviors on these covariations. Figure 1 displays the associations of interest in this research.

Leaders and Employee Emotions

Despite widespread beliefs that supervisors are a key source of bad moods at work, there is little empirical research documenting these effects. There is an extensive literature on workplace factors associated with employee well-being and stress (see Danna & Griffin, 1999, for a review), and a key assumption of this research is that supervisors affect employees’ emotional experiences. From a theoretical standpoint, there are at least two reasons why em- ployees may feel increased anxiety during supervisory interac- tions. First, supervisors are the individuals who directly evaluate performance, and thus interactions with supervisors may increase anxiety about performance, which can be directly evaluated by the supervisor during interactions. Second, Ryan and Deci (2000) provided compelling evidence that individuals have a need for autonomy, which tends to be limited in the workplace by super- visors. Interactions with supervisors allows for closer observations of employee behavior, which may leave employees feeling mon- itored and controlled (George & Zhou, 2001), leading to feelings of irritation. Furthermore, research by Diefendorff and Richard (2003) suggests that supervisors’ expectations may lead employees to constrain their emotional expressions, which also leads to neg- ative affect.

Empirical research is also consistent with the notion that super- visors may negatively influence employee emotions. Glaso and Einarsen (2006) found four affective factors relevant to the

1357

upervisor–subordinate relationship, three of which were negative (i.e., frustration, violation, and uncertainty). Fitness (2000) inter- viewed employees about their experiences of anger and found that unfair treatment by supervisors, which tended to remain unre- solved, was a key source of employee anger. A recent experience sampling study directly examined the link between mood and supervisory interactions (Miner, Glomb, & Hulin, 2005). This research revealed that employees rated 80% of their interactions with their supervisors as positive and only 20% as negative; however, the effects of negative interactions on employee mood were, in general, 5 times stronger than the effects of positive interactions. These findings suggest that even though most super- visory interactions are positive, the overall net effect of interac- tions with supervisors may be slightly negative because of the stronger effects of negative interactions on employee mood.1

Hypothesis 1: Employees experience more negative and fewer positive emotions when interacting with supervisors than when interacting with coworkers or clients and custom- ers.

In addition to the general mood effects of interactions with supervisors, we also expected supervisors’ typical behavioral pat- terns (i.e., leadership behaviors) to influence employees’ emo- tional experiences. Abusive supervisors (Tepper, 2000) might be expected to elicit frustration, anxiety, and anger. In contrast, su- pervisors who use transformational leadership behaviors may elicit feelings of happiness and enthusiasm in employees. Transforma- tional and charismatic leadership theories (Bass, 1985; Conger & Kanungo, 1998; House, 1977) are unique in the emphasis they place on the emotional aspects of leadership. Shamir et al. (1993) focused on emotional attachment to the leader and the emotional arousal of followers, and Bass (1985) suggested that transforma- tional leadership “has an intense emotional component” (p. 36). Testing these notions, Bono and Ilies (2006) recently linked ratings of transformational leadership to managers’ use of positive affect words (e.g., good, happy) in their written and verbal communica- tions. Lab studies also suggest that positive emotions experienced and expressed by leaders may be transferred to employees through the process of emotional contagion. Sy et al. (2005) reported that leader’s positive mood, which was induced in their study, influ- enced group affect and performance. Bono and Ilies extended this work by linking leaders’ expressions of enthusiasm to followers’ positive mood states. Whereas these results are promising, there has been no research examining the effects of leadership behaviors on employee mood throughout the day in a natural work setting.

Hypothesis 2: There is a positive association between super- visors’ transformational leadership behaviors and employees’ experiences of positive emotions throughout the workday.

Leaders and Employee Emotional Regulation

The preceding arguments imply direct effects of supervisors and their leadership behaviors on employees’ experienced emotions. A second potential influence of leader behaviors involves leaders as a coping resource for employees who regulate their emotions at work. Emotional regulation refers to processes by which individ- uals choose which emotions they express, relative to those they experience, in either a controlled or an automatic way (Gross, 1998). People habitually regulate their emotions and emotional displays to conform with norms and expectations of the workplace, as well as job role demands. Initial theoretical work by Hochschild (1979) and the more recent free trait theory (Little, 2000) both view emotional regulation as harmful to employees because it involves acting without authenticity. Empirical evidence supports this notion, as researchers have found that suppressing emotions has both physiological and cognitive costs, including cardiovascu- lar activation and decreased memory for social information (see Gross, Richards, & John, 2006, for a review), and is also associ- ated with psychological strain (e.g., stress, emotional exhaustion, and burnout) and physical complaints (Schaubroeck & Jones, 2000). Therefore, we expected the following:

Hypothesis 3a: Emotional regulation (both hiding negative emotions and faking positive emotions) is negatively associ- ated with job satisfaction within individuals.

Hypothesis 3b: Emotional regulation (both hiding negative emotions and faking positive emotions) is positively associ- ated with experienced stress within individuals.

Whereas these hypotheses are not unique in linking emotional regulation to stress and job satisfaction, existing research has examined only the association between employees’ general ten-

1 Although there is considerable evidence that positive and negative mood represent independent dimensions of affect experienced by individ- uals in general (Watson & Clark, 1997), when participants are asked to report their current mood, a single dimension (i.e., happiness vs. unhappi- ness) best represents mood data (Diener & Emmons, 1984). However, because we wished to examine the potential differential effects of regulat- ing positive and negative emotions, we considered positive and negative mood separately, expecting them to be negatively correlated.

dencies to regulate emotions at work and their general tendencies to experience stress and decreased job satisfaction. In contrast, we hypothesized and tested direct links between individual episodes of emotional regulation, job stress, and job satisfaction and their covariation within individuals. Using a repeated measures, within- subjects design, we sought to determine whether individuals’ lev- els of stress and job satisfaction fluctuate throughout the course of the workday in conjunction with emotional regulation. By exam- ining variability around each individual’s mean level, we con- trolled for differences between individuals in the tendency to experience, remember, and report emotional experiences and stress. This design also allowed us to examine the association between emotional regulation and stress over time.

In linking supervisors’ and managers’ leadership behaviors to emotional regulation, we drew from two theoretical perspectives. Self-determination theory (Ryan & Deci, 2000) and the goal self- concordance model (Sheldon & Elliott, 1999) both posit links between authentic self-expression and individual well-being. Hochschild (1983) suggested that when employees regulate their emotions at work, they experience feelings of depersonalization and separation from self; however, it appears that this is not always the case, as certain conditions may weaken or eliminate the neg- ative effects of emotional regulation. In a review of the literature, John and Gross (2004) reported that healthier patterns of social functioning and greater well-being were linked to emotional reg- ulation for individuals who changed the way they thought about an emotional event as compared with individuals who changed only their emotional expressions. In addition, Ashforth and Humphrey (1993) suggested that employees who identify with their work are more likely to feel authentic even when conforming to role expec- tations, such as demands for emotional regulation. This is where leadership can play a role. In both lab and field studies, Bono and Judge (2003) demonstrated that managers’ transformational lead- ership behaviors can influence employees’ identification with their work. They also found that transformational leadership was a positive predictor of the extent to which employees felt that their work activities were self-congruent and consistent with their own interests and values. This suggests that individuals who report to transformational leaders, compared with those who do not, will be more likely to identify with their work and to change the way they think about role-required emotional regulation, thereby leading them to experience less stress and greater job satisfaction.

Managers who engage in transformational leadership behaviors may also provide greater social support for their employees. Social support has been recognized as important to the experience of emotional management at work because providers of social sup- port (e.g., family, coworkers, and supervisors) are thought to enable employees to cope better with job stressors and increase their sense of control (Abraham, 1998; Zapf, 2002). Abraham reported that when employees experienced high social support, no association between emotional regulation and job satisfaction was found; however, under conditions of low social support, emotional labor negatively affected satisfaction. Transformational leaders are typically characterized as empathetic, and the behavioral dimen- sion of individualized consideration explicitly describes leaders as attending to and supporting the individual needs of followers. Such leaders may be able to help employees cope with emotional regulation in more effective and less psychologically draining ways by helping employees to understand why and how positive

emotional expressions contribute to work goals (e.g., Shamir et al., 1993). Because they engender trust, transformational leaders are also uniquely positioned to assist employees in coping with emo- tional labor. Therefore, we expected the following:

Hypothesis 4a: Supervisors’ transformational leadership be- haviors moderate the link between emotional regulation and job satisfaction, such that when supervisors engage in more transformational leadership behaviors, the negative associa- tion between emotional regulation and job satisfaction is weaker than when supervisors engage in less transformational leadership behaviors.

Hypothesis 4b: Supervisors’ transformational leadership be- haviors moderate the link between emotional regulation and stress, such that when supervisors engage in more transfor- mational leadership behaviors, the positive association be- tween emotional regulation and stress is weaker than when supervisors engage in less transformational leadership behav- iors.

Method

Participants

Participants were 57 employees of an ambulatory health care organization, randomly selected from nonmanagement employees participating in an organization-wide survey. As is common in health care organizations, participants were predominantly women (94%). They were, on average, 41 years old (SD 10 years). Eighty-six percent were Caucasian, 9% were African American, and 5% were of Asian origin. Participants’ average job tenure was 5.24 years. Participants worked in family practice clinics (n 14), administrative offices (n 10), and a billing office (n 33), and held a broad variety of jobs ranging from nurse, medical assistant, and lab technician in the family practice clinics; patient services support, case analyst, and account follow-up specialist in the billing office; and accountant, case manager, and human resource specialist in the administrative office.

Procedure

Participants provided two types of data: survey data and expe- rience sampling data. Paper surveys were used to gather general job satisfaction and stress data. Experience sampling data were obtained via a handheld computer that participants carried for 2 weeks and included momentary stress, job satisfaction, and affec- tive experiences. A third type of data—leadership behaviors of participants’ supervisors—was obtained from an organization- wide survey.

As part of an organization-wide survey that we administered, all employees (N 365; 309 nonmanagement and 56 manage- ment) reported on the leadership behaviors of their direct su- pervisor. Surveys, completed in small group sessions by 253 (73%) nonmanagement and 56 (100%) management employees, were anonymous (employees recorded only the name of their supervisor) and were returned to us in unmarked, sealed enve- lopes. Only aggregate data were provided to organization ex- ecutives. One month following the organizational survey, we invited 73 employees to participate in the study. Participants

were quasi-randomly selected, proportionately from each loca- tion, by entering the names of those who participated in the organization-wide survey into a spreadsheet, sorting by first name, and then choosing every other person on the list until the desired number of participants was achieved.

Fifty-eight of the 73 employees contacted (79%) affirmed they would attend an information session, 6 (8%) were interested but could not make it to a scheduled information session, and 9 (12%) indicated they were not interested in participating in this research. At the orientation session, participants were informed that the organization’s management had endorsed their participation in the study and that they were allowed to respond to surveys during normal working hours. They were also informed that they would receive $25 as payment for participation. After obtaining this information, 57 of the 58 employees agreed to participate.

After agreeing to enroll in the study, participants completed a survey, after which they were trained in the use of a personal digital assistant (PDA). For the experience sampling surveys, participants were signaled four times each day, Monday through Friday, for 2 weeks. The signal intervals were modified to accom- modate participants’ regular working hours (e.g., 7 a.m. to 3 p.m. or 8 a.m. to 4 p.m.). Participants were signaled once during each 2-hr segment of their 8-hr workday. Participants were given 10 min to respond to each signal, after which the survey was no longer available. They were told to respond to the surveys based on the emotions and attitudes they were experiencing “immediately be- fore the beep went off.” In addition, they were instructed to respond to as many signals as possible during the workday without compromising their job performance or service to patients or customers. Because we were not able to program the PDAs to stop signaling on weekends (i.e., Saturday and Sunday), we asked participants to leave the PDA either at work, in their car, or at home in a location where the signals would not be annoying. Data were collected for 10 working days.2

At each signal, participants were asked to report whether they were at work. As we were attempting to collect data based spe- cifically on emotional regulation in a social context, we next asked whether they were interacting with others when the signal sounded. Only surveys with yes responses to this item were in- cluded in this research. Next, they were asked to report on their level of stress and job satisfaction at that moment. Finally, they were asked to report on their affective experiences at that moment, including the extent to which they were feeling six emotions

(happiness, enthusiasm, optimism, anxiety, irritation, and anger) and the extent to which they were faking the three positive emo- tions (happiness, enthusiasm, and optimism) and hiding the three negative emotions (anxiety, irritation, and anger). Hereinafter, we refer to experienced emotions as emotions experienced or felt emotions and to faking and hiding emotions as emotional regula- tion. The 12 affective items (i.e., 6 experienced emotion items and 6 emotional regulation items) appeared in random order for each survey.

At the end of 2 weeks, participants attended a debriefing session, completed a second paper survey regarding general stress at work, returned their handheld computers, and were paid $25. At this time, they were also asked the name of their supervisor so that we could obtain reports on their supervisor’s leadership behaviors from the organizational survey. Table 1 displays the source and format of the data, along with the data collection timeline.

Measures

The leadership behaviors of participants’ supervisors were measured using the 20-item Mul- tifactor Leadership Questionnaire (MLQ – Form 5x; Avolio, Bass, & Jung, 1995).3 The MLQ is the most frequently used measure of transformational leadership and has exhibited high reliability and convergent and discriminant validity (Avolio et al., 1995). Re- sponses were evaluated on a 5-point scale ranging from 1 (not at all) to 5 (frequently, if not always). Consistent with other research- ers whose interest is the general construct of transformational leadership behaviors, we combined items to form a single score for

each supervisor (e.g., Lim & Ployhart, 2004).
Job satisfaction. We assessed momentary job satisfaction by

asking participants to rate their agreement with the statement “At this very moment, I am fairly satisfied with my job” on a 5-point scale (1 strongly disagree, 5 strongly agree). Ilies and Judge

2 The PDAs were programmed to signal twice during the evening hours as well. The optional evening surveys were used to collect pilot data for another project. Thirty-five individuals responded to evening surveys, which were not included in this study.

3 The MLQ, Form 5x (Copyright 1995 by Bernard Bass and Bruce Avolio) is used with permission of Mind Garden, 1690 Woodside Road, Suite 202, Redwood City, CA 94061.

Supervisors’ leadership behaviors.

(2002; Judge & Ilies, 2004) have found considerable variation in job satisfaction over the course of the workday, which our mo- mentary measure was designed to capture. We also obtained a measure of general job satisfaction, using five Brayfield–Rothe items (e.g., “Most days I am enthusiastic about my work”; see Bono & Judge, 2003; Brayfield & Rothe, 1951). Responses for these five items were on the same 5-point scale used for momen- tary job satisfaction.

Work status and interactions. Immediately after being sig- naled by the PDA, participants were asked to report whether they were currently at work (yes or no). Next, they responded to the following question: “When the beep went off, were you interacting with any of the following?” 1 (supervisor), 2 (coworker), 3 (cus- tomer/client), 4 (family/friends), 5 (no one), 6 (other).

Affective experiences. Although many taxonomies of affect and emotion exist in the literature, we were constrained by our method (i.e., surveys of only 1–2 min in length) to select a small number of affect terms in attempting to cover the full range of work-relevant affect and emotion. After reviewing the literature (Ortony & Turner, 1990; Russell & Feldman Barrett, 1999; Shaver, Schwartz, Kirson, & O’Connor, 1987) and conducting a pilot study of workers who reported how frequently they expressed each of 10 emotions at work, we selected three positive (happiness, enthusiasm, and optimism) and three negative (anxiety, anger, and irritation) work-relevant emotions. We excluded some basic emo- tions (e.g., fear and love) and low-activation affect terms (e.g., contentment and calm), deeming them of low relevance to our study. Participants reported the degree to which they felt the six selected emotions at each signal.

Because Glomb and Tews (2004) illustrated the importance of distinguishing between felt emotions and the act of emotional regulation, we also asked participants about faking (i.e., expressing an emotion they did not feel) the three positive emotions (happi- ness, enthusiasm, and optimism) and hiding (i.e., feeling an emo- tion they were not expressing) the three negative emotions (anxi- ety, irritation, and anger). Responses for these items were evaluated on a 7-point scale (1 none at all, 7 an intense amount) for feeling, hiding, and faking.

Stress. We assessed momentary stress by asking participants to respond to the statement “At this very moment, I am experi- encing stress.” This measure was intended to capture variability in stress throughout the day. We also obtained an overall, global measure of stress, using Motowidlo, Packard, and Manning’s (1986) four-item measure. Responses for both momentary and overall stress items were on the same 5-point scale used for job satisfaction.

Data Analyses

Prior to conducting our analyses of interest, we examined sev- eral properties of our data. We excluded data from 3 participants who had responded to fewer than three signals over the course of the 2 weeks (1 because of job termination, 1 because of PDA failure, and 1 for an unknown reason). Therefore, the total possible number of responses was approximately 2,160 (4 responses per day 10 workdays 54 participants 2,160); however, because of scheduling the drop-off and pick-up of PDAs, a few participants responded to surveys on Days 11 and 12, raising the total possible to 2,266). Participants responded to 1,983 signals (approximately

88% of the total possible), with an average of 37 responses per participant (ranging from 11 to 42). We excluded data from re- sponses when participants were not at work (n 287 episodes) and for interactions with family or friends (n 52 episodes), resulting in 1,653 work-related responses. Our goal was to exam- ine emotional regulation in a social context. However, in many cases participants were alone when the signal sounded (n 764; 48% of work-related responses).4 Thus, our final data set consisted of 889 responses (52% of total at-work responses; average of 15 per participant) from employees who were at work and involved in a work-related interaction at the time of the signal.

To form a single leadership score for the supervisor of each participant, we aggregated the leadership survey responses of all employees who completed a survey for the target supervisor (av- erage n 5 reports per supervisor). Aggregating leadership reports across followers was deemed justifiable in these data by a signif- icant ICC-1 value of .29 ( p .01) and an ICC-2 value of .72. The ICC-2 value can be interpreted as the reliability of the aggregated measure of leadership. An average rwg .84 across groups (James, Demaree, & Wolf, 1984; assuming a slight negative skew in the data [Bono & Judge, 2003]) further supports aggregation.

We also created several composite emotion variables. The three positive (enthusiasm, happiness, and optimism) and three negative (anger, irritation, and anxiety) emotions were highly correlated within individuals for both felt emotions and emotional regulation (average r .87, .77, .93, and .82, for positive experienced, negative experienced, positive faked, and negative hidden, respec- tively). Therefore, we computed four composite emotion variables: positive emotions experienced, negative emotions experienced, positive emotions faked, and negative emotions hidden. Principal- components analysis provided support for creating these compos- ites, as we found that a single factor explained most of the variance (ranging from 70%–82%) in each set of related emotions. We also found that the two types of emotional regulation were highly related (i.e., faking positive emotions and hiding negative emo- tions tended to co-occur; r .88, p .01). Because we were interested in possible differential effects of the two types of emo- tional regulation, we included faking positive and hiding negative in our analyses. However, we also formed a total emotional reg- ulation score by averaging the two types of emotional regulation.

Our hypotheses address two levels of analysis. Within-subject analyses must be used to link emotional regulation within an individual with momentary variation in stress and job satisfaction (Hypothesis 3). In these analyses, data from each individual are centered around that individual’s mean score, effectively control- ling for mean differences between individuals in emotional regu- lation, job satisfaction, and stress. Cross-level (between- and within-subjects) analyses are used to test our leadership hypothe- ses (Hypotheses 1, 2, and 4), which posit associations between supervisory behavior and employees’ emotional experiences. Therefore, we analyzed our data using multilevel modeling tech- niques (i.e., HLM 5; Raudenbush & Bryk, 2002).

4 A comparison of affective experiences when interacting with others versus alone revealed no significant differences in experienced positive and negative emotions or in emotional regulation.

LEADERSHIP AND EMPLOYEE EMOTIONS

1361

1362

BONO, FOLDES, VINSON, AND MUROS

Results

Means, standard deviations, and correlations are presented in Table 2, revealing that participants generally experienced more positive emotions than negative emotions at work and tended to hide negative emotions only slightly more than they faked positive emotions (mean hide negative 1.67, and mean fake positive 1.50). Mean levels of job satisfaction and stress were similar in the momentary, PDA data (Level 1) and in the global, overall measure (Level 2). We note a high correlation between negative mood and emotional regulation (r .78, .92, and .88, all ps .01, for faking positive, hiding negative, and total emotional regulation, respec- tively). These high correlations are consistent with the notion that positive emotions are valued (and negative emotions devalued) in organizations, leading employees to suppress negative emotions.

Leaders and Employee Emotions

Hypotheses 1 and 2 address the role of supervisors in employ- ees’ emotional experiences. We first examined the link between employees’ affective experiences and interaction partner (e.g., supervisor vs. customer–client). Analysis of variance results (see Table 3) revealed significant differences in affective experiences based on interaction partner. Specifically, participants reported the most positive emotions when interacting with their coworkers (mean positive emotions experienced 3.75) and fewest positive emotions when interacting with their supervisors (mean positive emotions experienced 3.05). This was true for each of the individual positive emotions, which we do not report in Table 3, and the positive emotion composite, F(3, 886) 5.74, p .01. Thus, Hypothesis 1 was supported for positive emotions; however, interaction partner was significantly related neither to the experi- ence of negative emotions nor to employees’ emotional regulation.

Next, we examined the direct effects of supervisors’ transfor- mational leadership behaviors on employees’ momentary emo- tional experiences, testing the theory that transformational leaders influence employee emotions. In this analysis we linked the su- pervisor’s typical leadership behaviors to employees’ emotions across all interactions, including those with customers and cowork- ers. Table 4 presents coefficients linking supervisors’ leadership behaviors to employees’ experienced emotions and emotional reg-

Table 2

ulation. As hypothesized (Hypothesis 2), employees who work for supervisors rated high on transformational leadership reported experiencing more positive emotions throughout the course of their workday ( .54, p .01). This held true for each of the individual positive emotions (results not reported in Table 4), as well as the positive emotions composite. Leadership behaviors were not linked to participants’ experience of negative emotions or to reports of emotional regulation.

Leaders and Employee Emotional Regulation

Means, Standard Deviations, and Intercorrelations Between Level 2 Variables and Aggregate of Level 1 Variables

In addition to main effects of supervisors’ leadership behaviors on employees’ affective experiences, we also hypothesized a moderating effect for transformational leadership. We examined the data to deter- mine whether emotional regulation covaries with stress and job satisfaction within individuals and over time. Results in Table 5 (upper portion: momentary data) addressed the link between mo- mentary emotional regulation, job satisfaction, and stress. Results revealed a significant, negative association between both types of emotional regulation and job satisfaction ( –.18 and –.18, p .01, for hiding negative and faking positive, respectively) and a significant positive link between both types of emotional regula- tion and stress ( .40 and .35, p .01, for hiding negative and faking positive, respectively). These results suggest that regardless of how often certain individuals regulate their emotions, individual episodes of emotional regulation are associated with increased stress (above the mean for a given individual) and decreased job satisfaction (below the mean for a given individual). This was true for both faking positive and hiding negative emotions, supporting

Hypotheses 3a and 3b.
A unique advantage of experience sampling research is that we

can examine the link between emotional regulation at one point in time (t) and stress or job satisfaction for that individual at the next random signal (t 1), allowing us to draw stronger causal infer- ences and to determine how long these effects last. Results in Table 5 (lower portion: lagged momentary data) showed that the association between emotional regulation and stress extended over time. Although weaker than the concurrent effects, results indi- cated a significant positive association between emotional regula- tion at one time point and stress approximately 2 hr later. No

Emotional regulation, job satisfaction, and stress.

experience increased stress and decreased job satisfaction. Further-

5 Several participants in our sample reported to the same supervisor, creating dependencies in the leadership data that were not addressed in our hierarchical linear modeling analyses and may have potentially produced biased tests of significance. Thus, we repeated our moderator analyses involving leadership (see Table 6) using only 1 participant per leader. This reduced our sample size (from 54 to 27) but eliminated the dependencies. All significant effects reported in Table 6 remained significant in this subsample.

more, although the effects of emotional regulation on job satisfac- tion appear to be short-lived, the effects of emotional regulation on stress are more long lasting (e.g., 2 hr). Managers who use trans- formational leadership behaviors buffer employees who regulate their emotions from decreased job satisfaction, but not from in- creased stress. Given these results, two important questions re- main: (a) How do these results contribute to our understanding of “the leader’s role in the production of moods and emotions” (Brief & Weiss, 2002, p. 289)?, and (b) How do these results contribute to our understanding of the role of managers in workplace emo- tional regulation?

With respect to leadership and emotions, results of this study demonstrate the powerful role that managers and supervisors have on employee emotions. Transformational leadership theory sug- gests emotional links between leaders and followers but offers little explanation of how such emotional links contribute to lead- ership effectiveness. Based on theories of primitive emotional contagion, lab studies (e.g., Barsade, 2002; Sy et al., 2005) have documented the fleeting effects of a leader’s positive mood on the mood of workers. Given evidence in this study that managers’ leadership behaviors have an ongoing influence on employee optimism and enthusiasm, it may be time to look beyond primitive emotional contagion as the primary process by which leaders influence follower emotions and organizational outcomes. Our

Table 6

Hierarchical Linear Modeling Estimates of the Moderating Effect of Transformational Leadership

results suggest that managers’ transformational leadership behav- iors may have broad, deep, and long-lasting effects on individual employees and the organization as a whole. Beyond their imme- diate effects on employee mood, the positive emotions elicited by transformational leaders have the potential to influence the overall work climate and customer satisfaction. Employees who are opti- mistic are more likely to invest the time to help others (Lee & Allen, 2002) and can be expected to persist in work tasks – even in the face of difficulty (Seligman & Schulman, 1986). In the same vein, research by Pugh (2001) linked positive emotional displays by employees to customer evaluations of service quality.

Our study also contributes to theory and existing research on emotional regulation. What we know from the existing literature is that individuals who tend to regulate their emotions are also less happy with their jobs and more stressed, although the link with job satisfaction varies across studies. Our results show that discrete episodes of emotional regulation are also associated with de- creased job satisfaction and increased stress, even for individuals who do not frequently regulate their emotions. This is an important finding given that emotional regulation theory is focused on the negative effects of sustained and frequent emotional regulation. However, it is also important to note that we did not find any time-lagged effects of emotional regulation on job satisfaction. Thus, the effects of discrete episodes of emotional regulation on

job satisfaction appear to be very short-lived (less than 2 hr). This is in contrast to the effects of emotional regulation on stress, which remained significant 2 hr later. Hochschild’s (1983) seminal re- search suggested that individuals who must frequently regulate their emotions on the job will experience depersonalization and stress. Although our results do not speak to cumulative effects over time, they do suggest that even for individuals who rarely regulate their emotions, doing so on a single occasion is associated with increased stress, which may last for several hours. An important topic for future research is to examine the long-term, sustained effects of emotional regulation on employee health.

In organizational research on emotional regulation, it is common to do as we did and consider both job satisfaction and stress as important outcomes of emotional regulation; however, our results suggest that different processes may underlie the effects of emo- tional regulation on these two outcomes. We found that emotional regulation was associated with stress 2 hr later, but the effects of emotional regulation on job satisfaction were more fleeting (i.e., we found no time-lagged effects). We also found that managerial behavior buffered employees from the effects of emotional regu- lation on job satisfaction but not on stress, suggesting that the processes linking emotional regulation with job satisfaction and stress may differ. The link between emotional regulation and job satisfaction may be cognitive in nature. One way that managers provide support for employees is by enhancing internalized self- regulation and drawing attention to outcomes associated with emotional regulation (e.g., greater service for customers). There- fore, when employees regulate their emotions, they may con- sciously decide that it is worth doing because it helps them to perform better in a job they care about and identify with. Thus, during emotional regulation they may feel momentarily unhappy with their job because it demands inauthentic emotional expres- sions, but these effects are fleeting (and do not occur for employ- ees who report to transformational managers) because employees understand how emotional regulation contributes to their job.

In contrast, the effects of emotional regulation on stress may be, at least in part, physiological. Empirical evidence shows that emotional regulation leads to cardiac arousal (see Gross & Lev- enson, 1993), which is linked to perceptions of stress (see Blas- covich & Tomaka, 1996). It may be that the association between emotional regulation and stress is stronger, lasts longer, and is less amenable to reduction through managerial support because it has a physiological basis. Clearly, we can only speculate about the processes by which emotional regulation affects stress and job satisfaction, but this is an important area for future research. If our theoretical propositions are correct, managers may be able to help employees more by reducing or eliminating the need for emotional regulation (perhaps by encouraging and supporting authenticity) rather than attempting to buffer them from the negative effects of emotional regulation after it occurs.

Our study highlights the need for follow-up research. First, it is important to establish causality by manipulating both emotional regulation and leadership behavior in the lab. Some work has been done in this regard by Bono and Vey (2007), who linked emotional regulation to increased heart rate and stress. Controlled laboratory settings would also be useful in assessing the longevity of the effects of emotional regulation episodes and sustained emotional regulation. Second, it is important to determine the direct cause of emotional regulation. Although our theoretical model suggests that emotional regulation causes stress, our methods did not allow us to establish causality. Indeed, it is possible that certain events in the workplace may cause both emotional regulation and stress. For example, in cases in which workers are asked to be pleasant and enthusiastic with customers, a rude customer may be the source of emotional regulation, which then leads to stress. Alternatively, a rude customer may be the source of both emotional regulation and stress, creating the illusion of a causal association between them. It is also plausible that an unpleasant interaction with a patient or customer would have both direct effects on stress and indirect effects, through emotional regulation. One possible next step

1366

BONO, FOLDES, VINSON, AND MUROS

would be to use experience sampling methods to catalogue events that lead to emotional regulation in the natural work environment and determine how these events affect both emotions and stress.

Our study is unique in investigating leadership as an important factor in employees’ experienced emotions and in the emotional regulation process; it has some methodological strengths and lim- itations worth noting. Concerning the former, we used a longitu- dinal design in conjunction with an experience sampling method- ology. In doing so, we were able to collect daily emotional regulation data across multiple locations and jobs as emotional regulation occurred. Thus, we avoided retrospective and recall biases that occur in reporting emotional experiences. Our within- person design also allowed us to confidently link specific events of emotional regulation to stress and job satisfaction, free from the biasing effects of individual differences. Furthermore, our mea- sures of leadership behavior and employee emotional experiences were relatively independent of each other. Our study also ad- dressed the call for more research to be conducted in health care settings (Ehler, Major, & Fletcher, 2003).

Despite these strengths, our study also has some limitations. First, although our participants held a variety of jobs, we did not have the sample size to examine subgroups of employees in service versus nonservice support jobs. Second, our sample was drawn from a single organization. Thus, it is possible that charac- teristics of the work climate, shared by all participants, influenced the experience of both positive and negative emotions and overall levels of emotional regulation. If the entire work climate was particularly supportive or abusive, our results may under- or over- estimate the moderating effects of leadership behaviors. Third, our sample—though representative of the larger organization and the health care industry—was largely female, and we cannot be certain that the associations we found would hold for men. Fourth, it is possible that the PDA signals may have annoyed participants, causing increased irritation and stress. If this is true, the effects were small because mean stress from PDA data was not higher than global reports of stress.

Considered as a whole, our results highlight the importance of supervisors in employees’ emotional experiences. Our results are provocative in that they suggest the possibility of different theo- retical explanations (and therefore the need for different manage- rial interventions) for the effects of emotional regulation on stress and job satisfaction. Our study also extends existing theory on leader–follower emotional links by looking beyond the primitive emotional contagion process. We hope this study will motivate future longitudinal research in naturalistic settings along with experimental research in the laboratory, including both psycholog- ical and physiological measures of employee moods, emotional regulation, and stress over time.

References

Abraham, R. (1998). Emotional dissonance in organizations: Antecedents, consequences, and moderators. Genetic, Social, and General Psychology Monographs, 124, 229–246.

Ashforth, B. E., & Humphrey, R. H. (1993). Emotional labor in service roles: The influence of identity. Academy of Management Review, 18, 88 –115.

Avolio, B. J., Bass, B. M., & Jung, D. I. (1995). Multifactor Leadership Questionnaire technical report. Redwood City, CA: Mind Garden.

Barsade, S. G. (2002). The ripple effects: Emotional contagion and its

influence on group behavior. Administrative Science Quarterly, 47,

644 – 675.
Bass, B. M. (1985). Leadership and performance beyond expectations.

New York: Free Press.
Blascovich, J., & Tomaka, J. (1996). The biopsychosocial model of arousal

regulation. Advances in Experimental Social Psychology, 28, 1–51. Bono, J. E., & Ilies, R. (2006). Charisma, positive emotions, and mood

contagion. Leadership Quarterly, 17, 317–334.
Bono, J. E., & Judge, T. A. (2003). Self concordance at work: Toward

understanding the motivational effects of transformational leaders.

Academy of Management Journal, 46, 554–571.
Bono, J. E., & Vey, M. A. (2007). Personality and emotional performance:

Extraversion, neuroticism, and self-monitoring. Journal of Occupational

Health Psychology, 12, 177–192.
Brayfield, A. H., & Rothe, H. F. (1951). An index of job satisfaction.

Journal of Applied Psychology, 35, 307–311.
Brief, A. P., & Weiss, H. M. (2002). Organizational behavior: Affect in the

workplace. Annual Review of Psychology, 53, 279–307.
Conger, J. A., & Kanungo, R. N. (1998). Charismatic leadership in

organizations. Thousand Oaks, CA: Sage.
Danna, K., & Griffin, R. W. (1999). Health and well-being in the work-

place: A review and synthesis of the literature. Journal of Management,

25, 357–384.
Diefendorff, J. M., & Richard, E. M. (2003). Antecedents and conse-

quences of emotional display rule perceptions. Journal of Applied Psy-

chology, 88, 284–294.
Diener, E., & Emmons, R. (1984). The independence of positive and

negative affect. Journal of Personality and Social Psychology, 47,

1105–1117.
Ehler, M. L., Major, D. A., & Fletcher, T. D. (2003). Applying I-O to

medicine: Making the case that it can be done and that it should be done.

The Industrial–Organizational Psychologist, 41, 50–54.
Erez, A. I., & Isen, A. M. (2002). The influence of positive affect on the components of expectancy motivation. Journal of Applied Psychology,

87, 1055–1067.
Fitness, J. (2000). Anger in the workplace: An emotion script approach to

anger episodes between workers and their superiors, coworkers, and

subordinates. Journal of Organizational Behavior, 21, 147–162. George, J. M., & Zhou, J. (2001). When Openness to Experience and Conscientiousness are related to creative behavior: An interactional

approach. Journal of Applied Psychology, 86, 513–524.
George, J. M., & Zhou, J. (2002). Understanding when bad moods foster creativity and good ones don’t: The role of context and clarity of

feelings. Journal of Applied Psychology, 87, 687–697.
Glaso, L., & Einarsen, S. (2006). Experienced affects in leader–subordinate

relationships. Scandinavian Journal of Management, 22, 49–73. Glomb, T. M., & Tews, M. J. (2004). Emotional labor: A conceptualization

and scale development. Journal of Vocational Behavior, 64, 1–23. Gross, J. J. (1998). Antecedent- and response-focused emotion regulation: Divergent consequences for experience, expression, and physiology.

Journal of Personality and Social Psychology, 74, 224–237.
Gross, J. J., & Levenson, R. W. (1993). Emotional suppression: Physiol- ogy, self-report, and expressive behavior. Journal of Personality and

Social Psychology, 64, 970–986.
Gross, J. J., & Levenson, R. W. (1997). Hiding feelings: The acute effects

of inhibiting negative and positive emotion. Journal of Abnormal Psy-