Introduction

Workplace inclusion and humor have received growing attention in the last decade. Research in this area has provided evidence that these two constructs are related to various individual attitudes, affects and performance indicators (Mesmer-Magnus et al. 2012; Blackhart et al. 2009; Lapalme et al. 2009; Wang and Kim 2013; Pearce and Randel 2004; Tremblay and Gibson 2016). However, why and how humor is linked to subordinates’ perceptions of inclusion, and subsequently to cooperative behavior, is still far from clear (Shore et al. 2011). Although some qualitative studies have raised the possibility that humor in the workplace can be an inclusionary or exclusionary work tool (Collinson 2002; Lennox-Terrion and Ashforth 2002), to our knowledge no study has empirically verified the mediating role of perception of inclusion in the influence of humor on organizational citizenship behavior (OCB). Drawing on the signaling argument (Derfler-Rozin et al. 2010; Thau et al. 2015), the current study suggests that team members use humor in the workplace as valuable information to validate their inclusionary status in their team.

Further, previous studies on humor have tended to emphasize positive humor practiced by the immediate supervisor (e.g., Avolio et al. 1999), and thus have largely ignored the influence of offensive humor (e.g., Martin et al. 2003; Huo et al. 2012). A recurring theme in the literature is that negative inclusive signals have a greater impact on attitudes and behavior than do positive signals (Baumeister et al. 2001). We know very little about the influence of positive and offensive humor on perception of employees’ inclusionary status. Third, with a few exceptions (e.g., Lennox-Terrion and Ashforth 2002; Scogin and Pollio 1980; Slåtten et al. 2011; Lehmann-Willenbrock and Allen 2014), studies on humor in the workplace have mainly investigated the individual level, ignoring the contagious effect of humor, the possibility of significant group-level variation in perceptions of humor, and that humor climates may directly or indirectly condition feelings of inclusion and other important individual outcomes such as citizenship performance. Finally, few humor studies have looked at the boundary conditions of humor effectiveness. Some scholars argue that humor may be a double-edged sword and interpreted in different ways (Meyer 2000). Building on trust theory (McAllister 1995) and social exchange and reciprocity perspectives (Blau 1964; Gouldner 1960), we posit that trust in the humor source is an important boundary condition of humor effectiveness.

The proposed model is shown in Fig. 1. By examining the pathways through which these constructs exert an influence on OCB, the present study enriches recent organizational humor and workplace inclusion/exclusion models (Cooper 2008; Romero and Cruthirds 2006; Robinson et al. 2013), illuminates the contents of the black box with regard to the relationship between humor behaviors and individual outcomes, and has the potential to improve our existing knowledge of the boundary conditions of efficacy of humor.

Fig. 1
figure 1

Meso-model of relationships between humor climates, inclusion, and OCB

Full size image

Theoretical Background and Hypotheses

Humor Climate and Psychological Inclusion

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 produces positive emotions in an individual, in a group, or in the organization. It is widely accepted that humor is not always used properly (Williams and Emich 2014), and that individuals can adopt a variety of humor styles. Martin et al. (2003) distinguish four humor styles, two positive and two negative, with the positive style comprising affiliative and self-enhancing types. The affiliative type fosters team spirit and a sense of belonging, while self-enhancing humor reduces stress levels and instils a positive outlook. Individuals that are deeply rooted in the affiliative style tend to say funny things, tell jokes, amuse others with their comments and stories, easily develop relationships, and have a strong ability to reduce interpersonal tensions (Puhlik-Doris 2004). This humor is never hostile and always seems to foster interpersonal relationships. Individuals who fully adopt the self-enhancing humor style are more inclined to have fun and laugh at life’s daily foibles. They have an overall positive view of the world, which allows them to manage stress more easily and to overcome challenges they face in their personal and professional lives. The aggressive and self-defeating humor styles described 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. This form of humor includes the use of humor to manipulate others without regard for its potential impact on others (Martin et al. 2003). People who are constantly making others laugh at their own expense are 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 others.

An important prerequisite of this study is that the humor constructs can be aggregated at the unit level. A growing body of research has provided evidence that team members’ attitudes and behaviors are affected not only by their own perceptions but also by the attitudes and behaviors of the leader and other group members (Yang et al. 2007). First, through their interactions with followers, leaders act as a key filter that helps subordinates make sense and interpret workplace events (Hackman 1992). Through their behavior, leaders communicate priorities and promote consensus on important goals and values (Luria 2008). Second, according to the interactive approach, horizontal relationships are another pathway to a strong climate (Gonzalez-Roma et al. 2002). Within a unit, members interact with one another and engage in collective sense-making; a tendency that may ultimately lead to the development of shared perceptions on how to evaluate exchange-triggering events (Naumann and Bennett 2000). Hackman (1992) argued that groups affect members’ attitudes and behavior by exposing members to “ambient stimuli,” i.e., aspects of the work environment that members encounter regularly in their life in the group. Ambient stimuli vary considerably between work groups. According to social processing theory (Salancik and Pfeffer 1978), people form attitudes and behaviors based on information collected in their immediate environment. Individuals discuss events that they have personally experienced or observed, and they exchange their interpretations of these events (Roberson 2006). Prior research suggests that perceptions of leadership and colleague behaviors are often shared within teams (Bass et al. 2003; Ehrhart 2004; Bowen and Ostroff 2004). Scholars assert that humor is fundamentally a group phenomenon. People are 30 times more likely to laugh in a group than in isolation (Johnson 2007), suggesting a contagious pattern in teamwork (Lehmann-Willenbrock and Allen 2014; Robinson and Smith-Lovin 2001). This suggests that the humor of the immediate supervisor can lead to the emergence of a common experience within a unit or group. There are two reasons for this: first, humor, positive or offensive, is often used in public to increase its effect and hence the possibility of convergence of views. Second, sometimes the humor comes in the form of a story or joke that is discussed and even repeated by individuals or sub-groups (Lehmann-Willenbrock and Allen 2014), which increases the possibility of convergence on the status of humor within the group (e.g., positive or offensive).

Like humor, there is no consensus on the nature and relevant theoretical frameworks of inclusion/exclusion (Shore et al. 2011). Schein (1971) defined inclusion as the degree to which an employee is an “insider in the organization.” Pelled et al. (1999) defined inclusion as “the degree to which an employee is accepted and treated as an insider by others in a work system” (p. 1014), whereas Blackhart et al. (2009, p. 207) defined social exclusion as “one person is put into a condition of being alone or is denied social contact” (p. 207). Other researchers used the term ostracism to describe the same exclusion phenomenon, which they define as “individual’s perception of being ignored or excluded at work” (Ferris et al. 2009, p. 1384). According to scholars, the need to belong, form, and maintain a minimum of interpersonal relationships is a universal human motivation (Baumeister and Leary 1995), and people are very sensitive to the degree to which they are accepted or excluded (Leary and Downs 1995). Brewer’s (1991, 2005) optimal distinctiveness theory posits that humans seek to balance two opposite needs: the need for assimilation and inclusion, and the need for differentiation. To fulfill the need for inclusion, people choose social identity with particular groups and seek acceptance in these groups (Shore et al. 2011).

Leader Humor Climate and Employee Inclusion

Why would leader humor climate positively or negatively affect employees’ perception of inclusion? Cooper (2008) suggests that humor can build or impoverish relationships through four mechanisms: affect/reinforcement, decreasing/increasing hierarchical salience, similarity/attraction and self-disclosure behavior. First, findings suggest that humorous individuals are perceived as more socially attractive (Wanzer et al. 1996). People who elicit positive affect by constructive use of humor are more likely to be viewed as an attractive and supportive force, whereas leaders who use offensive humor are more likely to elicit negative affect, and to be perceived as a repulsive force. Susa (2002) and Tremblay and Gibson (2016) showed that supervisors who use constructive humor styles are perceived as providing more support than those using some forms of offensive humor. Similarly, a recent study concludes that perceived support from supervisors is positively related to perception of insider status or inclusion (Lapalme et al. 2011).

Second, use of humor by leaders may make hierarchical distance more or less salient. Constructive humor may contribute to reducing social distance and status disparity and to promoting identification with leaders (Romero and Cruthirds 2006). Weak social distance involving the use of constructive humor would facilitate close relationships between follower and leader (Cooper 2008; Romero and Pescosolido 2008). A positive and strong leader humor climate suggests that humor is pervasive in the team and that the leader seeks to reduce the amount of within-group relationship differentiation or follower exclusion (Liden et al. 2006). Research demonstrates that employees are aware of the differential relationships their leaders form with other members (Erdogan and Bauer 2010). A constructive humor climate may signal that leaders seek to maintain an inclusive relationship with each member. Conversely, a strong offensive humor climate may be perceived as a signal that the leader seeks to maintain a high-quality relationship with some subordinates and a distant relationship with the rest. Leaders may use recurrent offensive jokes to signal that some team members are making insufficient contributions or underperforming, and consequently risk being excluded (Derfler-Rozin et al. 2010; Thau et al. 2015). Sharing jokes and anecdotes can lead to an improvement in the interaction between supervisor and subordinate and create a context that is conducive to the expression of emotions and opinions (Peter and Dana 1982). The use of positive humor can make team members feel safer about engaging in information exchange processes and sharing personal information, which facilitates followers’ inclusion (Heiss and Carmack 2012; Romero and Pescosolido 2008). The above considerations suggest that leaders who use constructive humor extensively may be perceived by followers as an attractive, approachable and supportive force that may foster the experience of inclusion. Conversely, leaders’ extensive use of an offensive humor style signals the wish to maintain status difference and power relations or dissatisfaction with employees’ contributions (Collinson 2002; Robinson et al. 2013). Thus, we propose that promotion of a leader offensive humor climate undermines followers’ perception of their relationship with the leader and of their degree of inclusion. We therefore predict that:

Hypothesis 1a

There is a positive relationship between a constructive leader humor climate and employee inclusion.

Hypothesis 1b

There is a negative relationship between an offensive leader humor climate and employee inclusion.

Building on Baumeister’s view (2001) that bad is stronger than good, the present study posits that an offensive humor climate exerts a stronger effect on perceived inclusion than a positive humor climate does. First, according to adaptation-level theory (Helson 1964) people are more sensitive to new conditions than to stable ones, such that they become more accustomed to positive events and have more difficulty adapting to a bad event. An important factor that contributes to this effect is that offensive behavior has a greater visibility and recall than constructive behavior does (Baumeister et al. 2001). The affect-as-information theory (Forgas 2000; Taylor 1991) suggests that negative information elicits more complex cognitive processing than positive information does. People who experience negative affect through offensive humor are more motivated to carefully scrutinize and analyze this behavior. They are also more likely to subsequently be influenced by this information than individuals who have experienced a positive affect through constructive humor (O’Leary-Kelly and Newman 2003). Therefore, we propose the following hypothesis:

Hypothesis 2

Offensive humor climate has a stronger influence on employee inclusion than a positive humor climate does.

Moderating Effect of Trust in Leaders

Humor behaviors are often ambiguous. Employees may interpret humor from leaders in different ways, depending on the motivations or the intentions of the humor sources. Some may interpret leader offensive humor as an attempt to humiliate or discredit employees, whereas other employees may interpret positive humor as an attempt to manipulate them. Building on theories of trust (Mayer et al. 1995) and the moderating perspective of trust (Dirks and Ferrin 2001), we propose that employees will react more positively to a constructive humor climate when they have high trust in their immediate supervisor, and react less severely to offensive humor when their supervisor is perceived as trustworthy.

Trust in a leader refers to employees’ confidence that their leader will act for the benefit of employees and not exploit their vulnerability (Pillai et al. 1999). Dirks and Ferrin (2002) distinguished two theoretical perspectives of trust: a relational perspective and a character perspective. The character perspective focuses on the perception of trust of the target character and how it impacts a follower’s vulnerability in the hierarchical relationship. Followers’ sense of vulnerability may be high, particularly in superior-subordinate relationships, where there is often power asymmetry between the subordinates and the formal leader. A high perception of leader trust increases confidence that the supervisor will not exploit followers’ vulnerability and that the leader’s intentions are positive. This perspective suggests that followers’ inferences on trust are based on a variety of leaders’ characteristics, such as integrity, dependability, fairness, benevolence, and ability (Dirks and Ferrin 2002).

The second perspective is relational-based trust. Relational-based trust reflects the importance of trust in interpersonal relationships, and is considered an important condition in the development of social capital within the unit (McAllister 1995; Ferres et al. 2004). Subordinates trust that their supervisor will support them, will not take advantage of them, and will act true to their word (Ferres et al. 2004). High supervisor trust may be perceived as the desire to establish high-quality relationships based on care and mutual consideration (Dirks and Ferrin 2002). The relationship-based perspective of trust, building on social-exchange theory (Blau 1964) and the reciprocity perspective (Gouldner 1960), suggests that when employees feel that their supervisor shows marked trust, care and consideration, this signals that the workplace promotes inclusion and encourages individuals to cooperate with other members of the team (Lau and Liden 2008). Building on the study by Dirks and Ferrin (2001), we posit that trust acts as a boundary condition for perceptions of the humor climate. These scholars state that trust may affect attitudes and behavior through two processes: (1) the assessment of the future behavior of another party with whom there is an interdependent exchange, and (2) the interpretation of the past actions of the party and the motives underlying those actions. According to the first process, people who attribute positive future behavior to one’s work partner (e.g., supervisor) are more likely to devote all their resources to their role integration and performance because they are confident that they will be treated fairly and that the party will not exploit their efforts. In contrast, if employees consider their manager to be undependable, much of their resources will be devoted to self-protection. In such cases, leaders’ efforts of inclusion may no longer be successful because a lack of trust in them may produce insecurity and anxiety. According to Dirks and Ferrin (2001), to minimize their risks, individuals who do not trust other parties are less likely to be motivated to cooperate and more likely to engage in antisocial behavior. Paradoxically, such a reaction is likely to lead to a self-fulfilling prophecy of exclusion due to the fact that these behaviors may be perceived as a lack of inclusion willingness that is more likely to accentuate exclusion (Twenge et al. 2001).

The second process of trust identified by Dirks and Ferrin (2001) is the interpretation of a partner’s action by the individual who trusts that partner. From this perspective, in a situation of high trust, an individual is more likely to respond favorably to the partner’s negative action. They suggest that the same action may be interpreted differently depending on the level of trust. People are more willing to give trustworthy people the benefit of the doubt in ambiguous situations. This suggests that a climate of offensive humor is perceived less severely when the leader is seen as trustworthy. Previous studies have concluded that individuals responded more positively to negative feedback from a trusted supervisor than a supervisor who was not trusted (Earley 1986; Fedor 1991). For example, Fedor (1991) showed that negative feedback from a trusted supervisor was perceived as more relevant, and subordinates were more motivated to make efforts to improve, compared with those who had little trust in their immediate supervisor. The literature suggests that although a climate of offensive humor may exist, which may signal risk of exclusion, individuals who have high trust in others are more likely to believe that their actions will be reciprocated and to be optimistic about their future inclusionary status (Derfler-Rozin et al. 2010). In contrast, when interpersonal trust in those that use offensive humor is weak, the probability of positive interactions with those individuals is low, reinforcing the employees’ feeling of being an outsider. These overall considerations suggest the following hypothesis:

Hypothesis 3

Perception of trust in a supervisor moderates the relationship between the leader humor climate and employees’ perception of inclusion, such that the relationship is stronger when the perception of trust is high rather than low. This relationship will be stronger when the leader offensive humor climate is high rather than low.

Psychological Inclusion and Citizenship Behavior

OCB is generally defined as an individual, discretionary, extra-role behavior not formally rewarded or punished by organizations that improves the overall efficiency and effectiveness of organizations (Organ 1988). The relationship between the perception of inclusion and OCB has been studied, but to our knowledge no researchers have specifically examined how inclusion mediates the cross-level influence of humor climate on OCB. Stamper and Masterson (2002) examined a sample of full and part-time workers from the United States and have provided evidence that psychological inclusion (PI) positively predicts OCB, and that PI has a mediating effect on the relationship between perceived organizational support (POS) and OCB. In a study of full-time employees from China, Chen and Aryee (2007) have shown that PI mediates the relationship between delegation and innovative behavior. Using a sample of Canadian agency workers, Lapalme et al. (2009) found that PI mediates the relationship between social support and interpersonal facilitation, while Wang et al. (2010) looked at the relationship between leader-member exchange (LMX) and OCB.

Building on social exchange theory (Blau 1964) and the reciprocity norm (Gouldner 1960), this body of research argues that individuals are more likely to reciprocate beneficial and inclusive actions by the supervisor by performing citizenship behaviors. Employees who feel fully included are more likely to see themselves as organizational citizens and are thus more willing to accept the relevant responsibilities (Stamper and Masterson 2002; Lapalme et al. 2009; Wang and Kim 2013). Consistent with our earlier argument that inclusion enhances the perception of being treated as a full member of the organization, this belief should predispose individuals to perform activities and assume roles that exceed their normal tasks (De Cremer and van Knippenberg 2002). Given that inclusion is considered an important human motive, PI would motivate employees to adhere to in-group norms and sanctions, cooperate with other members in the group or sanction those who deviate from group expectations (Brewer and Pierce 2005). Further, people who feel rejected are less likely to act in prosocial ways, such as cooperating with someone or providing help (Twenge et al. 2003). Socially excluded people tend to enter defensive states of cognitive deconstruction, such as focusing on the present rather than the future, thinking that life is meaningless, believing that time passes very slowly, and displaying chronic passivity and few emotions (Twenge et al. 2003). They are also more likely to display self-defeating behaviors such as taking irrational risks and making unhealthy choices such as consuming alcohol. Socially excluded people are also more likely to procrastinate regarding some task and non-tasks aspects (Twenge et al. 2002) and perform less extra-role behaviors (Ferris et al. 2009). Pearce and Randel (2004) argued that individuals who perceive themselves as excluded by others at work spend less time learning about co-workers, will feel less obliged to assist them, and will limit information or other resources needed to do the job. Given that discretionary behaviors are rarely rewarded or punished directly (Organ et al. 2006), excluded individuals are more likely to withhold or reduce discretionary behaviors and be less motivated to cooperate with others. Hence the following hypothesis:

Hypothesis 4

The perception of employee inclusion mediates the relationship between supervisor humor climates and citizenship behaviors.

The prediction that PI will mediate the relationship between the humor climate and OCB, and that trust will moderate the association between the humor climates suggests that trust in supervisors will conditionally influence the strength of the indirect link between humor climates and OCB through inclusion. This moderated mediation pattern suggests that the mediating effects of inclusion will be less strongly related with OCB when trust is low. We posit that trust will moderate the indirect effects of the leader humor climate on OCB through inclusion.

Hypothesis 5

The indirect effect of leader humor climate (constructive and offensive humor) on OCB through inclusion is weaker when trust is lower.

Methodology

Participants

To test our research hypotheses, we used a sample of respondents from a Canadian financial organization. Out of a total of 260 questionnaires sent out, 225 were returned to the researchers, representing an 86.3 % response rate. This was a voluntary, confidential, and anonymous exercise. The respondents were nested within 23 teams, each led by an immediate supervisor. The average number of respondents per team was 14, ranging from 5 to 20. The sample primarily consisted of women (89 %), and the average age of respondents was 41. The average seniority of the respondents was 10.8 years, 41 % of them held a university degree and 76 % worked full-time.

Measurement Instruments

For all indicators, respondents had to state their degree of agreement on a scale ranging from (1) Disagree completely to (7) Agree completely.

Leader’s Humor Styles

We used 13 items from the Humor Styles Questionnaire (HSQ) by Martin et al. (2003). We performed an exploratory factor analysis and found two clear interpretable solutions: positive (eight items) and negative (five items) leader humor styles. For the constructive humor style, examples of items used are: “My supervisor enjoys making people laugh” and “If my supervisor is feeling depressed, he can usually cheer himself up with humor.” For the negative humor style, examples of the items for this scale are: “If someone makes a mistake, my supervisor will often tease him about it”; “When telling jokes or saying funny things, my supervisor is usually not very concerned about how other people are taking it” and “Even if something is really funny to him, my supervisor will not laugh or joke about it if someone will be offended” (reverse). Reliability analyses were performed and indicated a very high level of internal consistency for constructive humor (α = .92) and adequate reliability (α = .70) for the offensive humor style. We computed the usual indices to provide empirical justification for aggregating constructive and offensive humor from the supervisor to the team level. The median level of constructive humor rwg(j), using a uniform null distribution, was .85, the ICCI was .16, and the ICC2 was .67. For offensive humor, the median level of rwg(j) was .78, the ICCI was .08, and the ICC2 was .50. The low ICC (2) for offensive humor may be explained mainly by the small group sizes in the sample (Bliese 2000); it is not unusual to have a lower ICC (2) value in studies (Yang et al. 2015). According to some scholars (e.g., Liao and Chuang, 2007), a low ICC2 value should not prevent aggregation if aggregation is justified by the theory and by an acceptable level of rwg(j). These statistics provide the justification for the aggregation.

Measurement of Perceived Supervisor Trust

Trust in the supervisor was measured using four items adapted from the scale introduced by Cook and Wall (1980). Two examples: “I can freely share and exchange ideas with my supervisor”; “I have high trust in my supervisor.” The Cronbach α reliability coefficient for the supervisor scale was .95.

Measurement of Perceived Inclusion

Perceived inclusion was measured using Stamper and Masterson’s (2002) six-item scale. Sample items include: “My work organization makes me believe that I am included in it,” “I feel I am an ‘insider’ in my work organization,” and “My work organization makes me frequently feel ‘left out’” (reverse coded). The reliability coefficient for the measure was 0.87.

Measurement of the Dependent Variable, OCB

To assess citizenship behavior, we used four items derived from Podsakoff et al. (2006) that represent cooperation (willingly donates time or energy to support someone else), loyalty (defends the reputation of the organization), agent of change behavior (improvement of the entire organization; proactive suggestions), and dedication. To assess these items, we used a 10-point Likert scale ranging from 1 (low intensity) to 10 (high intensity). The Cronbach α coefficient (0.71) indicates that the internal coherence of these four items is acceptable.

Control Variables

At the individual level, we controlled for respondent age and sex (female = 0 and male = 1) and at the team level we controlled for team size (number of direct reports) to minimize the potential confounding effects of several relationships proposed in our model. We controlled for age, sex, and tenure because they have been shown to relate to humor, perceived inclusion, and OCB (Hitlan et al. 2006; Kidder 2002; Cho and Mor Barak 2008). We controlled for team size because past research shows that team size may affect attitudes and concern for employee climate (Takeuchi et al. 2009). Our analyses indicated that none of these control variables exercised a significant influence, regardless of the model tested. To keep the number of parameters within acceptable limits, these control variables were removed from the analysis.

As our study was cross-sectional, the risk of a percept–percept bias may be present. Several precautions, recommended by Conway and Lance (2010) and Podsakoff et al. (2003), were taken to minimize them. First, we used constructs that have been validated by previous research. Secondly, the dependent variable OCB was measured by a different scale than its antecedent. Thirdly, we assured respondents of the anonymity and confidentiality of their responses Finally, as will be demonstrated in a later section, we have provided evidence that the variables in this study had an acceptable discriminant validity to multilevel.

Analytic Strategy

Our model is multilevel, consisting of variables at both the team level (leader humor climates) and individual level (e.g., perception of inclusion and citizenship behaviors). We used the IBM SPSS multilevel modelling version 21 (Heck et al. 2010) to test the hypotheses. Hierarchical linear models were tested by using REML estimations. This method provides better estimates when there are small numbers of groups (Raudenbush and Bryk 2002). It has been acknowledged that FIML estimates produced negatively biased level-1 and level-2 variance estimates under small size conditions (Peugh 2010). A Monte-Carlo simulation studies comparing REML and FIML found that there was no advantage to using one or the other of these estimation methods (Kreft and de Leeuw 1998). Before performing the analysis, we grand-mean centered the unit-level independent variables (e.g., constructive and offensive leader humor styles) and group-mean centered the individual-level independent variables (Hofmann and Gavin 1998). To estimate our 2-1-1 multilevel mediation model, we used Zhang et al. (2009) CWC (M) procedure in which the group-mean centered mediator is used and the group mean is reintroduced at level 2. To test the moderation effects, we introduced the group-level interaction between the types of humor climate and group-mean trust in supervisor. This approach, used by a growing number of scholars (e.g., Waldman et al. 2015), avoids confounding cross-level and between-group interactive effects. Finally, we used a moderated mediation approach to evaluate whether the indirect effect of humor climate constructs on OCB through perceived inclusion is more positive when trust is high using the Sobel (1982) test. To test the model fits, we used the −2 log likelihood ratio. This deviance test is an indicator of how well the model fits the data, and provides a more accurate estimate of variance differences (Bliese and Ployhart 2002). Models with a larger deviance statistic are worse than models with a lower deviance statistic. The deviance statistic is used to compare the goodness-of-fit of estimate models (Bliese and Ployhart 2002).

Results

Confirmatory Factor Analysis

With respect to common variance (Conway and Lance 2010), it is important to provide evidence that variables in the present study (independent, moderator, dependent) are empirically distinct. Our model proposes to use a multilevel dataset, thus single-level CFA analyses with nested data are problematic for three important reasons (Dyer et al. 2005; Dedrick and Greenbaum 2011). First, single-level CFA presumes that the data are independent, but this assumption is not realistic when a response to a construct comes from multiple subordinates, nested in teams or groups. Second, single-level CFA assumes that the nature of variables is isomorphic, that the variables exhibit the same dimension properties and thus do not differ across levels of analysis. Third, single-level CFA operates on a single covariance matrix or derives the factor structure by averaging the items’ responses at the group level. These two procedures both ignore the hierarchical structure of the data. We used a multilevel confirmatory factor analysis (MCFA) with Mplus version 5.1 (Muthén and Muthén 2007) on the multilevel data. MCFA decomposes the total sample covariance matrices into pooled within-group and between-group covariance matrices, and uses these two matrices to produce the factor structure at each level (Dedrick and Greenbaum 2011). With MCFA it is also possible to evaluate a different number of factors at each of the two levels. We used the comparative fit index (CFI), the root square error of approximation (RMSA), and the standardized root mean square residual (SRMR) as guides to evaluate the goodness-of-fit of the models. To compare alternative models, we used the Bayesian information criterion (BIC). The lower the value of the BIC, the better the fit of the models.

In the present study, we tested four multilevel measurement models (see Table 1). The first theoretical model included a 5-factor model in which we included two level 2 factors (constructive humor and offensive humor climate) and three leve1 factors (inclusion, OCB, and supervisor trust). The next model examined an alternative three-factor model that consisted of one level 1 factor and two level 2 factors (model 2). Model 3 examined a four-factor model that consisted of three level 1 factors and one level 2 factor. Finally, to test the discriminating validity of the measurement scales, we examined a model in which all models were combined to form a single factor in each level, which allowed us to identify common variance problems. According to the findings, our 5-factor theoretical measurement model fit the data well (χ 2 184.7, df 178, CFI .999, RMSEA .013, BIC 13,695.5), in addition to being significantly superior to the other models evaluated. The SRMR fit indices of each level revealed that the fit of the level 1 (within) part of the model was higher than the fit of level 2 or between part (SRMR within .064 vs. SRMR between .55). Relative to the hypothesized model, alternative models in which all level 1 factors were combined (model 2: ∆χ 2 624.8, df 185, CFI .740, RMSEA .103, SRMR within .12 vs. SRMR between .56, BIC 14,097.7), in which all level 2 factors were combined (model 3: ∆ χ 2 327.9, df 51, CFI .915, RMSEA .06, SRMR within .096 vs. SRMR between .624, BIC 13,806.3), and the model in which all level factors were combined into a single construct fit the data worse (model 4; χ 2 766.6, df 189, CFI .659, RMSEA .117, SRMR within .136 vs. SRMR between .624, BIC 14,217.8). The results of MCFA support the distinctiveness of study variables at multilevel and did not provide evidence for serious common method bias.

Table 1 Multilevel confirmatory factor analysis
Full size table

Hypothesis Testing

Table 1 shows the individual and team-level descriptive statistics, internal consistency reliability, and correlations among the study variables.

Hypotheses 1a and 1b suggested that PI would partially mediate the influence of constructive (H1a) and negative (H1b) leader humor styles on PI. Table 2 indicates that the aggregate leader offensive humor style is significantly related to PI (model 1; γ = −.55, <.05), whereas the constructive humor style failed to reach the normative level of significance (model 1; γ = .28, >.05) when we control for the within-group effect of leader humor styles. The stronger influence of offensive leader humor than constructive humor on PI provides support for hypothesis 2. To assess whether the indirect relationship between constructive and offensive leader humor styles and OCB through PI is significant, we used the Sobel Test. The average CWC (M) Sobel tests indicate that the indirect effects of constructive humor climate (z = 2.68, p < .01) and offensive humor climate (z =  −2.89, p < .01) were significant. Accordingly, PI partially mediates the influence of supervisor positive humor climate on OCB. Overall, regarding H5, the results fully support the indirect influence of humor climates on OCB through inclusion.

Table 2 Means, standard deviations, and correlations among variables
Full size table

Hypotheses 3 proposed that trust in the supervisor would moderate the relationship between humor climates and PI. Similar to Waldman et al. (2015), we introduced between-group interactions (PSHC X GST and OSHC X GST) and cross-level interactions (PSHC X IST and OSHC X IST) in model 2. In this model we entered leader humor styles, as well as individual-level and group-level trust in leader. Table 2 reveals that the interaction term constructive leader humor climate X supervisor trust for the inclusion model was significant (model 2: γ = .42, p < .05). We plotted this significant interaction graphically at one standard deviation below and above the mean (Aiken and West 1991). The plot shows (Fig. 2) that the beneficial effects of a positive leader humor climate on inclusion is stronger when trust in the supervisor is high rather than when it is low. However, the findings fail to provide support for the hypothesis that trust in the supervisor may mitigate the negative influence of offensive humor on inclusion (model 2: γ = −.12, p > .05). These moderating results provide only partial support for hypothesis 3. Although there was no specific hypothesis about the probable moderating role of trust in the influence of humor climate on OCB, a significant moderating effect was observed, but only when the between-group interactions of the leader humor style were removed. Unlike models of inclusion, trust seems to have a more significant moderating effect on OCB when there is a climate of offensive humor. Figure 3 show that a strong climate of offensive humor (γ = −3.3 p < .05) reduces OCB when trust is high, and increases OCB when trust is low. We observe that the OCB level is maximal when trust is high and the climate of offensive humor is low (Table 3). We return to these results in the “Discussion” section.

Fig. 2
figure 2

Interaction of leader humor climate and trust in predicting employee perception of inclusion

Full size image
Fig. 3
figure 3

Interaction of leader humor climate and trust in predicting employee OCB

Full size image
Table 3 Hierarchical linear modeling testing the mediating effect of inclusion between humor from supervisors on OCB and moderating effect of supervisor trust
Full size table

To evaluate the indirect effect of leader humor climates on OCB through PI at high and low levels of trust, we used the moderated mediation procedure using CWC (M) Sobel tests. Sobel tests indicate that for constructive leader humor, the indirect effect was not significant at a low level of supervisor trust (Z = .15, p > .05), but positive and significant at a high level of supervisor trust (Z = 2.7, p <. 01). For the offensive leader humor climate, the indirect effect was not significant at a low level of supervisor trust (Z = −.66, p > .05), but significant at a high level of supervisor trust (Z = −2.01, p < .05). These results fully support hypothesis 5.

Discussion

The main purpose of the current study was to explore the influence of leader humor climates on employee inclusion and OCB, and the role of supervisor trust in such relationships. Prior research has rarely examined how and why constructive and offensive humor climates can exert an inclusive influence on employee citizenship behavior. By incorporating the supervisor into the models as independent and moderating variables to test their effects on inclusion and OCB, this research takes into account what are probably the most influential sources shaping employee attitudes and behaviors (e.g., Chiaburu et al. 2013). Second, in exploring the cross-level influence of humor, we aimed to assess the utility of using micro- and macro-approaches, and, more specifically, the relevance of adopting an integrative multilevel view of the effect of a humor environment in predicting individual inclusion and citizenship behaviors.

Theoretical Implications

The first contribution of the present study is to extend humor literature to the team level. The findings suggest that extensive use of an offensive humor style by a supervisor may impoverish the follower relationship, and subsequently weaken the team members’ feeling of inclusion. Our results are consistent with the argument that a hostile humor climate may undermine the relationship between the leader and followers by signaling the leader’s intention to maintain status difference and power relations (Collinson 2002; Romero and Cruthirds 2006). The use of offensive humor may be viewed as an indication that the leader seeks to maintain high-quality relationships with only some subordinates, and thus is ready to foster variability in the quality of relationships with followers in the group (Liden et al. 2006) and promote a weak inclusion workplace environment. Multilevel analysis has shown that a constructive humor climate was no longer significantly related to inclusion when offensive humor behavior by the leader was entered in the model. This finding is consistent with a body of research suggesting that exploitative and aggressive behaviors make a stronger impression and have a stronger effect than do positive behaviors (Baumeister et al. 2001).

The present study also shows that inclusion represents a relevant mechanism to explain why leader humor styles may or may not elicit OCB. Employees who are accepted as insiders are more likely to reciprocate this beneficial treatment and the fulfillment of the need to belong by performing behaviors not explicitly required. According to inclusion scholars (Stamper and Masterson 2002; Lapalme et al. 2009), the perception of being fully included fosters citizenship status membership and the motivation to fulfill citizenship responsibilities. This positive link between inclusion and OCB is in the opposite direction from the social reconnection argument that suggests that people at risk of exclusion should be motivated to cooperate and make efforts to connect socially with others to improve their inclusionary status (Allen and Badcock 2003; Derfler-Rozin et al. 2010). The divergence in these results may be partly explained by the inclusion measure used. Whereas we have used a construct that captures a final state of inclusion/exclusion, this body of research has tested a social risk inclusion measure, a construct that prompts cooperative actions that may mitigate further exclusion.

Finally, we hypothesized that trust in the supervisor moderates the relationship between humor climate and inclusion. As expected, the relationship between leader constructive humor climate and inclusion is stronger when trust is high, confirming the enhancing role of trust, but disconfirming the possible “over-determination effect” of high trust (Dirks and Ferrin 2001). The enhancing effect of trust is consistent with the contention that humor from trusted sources is more likely to be interpreted as an indication that humorous people seek to establish and promote inclusive relationships. However, the present findings are not in line with the argument that high trust mitigates the detrimental effects of offensive humor on inclusion. We cannot rule out that an extensive offensive humor climate may be interpreted as a clear and definitive message of exclusion. This signal may be viewed as a strong situation (Meyer et al. 2009). As a result, whatever the level of trust, the inclusionary status probably would not change. However, the analysis shows that when trust is introduced in the models, the effect of humor on inclusion is no longer significant. The inclusive results for the offensive humor X trust interaction term suggests that when trust is high, whatever the inclusion signals, employees are confident that the risk of definitive exclusion is very low. Future research is needed to better understand the dynamic of offensive humor and trust.

Further, we found an intriguing moderating effect of trust in the relationship between an offensive humor climate and OCB. However this finding must be interpreted prudently because this significant moderating effect was only observed when between-group interactions of the leader humor style were removed. According to the findings, a positive humor climate increases motivation to exhibit OCB when employee trust is low, and decreases motivation when trust is high. These findings suggest that the use of offensive humor may be an efficient tactic to promote OCB, but only when employee trust is low. This result contradicts Cooper’s (2005) contention that positive humor would be a more effective strategy when the relationship with a target is low in terms of maturity and trust. The most plausible explanation for this unexpected finding is that, for people who remain in a group by obligation, OCB can be seen as a valid alternative to leaving this difficult or hostile environment. Some members may think that adopting extra-role behaviors, such as helping others or making constructive suggestions, will improve their relationship with their supervisor and their colleagues, and thus facilitate their inclusion. We can not rule out the idea that these extra-role behaviors are motivated by the intention to attract the attention of others (impression management) in the hope of being seen more favorably and better accepted (Bolino 1999; Chiaburu et al. 2015). Another explanation is that the employees’ risk of refraining from extra-role behavior and being punished is probably perceived as higher when employees are supervised by a distrusted leader with an offensive humor style. Untrustworthy leaders probably use offensive humor to ensure compliance. In contrast, when trusted leaders use offensive humor this may represent a violation of trust. In this case, lesser extra-role behavior is viewed as a retaliation tactic.Footnote 1 We cannot rule out that the relationship may be inverse, that OCBs predict inclusion and trust shown by others. In support of this argument, the LMX literature suggests that performance is a key determinant of the quality of exchange with the immediate supervisor (Nahrgang et al. 2009) and that a weak contribution to group goals may be one reason why groups put members at risk of exclusion (Robinson et al. 2013). Note that the highest level of OCBs is found among individuals exposed to a weak offensive humor climate and whose level of trust in their supervisor is high. This suggests that people who perceive consistent behaviors in trusted individuals are more motivated to cooperate and adopt extra-role behaviors because these consistent behaviors are more likely to reduce uncertainty in relational exchanges (Nahum-Shani et al. 2014).

Practical Implications

Our study finds that managers use different pathways to elicit citizenship behaviors, and provides insights into how and why humor climate may elicit such behaviors. First, managers must know that the use of offensive humor in the workplace may limit the beneficial effects of a positive humor climate, increase employee exclusion, and weaker performance. The positivist view that practicing humor always pays off should be put into perspective. Leaders who possess weak humor skills or have a strong tendency to prefer an offensive humor style gain little from managing by humor. Decision-makers must be aware that managers’ tolerance of offensive humor may have several drawbacks, such as more discrimination against minorities, higher turnover among newcomers, and lower cooperation among team members. Further, to promote a pervasive positive culture of humor (Mesmer-Magnus et al. 2012), HR departments must revise their recruitment strategies to ensure that they hire individuals with the right sense of humor. Our results also suggest that managers need to pay more attention to inclusion and ensure that all team members feel fully included. Leaders may promote inclusion by developing a close relationship with each member, treating each member fairly, and providing opportunities to influence decision-making (Findler et al. 2007). The present study suggests that trust in leaders is an important boundary condition for the efficacy of humor climates. Managers should be aware that a strong offensive humor culture is a risky inclusion strategy, particularly when relationships with employees are well established and based on high trust.

Limitations and Research Directions

Our study nonetheless has certain limitations. Because our sample is comprised of employees working in the same large financial organization, we cannot extrapolate our results to a larger scale. Second, given that the present study used a cross-sectional research design, we cannot firmly conclude that inclusion is the cause of OCB. The opposite may be true: OCB may be perceived as an indicator of inclusion. Third, risks associated with a common variance error may have influenced the relationships in our model. Considering that attitude variables and behaviors were measured simultaneously, and only by a single source (employees), there is a significantly high probability of a common variance error. As we have indicated in the “Methodology” section, a number of precautions have been taken to reduce such type of bias. However, the presumed upward bias in self-reported variables and that other-report measures produce superior sources is not shared by all scholars (Spector 2006; Conway and Lance 2010). A recent meta-analysis (Carpenter et al. 2014) has demonstrated that the mean difference in OCB between self and other raters is very small, and that self-rated and other-rated OCBs have a similar pattern of relationships with common correlates. In addition, results from moderated regressions are generally less sensitive to method variance bias (Evans 1985). Given that some of our hypotheses focused on the interactive effect of trust, our findings are probably relatively free of method bias effects.

Our results pave the way for future research. First, it would be interesting to submit our theoretical model to an empirical evaluation in a different context. We could then re-examine our research hypotheses to determine their robustness to new circumstances. This exercise should allow us to validate or refute this study’s results while evaluating their true generalization potential. Second, it would be interesting to study the causality of the relationships in the final model using a longitudinal design. As mentioned, the methodology used in this research does not allow us to conclude that a particular variable causes another. Thus, using a longitudinal design could considerably improve our level of knowledge and provide formal backing for the direction of such relationships. For example, we could examine whether extra-role performance has an equivalent effect on inclusion by coworkers and supervisors. To minimize the risks associated with common variance errors, it would be important to involve immediate supervisors in the process, for example, by asking them to evaluate the level of dissemination of humor and inclusion practices. Fourth, recent research has suggested that the motivations of those who are explicitly rejected differ from those who are ignored (e.g., Molden et al. 2009). It would be interesting to explore whether these two forms of exclusion have a differential impact on motivation to perform citizenship behaviors. Fifth, it would be worth considering other psychological mechanisms likely to come into play: the role of empowerment, of organizational commitment and of interpersonal justice. Finally, more studies are needed on the boundary conditions governing the effectiveness of humor. Lennox-Terrion and Ashforth (2002) proposed that the competition for scarce resources or rewards and task interdependence are important structural features that may influence the type of humor used and inclusion. According to the recent review by Robinson et al. (2013), newcomers are more likely to react differently to risks of exclusion or to rejection than are more senior employees. How these situational factors moderate the influence of exclusion on OCB and other behaviors, such as absenteeism and turnover, are potential fertile research avenues.

Conclusion

In conclusion, our results suggest that leader offensive humor climate is associated with a lower employee perception of inclusion, and subsequently lower extra-role performance (OCBs), whereas constructive leader humor climate is related to a higher display of OCB. However, the beneficial effects of humor climate depend on followers’ perception of degree of trust in their leader.