Holding Others in Contempt: the Moderating Role of Power in the Relationship Between Leaders’ Contempt and their Behavior Vis-à-vis Employees

Abstract:

The purpose of the present research was to investigate if and when leaders’ trait-like tendency to experience contempt would result in a lack of constructive attitudes and behaviors towards subordinates and an increase in destructive attitudes and behaviors towards subordinates. Previous research shows that increased power aligns individuals’ behavior with their trait-like tendencies. Accordingly, we hypothesized that leader contempt and power will interact to predict leaders’ people orientation, ethical leadership, dehumanization, and self-serving behavior. Across three studies, we indeed found that contempt was more negatively associated with leaders’ people orientation and ethical leadership, and more positively associated with dehumanization and leaders’ self-serving behavior, when the leader had higher levels of power rather than lower levels of power. These results are discussed in the context of corporate ethical scandals demonstrating leaders’ focus on personal gain to the detriment of the needs of their subordinates.

Recently, the producer of Twinkies, Hostess Brands, went bankrupt. The company’s top executives pointed to overly demanding and overpaid subordinates as the primary reason for not being able to stay afloat. Yet, Hostess Brands’ employees and the media paint a different story. A story in which corporate management felt contempt for its labor force and was serving its own needs over all others, showing little concern and regard for the workers who actually produced the products. Indeed, those in the powerful positions rewarded themselves with a 300% salary raise—lifting the salary of former company leader Brian Driscoll from $750,000 to $2,555,000—while simultaneously demanding that their labor force cut back on their earnings and benefits (cf. Cancella, 2012). This example suggests that powerful leaders’ contempt may result in both a shortfall of their constructive attitudes and behaviors towards subordinates and a surplus of their destructive attitudes and behaviors towards them. Clearly, if leader contempt can have such devastating effects, we should have a thorough understanding of it, because this may be the first step in attempts to mitigate negative consequences for subordinates and ultimately the organization at large (cf. Kalshoven, Den Hartog, & De Hoogh, 2011; Mayer, Kuenzi, Greenbaum, Bardes, & Salvador, 2009).

The present study is geared at testing the idea that leaders’ failure to care about subordinates (i.e., a lack of people orientation) and to provide ethical leadership, as well as their tendency to dehumanize subordinates and to act self-servingly at employees’ expense, is grounded in a trait-like tendency to experience feelings of contempt. However, we argue that whether leaders are likely to act upon their feelings of contempt may be a function of the amount of power they have within their position. Power has been found to increase the correspondence between traits and behavior (DeCelles, DeRue, Margolis, & Ceranic, 2012). Hence, we expect that the negative relationship between (a) leader contempt and people orientation and (b) leader contempt and ethical leadership, as well as the positive relationship between (c) leader contempt and dehumanization and (d) leader contempt and self-serving behavior, will be stronger when leaders have more power. The role that power plays in the effects of leader contempt on leaders’ attitudes and behaviors towards followers is all the more important given that higher power entails more control over valuable resources and thus a greater opportunity to affect employees’ outcomes. By drawing upon and integrating knowledge from two separate streams of research, contempt and power, this research aims to demonstrate that the combination of leaders’ contempt and power can have pernicious effects where leaders’ treatment of their employees is concerned.

LEADER EMOTIONS: A FOCUS ON CONTEMPT

Generally speaking, the leader’s task is to influence those who are in hierarchically subordinate positions to achieve a common goal (Yukl, 2010). This makes the social interaction between leader and subordinate a key aspect in the leadership process. In recent years, research has pointed to the role that leaders’ affective states may play in shaping their behavior towards subordinates (e.g., Côté & Hideg, 2011; Côté, Lopes, Salovey, & Miners, 2010; Barsade, Brief, & Spataro, 2003; Van Kleef, 2009). For instance, there is evidence that leaders’ display of positive, high energy emotions may lead to behavior that is seen as more charismatic (Bono & Ilies, 2006; Damen, Van Knippenberg, & Van Knippenberg, 2008b). Moreover, leaders’ expressions of emotions are thought to arouse emotions in subordinates through the mechanism of emotional contagion, where the displayed emotions of the leader are first mimicked and then felt by followers (Hatfield, Cacioppo, & Rapson, 1994). Moreover, research showed that those high in trait positive affectivity (those who are more likely to experience positive emotions and moods) are seen as displaying more transformational leadership behaviors and as being better leaders (Rubin, Munz, & Bommer, 2005; Staw & Barsade, 1993). Accordingly, emotions – either measured as immediate state experiences or as relatively stable trait-like inclinations to experience certain emotions (cf. Spielberger & Reheiser, 2009) – can direct and shape leader behavior. However, as state emotions are in the moment, short-term emotions, they are particularly predictive of immediate reactions in specific situations. In contrast, habitual or trait emotions, which are more stable tendencies to feel and act in certain ways (cf. Revelle & Scherer, 2009; Watson & Clark, 1984) are likely more strongly related to leadership styles and behaviors over time.

In the present research, we put forward the understudied other-condemning emotion of trait contempt as an emotion that may influence the way in which leaders behave vis-à-vis their subordinates. Contempt is characterized by a feeling of superiority over or disdain for others (Miller, 1997) and it includes feelings of condescension and disapproval of others (Izard, 1977). Notably, contempt is an emotion that tends to be prevalent in hierarchical relationships, including those that can be found in the workplace (Melwani & Barsade, 2011). Research on contempt is scarce (Pelzer, 2005). Yet, as any other emotion, we argue that some people are more likely to experience it than others, whether a result of genetic factors or a result of learning and socialization (Revelle & Scherer, 2009).

In line with a feeling-is-for-doing approach (Zeelenberg, Nelissen, Breugelmans, & Pieters, 2008), the intraphysic experience of contempt motivates the behavior of the person who experiences it. First, it may directly affect facial expressions – contempt is often displayed by a one-sided smirk (Matsumoto & Ekman, 2004). Second, it may instigate actions that are aimed at increasing social distance between the person who experiences contempt and the target of those feelings (Melwani, Mueller, & Overbeck, 2012). These could be actions that are aimed at reducing the status of the target individual in the social hierarchy (Keltner & Haidt, 1999; Morris & Keltner, 2000), or at socially excluding that person (Fischer & Roseman, 2007). Indeed, contempt has been found to result in ignoring the recipient of one’s contempt, treating the other person as inferior, or as someone who is unworthy of receiving attention (Ekman, 1994; Izard, 1977; Hutcherson & Gross, 2011).

Importantly, emotions are more than (readouts of) intrapersonal experiences; they serve a social function (Ekman, 1992; Keltner & Haidt, 1999; Keltner, Haidt, & Shiota, 2006). In fact, some have argued that the “primary function of emotions is to engender social influence” (Van Kleef, Van Doorn, Heerdink, & Koning, 2011: 115), and as such should be seen as acts of communication (Morris & Keltner, 2000). In other words, emotions’ social function is best served if the interaction partner is able to pick up on the emotion so that adaptive behavior can potentially follow. Likewise, others’ contempt is often perceived and felt by the target, which increases its impact on the target (Melwani et al., 2012).

CONTEMPT AND LEADER BEHAVIOR VIS-À-VIS SUBORDINATES

Given the behavior that is instigated by contempt, we argue that the extent to which a leader is contemptuous may result in a shortfall of constructive attitudes and behaviors towards subordinates. Specifically, we expect that those leaders with a higher tendency to experience contempt can be less likely to adopt a people orientation (i.e., genuinely caring about, respecting, and supporting subordinates and ensuring that their needs are fulfilled; cf. Kalshoven et al., 2011; Kanungo & Conger, 1993; Treviño, Brown, & Hartman, 2003). Leaders’ attitudes and orientation towards their subordinates can be considered to be of major importance because respect, caring, and support often go hand in hand with ethical conduct. For instance, ethical leaders are also described as “people-oriented, aware of how their actions impact others” (Resick, Hanges, Dickson, & Mitchelson, 2006: 347).

Broadening this perspective, we also contend that leaders with a higher tendency to experience contempt can be less likely to adopt an ethical leadership style. Ethical leadership not only captures being trustworthy and respecting subordinates, it also includes leaders’ communication of ethical codes of conduct and rewarding subordinates when they act in line with these codes. Specifically, ethical leadership has been defined as the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers through two-way communication, reinforcement, and decision-making (Brown, Treviño, & Harrison, 2005: 120). Investigating the effects of contempt on leaders’ people orientation and ethical leadership is important, because both have been positively associated with subordinates’ job satisfaction, organizational commitment, and organizational citizenship behavior (e.g., Kalshoven et al., 2011; Le Pine, Erez, & Johnson, 2002; Mayer et al., 2009).

Ethical leadership and a people orientation typically include behaviors such as taking time for personal contact, listening to what employees have to say, and making fair and balanced decisions (e.g., Brown et al., 2005; Kalshoven et al., 2011). Contempt appears to be at odds with these types of leader behaviors, as it is negatively associated with empathy, modesty, fairness, sincerity, and moral identity (Sanders, Wisse, Van Yperen, & Rus, 2015), and motivates behaviors targeted at socially excluding others (e.g., not talking to or avoiding all communication with the target of one’s contempt; cf. Fischer & Roseman, 2007). In addition, trait contempt has been positively associated with the probability of displaying abusive supervisory behaviors (Sanders et al., 2015). Moreover, recent research by Melwani and Barsade (2011) also hints at the possibility that contempt can motivate behavior that is characterized by lower concern for others and for behaving appropriately. As part of their research, they developed a business strategy simulation in which they tested the psychological and interpersonal consequences of receiving contemptuous feedback. The key findings showed that those who received contemptuous feedback more strongly responded with returned feelings of contempt than those who did not receive contemptuous feedback. Further, these feelings of contempt led to higher interpersonal aggression, often seen as inappropriate and reflecting a lack of concern for others.

However, we contend that not all leaders are equally likely to act upon their feelings of contempt. The degree to which leaders’ tendency to feel contemptuous will be reflected in their attitudes and behavior vis-à-vis subordinates may depend on the amount of power they have.

POWER

Social power is defined as the relative capacity to modify the behaviors and outcomes of others by providing or withholding resources (Fiske, 1993; Keltner, Gruenfeld, & Anderson, 2003; Magee & Galinsky, 2008). The leader role is characterized by having the authority to make decisions, to allocate resources, and to control punishments and rewards. As such, the leader role comes with a certain degree of position power, which is an important source of daily influence in organizations. Yet, while leaders may typically have more power than subordinates, not all leaders will have the exact same amount of power at their disposal within their leadership position (cf. Yukl & Falbe, 1992; Rus, Van Knippenberg, & Wisse, 2010). For instance, some leaders have the authority to reward, punish, or fire their subordinates, whereas other leaders may not have the power to do these things.

Previous research has shown that power has implications for how people process information, how they feel, and how they behave (Anderson & Berdahl, 2002). In contrast to the well-known adage that ‘Power corrupts and absolute power corrupts absolutely’ (Lord Acton, 1887), a growing body of research shows that power does not necessarily corrupt people but rather makes individuals more sensitive to their own subjective experiences, feelings, and preexisting tendencies (Weick & Guinote, 2008). Indeed, it has been argued and shown that power reduces the dependence on others, group norms, and rules, and allows individuals to act in accordance with their internal traits, states, and predispositions (Bargh, Raymond, Pryor, & Strack, 1995; Chen, Lee-Chai, & Bargh, 2001; Côté et al., 2011; DeCelles et al., 2012; Galinsky, Magee, Gruenfeld, Whitson, & Liljenquist, 2008; Maner & Mead, 2010; Piteša & Thau, 2013).

This principle also seems to hold for leaders with varying amounts of power. For instance, Rus et al. (2010) showed that for high-power leaders, but not for low-power leaders, internal beliefs were stronger predictors of behavior than external cues. They indicated that beliefs about effective leadership influenced high-power leaders’ self-allocations more than low-power leaders’ self-allocations. Moreover, Wisse and Rus (2012) showed that power accentuated the impact of leader self-construal on leader self-interested behavior, such that those with a salient personal self acted more self-servingly with higher levels of power. Recently, Niemann, Wisse, Rus, and Van Yperen (in press) found that supervisors and instructors seek less feedback when they experience high (vs. low) interpersonal uncertainty, particularly when they have high power. Based on these findings, we argue that leaders with preexisting tendencies that dampen concern for others are more likely to show a lack of people orientation and fail to display ethical leadership to the extent that they have more power. Specifically, we hypothesized that:

Hypothesis 1a. Leader contempt and power will interact to predict leaders’ people orientation, such that contempt will be more negatively associated with leaders’ people orientation when the leader has higher levels of power than when the leader has lower levels op power.

Hypothesis 1b. Leader contempt and power will interact to predict ethical leadership, such that contempt will be more negatively associated with ethical leadership when the leader has higher levels of power than when the leader has lower levels of power.

OVERVIEW OF THE PRESENT RESEARCH

The first two studies are geared at investigating if the interactive effect of contempt and power is associated with a lack of leaders’ constructive attitudes and behaviors towards subordinates. Study 1 provides a first test of our prediction that contempt is negatively associated with leaders’ people orientation when leaders have higher levels of power (Hypothesis 1a). For this study, we relied on multi-source data and assessed leaders’ own perceptions of their tendency to experience contempt and their degree of power, whereas we relied on subordinates’ ratings with regard to their leaders’ people orientation. In Study 2, we aimed to replicate and extend these findings in a sample of subordinates by testing both Hypothesis 1a and 1b. Specifically, in Study 2, we relied on subordinates’ perceptions of their leaders’ contempt, power, people orientation, and ethical leadership. In Study 3, we further extend our findings by focusing on leaders’ destructive attitudes and behaviors towards subordinates. Specifically, we will investigate if contempt is positively associated with leaders’ dehumanization and self-serving behavior at the expense of subordinates when leaders have higher levels of power, which will be discussed in the introduction of Study 3.

STUDY 1

Method

Respondents and procedure. We collected multi-source data from 49 Dutch leaders (82% response rate, 40.8% women, M = 41.94, SD = 11.50) and 165 of their subordinates (62.4% women, M = 32.67, SD = 9.07). On average, we collected data from 3.37 subordinates per leader. Leader’s average work experience as a leader was 7.31 years (SD = 6.99), they worked on average 38.75 hours a week (SD = 9.06), and supervised an average of 20.33 workers (SD = 24.75). Most of the leaders worked in health care (28.6%) or retail (24.5%) organizations, and 69.2 % held an Applied Sciences or University degree. Subordinates’ average organizational tenure was 4.73 years (SD = 6.50), and on average they worked 29.43 hours a week (SD =10.85). A total of 47.4% of the subordinates held an Applied Sciences or University degree.

Leaders were approached via e-mail and through phone calls to seek their participation. After making an appointment, we visited the company with paper-and-pencil questionnaires and asked those in a leadership position, and two to five of their subordinates (depending on team size and availability), to participate. We kept the questionnaire short because people often filled out the questionnaires at work. Moreover, we stressed that participation was voluntary and that responses would be treated confidentially. We used a coding system in order to match subordinates’ answers to those of their leader. Sealed envelopes, each including one filled-out questionnaire, were collected one week later. Only completely filled-out questionnaires were included in the analyses.

Measures. Means, standard deviations, correlations, and reliabilities are displayed in Table 1. All items are displayed in the Appendix.

Table 1: Means, Standard Deviations, Reliabilities, and Intercorrelations for Study 1, 2, and 3

Note. Cronbach’s alphas are displayed on the diagonal.

* p < .05

** p < .01

*** p < .001

Contempt. Leaders’ self-reported contempt was measured with five items using a seven-point Likert scale (1 = does not apply to me at all, 7 = does apply to me). All items were averaged into a single contempt score.

Power. Leaders’ self-reported position power was measured with nine items of the Yukl and Falbe (1991) Position Power Scale (the subscales coercive, reward, and legitimate power; cf. Rus et al., 2010) using a seven-point Likert scale (1 = strongly disagree, 7 = strongly agree).

People orientation. The degree to which leaders showed true concern for their subordinates was measured with the seven-item people orientation scale (Kalshoven et al., 2011) using a five-point Likert scale (1 = never, 5 = always).

Results

Preliminary analyses. We assessed the appropriateness of aggregating subordinates’ perceptions of their leaders’ people orientation by calculating the R _wg, ICC(1), and ICC(2) values (Bliese, 2000). The R _wg statistic (measuring the agreement between subordinates belonging to the same leader) was .82, which is higher than the generally recommended cutoff value of .70 (James, 1988). ICC(1) (assessing the interrater reliability) was .44, thus exceeding the .12 cutoff criterium (Glick, 1985). The ICC(2) (providing an estimate of the reliability of the group means) was .72, and thus surpassed the recommended cutoff value of .70 (Bliese, 2002). On the basis of these results, we believe that aggregation to the group level is justified.

Multilevel confirmatory factor analyses using Mplus were conducted to test whether our study variables are not only theoretically but also empirically distinct. A model with two factors on the between-person level (i.e., contempt and power) and one factor on the within-person level (i.e., people orientation) had an acceptable fit, χ²(90) = 224.54, p < .001; CFI = .88, RMSEA = .095, SRMR_within = .05, SRMR_between = .12, and fitted the data better than a model including a single factor on both the between-person and within-person level, χ²(91) = 278.86, p < .001; CFI = .83, RMSEA = .11, SRMR_within = .05, SRMR_between = .17; Δχ²(1) = 54.32, p < .001 (cf. Kline, 2010; Marsh, Hau, & Wen, 2004). These results testify to the empirical distinguishability of our measures.

Hypothesis testing.Hypothesis 1a was that leader contempt and power will interact to predict leaders’ people orientation, such that contempt will be more negatively associated with leaders’ people orientation when the leader has higher levels of power than when the leader has lower levels of power. To test this hypothesis, we first centered the scores of our independent measures (Aiken & West, 1991). In Step 1 of the hierarchical regression analysis, we included the main effect terms for our independent variables (leaders’ contempt and power), and in Step 2 we added the two-way interaction term. All regression results are displayed in Table 2^1,2 . Step 1, did not show a significant main effect for leaders’ feelings of contempt (p = .25) or leaders’ power (p = .37) on their people orientation. In line with our hypothesis, Step 2 explained a significant proportion of variance and it revealed our predicted contempt × power interaction (p = .04; see Table 2). Simple slopes analyses showed that with higher levels of power (1 SD above the mean), leaders’ contempt was negatively associated with their people orientation, b = -0.63, SE _b = 0.27, t(45) = 2.39, p = .02. With lower levels of power (1 SD below the mean), leaders’ contempt did not show a relationship with their people orientation, b = 0.19, SE _b = 0.27, t(45) = 0.70, p = .49 (see Figure 1). These findings provide support for Hypothesis 1a.

Table 2: Summary of Regression Analysis for Contempt and Power Predicting People Orientation in Study 1

Note. N = 49. R ² represents the adjusted R ².

* p < .05.

Figure 1: Leaders’ People Orientation as a Function of Contempt and Power in Study 1.

STUDY 2

Method

Respondents and procedure. A total of 194 workers from the United States (35.10% women, M _age = 31.32, SD = 8.84) participated in our online field study. Respondents’ average organizational tenure was 3.30 years (SD = 2.92). Most of the respondents (80.92%) worked in technology, business and finance, human services, industry, or education, and 85.5% held an Applied Sciences or University degree. We recruited workers using Amazon’s Mechanical Turk Website and paid them $0.80 US for their participation. Internet recruitment methods have become more popular among researchers now that more and more people are online and their use has been approved by the American Psychological Association’s Board of Scientific Affairs’ Advisory Group (Kraut et al., 2004). Notably, research has shown that data obtained with Mechanical Turk is at least as reliable as data obtained via traditional methods (Buhrmester, Kwang, & Gosling, 2011; Mason & Suri, 2012; Paolacci, Chandler, & Ipeirotis, 2010).

Measures. Means, standard deviations, correlations, and reliabilities are displayed in Table 1.

Contempt. To assess subordinates’ perceptions of their leaders’ contempt we used the same five items as in Study 1. The items were adapted to a sample of subordinates (i.e., we used “My supervisor…” instead of “I…”). Responses were assessed using a seven-point Likert scale (1 = does not apply to my supervisor, 7 = does apply to my supervisor).

Power. Leaders’ position power as perceived by subordinates was measured with the same items as in Study 1. The items were adapted to a sample of subordinates (i.e., we used “My supervisor…” instead of “I…”). Responses were assessed using a five-point Likert scale (1 = strongly disagree, 5 = strongly agree).

People orientation. The degree to which leaders showed true concern for their subordinates was measured with the same items as in Study 1 using a five-point Likert scale (1 = strongly disagree, 5 = strongly agree).

Ethical leadership. The extent to which supervisors demonstrated ethical leadership was assessed with the 10-item Ethical Leadership Scale (Brown et al., 2005) using a five-point Likert scale (1 = strongly disagree, 5 = strongly agree).

Results

Preliminary analyses. Prior to conducting a hierarchical regression analysis, we performed a confirmatory factor analysis (CFA) on our predictor variable items (i.e., contempt and power) as well as our dependent variable items (i.e., people orientation and ethical leadership). We defined and compared six different factor structures, ranging from a one-factor model in which all items were indicative of one larger factor, to a four-factor model in which each of the study variables was indicative of its own factor. The four-factor model had a better fit to the data, χ²(428) = 831.75, p < .001; CFI = .89, RMSEA = .07, SRMR = .08, than all alternative models with fewer factors. When taking into account the high correlation between people orientation and ethical leadership, it was not surprising that the best fitting model of these alternatives was a three-factor model in which contempt and power loaded on their own factor and people orientation and ethical leadership together loaded on a single factor. Yet, this three-factor model, χ²(431) = 926.03 p < .001; CFI = .87, RMSEA = .08, SRMR = .09, still fitted the data less well than the four-factor model in which all study variables loaded on their own factor, ∆χ²(3) = 94.28, p < .001. This indicates that the variables are not only theoretically but also empirically distinct from each other, which justifies the decision to conduct the analyses on people orientation and ethical leadership separately³.

Because we relied on single-source data, we assessed the degree to which the correlations among our study variables are an artifact of common method variance using Harman’s single-factor test (Podsakoff & Organ, 1986). An unrotated exploratory factor analysis revealed multiple factors, thereby reducing concerns regarding common source bias.

Hypothesis testing

People orientation. To test Hypothesis 1a, we conducted a hierarchical regression analysis with people orientation as the dependent variable. All regression results are displayed in Table 3. Step 1, explained a significant proportion of variance in leaders’ people orientation and revealed that contempt was negatively associated with leaders’ people orientation (p < .001). No significant main effect of power was found (p = .31). In line with our prediction, we found a significant contempt × power interaction effect (p = .03; see Table 3). Supporting Hypothesis 1a, subsequent simple slopes analyses showed that with higher levels of power (1 SD above the mean), contempt was strongly negatively associated with people orientation, b = -0.30, SE _b = 0.05, t(190) = 6.61, p < .001, whereas with lower levels of power (1 SD below the mean), contempt showed a weaker negative association with leaders’ people orientation, b = -0.15, SE _b = 0.05, t(190) = 3.12, p = .002.

Table 3: Summary of Regression Analysis for Contempt and Power Predicting People Orientation and Ethical Leadership in Study 2

Note. N = 194. R ² represents the adjusted R ².

* p < .05

** p < .01

*** p < .001

Ethical leadership. To test Hypothesis 1b, we conducted a hierarchical regression analysis with ethical leadership as the dependent variable (see also Table 3). Step 1, explained a significant proportion of variance in ethical leadership and revealed that contempt was negatively associated with ethical leadership (p < .001), whereas power was positively associated with ethical leadership (p < .001). Yet, these main effects were qualified by the anticipated contempt × power interaction (p = .02; see Table 3). Simple slopes analyses showed that with higher levels of power (1 SD above the mean), contempt was strongly negatively associated with ethical leadership, b = -0.26, SE _b = 0.03, t(190) = 7.74, p < .001. With lower levels of power (1 SD below the mean), contempt showed a weaker negative association with leaders’ people orientation, b = -0.14, SE _b = 0.04, t(190) = 3.97, p < .001. In sum, we found support for Hypothesis 1a and 1b by showing that the negative relationship between leaders’ contempt and their people orientation as well as the negative relationship between leaders’ contempt and ethical leadership was stronger when leaders’ position power was higher^2,4 .

INTRODUCTION STUDY 3

Study 1 and Study 2 show that the integrative effect of leader trait contempt and power may result in a lessening of constructive attitudes and behaviors towards subordinates. Specifically, leader contempt was negatively related to leader people orientation and ethical leadership when leader power was high rather than low. However, we contend that contempt does not only instigate less positive attitudes and behaviors towards subordinates, but that it may also engender more negative attitudes and behaviors towards subordinates. To this end, our third study will focus on leaders’ dehumanization and their self-serving leadership.

Dehumanization has often been studied in the context of intergroup violence and conflict. Representing others as subhuman expedites and justifies aggression towards these persons (Waytz & Schroeder, 2014), which can clearly be harmful. Dehumanization refers to the deindividualization of people by divesting them of human qualities (cf. Haslam, 2006; Kelman, 1976). Once dehumanized, people are no longer viewed as deserving basic human consideration (Moore, Detert, Treviño, Baker, & Mayer, 2012), nor do they evoke compassion or empathic concern (cf. Haslam, 2006). Instead, dehumanized individuals are morally excluded – meaning that they are placed “outside the boundary in which moral values, rules, and considerations of fairness apply” (Opotow, 1990: 1) – and, consequently, are cutoff from moral regard (Deutsch, 1990). In addition, it has been argued that the dehumanization of victims can be functional in disengaging oneself from self-sanctions and self-condemnation (Bandura, 2002). That is, inflicting harm upon a person who is regarded as lacking fundamental human features is less likely to result in self-censure than inflicting harm upon a person who is viewed as a human being with dignifying qualities. Indeed, an experimental study revealed that individuals deprived of humanness were treated in a particularly aggressive manner (Bandura, Underwood, & Fromson, 1975).

For the purpose of the present research, it is important to note that feelings of disconnectedness can prompt dehumanization (Opotow, 1990). The social function of contempt is to avoid contact with, and, therefore, to be disconnected from the recipient of one’s contempt. It has also been argued that it is the depersonalization and objectification of others that accompanies feelings of contempt, which increases the likelihood that those with higher levels of contempt commit hurtful acts towards others (e.g., Izard, 1977; Melwani & Barsade, 2011). Based on this, we argue that leader contempt may be positively associated with leaders’ dehumanization. As in Studies 1 and 2, we expect this relationship to be particularly strong when the leader has high power.

Just like leaders who dehumanize others, self-serving leaders, who place their own well being above the needs of their subordinates, can have a severe impact on the people who work for them (cf. Camps, Decoster, & Stouten, 2012). By regarding others as inferior and unworthy of any consideration, leader contempt may be positively related to self-serving behavior. Self-serving leader behavior, which is characterized by leaders’ focus on personal gain while losing sight of the needs and interests of subordinates, has been argued to lie at the heart of many ethical scandals (cf. Van Gils, Van Quaquebeke, & Van Knippenberg, 2010). Indeed, behavior that sprouts from egoism and demonstrates a lack of responsiveness to the needs and interests of others is commonly considered to be unethical (Howell & Avolio, 1992). In line with recent findings, showing that a tendency to experience contempt in everyday life is negatively associated with greed-avoidance, whereas it is positively associated with callousness (e.g., not caring about hurting others in the pursuit of one’s own needs) and abusive supervision (Sanders et al., 2015), we posit that leaders with higher levels of contempt are inclined to act self-servingly to the detriment of subordinates’ needs, especially when they have higher levels of power. Hence, we hypothesized that:

Hypothesis 2a. Contempt will be more positively associated with leaders’ dehumanization when the leader has higher levels of power than when the leader has lower levels of power.

Hypothesis 2b. Contempt will be more positively associated with leaders’ self-serving behavior when the leader has higher levels of power than when the leader has lower levels of power.

Thus, Study 3 extends the findings of Study 1 and 2 by examining whether the interactive effects of contempt and power also predict destructive attitudes and behaviors towards subordinates. In addition, in Study 3 we measured our variables non-concurrently (at three different points in time), thereby reducing common method variance that can be attributed to the measurement of constructs at the same point in time. Furthermore, we used Study 3 to test whether contempt (in interaction with power) explained variance above and beyond dark side personality traits (i.e., psychopathy, Machiavellianism, and narcissism).

STUDY 3

Method

Respondents and procedure. Data from individuals holding a leadership position (recruited via Mechanical Turk) were collected at three different points in time. After respondents were informed about the study’s requirements and the monetary rewards attached to completing each of the surveys (Time 1: $0.60; Time 2: $1.00; Time 3: $1.00; with an additional $0.50 bonus for completing all three surveys), we asked screening questions pertaining to respondents’ position within the company and the number of subordinates they supervised. Leaders, who completed the survey at Time 1, were one week later invited via a personal message to complete a survey at Time 2. In turn, those who completed the survey at Time 2 were one week later invited to complete a survey at Time 3. At Time 1, we assessed demographic indicators and leaders’ self-reported contempt. At Time 2, we assessed leaders’ self-reported power, and at Time 3 we assessed leaders’ propensity to dehumanize other people and their self-serving behavior at the expense of subordinates. Reminder messages were sent to those without complete surveys to maximize response rates (Dillman, 2000).

Of the 250 respondents who completed the survey at Time 1, 167 individuals responded to the second survey (66.80% retention rate), and 137 individuals (42.3% women, M _age = 35.03, SD _age = 10.03) responded to the third survey (82.04% retention rate). The demographic data collected at Time 1, revealed that respondents’ average work experience in a supervisory position was 4.54 years (SD = 4.39) and on average they supervised 10.55 workers (SD = 13.79). Most of the leaders worked in the industry of business and finance (19.7%), education (19.0%), or human services (16.8%), and 75.2% held an Applied Sciences or University degree.

Measures. Means, standard deviations, correlations, and reliabilities are displayed in Table 1.

Contempt. Leaders’ self-reported contempt, assessed at Time 1, was measured with the same items as in Study 1.

Power. Power, assessed at Time 2, was measured with the same items as in Study 1 using a five-point Likert scale (1 = strongly disagree, 5 = strongly agree).

Dehumanization. Dehumanization, assessed at Time 3, was measured with the three-item dehumanization subscale of the propensity to morally disengage scale (Moore et al., 2012) using a five-point Likert scale (1 = strongly disagree, 5 = strongly agree).

Leader self-serving behavior. One way for leaders to act self-servingly to the detriment of subordinates’ needs is by securing higher monetary benefits for themselves, leaving less money left to divide among subordinates. To this end, we assessed leaders’ self-serving behavior to the disadvantage of subordinates needs by relying on a short scenario in which leaders could decide on the percentage of money they would self-award as a bonus (cf. Wisse & Rus, 2012). Specifically, respondents were asked: “Suppose, you as a leader, can payout bonuses to yourself and 9 of your employees. Please indicate what percentage of the available amount of money for bonuses you will allocate to yourself”. Higher scores indicated that leaders’ self-awarded a higher percentage of the available amount of money to themselves and that a smaller percentage of the money was left over to pay out bonuses for subordinates. An equal distribution of bonuses would result in awarding everyone (including the leader him/herself) a bonus of ten percent.

Results

Preliminary analyses. A CFA – comparing six different factor structures – on our predictor variable items (i.e., contempt and power) as well as our dependent variable items (i.e., dehumanization and self-serving behavior) revealed that a four-factor model had a better fit to the data, χ²(184) = 346.58, p < .001; CFI = .88, RMSEA = .08, SRMR = .08, than alternative models with fewer factors. The best fitting model of these alternatives (a three-factor model in which contempt and power loaded on their own factor and dehumanization and self-serving behavior together loaded on a single factor), χ²(186) = 352.16, p < .001; CFI = .87, RMSEA = .08, SRMR = .08, did not fit the data better. The superiority of the four-factor model over the three-factor model failed to reach significance (∆χ²(2) = 5.58, p = .06), but the CFA seems to provide some support for the theoretical notion that all factors are empirically distinct.

In Study 3, we relied on single-source data, and, therefore, we performed Harman’s single-factor test. An unrotated exploratory factor analysis revealed neither a single factor nor one general factor that accounted for the majority of variance, thereby alleviating potential concerns about single source bias.

Hypothesis testing

Dehumanization. To test Hypothesis 2a, we conducted a hierarchical regression analysis with dehumanization as the dependent variable. All regression results are displayed in Table 4. The findings revealed a main effect of contempt (p < .001). Leaders who score higher on contempt are more likely to dehumanize other people. No main effect was found for power (p = .27). The contempt × power interaction on dehumanization was not statistically significant (p = .08; see Table 4), but simple slopes analyses pointed in the predicted direction. For those leaders with higher levels of power, contempt was strongly positively associated with dehumanization, b = 0.54, SE _b = 0.16, t(133) = 3.44, p = .001. However, for those leaders with lower levels of power, contempt was not related to dehumanization, b = 0.18, SE _b = 0.12, t(133) = 1.50, p = .14.

Table 4: Summary of Regression Analysis for Contempt and Power Predicting Dehumanization and Self-Serving Behavior in Study 3

Note. N = 137. R ² represents the adjusted R ².

* p < .05

** p < .01

*** p < .001

Leader self-serving behavior. To test Hypothesis 2b, we conducted a hierarchical regression analysis with self-serving behavior as the dependent variable (see also Table 4). Again, we found a main effect of contempt (p < .001). Leaders who score higher on contempt are more likely to act self-servingly. No main effect was found for power (p = .26). More importantly, we found a significant contempt x power interaction (p = .03; see Table 4). Simple slopes analyses indicated that with higher levels of power, contempt was strongly positively associated with leaders’ self-serving behavior, b = 11.54, SE _b = 2.95, t(133) = 3.91, p < .001. However, with lower levels of power, leaders’ contempt was not related to their self-serving behavior, b = 2.92, SE _b = 2.30, t(133) = 1.27, p = .21 (see Figure 2). Taken together, we found support for Hypothesis 2b by showing that the positive relationship between leaders’ contempt and self-servingness was stronger when leaders’ position power was higher. In addition, simple slopes analyses revealed, as predicted in Hypothesis 2a, that the positive relationship between leaders’ contempt and dehumanization was stronger when leaders had higher levels of power rather than lower levels of power⁵.

Figure 2: Percentage of Money Self-Awarded as a Function of Contempt and Power in Study 3.

Auxiliary analyses. Increased research attention has been devoted to the impact of dark side personality traits on leader behavior (cf. Krasikova, Green, & LeBreton, 2013). As trait contempt may show conceptual overlap with these personality traits, we also conducted analyses controlling for these dark side personality traits and measured leaders’ self-reported psychopathy (α = .81), Machiavellianism (α = .82), and narcissism (α = .82) at Time 1 using a twelve-item scale (1 = strongly disagree, 5 = strongly agree; Jonason & Webster, 2010). Contempt was positively associated with each of the dark triad traits (r = .59 with psychopathy, r = .51 with Machiavellianism, and r = .36 with narcissism). Including these theoretical control variables in the analyses provides the opportunity to demonstrate that high power leaders’ feelings of contempt are positively related to their dehumanization and self-serving behavior beyond the influence of dark side personality traits. Indeed, when controlling for psychopathy, Machiavellianism, and narcissism the findings did not substantially change in direction or level of significance.

DISCUSSION

Central themes in most corporate scandals are a lack of ethical leadership and concern for others, a tendency to dehumanize others and a ruthless focus on self-gain (cf. Van Gils et al., 2010). In light of the plentiful negative consequences of such attitudes and behaviors it seems essential to identify the determinants of such lack of regard of subordinates on the part of the leader. In the present research, we show that contempt limits leaders’ people orientation and – more broadly – hinders their display of ethical leadership, whereas it fosters their dehumanization of others and acts of self-interest at the expense of subordinates. Moreover, we argued and demonstrated that these relationships depend on the amount of perceived power. Across three field studies, we consistently showed that particularly with higher levels of power, contempt is negatively associated with leaders’ people orientation and their display of ethical leadership, whereas contempt is positively associated with powerful leaders’ dehumanization and self-serving behavior. As such, these results are in line with research from the power approach theory which indicates that power frees people from situational constraints (e.g., Rus et al., 2010; Whitson, Liljenquist, Galinsky, Magee, Gruenfeld, & Cadena, 2013) and aligns individuals’ behavior with their internal feelings, preferences, and traits (e.g., Piteša & Thau, 2013; Weick & Guinote, 2008). We further learned that our findings were not affected by including age, gender, level of education, and dark side personality traits (psychopathy, Machiavellianism, narcissism) as control variables. This indicates that the combined effects of contempt and power on people orientation, ethical leadership, dehumanization, and self-serving behavior occur independent of age, gender, level of education, and leaders’ dark side personality traits.

Theoretical Implications

An investigation of the influence of contempt on leader behavior is particularly timely given that emotions are more and more recognized as being an integral part of daily interactions in the workplace (e.g., Anderson & Jones, 2000; Barsade et al., 2003). The present research is, to our knowledge, one of the first to highlight the important role of leaders’ trait-like tendency to experience contempt in the leader-subordinate relationship. By empirically testing the influence of contempt on leaders’ people orientation, ethical leadership, dehumanization, and self-serving behavior we showed that dispositional and discrete emotions might play an important role in predicting leadership styles that are characterized by more or less concern for subordinates. As such, the findings of our study adds further credence to the idea that it is valuable to study trait based discrete emotions in addition to current approaches in which dispositional affect is examined through summarizing the wide variety of possible human affective experiences into a few critical underlying dimensions (Barsade & Gibson, 2007). A focus on discrete emotions may lead to the uncovering of relationships that would not have been revealed by a focus on general experiences of negative and positive affect alone (see Angie, Connelly, Waples, & Kligyte, 2011 for a meta-analytic review on discrete emotions). Furthermore, the present research contributes to literature examining the influences of trait emotions. So far, this stream of literature has especially focused on trait anxiety and trait anger (e.g., Spielberger & Reheiser, 2009), we add to this literature by demonstrating that trait-like contempt may have implications for leaders’ attitudes and behaviors vis-à-vis their subordinates.

Previous theorizing suggests that leaders’ experience of positive other-directed emotions (e.g., gratitude, admiration, sympathy) motivates leaders to act on their other-regarding values (e.g., treat others with respect; Michie & Gooty, 2005). The present research adds to this line of reasoning by showing that in contrast to positive other-directed emotions, the negative other-directed emotion of contempt is negatively associated with powerful leaders’ people orientation and ethical leadership and positively associated with their dehumanization and self-serving behavior. Notably, within the clusters of positive (e.g., gratitude, admiration, sympathy) and negative (e.g., contempt, anger, disgust) other-directed emotions, some discrete emotions may have better predictive value for certain outcomes than others (Angie et al., 2011). In the present research, we expected the negative other-directed emotion of contempt to be particularly likely to have an influence on leaders’ behavior vis-à-vis subordinates, because it arises to a larger extent in hierarchical and formal relationships – of which the leader-subordinate relationship is an example – than the negative other-directed emotions of anger and disgust (Hutcherson & Gross, 2011). Furthermore, leaders’ may be particularly apt to experience contempt (Pelzer, 2005) due to their higher status position. Such a position implies superiority and dominance over others (Rozin et al., 1999; Tiedens, 2001), which may elicit appraisal styles that are conducive to the experience of contempt (Revelle & Scherer, 2009; Van Reekum & Scherer, 1997). However, future research is needed to test the idea that the negative other-directed emotion of contempt indeed explains variance above and beyond the negative other-directed emotions of anger and disgust with respect to predicting leader behavior vis-à-vis subordinates (cf. Sanders et al., 2015).

Given the assumed link between hierarchical position and contempt – Melwani and colleagues (2012) found for instance that individuals displaying contempt were more often categorized as leaders – one may also assume that (perceived) contempt and power covary likewise. Testifying to the idea that contempt is an emotion that is argued to be tied to hierarchy (Rozin et al., 1999) but not necessarily to higher levels of power within hierarchical positions, we did not find that (perceived) contempt was inherently related to higher levels of power (see Table 1). Rather, we found that power moderates the relationship between contempt and leaders’ treatment of their subordinates, such that particularly those leaders with higher levels of power act upon and in line with their feelings of contempt.

Practical Implications

From a practical point of view, our findings suggest that it might be wise to screen leaders on their trait-like tendency to experience feelings of contempt, and to be careful with awarding contemptuous leaders higher levels of power. In contrast, less contemptuous leaders could more ‘safely’ be granted higher levels of power within their leadership positions. Fair to note, the role of power increasing a focus on individuals’ feelings and preexisting dispositions may not be limited to contempt. As such, it is wise to not exclusively screen leaders’ on their tendency to experience feelings of contempt, but to screen leaders’ for other preexisting tendencies that may be associated with less favorable behaviors as well.

Additionally, previous research has shown that feelings of contempt are particularly intense for those that are deemed incompetent or unintelligent (Hutcherson & Gross, 2011). As such, altering unjustified perceptions of subordinates’ incompetence or unintelligence may reduce leaders’ contemptuous feelings. One way to accomplish this is to highlight the different competencies that subordinates may have and to establish an organizational culture in which subordinates are allowed to make errors (i.e., an error management culture; Van Dyck, Frese, Baer, & Sonnentag, 2005), without having to be afraid to be disqualified as a person. Another way to reduce leaders’ feelings of contempt is to limit leaders’ in their experience of contempt by clarifying feeling rules in the work place (cf. Hochschild, 2008). Moreover, increasing leaders’ experience of other-directed positive emotions (e.g., gratitude, admiration, sympathy) by developing or training these emotions may help to boost leaders’ people orientation and ethical leadership and reduce their dehumanization and self-serving behavior at the expense of subordinates (Michie & Gooty, 2005).

Strengths, Limitations, and Future Directions

Although we found consistent support for our hypotheses across three field studies, conducted in two different countries (i.e., the United States and the Netherlands), our research is not without its limitations and our results should be interpreted in light of these limitations. First, in Study 2 and Study 3 we relied on single-source data. Yet, in Study 1 we relied on multi-source data and found a similar pattern. Furthermore, we aimed to address this limitation by conducting a test to detect common method variance and found that it might not play a substantial role. Taken together, this increases the confidence in our findings, but it should be noted that ex post statistical methods for detecting common method bias are not always completely accurate (cf. Podsakoff et al., 2003), and that the collection of multi-source data is preferred in future studies.

Second, the nature of our data does not allow us to draw firm conclusions about causal relationships. The design of Study 3 in which trait contempt (measured at Time 1) in interaction with power (measured at Time 2) predicts leaders’ self-serving behavior (measured at Time 3) a few weeks later, suggests that trait contempt can indeed be viewed as an antecedent of leader behavior. The interactive effect of contempt and power on dehumanization did not reach statistical significance (p = .08), but given that the simple slopes revealed the predicted pattern of results, we take these findings as further evidence that trait contempt may manifest itself in powerful leaders’ orientation towards others. Of course, time-lagged data cannot as strongly provide causal evidence for the order of effects as a well-controlled experiment. Although manipulating trait contempt may prove difficult, an experimental design could investigate the effects of more transitory experienced feelings of contempt (instead of a trait-like tendency to experience feelings of contempt) on leaders’ attitudes and behavior. Previous research demonstrated that the measurement of trait-like tendencies to experience feelings of contempt and the manipulation of immediate experiences of contempt yields similar results (cf. Sanders et al., 2015), and as such future research employing an experimental design could provide stronger evidence for causal paths.

Third, leader emotions are not only found to play an important role in predicting leader behavior and leadership styles they are also found to play a crucial role in shaping follower outcomes. For instance, Damen, Van Knippenberg, and Van Knippenberg (2008a) found that leader emotional displays in interaction with follower affect shaped follower performance. Through the process of emotional contagion (Hatfield et al., 1994), the emotions felt and displayed by the leader may affect the emotions experienced by followers. As such, leader contempt may breed contempt on the part of the follower (Melwani & Barsade, 2011). In line with the findings presented in the current research, this may, in turn, negatively influence the ethicality of the follower. Furthermore, leaders’ contempt may indirectly influence follower behavior via their lack of ethical leadership and demonstration of self-serving behavior at the expense of the subordinate. Theories such as social learning theory (Bandura, 1986) and social exchange theory (Homans, 1961) provide an explanation for why subordinates may copy the behavior of their leader (e.g., via role-modeling) or respond with negative treatment in reaction to negative treatment (e.g., following a negative reciprocity principle). Highly contemptuous leaders’ lack of people orientation and ethical leadership as well as their dehumanization and demonstration of self-serving behavior may, therefore, result in unfavorable behavior on the part of the follower, such as supervisor-directed or organizational deviance. Future research could be geared at testing a broader model, in which leader contempt has implications for the emotions experienced by subordinates, which, in turn, may have implications for subordinates’ behavior and performance.

Fourth, the studies presented in the current paper do not provide empirical evidence pertaining to the underlying mechanisms that may explain why power and contempt interact to predict leaders’ constructive (e.g., people orientation, ethical leadership), as well as their destructive, attitudes and behaviors (e.g., dehumanization, self-serving behavior). In line with recent research demonstrating power’s capacity to increase the likelihood that people will act upon their own preferences via a heightened self-focus (i.e., awareness of one’s own personal thoughts and feelings; Piteša & Thau, 2013) or an increased reliance on accessible constructs (Guinote, Weick, & Cai, 2012), we surmise that the stronger negative association of contempt with leaders’ people orientation and ethical leadership and the stronger positive association of contempt with leaders’ dehumanization and self-serving behavior when the leader has higher levels of power rather than lower levels of power, is due to high power individuals’ increased focus on their own feelings and/or reliance on accessible constructs. Yet, future research should explicate whether this is indeed the case.

On a related note, the question why some leaders are more likely to experience contempt than others is one that deserves more attention in future research. Some argue that individual differences in trait emotions can be the result of genetic factors and/or a result of learning and socialization processes (Revelle & Scherer, 2009). This implies that the tendency to experience contempt could sprout from broader personality traits (e.g., a lack of communal orientation; Clark, Quellette, Powell, & Milberg, 1987) or from certain characteristics in the environment that would foster specific styles of causal attribution or appraisal (see Connelly, Helton-Fauth, & Mumford, 2004).

Returning to the example of Hostess Brands, both subordinates working for the company as well as the media blamed the company’s executives for showing contempt for the labor force and not caring about the workers. This example illustrated the co-occurrence of powerful leaders contempt and their lack of people orientation and ethical leadership. The present set of studies underscore this anecdotal evidence by showing that leaders’ lack of people orientation and ethical way of leading may be grounded in their feelings of contempt and may manifest itself in particular when leaders have the power to act upon these feelings. Furthermore, we demonstrated that contempt might even foster powerful leaders’ dehumanization and self-serving behavior at the expense of their subordinates. We hope that the findings presented in the current paper inspires more research investigating the role of contempt in the leader-subordinate relationship, as well as research on power’s revealing function.