The Role of Humor in the Relationship Between Transactional Leadership Behavior, Perceived Supervisor Support, and Citizenship Behavior

Transcription

1 613018JLOXXX / Journal of Leadership & Organizational StudiesTremblay and Gibson research-article2015 Article The Role of Humor in the Relationship Between Transactional Leadership Behavior, Perceived Supervisor Support, and Citizenship Behavior Journal of Leadership & Organizational Studies 2016, Vol. 23(1) The Authors 2015 Reprints and permissions: sagepub.com/journalspermissions.nav DOI: / jlo.sagepub.com Michel Tremblay 1 and Megan Gibson 2 Abstract This study, building on uncertainty management theory, examines the role of humor use by the supervisor and team members in the relationship between leader behaviors, perceived supervisor support, and citizenship behavior. Data were collected from a sample of 284 employees working in nine small organizations. The results show that weak contingent reward leaders are viewed as more supportive when they use constructive and self-defeating humor styles extensively as opposed to aggressive humor, whereas skillful contingent reward leaders are perceived as less supportive when they use constructive and self-defeating humor extensively, and more supportive when they favor an aggressive humor style. Laissez-faire leaders are viewed as less supportive when they use aggressive humor extensively. The results provide only partial support for the buffer effect of constructive humor and the undermining influence of aggressive humor style. Finally, whereas offensive coworker humor is negatively related to organizational citizenship behavior, the results do not provide significant evidence that coworker humor moderates the influence of perceived supervisor support on organizational citizenship behavior. We conclude by discussing the theoretical contributions and practical implications of our findings. Keywords humor, leadership, perceived supervisor support, citizenship behavior Introduction Humor as a business topic is gaining increasing importance among academics (Mesmer-Magnus, Glew, & Viswesvaran, 2012; Westwood & Johnston, 2013). Organizational humor has frequently been associated with leadership (Avolio, Howell, & Sosik, 1999; Decker & Rotondo, 2001; Hughes, 2009; Vecchio, Justin, & Pearce, 2009), and is even considered one of the primary components in the manager subordinate relationship (Cooper, 2008). Leadership and humor share a common point: both include constructive and offensive behaviors (Avolio, Bass, & Jung, 1999; Martin, Puhlik-Doris, Larsen, Gray, & Weir, 2003). Supervisor humor styles range from constructive humor styles, which encompass affiliative and self-enhancing humor, to offensive humor styles, which include aggressive and self-defeating humor styles (Martin et al., 2003). Similarly, leadership behaviors have been distinguished in two ways: constructive leadership behaviors, which encompass transformational and contingent reward leader behavior (CRLB), and offensive leader behavior, which generally includes laissez-faire leader behavior (LFLB) and passive management by exception (Lee & Jensen, 2014; Skogstad, Einarsen, Torsheim, Aasland, & Hetland, 2007). Research in leadership suggests that supervisor leadership styles often coexist with humor styles (Avolio, Howell, et al., 1999; Vecchio et al., 2009). For example, a supervisor may display high CRLB in response to good employee performance and adopt an aggressive humor style, while another contingent reward leader may likely use a constructive humor style. Furthermore, a laissez-faire leader may be more likely to use an aggressive humor style, whereas another passive leader may be more likely to prefer a constructive humor style. Despite the increased attention devoted to the effect of conflicting leader behaviors on employee outcomes (Duffy, Ganster, & Pagon, 2002; Nahum-Shani, Henderson, Lim, & Vinokur, 2014; Thau, Bennet, Mitchell, & Marrs, 2009), to best of our knowledge, no study has specifically examined how a mixture of constructive and offensive humor and leadership styles affects perceived supervisor support (PSS) and motivation to display organizational citizenship behavior 1 HEC Montréal, Montreal, Quebec, Canada 2 Cardinal Health Canada, Pointe-Claire, Quebec, Canada Corresponding Author: Michel Tremblay, HEC Montréal, 3000, Côte-Sainte-Catherine Road, Montreal, Quebec, Canada, H3T 2A7. michel.tremblay@hec.ca

2 40 Journal of Leadership & Organizational Studies 23(1) (OCB). Building on uncertainty management theory (Uchino et al., 2012; van den Bos, 2001), we argue that inconsistency in humor and leadership styles will result in high employee uncertainty and subsequently in weaker positive reactions than consistent humor and leadership behaviors. More specifically, we will examine the influence of constructive and offensive humor styles in the relationship between CRLB, LFLB, and PSS, and subsequently on OCB. Several reasons explain the choice of CRLB and LFLB. First, while transformational leadership is probably the dimension that has received the most attention from scholars, some researchers have recently argued that the positive effects of CRLB on employee attitudes and behavior have been significantly underestimated (N. Podsakoff, Podsakoff, & Kuskova, 2010; Wang, Oh, Courtright, & Colbert, 2011). Furthermore, there is strong evidence to suggest that LFLB produces negative outcomes (Bass & Riggio, 2006; Chênevert, Vandenberghe, & Tremblay, 2015; Kelloway, Mullen, & Francis, 2006; Wang et al., 2011), and that laissez-faire behavior is quite prevalent in the workplace (Aasland, Skogstad, Notelaers, Nielson, & Einarsen, 2010), and its relationship to humor has rarely been studied. Second, CRLB and LFLB represent the two ends of the transactional leadership continuum (Judge & Piccolo, 2004), but more important, these behaviors have been associated in the literature with a higher level of uncertainty and ambiguity (e.g., N. Podsakoff et al., 2010; Skogstad et al., 2007; Tremblay, Vandenberghe, & Doucet, 2013). Despite the growing interest in workplace humor, very few studies have identified the mechanisms through which humor can be manifested. Gkorezis et al. (Gkorezis, Hatzithomas, & Petridou, 2011; Gkorezis, Petridou, & Xanthiakos, 2014) provided evidence that the use of humor by leaders may affect psychological empowerment and leader member exchange relationships. However, they did not evaluate the effect of such mediators on individual performance, nor the main and moderating effects of humor styles on performance. Arendt (2006) tested the mediated role of follower sense of humor and self-efficacy in the relationship between the leader s use of humor and creative performance. Although these studies helped improve our understanding of the antecedents and outcomes of humor, other important mediators and outcomes were ignored. It is increasingly acknowledged that leaders exert their influence through a number of psychological mechanisms and behaviors. PSS and OCB are among the factors most likely to increase team and organizational performance (Bass & Riggio, 2006; Organ, Podsakoff, & McKenzie, 2006; N. Podsakoff, Whiting, Podsakoff, & Blume, 2009). While the impact of leadership behavior on citizenship performance via PSS has received both theoretical attention and empirical support (e.g., Eisenberger & Stinglhamber, 2011; Organ et al., 2006), very few studies have examined whether and how leaders constructive and offensive humor styles might influence such relationships. Humor scholars assert that workplace humor is not the sole preserve of the formal hierarchy. Informally, humor may be present within work teams regardless of the quality of leaders, and humor may represent a significant facet of the culture (Clouse & Spurgeon, 1995; Holmes & Marra, 2002; Romero & Pescosolido, 2008). Recent research on the PSS citizenship behavior relationship (e.g., Chiaburu, Lorinkova, & Van Dyne, 2013; Wang et al., 2011) has demonstrated that some important variations exist across studies, suggesting that some contextual variables may act as potential moderators. One important question is how coworkers constructive and offensive humor plays an influencing role in the relationship between PSS and OCBs. Building on uncertainty theory and social support theory (S. Cohen & Wills, 1985; Duffy et al., 2002; Eisenberger, Stinglhamber, Vandenberghe, Sucharski, & Rhoades, 2002), we expect the link between PSS and OCB to be positive and stronger when coworkers use constructive humor (enhancing effect), and probably negative when followers perceive both low support and high offensive humor usage by coworkers. The recurrent negative outcomes, generally associated with undermining behavior such as offensive humor (e.g., Martin et al., 2003; Mesmer-Magnus et al., 2012), point to the relevance of verifying whether inconsistent signals in a cross-domain may interact (Duffy et al., 2002), and more particularly, whether the use of offensive humor by supervisor (same domain) has a stronger undermining effect on the influence of PSS on OCB than does the use of offensive humor from coworkers (cross-domain). The current study can make some important contributions to the literature. First, we believe that this study contributes to improving the understanding of the role of humor in leadership effectiveness, and especially of the impact of coexistence of constructive and offensive humor and leadership styles. Our research can also expand knowledge of the situational factors modifying the links between transactional leadership and individual outcomes. The second contribution of this study is to deepen our knowledge of the effect of coworkers humor and of the role of inconsistent signals across domains. To our knowledge, no study has assessed how support from supervisor can buffer the negative influence of offensive humor from coworkers on citizenship performance. Third, by exploring two key sources of constructive and offensive humor in the workplace, the current study can enhance our understanding of the relative importance of humor by leaders and coworkers. Given that the relationship between PSS and OCB has already been established, this study focuses on the moderating role of humor in the workplace. The general model used in our study is presented in Figure 1. Humor and Leader Styles There is no real consensus on what constitutes humor. Romero and Cruthirds (2006) use the term organizational humor to describe any form of communication that provokes emotions in an individual, group, or organization. It

3 Tremblay and Gibson 41 Figure 1. General research model. Note. OCB = organizational citizenship behavior. is widely accepted that individuals can adopt a variety of humor styles. Martin et al. (2003) distinguish four humor styles, two positive and two negative, and focused on self versus others. The positive styles are affiliative and selfenhancing. The former is credited with fostering team spirit and a sense of belonging among team members, while the latter is known to reduce the level of stress experienced by the user of humor, making it possible to maintain a positive outlook. The aggressive and self-defeating humor styles proposed by Martin et al. (2003) represent the dark side of humor. The aggressive style is associated with the use of sarcasm, ridicule, and humor that is demeaning to others. Someone who is constantly making others laugh at his own expense is said to have a self-defeating style. This form of humor, which consists in laughing at oneself, is often used as a means to seek acceptance or love from a group. This study will focus on the humor styles used by a transactional leader. In this framework, the contingent reward leader is described as one who succeeds in motivating employees by skillfully using contingent rewards. This leader sets clear and precise objectives, clarifies what is expected from employees, provides constant feedback, and rewards his subordinates according to their outcomes (N. Podsakoff et al., 2010). The transactional dimension also includes active and passive management by exception (Judge & Piccolo, 2004). Active management by exception involves monitoring the follower s performance and administering some form of interventions when the follower s performance is problematic, whereas passive leadership involves reacting to correct problems, waiting until the problem amplifies or hoping that the problem will resolve itself. The third type is the laissez-faire leader. This leader avoids making decisions and is rarely available when employees need him or her (Judge & Piccolo, 2004). This style of leadership also implies not meeting the legitimate expectations of the subordinates (Skogstad et al., 2007). We will not examine active management by exception because we found no theoretical justification for including it in our study. Research has shown that laissez-faire leadership and passive leadership are strongly correlated: both of these leadership styles are negatively related to the same dependent variables, and generally have formed a single passive leadership dimension in studies (Hinkin & Schriesheim, 2008; Judge & Piccolo, 2004). Consequently, the present study will focus on the laissez-faire dimension. Theoretical Background and Hypotheses Theories of Management Uncertainty and Social Support. The uncertainty perspective is based on the notion that inconsistency ties give rise to relational uncertainty, which leads to a more negative outcome. Uchino et al. (2012) posit two broad types of relationships: positive and negative. These relationships can be separable or occur together. They classify relationships as socially positive (high positivity/low negativity), socially averse (low positivity/high negativity), indifferent (low positivity/low negativity), and ambivalent (high positivity/high negativity). According to Uchino et al. (2012), ambivalent relationships, reflecting high inconsistent behaviors, may be harmful because such relationships or behaviors may be unpredictable and stressful. From the uncertainty-management perspective, the mixture of supportive and undermining styles from the same supervisor reflects inconsistency behaviors (Uchino et al., 2012). The use of an aggressive humor style by a high contingent reward leader or by a laissez-faire leader is more likely to be perceived by subordinates as inconsistent or ambivalent behavior. Such inconsistency creates uncertainty regarding the supervisor s intentions and trustworthiness. If not well managed, such uncertainty might affect followers attitudes and well-being (a) by threatening the coherence of their perceived self-evaluation (De Cremer, 2003; Lind & van den Bos, 2002), (b) by reducing their ability to predict and exert control over their work environment (Nahum-Shani

4 42 Journal of Leadership & Organizational Studies 23(1) et al., 2014; Thau et al., 2009), and (c) by signaling a failure to maintain a good relationship with one s supervisor (Nahum-Shani et al., 2014). Consistent with the uncertainty management theory, Thau et al. (2009) found that a positive relationship between an abusive supervisor and organizational deviance was stronger when the authoritarian management style was low (high uncertainty) rather than high (low uncertainty). Uchino et al. (2012) found that being in ambivalent relationships (positive and negative behaviors) was the only significant predictor of physical health indicators. Duffy et al. (2002) provided evidence that high undermining and support from the same source was associated with negative outcomes. Nahum-Shani et al. (2014) argue that supervisor support exacerbates the adverse impact of supervisor undermining on employee health and/or on employees with low self-esteem. Social support is a metaconstruct referring to a set of functional resources (S. Cohen & Wills, 1985). Social support theory suggests that (a) informational support provides individuals with relevant data or knowledge that helps them understand and cope with problems; (b) instrumental support is a provision of financial, material, and technical resources that directly help employees perform; (c) emotional support or self-esteem support is information provided to signal that a person is accepted and appreciated; and (d) appraisal support is considered as a perceptual belief that one has the ability, means, or aptitude to overcome the perceived difficulty. Perceived organizational support (POS) theory (Eisenberger & Stinglhamber, 2011) suggests that employees form a global belief concerning the extent to which the organization values their contributions and cares about their well-being. POS is interpreted as assurance that aid will be available from the organization when needed. According to the POS theory, resources or support received are more highly valued if they are provided in a discretionary manner and fulfil subordinates important socioemotional needs and motives (Rhoades & Eisenberger, 2002). Because supervisors act as agents of the organization, employees tend to see their behaviors as indicative of the extent to which the organization values their contributions and cares about their well-being (Eisenberger, Huntington, Hutchison, & Sowa, 1986). The organizational literature generally suggests that supervisor support signals to employees they are cared for and appreciated. Supervisors behaviors are an important source of guidance, assistance, and feedback to subordinates as they accomplish their tasks at the workplace. Because supervisors normally have more frequent day-to-day contact with subordinated employees, they have more opportunities than the organization to demonstrate support (or lack thereof) to employees (Maertz, Griffeth, Campbell, & Allen, 2007). Social exchange theorists argue that when supervisors provide a useful discretionary support, the subordinates are more likely to reciprocate by performing extrarole actions (Shanock & Eisenberger, 2006). Leadership, Perceived Supervisor Support, and Influence of Leaders Humor Styles. PSS is generally defined as the feeling that a leader values followers contributions and cares about their well-being (Eisenberger & Stinglhamber, 2011; Kottke & Sharafinski, 1988). The impact of leadership on PSS has been well documented (Rhoades & Eisenberger, 2002). According to the path-goal theory (House, 1971), important supportive behavior includes the leader s clarification of what is expected from the subordinates and how they should accomplish their work effectively (Organ et al., 2006). When a contingent reward leader is able to clarify goals and expectations of the employees, this increases employees role clarity. Similarly, when contingent reward leaders allocate rewards fairly, this increases the subordinates feelings that their contributions are valued (N. Podsakoff et al., 2010; Tremblay et al., 2013). Humor scholars (Decker & Rotondo, 2001; Vinton, 1989) assert that leaders who use constructive humor are more effective at assigning and explaining tasks and making guidelines more acceptable. They can make the presentation of objectives more interesting and provide a persuasive message regarding the importance of accomplishing them (Lyttle, 2001). A mixture of high CRLB and constructive humor styles is likely to be perceived by subordinates as consistent supportive behaviors. However, a mixture of high CRLB and offensive humor behavior can produce the opposite effect (Decker & Rotondo, 2001). Offensive behaviors by a high contingent reward leader could be perceived as ambiguous signals, and an ambivalent relationship (Uchino et al., 2012), indicating that the support received was not deserved. Some studies found that supportive behaviors from supervisors exacerbated the harmful effects of supervisors offensive behaviors (Duffy et al., 2002; Hobman, Restubog, Bordia, & Tang, 2009). Receiving support in the context of a negative relationship is perceived by the target person as unexpected and conflicting, increasing the salience of offensive behavior and its adverse implications. Research also indicates that negative events are weighted more heavily in social judgment than positive experience or events (Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001), which explaining why undermining events such as offensive or inappropriate humor may produce more intense reactions in the presence of inconsistent supervisor behaviors (Uchino et al., 2012). We therefore formulate the following hypotheses: Hypothesis 1: Constructive humor demonstrated by leaders moderates the relationship between CRLB and PSS, such that the relationship will be positive and PSS will be higher under conditions of high CRLB and high constructive humor styles. Hypothesis 2: Offensive humor demonstrated by leaders moderates the relationship between CRLB and PSS, such that the relationship will be negative and PSS will be lower under conditions of high CRLB and high offensive

5 Tremblay and Gibson 43 humor styles than under conditions of high CRLB and low offensive humor styles, and even lower than under conditions of low CRLB and high offensive humor styles. Laissez-faire leaders are frequently described as indifferent; they refrain from intervening and fail to support and monitor performance (Kelloway et al., 2006). According to Skogstad et al. (2007), laissez-faire leadership style implies not only a lack of presence but also not meeting the legitimate expectations of subordinates and addressing their concerns. Exposure to laissez-faire leaders has been negatively associated with subordinate satisfaction and with leader effectiveness (Judge & Piccolo, 2004) and with high role conflict and ambiguity (Skogstad et al., 2007). Subordinates exposed to such poor leadership are more likely to experience high uncertainty. For laissez-faire leaders, positive humor may not have an enhancing role because these supervisors are known to be poor supportive leaders. Their use of constructive humor is likely to produce high uncertainty and an ambivalent relationship. Use of positive humor to avoid responsibilities can be viewed as an attempt to shift attention away from the work to be done (Avolio, Howell, et al., 1999). Offensive humor would certainly have an adverse impact on this leader s supportive credibility. An laissez-faire leader who uses offensive humor may be perceived as a clown in the workplace, and, as Avolio, Howell, et al. (1999) have said, might be seen as adding insult to injury (p. 221). Consistent with Uchino et al. s (2012) classification, such a relationship may be viewed as adverse. We therefore propose the following hypotheses: Hypothesis 3: Constructive humor shown by leaders moderates the relationship between LFLB and PSS such that the relationship will be negative and PSS will be lower under conditions of high LFLB and high constructive humor styles than under conditions of high LFLB and low constructive humor styles. Hypothesis 4: Offensive humor demonstrated by leaders moderates the relationship between LFLB and PSS, such that the relationship will be negative and PSS will be lower under conditions of high LFLB and high offensive humor styles, than under conditions of high LFLB and low offensive humor styles, and will be even lower than under conditions of low LFLB and high offensive humor styles. Perceived Supervisor Support, Organizational Citizenship Behavior, and Impact of Coworkers Humor. According to Organ et al. (2006), OCB represents an individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and in aggregate, promotes the efficient and effective functioning of the organization (p. 3). Primarily, this relationship centers on Blau s (1964) social exchange theory and Gouldner s (1960) reciprocity norm, both of which stipulate that support actions received from the organization or from its representatives create a sense of obligation to reciprocate the favors, which is expressed through behaviors beneficial to the supportive source. While the PSS OCB relationship has been well documented (e.g., LePine, Erez, & Johnson, 2002), the role of coworkers humor in this relationship has never been examined. Clouse and Spurgeon (1995) argue that positive humor in a team may minimize differences in status among employees, facilitating socialization, inclusion, and cooperation. Building on the cross-domain perspective (e.g., Duffy et al., 2002) and the stress-buffering hypothesis (S. Cohen & Wills, 1985), these theoretical frameworks suggest that social support in one domain (e.g., supervisor) can compensate for social undermining in another social domain (e.g., coworkers). This buffering hypothesis states that a less desirable exchange or relationship in one domain may be compensated by a supportive relationship in another domain. We believe that team members working in a humorous team context will be motivated to perform OCB even if they receive weak support from their supervisor, because the humorous work environment may fulfil important needs such as inclusion and belonging. Consistent with the claim that humor may play an important role in buffering the negative impact of job-related stress or threatening events, Karl and Peluchette (2006) found that the negative impact of emotional exhaustion on job satisfaction was significantly lower for those who experienced a high level of fun at work. A lack of support from the supervisor may be experienced as a stressful situation, so it is possible that high constructive humor among coworkers may mitigate the negative impact of low supervisor support on employee OCB. Hence the following hypothesis: Hypothesis 5: Constructive humor among team members moderates the relationship between PSS and OCB, such that the relationship will be positive and the level of OCB will be higher under conditions of low supervisor support and high constructive humor among team members, than under conditions of low supervisor support and low constructive humor among team members. We cannot rule out the possibility that highly supportive leadership may compensate for coworker high offensive humor because that leader may provide the resources and rewards to help and motivate subordinates to perform citizenship behavior. This favorable treatment from a supervisor should produce a felt obligation to help such supervisor in reaching goals, notably by performing behaviors that go beyond prescribed responsibilities (Shanock & Eisenberger, 2006). However, offensive humor or other comparable forms of mistreatment in a workgroup is more likely to lead to cynicism among team members, to reinforce power and status differences, to exclude certain individuals and to

6 44 Journal of Leadership & Organizational Studies 23(1) offend others, and contributes to the degradation of the work environment, and a decrease in OCB (Cruthirds, Wang, & Romero, 2013; Penhalion, Louis, & Restubog, 2013; Susa, 2002). These outcomes could lead to a deterioration of interpersonal relationships and create more conflicts, which can strongly mitigate the positive impact of PSS on OCB. These conflicting arguments preclude our proposing specific hypotheses regarding the direction of the moderating effect of coworkers offensive humor in relationship between PSS and OCB. Research Question 1: Does supervisor support buffer or exacerbate the adverse effect of team members offensive humor use on citizenship behavior? Method Sample A sample of employees belonging to different unit levels in nine small organizations was invited to participate in this study. Over 450 questionnaires were distributed, and 284 participants agreed to complete our survey, representing a 63% response rate. This response rate is higher than the average rate for studies that utilize individual data (Baruch & Holtom, 2008). Two thirds of the respondents were women (66%); the average age of the participants was 34, 62% of them had been with their organization for more than 2 years, and 15% had worked for the same supervisor for more than 2 years. Measurement of Variables Leadership Behaviors. Leadership behaviors were extracted from P. M. Podsakoff, MacKenzie, Moorman, and Fetter s (1990) scale, and rated on a 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). The contingent reward leadership dimension was measured using four items (e.g., congratulates me for my significant contributions ). The Cronbach alpha coefficient of the construct was.88. The laissez-faire dimension was evaluated using three items (e.g., avoids making decisions ). However, the.68 Cronbach alpha coefficient for this measure was somewhat low. Supervisor s Humor Styles. We used 13 items from the Humor Styles Questionnaire by Martin et al. (2003). We performed an exploratory factor analysis with varimax rotation and we found three clear interpretable solutions. The first six-item constructive humor factor that emerged explained 30.1% of the variance (eigenvalue = 0.36), and included the three affiliative and the three self-enhancing items. For the constructive humor style, examples of items used are My supervisor enjoys making people laugh (affiliative) and If my supervisor is feeling depressed, he can usually cheer himself up with humor (self-enhancing). Reliability analyses were performed and indicated an acceptable level of internal consistency for constructive humor (α =.80). The second supervisor humor factor included three items and explained 21.2% of the variance (eigenvalue = 3.6). This construct is related to an aggressive humor style, and the reliability analyses show a high level of internal consistency for this offensive component humor style (α =.85). Examples of the items for this scale are If someone makes a mistake, my supervisor will often tease him about it and When telling jokes or saying funny things, my supervisor is usually not very concerned about how other people are taking it. The third supervisor humor factor comprised three items and explained 11.1% of variance (eigenvalue = 1.4). This construct captured the self-defeating dimension, and the reliability analyses indicated acceptable internal consistency for this offensive component humor style (α =.71). A typical example of a self-defeating humor style was my supervisor lets people laugh at him or make fun at his expense more than he should. Colleagues Humor. To assess humor within a work team, we used the Organizational Humor Scale by Susa (2002) to create a 5-point Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). Because of space constraints, we chose a total of six of the original items to measure humor, broken down as follows: constructive humor (three items, e.g., my colleagues use humor to break work tension ; my coworkers use humor in a balanced way without harming the work ; humor by my colleagues improves the working atmosphere in my team ) and offensive humor (three items, e.g., my coworkers used offensive and inappropriate humor ; in my team, when someone makes a mistake, sometimes the others mock him ; people in my team sometimes use humor that demeans others ). Reliability analyses indicated an acceptable level for constructive team member humor (α =.70) and a relatively high level for offensive team member humor (α =.82). Perceived Supervisor Support. To assess perceived support from the immediate supervisor, we adapted three statements from POS by replacing the word organization with the term immediate supervisor (cf. Eisenberger et al., 2002) and used the 7-point Likert-type scale described above (e.g., My immediate supervisor cares about my wellbeing ). The internal consistency of this construct is very high (α =.91). Organizational Citizenship Behavior. To assess citizenship behavior, we used three items derived from P. M. Podsakoff, Bommer, Podsakoff, and MacKenzie (2006) that represent cooperation (willingly donates time or energy to support someone else), loyalty (defends the reputation of

7 Tremblay and Gibson 45 the organization), and proactive behavior (improvement of the entire organization; proactive suggestions). To assess these items, we used a 10-point Likert-type scale ranging from 1 (low intensity) to 10 (high intensity). The Cronbach alpha coefficient (.71) indicates that the internal consistency of these items is acceptable. Control Variables. We controlled for the effects of gender, number of years of service in the organization, age, and number of years working with the current immediate supervisor, because these variables are likely to influence the variables of our model. Previous research suggests that age, gender, and tenure may partly account for variations in the use of humor styles and their effectiveness (Gkorezis, Hatzithomas, & Petridou, 2011; Martin & Kuiper, 1999; Markey, Suziki, & Marino, 2014). Analysis We carried out a confirmatory factorial analysis (AMOS, 18.0) to evaluate the discriminatory validity of the study variables, and compared our theoretical model with different alternative models in which additional paths were relaxed. To test the moderating hypotheses, we produced several interaction terms representing two separate interactions: the interaction between leadership dimensions and supervisors humor styles and the interaction between PSS and colleagues humor styles. There are various techniques for testing the moderating effect of a variable using structural equation models (Cortina, Chen, & Dunlap, 2001). However, most of them either dichotomize continuous variables by the median split, with potential loss of information (MacCallum, Zhang, Preacher, & Rucker, 2002; Yaffe & Kark, 2011), or do not provide an adequate estimate of improvement of fit values. The procedure proposed by Mathieu, Tannenbaum, and Salas (1992) is the most highly recommended (Cortina et al., 2001). Consistent with the recommendations of J. Cohen et al. (2003), we mean-centered all the moderators. To evaluate the mediating effect of supervisor humor in the relationship between leadership and citizenship behavior, we proceeded with the bootstrap approach proposed by Preacher, Rucker, and Hayes (2007). This method is considered the most powerful technique for testing the size and significance of indirect effects. Confirmatory Factor Analyses. To assess the discriminant validity of our measurement scales, we performed a confirmatory factor analysis of our eight-factor model, in addition to testing several alternative models. As can be seen in Table 1, our eight-factor model presents acceptable indices (χ 2 = , degrees of freedom [df] = 322, goodness-of-fit index [GFI] =.82, comparative fit index [CFI] =.90, Tucker Lewis index [TLI] =.89, root mean square error of approximation [RMSEA] =.07), in addition to being significantly superior to other models evaluated. An RMSEA of.07 suggests a reasonable fit of the model to the data (Browne & Cudeck, 1993). We tested a five-factor model (Model 2) in which we merged all leader humor and coworker humor variables. Next, we tested a five-factor model (Model 3) in which we merged all constructive and offensive leader and coworker humor variables. We also tested a four-factor model (Model 4), in which we merged all constructive and offensive leader and coworker humor variables, and the two leadership styles. We tested a three-factor model (Model 5), in which we merged all constructive and offensive leader and coworker humor variables, the two leadership styles and the PSS measure. Finally, we tested a one-factor model (Model 6) that included all variables. The results show that our eight-factor theoretical measurement model is significantly superior to the other models evaluated (Model 2: χ 2 = , Δχ 2 = , p <.01, df = 340, GFI =.65, CFI =.68, TLI =.64, RMSEA =.12; Model 3: χ 2 = , Δχ 2 = , p <.01, df = 340, GFI =.71, CFI =.76, TLI =.73, RMSEA =.10; Model 4: χ 2 = , Δχ 2 = , p <.01, df = 344, GFI =.69, CFI =.73, TLI =.71, RMSEA =.11; Model 5: χ 2 = , Δχ 2 = , p <.01, df = 347, GFI =.67, CFI =.70, TLI =.67, RMSEA =.11; Model 6: χ 2 = , Δχ 2 = , p <.01, df = 350, GFI =.50, CFI =.43, TLI =.39, RMSEA =.16). Group Effect. Because respondents attitudes and behaviors are nested in their respective business units, respondents from the same organization may show similar work-related attitudes and citizenship behaviors. A one-way analysis of variance showed no significant differences in the mean for CRLB (F = 0.58, ns), LFLB (F = 1.9, ns), PSS (F = 0.27, ns), and OCB (F = 0.36, ns) across the organizations studied. We performed ICC1 (intraclass correlation) to assess the group-level effect. ICC1 scores, which assess the amount of variance explained by the business-level effects or variance attributable to differences between organizations, did not exceed.03. In general, coefficients from.05 to.09 indicate a weak group effect (Klein et al., 2000). These results show that the individual level is the most relevant to our analyses. Results Table 2 presents the averages, standard deviations, and correlations between the study variables. Hypothesis Testing The structural equation modeling results are presented in Figure 2. They show that PSS was positively predicted by CRLB (β =.72, p <.01) and negatively predicted by LFLB (β =.09, p <.01). PSS also positively predicted OCB (β =.19, p <.01). Our theoretical model suggesting that

8 46 Journal of Leadership & Organizational Studies 23(1) Table 1. Confirmatory Factor Analysis Results. Models χ 2 Δχ 2 df GFI CFI TLI RMSEA 1. Eight-factor theoretical measurement model Five-factor model (merge of all supervisor humor ** styles and of all coworker humor styles) 3. Five-factor model (merge of all positive supervisor ** and coworker constructive humor styles and of all offensive humor styles) 4. Four-factor model (same as Model 3, but the two ** leadership dimensions were merged) 5. Three-factor model (same as Model 4, but ** leadership styles and supervisor support were merged) 6. One-factor model ** Note. df = degrees of freedom; GFI = goodness-of-fit index; CFI = comparative fit index; TLI = Tucker Lewis index; RMSEA = root mean square error of approximation. *p <.05. **p <.01. Table 2. Averages, Standard Deviations, and Correlations Between Variables. Variables M SD Contingent reward leadership (.88) 2. Laissez-faire leadership ** (.68) 3. Constructive humor style (positive) **.12* (.80) 4. Aggressive humor style (negative) **.16**.04 (.85) 5. Self-defeating humor style (negative) **.11.40**.09 (.71) 6. Supervisor support **.40**.40**.36**.19** (.91) 7. Offensive humor (work team) **.22**.01.29**.16**.19** (.82) 8. Constructive humor (work team) **.02.21**.18**.01 (.70) 9. Citizenship behavior **.02.16**.15**.06.24**.19**.17** (.71) Note. The Cronbach alpha coefficients are presented in the diagonal within parenthesis. **Correlation significant at.01. *Correlation significant at.05. PSS mediates the relationship between leader behaviors and OCB was tested by the procedure proposed by Preacher and Hayes (2004). We ran the bootstrap resampling method with replacements to generate 1,000 samples of the entire data set. The indirect relationship for CRLB (.09 and.31) and LFLB (.21 and.05) through PSS was significant, with a 95% confidence interval (CI) for OCB. Because the CI did not include zero, these relationships are significant. These results indicate that PSS mediates the influence of CRLB and LFLB on OCB. Due to the size of our sample and the complexity of estimating interactions with continuous latent variables in structural equation modeling (Preacher et al., 2007; Yaffe & Kark, 2011), each moderation hypothesis was examined in a different structural equation analysis. The first step of the procedure introduced by Mathieu et al. (1992) involves the creation of a single indicator of latent product for each of the latent variables. The potential moderating effects of each moderator were evaluated by comparing the model with and without the path from the latent product to the criterion variable. A significant χ 2 suggests that the two models are significantly different and the variable of interest moderates the hypothetical relationship (Cortina et al., 2001). Hypothesis 1 predicted that the positive relationship between CRLB and PSS should be higher when the supervisor used a positive humor style. The comparison between the model with interaction and the model without the path of the interaction to PSS shows that the fit of the model with interaction is better (χ 2 = 74.3 and 85.8, Δχ 2 = 11.5, p <.01). Contrary to the enhancing effect of constructive humor anticipated by the first hypothesis, the relationship between CRLB and PSS becomes weaker (β =.18, p <.01) when the supervisor makes extensive use of constructive humor (low constructive humor, β =.77, vs. high constructive humor, β =.71). Figure 3a shows that high contingent reward leaders who use constructive humor less are perceived as more supportive than leaders who use this form of humor extensively. Figure 3a also shows that the PSS is higher under conditions of weak CRLB and high constructive humor styles than under conditions of weak CRLB and low use of constructive humor by the leader. The lowest level of PSS was observed under conditions of low CRLB

9 Tremblay and Gibson 47 Figure 2. Final model results. Note. CRLB = contingent reward leader behavior; LFLB = laissez-faire leader behavior; PSS = perceived supervisor support; OCB = organizational citizenship behavior. Figure 3. Moderating affect of (a) constructive leader humor style in relationship between contingent reward leader behaviors and perceived supervisor support; (b) self-defeating leader humor style in relationship between contingent reward leader behaviors and perceived supervisor support; (c) aggressive leader humor style in relationship between contingent reward leader behaviors and perceived supervisor support; (d) aggressive leader humor style in relationship between laissez-faire leader behaviors and perceived supervisor support.

10 48 Journal of Leadership & Organizational Studies 23(1) and low constructive humor styles. This result suggests that weak contingent reward leaders are viewed as less supportive when they use constructive humor less frequently, and are viewed as more supportive when they use this type of humor extensively. Thus, these findings suggest that constructive humor has a buffer effect for leaders who have weak contingent reward skills, and a reducing effect for those with high contingent reward skills. Furthermore, the combined effect of CRLB with selfdefeating humor results in a negative link with PSS (β =.21, p <.01). The comparison between the models shows that the fit of the model with interaction is better than the model without interaction (χ 2 = 74.3 and 85.8, Δχ 2 = 11.5, p <.01). Figure 3b suggests that the CRLB PSS relationship is less positive when a leader uses self-defeating humor extensively (low β =.83 vs. high β =.68, p <.01). Figure 3b shows that PSS is higher under conditions of weak CRLB and high self-defeating humor styles than under conditions of weak CRLB and low use of this humor style. In contrast, Figure 3b suggests that high humorous CRLs are perceived as less supportive than those who use self-defeating humor less frequently. The lowest level of PSS was observed under conditions of low CRLB and a low self-defeating humor style. This result suggests that weak CRLs are viewed as less supportive by their followers when they refrain from using a self-defeating humor style than when they use of this type of humor extensively. Contrary to what we had anticipated, these findings suggest that self-defeating humor has a buffer effect rather than an exacerbating effect for leaders with weak contingent reward skills, but an undermining effect for leaders who use self-defeating humor extensively. Moreover, CRLB and aggressive humor have a significant positive joint effect on PSS (β =.10, p <.05), contrary to what we had anticipated with respect to Hypothesis 2. The comparison between the models shows that the fit of the model with interaction is better (χ 2 = 55.0 and 60.4, Δχ 2 = 5.4, p <.01). Figure 3c shows that the CRLB PSS relationship is more positive when a leader uses an aggressive humor style extensively (low β =.73 vs. high β =.80, p <.01). Figure 3c shows that PSS is higher under conditions of weak CRLB and a low aggressive humor style than under conditions of weak CRLB and a high use of this humor style. This result suggests that weak contingent reward leaders are viewed as less supportive by their followers when they use more aggressive humor rather than less. In contrast, Figure 3c suggests that contingent reward leaders who favor aggressive humor are perceived as more supportive than those who use aggressive humor less. Overall, the results suggest that the use of aggressive humor has an exacerbating effect for weak CRLB, and an enhancing effect for skillful CRLB. Building on uncertainty management theory, Hypothesis 3 predicted that PSS will be lower under conditions of high laissez-faire leadership and high constructive humor. Given that no significant interaction was found, Hypothesis 3 was not supported. Hypothesis 4 predicted that the negative relationship between LFLB and PSS would be stronger for leaders using high self-defeating and aggressive leader humor styles. Figure 2 shows that interactions were significant only for the aggressive humor style (β =.13, p <.01). A comparison of the models illustrates that the fit of the model with interaction is better than that of the model without interaction (χ 2 = 9.07 and 16.2, Δχ 2 = 7.5, p <.01). Figure 3d indicates that the negative relationship between LFLB and PSS is higher with a high degree of aggressive humor style, partially confirming Hypothesis 4. Specifically, the lowest level of PSS was found under conditions of high laissez-faire leadership and high aggressive humor styles. This finding suggests that aggressive humor exacerbates the negative influence of laissez-faire leadership on PSS (low aggressive, β =.34, high aggressive, β =.44, p <.01). Note that the lowest PSS was reported for the aggressive laissez-faire leaders, as opposed to followers supervised by weak laissez-faire leaders, regardless of whether they use an aggressive humor style. Our theoretical model explicitly proposed that the indirect effect of transactional leadership behaviors on OCB through PSS should be moderated by supervisor humor styles. We used Hayes s (2011) process path analysis tool, which combines moderation and mediation, and the bootstrapped 1,000 random samples, which allows us to generate 95% bias-corrected CIs for all moderated indirect effects anticipated in our model. Our analysis shows that the indirect effect of contingent reward behavior on OCB through supervisor support was not moderated by any supervisor humor style. Analyses also show that the indirect effect of passive leadership on OCB through supervisor support was both significant at low (estimate =.03, 95% CI [.12,.01]) and high levels of passive leadership (estimate =.05, 95% CI [.11,.01]). However, the difference between the two levels of passive leadership was not significant (estimate =.01, 95% CI [.05,.03]). Overall, the moderated mediation analysis provides little evidence that the indirect effect of transactional leadership behaviors on OCB through supervisor support varies across levels of constructive and offensive humor styles. Moderating Impact of Offensive Humor Among Coworkers Figure 2 indicates that offensive humor from coworkers negatively predicts employee OCB (β =.17, p <.01), whereas constructive humor was not significantly related. However, neither constructive nor offensive coworker humor has a significant moderating influence on the relationship between PSS and OCB. Hypothesis 5 was thus not supported, while for the research question 1, we have failed to provide a clear and conclusive response.

11 Tremblay and Gibson 49 Discussion This study provides evidence that not all humor styles are beneficial, and justifies the inclusion of humor in interventions designed to improve leaders effectiveness and to foster OCB. Below we discuss the results relating to constructive and offensive humor used by the immediate supervisor, followed by the role of humor used by team members. Influence of Constructive Humor by the Immediate Supervisor Contrary to Hypothesis 1, we found that the relationship between CRLB and PSS became less positive with high use of constructive humor by the leader. This finding mirrors those of Avolio, Howell, et al. (1999) and Vecchio et al. (2009). Several reasons can explain this intriguing result. First, constructive humor as a self-enhancing style may be viewed as a tactic aimed at avoiding the sources of stress and taking situations lightly. These perceptions can create a deep sense of a lack of support when the need arises. Second, CRLBs such as setting clear objectives, and providing negative performance feedback could be undermined if the leader wastes too much time telling jokes during work hours (London, 2002). Our study demonstrates that extensive use of constructive humor styles appears to compensate for the fact of being supervised by a poor contingent reward leader. However, the support provided by such leaders is far from reaching the level of support provided by humorous or nonhumorous skillful contingent reward leaders. Our findings also suggest that constructive humor does not exacerbate the impact of LFLB, nor do they provide evidence that constructive humor may buffer the negative influence of poor leaders. Because humorous laissez-faire leaders cannot make employees overlook their deficiencies as a leader, they will not be able to provide support to others. In this respect, we contradict the significant beneficial effect of humor for poor leaders found by Avolio, Howell, et al. (1999), and the exacerbating effect of inconsistent behaviors (e.g., Nahum-Shani et al., 2014; Thau et al., 2009). This discordance is probably mainly attributable to the nature of the variables assessed: We evaluated humor effectiveness on PSS among followers, whereas Avolio, Howell, et al. (1999) evaluated the influence of humor on a measure of perception of leadership skills. Furthermore, it is possible that inconsistent signals, in this case constructive humor behaviors, are not strong enough to generate sufficient uncertainty to exacerbate undermining behaviors such as laissez-faire behavior. It is also possible that some undermining behaviors generate an uncertainty level such that additional ambiguity, such as adoption of inconsistent behavior, may have no amplifying impact. Such a situation can be qualified as overdetermined (Dirks & Ferrin, 2001). Accordingly, when cues from behavior are particularly strong, outcomes become dominated by such factors. Therefore, other moderate signals will unlikely facilitate, hinder, or exacerbate the effects of these strong cues. The presence of laissez-faire leadership would thus have such a strong negative overdetermined effect that any other positive behavior by the leader would not change the perceptions generated by this leadership style. Influence of Offensive Humor Style by the Immediate Supervisor We found a significant negative link between leaders aggressive humor style and PSS. This finding is in line with recent work on aggressive humor (Huo, Lam, & Chen, 2012). However, contrary to what we had anticipated in respect to Hypothesis 2, skillful contingent reward leaders are perceived as more supportive when they use aggressive humor extensively. This result requires elaboration. Vinson (2006) believes that aggressive humor, which targets a common enemy or specific deviant behavior, can have beneficial effects. Aggressive humor against an enemy, such as a competitor or another team, can be perceived as a uniting factor (Waples & Connelly, 2008). Moreover, aggressive humor is more likely to be seen as legitimate if it is used by a skilled contingent reward leader. These leaders often inspire trust, and their aggressive humor can be perceived as serving the common interest and individual well-being (e.g., correcting mistakes and helping employees reach their objectives). Aggressive humor allows a contingent reward leader to reprimand the least efficient employees or those who adopt antisocial behaviors (Cruthirds et al., 2013; Holmes & Marra, 2002). In these cases, aggressive humor may not only be accepted but may also be preferred to disciplinary sanctions. Thus, aggressive humor does not appear to be negative in all situations. However, our findings suggest that aggressive humor has an undermining effect for both weak CRLB and high LFLB. Scholars have stated that respect for an individual determines whether or not aggressive humor will be appreciated and interpreted as being funny (Duncan & Feisal, 1989; Terrion & Ashforth, 2002). As followers generally demonstrate little trust for poor leaders or for those who do not provide adequate rewards or manifest their gratitude (Doucet, Fredette, Simard, & Tremblay, 2015; Tremblay et al., 2013), logic would dictate that weak leaders using aggressive humor are rarely appreciated, and consequently rarely perceived as a caring person. Moreover, we found that contingent reward leaders are perceived as being less supportive when they use a selfdefeating humor style. Notably, the level of support increases more significantly when such leaders use self-defeating humor less rather than more. Making others laugh at one s own expense could weaken the leader s capacity to reward