The Impact of Leader Humor on Subordinate Job Satisfaction

Abstract

Conventional wisdom from the popular and scholarly literatures consistently suggests that positive humor by leaders can be beneficial, but that negative humor should be avoided at all costs. To explore the boundaries of that conventional wisdom, we draw on leadership and humor theory to develop and test a conceptual model describing the relationships between leader humor, leader–subordinate relationship quality, the subordinate’s tenure with the leader, and subordinate job satisfaction. Analysis of multilevel data from 241 subordinates nested within 70 leaders in 54 organizations revealed that the relationship between leader humor and job satisfaction was dependent on the quality of the leader–subordinate relationship, and not the positive/negative tone of the leader’s humor. Specifically, both positive and negative (i.e., affiliative and aggressive) leader humor styles were positively associated with job satisfaction when the relationship was positive, but both types were negatively associated with job satisfaction when the relationship was negative. Our results also suggested that the effects of positive humor increased with increasing subordinate tenure. We discuss the practical implications of these findings, including the importance of understanding the relational context of humor.

Keywords

humor leadership humor styles job satisfaction LMX theory

Humor is woven into most domains of human life, including work contexts (Martin, 2007). The conventional wisdom touted in both the popular and scholarly literatures suggests that positive humor can be an effective tool for leaders, but that negative forms of humor will result in negative effects (A. Davis & Kleiner, 1989; Malone, 1980; Tarvin, 2012). Indeed, in line with conventional wisdom, scholars agree that there is a fundamental distinction between positive and negative humor (e.g., Robert & Yan, 2007), and there is some empirical support for a relationship between positive leader humor, positive subordinate outcomes, and positive perceptions of leaders (Mesmer-Magnus, Glew, & Viswesvaran, 2012).

But the limited literature on leader humor leaves a number of unresolved issues. First, past empirical research on leader humor is almost exclusively correlational, and has relied heavily on single-source designs in which subordinates provide all of the responses. As such, it is difficult to know whether observed correlations between positive leader humor, outcomes, and perceptions of the leader are driven by true associations, or by participants’ attempts to respond to surveys in a way that is consistent with conventional wisdom about humor (e.g., I like my supervisor, so she must use positive humor).

In addition, conventional wisdom tells us very little about the process that drives humor’s positive and negative effects. In this regard, Cooper’s (2002, 2008) work suggests a mediational model in which positive and negative humor influence positive and negative subordinate outcomes through their influence on the quality of the relationships that develop between leaders and subordinates. The emphasis on relationships as a mediating mechanism through which leader humor can affect subordinate outcomes represents an important insight that we build on in this article.

However, as we detail below, neither conventional wisdom nor the mediational model account for a number of findings in the empirical literature. In this study, we develop and test an alternative conceptual model that can account for those unexplained research findings. Extending Cooper’s (2008) relational approach, our model theorizes that leader–subordinate relationships mediate the relationship between leader humor and subordinate outcomes, and that the subordinate’s tenure with the leader moderates the relationship between leader humor and leader–subordinate relationship quality. However, our model’s most central prediction is that leader–subordinate relationship quality also affects the way in which a leader’s humor is perceived and interpreted. Specifically, we build on humor theory (Wyer & Collins, 1992) to argue that relationship quality is a key contextual variable that moderates the relationship between leader humor and job satisfaction. We propose that subordinates with a good relationship with their leader tend to perceive their leader’s humor positively, resulting in higher job satisfaction, whereas subordinates with a poor relationship with their leader will view their leader’s humor negatively, resulting in lower job satisfaction. Contrary to conventional wisdom, our conceptual model suggests that even leader humor that is explicitly negative in tone can be positively related to subordinate job satisfaction if the leader–subordinate relationship is positive, whereas even positive humor can be negative when the leader–subordinate relationship is negative. In the next sections, we review the literature pertaining to the conventional wisdom and mediational approaches, and then we develop arguments for the key distinctions made in our alternative conceptual model (see Figure 1), which we then examine empirically.

Figure 1.

Conventional wisdom, mediational, and proposed conceptual models.

Conventional Wisdom and the Mediational Model

If you ask people casually about humor in the workplace, the opinions expressed almost invariably indicate an appreciation of positive humor, coupled with caveats like “but some people take it too far.” This conventional wisdom about positive and negative humor is a major theme in the popular and practitioner literature. For example, Tarvin (2012) generally espoused positive humor, but recommended avoiding negative humor to avoid ostracizing others and to maintain a supportive environment. Similarly, Goudreau (2012) suggested that the benefits of humor only result when humor is empathic and has positive intent, and that negative humor undermines effectiveness and employee well-being. The basic structure of the conventional wisdom approach is depicted in Figure 1A.

These sentiments are largely echoed by scholars. Although empirical studies have focused predominantly on positive humor (Mesmer-Magnus et al., 2012), researchers take great pains to caution readers about the destructive nature of negative humor (e.g., Pundt & Herrmann, 2015). They typically stress the inherent dualities of humor and how it can be a “double-edged sword” (Malone, 1980), or that humor can be “misused” to humiliate or intimidate (Clouse & Spurgeon, 1995). Indeed, scholars generally support the conventional wisdom that negative humor is associated with negative outcomes such as poor work attitudes, decreased morale, and damaged interpersonal relationships (Romero & Cruthirds, 2006).

Empirical findings summarized in a recent meta-analysis provide some support for conventional wisdom with regard to the relationships between positive leader humor, positive ratings of leaders’ behavior and performance, and work attitudes and behaviors (Mesmer-Magnus et al., 2012). For example, Priest and Swain (2002) found that good leaders tended to be rated high in positive humor use, and Decker (1987) found that employees who rated their supervisors higher on sense of humor had higher job satisfaction.

But prior research supporting conventional wisdom has been almost entirely correlational and has relied almost exclusively on subordinate responses to questionnaires as the sole source of data. The concern with such designs is that participants might complete questionnaire measures in a manner that validates their perceptions of themselves, and/or confirms conventional wisdom about humor’s effects: good things, like positive humor, good leadership, and job satisfaction go together, as do negative things. Indeed, Decker, Yao, and Calo (2011) noted, “it is not clear whether these reports reflect reality or if the respondents distorted their perceptions” (p. 50).

Although conventional wisdom about positive and negative humor has provided a useful starting place for researchers, it does not contribute substantially to our understanding of the underlying processes by which positive humor affects subordinates. In this regard, Cooper’s (2002, 2008) work suggests that positive and negative humor by leaders affects subordinates and their outcomes by virtue of their ability to stimulate positive or negative affect within the leader–subordinate dyad, which influences the quality of the leader–subordinate relationship. Cooper’s (2002) work also leverages leader–member exchange (LMX) theory’s notion that subordinate attitudes and behaviors are influenced by the leader–subordinate relationship. LMX theory suggests that when the relationship is high quality and positive, as characterized by trust, respect, and mutual obligation, the leader allows more latitude in subordinate behavior and in negotiating his or her work role, and provides a more positive environment for subordinates (Graen & Uhl-Bien, 1995). Subordinates reciprocate with positive work-related behaviors and attitudes, including job satisfaction (Liden, Sparrowe, & Wayne, 1997). In negative relationships, there is limited trust and respect, and obligations are based exclusively on an exchange of instrumental rewards for effort. Such negative relationships undermine work-related attitudes.

The integration of Cooper’s (2008) relational process model of humor with LMX theory suggests a mediated process model (depicted in Figure 1B) by which leader humor affects subordinate attitudes (e.g., job satisfaction) through its impact on the leader–subordinate relationship. Supporting the first part of this model, Pundt and Herrmann (2015) found that positive (i.e., affiliative) humor was positively associated with LMX quality, whereas negative (i.e., aggressive) humor was negatively associated with LMX quality. Cooper (2002) also examined the basic mediated relationship, and her results indicated that the relationships between leader humor frequency and both turnover intentions and organizational citizenship behaviors (OCB) were mediated by some dimensions of LMX. Similarly, Gkorezis, Petridou, and Xanthiakos (2014) found that LMX quality mediated the relationship between positive humor and (lower) organizational cynicism. Therefore, some limited initial support for this basic mediated process has been obtained previously, and we suggest that this model provides a useful starting point for integrating humor and leadership theory.

However, a number of counterintuitive findings about positive and negative humor by leaders cannot be explained by conventional wisdom or by the mediational model. For example, Sala (2000) found that high-performing leaders used more positive humor, but also more negative humor (i.e., derogating peers and subordinates) than those with average performance. Also, Cooper (2002) found that leader humor frequency and negative hedonic tone interacted in an unexpected way, such that subordinates reported more positive LMX quality when the leader’s humor was high in frequency and negative in tone than when it was low in frequency and negative in tone. Other research supports the notion that positive leader humor is related to positive relationships and subordinate outcomes, but does not strongly support the reverse predictions for negative humor. For example, Decker and Rotondo (2001) found that although positive humor was correlated with positive leader behaviors and outcomes, negative humor was uncorrelated with positive leader behavior and outcomes. Similarly, Pundt and Herrmann (2015) found that although positive (i.e., affiliative) humor was related to identification with the leader, negative humor was unrelated to identification with the leader.

Although such findings suggest that the mediational model may not adequately explain all previous findings in the literature, we will examine the mediated model as a point of departure for examining our proposed conceptual model. Importantly, our examination of the mediational model represents a more rigorous test than earlier studies, in that our study is the first to examine the full mediated process using both negative and positive forms of leader humor, and independent sources of data (i.e., both leader and subordinate).

Proposed Conceptual Model

The primary goal of this article is to develop and test a conceptual model that proposes two important extensions of the mediational model. As depicted in Figure 1C, we propose that (a) a subordinate’s tenure with his or her supervisor will moderate the relationship between leader humor and leader–subordinate relationship quality (Hypothesis 1), and (b) the quality of the leader–subordinate relationship will moderate the relationship between leader humor and subordinate outcomes (Hypothesis 2).

Hypothesis 1 proposes that tenure with the leader will amplify the positive effects of positive leader humor, but will dampen the negative effects of negative humor. This prediction is derived from humor theory and LMX theory. With regard to positive humor, Robert and Wilbanks’s (2012) wheel model of humor suggests that positive humor tends to have increasingly strong effects within dyads or social groups over time. Specifically, they suggest that positive affect and positive relationships create a social climate in which humor is more likely to be accepted, and to have a stronger affective punch. This occurs in part because individuals learn how and when to use humor that is pitched at the right level for the audience, and in part because individuals develop a stronger learned association between the use of humor and experienced positive affect. In other words, a positively toned joke that is used early in a relationship is likely to result in a less positive response by a target than the same joke used in a more established relationship (i.e., with higher tenure). This idea is also consistent with Wyer and Collins’s (1992) notion that people pay more attention to what makes humor funny as they become less concerned about things like an individual’s motivation (i.e., ulterior motives) for using humor.

This pattern is also consistent with the LMX theory notion that during the earlier phases of leadership building (i.e., the stranger and acquaintance stages; Graen & Uhl-Bien, 1995), the relationship between the leader and subordinate is constrained to the more formal economic exchanges between the leader and subordinate. It is not until later, during the “mature partnership” stage, that emotional connections based on respect, trust, and obligation develop between individuals. As a result, we propose that positive humor is unlikely to be viewed as appropriate or normative during the earlier phases of a relationship, but will become more normative and acceptable over time as a relationship matures.

With regard to negative humor, we predict the opposite pattern. Although positive humor might not be fully appreciated during the early stranger and acquaintance stages of LMX development, aggressive humor might be viewed as a signal that the leader does not want the relationship to proceed into the mature partnership stage. Aggressive humor by leaders is often perceived as an attempt to accentuate hierarchical differences between leaders and subordinates, and to create social distance (Robert & Yan, 2007). In this regard, if subordinates are waiting for an “offer” of an improved relationship from their leader (Graen & Uhl-Bien, 1995), aggressive humor might indicate that such an offer is not forthcoming. However, as tenure increases, relationships between leaders and subordinates tend to stabilize (Nahrgang, Morgeson, & Ilies, 2009). As such, day-to-day negative humor becomes less diagnostic for determining the nature of the leader–subordinate relationship, and the subordinate might already have well-established beliefs about their leader’s motivations. A negatively toned joke by a supervisor early in a relationship might have a devastating effect on the subordinate’s perception of the relationship, whereas the same joke might be brushed off by a high-tenure target in a more stable relationship with the humorist. We propose the following:

Hypothesis 1: Affiliative humor has an increasingly positive effect on leader–subordinate relationships as tenure increases, but the negative effect of aggressive humor on leader–subordinate relationships decreases as tenure increases.

The remaining aspect of our conceptual model is the most central to this study. We argue that although leader humor serves to influence the quality of leader–subordinate relationships over time, humor might also be interpreted in the context of the existing relationship. We predict that when relationships are positive, leader humor with positive tone, and even negative tone, will be associated with more positive subordinate outcomes. However, when the relationship is negative, both positive and negative forms of leader humor will be associated with negative outcomes. In other words, we propose that the impact of leader humor depends on the quality of the existing relationship rather than the positive or negative tone of the humor.

This hypothesis is rooted in observations and theory from the humor literature. First, although humor is primarily a form of interpersonal communication (Lynch, 2002), it can be quite challenging to decipher. There are many different motivations for using humor in interpersonal contexts (Robert & Yan, 2007), and humor can fall along the broad continuum that ranges from extremely positive and prosocial, to extremely negative and destructive (Martin, Puhlik-Doris, Larsen, Gray, & Weir, 2003). Moreover, the use of incongruity within humor allows its underlying message to be quite subtle and nuanced (Martin, 2007), which is complicated further by the fact that humorists can use many different humor forms (e.g., sarcasm, irony, understatement, overstatement, and wordplay; Norrick, 2003).

Thus, interpreting humor and the humorist’s underlying intent and motivations is often difficult. Those who experience humor must rely on their complex understanding of the social context to determine its meaning (M. S. Davis, 1993). Wyer and Collins’s (1992) theory of humor elicitation suggests that individuals who are exposed to humor engage in pragmatic information processing of humor to “try to understand why the communication might have been conveyed and, therefore, what the communicator actually intended to say” (p. 672). In particular, they argue that contextual cues are central to determining the intent of the humorist (e.g., positive or negative), and influence whether the humor is perceived as amusing. The notion that contextual cues are important for evaluating humor parallels recent findings in leadership research, which indicate that subordinates’ cognitions and attributions about leaders’ intentions play an important role in determining how subordinates perceive their leaders (Eberly & Fong, 2013).

With respect to leader humor, we borrow from LMX theory and Cooper’s (2002, 2008) insights into the importance of leader–subordinate relationships in driving humor’s effects to suggest that the existing relationship between a leader and subordinate is a crucial contextual cue for subordinates as they process leader humor. We suggest that a subordinate’s perception of his or her relationship with his or her leader can serve as a useful backdrop against which a leader’s behavior can be interpreted, because it represents their accumulated experience with regard to past interactions. As such, it will shape subordinates’ beliefs about whether the motives and intentions behind a leader’s interpersonal behavior are positive or negative.

LMX theory (e.g., Graen & Uhl-Bien, 1995) suggests that high-quality leader–subordinate relationships are more social in nature, and involve an emotional connection. In such contexts, leader humor is more likely to be seen as normal social discourse. Moreover, leaders will receive the benefit of the doubt when subordinates attempt to interpret humor that could be ambiguous. Indeed, because people with a positive leader–subordinate relationship are likely to perceive their leaders’ motivations and intentions as positive, even humor that is perceived by outside observers as negative might be perceived positively by subordinates with good leader–subordinate relationships. In contrast, when leaders and subordinates have a low-quality relationship, their interactions focus on the formal aspects of the job and task completion in exchange for compensation. Given the lack of emotional connection between leader and subordinate, in poor relationship contexts subordinates are likely to be skeptical about leaders’ motivations and intentions. Humor behavior, even that which is outwardly positive in tone, might be perceived as frivolous, unnecessary, and a waste of the subordinate’s time (Avolio, Howell, & Sosik, 1999), or even manipulative or deceptive.

To illustrate, imagine a leader making an ambiguous statement to a subordinate; “You could be a bit more of a perfectionist.” Tone and facial expression might convey that the statement involves humor (Martin, 2007), but the subordinate’s reaction might range from amused to offended, depending on his or her attributions about the leader’s behavior. If the leader and subordinate have a good relationship, the subordinate is likely to perceive the “perfectionism” comment as sarcastic, and perhaps even as a veiled compliment. However, in the context of a poor leader–subordinate relationship, the subordinate might perceive that even though the comment was made in a manner that would allow the leader to mask his or her true intent (e.g., “I was only joking”), the message either accurately reflects its semantic content (i.e., the subordinate should pay more attention to detail), or uses understatement to drive humor (i.e., the subordinate needs to pay a lot more attention to detail).

This perspective on the importance of context within the broad gray area of interpersonal humor is supported by research on teasing. This literature has found that even behaviors that appear to be objectively negative in tone to outsiders can be perceived positively by insiders, and can have positive effects on the audience. For example, Scogin and Pollio (1980) found that people who targeted others with aggressive teasing humor tended to be targets of teasing themselves, and they suggested that this was indicative of bonding and a sense of inclusion that supports group harmony. Similarly, Terrion and Ashforth (2002) found that incidences of aggressive “put-down” humor were actually an indicator of growing trust and cohesiveness.

Based on the integration of humor theory and leadership theory, our prediction adds an important alternative perspective to the straightforward nature of conventional wisdom, and potentially helps to explain some of the results obtained in the previous literature that are not consistent with the mediational model (e.g., where negative humor is associated with good or neutral outcomes). Specifically, we propose the following:

Hypothesis 2: The relationship between leader humor and subordinate job satisfaction will depend on the quality of the leader–subordinate relationship. Specifically, both positive and negative humor will improve job satisfaction when the leader–subordinate relationship quality is high, but both positive and negative humor will decrease job satisfaction when the relationship quality is low.

Method

Procedure

To test our hypotheses, multilevel data were collected from mid-level managers (leaders) and their direct subordinates in 54 organizations. The leader sample was obtained from two executive education programs at a large university in Hong Kong. Two sets of questionnaires were developed: one for leaders and one for their subordinates. Leaders completed and returned the leader questionnaire, and were asked to distribute a cover sheet and questionnaire, at random, to at least four subordinates. Each subordinate questionnaire included a unique identification number so that it could later be linked to leader data. Subordinates returned questionnaires directly to the research team by mail to ensure confidentiality. A total of 345 subordinate questionnaires were distributed and 241 questionnaires were returned, for a response rate of 70%. Subordinate respondents were nested within 70 leaders, and the average number of subordinate responses for each leader was 3.44 (SD = 2.8).

Sixty-three percent of subordinates and 54% of leader respondents were female. The average age of subordinates was 31.4 years (SD = 7.3), whereas the mean age of leaders was 37.8 years (SD = 7.5). The average organizational tenure for subordinate respondents was 4.7 years (SD = 4.8), subordinates’ tenure with their current leader was 3.14 years (SD = 3.55), and the average leader tenure was 8.2 years (SD = 6.5).

Translation

Items were translated from English into Chinese. A back-translation was done to ensure that both English and Chinese versions were comparable (Brislin, Lonner, & Thorndike, 1973). Finally, two Chinese native speakers checked the translations for accuracy.

Measures

Job satisfaction

A three-item scale on the subordinate questionnaire was used to assess subordinate job satisfaction (Cammann, Fichman, Jenkin, & Klesh, 1979). Responses to the items were recorded on a 1 (strongly disagree) to 7 (strongly agree) scale. A sample question was “In general, I enjoy working here” (α = .86). There is considerable empirical support for the positive relationship between leader–subordinate relationship quality and job satisfaction in the LMX literature (e.g., Dulebohn, Bommer, Liden, Brouer, & Ferris, 2012), and job satisfaction has been one of the more well-studied outcomes of leader humor (Mesmer-Magnus et al., 2012).

Leader–subordinate relationship quality

A seven-item measure developed by Graen, Novak, and Sommerkamp (1982) measured leader–subordinate relationship quality on the subordinate questionnaire, using a 5-point Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). A sample item from this scale is “My supervisor understands my problems and needs” (α = .75).

Leader humor

Leader humor behavior was measured on the leader questionnaire using Martin et al.’s (2003) Humor Styles Questionnaire (HSQ). The HSQ measures two interpersonal humor styles, which are labeled affiliative (positive) and aggressive (negative). It also measures two intrapersonal styles (i.e., self-enhancing/positive and self-defeating/negative), which involve humor behavior used to regulate one’s own cognitions and emotions rather than humor directed toward others. Given that the focus of the current study was on examining how humor behaviors influence relationships, and how the relationship between a leader and subordinate can influence the perceptions of humor used in interpersonal contexts, we followed Pundt and Herrmann (2015) in focusing on the affiliative and aggressive styles to define positive and negative humor, respectively, in the current study. All four humor styles are measured with eight items pertaining to the respondent’s use of humor using a 7-point Likert-type scale with options ranging from 1 = totally disagree, to 7 = totally agree. Sample items from the affiliative humor scale are “I have a tendency to share humor with others,” and “I usually don’t like to tell jokes or amuse people” (reverse coded). Sample items on the aggressive humor style scale are “I have a tendency to use humor to disparage others,” and “People are never offended or hurt by my sense of humor” (reverse coded). It is important to emphasize that the aggressive humor scale is not designed to measure light-hearted or friendly teasing. As Martin et al. (2003) noted, the scale measures sarcasm, ridicule, manipulation, and disparagement. The reliability and validity of the HSQ have been assessed and supported in research contexts outside of North American samples and in cross-cultural studies (e.g., Kazarian & Martin, 2004), and specifically in Chinese samples (Chen & Martin, 2007). In the current sample, the reliability for the affiliative humor scale was α = .81, and for aggressive humor it was α = .65.

Tenure with supervisor

This continuous variable was measured using self-report responses by subordinates, who were asked to indicate the amount of time (in years) they had worked for their current supervisor.

Control variables

Several variables were included as controls in our analyses. Because humor takes place in the context of the social relationship between the leader and subordinate that might vary systematically over time, we included the tenure with supervisor variable (see above) as a control in our analyses of the mediational model. For Hypothesis 1, tenure was predicted to be a moderator. In addition, because job satisfaction, humor use, and responses to humor are likely linked to affect, especially positive affect (Robert & Wilbanks, 2012), we controlled for subordinate trait positive affectivity (labeled Subordinate PA in our tables) using subordinate responses to Watson, Clark, and Tellegen’s (1988) PANAS measure. Although our proposed model focused on the impact of leader humor behavior rather than leader characteristics, we controlled for the sex of the leader to account for the possibility that subordinate responses to different leadership styles might vary based on the leader’s sex (Eagly & Johannesen-Schmidt, 2001).

Our hypothesis tests focus on the interpersonal affiliative and aggressive humor styles, because interpersonal humor is explicitly intended to affect others. However, we also included the two intrapersonal humor styles (i.e., self-enhancing, and self-defeating) as control variables in our analysis. Although these two styles are thought to help individuals manage their own emotions, it is possible that subordinates still detect these types of humor by leaders, and might be affected by them. For example, a leader’s use of self-defeating humor (e.g., “I guess I’m getting less and less useful around here!”), although designed to help the leader manage his or her own emotions, might still affect subordinates’ perceptions. Thus, we treated the self-enhancing and self-defeating humor styles as control variables.

Analysis

Our data were from subordinates nested within leaders. Level 1 variables included subordinate responses regarding leader–subordinate relationship quality, job satisfaction, and two control variables (i.e., tenure with leader and subordinate positive affectivity). At Level 2 were leaders’ responses regarding humor style (i.e., affiliative, aggressive, self-enhancing, and self-defeating). The nested nature of the data required the use of multilevel analytical methods. Prior to examining Hypotheses 1 and 2, we examined the basic mediational model as a point of departure from previous literature. Consistent with prior research examining cross-level mediation (Preacher, Zyphur, & Zhang, 2010), we utilized multilevel structural equation modeling (MSEM) using the Mplus statistical package (Muthén & Muthén, 1998-2012) to test the mediational model. Following recommendations from Preacher, Zhang, and Zyphur (2011), we analyzed a 2-1-1 mediation model to examine whether leader humor affects subordinate job satisfaction through leader–subordinate relationship quality. Our analysis consisted of a one-step procedure, simultaneously testing direct and indirect pathways of leader humor for job satisfaction through leader–subordinate relationship quality. Although similar in nature to commonly used mediation techniques (Baron & Kenny, 1986), this procedure bypasses the first step of examining the direct effect of the predictor variable on the outcome variable, as that criterion for establishing mediation is increasingly questioned (Croon & van Veldhoven, 2007).

Next, hierarchical linear modeling (Raudenbush & Bryk, 2002) was used to test Hypotheses 1 and 2, which involved cross-level interaction effects. These included the interaction of leader humor and tenure (with current supervisor) in predicting leader–subordinate relationship quality (Hypothesis 1), and the moderating role of leader–subordinate relationship quality in the relationship between leader humor and subordinate job satisfaction (Hypothesis 2).

We followed the procedures described as best practices by Aguinis, Gottfredson, and Culpepper (2013) for analyzing cross-level moderation effects. Specifically, the moderation effects were estimated using a four-step multilevel model building process. Those steps included tests of (a) a null model, (b) a random intercept and fixed slope model (RIFSM), (c) a random intercept and random slope model (RIRSM), and (d) a cross-level interaction model.

Results

Mediational Model Results

Descriptive statistics and correlations among focal variables in our study appear in Table 1. Prior to testing our hypothesis, we confirmed the need for a multilevel analysis of our mediational model by examining the intraclass correlation coefficient (ICC). The ICC for job satisfaction was .16, suggesting that 16% of the variance in subordinate job satisfaction can be accounted for at the leader level (Level 2), and that cross-level investigation is warranted (Mathieu, Aguinis, Culpepper, & Chen, 2012).

Table 1.

Means, Standard Deviations, and Bivariate Correlations Among Study Variables.

Variable	M	SD	1	2	3	4	5	6	7
1. Subordinate PA	3.60	0.73
2. Tenure with supervisor	3.14	3.55	.10
3. Self-enhancing humor	4.25	0.79	.25**	−.27**
4. Self-defeating humor	3.38	0.78	.01	−.12	.20**
5. Affiliative humor	4.55	0.82	.17*	−.13*	.52**	.45**
6. Aggressive humor	3.00	0.74	.01	−.12	.09	.36**	.08
7. Leader–subordinate relationship quality	3.73	0.49	.40**	.21**	.06	.05	.13*	.01
8. Job satisfaction	5.15	1.22	.30**	.22**	.09	−.06	.08	−.08	.51**

Note. Variables in bold indicate Level 2 measures.

p < .05. **p < .01.

The mediational relationship that we tested as a point of departure from the existing literature was not supported by the results of our MSEM analysis. Interestingly, although leader–subordinate relationship quality was related to job satisfaction, neither positive nor negative leader humor had direct effects on subordinate job satisfaction (outcome variable), or on leader–subordinate relationship quality (mediator variable). In addition, there were no indirect effects of leader humor on the subordinate job satisfaction measure through leader–subordinate relationship quality. Thus, the data did not provide support for the mediational model.

Results for the Moderation Hypotheses

Hypothesis 1 predicted an interaction between leader humor and tenure with supervisor in predicting leader–subordinate relationship quality, and Hypothesis 2 predicted an interaction between leader humor and leader–subordinate relationship quality in predicting subordinate job satisfaction. We conducted a four-stage model building process to test both hypotheses, separately. The same procedure was used to examine multilevel moderation relationships proposed in Hypotheses 1 and 2, and results of those analyses can be found in Tables 2 and 3, respectively. Sex of the leader was included as a control variable in our initial analyses, but it did not have a significant effect, nor did it substantively change our results relative to models that did not include leader sex as a control. We report results of models that did not include leader sex. Variables in each of the models were group mean centered, following recommended practices (Aguinis et al., 2013; Hofmann & Gavin, 1998).

Table 2.

Results of Multilevel Modeling Analysis for Predicting Leader–Subordinate Relationship Quality.

Predictor variables	Leader–subordinate relationship quality
Predictor variables	Null model	Random intercept and fixed slope	Random intercept and random slope	Cross-level interaction
Fixed effects
Level 1
Intercept	3.71** (0.06)	3.72** (0.06)	3.72** (0.06)	3.72** (0.06)
Subordinate PA		0.23** (0.08)	0.29** (0.08)	0.23** (0.08)
Tenure with supervisor		0.01 (0.02)	0.02 (0.02)	0.03 (0.02)
Level 2
Affiliative humor		0.06 (0.10)	0.06 (0.10)	0.06 (0.10)
Aggressive humor		0.13 (0.07)	0.12 (0.07)	0.13 (0.07)
Self-enhancing humor		0.18 (0.11)	0.17 (0.10)	0.17 (0.11)
Self-defeating humor		−0.11 (0.07)	−0.10 (0.06)	−0.11 (0.07)
Cross-level interaction
Tenure with supervisor × Affiliative humor				0.04* (0.02)
Tenure with supervisor × Aggressive humor				−0.03 (0.02)
Random effects
Residual	0.140	0.112	0.1110	0.106
Intercept	0.071	0.058	0.0590	0.060
Slope			0.0004	0.002
Intercept–slope covariance			−0.0040	0.007
Additional information
ICC	0.34
Deviance	98.78	79.81**	79.46	76.63

Note. Values in parentheses are robust standard errors of each regression coefficient. Significant deviance values indicate better fit than the previous step in the model building process. ICC = intraclass correlation coefficient.

p < .05. **p < .01.

Table 3.

Results of Multilevel Modeling Analysis for Predicting Subordinate Job Satisfaction.

Predictor variables	Job satisfaction
Predictor variables	Null model	Random intercept and fixed slope	Random intercept and random slope	Cross-level interaction
Fixed effects
Level 1
Intercept	5.10** (.17)	4.88** (0.27)	4.88** (0.50)	4.93** (0.27)
Subordinate PA		0.62** (0.16)	0.62** (0.16)	0.58** (0.15)
Tenure with supervisor		0.08 (0.05)	0.08 (0.05)	0.07 (0.06)
Leader–subordinate relationship quality		0.71** (0.20)	0.70** (0.20)	0.51** (0.17)
Level 2
Affiliative humor		−0.10 (0.26)	−0.10 (0.27)	−0.10 (0.26)
Aggressive humor		−0.03 (0.16)	−0.03 (0.16)	0.07 (0.18)
Self-enhancing humor		0.66** (0.23)	0.55* (0.22)	0.66** (0.23)
Self-defeating humor		−0.33 (0.18)	−0.38* (0.21)	−0.34 (0.18)
Cross-level interaction
Leader–subordinate relationship quality × Affiliative humor				1.11* (0.59)
Leader–subordinate relationship quality × Aggressive humor				0.60** (0.23)
Random effects
Residual	1.113	0.776	0.775	0.726
Intercept	0.595	0.240	0.240	0.250
Slope			0.005	0.001
Intercept–slope covariance			−0.030	0.007
Additional information
ICC	0.62
Deviance	265.28	234.61**	234.59	230.43

Note. Values in parentheses are robust standard errors of each regression coefficient. Deviance statistics indicate fit values for each model. Significant deviance values indicate better fit than the previous step in the model building process. ICC = intraclass correlation coefficient.

p < .05. **p < .01.

The first step in the model building process for examining Hypothesis 1 involved testing a null model for leader–subordinate relationship quality to determine whether multilevel methods are necessary for examining variability in this Level 1 variable. Results from the null model indicated a significant intercept for leader–subordinate relationship quality as an outcome variable (γ₀₀ = 3.71, p < .01). In addition, the ICC for leader–subordinate relationship quality was .34, indicating that 34% of the variance in the dependent variable can be attributed to leader-level effects. These steps indicate that examining leader humor as a Level 2 variable is warranted. The second step in the model building process, referred to as the RIFSM (Aguinis et al., 2013), tested the direct effect of tenure with supervisor (single-level effect) and leader humor (cross-level effect) on leader–subordinate relationship quality, respectively. Results from this step showed no relationship between tenure with supervisor and leader–subordinate relationship quality (γ₁₀ = .01). In addition, the relationships between leader humor and leader–subordinate relationship quality had positive coefficients, but were not significant (Affiliative, γ₀₁ = .06; Aggressive, γ₀₂ = .13). The model in Step 3, referred to as the RIRSM, added a random slope parameter to test whether the relationship between tenure with supervisor and leader–subordinate relationship quality varied across leaders. Results from the model were similar to step two, and the model provided no better fit (χ² = 0.34, ns). We discuss the implications of this at the end of this section.

The final step in the model building process, referred to as the cross-level interaction model, tested whether tenure with the supervisor moderated the relationship between leader humor and leader–subordinate relationship quality. Results from this analysis provided partial support for Hypothesis 1. Specifically, the coefficient for the interaction was significant and positive (γ₁₁ = .04, p < .05), suggesting that although the direct relationship between affiliative leader humor and leader–subordinate relationship quality was positive but non-significant, tenure with the supervisor significantly strengthened that relationship. However, Hypothesis 1 was not supported with respect to aggressive leader humor, as tenure did not moderate the relationship between aggressive humor and relationship quality (γ₁₂ = −0.03, ns).

Hypothesis 2 predicted that leaders’ positive and negative humor would interact with subordinates’ perceptions of leader–subordinate relationship quality in predicting subordinate job satisfaction. To test this hypothesis, we followed the same four-step procedure outlined above. A test of the null model revealed significant variability in job satisfaction that can be attributed to Level 2 predictors (γ₀₀ = 5.10, p < .001). The ICC for the model was .62, indicating that 62% of employee job satisfaction is determined at the leader level, justifying multilevel analysis. The RIFSM tested in Step 2 showed a positive relationship between leader–subordinate relationship quality and employee job satisfaction (γ₁₀ = 0.71, p < .001). However, neither affiliative (γ₀₁ = −0.10, ns) nor aggressive humor (γ₀₂ = −0.03, ns) were directly related to job satisfaction. This model did however provide a better fit to the data than the null model (χ² = 30.67, p < .01). Step 3 tested the RIRSM, and produced results similar to those found in Step 2, with only leader–subordinate relationship quality affecting job satisfaction (γ₁₀ = 0.70, p < .001). The fit of this model did not improve on the model from Step 2 (χ² = 0.01, ns). Finally, Step 4 tested the cross-level interaction model, providing a test of the predicted moderation relationship from Hypothesis 2. As Table 3 shows, results from this analysis provided support for our hypothesis. Specifically, increases in the use of affiliative humor strengthened the relationship between leader–subordinate relationship quality and job satisfaction (γ₁₁ = 1.11, p < .05). Similarly, and consistent with the counterintuitive relationship posited in Hypothesis 2, leader use of aggressive humor strengthened the relationship between relationship quality and subordinate job satisfaction (γ₁₂ = 0.60, p < .01). Although the results from this model support the predictions in Hypothesis 2, the full model did not significantly improve on the fit from the previous model (χ² = 4.15, ns).

Conceptually, the Level 2 humor variables can be seen as affecting the slope of the relationship between the Level 1 variables. We have graphed the results in Figure 2 in a manner that enables one to visualize the results relative to our theorizing, which focused on leader–subordinate relationship quality as a contextual variable. Consistent with our hypothesis, the plots of the interactions indicate that there was a positive relationship between leader affiliative and aggressive humor styles and subordinate job satisfaction when the leader–subordinate relationship quality was high. Also consistent with our prediction, the relationship between leader affiliative and aggressive humor and subordinate job satisfaction was negative when leader–subordinate relationship quality was low.

Figure 2.

Interactive effects of humor style and leader–subordinate relationship quality on job satisfaction.

To confirm the visual pattern observed in these interaction plots (i.e., Figure 2) we evaluated the simple slopes following the process described by Preacher, Curran, and Bauer (2006). Analyses indicated that the relationships between both affiliative(simple slope = 0.706 (0.331), t = 2.13, p < .05) and aggressive (simple slope = 0.495 (0.232), t = 2.13, p < .05) humor styles and subordinate job satisfaction were positive and significant when the leader–subordinate relationship quality was high (+1 SD). Similarly, when the leader–subordinate relationship quality was low (−1 SD), the slopes of the relationships between affiliative (simple slope = −0.584 (0.273), t = −2.14, p < .05) and aggressive humor styles(simple slope = −1.042 (0.489), t = −2.13, p < .05) and job satisfaction were significant and negative.

It is important to note that comparisons of models in Steps 2 to 3 for tests of both Hypotheses 1 and 2 did not indicate a significant improvement in fit. This is sometimes interpreted to suggest that slopes of the relationships between Level 1 variables do not vary across groups (Level 2). However, in both cases, we followed the recommendation of Aguinis et al. (2013), which is “to proceed with the cross-level interaction test even when the null hypothesis of no slope variance is retained” (p. 1502), given that our test of the interaction was guided by theory. This recommendation is also consistent with Snijders and Bosker’s (2012) observation that the interpretation of the random slope model fit and changes in fit can be complicated, as sometimes even the addition of significant explanatory variables can result in the appearance of unchanged or worse model fit. Similarly, LaHuis and Ferguson (2009) noted that

the significance tests for slope variance components do not always reflect what is happening in terms of how Level 2 variables are related to the slope. This suggests that these tests may not be useful in determining when significant cross-level interactions are possible. (pp. 430-431)

Moreover, particularly with regard to the disordinal (i.e., crossover) interaction patterns obtained for Hypothesis 2, the cross-level interaction pattern confirmed by the simple slope tests was striking rather than subtle, increasing our confidence that interpreting the interaction pattern is meaningful.

Discussion

This study was designed to build on conventional wisdom about the effects of leader humor, and to extend Cooper’s (2002, 2008) insights into the importance of the leader–subordinate relationship. Leveraging humor theory and LMX theory, we developed a conceptual model suggesting that humor mediates the relationship between leader humor and subordinate outcomes such as job satisfaction, but that it also represents an important contextual condition that moderates the impact of leader humor. In addition, based on LMX theory (Graen & Uhl-Bien, 1995) and humor theory (i.e., Robert & Wilbanks, 2012), we proposed that tenure would moderate the relationship between leader humor and leader–subordinate relationship quality such that tenure would amplify the positive effects of positive humor, but dampen the negative effects.

Although our conceptual model was inspired by conventional wisdom about humor’s effects, and the mediational model derived from Cooper’s (2002, 2008) work, our results paint a very different picture of leader humor and its effects. In contrast to conventional wisdom, we found that the effects of leaders’ aggressive and affiliative humor on subordinates’ job satisfaction had little to do with the positive or negative tone of the humor. Instead, subordinate job satisfaction was driven primarily by subordinates’ perceptions of the quality of the leader–subordinate relationship.

Consistent with humor theory (Wyer & Collins, 1992), we argue that the quality of the leader–subordinate relationship represents an important backdrop against which subordinates interpret leader humor. If subordinates’ relationships with their supervisors are positive, they are likely to view humor that the leader intended to be positive (e.g., affiliative humor) in a positive manner, as such behavior is entirely consistent with past behavior and expectations for the leader. However, our results indicate that even when leaders use humor that they themselves classify as aggressive, subordinates who perceive a positive relationship with their leader are likely to respond positively. We propose that because much humor falls within a broad gray area between clearly positive and clearly negative, individuals fall back on their accumulated experience with their leader to interpret their behavior. When past experience indicates a positive leader–subordinate relationship, even aggressive humor can be interpreted in a positive light. This explanation is consistent with research suggesting that humor that is perceived as aggressive to outside observers (e.g., put-downs and deprecation) can be associated with social acceptance and in-group status by insiders who understand the social context (Scogin & Pollio, 1980; Terrion & Ashforth, 2002).

In contrast, when the leader–subordinate relationship is negative, both positive and negative forms of humor might be perceived negatively. This is fairly straightforward for negative forms of humor (e.g., aggression), in that subordinates who perceive a negative relationship might be especially likely to expect negative behavior, and to interpret negative forms of humor such as aggression in a negative light. However, in the context of a negative relationship, characterized by a focus on economic exchange between the leader and subordinate, we found that even humor that leaders intend to be positive might have negative effects. Such behavior might be viewed as an inappropriate use of time, a distraction (Avolio et al., 1999), or even a misguided attempt by the leader to ingratiate himself or herself (Cooper, 2005).

The theoretical notion that the leader–subordinate relationship is an important contextual variable might help explain a number of previous empirical results that cannot be explained by the simple proposition that positive humor is good and negative humor is bad, or the mediational model. For example, Decker and Rotondo (2001) found that although positive humor was related to positive leader behaviors, negative humor was unrelated to positive humor behaviors. This result might be expected if subordinates and leaders in their sample already had positive relationships that could attenuate the impact of negative humor. Similarly, Cooper (2002) found that when leader humor was high in frequency and negative in tone, leaders’ relationships with subordinates were more positive than when they were low in frequency and negative in tone, and Sala (2000) found that leaders who used more negative humor were actually rated as more effective than leaders who used less negative humor. Again, extrapolating from our model, it is possible that if the relationships between leaders and subordinates in Cooper’s and Sala’s studies were generally positive, then even humor with an ostensibly negative tone (e.g., teasing, put-downs) might have been interpreted positively by subordinates.

We also explored the issue of whether the subordinates’ tenure with their supervisor might moderate the relationship between positive and negative leader humor and leader–subordinate relationship quality (Hypothesis 1). We predicted that tenure would amplify the positive effects of positive humor, and dampen the negative effects of negative humor. Our results provide partial support for the hypothesis. With regard to positive (i.e., affiliative) humor, subordinate tenure significantly moderated the relationship between leader humor and leader–subordinate relationship quality such that the relationship was significant for high-tenure subordinates. This finding generally supports our contention that although positive humor might not be particularly effective in the early relationship-building stages of the leader–subordinate relationship, positive humor might have more of an effect on longer tenure individuals who are less concerned about extraneous information about the humor (e.g., the leader’s motivation; Wyer & Collins, 1992). Positive humor might also initiate a spiraling cycle of positive affect that improves the relationship climate and supports additional humor and positive affect (Robert & Wilbanks, 2012).

However, we did not find support for our hypothesis with respect to negative (aggressive) humor. Aggressive humor was unrelated to subordinates’ perceptions of leader–subordinate relationship quality, and this (lack of) relationship was not moderated by the subordinate’s tenure with the leader. One possible explanation for the failure to support this part of the hypothesis is that we might have had relatively few individuals in the sample with tenure that was sufficiently low to detect an effect. Specifically, our hypothesis was driven by the notion that negative humor that occurs early in the relationship-building stages (i.e., stranger and acquaintance stages; Graen & Uhl-Bien, 1995) might be particularly harmful because it signals a lack of desire on the part of the leader to have the relationship progress into a “mature partnership” (Graen & Uhl-Bien, 1995). Although such behavior might be less harmful after the relationship has reached a point of stability, researchers have found that the leader–member relationship can reach a stable plateau relatively quickly (Nahrgang et al., 2009). Given that our participants had generally been with their leader for quite a long time on average (M = 3.1 years), it is possible that we did not have enough people in our sample with sufficiently low tenure to detect the negative effects of negative humor early in the relationship.

Our results also failed to support mediation of the relationship between leader humor and job satisfaction by leader–subordinate relationship quality, which we tested as a point of departure from the earlier literature (e.g., Cooper, 2002; Gkorezis et al., 2014; Pundt & Herrmann, 2015). This was somewhat surprising because certain parts of that mediated relationship have been supported in earlier research. However, this was the first test of the mediated model that used both leader and subordinate sources of data, and examined both positive and negative leader humor. As such, one possible explanation for our failure to replicate earlier findings is that the effects observed in earlier research might have been dependent on the fact that the data were obtained from a single source; that is, participants’ responses might have been influenced by the pervasive belief that humor is a positive characteristic (Martin, 2007), such that people who reported having a good relationship with their leader might have been inclined to report that their leader uses positive humor. This would be consistent with our argument that people might perceive their leader’s humor in the context of their relationship with the leader. Another possibility is that because the subordinates in our sample had typically been with their supervisor for a fairly long period of time, and their relationships with their leaders had likely stabilized (Nahrgang et al., 2009), the influence of day-to-day leader humor on subordinates’ perceptions of leader–subordinate relationship quality might not have been sufficient to exert an indirect influence on job satisfaction.

Practical Implications

The vast majority of the practitioner, popular press, and academic literature on humor in organizations advises leaders that they should use positively-toned humor, and strongly cautions against the use of negative humor (e.g., Mesmer-Magnus et al., 2012; Tarvin, 2012). Some of our results are consistent with this approach. In particular, our results suggested that affiliative leader humor might have positive effects on job satisfaction for subordinates with higher tenure. At the same time, other results seem to run counter to conventional wisdom, and suggest that we might not need to be so pessimistic about negative forms of humor. For example, analyses for Hypothesis 1 indicated that leader aggressive humor was unrelated to reports of leader–subordinate relationship quality, and that that relationship was not moderated by tenure. This finding suggests that cautions against the use of negative forms of humor might be overblown.

Indeed, our most dramatic results imply that leaders should adopt a rather different perspective than what is offered by conventional wisdom. If leaders want to integrate humor into their interactions with subordinates, they should first assess whether or not their subordinates are likely to interpret their humorous overtures positively by asking themselves whether the quality of their relationship with their subordinate will support their humor. Based on LMX theory, this would be characterized by an emotional connection involving trust, mutual respect, and a sense of obligation, perhaps as exemplified by employees’ motivation to go beyond their formal work role, and a willingness to provide the leader with assistance and/or honest and constructive criticism (Graen & Uhl-Bien, 1995).

Although accurately interpreting subordinates’ perceptions of the relationship might not be simple, if leaders are able to identify subordinates with whom they have a good relationship, the good news is that they might not have to be too concerned about the tone of their humor. Even when leaders themselves believe that the tone of their humor is aggressive (e.g., hurtful, offensive, and inappropriate; Martin et al., 2003), if their subordinates perceive a positive relationship with their leader, even aggressive humor might be positively related to job satisfaction. Although this finding implies that sometimes “negative humor” can have positive effects, it is important to note that a high-quality leader–subordinate relationship does not necessarily give leaders a blank check to use any type of humor in any context. For example, jokes that leverage racial or sexual stereotypes might cross lines of propriety with some subordinates. It is also likely that there are limits to the amount of negative/aggressive humor that subordinates will accept or be able to interpret in a positive manner.

Relatedly, our findings imply that humor use might be a poor strategy for repairing a poor leader–subordinate relationship. A leader’s use of humor to try to win subordinates over when they already do not like the leader might backfire, even if the leader tries to use humor he or she believes to be affiliative (e.g., making people laugh, saying witty things, telling funny stories; Martin et al., 2003). This idea is consistent with Fleming’s (2005) suggestion that humor “programs” designed to insert humor into the workplace to raise morale and positive affect will likely be met with cynicism. Leaders who wish to repair a damaged relationship with a subordinate might first find it necessary to repair the relationship through other strategies, such as providing positive reinforcement, praise, encouragement, and constructive feedback.

Limitations and Future Directions

A number of methodological features of this study strengthen our confidence in the results. First, the sample was obtained from working adults and their leaders from across a range of different organizations, supporting the generalizability of the findings. In addition, our hypotheses were tested using data obtained from both leaders and subordinates, which addresses concerns about single-source biases that have been an important shortcoming in much of the previous research on humor and leadership. Our analyses also included the two focal interpersonal humor variables (i.e., affiliative and aggressive leader humor styles), as well as Martin et al.’s (2003) other two intrapersonal humor style variables (i.e., self-enhancing and self-defeating) simultaneously in a single model, and controlled for subordinate positive affectivity (Robert & Wilbanks, 2012). This provided a particularly strong test of the effects of leader humor style as a moderator (i.e., affiliative or aggressive). Essentially, our analyses examined the impact of each humor style while controlling for the others, and controlled for the possible overarching influence of positive affect on all relationships in the model.

Those strengths notwithstanding, we also note a number of limitations. First, though the use of leader reports of their own humor styles enabled us to examine our models using multi-source data, the leader–subordinate relationship quality and job satisfaction variables were still obtained from subordinate reports, making the models prone to problems associated with self-report data (Podsakoff & Organ, 1986). In addition, there is a possible disconnect between the tone that leaders think they are using (e.g., affiliative vs. aggressive) and the tone that is perceived by subordinates. Thus, using leader reports of humor use as a second source of data involved a trade-off between the avoidance of purely single-source designs, and the possible lack of correspondence between leader and subordinate perceptions. Indeed, this possible disconnect might have contributed to the fact that we did not replicate the results of previous studies, which found direct or mediated relationships between leader humor, leader–subordinate relationships, and subordinate outcomes. Certainly the use of additional sources of data in future research, including additional dependent variables, would increase confidence in the stability and replicability of our findings. For example, the use of observational measures of leader humor could be an alternative to subordinate reports or leader reports of humor use.

Similarly, leader reports of leader–subordinate relationship quality might be used instead of or in addition to reports by the subordinate. Relatedly, we asked the leaders in our sample to indicate their humor styles in general, rather than asking them to report on their use of different types of humor with each subordinate. Given that the focus of LMX theory is on the unique relationships that develop between each leader and subordinate, such measures might uncover patterns of humor use that differ based on whether the subordinate is considered out-group or in-group. However, it remains to be seen whether participants’ memories are sufficiently accurate to make such fine-grained distinctions about their humor use in hindsight.

We also note that our data are cross-sectional, which limits our ability to make causal inferences. Although we draw on humor theory to support our contention that humor use precedes leader–subordinate relationship quality (e.g., Cooper, 2008; Wyer & Collins, 1992), we cannot rule out alternative causal orderings, or more complex causal relationships. In particular, it is likely that there is a dynamic component to the relationship between leader humor and relationship quality (cf. Robert & Wilbanks, 2012). Although leader humor probably influences leader–subordinate relationship quality, over time, relationship quality probably has a reciprocal effect on the leader’s use of humor. Specifically, the feedback that leaders receive from subordinates about the quality of their relationship in general, and responses to humor in particular (e.g., smiles and laughter), will likely lead leaders to tweak and adjust their humor.

The dynamic nature of this relationship is not inconsistent with our substantive conclusion that the leader–subordinate relationship provides an important contextual condition for subordinates to interpret leader humor. Indeed, conceptualizing the relationship between leader humor and leader–subordinate relationship quality as dynamic is helpful for clarifying how leader–subordinate relationship quality can be both an outcome of leader humor, and a moderator of the relationship between leader humor and job satisfaction. Specifically, our moderation hypothesis is built on the notion that at any given point in time, a subordinate will interpret his or her leader’s humor based on his or her perception of the existing relationship: If the relationship is good, the effects of the humor are positive, and vice versa. However, the repeated use of certain types of humor might also have a cumulative effect on subordinates’ perception of the relationship. In this regard, it is reasonable to view leader humor as having both chronic effects on relationship quality, and acute effects on subordinate outcomes that are driven by subordinates’ current perceptions of relationship quality.

Conceptualizing the humor–relationship quality relationship as dynamic also suggests additional avenues for future research. For example, a dynamic relationship between humor use and relationship quality suggests that there might be a qualitative difference between the types of humor that are used early in a relationship, and those that are used in established relationships. For example, it is possible that leaders and subordinates who wish to develop positive relationships experiment with the use of humor, perhaps starting with small and subtle forms of “safe” humor that can be useful in gauging the target’s reaction (e.g., self-deprecation, mild teasing). Over time, though, producers might use feedback from targets to take more calculated risks with humor use, choosing edgier humor that has a higher likelihood of producing a positive reaction, but also a higher likelihood of causing offense (e.g., more aggressive put-down humor or sarcasm). Given a dynamic process in which leaders and subordinates adjust and calibrate their behaviors and responses to one another, over time we might also expect to see a closer correspondence between producers’ and targets’ perceptions of each party’s humor and its impact in the dyadic relationship. Such synchronicity might be especially likely when producers and targets are high in emotional intelligence or self-monitoring, and thus more likely to be tuned in to the reactions of others (Robert & Wilbanks, 2012). An examination of these types of dynamic process predictions would require longitudinal or experimental designs that would allow for a clearer temporal ordering of the constructs (i.e., established relationship, leader humor, subordinate response), but are exciting avenues for future research.

Another potential limitation of this study is that the data were gathered in a single culture: Hong Kong. This potentially limits the generalizability of our results, and suggests that it might be valuable to replicate this study in other cultures. However, some evidence from the literature provides confidence that our results might be robust. First, Martin et al.’s (2003) HSQ measure has recently been translated and validated in a number of Chinese samples, suggesting that this key measure is comparable across cultures. With regard to LMX, a recent meta-analysis by Rockstuhl, Dulebohn, Ang, and Shore (2012) found that leader–subordinate relationship quality tends to be more important in horizontal individualist (i.e., Western) societies than in vertical collectivist (e.g., Asian) societies. This suggests that the results we obtained in the current study, which was conducted in a relatively vertical collectivist culture, might be magnified, rather than diminished, if replicated in a more horizontal individualist culture.

Conclusion

Humor is a ubiquitous human behavior that infiltrates most social contexts, including organizations. Nevertheless, the topic of humor in organizations has garnered surprisingly little attention from researchers, particularly with respect to theoretical explanations for humor’s effects. Indeed, much of the existing literature on humor in organizations is fairly speculative, correlational, and rooted in popular beliefs about humor’s role in human interaction. This might be due to the fact that because humor is such a familiar aspect of interpersonal interaction in everyday life, people already believe it is well-understood (Martin, 2007). In this research, we attempted to move the topic of humor in organizations forward by proposing a conceptual model describing the relationship between leader humor, leader–subordinate relationship quality, and subordinate job satisfaction. Contrary to pervasive conventional wisdom, our results suggested that although leaders might indeed be able to leverage humor at work, the effectiveness of those efforts might depend more on the quality of the relationship that already exists with their subordinate than on the tone of their humor.

Footnotes

Acknowledgements

We would like to thank Art Jago for his helpful comments on an earlier draft, and Kris Preacher for analytical advice and direction.

Associate Editor: Lisa Finkelstein

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Christopher Robert (PhD, University of Illinois Urbana–Champaign) is an associate professor in the Department of Management in the Robert J. Trulaske College of Business at the University of Missouri. His research interests include cross-cultural management, humor in the workplace, negotiation and conflict, and groups and teams.

Timothy C. Dunne (PhD, University of Missouri) is an assistant professor of Management at the Jones College of Business, Middle Tennessee State University. His current research focuses on leadership processes, conflict management, and team dynamics.

Joyce Iun (DBA, The Hong Kong Polytechnic University) is a professional consultant at the Chinese University of Hong Kong. Her research interests include humor, leader–member exchange, and employee well-being. She works at various career counseling and e-learning projects.

References

Aguinis

Gottfredson

R. K.

Culpepper

S. A.

(2013). Best-practice recommendations for estimating cross-level interaction effects using multilevel modeling. Journal of Management, 66, 805-824.

Avolio

B. J.

Howell

J. M.

Sosik

J. J.

(1999). A funny thing happened on the way to the bottom line: Humor as a moderator of leadership style effects. Academy of Management Journal, 42, 219-227.

Baron

R. M.

Kenny

D. A.

(1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Brislin

R. W.

Lonner

W. J.

Thorndike

R. M.

(1973). Cross-cultural research methods. New York, NY: John Wiley.

Cammann

Fichman

Jenkin

Klesh

(1979). The Michigan Organizational Assessment Questionnaire. Unpublished manuscript, University of Michigan.

Chen

G.-H.

Martin

R. A.

(2007). A comparison of humor styles, coping humor, and mental health between Chinese and Canadian university students. Humor: International Journal of Humor Research, 20, 215-234.

Clouse

R. W.

Spurgeon

K. L.

(1995). Corporate analysis of humor. Psychology: A Journal of Human Behavior, 32(3-4), 1-24.

Cooper

C. D.

(2002). No laughing matter: The impact of supervisor humor on leader-member exchange (LMX) quality (Unpublished doctoral dissertation). University of Southern California, Los Angeles.

Cooper

C. D.

(2005). Just joking around? Employee humor expression as an ingratiatory behavior. Academy of Management Review, 30, 765-776.

10.

Cooper

C. D.

(2008). Elucidating the bonds of workplace humor: A relational process model. Human Relations, 61, 1087-1115.

11.

Croon

M. A.

van Veldhoven

M. J. P. M.

(2007). Predicting group-level outcome variables from variables measured at the individual level: A latent variable multilevel model. Psychological Methods, 12, 45-57.

12.

Davis

Kleiner

B. H.

(1989). The value of humour in effective leadership. Leadership & Organization Development Journal, 10, 1-3.

13.

Davis

M. S.

(1993). What’s so funny? The comic conception of culture and society. Chicago, IL: University of Chicago Press.

14.

Decker

W. H.

(1987). Managerial humor and subordinate satisfaction. Social Behavior and Personality, 15, 225-232.

15.

Decker

W. H.

Rotondo

D. M.

(2001). Relationships among gender, type of humor, and perceived leader effectiveness. Journal of Managerial Issues, 13, 450-465.

16.

Decker

W. H.

Yao

Calo

T. J.

(2011). Humor, gender, and perceived leader effectiveness in China. SAM Advanced Management Journal, 76, 43-53.

17.

Dulebohn

J. H.

Bommer

W. H.

Liden

R. C.

Brouer

R. L.

Ferris

G. R.

(2012). A meta-analysis of antecedents and consequences of leader-member exchange: Integrating the past with an eye toward the future. Journal of Management, 38, 1715-1759.

18.

Eagly

A. H.

Johannesen-Schmidt

M. C.

(2001). The leadership styles of women and men. Journal of Social Issues, 57, 781-797.

19.

Eberly

M. B.

Fong

C. T.

(2013). Leading via the heart and mind: The roles of leader and follower emotions, attributions and interdependence. Leadership Quarterly, 24, 696-711.

20.

Fleming

(2005). Workers’ playtime? Boundaries and cynicism in a “culture of fun” program. Journal of Applied Behavioral Science, 41, 285-303.

21.

Gkorezis

Petridou

Xanthiakos

(2014). Leader positive humor and organizational cynicism: LMX as a mediator. Leadership & Organization Development Journal, 35, 305-315.

22.

Goudreau

(2012, February 21). Are funny people more successful in business. Forbes. Retrieved from http://www.forbes.com/sites/jennagoudreau/2012/02/21/funny-people-more-successful-in-business-humor-workplace/2/

23.

Graen

G. B.

Novak

Sommerkamp

(1982). The effects of leader-member exchange and job design on productivity and satisfaction: Testing a dual attachment model. Organizational Behavior and Human Performance, 30, 109-131.

24.

Graen

G. B.

Uhl-Bien

(1995). Relationship-based approach to leadership: Development of leader-member exchange (LMX) theory of leadership over 25 years: Applying a multi-level multi-domain perspective. Leadership Quarterly, 6, 219-247.

25.

Hofmann

D. A.

Gavin

M. B.

(1998). Centering decisions in hierarchical linear models: Implications for research in organizations. Journal of Management, 24, 623-641.

26.

Kazarian

Martin

(2004). Humor styles, personality and well-being among Lebanese university students. European Journal of Personality, 18, 209-219.

27.

LaHuis

D. M.

Ferguson

M. W.

(2009). The accuracy of significance tests for slope variance components in multilevel random coefficient models. Organizational Research Methods, 12, 418-435.

28.

Liden

R. C.

Sparrowe

R. T.

Wayne

S. J.

(1997). Leader-member exchange theory: The past and potential for the future. In Ferris

G. R.

(Ed.), Research in personnel and human resources management (pp. 47-119). New York, NY: Elsevier.

29.

Lynch

(2002). Humorous communications: Finding a place for humor in communication research. Communication Theory, 12, 423-445.

30.

Malone

P. B.

(1980). Humor: A double-edged tool for today’s managers? Academy of Management Review, 5, 357-360.

31.

Martin

R. A.

(2007). The psychology of humor: An integrative approach. Amsterdam, The Netherlands: Academic Press.

32.

Martin

R. A.

Puhlik-Doris

Larsen

Gray

Weir

(2003). Individual differences in uses of humor and their relation to psychological well-being: Development of the Humor Style Questionnaire. Journal of Research in Personality, 37, 48-75.

33.

Mathieu

J. E.

Aguinis

Culpepper

S. A.

Chen

(2012). Understanding and estimating the power to detect cross-level interaction effects in multilevel modeling. Journal of Applied Psychology, 97, 951-966.

34.

Mesmer-Magnus

Glew

D. J.

Viswesvaran

(2012). A meta-analysis of positive humor in the workplace. Journal of Managerial Psychology, 27, 155-190.

35.

Muthén

L. K.

Muthén

B. O.

(1998-2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Author.

36.

Nahrgang

J. D.

Morgeson

F. P.

Ilies

(2009). The development of leader-member exchanges: Exploring how personality and performance influence leader and member relationships over time. Organizational Behavior and Human Decision Processes, 108, 256-266.

37.

Norrick

N. R.

(2003). Issues in conversational joking. Journal of Pragmatics, 35, 1333-1359.

38.

Podsakoff

P. M.

Organ

D. W.

(1986). Self-reports in organizational research: Problems and prospects. Journal of Management, 12, 531-544.

39.

Preacher

K. J.

Curran

P. J.

Bauer

D. J.

(2006). Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31, 437-448.

40.

Preacher

K. J.

Zhang

Zyphur

M. J.

(2011). Alternative methods for assessing mediation in multilevel data: The advantage of multilevel SEM. Structural Equation Modeling: A Multidisciplinary Journal, 18, 161-182.

41.

Preacher

K. J.

Zyphur

M. J.

Zhang

(2010). A general multilevel SEM framework for assessing multilevel mediation. Psychological Methods, 15, 209-233.

42.

Priest

R. F.

Swain

J. E.

(2002). Humor and its implications for leadership effectiveness. Humor, 15, 169-189.

43.

Pundt

Herrmann

(2015). Affiliative and aggressive humour in leadership and their relationship to leader-member exchange. Journal of Occupational and Organizational Psychology, 88, 108-125.

44.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods. London, England: SAGE.

45.

Robert

Wilbanks

J. E.

(2012). The Wheel Model of humor: Humor events and affect in organizations. Human Relations, 65, 1071-1099.

46.

Robert

Yan

(2007). The case for developing new research on humor and culture in organizations: Towards a higher grade of manure. In Martocchio

J. J.

(Ed.), Research in personnel and human resource management (Vol. 26, pp. 205-267). Amsterdam, The Netherlands: Elsevier.

47.

Rockstuhl

Dulebohn

J. H.

Ang

Shore

L. M.

(2012). Leader-member exchange (LMX) and culture: A meta-analysis of correlates of LMX across 23 countries. Journal of Applied Psychology, 97, 1097-1130.

48.

Romero

E. J.

Cruthirds

K. W.

(2006). The use of humor in the workplace. Academy of Management Perspectives, 20(2), 58-69.

49.

Sala

(2000). Relationship between executives’ spontaneous use of humor and effective leadership. Dissertation Abstracts International, Section B: The Sciences and Engineering, 61(3-B), 1683.

50.

Scogin

F. R.

Pollio

H. R.

(1980). Targeting and the humorous episode in group process. Human Relations, 33, 831-852.

51.

Snijders

T. A. B.

Bosker

R. J.

(2012). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London, England: SAGE.

52.

Tarvin

(2012). Humor that works: 501 ways to use humor. New York, NY: Humor That Works.

53.

Terrion

J. L.

Ashforth

B. E.

(2002). From I to “we”: The role of putdown humor and identity in the development of a temporary group. Human Relations, 55, 55-88.

54.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063-1070.

55.

Wyer

R. S.

Collins

J. E.

(1992). A theory of humor elicitation. Psychological Review, 99, 663-688.