How do you really feel? Effect of leaders' perceived emotional sincerity on followers' trust of

: We extend research on leadership and emotions by addressing two previously under-researched areas. Prior research has focused primarily on the valence of leaders' displayed emotion and on followers' affective reactions to those displays. In contrast, we examined followers' cognitive reactions to the perceived sincerity of leaders' displayed emotion. Study 1 found that American workers' trust in a leader was related to their perceptions of that leader's emotional sincerity. Study 2 replicated these results among Chinese workers, and further indicated the mechanisms through which perceived emotional sincerity influenced trust and showed how trust influenced performance. The findings demonstrate the importance of including emotional sincerity in studies of leader affect, and suggest the value of adding a cognitive perspective to the current focus on followers' affective reactions to their leaders' emotions.


Introduction
On January 7, 2008, potential presidential candidate Hilary Clinton gave a speech in New Hampshire in which she was described as "tearing up" (Breslau, 2008). Those tears became a top news story, and debate raged about whether they were genuine and what that revealed about her leadership potential (Krone, 2008). Beliefs about Clinton's emotional sincerity seemed to be as important as the emotion itself, and attracted far more attention than her speech's content. Many observers apparently felt the display was sincere and that judgment of emotional sincerity caused them to change their assessment of Clinton's character (Goldenberg, 2008).
This example illustrates the importance of leaders' emotional sincerity. However, the topic is under-researched (Gooty et al., 2010, Van Kleef et al., 2012, because the study of emotions in leadership is still relatively new (Eberly & Fong, 2013). Most prior research has focused on leaders' expressed emotion or their experienced emotion, but not the relationship between the two (Gooty et al., 2010). Nonetheless, several scholars have suggested the importance of followers' beliefs about their leaders' emotional sincerity (i.e., match between experience and expression), recognizing that a leader's sharing of true feelings, in contrast to displaying insincere ones, may fundamentally change followers' assessments of that leader (Eberly and Fong, 2013, Gardner et al., 2009, Van Kleef et al., 2012. The Clinton example highlights this importance, not just in the fact that more attention was paid to her emotional sincerity than to any other aspect of the speech, but also in the results that her apparent sincerity produced. Clinton displayed a negative emotion, which generally leads to poor reactions from followers (Gooty et al., 2010), but she nonetheless attracted many new supporters and won the New Hampshire vote despite polls predicting a loss (Goldenberg, 2008, November 10, Krone, 2008. In this case, Clinton's perceived emotional sincerity seems to have been more important than valence, enhancing her effectiveness with followers.
To advance the study of leadership and emotions, we examined followers' perceptions of their leaders' emotional sincerity in two studies. The results show that followers' trust is influenced by their leaders' apparent emotional sincerity, which makes four important contributions. First, the results demonstrate how perceptions of a leader's emotional sincerity influence follower trust. Second, our data reveal the mechanisms of that influence. Third, we provide a measure to use in future studies of perceived emotional sincerity. And finally, our data suggest the value of including a cognitive perspective in the study of leadership and emotion.

Theory and hypotheses
Our research is based on the assumption that followers will make judgments about their leader's emotional sincerity and those judgments will lead to inferences about the leader's trustworthiness, which in turn will influence followers' trust in the leader and their subsequent behavior. In this section, we define emotional sincerity and consider why leaders may not always be sincere. We then explain our focus on followers' perceptions of emotional sincerity. Following this explanation, we use the Mayer, Davis, and Schoorman (1995) model of trust to explain each of the steps in the chain that links perceptions of emotional sincerity to trust and subsequent performance.

Leaders' emotional sincerity
Emotions have two components: experiential and expressional (Elfenbein, 2007). The experiential component concerns how an emotion is felt by an individual; the expressional component is the external display of that feeling, concerning how individuals reveal the emotions they experience. The two components are distinct (Fridlund, 1992), meaning that the emotion a person displays need not match the emotion being experienced (i.e., an emotion display may lack sincerity).
Leaders' emotion displays may not always align with their emotion experience. The influential power of expressed emotion may motivate leaders to intentionally modify their displays for strategic reasons (Humphrey, Pollack, & Hawver, 2008). For example, a leader may intentionally amplify negative emotions, such as anger, to motivate followers to work harder (Van Kleef et al., 2009). Likewise, a leader may be motivated by impression management concerns to display particular emotions, such as faking positive affect when interacting with followers to seem more charismatic (Bono & Ilies, 2006). Whenever leaders would prefer to display a particular emotion, but are not experiencing that emotion, they may be inclined to strategically modify their emotion displays. Consistent with this possibility, research suggests that leaders do frequently engage in strategic displays of emotion, and do so more frequently than non-leaders (Glaso & Einarsen, 2008).
In general, leaders may engage in three types of emotion display, each with a different degree of sincerity (Ashforth andHumphrey, 1993, Gardner et al., 2009). The first type of emotion display is a genuine one, in which the leader expresses the emotion being felt internally. For example, a leader who is excited about a new project may display her internal emotion experience. The congruence between the internal emotion experience and the external emotion display makes the display wholly sincere. In contrast, if the leader introducing the new project did not feel excited, but wished to display excitement to followers, that leader would have to regulate her emotion display using either deep acting or surface acting (Grandey, 2003). Deep acting involves efforts to reassess the situation or otherwise move oneself toward feeling the desired emotion. In our example, the leader may search for reasons to feel more excited about the project. Surface acting lacks the effort to feel the required emotion, and involves simulating the emotion by making a fake behavioral display (e.g., the leader would put on an excited face despite not feeling so). Empirical evidence confirms that observers distinguish among these three different types of display (i.e., genuine emotion, deep acting, surface acting), and that observers' responses are typically more negative as the display becomes more insincere (Cote et al., 2013, Diefendorff et al., 2005, Grandey, 2003.

Followers' perceptions of emotional sincerity
Emotional sincerity is a purely intra-psychic phenomenon; if people feel that they are genuinely expressing their emotions, then they are (Harter, 2002). However, as followers lack direct knowledge of leaders' internal states, they must make their own judgments about a leader's emotional sincerity. That is, they must determine whether or not they perceive an emotion display as sincere. It is this judgment that influences their subsequent reactions. If a leader feels sincere, but is not perceived so by followers, the feeling of sincerity is unlikely to have any direct effect on followers (Erickson, 1994).
Many factors may influence the sincerity that followers perceive in a leader's displayed emotion. Not least of these factors is the display's actual sincerity, since insincere emotion displays are noticeably different from sincere ones (Frank, Ekman, & Friesen, 1993), and likely to be revealed through nonverbal clues (Mehrabian & Wiener, 1967). In addition, a variety of leader characteristics, follower characteristics, and contextual factors are likely to influence followers' perceptions of emotional sincerity. For example, gender (Eagly, 2005) and demographic similarity (Pittinsky & Tyson, 2005) may be potentially important moderators, influencing both how sincere leaders actually are and how sincere they are perceived to be. Culture may be another important factor, as different groups think about emotion differently and may express the same emotions in different ways (Ekman et al., 1987). However, regardless of the cause or the level of accuracy, followers will form some judgment about their leaders' emotional sincerity, since interpretations and attributions are the way that we make sense of the world (Heider, 1944). Therefore, followers will form and act on beliefs about their leaders' emotional sincerity.
Moreover, the beliefs that followers form about leaders are likely to be global in nature (Dasborough & Ashkanasy, 2002), such that followers will regard their leader as emotionally sincere or insincere in general. The fundamental goal when any person observes another is to draw conclusions about the actor's motives and dispositions (Ybarra, 2002), and these conclusions are strongly biased toward trait-like explanations (Gilbert & Malone, 1995). In other words, observers are most likely to assume that another's behavior reflects stable personal traits. As a result, followers' beliefs about a leader's emotional sincerity are likely to have a trait-like character. While emotional sincerity is obviously a variable state, such that a leader can be more or less sincere in any given display, followers are prone to make global judgments about emotional sincerity. Those judgments may be changed by particular emotion displays, but the bias will be toward thinking of sincerity in trait terms.
The Clinton example that opened the article demonstrates this bias. Clinton was generally perceived as emotionally controlled and non-expressive before the New Hampshire speech, but her display of emotion that day led many to subsequently change their assessment of her (Goldenberg, 2008). Reports suggest that observers modified their trait-like assessments of Clinton; rather than thinking that she was someone who usually hid her emotions but showed them that one time, the audience seemed to have updated their view of her stable personality to consider her an expressive and sincere person. As this example highlights, and attribution research implies (Gilbert andMalone, 1995, Ybarra, 2002), followers likely form global assessments of their leaders' emotional sincerity. These assessments are the focus of our research.
Follower perceptions of a leader's emotional sincerity should not be confused with related, but distinct, constructs such as authentic leadership, emotional labor, integrity and trust (Palanski & Yammarino, 2009). Authentic leadership is defined to include emotional sincerity, though it also includes far more, being defined as a multidimensional construct that includes transparency, information processing, self-awareness and morality (Gardner, Cogliser, Davis, & Dickens, 2011). While followers may need to consider leaders emotionally sincere before they will judge them authentic, leaders may be emotionally sincere without meeting all the other criteria of authentic leadership (Caza & Jackson, 2011). Similarly, emotional labor and the closely related construct of emotion regulation are antecedent to emotional sincerity (Grandey, 2000). If a job demands the display of specific emotions, it requires emotional labor, and this requirement may be satisfied by the use of emotion regulation and potentially lead to a lack of emotional sincerity, but these actions precede the judgment of whether a leader exhibits sincerity. In contrast, attributions of integrity and trust in a leader are consequences of perceived emotional sincerity. As described below, followers' beliefs about their leaders' emotional sincerity are information that followers use to make judgments about the leaders' trustworthiness.

Consequences of perceived emotional sincerity
While there has been relatively little study of perceived emotional sincerity in leadership, there is evidence that suggests its importance. Interestingly, however, the available data and interpretation imply both positive and negative results from perceptions of emotional sincerity in leaders. Humphrey (2012) argues that the skilled use of emotion regulation makes leaders more effective by improving leader-follower relations and increasing the leader's charisma, while Bass (1990) similarly suggests that emotion control is among the most valuable skills for effective leaders. Because those judged to have more self-control are perceived as more trustworthy (Righetti & Finkenauer, 2011), leaders may benefit from being seen to have the capacity to control their emotions, rather than being controlled by them. In fact, individuals who engage in emotion regulation have been shown to be more effective in managing teams toward greater performance (Jiang, Zhang, & Tjosvold, 2013). Based on these observations, one might expect that there are benefits to leaders who are seen to be able to display the "right" emotion, rather than a sincere one.
Nonetheless, there is perhaps stronger evidence that sincere emotion displays produce better responses. For example, Grandey (2003) found that employees with sincere emotion displays were rated more positively by coworkers. Newcombe and Ashkanasy (2002) found that leaders were judged less favorably when followers perceived incongruence between the valence of leaders' feedback and displayed emotion, which may suggest that perceptions of emotional insincerity produce negative responses. Similarly, Fisk and Friesen (2012) found that followers who perceived their leader as more frequently displaying sincere emotions reported higher satisfaction. As such, while we recognize that there may be disadvantages associated with sincerity in specific situations, we expect that, on balance, followers' trait-like perceptions of emotional sincerity in a leader will lead to more positive responses. Specifically, as described below, we predict that perceptions of emotional sincerity will foster trust among followers.

Perceived emotional sincerity and trust
In this paper, we follow Mayer and colleagues (Mayer et al., 1995, Schoorman et al., 2007, who define trust as a psychological state in which one has positive expectations of the trusted person, and so is willing to be vulnerable to the actions of that person. Trust of this sort is among the most important of follower responses to leadership, and has figured prominently in empirical studies for decades (Dirks & Ferrin, 2002). In fact, if one defines leadership as influencing others to achieve shared objectives (Yukl, 2012), followers' trust is almost indispensable. Leaders who are not trusted by followers will be far less effective (Burke, Sims, Lazzara, & Salas, 2007).
It has been noted that Mayer and colleagues' model of trust is primarily cognitive, focusing on followers' evaluations of signals from a leader (Schoorman et al., 2007). For this reason, it was the most appropriate model for our study, which concerns followers' cognitive reactions to leaders' perceived emotional sincerity. In particular, we predict that followers will use their beliefs about a leader's emotional sincerity as a source of information from which to make judgments about that leader, and particularly how much to trust him or her.
Followers will trust a leader to the extent that they perceive that leader as being trustworthy, and their assessment of trustworthiness is based on three qualities: the leader's ability, the leader's integrity, and the leader's benevolence (Mayer et al., 1995). The more of these qualities a leader is believed to have, the more a follower will trust that leader. Because our concern here is with the consequences of perceived emotional sincerity, we address only the second and third qualities. It is not obvious that perceptions of emotional sincerity in a leader will have any consistent effect on followers' beliefs about that leader's ability (i.e., the possession of skills and knowledge required to succeed in a particular domain; Mayer et al., 1995). In contrast, as we describe below, perceptions of emotional sincerity can be expected to influence follower beliefs about their leader's integrity and benevolence.
The first quality that influences follower trust is leader integrity. Integrity has been defined in many ways (Palanski & Yammarino, 2009), but in the model of Mayer et al. (1995), it refers to the quality of being consistent in adhering to a set of values. More specifically, followers will trust a leader if they believe that leader consistently adheres to a set of values that the followers admire and approve. We recognize that others have argued against defining integrity in terms of follower perceptions (e.g., Palanski & Yammarino, 2009), but since our concern here is with understanding follower reactions, it is their perceptions that are most important. Followers who perceive their leader as having integrity believe that they know what to expect from that leader and that the leader will behave in an appropriate fashion. If a leader reliably behaves in a fashion that followers judge to be appropriate, followers will be more willing to risk making themselves vulnerable to that leader. The relationship between perceived integrity and trust is wellestablished (Colquitt, Scott, & LePine, 2007), and we expected to replicate it in our data.
What has not previously been examined is the effect of perceived emotional sincerity on follower beliefs about a leader's integrity. However, we predict that leaders whom followers perceive as emotionally sincere are likely to also be perceived as having integrity. Emotional sincerity is honestly expressing one's internal affective experience. As such, a follower who perceives a leader to be emotionally sincere believes that leader maintains consistency between feelings and actions. One might assume that those who are consistent and honest about their emotions are likely to be consistent and honest in general, and hence have the quality of integrity. Moreover, openness and honesty are generally considered desirable and normatively appropriate behaviors, so that people displaying them are more likely to be judged as having integrity and good character (Peterson & Seligman, 2004). Indeed, evidence shows that individuals tend to judge emotionally sincere others as open and honest, whereas those who feign emotions seem manipulative and dishonest (Cote et al., 2013). Likewise, insincere leaders are perceived as hypocritical, disingenuous and less credible (Gardner et al., 2009). Underscoring the power of sincerity, it should be noted that even when the emotion displayed is positive (e.g., smiling), a perception of insincerity can make followers question the leader's motives and lead to negative attributions (Frank et al., 1993). As such, we expect that leaders perceived as emotionally sincere will be judged as having more integrity, and thus be more trusted.

Hypothesis 1.
Followers who perceive a leader as emotionally sincere will perceive that leader as having more integrity.

Hypothesis 2. Replication
Followers who perceive a leader as having more integrity will have greater trust in that leader.
The second quality that influences follower trust is benevolence, or more precisely, followers' belief that the leader has benevolent intentions toward them (Mayer et al., 1995). If followers believe that a leader wants to help and benefit them, they will be more willing to submit themselves to that leader's influence. Consistent with this reasoning, the positive relationship between attributions of benevolence and trust is well-established in prior research (Colquitt et al., 2007). However, it should be noted that there is sometimes empirical confusion between integrity and benevolence as individual traits (Schoorman et al., 2007). Moreover, as Mayer et al. (1995) noted, attributing benevolence to another implies a belief that there is some specific attachment between trustee and trustor. Combining these facts with Dirks and Ferrin's (2002) meta-analytic results showing that the two fundamental pathways through which leaders gain follower trust are personal character and the leader-follower relationship, we focus here on benevolence operationalized as a positive, high quality relationship between leader and follower.
As described in the extensive literature of leader-member exchange theory (LMX; Erdogan & Bauer, 2014), leaders and followers typically begin with a contractual relationship, but as their dyadic tie strengthens, it can move toward a relationship based on social exchange. Social exchange relationships are distinguished by their non-contractual nature, in which valued resources are exchanged in a fashion that creates feelings of mutual goodwill, obligation and reciprocity (Wayne, Shore, & Liden, 1997). In other words, when in a high quality leaderfollower relationship, the follower supports and cares for the leader, and expects to receive the same treatment in return. A follower in a high LMX relationship will thus assume that the leader has benevolent intentions, and so will be more willing to trust that leader.
Follower perceptions of leaders' emotional sincerity will influence beliefs about leader benevolence, in the form of leader-follower relationship quality, because sincerity is interpreted as a signal about the actor's relationship intentions (Van Kleef et al., 2009). Insincere displays of emotion are often interpreted as revealing a lack of desire to build a close and open relationship (Liu & Perrewe, 2006). As a result, followers who perceive their leader as emotionally insincere may feel that their relationship with that leader is distant and superficial, that the leader has less benevolent intentions toward them (Griffith, Connelly, & Thiel, 2011). In contrast, individuals' willingness to share what they are truly feeling can suggest that the relationship is strong and that the leader feels benevolently toward the follower (Butler et al., 2003). Consistent with this reasoning, evidence shows that leaders and followers who engage in more sincere emotional exchanges report higher quality relationships (Glaso & Einarsen, 2008). As such, followers who perceive their leaders as more emotionally sincere should also judge their relationship with that leader more positively, implying an expectation of benevolence from the leader.
Hypothesis 3. Followers who perceive a leader as emotionally sincere will perceive a higher quality relationship with that leader (i.e., greater benevolence).

Hypothesis 4. Replication
Followers who perceive a high quality relationship with a leader (i.e., more benevolence) will have greater trust in that leader.

Trust and performance
There is established meta-analytic evidence supporting the relationship between trust and performance in general (Colquitt et al., 2007), and specifically between trust in a leader and follower performance (Dirks & Ferrin, 2002). When followers trust their leaders, they are more satisfied with those leaders (Fulmer & Gelfand, 2012) and subsequently devote more time and attention toward the leaders' ends (Colquitt et al., 2007, Lapierre, 2007. Followers who trust their leader are more willing to adopt that leader's goals and vision, and therefore can be better contributors toward organizational ends (Dirks, 2000). Trust allows followers to concentrate all of their attention on work tasks, without needing to worry about protecting themselves from leaders (Mayer & Gavin, 2005). As such, followers who trust their leaders should be more effective in their work (Brower, Lester, Korsgaard, & Dineen, 2009), both in formally required or in-role tasks, as well as in discretionary or extra-role efforts.

Hypothesis 5. Replication
Followers who trust their leader will have greater performance, both in-role (H5a) and extra-role (H5b).

Item selection
We were not aware of any existing measure to assess one's perception of another's emotional sincerity, but there were several self-report scales for closely related constructs, including emotional labor (Brotheridge andLee, 2003, Glomb andTews, 2004), surface and deep acting (Grandey, 2003), and personal authenticity (Kernis andGoldman, 2006, Sheldon et al., 1997). To maximize coverage of the construct domain and variety in phrasing, we used the items from these scales as inspiration for ours. Items were modified to reflect the differences between those constructs and emotional sincerity, to be observations of others rather than self-report, and to focus on leaders. The resulting items were given to ten researchers and practitioners in a sorting task, along with the items from measures of the closely related constructs of authentic leadership (Neider & Schriesheim, 2011) and perceived leader integrity (Craig & Gustafson, 1998). We retained the six items which were associated with the proper construct definition by at least 90% of the group: (1) "My manager's emotions are credible"; (2) "My manager shows his/her true feelings when dealing with me"; (3) "My manager is sincere about his/her emotions"; (4) "My manager pretends or puts on an act about his/her emotions" (reverse score); (5) "My manager fakes his/her emotions and feelings" (reverse score) and (6) "My manager uses fake emotions" (reverse score). These items were then used in the confirmatory assessment describe below.

Participants and procedure
We used an online participant recruiting firm because our aim was to have as diverse and representative a sample as possible. Online recruiting firms such as the one we used have been shown to be at least as good as traditional recruitment techniques in terms of sample representativeness, demographic diversity, and data reliability, while being more efficient (Buhrmester et al., 2011, Paolacci et al., 2010. Desiring a sample size of at least 100 (Gorsuch, 1983), and anticipating some low quality responses, we collected data from 160 respondents. We excluded 15 respondents because they had less than five years of work experience, and another 34 because they failed to answer the survey's data quality control items correctly (i.e., we included three items in the survey that were of the form "enter 3 for this question" and any respondent who entered a number other than 3 was assumed to not be reading carefully, and excluded).
Our final sample consisted of 111 respondents (45.9% female). All but two reported that English was their first language. They had a mean age of 47.6 years (SD = 14.0) and an average of 27.0 years of work experience (SD = 13.1). The survey asked the respondents to complete the measures below about their work and their immediate supervisor, with whom they had worked for an average of 11.3 years (SD = 10.9). The majority (66.7%) reported that their immediate supervisor was female.

Measures
Study 1 assessed the convergent validity of the new measure of a leader's perceived emotional sincerity, explored its criterion validity with relevant outcomes (including trust in leader, job satisfaction, and perceived leader integrity), and considered its incremental explanatory power over the closely-related leadership styles of authentic leadership and transformational leadership (a leadership style in which followers are made aware of the value of their work and motivated to transcend self-interest; Podsakoff, MacKenzie, Moorman, & Fetter, 1990). We used 5-point Likert-type scales to measure all variables (1 = "strongly disagree"; 5 = "strongly agree").
Follower ratings of their leaders' perceived emotional sincerity used the six items described above, and had good reliability (α = .95). Trust in leader (α = .94) was measured using the items from Casimir, Waldman, Bartram, and Yang (2006). Sample items were "my manager can be relied on to uphold my best interests" and "I feel quite confident that my manager will always try to treat me fairly." Satisfaction with leader (α = .95) used Hackman and Oldham's (1974) scale (e.g., "I am satisfied with the amount of support and guidance I receive from my supervisor"). Job satisfaction (α = .74) also used a Hackman and Oldham (1974) scale, which included the item "I often think of quitting the job" (reverse score). Leader integrity (α = .89) was measured by a three-item, short-version of the perceived leader integrity scale of Craig and Gustafson (1998), which focuses on the display of normatively desirable behavior and is most consistent with the definition of integrity used in this research. Sample items include "my manager would lie to me" (reverse score) and "my manager would allow someone else to be blamed for his/her mistake" (reverse score). Authentic leadership (α = .95) was measured by Neider and Schriesheim's (2011) 16-item scale (e.g., "my manager seeks feedback to improve interactions with others" and "my manager makes decisions based on his/her core beliefs"). Transformational leadership (α = .95) was assessed using the 12 items reflecting the "core" qualities in Podsakoff et al. (1990) scale: identifying and articulating a vision ("My manager paints an interesting picture of the future"), providing an appropriate role model ("My manager leads by example") and fostering acceptance of group goals ("My manager encourages employees to be team players").
We included control variables for follower age, gender, work experience, and tenure with the leader, as well as the leader's gender. Also, since our measures were all collected from a single source, we included a five-item negative affectivity scale developed by Watson and Clark (1994), in which respondents rated the extent to which they felt negative affect such as anger or sadness (α = .86). The inclusion of this scale allowed us to use the partial correlation procedure described by Podsakoff, MacKenzie, Lee, and Podsakoff (2003) to control for the effect of common method bias.

Study 1: results and discussion
As revealed in Table 1, many of the constructs were highly correlated, reflecting their conceptual similarity and strong relationships (Cooper et al., 2005, Gardner et al., 2011, Palanski and Yammarino, 2009, as well as the fact that all data came from a single source (Podsakoff et al., 2003). However, the correlations were not dissimilar in size to those observed in other studies with similar constructs (Braun et al., 2013, Connelly and Ruark, 2010, Podsakoff et al., 1990. Moreover, as indicated below, all available evidence suggested the discriminant validity of the measures. A confirmatory factor analysis (using LISREL 8.70) of the six-item perceived emotional sincerity (PES) scale fit well (Χ 2 = 12.22, df = 9; RMSEA = .05; CFI = 1.00; TLI = 1.00). All path loadings exceeded .78 and the average variance extracted was .79, indicating convergent validity. To assess the discriminant validity of the PES measure, another confirmatory factor analysis was conducted using all seven constructs. The seven-factor model fit well (Χ 2 = 537.97, df = 303; RMSEA = .08; CFI = .99; TLI = .98), with all items loading on the appropriate latent variable. The average variance extracted was high (.75 to .81) and larger than the squared correlations among variables, satisfying the criteria for discriminant validity (Fornell & Larcker, 1981). To further test the discriminant validity of the PES measure, we compared the sevenfactor model with several rival models in which we combined PES with another construct as one factor (see Table 2). All of the rival models had significantly worse fit than the seven-factor model, further supporting the discriminant validity of the PES measure.
Finally, to assess and control for possible common method bias in the data, we took two complementary steps (Podsakoff et al., 2003). Harman's one-factor test in an unconstrained exploratory factor analysis (using SPSS version 21) revealed a multi-factor solution, the first factor of which accounted for only 28.4% of the total variance. In addition, as described below, we included negative affectivity as a control for the influence of common method bias (Podsakoff et al., 2003).
In SPSS, we used hierarchical regression to test the criterion and incremental validity of the PES scale. As shown in Table 3, including perceived emotional sincerity significantly improved explanatory power for all but one of the outcomes. That is, after controlling for demographics, negative affectivity (i.e., common method bias), authentic leadership and transformational leadership, perceived emotional sincerity still had a significant and positive relationship with attributed integrity (β = .29, p < .01), trust in the leader (β = .26, p < .01), and satisfaction with one's manager (β = .16, p = .02). However, PES did not consistently predict overall job satisfaction (β = .09, p = .45), which is appropriate as a leader's emotional sincerity is unlikely to be a key predictor of general satisfaction. In addition, Kmenta's (1997) t-test found no significant difference between the coefficient effect sizes of (a) PES and authentic leadership in relation to leader integrity (t = .70, p = .49), trust in leader (t = 1.31, p = .19) and satisfaction with leader (t = 1.30, p = .20), or (b) PES and transformational leadership in relation to the outcomes (t = 1.45, .46, 1.04; p = .15, .65, and .30, respectively). These results suggest that PES is just as important a predictor of follower responses to a leader as are authentic leadership and transformational leadership.    (Neter, Kutner, Nachtsheim, & Wasserman, 1996). There was no evidence of problematic multicollinearity in these models. ⁎ p < .05. This sample of experienced workers, who had considerable tenure with their current manager, supported the convergent, discriminant and predictive validity of the new measure of a leader's perceived emotional sincerity. Workers' global judgments about their leaders' emotional sincerity predicted their trust and satisfaction. Moreover, perceived emotional sincerity was important beyond the effects of authentic and transformational leadership. In addition, the fact that each of the three ratings of the leader (transformational, authentic, emotional sincerity) had a unique pattern of relationships with the outcomes further supports their discriminant validity. As such, these findings demonstrate the importance of followers' beliefs about their leader's emotional sincerity. However, while the use of single-source self-report data provided a particularly stringent test of discriminant validity, it could not reveal the mechanisms or behavioral implications associated with perceived emotional sincerity. Therefore, we conducted Study 2 to address these questions.

Participants and procedure
We surveyed the employees of three entertainment and service companies in northwest China. The companies were organized in functional teams with distinct roles, such as tea servicing and cuisine preparation. The teams offered several advantages to our study. First, leaders were influential in these teams: teams were relatively small; there was little formal training for employees, so leaders were vital; and team leaders were directly involved in all aspects of work, from task assignment to performance management. Second, while China is modernizing quickly, it continues to be influenced by traditional values, which include power distance, submission to authority, and strong rules about emotion displays (Farh, Hackett, & Liang, 2007). Traditional Chinese followers may be less critical of their leaders (Zhang, Bai, Caza, & Wang, 2014) and are typically expected to be loyal to supervisors without expecting anything in return (Jiang & Cheng, 2008). As such, in addition to the generalizability benefit of collecting data in a new country, China provided a particularly conservative test of the importance of perceived emotional sincerity. If Chinese workers do not expect as much openness and sincerity as American workers, the effect of perceived emotional sincerity would be reduced in China.
After promising confidentiality and securing individual consent, we provided one of two kinds of paper surveys to participants. Followers (i.e., workers) completed all measures below, except the two performance measures. These measures were in the second questionnaire, which was completed by leaders (i.e., team managers) to reduce same-source and self-report biases. To protect confidentiality, we used employee identification numbers to match the leader and follower data, and we collected follower questionnaire on site immediately after completion. The study was endorsed by each firm's CEO, and surveys were completed during work hours.
In total, we received responses from 211 leader-follower dyads, representing 47 leaders, with an average of 4.5 followers per leader (overall response rate 90.6%). The average age of leaders was 26.5 years (SD = 4.9), most were male (61.7%), and their average job tenure was 4.0 years (SD = 2.8). Followers had an average age of 24.1 years (SD = 4.1), with an average job tenure of 1.6 years (SD = 1.8). Most followers were female (76.3%). All respondents were full-time employees; most had no college education.

Measures
A native speaker translated all of the materials from English, and then another native speaker translated them back to English to ensure clarity and consistency. All measures used 7-point Likert scales of agreement. Perceived emotional sincerity (α = .94) and trust in leader (α = .88) were measured with the same items as in Study 1. Perceived leader integrity (α = .94) was measured with the full 8-item scale of Craig and Gustafson (1998) used in Study 1. Benevolence, as relationship quality (α = .89), was measured with Graen, Liden, and Hoel's (1982) 7-item LMX scale (e.g., "my leader would 'bail me out' at his/her expense"). The in-role performance measure (α = .86) was taken from Janssen and Van Yperen (2004). Sample items included "this worker always completes the duties specified in his/her job description" and "this worker fulfills all responsibilities required by his/her job." Extra-role performance (α = .91) was measured with the organizational citizenship behavior scale developed by Farh, Earley, and Lin (1997) for use in China (e.g., "this worker is willing to help colleagues solve work-related problems" and "this worker takes his/her job seriously and rarely makes mistakes").
To allow an additional test of our PES measure's discriminant and incremental validity, followers also rated their leader's authentic leadership with the same scale as in Study 1 (α = .91). In addition, we collected demographic information on followers, leaders, and their organizational context.

Analytic strategy
Our hypotheses and data were all at the individual level, but the responses were not independent, because followers were grouped in teams under leaders. Moreover, group membership proved to be a significant predictor of some outcomes (see Table 4). Therefore, we used multilevel structural equation modeling (in EQS 6.1), which offered several advantages: it accounted for the nested structure of the data and separated the between-group variance from the within-group variance; it simultaneously tested all hypotheses; it accounted for potential covariance among dependent variables; and it explicitly modeled variables as latent constructs (Preacher, Zyphur, & Zhang, 2010). To reduce model complexity, we treated each dimension of Farh et al. (1997) scale as an indicator. We assessed several potential alternative measurement models, and then used a comparative approach to test rival structural models against our predictions (Bollen & Long, 1992).    Table 5 presents a summary of the variables measured. This summary reveals that none of the demographic or organizational variables were significantly related to the variables of interest. This pattern of results replicated those in Study 1, where demographic variables were nonsignificant controls. As a result, in the interests of parsimony and clarity, we followed the advice to exclude control variables for which we had neither theoretical nor empirical rationale (Aguinis andVandenberg, 2014, Becker, 2005).
As shown in Table 6, a multilevel confirmatory factor analysis including all six study variables and the control variable of authentic leadership (i.e., baseline model) fit the data well (Χ 2 = 1655.45, df = 1438; RMSEA = .03; CFI = 1.00; TLI = 1.00). 4 All items had significant loadings above .60 on the correct factor, indicating convergent validity. The average variance extracted ranged from .52 to .95 and always exceeded the largest squared correlation, indicating discriminant validity (Fornell & Larcker, 1981). In addition, the baseline model had a better fit than all of the alternative measurement models (rival models 1 to 9). These results suggest the validity of the measures used in Study 2, and the distinctiveness of the PES construct. To test our hypotheses, we compared the fit of our predicted model with that of several rivals. As shown in Table 7, our predicted six-variable (i.e., baseline) model had a good fit with the data (Χ 2 = 1449.87, df = 1176; RMSEA = .03; CFI = 1.00; TLI = 1.00). We rejected the change-oforder models (rivals A and B) for having worse fits, which provides some evidence against reverse causality and in support of our predictions. However, we found evidence of additional direct paths from perceived emotional sincerity to trust and to in-role performance. The best fitting model was one that included all of our predictions, but also two additional effects from perceived emotional sincerity (rival D), as shown in Fig. 1. While the predictor and criterion scores were from different sources, common method bias still may have inflated the relationships among variables reported by followers. To assess this possibility we used Harman's one-factor test, which suggested a multi-factor solution in which the first factor accounted for only 12.0% of the total variance. In an additional test, we included a method variance latent factor in the structural model, and found that its inclusion did not substantively change any results or improve the model fit (rival F in Table 7). Based on the similar results from these two different tests, we judged that common method bias was not a significant threat in these data (Podsakoff et al., 2003) and adopted the results in Fig. 1.
We used Sobel (1982) tests to further assess mediation. The results showed that the indirect effects from PES to trust were significant through both leader integrity (Z = 2.85, p < .01) and relationship quality (Z = 4.28, p < .01). The indirect effects from leader integrity to performance through trust were also significant, both for in-role performance (Z = 2.27, p = .02) and for extrarole performance (Z = 3.30, p < .01). The same was true for relationship quality, which had significant effects through trust on both in-role performance (Z = 2.48, p = .01) and extra-role performance (Z = 4.07, p < .01). Moreover, all of the three-stage mediation tests had significant results: PES → leader integrity → trust → in-role performance (Z = 2.01, p = .04); PES → integrity → trust → extra-role performance (Z = 2.62, p = .01); PES → relationship quality → trust → in-role performance (Z = 2.38, p = .02); PES → relationship quality → trust → extra-role performance (Z = 3.65, p < .01). These results are supportive of the mediation sequence in Fig. 1.
As in Study 1, we also assessed the incremental predictive power of perceived emotional sincerity relative to authentic leadership. In a new model (G in Table 7), we included authentic leadership as an additional independent variable. In this model, perceived emotional sincerity was a significant predictor of both attributed leader integrity (β = .21, p = .01) and relationship quality (β = .54, p < .01), while authentic leadership predicted relationship quality (β = .55, p < .01), but not attributed leader integrity (β = .19, p = .06). Therefore, in these data, as in Study 1, perceived emotional sincerity added incremental predictive power.
In China, as in the United States, followers appear to be concerned with their leader's emotional sincerity. Similar to Study 1, the data revealed that global perceptions of emotional sincerity were associated with trust in one's leader. Furthermore, Study 2 extended the results of Study 1 by providing evidence consistent with the hypothesized mediation process and the performance consequences of perceived emotional sincerity.

General discussion
Our research was motivated by the desire to learn whether followers' perceptions of their leaders' emotional sincerity were important, and if so, how. In the first study, we used a sample of experienced American employees and found that followers who believed their leader was more emotionally sincere reported greater trust in that leader. Study 2 supported and extended these findings, showing that Chinese employees also feel greater trust in leaders they believe are emotionally sincere, and that this increased trust improves their work performance. These results contribute to our understanding of the role of emotions in leadership, as described below.

Theoretical and methodological contributions
We consider these studies to be a first step toward a more complete theory of emotional sincerity in leadership. The results demonstrate the importance of perceived emotional sincerity by showing that followers' global judgments about their leader's emotional sincerity are related to their attitudes and behavior. Specifically, when a leader is perceived as emotionally sincere, followers have greater trust in that leader and they deliver better work performance. In short, leaders' emotional sincerity matters to followers.
Studies of leader affect, therefore, may benefit from including emotional sincerity. There is already strong evidence that leaders' displayed emotions influence followers, but that evidence has implicitly assumed the emotions were sincere, because it has overwhelmingly focused on either the leader's internal emotion experience or the external emotion display, without considering the relationship between the two (Eberly andFong, 2013, Gooty et al., 2010). However, one cannot assume that leaders are always emotionally sincere (Glaso & Einarsen, 2008), nor that followers always perceive them as such (Fisk & Friesen, 2012). Therefore, we suggest that future studies need to take account of sincerity, whether of the leader's emotions in general, as we did here, or of the discrete emotion display being studied (Gooty et al., 2010).
To support the inclusion of perceived emotional sincerity in future studies, we developed a scale for its measurement. In a recent review of the emotions and leadership literature, Gooty et al. (2010) suggest that psychometrically sound measures of emotion-related constructs are lacking and need development. We have responded to this call by developing and validating a brief scale to measure follower perceptions of a leader's emotional sincerity. The scale performed well among workers in both the United States and China. It showed good convergent and discriminant validity, appropriate relationships with related constructs, and important predictive power, including incremental explanation beyond the effect of related leadership behaviors. Our measure is a foundation for future study.
More generally, our findings have implications for how we study follower responses to leader emotion. Previous research has focused primarily on the emotional responses of followers (Gooty et al., 2010), either through direct transmission of leaders' emotions (Bono andIlies, 2006, Erez et al., 2008) or the affect elicited by leaders' emotion displays (Gaddis, Connelly, & Mumford, 2004). Relatively few studies have adopted a cognitive perspective and examined the inferences that followers make in response to their leaders' emotion displays (though see Eberly andFong, 2013, Van Kleef et al., 2009). We contributed to this latter perspective by examining how followers use leaders' emotional sincerity as information with which to make judgments about the leaders' trustworthiness. Using Mayer et al. (1995) model of trust, we showed that followers appear to use a leader's emotion displays as information about whether or not that leader can be trusted. We do not deny that followers' affective responses are important, but our data and others' suggest that cognitive responses are also pertinent. Therefore, future studies should consider both affective and cognitive responses (e.g., Van Kleef et al., 2009).
With regard to combining both cognitive and affective responses, we speculate that in addition to followers' cognitive responses to the perception of emotional sincerity, followers may also have important affective responses to leader sincerity. For example, a leader's seemingly sincere display of anger may provoke different emotional responses than would a leader who seemed to be faking anger. While our data did not include measures of such affective reactions, we may have found indirect evidence of them.
In addition to supporting our predictions, the data also revealed the presence of direct paths from perceived emotional sincerity to trust and to performance. That is, as we predicted, there was an apparently conscious, cognitive response to perceived emotional sincerity, in that leaders judged more sincere were also judged more trustworthy. However, after taking account of this response, there were two additional positive effects associated with perceived emotional sincerity: leaders perceived as emotionally sincere were trusted more and received better in-role performance, above and beyond the effects associated with trustworthiness. It may be that these effects reflect followers' affective reactions to their leaders' sincerity. If sincerity elicits positive affect (e.g., sincerity may generate a feeling of warmth and closeness to the leader), that positive affect may influence followers' subsequent attitudes toward the leader (i.e., trust), as well as how they approach their work (Fredrickson, 2001).
This possibility raises interesting questions about interactions between perceived sincerity and valence. For example, how would followers' reactions to an apparently sincere display of negative emotion differ from their reactions to an equivalent negative display that was judged to be insincere? Based on the evidence that insincere positive emotion displays produce less positive (i.e., more negative) responses than sincere ones (Frank et al., 1993), it is our tentative hypothesis that while the valence of a leader's negative emotion display will produce negative responses in followers, those responses will be less negative when the display is judged as sincere, rather than insincere. That is, followers' negative reactions to insincerity will add to their negative reaction to the valence. However, this hypothesis is purely speculative; we are not aware of evidence that directly addresses such issues.
Another possible interpretation of the unmediated paths we found involves the content of the sincere emotion displays. Evidence suggests that positive emotion displays are more likely to be perceived as sincere (Dasborough andAshkanasy, 2002, Eberly andFong, 2013). If that is the case, then it may be that leaders who were rated as more emotionally sincere also displayed more positive emotions, and those positive emotions may have been transmitted to followers and explain the additional effects observed in our study. Our data cannot distinguish between these two explanations, but they do highlight the many exciting possibilities revealed by taking account of emotional sincerity. Since we have demonstrated the importance of perceived emotional sincerity and provided a measure for its assessment, we hope that future work will examine the interrelationships among valence, sincerity and discrete emotions in leadership.

Limitations and future directions
While our research had some valuable strengths, including consistent findings across two studies and two countries, it also included limitations that suggest promising directions for future study. For one, because our data were cross-sectional and based on well-established leader-follower relationships, we cannot comment on how perceptions of emotional sincerity develop. Consistent with previous findings (Dasborough & Ashkanasy, 2002), our data suggest that followers form meaningful global judgments about their leaders, but it will be useful to learn how these judgments form.
In particular, what are the leader and follower characteristics that influence perceptions of emotional sincerity? For example, it appears that, all other things being equal, positive emotion expressions are more likely to be perceived as sincere (Dasborough andAshkanasy, 2002, Eberly andFong, 2013), which suggests that those in more positive environments and those who have greater dispositional positive affect may be perceived as more sincere. Similarly, one could imagine that greater personal familiarity or dyadic similarity (e.g., gender, race, age) might influence assessments of a leader's emotional sincerity.
There may also be contextual factors that influence judgments of emotional sincerity and responses to it. While acting on their general beliefs about a leader's emotional sincerity, followers may also consider the specific emotion being displayed and the setting where it occurs. For example, Gaddis et al. (2004) found that the normally poor response to negative emotion displays did not pertain in challenging contexts where such emotions were merited. As such, a fine-grained examination of specific, episodic responses may provide great insight. Do particular emotions seem more or less sincere in a particular context? Do some types of displays have more or less effect on global judgments?
Finally, being a first study of emotional sincerity, our research examined only main effects. That is, our hypotheses and findings concerned how perceived emotional sincerity influences follower outcomes through a chain of causal mediators. While we found evidence of significant effects, it may be that perceived emotional sincerity is also an important moderator in leadership. For example, might the emotional sincerity attributed to leaders influence their success in other leadership behaviors (Gardner et al., 2009)?

Practical implications and conclusion
For managers, our results offer cautionary evidence against the advice that leaders should use their emotions strategically. Managers appear to engage in more emotion regulation than any other employee group (Glaso & Einarsen, 2008), and leaders receive advice about strategic emotion displays (Humphrey et al., 2008). However, we question the value of leaders using displayed emotion as a tool. While an insincere display of excitement or anger may produce the desired immediate results, one must wonder whether the potential long-term reduction in trust and performance justifies the action. If a leader cannot be certain that an emotion display will seem sincere, it may be better not to take the risk.
Moreover, the implications of these findings should not be considered in isolation, but rather as a complement to previous results. While our findings demonstrate that perceived emotional sincerity is an important consideration for followers, it is only one consideration. As an illustration, we return to the Clinton example from the start of the article. As we noted, most observers apparently judged Clinton's emotion as sincere (Goldenberg, 2008, November 10, Krone, 2008, and Clinton won an unexpected victory in the New Hampshire vote. However, when asked for a response to Clinton's emotion display, competitor John Edwards said, "What we need in a commander-in-chief is strength and resolve," implying that someone who cried on the campaign trail lacked strength (CNN, 2008). This comment is important, because if Clinton had only shown her feelings of fatigue and worry, without also demonstrating the strength and creativity to deal with them, the emotion display could be interpreted as the inability to lead despite its sincerity. Therefore, the results of our research should be taken to demonstrate that emotional sincerity is important in leading others, but it is clearly not the only factor that matters. A leader who is perceived as sincere but otherwise inept is unlikely to maintain follower support. As such, we look forward to seeing future research clarify the links between emotional sincerity and other elements of the leader-follower dynamic.