Power and Emotional Contagion: the Role of Attention, Relational Identification and Trust

Home , Emotional contagion, Group affective tone

Power and Emotional Contagion: The Role of Attention, Relational Identification and Trust

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Emily Knecht Tarr, B.A.

Graduate Program in Labor and Human Resources

The Ohio State University

2016

Dissertation Committee:

Steffanie L. Wilk, Advisor

Robert B. Lount, Jr.

Bennett Tepper

Copyrighted by

Emily Knecht Tarr

2016

Abstract

This dissertation examines how attention plays a role in power relationships and individuals’ susceptibility to emotional contagion. By integrating separate literatures on the power—attention relationship and the attention—emotional contagion relationship, I hypothesize and test that the relationship between power and emotional contagion is mediated by attention. Moreover, I examine variables that may affect how power relates to attention: relational identification and trust based on different factors, but propose that these variables work differently for high- and low-powered people. These moderators may further explain the relationship between power and emotional contagion, suggesting a moderated mediation model. This model is tested using two samples, one composed of working professionals and the other of students in a laboratory setting. Results support the mediation hypothesis, but only with negative emotions. Moderation relationships were supported for high-power individuals, but not for low-power individuals.

Dedicated to my husband, Nate, and my parents for giving me all the love and support I

could have hoped for throughout my studies.

iii

Acknowledgments

I would like to sincerely thank my advisor, Steffanie Wilk, for her invaluable help in the completion of this dissertation. I would also like to thank my other committee members,

Robert B. Lount, Jr. and Bennett Tepper for their advice and support that made this dissertation what it is. Thank you also to Sarah Doyle for help on statistical analyses and unending support throughout the dissertation process.

Vita

June 2007 ...... Worthington Kilbourne High School

2011...... B.S. Psychology, Spanish, University of

Michigan

2011 to present ...... Graduate Teaching Associate, Department

of Management and Human Resources, The

Ohio State University

Fields of Study

Major Field: Labor and Human Resources

Table of Contents

Abstract...... ii

Acknowledgments...... iv

Vita...... v

Table of Contents...... vi

List of Tables ...... ix

List of Figures...... xi

Introduction...... 1

Attention and Power...... 4

Attention and Influence...... 8

Hypotheses...... 10

The mediating role of attention in the power—emotional contagion relationship ...... 10

Moderators that Decrease Attention for Low-Power Individuals ...... 15

Moderated Mediation of Trust in Benevolence and Integrity for Low-Power

Individuals...... 17

Moderators that Increase Attention for High-Power Individuals...... 18

Moderated Mediation of Trust Based on Ability and Relational Identification for High-

Power Individuals...... 22

Methods Overview...... 24

Study 1 ...... 28

Methods...... 28

Measures...... 29

Results ...... 33

Discussion ...... 36

Study 2 ...... 38

Methods...... 38

Measures...... 41

Results ...... 48

Discussion ...... 53

General Discussion ...... 54

Theoretical Contributions...... 54

Methodological Contributions...... 58

Limitations ...... 60

Practical Implications...... 63

vii

References...... 65

Appendix A: Calculation of the Emotional Contagion Measure...... …...... 78

Appendix B: Tables and Figures...... 80

viii

List of Tables

Table 1: Means, standard deviations, and intercorrelations for variables in Study 1.

(Continued) ...... 80

Table 2: Results for regression analyses testing the main effect of power on attention... 82

Table 3: Results for regression analyses testing the interaction between power and benevolence predicting attention ...... 83

Table 4: Results for regression analyses testing the interaction between power and integrity predicting attention...... 84

Table 5: Results for analyses testing the curvilinear moderation between power and ability predicting general attention...... 85

Table 6: Results for analyses testing the curvilinear moderation between power and ability predicting attention to emotions...... 86

Table 7: Results for regression analyses testing the interaction between power and relational identification predicting attention...... 87

Table 8: Means, standard deviations, and intercorrelations for variables in Study 2. Note

N = 121. (Continued) ...... 88

Table 9: Results of analyses testing the direct effect of power on emotional contagion.. 90

Table 10: Results of analyses testing the indirect effect of power on the three measures of attention...... 91

Table 11: Results of analyses testing the indirect effect of attention on emotional contagion...... 92

Table 12: Results of analyses examining the interaction between power and benevolence predicting the three measures of attention...... 93

Table 13: Results of analyses examining the interaction between power and integrity predicting the three measures of attention...... 94

Table 14: Results of analyses examining the curvilinear interaction between power and ability predicting self-reported attention...... 95

Table 15: Results of analyses examining the curvilinear interaction between power and ability predicting average gaze length...... 96

Table 16: Results of analyses examining the curvilinear interaction between power and ability predicting attention percentage...... 97

Table 17: Results of analyses examining the interaction between power and similarity predicting the three measures of attention...... 98

Table 18: Examples of emotional contagion scores...... 99

List of Figures

Figure 1: Proposed theoretical model...... 100

Figure 2: Proposed interaction of Hypothesis 5...... 101

Figure 3: Proposed interaction of Hypotheses 3a and 3b...... 102

Figure 4: Plot showing the curvilinear interaction between power and ability predicting attention to emotions……………………………………………………………………103

Figure 5: Plot showing the moderated effect of relational identification on the power- attention relationship...... 104

Introduction

It is widely recognized that social influence is a primary consequence of power

(Cartwright, 1965; French & Raven, 1959; Kipnis, 1972, 1976; Lewin, 1951). Powerful individuals have an increased ability to change the behavior, thoughts, and feelings of those over whom they hold power (Asch, 1955; Cialdini & Trost, 1998; French & Raven,

1959). One type of social influence that powerholders have is emotional influence (e.g.,

Sy, Côté, & Saavedra, 2005; Van Kleef, De Dreu, & Manstead, 2004). Emotional contagion or the “sending” and “catching” of moods among people (Hatfield, Rapson, &

Cacciopo, 1994) is a form of emotional influence that may be affected by power relationships. Given the relationship between power and influence over others, low- power people should be more likely to be influenced by—or catch—the emotions of a person who has power over them while powerholders should be more likely to influence the emotions of others (i.e., send emotion) and relatively unlikely to catch the emotions of others.

An individual’s level of power also has psychological and behavioral effects

(Keltner, Gruenfeld, & Anderson, 2003), which may in part explain the relationship between power and emotional influence. One such behavioral effect of power is social attention, such that individuals who hold great power are unlikely to pay attention to

1 others, while individuals with low power pay great attention to those that have power over them (Keltner & Robinson, 1997; Keltner, et al., 2003; Neuberg & Fiske, 1987).

Given that paying attention to others opens one to another’s influence, the relationship between power and influence may, in part, arise out of the differential behavioral (i.e., attentional) tendencies of high- and low-power individuals. However, if the direction of one’s attention can be changed such that high-power individuals pay greater attention to low-power people, and low-power individuals pay less attention to powerholders, the emotional influence one has over another may not necessarily follow from the amount of power one holds. The purpose of this dissertation, therefore, is to examine how a focal person’s level of power affects his/her behavior, specifically attention to another (target), and how that attention affects the emotional influence that the target has over the focal person. This dissertation will also propose and test moderators to the power—attention relationship, which are hypothesized to change the direction of this relationship.

Attention is driven by people’s goals, resulting in great attention paid to objects or people that help one achieve one’s goals (Dijksterhuis & Aarts, 2010). According to

Power Approach Theory (Keltner et al., 2003), high-power people are motivated to approach rewards. If there are situations in which high-power people view a low-power counterpart as able to provide them with some sort of reward, such as when they relationally identify with them or have very high or very low trust based on their ability, this may garner more attention to that low-power person, potentially opening them to emotional contagion from that person. Conversely, low-power individuals are motivated to avoid negative outcomes and punishments, which, by definition are directly controlled

2 by those that have power over them (Keltner et al., 2003). However, if there are times when low-power people feel assured that powerholders will not allow negative outcomes to occur, such as when they trust in them based on perceptions of their benevolence and integrity, they may pay less attention to high-power people, resulting in a reduced likelihood of catching their emotions. This implies that attention and thus emotional influence over others may not always follow the amount of power one holds.

This dissertation makes two contributions to the literature. First, I expand on prior work on the relationship between power and attention (Fiske, 1993; Fiske & Dépret,

1996; Goodwin, Gubin, Fiske, & Yzerbyt, 2000; Keltner & Robinson, 1997; Ebenbach &

Keltner, 1998) to show that the attention that high- and low-power people pay to each other affect how likely they are to be emotional influenced. Thus, I test whether attention mediates the power-emotional contagion relationship. Second, using Power Approach

Theory (Keltner et al., 2003), I will examine when the negative relationship between power and emotional contagion, which is suggested in the literature (Anderson, Keltner,

& John, 2003; Hsee, Hatfield, Carlson, & Chemtob, 1990; Spoor & Kelly, 2009), may be lessened or strengthened. As goals are presumed to underlie the classically-theorized patterns of attention for low- and high-power individuals, I examine factors that alter the way low- and high-power individuals see their counterparts as relevant to their goals.

Accordingly, I present situations in which high-power people are more likely to focus their attention on and receive greater emotional influence from low-power others, and in which low-power people will decrease their attention to and thus be less susceptible to the emotional influence of powerholder

Theoretical Development

Attention and Power

Attention is marked by orientation toward and processing of specific stimuli (e.g.,

Posner, 1980; Dijksterhuis & Aarts, 2010; Gardner, Dunham, Cummings, & Pierce,

1989). By attending to various stimuli, individuals are able to gain key information that helps them navigate their environment. For instance, paying attention to others can provide infants and other young animals with information about what foods to eat, how to avoid predators, and what appropriate social behaviors are (e.g., Choleris & Kavaliers,

1999). Given that humans live in a social environment, other people are often the targets of attention as they may provide information that helps guide one’s behavior and learning processes (Foulsham, Cheng, Tracy, Henrich, & Kingstone, 2010). Moreover, paying attention to others enables people to make better decisions about cooperation, competition, and communication with these people and others around them (Range,

Horn, Bugnyar, Gajdon, & Huber, 2008). Thus, we tend to pay attention to objects, events, and people that provide us with the most useful information for navigating our environment (Yarbus, 1967; Mackworth & Morandi. 1967; Antes, 1974; Buswell, 1935;

Henderson & Hollingworth, 1999; Foulsham et al., 2010).

However, given that our environment is filled with infinite stimuli and our attention is a limited resource (Kahneman, 1973), it is impossible to attend to every stimulus in our surroundings. As a result, attention is selective, such that we concentrate on certain objects at the expense of others. In fact, that we selectively process certain information while ignoring other information is considered one of the essential functions of attention (Desimone & Duncan, 1995; Dijksterhuis & Aarts, 2010). Thus, much research has sought to determine what may influence where individuals direct their attention (e.g., Loftus & Mackworth, 1978). At a basic level, whether certain stimuli are attended to or not may be driven by bottom-up processes, which reflect sensory stimulation and guide attention toward salient stimuli such as those that are bright or moving quickly (Koch & Ullman, 1985; Corbetta & Shulman, 2002). However, the majority of attention appears to be driven by top-down processes, such as prior knowledge, expectations, or current goals (Corbetta & Shulman, 2002; Dehaene,

Changeux, Naccache, Sackur, & Sergent, 2006; Koch & Tsuchiya, 2006; Land &

Hayhoe, 2001). One top-down factor that has been shown to affect the allocation of attention is one’s social power, or one’s relative capacity to influence others’ outcomes based on having control over valued resources (Keltner et al., 2003).

The key mechanism that helps explain why power relates to attention is goals.

Goals influence both of the key parts of attention: to whom you pay attention (i.e., direction or orientation) and the processing of information about them. Considerable research has demonstrated that individuals use attention as a resource to pursue goals (see

Dijksterhuis & Aarts, 2010 for a review) and has illustrated that individuals pay more

5 attention to incoming information that is relevant for goal attainment than information that is irrelevant (e.g., Aarts, Dijksterhuis, & De Vries, 2001). Although all individuals use attention in the service of goal pursuit, whether one has high or low power greatly affects what those goals are (e.g., Keltner et al., 2003). Thus, power affects attention due to the different goals of those with high versus low power.

Low-Power and Attention. Individuals who are in a position of low power live in an uncertain environment, in which they are subject to threats, punishments, and others’ evaluations of them (Keltner et al., 2003; Fiske, 1993; Steele & Aronson, 1995).

This environment triggers a prevention focus (Higgins, 1997, 1998), meaning that low- power people have a goal to avoid negative outcomes and undesirable end-states (Keltner et al., 2003). More specifically, they seek to avoid punishment and uncertainty and are highly sensitive to threats (Keltner et al., 2003; Keltner & Robinson, 1997). This prevention focus is associated with the activation of the behavioral inhibition system

(BIS), which makes low-power individuals vigilant to those in their environment who could benefit or harm them (Keltner & Robinson, 1997; Keltner, et al., 2003). Thus, because of the uncertainty they face, low-power individuals pay great attention to their environment in order to avoid potential negative outcomes.

An especially important target for the attention of low-power individuals is a person who holds power over them. By definition, a person who has power over another is in control of that individual’s outcomes and determines whether they receive the punishments or negative outcomes that low-power people are motivated to avoid.

Specifically, low-power individuals focus attention on high-power individuals to gain a

6 sense of predictability regarding the intentions and actions of powerholders (e.g., Chance,

1967; Ellyson & Dovidio, 1985; Emory, 1988; Erber & Fiske, 1984). Being able to predict the behavior and intentions of powerholders helps minimize the uncertainty that low-power individuals face and seek to avoid, even though the powerful may not necessarily behave in corruptive ways that are harmful to those they have power over

(e.g., Kipnis, 1972; Chen, Lee-Chai, & Bargh, 2001).

The activation of the BIS for low-power people also means that they are likely to engage in more thoughtful and effortful information processing (Higgins, 1997), a key indicator of increased attention (Posner, 1980; Dijksterhuis & Aarts, 2010; Gardner et al.,

1989; Desimone & Duncan, 1995). Neuberg & Fiske (1987) found that individuals who were resource-dependent (low-power) used individuating processes (i.e., using idiosyncratic characteristics or attributes rather than social categories) to form an impression of a target and thus paid more attention to that target. The authors surmised that low-power individuals use this type of processing to make the most accurate impression of a target, so that they can better predict that person’s behavior and gain greater control over their own outcomes. Similarly, research on motivated information processing theory (De Dreu & Carnevale, 2003) suggests that the more we are motivated to maintain a rich and accurate understanding of certain targets, the more we engage in effortful and systematic processing of information about those targets (De Dreu &

Carnevale, 2003; Kruglanski, 1989; Kruglanski & Ajzen, 1983). Since low-power people are motivated to gain information about powerholders in order to reduce the uncertainty they face, they pay great attention to these powerholders.

High-Power and Attention. Since powerholders live in a reward-rich environment and may feel that they can act without interference or serious social consequences (Weber, 1947), having power activates the behavioral activation system

(BAS; Keltner et al., 2003). As a result, high-power individuals have approach-related goals, or those that have great promise for rewards (Higgins, 1997). The BAS regulates behavior, cognition, and affect that helps the individual pursue and obtain goals related to attainable rewards (Keltner et al, 2003; Anderson & Berdahl, 2002). Attention is one of these resources regulated by the BAS, and since it is a limited resource (Kahneman,

1973), it is predominately fixated on stimuli that provide value. Thus, unless low-power individuals are instrumental in goal achievement or are able to provide some sort of reward, the powerful are unlikely to attend to them.

Moreover, the BAS is associated with quick and automatic cognition, suggesting that those with high power use heuristics rather than individuating information in processing those around them, especially those with low power (Keltner et al., 2003). For example, Goodwin et al., (2000) showed that power increases stereotyping as a result of powerholders’ careless social attention to others. This tendency for automatic social cognition and use of heuristics suggests another explanation for why the powerful tend to pay less attention to others.

Attention and Influence

While attention to certain others provides us with useful information and helps us achieve our goals, it also may open us up to influence from those others. Influence is defined as bringing about change in another (e.g., Cartwright, 1965; March, 1955; Simon,

1957; Cialdini & Goldstein, 2004). There are different types of influence, some of which are more intentional than others. For instance, individuals may try to influence others by attempting to persuade them to come to an agreement in negotiation (e.g., Van Kleef et al., 2004) or to buy a product by offering them a free sample, thus inducing the norm of reciprocity (Cialdini, 2009). However, social influence may also occur in more subtle ways, such as conforming to a popular opinion even if that opinion may not be internally held (Asch, 1952, 1956). While it is not necessary that an individual is aware of the influence they may be receiving, it is necessary that they are paying attention to the influencing agent in order to be influenced by it.

In sum, since individuals’ goals impact where they direct their attention

(Dijksterhuis & Aarts, 2010), and one’s level of power affects one’s goals, whether individuals have high or low power may determine how they direct their attention. This in turn may affect their susceptibility to influence from others. One form of social influence that has been suggested to result from attention is emotional contagion, whereby people “catch” the emotions of other people (Hatfield et al., 1994). Catching the emotions of others implies that an individual’s emotions change following an interaction with others. In the sections below, I discuss how low- and high-power people may be differentially susceptible to emotional contagion from others because of the differences in how they pay attention to others.

Hypotheses

The mediating role of attention in the power—emotional contagion relationship

Emotional contagion is a two-stage process whereby first, individuals mimic the expressions, vocal tones, or postures (reliable manifestations of affect; see Cacioppo,

Petty, Losch, & Kim, 1986) of others, and second, come to experience congruent moods through the process of afferent feedback (Hatfield et al., 1992, 1994; Neumann & Strack,

2000), or physiological feedback from muscular, visceral, and glandular responses that induces subjective feelings (e.g., Adelmann & Zajonc, 1989). Research guided by the perception-behavior link (Bargh, Chen, & Burrows; Dijksterhuis & Bargh, 2001;

Neumann & Strack, 2000) suggests that simply perceiving another’s behaviors, facial expressions, or vocal tones elicits corresponding behavior in the perceiver. However, mimicry is unlikely to occur if attention is not paid to the other, as one will not notice or process the behaviors and emotional expressions of the target. This suggests that the more attention an individual pays to others, the more he or she is likely to mimic them and thus the more likely he or she is to receive emotional contagion from them.

Accordingly, Hatfield et al. (1994) propose that individuals who pay particularly great attention to other people are especially likely to catch their emotions. Moreover,

Neumann and Strack (2000: 212) offer a definition of emotional contagion that involves

10 the “observation of another person’s public display of mood,” suggesting that in order to receive emotional contagion, one must be observing (i.e., attending to) another’s emotional displays. Although it is a central component, the role of attention in the emotional contagion process is often assumed rather than directly tested.

For example, prior studies (Hsee, et al., 1990; Anderson et al., 2003; Spoor &

Kelly, 2009) have examined the relationship between power and emotional contagion assuming that attentional processes link the two. Specifically, they hypothesized a negative relationship between power and emotional contagion and discussed attentional differences as a reason why this may occur. However, the design and findings of these studies suggest opportunities to further understand the power-emotional contagion relationship. First, the findings of these studies are mixed and sometimes contradictory to the authors’ theoretical predictions. While Anderson, et al. (2003) did find results consistent with the predicted negative relationship, Hsee, et al. (1990) and Spoor and

Kelly (2009) found opposite results: that high-power partners were more likely to catch the emotions of their low-power counterparts. Second, although these authors discuss attention as possibly playing a role in the relationship between power and emotional contagion, they do not specifically test attention in their studies. Moreover, Hsee and colleagues (1900) and Spoor and Kelly (2009) suggest their inconsistent results may be due to methods that inadvertently affected the goals and thus attention of high- and low- power individuals. In an attempt to resolve the inconsistent findings from previous studies examining the relationship between power and emotional contagion, the current

11 study proposes and measures attention to a counterpart and focuses on how power and goals are manipulated.

Attention is hypothesized to mediate the relationship between power and emotional contagion. Powerholders are unlikely to orient their attention to those with low power, but even if they do, they are likely to process this information in an effortless and heuristic way (Keltner et al., 2003), which constitutes little attention. This effortless and heuristic information processing is unlikely to lead to mimicry and thus unlikely to result in emotional contagion. Since this type of information processing reflects little effort and thought, individuals are not likely to gain a full or accurate picture of others’ emotional expressions and thus may not fully mimic the other. In support of this, studies have found that those with high-power are comparatively worse at accurately judging others’ emotions (Snodgrass, 1985; Hall, 1979), and that those who engaged in heuristic processing were less influenced by others (Chaiken, 1980). Thus, since powerholders are likely to pay little attention to those with low power, and what attention they do give will be more heuristic, I propose that they are unlikely to be susceptible to emotional contagion from them.

In contrast, low-power people are likely to both orient their attention to those who have power over them and process this information in an effortful and systematic way

(Keltner et al., 2003), which constitutes great attention. Additionally, this effortful and systematic information processing may be especially likely to lead to mimicry and thus emotional contagion because this type of information processing involves considerable effort to comprehend and evaluate incoming information. Thus, systematic information

12 processing is likely to lead to a more accurate or complete perception of others emotional expressions. When they then mimic these expressions, the emotions they adopt are likely to be an accurate reflection of the target’s. Since low-power people pay great attention to those who have power over them, they are likely to be susceptible to emotional contagion from them.

Hypothesis 1: Attention paid to a target individual mediates the relationship

between one’s level of power and emotional contagion received from the target.

Specifically, those higher (lower) in power are less (more) likely to receive

emotional contagion by someone of lower (higher) power because of the lower

(greater) amount of attention paid to that target.

Nascent research (Overbeck & Park 2001, 2006; De Dreu & Van Kleef, 2004;

Vescio, Snyder, & Butz, 2003) has begun to uncover boundary conditions (i.e., moderators) to the power-attention relationship. This suggests that there may be situations where the power-attention relationship does not follow the traditionally- predicted pattern. More specifically, Vescio and colleagues (2003) found that high-power actors paid more attention to information about “subordinates” when that information would help them complete their given task. Overbeck and Park (2006) found that attention to low-power individuals increased when powerholders were told that they were a leader for an organization that employed a people-centered mentality (as opposed to a productivity-centered mentality), which stressed the importance of making workers feel engaged and included.

Drawing from these findings, I suggest that since attention is affected by goals

(Dijksterhuis & Aarts, 2010), moderators that affect either the goals that individuals have, or that affect how they see a high- or low-power counterpart in relation to goal achievement, may affect attention. These moderators may either be related to the task or related to the relationship with the relevant other. The extant research (e.g., Overbeck &

Park 2001, 2006; De Dreu & Van Kleef, 2004; Vescio et al., 2003) tends to focus on task-related moderators, but the finding from Overbeck and Park (2006), hints that relational moderators may affect attention as well.

Moreover, the extant research tends to focus only on powerholders and not on low-power individuals. For example, Overbeck and Park (2006) found that powerholders paid more attention to people when doing so helped them achieve explicit goals of the task, but were unable to draw conclusions from their data about low-power individuals.

Expanding on this research, I propose moderators that are more relational in nature and that address the attention of both low-power and high-power individuals.

Specifically, I propose two relational moderators that I suggest will decrease the attention of low-power individuals to those that hold power over them—trust based on benevolence and trust based on integrity—and thus decrease the emotional contagion received from powerholders. I also suggest two relational moderators that may increase the attention of powerholders to their low-power counterparts—trust based on ability and relational identification—and thus increase the emotional contagion received from low- power individuals.

Moderators that Decrease Attention for Low-Power Individuals

As noted above, low-power people will pay attention to those who have power over them because of their vulnerability to powerholders (Keltner et al., 2003). Part of their vulnerability relates to dependence on others for their outcomes such that low-power individuals have a prevention focus, meaning they are highly motivated to avoid negative outcomes (Keltner et al., 2003; Higgins, 1997). However, trusting a person who has power over oneself may increase one’s willingness to accept the vulnerability of being at the mercy of the powerful (e.g., Mayer & Gavin, 2005). This, in turn, may render less relevant the need to pay attention to the powerful in an attempt to predict their behaviors and intentions.

When an individual trusts a given target, that individual is willing to be vulnerable to that target based on positive expectations about the target’s behavior, independent of an ability to monitor that target. The positive expectations one has of a trusted other are based on one’s perception of the other’s trustworthiness, as comprised of three factors: ability, benevolence, and integrity (Mayer, Davis, & Schoorman, 1995). The relative importance of these three factors is context-dependent, meaning that there are situations where some factors are more important than others. When a trustor perceives an individual as having high ability, that trustor sees the individual as having skills and competencies in a domain, which allow the individual to have relative influence in that domain. A trustor perceives an individual to have high benevolence when the trustor thinks the individual has a positive orientation toward the trustor, or a general desire to do good by the trustor. Finally, the trustor perceives an individual as having high integrity

15 when the trustor believes the individual acts in accordance with a set of values or principles that the trustor deems appropriate (Mayer et al., 1995).

Low-power individuals are especially concerned with how powerholders control their outcomes, resources, and punishments, which suggests that a powerholder’s benevolence and integrity may be especially important in the decisions low-power individuals make about trusting those with power. If a low-power actor holds high trust in a powerful individual that is based on perceptions of high integrity and benevolence, they will have positive expectations that the powerful person will control their resources in a positive and fair way. Thus, low-power actors are less likely to see the powerful person as a threat and therefore may pay less attention to them.

However, in the absence of trust based on a powerholder’s benevolence and integrity, low-power individuals may engage in behaviors to try to manage the uncertainty of the situation. McAllister (1995) proposes that monitoring is one such response to low trust, such that one may monitor the untrustworthy individual in an attempt to gain more certainty that he or she will engage in desired behaviors. Monitoring entails watching or paying close attention to the untrustworthy individual, thus suggesting that low trust may engender increased attention to that person. This suggests that low- power individuals who do not perceive their high-power counterparts to have high benevolence and integrity (i.e., the trustworthiness factors most relevant to them) will be especially attentive to those with power over them.

Hypothesis 2a: Trust based on benevolence moderates the relationship between

relative power position and attention, such that for low-power individuals, greater

(lower) trust based on benevolence will result in decreased (increased) attention to

their high-power counterparts.

Hypothesis 2b: Trust based on integrity moderates the relationship between

relative power position and attention, such that for low-power individuals, greater

(lower) trust based on integrity will result in decreased (increased) attention to

their high-power counterparts.

Moderated Mediation of Trust in Benevolence and Integrity for Low-Power

Individuals

Since less powerful individuals are presumed to be paying attention to the powerful due to their goal of gaining predictability over the thoughts and actions of the powerful, they may be relatively more likely to receive emotional contagion from the powerful. However, if their attention to those who have power over them is attenuated, such as when they have a trusting relationship based on perceptions of integrity and benevolence with those individuals, the likelihood of receiving emotional contagion from the powerful may be diminished.

In contrast, not trusting the powerful based on their integrity and benevolence will leave those with low power in a state of uncertainty, during which they seek predictability and control by attending to high-power people, and through this attention, open themselves to emotional contagion. Thus, when the less powerful trust the integrity and benevolence of the powerful they may no longer be motivated to pay attention to them. In such situations, less-powerful individuals are unlikely to catch the emotions of the powerful.

Hypothesis 3: The mediated effect of power on emotional contagion through

attention will be moderated by trust based on integrity and benevolence. When

trust based on integrity and benevolence is high, lower power will be related to

lower emotional contagion through a decrease in attention.

Moderators that Increase Attention for High-Power Individuals

Keltner et al. (2003) argue that having high power activates an approach system, which causes powerful individuals to direct their attention to rewards and means for obtaining those rewards (Higgins, 1997). As a result, high-power individuals tend to construe others through a lens of self-interest, or, more specifically, they “attend to others in terms of how they enable the power holder to satisfy current goals and desires”

(Keltner et al., 2003: 272; Gruenfeld, Inesi, Magee, & Galinsky, 2008). By not paying attention to low-power people, powerholders are potentially taking a risk. This risk is in either ignoring a low-power person who may undermine their goal achievement, or in missing out on an opportunity for a low-power person to help them achieve their goals.

It would be in powerholders’ self-interest to focus on those in low power who can help them achieve goals, such as would be the case when powerholders hold high ability- based trust in a lower-power person. Since trust becomes relevant in a risky situation when individuals are vulnerable to another (Mayer et al., 1995), powerholders’ trust in low-power individuals may affect powerholders’ attention, but since high-power individuals are especially concerned with goal achievement, such as performing well on tasks, ability-based trust may be especially relevant to them. The increase in attention to low-power people in whom they hold high ability-based trust may help them mitigate the

18 risk of missing out on a resource that could help them achieve their goals. If a powerful person deems a low-power individual as having high ability, or skills and capacities that allow that person to have influence (Mayer et al., 1995), the powerholder may see the low-power person as capable of helping achieve a goal and thus increase his or her attention to the low-power person.

However, powerholders may also be taking a risk by not paying attention to low- power people in whom they hold low ability-based trust. When powerholders hold low trust based on ability in a low-power individual, they face vulnerability in paying little attention to that person, as they may be ignoring someone who will interfere with goal attainment. This vulnerability may be especially acute if there is some degree of outcome interdependence between them, which is often the case in organizations (Kanter, 1979;

Dirks & Ferrin, 2002; Mayer et al., 1995; Werbel & Henriques, 2009). As suggested by

McAllister (1995), increasing attention to a person who is not trustworthy is a way that individuals can manage the vulnerability produced by low levels of trust. Thus, given that high-power people are especially motivated to achieve rewards, they may pay more attention to a low-power person if they see that person as having low ability that may prevent them from achieving their performance-related goals.

The vulnerability that powerholders face in not paying attention to low-power individuals is more likely to arise at very high or very low levels of trust based on ability, and thus moderate levels of ability-based trust are less likely to affect the attention of powerholders. Since powerholders use their resources, such as attention, in service of pursuing rewards, such as performance, they are less likely to pay attention to low-power

19 people in whom they hold moderate levels of trust based on ability, as they are less likely to see these individuals as able to aid in or interfere with goal achievement.

Hypothesis 4: The interaction between trust based on ability and power has a

curvilinear relationship on attention, such that high-power individuals will pay

greater attention to low-power counterparts in whom they hold very low or very

high trust based on ability, but will pay less attention to low-power counterparts in

whom they hold moderate levels of trust in ability.

When individuals identify with a given target, it shapes their motivations, goals, and behaviors toward that target (Brewer & Gardner, 1996). When a high-power person relationally identifies with a low-power individual, he or she may see a relationship with that person as rewarding and thus increase the attention allocated to that low-power person. Identification is thought to be relevant in the context of organizations. Not only do individuals bring their own identities into the workplace, but the organization (and groups within it) serve as prevailing targets of identification (Ashforth, Harrison, &

Corley, 2008), and these identities inform how individuals think, feel, and act (Sluss &

Ashforth, 2008). Identification is concerned with how one defines oneself, and one’s various identities are based on one’s self-concept (Brewer & Gardner, 1996). Self- concepts can exist at three levels: the individual level (personal self-concept), the interpersonal level (relational self-concept), and the group level (collective self-concept;

Brewer & Gardner, 1996). At the individual level, one focuses on what makes oneself unique and different from others and is motivated by self-interest. Conversely, both the

20 interpersonal and collective levels exist on the predication of oneself in relation to others.

However, these two levels are differentiated by whether one’s relationship with others is based on personalized or impersonal bonds. Identification with a collective implies impersonal bonds that are derived from common identification with a group or social category, and does not require that one have personal relationships with members of that group. For example, one might identify with a particular political party. In contrast, relational identification involves personalized bonds and is derived from a relationship with a specific significant other. Relational identification with a given target, then, is the extent to which an individual considers the relationship with that target a part of their self-concept.

In the case of relational identification, individuals see themselves as fundamentally interrelated with another person, and therefore ensuring one’s own well- being means ensuring the well-being of one’s relationship with that person. Moreover, when one relationally identifies with another, one becomes concerned for the interests and well-being of the other, and is substantially motivated to maintain a positive relationship with that person (Brewer & Gardner, 1996; Andersen & Chen, 2002; Markus

& Kitayama, 1991)1,2.

1 Importantly, while relational identification and trust based on benevolence share some similarities, they are distinct constructs. More specifically, with trust based on benevolence, individuals’ uncertainty may be tempered because they feel that the trustee is concerned with the trustor’s interests and well-being. Although they may desire to maintain a positive relationship with this person, trusting in them based on benevolence does not make them motivated to do so. In contrast, when individuals relationally identify with a target, they are concerned about the interests and well-being of that target. Moreover, relationally identifying with a target means that maintaining a positive relationship with this target is a primary goal.

2 It is important to note that power and relational identification are conceptually distinct. That is to say that simply being in a high- or low-power position does not imply nor precipitate relational identification with a 21

When a high-power individual identifies with a low-power target, maintaining a positive and meaningful relationship with that target is likely to be rewarding. This relationship maintenance will become an active goal as powerholders have a goal of approaching things that are potentially rewarding. Ensuring such a relationship requires interaction with and a strong understanding of how that person is feeling, which necessitates increased attention toward that person. Therefore, high-power individuals who relationally identify with a low-power target are likely to pay greater attention to that target in order to help them maintain a positive relationship with them In sum, for high- power individuals, attention to low-power counterparts will increase as relational identification with that counterpart increases.

Hypothesis 5: Relational identification moderates the relationship between power

and attention, such that for high-power individuals, greater (lower) relational

identification with a low-power individual will increase (decrease) attention to

that individual.

Moderated Mediation of Trust Based on Ability and Relational Identification for

High-Power Individuals

Since powerholders typically lack the motivation to attend to those with less power, one may assume that they are generally unlikely to catch their emotions. In the counterpart. For instance, Sluss et al. (2012) found considerable variation in the extent to which subordinates relationally identified with their supervisors (i.e., those in a high power position) in two separate samples. Moreover, although the research on high-power individuals’ relational identification with low-power individuals is scant, LMX theory suggests that leaders (i.e., those with high power) do not have the same relationship with each of their subordinates (i.e., low-power individuals; Dansereau, Graen, & Haga, 1975; Graen, Novak, & Sommerkamp, 1982; Liden & Graen, 1980). Research suggests that LMX may predict relational identification (Carmeli, Atwater, & Levi, 2011) and thus powerholders may relationally identify with some individuals over whom they hold power, but not others. This would imply that simply holding power over an individual does not mean that one will or will not relationally identify with that individual. 22 absence of motivation to attend to low-power individuals, the powerful will be relatively unlikely to pay attention to them because they view them as unable to provide anything of value. Without paying attention to the less powerful, high-power individuals will be unlikely to receive emotional contagion from them. However, their susceptibility to emotional contagion from low-power people may increase when they have substantial motivation to direct their attention to low-power individuals. This motivation is more likely when they have high levels of relational identification with or have very low or very high trust based on ability in low-power individuals. Thus, if power has an influence on emotional contagion through attention (Hypothesis 1), this relationship is moderated by relational identification and trust based on ability for high-power individuals.

Hypothesis 6a: The mediated effect of power on emotional contagion through

attention will be moderated by trust based on ability. When trust based on ability

is very low or very high, greater power will be related to emotional contagion

through an increase in attention.

Hypothesis 6b: The mediated effect of power on emotional contagion through

attention will be moderated by relational identification. When relational

identification is high, greater power will be related to emotional contagion

through an increase in attention.

Methods Overview

In approaching the methods for the current study, it is important to note the mixed results from prior studies examining the power-emotional contagion relationship. While

Anderson et al. (2003) found results congruent with the predictions of the current study

(i.e., that low-power individuals, compared to high-power individuals, will be more susceptible to emotional contagion), both Hsee et al. (1990) and Spoor and Kelly (2009) found the opposite results. These conflicting findings may be partially explained by experimental procedures that inadvertently affected the attention of subjects, which was an unmeasured variable in the study. For instance, in Hsee et al.’s (1990) study, power was manipulated by telling subjects that those in the high-power condition had the ability to administer electrical shocks to their low-power counterparts. As a result, those in the low-power conditions showed high levels of anxiety, which may have interfered with their ability to pay attention to the powerful, as stress and anxiety have been shown to significantly impair individuals’ capability to attend to and process information (e.g.,

Mandler, 1982, 1984; Easterbrook, 1959). The power manipulation in Spoor and Kelly’s

(2009) study may also have inadvertently affected their results. Their specific manipulation of power was to have participants answer a face-valid questionnaire on leadership skills and then inform subjects that based on their answers, one has been

24 chosen to be the leader for a task they were to complete together. Leaders were then handed the decision sheet and pencil for the task. The authors suggest that their manipulation of power may not have been strong enough. Moreover, they may have confounded power with leadership competency/ability. Therefore, it might not have induced the expected patterns of attention, which led to the unexpected findings for emotional contagion.

This suggests that the manipulation of power is particularly important in examining the relationship among power, attention, and emotional contagion. In the current study, power is defined as the relative capacity to modify a target’s attitudes or feelings, such as their emotional state, due to having control over valued resources and the capacity to administer rewards and punishments (Keltner et al., 2003; Emerson,

1962). In the first study, subjects were provided with this definition to use as a basis for picking a target about whom they would answer questions. In the second study, power was manipulated by giving all control over a task to one subject. While this subject may not have been able to administer punishments, he or she had control over making decisions and accepting input from the partner.

Although many extant studies examining power use this definition, they primarily examine the psychological effects that power has on an individual’s behavior, cognitions, and affect (e.g., Galinsky, Gruenfeld, & Magee, 2003; Anderson & Berdahl, 2002). For this reason, many studies manipulate power by priming participants with a power mindset by asking them to recall a time in which they had more or less power (Galinsky et al.,

2003; Anderson & Galinsky, 2006; Chen et al., 2001; Fast, Sivanathan, Mayer, &

Galinsky, 2012, Experiments 1-4; Inesi, 2010; Tost, Gino, & Larrick, 2012). However, the current study seeks to examine how having power over another subject affects individuals in a relational context (i.e., with other people over whom they have power or who have power over them) and thus simply priming a power mindset is insufficient for the current study. Studies that have examined the relational effects of power have done so by assigning one participant to a manager role and the other to a subordinate role allegedly based on subjects’ scores on a leadership skill questionnaire (e.g., Spoor &

Kelly, 2009). However, this may confound power with ability and thus a stronger power manipulation may be created by having one participant actually control the decision- making in a situation as it creates an asymmetrical control over resources (i.e., decisions), a key element in the definition of power (Keltner et al., 2003).

In order to test my hypotheses, I have devised two studies. The first study collected survey data from individuals working in organizations. While this study is particularly suited to testing the relationship between power and attention as well as moderators for this relationship (i.e., Hypotheses 2a, 2b, 4, and 5), it is not suitable for testing emotional contagion, as contagion is a process that occurs in a momentary time period. Thus, a second study was conducted in a laboratory setting using two subjects, who will be randomly assigned to a high- or low-power role. This second study measured all parts of the model and is an appropriate methodology for testing emotional contagion between high- and low-power individuals (e.g., Spoor & Kelly, 2009).

One particular strength of employing two studies is that attention could be measured differently in each. In the survey study, attention was measured by a self-report

26 questionnaire. In the laboratory study, attention was measured two ways. First, coders rated the attentional behavior of subjects. Second, attention was measured by having subjects report the amount of attention they paid to different parts of the task and their partner.

Study 1

Methods

The purpose of Study 1 was to test the moderation hypotheses (Hypothesis 2,

Hypothesis 4, and Hypothesis 5). Data were collected from respondents using Qualtrics

Panels at three time periods, each one month apart. At time 1, N = 167, at Time 2, N =

115, and at Time 3, N = 84. The use of Qualtrics Panels ensured that repondents varied widely in job positions, and types of organizations, as well as personal and demographic characteristics, thus lending the results great external validity (see Brandon, Long,

Loraas, Mueller-Phillips, and Vansant, 2014). Moreover, Qualtrics utilizes by-invitation- only online panel recruitment, thus avoiding self-selection and professional survey takers.

Participants were instructed to choose a target co-worker with whom they have contact at least on a daily basis (on average) and answer a variety of measures with that target in mind. Half of the participants were instructed to choose a target who has power over oneself, while the other half were instructed to choose a target over whom s/he holds power.

Respondents completed the following measures in relation to their chosen target: attention, trust based on ability, benevolence, and integrity, relational identification, and power (as well as controls discussed below). All variables were collected at all three

28 times. This was done in order to gauge if any of these variables change over time.

However, as will be discussed, there was very little change over time and thus these analyses testing change were not conducted. The use of three time points that span over two months also allows for a meaningful amount of time between the measure of the independent (power) and dependent (attention) variables.

Measures

Attention. In Study 1, attention to a target was measured using two self-report scales—one that measured general attention and one that measured attention to emotions3. The scale to measure general attention was a modified version of the three items from Rothbard’s (2001) engagement scale: (1) “I focus a great deal of attention on this person.” (2) I concentrate a lot on this person.” (3) “I pay a lot of attention to this person.” Additionally, subjects answered a sliding-scale question, modified from Gardner et al. (1989): “In a typical day, please indicate how much attention you allocate to this person,” with the two ends of the scale reading almost none and almost all the time (α =

0.88) Four items adapted from Reiffe, Oosterveld, Miers, Terwogt, and Ly’s (2008) measure of attending to others’ emotions were used: (1) “When this person is in a bad mood, it catches my attention,” (2) It is important to know how this person is feeling,” (3)

“When this person is in a good mood, I am more attentive to him/her,” and (4) “I usually know how this person is feeling” (α = 0.76).

3 A confirmatory factor analysis was performed and showed poor model fit for a 1-factor model (RMSEA = 0.128; SRMR = 0.061). A 2-factor model indicates acceptable fit (RMSEA = 0.043; SRMR = 0.043), and thus I made the decision to use these as separate measures. 29

Power. Although respondents were instructed to pick a target that either they had power over or that had power over them, the purpose of this was to get variance in power, and thus a continuous measure of power was used in the analyses. The mean (2.99) and standard deviation (1.04) of this variable show that there is considerable variance., thus justifying using a continuous measure of power. Hinkin and Schriesheim’s (1989) measures of coercive and reward power were used and combined for an overall 8-item power measure (α = 0.90). These two scales were chosen, rather than the other measures

(legitimate, referent, and expert) because they fit the definition of power used in this paper: “one’s relative control over valued resources and the capacity to administer rewards and punishments” (see page 24; Keltner et al., 2003; Emerson, 1962). Since this definition encompasses both administering rewards (reward power) and administering punishments (coercive power), there is a theoretical basis to justify the combination of these two scales to measure power. The scale was prompted with “I believe this person can...” and the items are as follows: (1) “Increase my pay level,” (2) “Make work difficult for me,” (3) “Influence my getting a pay raise,” (4) Make things unpleasant here for me,”

(5) “Influence my getting a promotion,” (6) “Make being at work distasteful,” (7)

“Provide me with special benefits,” and (8) “Give me undesirable job assignments.”

Relational Identification. For Study 1, relational identification was measured using a four-item measure (α = 0.89) from Sluss, Ployhart, Cobb, and Ashforth (2012).

Items on this scale include the following: (1) “My work relationship with ______is important to how I see myself,” (2) “If someone critisized my relationship with ______, it would feel like a personal insult,” (3) “My relationship with ______is vital to the

30 kind of person I am at work,” (4) “My work relationship with ______reflects the kind of person I am.”

Trust. Trust based on benevolence, integrity, and ability were measured using

Mayer and Davis’s (1999) scales. The measure for benevolence-based trust consists of the following five items (α = 0.92): (1) “This person is very concerned about my welfare,” (2) “My needs and desires are very important to this person,” (3) “This person would not knowingly do anything to hurt me,” (4) “This person really looks out for what is important to me,” and (5) “This person will go out of his/her way to help me.” The measure of integrity-based trust consists of the following six items (α = 0.94): (1) “This person has a strong sense of justice,” (2) “I never have to wonder whether this person will stick to his/her word,” (3) “This person tries hard to be fair in dealings with others,” (4)

“This person’s actions and behaviors are not very consistent,” (5) “I like this person’s values,” and (6) “Sound principles seem to guide this person’s behavior.” The measure of ability-based trust consists of six items (α = 0.94): (1) “This person is very capable of performing his/her job,” (2) “This person is known to be successful at the things he/she tries to do,” (3) “This person has much knowledge about the work that needs to be done,”

(4) “I feel very confident about this person’s skills,” (5) “This person has specialized capabilities that can increase our performance,” and (6) “This person is well-qualified.”

Control Variables. Since theory on relational identification suggests that it develops over time and through recurring interactions (Sluss & Ashforth, 2008), length of relationship with one’s chosen target will be controlled for (measured by one question:

“How long have you known this individual?”), as well as frequency of interaction

(measured by one question: “On average, how frequently do you have interactions with this person?” Response choices: 1 = Once a week, 2 = Several times a week, 3 = Once a day, 4 = 2-3 times a day, 5 = 4-5 times a day or more). Tenure and gender were also controlled as this may affect attention (e.g., Koch & Ullman, 1985), or perceptions of power (French & Raven, 1959; Henley & LaFrance, 1984).

Whether one has a formal power relationship with their counterpart was also controlled by asking the question, “Do you/does this person hold a formal position of authority over this person/you (for example, supervisor or boss)?” This variable was dummy-coded (1 = yes, 2 = no) in analyses4.

Finally, monitoring was controlled in order to parse out this variable from attention. While monitoring may result in increased attention, it may go beyond the definition of attention as it involves “observation, examination, or recording of employee work related behaviors” (Stanton, 2000: 87). Thus, this may include attention to work but not to the person, and in that way, not lead to emotional contagion. By controlling for monitoring, I am able to better determine whether their attention is directed more toward the target versus toward the target’s work. To measure monitoring, I used 5 items adapted from de Jong and Elfring (2010): (1) “I carefully monitor this person’s behavior and actions,” (2) “I check whether this person is doing what is expected of him/her,” (3) “I watch this person while he/she works,” (4) “I monitor this person’s work progress,” and

(5) “I attend to this person in order to monitor his/her actions” (α = 0.92).

4 This variable was only measured at the third time point. 32

Statistical Analysis. A univariate regression analysis will be used to test these hypotheses. Analyses were run first with the independent variable (power) at Time 1, the moderator at Time 3, and attention (of each type) at Time 3. Testing the moderator at

Time 3 allows for a test of how concurrent perceptions of the relational variables (used as moderators) affect attention at that time. As discussed above, theory suggests that attention is affected by current (i.e., immediately-relevant) goals (Dijksterhuis & Aarts,

2010). I argue that the proposed moderators affect a focal person’s current goals and thus his or her attention. Thus, since current or relevant goals may change between measurement times, using measurements of the moderators at the same time as attention is a better test of whether the moderators affect attention.

Although the methods of this study allow for testing change across time (i.e., whether a change in power predicts a change in attention), the variables are very highly correlated across times, and thus there is not meaningful variance to predict change.

Results

The purpose of Study 1 was to examine the role of moderators on the relationship between power and attention, thus testing Hypotheses 2a, 2b, 4, and 5. Means, standard deviations, and intercorrelations for the variables in this study can be found in Table 1.

Although the negative relationship between power and attention was not hypothesized in this paper, a regression analysis showed a significant negative relationship between power at Time 1 and general attention at Time 3 (β = -0.27, p < .01) and a marginally significant relationship between attention to emotions at Time 2 (β = -

0.12, p = 0.08). Results can be seen in Table 2.

Hypothesis 2 makes predictions about moderators that will decrease attention for low-power individuals. Hypothesis 2a predicted that an individual’s perception of his or her counterpart’s benevolence would moderate the relationship between power and attention such that attention would decrease with greater levels of benevolence for low- powered individuals, but not affect the attention of high-powered individuals. A regression analysis was run to first test the interaction between power and benevolence

(at Time 3) predicting general attention (β = 0.02, p = 0.69) and attention to emotions (β

= -0.06, p = 0.29) at Time 3 (see Table 3). Neither of these analyses were significant, thus

Hypothesis 2a is not supported.

Hypothesis 2b predicted that an individual’s perception of his or her counterpart’s integrity would moderate the relationship between power and attention such that attention would decrease with greater levels of integrity for low-powered individuals, but not affect the attention of high-powered individuals. A regression analysis was run to first test the interaction between power and integrity (at Time 3) predicting general attention

(β = 0.00, p = 0.99) and attention to emotions (β = -0.05, p = 0.34) at Time 3 (see Table

4). Neither of these analyses were significant, thus Hypothesis 2b is not supported.

Hypotheses 4 and 5 made predictions about moderators that would increase the attention of high-power individuals. Hypothesis 4 predicted that power and trust based on ability would interact in a curvilinear way to predict attention such that for high-power individuals, attention would be greatest at very low or very high levels of trust based on ability and lower at moderate levels of trust based on ability. To test Hypothesis 4 I first tested whether ability at Time 3 had a curvilinear relationship with both types of attention

(at Time 3), and found a non-significant result for both general attention (β = -0.02, p =

0.72) and attention to emotions (β = -0.06, p = 0.36). To examine whether the curvilinear relationship between ability on attention depends on power, I included a linear interaction term between power and ability in addition to a second interaction term of power and trust based on ability squared. This model tests whether there is a curvilinear relationship for trust based on ability on attention that looks different for high and low levels of power

(i.e., either one concave and one convex, or one curvilinear and one linear, etc.). The results of this analysis, which can be seen in Tables 5 and 6, show a marginally significant interaction predicting general attention (β = 0.12, p = 0.08), and a significant interaction between power and ability-based trust squared predicting attention to emotions (at Time 3) (β = 0.12, p = 0.05). I plotted this interaction to see the shape of the curve for high- and low-powered individuals. As can be seen in Figure 4, for low-power people, attention increases as perceptions of ability increase, but this increase seems to level off at average levels of trust based on ability (note, overall average = 4.1). In contrast, high-power people seem to pay high attention to people with very low to or very high levels of trust based on abilility, but pay little attention to people in between.

Overall, this supports Hypothesis 4.

Hypothesis 5 predicted that relational identification would moderate the relationship between power and attention, such that attention would increase with high levels of relational identification for high-powered individuals, but not affect the attention of low-powered individuals. A regression analysis was run using the two dependent variables (general attention and attention to emotions) and the moderator (relational

35 identification) as measured at Time 3 and the independent variable (power) as measured at Time 1. The results of these analysis, displayed in Table 7, showed a non-significant interaction predicting general attention (β = -0.03, p = 0.65), but a significant interaction predicting and attention to emotions (β = -0.12, p < 0.01.

In order to interpret the form of the interaction, I used the method suggested by

Preacher, Curran, and Bauer (2006) and plotted the relationship between power (at Time

1) and relational identification (at Time 3) ratings at one standard deviation above and below the mean (see Figure 5). In order to test Hypothesis 5, I tested for a significant difference for high-power individuals between one standard deviation below and one standard deviation above the mean of relational identification using methods from

Preacher et al. (2006). This difference is not significant (slope = 0.12, t = 0.73, p > .05).

Thus, while there is a difference in attention between low and high relational identification for high-power individuals, this difference is not significant and does not fully support Hypothesis 5.

Discussion

The purpose of Study 1 was to test the role of moderators on the relationship between power and attention. The results of this study suggest times in which the normal attentional patterns of low- and high-power individuals may not hold. While I predicted that trust based on benevolence and integrity would lessen the attention that low-power people pay to powerholders, neither of these predictions was supported.

For high-power people, I predicted that their attention would increase at very low and very high levels of trust based on ability and with greater relational identification

36 with a low-power person. Analyses supported the prediction for ability-based trust but did not fully support the prediction for relational identification. The shape of the curvilinear interaction with ability-based trust showed that high-power individuals pay low attention to a low-power target in whom they hold moderate levels of ability-based trust, but greater attention to to a low-power target in whom they hold especially low or especially high levels of ability-based trust. While not predicted, low-power individuals’ attention was also affected by ability-based trust and relational identification such that attention increased with increasing levels of both, but tapered off at average levels of ability-based trust.

These relationships will be further explored in Study 2 and in addition, their effects on emotional contagion will be tested. Given that emotional contagion is a momentary process (Hatfield et al., 1994), the survey methodology used in Study 1 is not conducive to testing emotional contagion. Thus, the chief purpose of Study 2 is to test the mediation of attention on the power—emotional contagion relationship.

Study 2

Methods

Data were collected in the laboratory using a sample of students enrolled in a large introductory business class. Subjects were placed in dyads and asked to engage in a task (more on this below). Data were collected from 62 dyads, however, three people had to be removed due to technical difficulties5, thus resulting in a total N of 121. Subjects were 50.4% male, 49.6% female, an average age of 20, and 80.1% White. They were assigned either the high- or low-power role, and both subjects received a mood manipulation (positive affect or negative affect) prior to engaging in the task. This was done in order to induce differences in beginning mood. The dependent variable of interest—emotional contagion—occurs when one individual’s mood moves toward the mood of the other. Thus, in order to observe whether this occurred, it is necessary that the two individuals do not begin in the same mood. Given the outcome of interest (i.e., emotional contagion—how much does one person’s mood “move toward” the mood of the other person), only two conditions were tested: (1) High-power induced with positive affect (PA), low-power induced with negative affect (NA); (2) High-power induced with

NA, low-power induced with PA. The conditions that are unnecessary to test are those

5 The subjects were removed because they were outside of the camera view for a portion of the interaction, and thus attention could not be coded. 38 where both subjects receive the same mood manipulation6, as they will be unsuitable to examine whether one individual’s mood moves more toward the other’s.

After subjects arrived, an experimenter explained that they will first engage in a film-rating task, then sent to another room where they will engage in a discussion task with a partner. In order to enhance the cover story, subjects were told that they are engaging in two separate and unrelated studies.

The purpose of the film-rating task was to deliver the mood manipulation.

Although the film-rating data was not used in the experiment, it ensured that participants paid sufficient attention to the film and increase the cover story of the task. Subjects watched and “rated” one of two clips from films, lasting approximately 6 minutes each.

One film was a humorous clip from “Mrs. Doubtfire” (used to induce PA) and the other was a sad clip from “Steel Magnolias” (used to induce NA). This method and length of clips has been used in prior studies manipulate mood (e.g., Saavedra & Earley, 1991; Sy et al., 2005; Spoor & Kelly, 2009). After watching the clip, subjects completed an assessment of their current mood.

At this time, the two subjects were sent to a smaller, separate room where the experimenter explained that one subject has been randomly assigned to the high-power role for the next part of the study. The experimenter explained that s/he randomly determined that one subject (subject A) will be the leader, and this person has all the power and control over the discussion task such as the decisions they make and how the discussion proceeds. A similar manipulation for power involving the assignment of a

6 Specifically, the conditions that will be unnecessary to test are: (1) Both high-power and low-power are induced with PA, and (2) Both high-power and low-power induced with NA. 39 subject as the “leader” in a task and giving that person control over decision making was used by Waytz, Chou, Magee, and Galinksy(2015). Importantly, this manipulation of power is stronger than the one used by Spoor and Kelly (2009) as it involves more than simply assigning titles of “leader” and “follower” to subjects. It is important that subjects understood that the “leader” role was determined randomly, so that they did not believe it had to do with one’s personal characteristics such as ability. Moreover, it was important that the assignment of high- and low-power roles occurs prior to the partner discussion task in order to be able to study subjects’ patterns of attention and emotional contagion during the group discussion task.

Following this explanation, the experimenter gave the instructions for this task.

The task was to complete the winter survival task (Johnson & Johnson, 1994) in which they have to discuss and determine the importance of fifteen items for survival following a plane crash in Canada during mid-winter. This task was used by Spoor and Kelly

(2009) in their study examining emotional contagion.

At the conclusion of the group discussion task, subjects answered a series of questions to assess their mood, trust in their counterpart, perceptions of the counterpart’s ability, benevolence, and integrity, and how much they identify with their counterpart (as well as the control measures discussed below). The subjects also answered questions to gauge their attention to their counterpart. Following the completion of the measures, participants were debriefed and dismissed.

The advantages to this second study arise around the ability to have greater control over various factors that may confound the results. Specifically, by manipulating

40 mood prior to the task, emotional contagion can be better tested by ensuring that both participants will not be simply in a neutral mood. One disadvantage to this particular study is that it may not be suitable to testing relational identification7. Relational identification occurs when an individual partially defines oneself in terms of a given relationship (Sluss & Ashforth, 2007), which presumes that one has recurring connections or interactions with the relationship partner (Sluss & Ashforth, 2008). Thus, there may not be sufficient time or information about the individual to develop relational identification. These issues are not likely to arise when examining the effect of trust.

Although some trust theory presumes that one needs prior interactions with a target in order to determine one’s level of trust in that target, others suggest and find that individuals can form a judgment about trust in another almost instantaneously, with little to no prior knowledge of the target (e.g., Kramer & Lewicki, 2010; Lewicki, Tomlinson,

& Gillespie, 2006; Meyerson, Weick, & Kramer, 1996; Malhotra & Murninghan, 2002;

Pillutla, Malhotra, & Murnighan, 2003). These shortcomings regarding relational identification are addressed by the measures and methods of Study 1 by using respondents with longer-term, established, relationships.

Measures

Emotional Contagion. In order to measure emotional contagion, subjects needed to have a measure of emotions prior, during, and after the discussion task. Measures were collected in two ways: self-report and other-observed. First, subjects themselves completed a measure immediately before and immediately following the task. This

7 A proxy for relational identification will be used in Study 2. This proxy is discussed in detail below in the Measures section. 41 measure was the PANAS (Watson, Clark, & Tellegen, 1988), which consists of twenty emotional descriptors (e.g., enthusiastic, inspired, ashamed, irritated). To complete the measure, subjects indicated to what extent they were currently feeling each, from 1, “very slightly or not at all,” to 5, “extremely.” PA and NA are determined by adding the values given to respective positive and negative descriptors. In previous studies examining emotional contagion (e.g., Hsee et al., 1990; Spoor & Kelly, 2009; Barsade, 2000), shorter, more generic scales were used to measure affect, but the use of the 20-item

PANAS allows for a more complete picture of subjects’ moods.

Second, emotions were also measured using coding by three outside observers.

This coding procedure followed Barsade (2002). Each interaction was videotaped and subsequently coded by coders blind to the purpose of the study. The three coders were extensively trained and reached an inter-rater reliability of 83% by the end of the training, which included independently coding videos and then coming together to discuss rationale behind their ratings. Coders then independently rated one-third of the remaining videos8. They rated the affect of participants by watching participants’ facial expressions, body language, and verbal tone to rate their affect on the eleven items used by Barsade

(2002): sad, pleasant, unhappy, interested, pessimistic, happy, gloomy, lethargic, optimistic, depressed, and warm on a scale of 1, “very slightly or not at all,” to 7,

“extremely.” This scale was chosen for the coded emotions because it keeps with the rating methods used by prior emotional contagion studies (i.e., Barsade, 2002) as well as

8 In order to ensure that coders remained consistent once they independently rated videos, I conducted ANOVAS on the average ratings for each item by coder. None of the analyses were significant, indicating no significant differences by coder. 42 studies examining observers’ accuracy in perceiving others’ emotions. More specifically, while evidence shows individuals are aptly able to report their own emotions measured by the PANAS, no research exists examining observers’ accuracy in measuring these emotions. Some research suggests that individuals may be able to accurately perceive certain emotions better than others (Montagne, Kessels, De Haan, & Perrett, 2007) and thus, using a scale that has been previously used to measure others’ ratings of individuals’ emotions is more appropriate. Due to very low levels of variance on two of the NA items—sad and depressed—these items were removed and not used in analyses.

Videos were split into thirds, and coders rated affect separately for each third of the video. Coders rated one subject throughout the duration of the interaction so that they could use the subject’s prior emotional display as a basis for change over the interaction.

The average length of the interactions was 9 minutes and 3 seconds, so the average length of the thirds was 3 minutes and 1 second, which is similar to the coding strategy followed by Barsade (2002).

Emotional contagion was calculated by subtracting the absolute value of the difference between the subject’s ending affect and the subject’s initial affect from the absolute value of the difference between subject’s and partner’s initial affect. A detailed description and examples regarding this calculation can be found in Appendix A. This was calculated using the subject’s self-report measures and the partner’s observer-rated measures because it matches the situation that subjects encoutered: subjects interact with their partner and observe their moods. According to emotional contagion theory (Hatfield et al., 1994), subjects should observer and mimic partners’ emotional expressions and

43 then come to feel similar emotions through afferent feedback. Thus, using observer-rated measures of the partner’s emotions mirrors subjects’ experience and using self-report measures of their own emotions tells us about their emotional experience that may not be outwardly expressed.

While a variable calculated partially by a difference score is not ideal, this method is appropriate as emotional contagion is defined as a change in a focal person’s emotions following an interaction with another individual (Hatfield et al., 1994).

Moderators. The moderators of trust in ability, benevolence, and integrity were measured following the task using the same scales as in Study 1. However, considering the nature of Study 2, the scale used to measure relational identification in Study 1 does not seem appropriate, as it seems to ask about a relationship that is long-term and ongoing. Thus, a proxy was be used: similarity. Based on theory on social identification

(e.g., Tajfel, 1981), when individuals identify with a target, they are likely to see themselves as similar to that target. Thus, the following four-item scale from Back and

Lips (1998) was used to measure identification: (1) “I feel that I am similar to this person.” (2) “I identify with this person,” (3) “I feel a sense of sameness with this person,” and (4) “This person reminds me of myself” (α = 0.91).

Theoretically, relational identification is predicated on feeling “more connected and less differentiated” from another (Markus & Kitayama, 1991: 217) and on perceiving a sense of “oneness” with another. The above measure (Back & Lips, 1998) captures these aspects and thus represents an appropriate proxy for relational identification.

Attention. Attention was measured by coder ratings as well as self-report. For coded ratings, attention is operationalized by gauging subjects’ direction of gaze.

Importantly, it is recognized that eye movements are a reliable and clear manifestation of where individuals allocate their attention in a scene (Henderson, 2003). It is appropriate to use coders to rate subjects’ attention as the notion that individuals are adept at assessing the direction of others’ eye movements (Von Grünau & Anston, 1995), and thus attention, is supported both by theoretical models (Baron-Cohen 1994, 1995a,

1995b; Langton, 2000) and evidence from human subjects (e.g., Langton, 2000). In fact,

Baron-Cohen (1994, 1995a, 1995b) and Langton (2000) suggest that because attention is directed toward that which is most informative (see discussion above in “Attention &

Power” section), being able to assess the direction of others’ attention is an adaptive, and thus innate, behavior as it facilitates detection and response to informative cues in the environment. A coder rated subjects’ attention to their counterparts using two metrics.

The first is the total amount of time focused on the counterpart assessed by when a subject was looking at the face and/or body of the counterpart, as a proportion of the total length of the interaction (heretoforth labeled “attention percentage”). A coder used a stopwatch to record this time and stopped and started it over the duration of the interaction. The second metric is mean gaze duration (heretoforth labeled as “average gaze length”), which provides information about the intensity of a subjects’ attention on another. Similar metrics have been used in other studies using eye gaze as an operationalization of attention (e.g., Foulsham et al., 2010; Birmingham, Bischof, &

Kingstone, 2008).

Attention was also measured using a self-report measure. Subjects were asked what proportion of their attention they allocated to different targets by having them divide a circle into pieces that represent how much they paid attention to each stimulus.

Specifically, the instructions read: “This circle represents 100% of your attention during this previous task. Please move the pieces of the circle to represent the percentage of time you spent paying attention to each of the following.” The options included the other person in the group, the task, and ‘other.’

Control Variables. Five variables were used as controls: gender, prior knowledge of partner, susceptibility to emotional contagion, and task and outcome interdependence.

For the first two controls, subjects were asked to report their gender (dummy-coded such that 1 = male and 2 = female) and whether they knew their counterpart prior to meeting them that day (dummy-coded such that 0 = no and 1 = yes), as these variables may affect attention (e.g., Koch & Ullman, 1985) or how much one identifies with another (e.g.,

Sluss et al., 2012). Susceptibility to emotional contagion was measured using five items from Doherty’s (1997) scale. Although the original scale consists of fifteen items (three items for each emotion—sadness, happiness, love, fear, and anger), subjects answered the highest loading item for each emotion. Thus the five questions they will answer are: (1)

“If someone I’m talking to begins to cry, I get teary-eyed” (sadness), (2) “When I look into the eyes of the one I love, my mind is filled with thoughts of romance” (love), (3) “I tense when overhearing an angry quarrel” (anger), (4) “Being around happy people fills my mind with happy thoughts” (happiness), (5) “I notice myself getting tense when I’m around people who are stressed out” (fear) (α = 0.54).

Feelings of both task and outcome interdependence were controlled for, as the extent to which one perceives interdependence with another can affect one’s attention

(Tjosvold, 1985; Lammers, Stoker, & Stapel, 2009). Perceived interdependence was measured by adapting van der Vegt, Emans, and van de Vliert’s (1998) scales of both task and outcome interdependence. The following items were used to measure outcome interdependence: (1) “The things that this person wants to accomplish and the things I want to accomplish are compatible,” (2) “It is advantageous for me when this person succeeds,” (3) “My concerns and those of this person are clashing” (α = 0.64). The following items were used to measure task interdependence: (1) “To what extent did you depend on this person for information and advice?” (2) “To what extent did you depend on the presence, help and support of this person?” and (3) “To what extent did you depend on this person for doing the task well?” (α = 0.84).

Statistical Approach. For all analyses, power was dummy coded such that low power = 0 and high power = 1. Thus, in testing a regression, for which the equation is Y = i + bX, I tested for a difference in Y as a function of values on X (which denote low versus high power). For example, when X = 0 (low power), Y will be the value of the dependent variable, i will represent the mean of Y for low power subjects, and b will represent the change coefficient, or the difference on Y between low and high power subjects

(Edwards, 2012).

Although some laboratory studies do not use control variables, it is important to include them in analyses in the current study as both attentional and emotional contagion processes may be influenced by demographic and task variables (Fousham et al., 2010;

Dijksterhuis & Aarts, 2010; Barsade, 2002). Thus, following the methods of Barsade

(2002), I control for the demographic and task variables listed above in the analyses.

However, there is some debate in the literature about the use of control variables in analyses of laboratory studies. For example, Spoor and Kelly (2009) and Anderson et al.

(2003) do not use controls in their analyses. Therefore, I also include results with no controls in footnotes.

Results

The purpose of Study 2 was to examine the whole model, but chiefly to test the relationship between power and emotional contagion as mediated by attention. Means, standard deviations, and correlations for all variables can be found in Table 8. Since attention was measured in two ways (i.e., self-report and coded), it is important to discuss how these measures relate to each other in order to understand if there are differences in how individuals thought their attention was allocated (self-report) compared to coded attention. Since emotional contagion is thought to occur outside of conscious awareness

(Hatfield et al., 1994), the measurement of attention in both ways will help clarify whether the type of attention used in emotional contagion is within and/or outside of awareness. The two coded measures of attention—attention percentage and average gaze length—correlate highly at 0.79 (p < .01), but self-reported attention does not significantly correlate with attention percentage (0.01) or average gaze length (-0.01).

These low correlations between the coded and self-report measures of attention suggest that individuals may not be accurate at reporting their own attention allocation.

Manipulation checks. Both power and initial mood (positive and negative affect) were manipulated in this study. There was a significant difference in perceptions of power in the partner discussion task (F(1, 118) = 108.73, p < .01), thus demonstrating a successful power manipulation. Participants who watched the sad movie clip reported significantly more initial negative affect than those who watched the happy movie clip

(F(1, 118) = 60.67, p < .01), and participants who watched the happy movie clip reported significantly more initial positive affect than participants who watched the sad movie clip

(F(1, 118) = 52.57, p < .01). Thus, initial positive and negative affect were successfully manipulated. However, this manipulation only worked with self-report measures of emotion. There was no significant difference in coded initial PA (F(1, 118) = 0.03, p =

0.86) or coded initial NA (F(1, 118) = 0.02, p = 0.88).

The mediating role of attention on the power—emotional contagion relationship. To test Hypothesis 1, that attention mediates the relationship between power and emotional contagion, I first tested the direct effect of power on emotional contagion by regressing emotional contagion on power (dummy-coded). This analysis was marginally significant for PA contagion (β = -0.20, p = 0.07), but not significant for

NA contagion (β = 0.03, p = 0.72)9 as shown in Table 9.

To test the indirect effects, a regression analysis was run to test the relationship between power (dummy-coded) and attention, the results of which support a negative relationship between power and attention as measured by percentage (β = -0.11, p < .01) and average gaze length (β = -1.26, p < .01). However, there was no significant

9 These results are similar when removing all controls: no significant relationship between power and PA contagion (β = -0.17, p = 0.11) or NA contagion (β = -0.03, p = 0.78). 49 relationship between power and self-reported attention (β = 1.14, p = 0.66; see Table

10)10. In order to test the relationship between attention and emotional contagion, PA contagion and NA contagion were independently regressed onto the three measures of attention. There were no significant relationships between any of the measures of attention and PA contagion (attention percentage: β = -0.64, p = 0.18; average gaze length: β = -0.03, p = 0.42; self-report attention: β = 0.00, p = 0.36)11. There were no significant relationships between average gaze length (β = 0.01, p = 0.82) or self-report attention (β = 0.00, p = 0.88)12 and NA contagion, but there was a marginally significant positive relationship between attention percentage and NA contagion (β = 0.84, p = 0.07; see Table 11).

To explore whether there is support for mediation for the three measures of attention, I conducted 5,000 Monte Carlo intervals (Mackinnon, Lockwood, & Williams,

2004; Bauer, Preacher, & Gil, 2006). For PA contagion, there was no significant mediation for any of the measures of attention (attention percentage: -0.03 to 0.20; average gaze length: -0.05 to 0.15, self-report attention: -0.03 to 0.02). For NA contagion, there was marginally significant mediation for attention percentage (90% CI: -0.19 to -

10 The results of this analysis are similar with no controls: there is a negative relationship between power and attention percentage (β = -0.11, p < .01) and between power and average gaze length (β = -1.19, p < .01), but no significant relationship between power and self-reported attention (β = 0.65, p = 0.79). 11 There were also no significant results for these analyses when removing all controls (attention percentage: β = -0.10, p = 0.82; average gaze length: β = 0.00, p = 0.91; self-report attention: β = 0.00, p = 0.36). 12 When removing controls from these analyses, there are still non-significant relationships between average gaze length (β = -0.01, p = 0.70) or self-report attention (β = 0.00, p = 0.74) and NA contagion, however the relationship between attention percentage and NA contagion becomes non-significant (β = 0.55, p = 0.19). 50

0.01)13, but no significant mediation for average gaze length (-0.11 to 0.08) or self-report attention (-0.02 to 0.03).

Moderating effects on the power-attention relationship. Hypothesis 2a predicted that power and trust based on perceptions of the partner’s benevolence would interact to predict attention. The interaction between power and benevolence predicting attention is not significant for any of the measures of attention (self-reported attention: β

= 2.69, p = 0.34; average gaze length: β = 0.36, p = 0.21; percentage of attention: β =

0.01, p = 0.58; see Table 12)14. Hypothesis 2b predicted that power and trust based on perceptions of the partner’s integrity would interact to predict attention. The interaction between power and integrity predicting attention is not significant for any of the measures of attention (self-reported attention: β = 5.00, p = 0.19; average gaze length: β

= -0.18, p = 0.64; percentage of attention: β = -0.04, p = 0.24; see Table 13)15. Thus,

Hypotheses 2a and 2b were not supported with these data.

Hypothesis 4 predicted a curvilinear interaction between power and ability. I first tested whether trust based on perceptions of the partner’s ability had a curvilinear relationship with the three measures of attention, and found a non-significant result for all three measures of attention (Second column in Tables 14, 15, and 16; self-reported attention: β = 1.34, p = 0.13; average gaze length: β = 0.05, p = 0.55; attention

13 This confidence interval is a 90% confidence interval, which does not include 0, suggesting a marginal significant mediation relationship. This marginal significance is likely due to the marginal significance of the indirect effect of attention on NA emotional contagion. A 95% confidence interval does include 0 (- 0.22 to 0.00), meaning this mediation has a p-value between 0.05 and 0.10. 14 These analyses are also non-significant when removing all controls (percentage of attention: β = 0.01, p = 0.63;average gaze length: β = 0.43, p = 0.14; self-reported attention: β = 2.65, p = 0.33). 15 These analyses remain non-significant when removing all controls (percentage of attention: β = -0.04, p = 0.14; average gaze length: β = -0.10, p = 0.78; self-reported attention: β = -0.02, p = 0.24). 51 percentage: β = 0.00, p = 0.94)16,17. To examine whether the curvilinear relationship between trust based on ability on attention depends on power, I included a linear interaction term between power and trust based on ability in addition to a second interaction term of power and ability squared. This model tests whether there is a curvilinear relationship for trust based on ability on attention that looks different for high and low levels of power (i.e., either one concave and one convex, or one curvilinear and one linear, etc.). The results of this analysis, which can be seen in column 3 in Tables 14,

15, and 16, show no significant interaction predicting self-reported attention (β = 0.99, p

= 0.59), average gaze length (β = -0.21, p = 0.25), or attention percentage (β = -0.02, p =

0.26)18. Overall, these results do not support Hypothesis 4 with this sample.

Hypothesis 5 predicted an interaction between power and relational identification to predict attention. To test this hypothesis, a regression analysis was run in which power was interacted with the proxy variable of similarity. The results of this analysis are not significant in predicting self-reported attention (β = 0.46, p = 0.84), average gaze length

(β = 0.21 p = 0.35), or percentage of attention (β = 0.01, p = 0.59). Results of these analyses can be seen in Table 17. Thus, Hypothesis 5 was not supported in this sample.

16 These analyses remain non-significant when removing all controls (percentage of attention: β = -0.02, p = 0.24; average gaze length: β = -0.19 p = 0.26; self-reported attention: β = 0.94, p = 0.56). 17 Although not hypothesized, I also tested whether there was a linear interaction between power and ability predicting attention. This analysis was not significant for any of the attention measures (self-reported attention: β = 1.99, p = 0.48; average gaze length: β = -0.06, p = 0.84; attention percentage: β = -0.03, p = 0.19). 18 These analyses remain non-significant when removing all controls (percentage of attention: β = 0.01, p = 0.46; average gaze length: β = 0.29 p = 0.21; self-reported attention: β = 0.00, p = 0.99). 52

Moderated Mediation Hypotheses. As none of the moderated relationships were significant, it is unsuitable to test for moderated mediation. Thus, Hypotheses 3 and 6 were not supported.

Discussion

The data from Study 2 were especially suited to test the mediation of attention on the power—emotional contagion relationship. Analyses supported this mediation, but only for NA contagion and only with the observer-rated attention measure of attention percentage. Moreover, this mediation was marginally significant, likely due to the marginally significant indirect effect of attention on NA contagion. The fact that self- report attention was not a significant mediator and did not have a significant relationship with any of the PA or NA contagion measures supports the idea that the type of attention required for emotional contagion is non-conscious and outside of awareness (Hatfield et al., 1994).

Study 2 also tested the moderators to the power—attention relationship, none of which were significant. As suggested above, since the moderators are indicators of positive relationships, they may take more time to develop and thus initial snap judgments of these variables were not enough to affect attention in this short amount of time.

General Discussion

Theoretical Contributions

With this dissertation, I integrate two streams of power theory: one that examines how power affects the behavior and goals of those who do or do not have it (i.e., the

Power Approach Theory; Keltner, et al., 2003) and another that examines how those who have power influence those who do not (i.e., the bases of power; French & Raven, 1959).

More specifically, power approach theory suggests that the respective goals of high- and low-power individuals induce patterns of attention such that high-power people tend to pay little attention to those with low power, and low-power individuals pay great attention to those that hold power over them. The latter stream of power theory contends that those with power hold a greater potential to change the beliefs, attitudes, or behaviors of others (Bruins, 1999; Bugental, Blue, & Cruzcosa, 1989; Buss, Gomes, Higgins, &

Lauterbach, 1987; Kipnis, Schmidt, & Wilkinson, 1980; Yukl & Falbe, 1990). The results of my dissertation suggest that the goals and behaviors induced by one’s relative level of power compared to another may help explain the patterns of influence between them, thus linking these two streams of power research. Specifically, I found that attention (i.e., one of the behaviors affected by goals) mediates the relationship between power and emotional contagion, a form of influence.

This finding helps unpack the relationship between power and emotional contagion by examining the role of attention, which has been discussed in theory on emotional contagion (Hatfield et al., 1994) but has remained untested. By measuring attention and testing its effect on emotional contagion, the inconsistent results from studies examining the relationship between power and emotional contagion (Hsee et al.,

1990; Spoor & Kelly, 2009; Anderson et al., 2003) may be disentagled. The results from this dissertation suggest that because low-power individuals pay great attention to the powerful, they may be especially susceptible to emotional contagion from them. In contrast, because high-power individuals are unlikely to attend to those with low power they are unlikely to receive emotional contagion from those low-power individuals.

While results supported this hypothesis, they did so only for emotional contagion with negative emotions. This finding supports results from prior research on emotional contagion that people are more likely to catch negative than positive emotions (Spoor &

Kelley, 2009; Bartel & Saavedra, 2000; Bakker, Le Blanc, & Schaufeli, 2005). This may be because negative emotions are theorized to be of more informational value to individuals by communicating potential threats (Rolls, 1992; Spoor & Kelly, 2004). This suggests that emotional contagion may be functional (Anderson et al., 2003), in that the greater emotional similarity that results may benefit partners by helping coordinate reponses to environmental threats or opportunities (Festinger, 1951; Hatfield et al., 1994), or by fostering the well-being of a relationship (LaFrance & Ickes, 1981; Locke &

Horowitz, 1990; Rosenblatt & Greenberg, 1991; Schachter, 1959).

My dissertation also contributes to power approach theory (Keltner, et al., 2003) and a nascent body of literature that examines boundary conditions to the power— attention relationship (e.g., Overbeck & Park, 2001, 2006). I hypothesized specific situations in which low-power individuals may pay relatively little attention to the powerful, and in which high-power individuals may pay more attention to those with low power. While I hypothesized that low-power individuals’ attention to powerholders would decrease when they held greater levels of trust in them based on benevolence and integrity, these hypotheses were not supported. I also hypothesized two situations in which high-power individuals’ attention to those over whom they hold power would increase—when they relationally identify with them and when they hold either very low or very high trust in them based on ability. Results supported the hypothesis about ability-based trust, and while results showed an increase in high-power individuals’ attention with high levels of relational identification, this increase was not significant.

The results for high-power individuals support Overbeck and Park’s (2006) assertion that powerholders use their attention flexibly, meaning they are able to use attention as a resource to achieve their current goals. Conversely, the non-significant results for low-power individuals may suggest that these individuals do not hold this same capacity to use their attention flexibly. Instead, even if they hold high trust in powerholders based on perceptions of their benevolence and integrity, this may not temper their primary goal of avoiding negative outcomes.

Research thus far has identified task-specific goals as moderators to the power— attention relationship, but the findings from my dissertation support additional variables

56 that are relational in nature as moderators. This suggests that certain features of the relationship that high-power individuals have with people over whom they hold power may shape their behaviors toward those individuals, such as attention, by affecting whether they see these individuals as having the potential to provide them with rewards.

Prior research has found that power increases the likelihood of objectifying people

(Gruenfeld et al., 2008) such that they approach useful targets “regardless of the value of their other human qualities” (111), painting a negative picture of powerholders’ approach to others. However, the finding from my study that relational identification with a low- power target increases attention toward that target suggests that powerholders are not simply blind to these “human qualities.” Instead, maintaining a positive relationship with that person based on personalized bonds with him or her, may be seen as rewarding to powerholders. In this case, they would no longer be objectifying the low-power target, and instead would be concerned with that individual’s needs, interests, and experiences.

Consequently, this attention toward the low-power individual may result in increased influence from him or her.

Finally, the results which suggest that the attention of powerholders can be changed such that they pay greater attention to low-power individuals, but the attention of low-power individuals may be more resistent to change, may have implications for the likelihood of receiving influence from others. If the attention that high-power individuals pay to low-power others can be changed, and this attention determines whether one is likely to catch emotional contagion from one another, the power—emotional contagion relationship may be subject to the variables that impact attention. These results shed light

57 on to times in which high-power people may be more likely to catch the emotions of low- power counterparts. Specifically, given that increased attention to targets increases negative emotional contagion from that target (as found in Study 2), it may be that high- power people are likely to catch low-power individuals’ negative emotions when they relationally identify with them, or when they have very low or very high trust in them based on ability.

Methodological Contributions

This dissertation also makes a methodological contribution to the measurement of attention. Within the extant literature on power and attention, attention has been measured by examining individuals’ use of stereotype-consistent or stereotype- inconsistent information in forming impressions of others (e.g., selecting candidates for a job position) (Overbeck & Park, 2006). This methodology is valuable in that it has begun to illustrate how high- and low-power individuals differ in attending to others. However, the methods of the current studies may illuminate the way high- and low-power individuals attend to others in a real-life context in which they are interacting with one another, which may be more applicable to work contexts. In Study 1, attention was measured using a self-report scale which tapped both general attention to a counterpart and attention to that counterpart’s emotions within the workplace environment. Study 2 used a behavioral measure for attention, by having a coder rate the amount of attention, measured by percentge of attention participants paid to their partners during their interaction with them, as well as the intensity of their attention, gauged by the average length of their gazes toward the partner. It is interesting to note that within Study 2, coded

58 attention had a very low correlation with participants’ self-reported attention19, and only coded attention had a significant relationship with emotional contagion. This suggests that the type of attention required for emotional contagion may be somewhat subconscious or not easily recalled by individuals.

Additionally, this dissertation used a novel way of measuring and calculating emotional contagion. Prior studies (Spoor & Kelly, 2009; Barsade, 2002; Sy, et al., 2005) have calculated emotional contagion by looking at the difference between only the subject’s beginning and ending emotions. However, the method used in the current study better fits the definition of emotional contagion as an individual’s move toward another’s emotions (Hatfield et al., 1994) by taking into account the specific starting point of the partner, thus gauging how close subjects moved toward their partner. Moreover, controlling for the partner’s change in emotions over the interaction paints a more complete picture of the emotional information the subject was exposed to from the partner, to which they had the potential to pay attention and thus become emotionally contaiged.

Finally, this dissertation used a novel way of calculating emotional contagion by using the subject’s self-report emotions and the partner’s displayed emotions. Prior studies on emotional contagion have measured subjects’ emotions using both self-report and observer-rated methods. However, the current study builds upon the methods of prior studies by using observer-rated emotions as a measure of the emotional expressions of partners. This method is directly linked to the three-step process outlined in emotional

19 The correlation between attention percentage and self-reported attention is 0.01. The correlation between average gaze length and self-reported attention is -0.01. 59 contagion theory whereby individuals pay attention to a target, mimic that target’s emotional expressions, and come to experience similar emotions through afferent feedback (Hatfield et al., 1994). Using observer-rated emotions mirrors the situation experienced by the subject, as it measures the emotional expressions to which the subject was exposed and thus would mimic (if paying attention). Importantly, research has found that even though emotional displays often match felt or experienced emotions, this is not always the case (Hochschild, 1979; Rafaeli & Sutton, 1991; Grandey, 2003). Thus, a method that calculates emotional contagion using the partner’s self-reported emotions may not accurately mirror the partner’s emotional displays to which the subject was exposed. This is supported by the low correlations between self- and other-ratings of emotions found in Study 2.

Limitations

There were several limitations to the current study. First, the conclusions that can be drawn about emotional contagion come from data collected in a laboratory study with student subjects in which power relationships were manipulated. While power was successfully manipulated such that subjects in the high-power condition perceived themselves to have had significantly more power in the task than low-power subjects, the nature of the power differentials that these subjects experienced may be different than that experienced by respondents in Study 1. Subjects in Study 2 knew that their participation in this study was limited to one hour, and thus any power one subject held over the other was time-limited. While this study was especially suited for testing the relationship of attention on emotional contagion, the power relationships may not mirror

60 those that are experienced in organizations, which tend to be longer-term. Additionally, power was not manipulated by affecting one subject’s capacity to administer punishments and rewards to another subject, thus not perfectly matching the definition of power used in Study 1. However, high-power subjects were given complete control over the discussion task and all decision-making that occurred in that task. Assuming that all subjects desire to have an input in the task, this implies that high-power subjects were able to “reward” low-power subjects by allowing them to be involved in the decision making or “punish” them by not taking their input into consideration.

Moreover, these relationships in the laboratory were quite transient, as the average time spent with their partners was about nine minutes. While the results of this study suggest that this relatively short length of time is enough for emotional contagion processes to occur, this may not have been long enough to form reliable judgments of the variables proposed to be moderators (relational identification, trust based on ability, benevolence, and integrity) in a way that would have affected goals and thus attention.

More specifically, trust based on ability was proposed to change the way powerholders viewed low-power individuals such that they would see them as either able to hinder or help goal achievement, which in this case, was successfully ordering the survival items.

While subjects may have been able to make a quick judgment about whether they trust their partner based on their ability, this may not have had a substantial effect on how they see the partner in terms of hindering or aiding in goal achievement. Instead, it may be that ability-based trust may change based on specific experiences with targets. Thus, with a longer relationship history, these judgments will be made on more information under a

61 variety of contexts. It is also a possibility that the proxy variable used in place of relational identification may not have induced attention in a similar way as relational identification. Specifically, although similarity and relational identification are similar constructs, individuals who are similar to another may desire to have a relationship with the target, but maintaining a positive relationship with that target is not a primary goal, as is the case with relational identification. In contrast, subjects in Study 1 had ongoing power differential relationships (a mean of 7.32 years) with the individuals about whom they answered questions, and this likely provided subjects with ample time to make judgments of these variables.

Due to the non-significant results of the interaction hypotheses in Study 2, I was unable to test the full moderated mediation model. However, if the results from Study 1 may be combined with the results of Study 2, I can conclude that powerholders may pay greater attention to low-power people, and thus may be more likely to catch their negative emotions, when they relationally identify with them or hold high or low levels of ability-based trust in them.

A concern may also arise around the use of self-reported attention in Study 1 in light of the very low correlations between self- and observer-rated attention in Study 2.

Moreover, the results from Study 2 suggest that the type of attention necessary for emotional contagion to occur is relatively subconscious. Thus, the interactive relationships found to predict attention in Study 1 may not reflect the type of attention required for emotional contagion. However, it is important to note that different self- report measures were used in each study. In Study 2, subjects were asked to estimate the

62 percentage of attention they paid to their partner in relation to the task. In Study 1, respondents rated how much they agreed to statements on measures asking about whether they pay attention to a target in general, and to that target’s emotions. Moreover, the moderated relationships found in Study 1 were only significant in predicting attention to emotions, which may be more conducive to emotional contagion than general attention.

As Study 2 subjects were only asked to report general attention, it is unclear whether the results may have been different if they completed the measures used in Study 1.

Practical Implications

Practical implications may arise primarily about high-power actors, as results were supported for these individuals, but not low-power individuals. The findings from these studies may be particularly important to understand in the context of leadership, where power differences exist between leaders and their subordinates. As emotional contagion is a form of influence over others (Schacter, 1959; Levy & Nail, 1993), understanding the role that paying attention to others plays in emotional contagion may help clarify the way influence functions within leadership relationships. This study supports the line of research suggesting that influence follows power and leadership

(Cartwright, 1965; French & Raven, 1959; Kipnis, 1972, 1976; Lewin, 1951), but this influence seems to be limited to negative emotions in the context of emotional contagion.

While much of the leadership literature centers around how leaders can have a positive impact on followers (e.g., Dirks & Ferrin, 2002; George, 1996), results from these studies suggest that leaders may need to be especially careful in the negative emotions they display, as followers may be likely to catch them.

From the perspective of the leader, paying attention to one’s subordinates is presumably a requisite for effective leadership (e.g., Wilemon & Cicero, 1970; Yukl,

1989), but according to theory on power, by having power over their subordinates, leaders are unlikely to attend to them. The results from this study suggest that leaders may pay more attention to their followers when they see them as able to provide them with something rewarding, or if they see them as an obstacle to achieving something of reward. This line of reasoning may help explain some of the prior research on the abuse of power such as through sexual objectification/sexual harassment (e.g., Bargh,

Raymond, Pryor, & Strack, 1995). Specifically, powerholders may see an attractive subordinate as potentially able to provide them with rewards, which would increase their attention (although unwanted) toward that subordinate.

By paying more attention to their subordinates, leaders may be better able to address their concerns, making them more effective leaders (Wilemon & Cicero, 1970;

Yukl, 1989) by opening them up to influence from subordinates. There are times when this may constitute effective leadership, such as making a decision based on input given by a subordinate (thus receiving influence from them). However, there may be negative outcomes. Specifically, as this study finds that paying attention to others increases one’s likelihood of adopting their negative moods. While this mood convergence may be functional and strengthen the relationship (LaFrance & Ickes, 1981; Locke & Horowitz,

1990; Rosenblatt & Greenberg, 1991; Schachter, 1951), it may create a cycle of negativity in the relationship.

References

Aarts, H., Dijksterhuis, A., & De Vries, P. (2001). On the psychology of drinking: Being thirsty and perceptually ready. British Journal of Psychology, 92, 631-642.

Andersen, S.M., & Chen, S. (2002). The relational self: An interpersonal social-cognitive theory. Psychological Review, 109(4), 619-645.

Adelmann, P.K., & Zajonc, R.B. (1989). Facial efference and the experience of emotion. Annual Review of Psychology, 40(1), 249-280.

Anderson, C., & Berdahl, J.L. (2002). The experience of power: Examining the effects of power on approach and inhibition tendencies. Journal of Personality and Social Psychology, 83, 1362-1377.

Anderson, C., & Galinsky, A.D. (2006). Power, optimism, & risk-taking. European Journal of Social Psychology, 36, 511-536.

Anderson, C., Keltner, D., & John, O.P. (2003). Emotional convergence between people over time. Journal of Personality and Social Psychology, 84(5), 1054-1068.

Antes, J.R. (1974). The time course of picture viewing. Journal of Experimental Psychology, 103(1), 62-70.

Asch, S.E. (1952). Effect of a group pressure on the modification and distortion of judgments. In G.E. Swanson, T.M. Newcomb, & E.L. Hartley (Eds.), Readings in social psychlogy (2nd ed., pp 2-11). New York: Holt.

Asch, S.E. (1955). Opinions and social pressure. Scientific American, 193, 33-35.

Ashforth, B.E., Harrison, S.H., & Corley, K.G. (2008). Identification in organizations: An examination of four fundamental questions. Journal of Management, 34(3), 325-374.

Back, S., & Lips, H.M. (1998). Child sexual abuse: Victim age, victim gender and

observer gender as factors contributing to attributions of responsibility. Child Abuse & Neglect, 22(12), 1239-1252.

Bakker, A.B., Le Blanc, P.M., & Schaufeli, W.B. (2005). Burnout contagion among intensive care nurses. Journal of Advanced Nursing, 51(3), 276-287.

Bargh, J.A., Chen, M., & Burrows, L. (1996). The automaticity of social behavior: Direct effects of trait concept and stereotype activation on action. Journal of Personality and Social Psychology, 71, 230-244.

Bargh, J.A., Raymond, P., Pryor, J.B., & Strack, F. (1995). Attractiveness of the underling: An automatic power→sex association and its consequences for sexual harassment and aggression. Journal of Personality and Social Psychology, 68(5), 768-781.

Baron-Cohen, S. (1994). How to build a baby that can read minds: Cognitive mechanisms in mind-reading. Cahiers de Psychologie Cognitive, 13, 513-552.

Baron-Cohen, S. (1995a). Mindblindness: An essay on autism and theory of mind. Cambridge, MA: MIT Press/Bradford Books.

Baron-Cohen, S. (1995b). The Eye Direction Detector (EDD) and the Shared Attention Mechanism (SAM): Two cases for evolutionary psychology. In C. Moore & P. Dunham (Eds.), Joint attention: Its origins and role in development (pp. 41-59). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Barsade, S.G. (2002). The ripple effect: Emotional contagion and its influence on group behavior. Administrative Science Quarterly, 47, 644-675.

Bartel, C.A., & Saavedra, R. (2000). The collective construction of work group moods. Administrative Science Quarterly, 45(2), 197-231.

Bauer, D.J., Preacher, K.J., & Gil, K.M. (2006). Conceptualizing and testing random indirect effects and moderated mediation in multilevel models: New procedures and recommendations. Psychological Methods, 11(2), 142-163.

Birmingham, E., Bischof, W.F., & Kingstone, A. (2008). Gaze selection in complex social scenes. Visual Cognition, 16, 341-355.

Brandon, D.M., Long, J.H., Loraas, T.M., Mueller-Phillips, J., & Vansant, B. (2014). Online instrument devliery and participant recruitment services: Emerging opportunities for behavioral accounting research. Behavioral Research in Accounting, 26(1), 1-23.

Brewer, M.B., & Gardner, W. (1996). Who is this “we”? Levels of collective identity and self representation. Journal of Personality and Social Psychology, 71(1), 83-93.

Bugental, D.B., Blue, J., & Cruzcosa, M. (1989). Perceived control over caregiving outcomes: Implications for child abuse. Deveopmental Psychology, 25(4), 532- 539.

Buswell, G.T. (1935). How people look at pictures: A study of the psychology and perception in art. Oxford: University of Chicago Press.

Buss, D.M., Gomes, M., Higgins, D.S., & Lauterbach, K. (1987). Tactics of manipulation. Journal of Personality and Social Psychology, 52(6), 1219-1229.

Cacioppo, J.T., Petty, R.E., Losch, M.E., & Kim, H.S. (1986). Electromyographic activity over facial muscle regions can differentiate the valence and intensity of affective reactions. Journal of Personality and Social Psychology, 50(2), 260-268.

Carmeli, A., Atwater, L., & Levi, A. (2011). How leadership enhances employees’ knowledge sharing: The intervening roles of relational and organizational identification. The Journal of Technology Transfer, 36(3), 257-274.

Cartwright, D. (1965). Influence, Leadership, Control. In J.G. March (Ed.) Handbook of Organizations (pp. 1-47). Chicago: Rand McNally.

Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39(5), 752-766.

Chance, M.R.A. (1967). Attention structure as the basis of primate rank order. Man, 2, 503-518.

Chen, S., Lee-Chai, A.Y., & Bargh, J.A. (2001). Relationship orientation as a moderator of the effects of social power. Journal of Personality and Social Psychology, 80(2), 173-187.

Choleris, E., & Kavaliers, M. (1999). Social learning in animals: Sex differences and neurobiological analysis. Pharmacology Biochemistry and Behavior, 64(4), 767- 776.

Cialdini, R.B. (2009). Influence: Science and practice (Vol. 4). Boston: Pearson Education.

Cialdini, R.B., & Goldstein, N.J. (2004). Social Influence: Compliance and Conformity. Annual Review of Psychology, 55, 591-621. 67

Corbetta, M. & Shulman, G.L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nature Reviews Neuroscience, 3(3), 201-215.

Dansereau, F., Graen, G., & Haga, W.J. (1975). A vertical dyad linkage approach to leadership in formal organizations. Organizational Behavior and Human Performance, 13, 46-78.

De Dreu, C.K.W., & Carnevale, P.J. (2003). Motivational bases of information processing and strategy in conflict and negotiation. In M.P. Zanna (Ed.), Advances in experimental social psychology (Vol. 35, pp. 235-291). New York: Academic Press.

De Dreu, C.K.W., & Van Kleef, G.A. (2004). The influence of power on the information search, impression formation, and demands in negotiation. Journal of Experimental Social Psychology, 40(3), 303-319. de Jong, B.A., & Elfring, T. (2010). How does trust affect the performance of ongoing teams? The mediating role of relexivity, monitoring, and effort. Academy of Management Journal, 53(3), 535-549.

Dehaene, S., Changeux, J.P., Naccache, L., & Sackur, J., & Sergent, C. (2006). Conscious, preconscious, and subliminal processing: A testable taxonomy. Trends in Cognitive Sciences, 10(5), 204-211.

Desimone, R. & Duncan, J. (1995). Neural mechanisms of selective visual attention. Annual Review of Neuroscience, 18(1), 193-222.

Dijksterhuis, A., & Aarts, H. (2010). Goals, attention, and (un)consciousness. Annual Review of Psychology, 61, 467-490.

Dijksterhuis, A. & Bargh, J.A. (2001). The perception-behavior expressway: Automatic effects of social perception on social behavior. In M.P. Zanna (Ed.), Advances in Experimental Social Psychology (pp. 1-40). San Diego, CA: Academic.

Dirks, K.T., & Ferrin, D.L. (2002). Trust in leadership: Meta-analytic findings and implications for research and practice. Journal of Applied Psychology, 87(4), 611- 628.

Doherty, R.W. (1997). The emotional contagion scale: A measure of individual differences. Journal of Nonverbal Behavior, 21(2), 131-154.

Easterbrook, J.A. (1959). The effect of emotion on cue-utilization and the organization of behavior. Psychological Review, 66, 183-201. 68

Ebenbach, D.H., & Keltner, D. (1998). Power, emotion, and judgmental accuracy in social conflict: Motivating the cognitive miser. Basic and Applied Social Psychology, 20, 7-21.

Edwards, M. (2012). Lecture on Multiple Regression with Categorical Independent Variables. Personal Collection of M. Edwards, The Ohio State University, Columbus, OH.

Emerson, R.M. (1962). Power-dependence relations. American Sociological Review, 27(1), 31-41.

Emory, R. (1988) Social geometry and cohesion in three primate species. In M.R.A. Chance (Ed.), Social fabrics of the mind (pp. 47-60). East Sussex, England: Erlbaum.

Erber, R., & Fiske, S.T. (1984). Outcome dependency and attention to inconsistent information. Journal of personality and social psychology, 47(4), 709-726.

Fast, N.J., Sivanathan, N., Mayer, N.D., & Galinsky, A.D. (2012). Power and overconfident decision-making. Organizational Behavior and Human Decision Processes, 117(2), 249-260.

Festinger,

Fiske, S.T. (1993). Controlling other people: The impact of power on stereotyping. American Psychologist, 48, 621-628.

Fiske, S.T., & Dépret, E. (1996). Control, interdependence and power: Understanding social cognition in its social context. European Review of Social Psychology, 7(1), 31-61.

Foulsham, T., Cheng, J.T., Tracy, J.L, Henrich, J, & Kingstone, A. (2010). Gaze allocation in a dynamic situation: Effects of social status and speaking. Cognition, 117, 319-331.

French, J.R., & Raven, B. (1959) The bases of social power. In D. Cartwright (ed.), Studies in Social Power (pp. 150-167). Ann Arbor, MI: Institute for Social Research.

Galinsky, A.D., Gruenfeld, D.H., & Magee, J.C. (2003). From power to action. Journal of Personality and Social Psychology, 85(3), 453-466.

Gardner, D.G., Dunham, R.B., Cummings, L.L., & Pierce, J.L. (1989). Focus of attention 69

at work: Construct definition and empirical validation. Journal of Occupational Psychology, 62, 61-77.

George, J.M. (1996). Group affective tone. In M. West (ed.), Handbook of Work Group Psychology (pp. 107-116). Sussex, UK: Wiley.

Goodwin, S.A., Gubin, A., Fiske, S.T., & Yzerbyt, V.Y. (2003). Power can bias impression processes: Stereotyping subordinates by default and by design. Group Processes & Intergroup Relations, 3(3), 227-256.

Graen, G., Novak, M.A., & Sommerkamp, P. (1982). The effects of leader-member exchange and job design on productivity and satisfaction: Testing a dual attachment model. Organizational Behavior and Human Performance, 30, 109- 131.

Grandey, A.A. (2003). When “the show must go on”: Surface acting and deep acting as determinants of emotional exhaustion and peer-rated service delivery. Academy of Management Journal, 46(1), 86-96.

Gruenfeld, D.H., Inesi, M.E., Magee, J.C., & Galinsky, A.D. (2008). Power and the objectification of social targets. Journal of Personality and Social Psychology, 95, 111-127.

Guinote, A. (2007). Power and goal pursuit. Personality and Social Psychology Bulletin, 33(8), 1076-1087.

Hall, J.A. (1979). Gender, gender roles, and nonverbal communication skills. In R. Rosenthal (Ed.) Skill in nonverbal communication: Individual Differeces (pp. 47- 80) Cambridge: Cambridge University Press.

Hatfield, E., Cacioppo, J.T., & Rapson, R.L. (1992). Primitive emotional contagion. In M.S. Clark (Ed.), Review of personality and social psychology: Vol. 14. Emotions and social behavior (pp. 151-177). Newbury Park, CA: Sage.

Hatfield, E., Cacioppo, J.T., & Rapson, R.L. (1994). Emotional Contagion. New York: Cambridge University Press.

Henderson, J.M. (2003). Human gaze control during real-world scene perception. Trends in Cognitive Sciences, 7(11), 498-504.

Henderson, J.H. & Hollingworth, A. (1999). The role of fixation position in detecting scene changes across saccades. Psychological Science, 10(5), 438-443.

Henley, N.M., & LaFrance, M. (1984). Gender as culture: Difference and dominance in 70

nonverbal behavior. In A. Wolfgang (Ed.) Nonverbal behavior: Perspectives, applications, intercultural insights (pp. 351-371). Ashland, OH: Hogrefe & Huber Publishers.

Higgins, E.T., (1997). Beyond pleasure and pain. American Psychologist, 52(12), 1280 1300.

Higgins, E.T. (1999). Promotion and prevention as motivational duality: Implications for evaluative processes. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology (pp. 503-525). New York: Guilford Press.

Hinkin, T.R., & Schreisheim, C.A. (1989). Development and application of new scales to measure the French and Raven (1959) bases of social power. Journal of Applied Psychology, 74(4), 561-567.

Hochschild, A.R. (1979). Emotion work, feeling rules, and social structure. American Journal of Sociology, 85(3), 551-575.

Hsee, C.K., Hatfield, E., Carlson, J.G., & Chemtob, C. (1990). The effect of power on susceptibility to emotional contagion. Cognition and Emotion, 4(4), 327-340.

Inesi, M.E. (2010). Power and loss aversion. Organizational Behavior and Human Decision Processes, 112(1), 58-69.

Johnson, D.W., & Johnson, F.P. (1994). Joining together: Group theory and group skills (5th ed.). Boston: Allyn & Bacon.

Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.

Kanter, R.M. (1979). Power failure in management circuits. Harvard Business Review, 57(4), 65-75.

Keltner, D., Gruenfeld, D.H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110(2), 265-284.

Keltner, D., & Robinson, R.J. (1997). Defending the status quo: Power and bias in social conflict. Personality and Social Psychology Bulletin, 23(10), 1066-1077.

Kipnis, D. (1972). Does power corrupt? Journal of Personality and Social Psychology, 24, 33-41.

Kipnis, D. (1976). The powerholders. Chicago: University of Chicago Press.

Kipnis, D. Schmidt, S.M., & Wilkinson, I. (1980). Intraorganizational influence tactics: 71

Explorations in getting one’s way. Journal of Applied Psychology, 65(4), 440- 452.

Koch, C., & Ullman, S. (1985). Shifts in selective visual attention: Towards the underlying neural circuitry. Human Neurobiology, 4(4), 219-227.

Kramer, R.M., & Lewicki, R. (2010). Repairing and enhancing trust: Approaches to educing organizational trust deficits. Academy of Management Annals, 4, 245- 277.

Kruglanski, A.W. (1989). Lay emistemics and human knowledge: Cognitive and motivational bases. New York: Plenum.

Kruglanski, A.W., & Ajzen, I. (1983). Bias and error in human judgment. European Journal of Social Psychology, 13, 1-44.

LaFrance, M. & Ickes, W. (1981). Posture mirroring and interactional involvement: Sex and sex typing effects. Nonverbal Behavior, 5, 139-154.

Lammers, J., Stoker, J.I., & Stapel, D.A. (2009). Differentiating social and personal power: Opposite effects on stereotyping, but parallel effects on behavioral approach tendencies. Psychological Science, 20(12), 1543-1548.

Land, M.F., & Hayhoe, M. (2001). In what ways do eye movements contribute to everyday activies?. Vision Research, 41(25), 3559-3565.

Langton, S.R.H. (2000). The mutual influence of gaze and head orientation in the analysis of social attention direction. The Quarterly Journal of Experimental Psychology Section A: Human Experimental Psychology, 53(3), 825-845.

Levy, D.A., & Nail, P.R. (1993). Contagion: A theoretical and empirical review and reconceptualization. Genetic, Social, and General Psychology Monographs, 119, 233-284.

Lewicki, R.J., Tomlinson, E.C., & Gillespie, N. (2006). Models of interpersonal trust development: Theoretical approaches, empirical evidence, and future directions. Journal of Management, 32, 991-1022.

Lewin, K. (1951). Problems of research in social psychology. In D. Cartwright (Ed.), Field theory in social science: Selected theoretical papers (pp.155-169). Westport, CT: Greenwood.

Liden, R.C., & Graen, G. (1980). Generalizability of the vertical dyad linkage model of leadership. Academy of Management Journal, 23, 451-465. 72

Locke, K.D., & Horowitz, L.M. (1990). Satisfaction in interpersonal interactions as a function of similarity in level of dysphoria. Journal of Personality and Social Psychology, 58, 823-831.

Loftus, G.R., & Mackworth, N.H. (1978). Cognitive determinants of fixation location during picture viewing. Journal of Experimental Psychology: Human Perception and Performance, 4(4), 565-572.

MacKinnon, D.P., Lockwood, C.M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate behavioral research, 39(1), 99-128.

Mackworth, N.H., & Morandi, A.J. (1967). The gaze selects informative details within pictures. Perception & Psychophysics, 2(11), 547-552.

Malhotra, D., & Murnighan, J.K. (2002). The effects of contracts on interpersonal trust. Administrative Science Quarterly, 47(3), 534-559.

Mandler, G. (1982). Stress and thought processes. In L. Goldberger & S. Breznitz (Eds.), Handbook of Stress. New York: Free Press/Macmillan.

Mandler, G. (1984). Mind and body: Psychology of emotion and stress. New York: Norton.

March, J.G. (1955). An Introduction to the theory and measurement of influence. American Political Science Review, 49(2), 431-451.

Markus, H.R. & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98(2), 224-253.

Mayer, R.C. & Davis, J.H. (1999). The effect of the performance appraisal system on trust for management: A field quasi-experiment. Journal of Applied Psychology, 84, 123-136.

Mayer, R.C., Davis, J.H., & Schoorman, F.D. (1995). An integrative model of organizational trust. Academy of Management Review, 20(3), 709-734.

Mayer, R.C., & Gavin, M.B. (2005). Trust in management and performance: Who minds the shop while the employees watch the boss? Academy of Management Journal, 48(5), 874-888.

McAllister, D.J. (1995). Affect- and cognition-based trust as foundations for

interpersonal cooperation in organizations. Academy of Management Journal, 38(1), 24-59.

Meyerson, D., Weick, K.E., & Kramer, R.M. (1996). Swift trust and temporary groups. In R.M. Kramer and T.R. Tyler (Eds.), Trust in organizations: Frontiers of theory and research (pp. 166-195). Thousand Oaks, CA: Sage.

Montagne, B. Kessels, R.P.C., De Haan, E.H.F., & Perrett, D.I. (2007). The emotion recognition task: A paradigm to measure the perception of facial emotional expressions at different intensities. Perceptual and Motor Skills, 104, 589-598.

Neuberg, S.L., & Fiske, S.T. (1987). Motivational influences on impression formation: Outcome dependency, accuracy-driven attention, and individuating processes. Journal of Personality and Social Psychology, 53(3), 431-444.

Neumann, R. & Strack, F. (2000). “Mood contagion”: The automatic transfer of mood between persons. Journal of Personality and Social Psychology, 79(2), 211-223.

Overbeck, J.R., & Park, B. (2001). When power does not corrupt: Supervior individuation processes among powerful perceivers. Journal of Personality and Social Psychology, 81(4), 549-565.

Overbeck, J.R., & Park, B. (2006). Powerful perceivers, powerless objects: Flexibility of powerholders’ social attention. Organizational Behavior and Human Decision Processes, 99, 227-243.

Pillutla, M., Malhotra, D., & Murnighan, J.K. (2003). Attributions of trust and the calculus of reciprocity. Journal of Experimntal Social Psychology, 39, 448-455.

Posner, M.I. (1980). Orienting of attention. Quarterly Journal of Experimental Psychology, 32(1), 3-25.

Preacher, K.J., Curran, P.J., & Bauer, D.J. (2006). Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31(4), 437-448.

Rafaeli, A. & Sutton, R.I. (1991). Emotional contrast strategies as means of social influence: Lessons from criminal interrogators and bill collectors. Academy of Management Journal, 34(3), 749-775.

Range, F., Horn, L., Bugnyar, T., Gajdon, G.K., Huber, L. (2009). Social attention in keas, dogs and human children. Animal Cognition, 12(1), 181-192.

Rieffe, C., Oosterveld, P., Miers, A.C., Terwogt, M.M., & Ly, V. (2008). Emotion 74

awareness and internalising symptoms in children and adolescents: The Emotion Awareness Questionnaire revised. Personality and Individual Differences, 45(8), 756-761.

Rosenblatt, A. & Greenberg, J. (1988). Depression and interpersonal attraction: The role of perceived similarity. Journal of Personality and Social Psychology, 55, 112- 119.

Rothbard, N.P. (2001). Enriching or depleting? The dynamics of engagement in work and family roles. Administrative Science Quarterly, 46(4), 655-684.

Saavedra, R., & Earley, P.C. (1991). Choice of task and goal under conditions of general and specific affective inducement. Motivation and Emotion, 15(1), 45-65.

Schacter, S. (1959). The Psychology of Affiliation. Stanford, CA: Stanford University Press.

Simon, H.A. (1957). Models of Man. New York: Wiley.

Sluss, D.M., & Ashforth, B.E. (2007). Relational identity and identification: Defining ourselves through work relationships. Academy of Management Review, 32(1), 9- 32.

Sluss, D.M., & Ashforth, B.E. (2008). How relational and organizational identification convergence: Processes and conditions. Organization Science, 19(6), 807-823.

Sluss, D.M., Ployhart, R.E., Cobb, M.G., & Ashforth, B.E. (2012). Generalizing newcomers’ relational and organizational identifications: Processes and prototypicality. Academy of Management Journal, 55(4), 949-975.

Snodgrass, S.E. (1985). Women’s intuition: The effect of subordinate role on interpersonal sensitivity. Journal of Personality and Social Psychology, 49, 146- 155.

Spoor, J.R., & Kelly, J.R. (2009). Mood convergence in dyads: Effects of valence and leadership. Social Influence, 4(4), 282-297.

Stanton, J.M. (2000). Reactions to employee performance monitoring: Framework, review, and research directions. Human Performance, 13(1), 85-113.

Steele, C.M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69(5), 797-811.

Sy, T., Côté, S., & Saavedra, R. (2005). The contagious leader: Impact of the leader’s mood on the mood of group members, group affective tone, and group processes. Journal of Applied Psychology, 90(2), 295-305.

Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. New York: Cambridge University Press.

Tjosvold, D. (1985). Power and social context in superior-subordinate interaction. Organizational Behavior and Human Decision Processes, 35, 281-293.

Tost, L.P., Gino, F., & Larrick, R.P. (2012). Power, competitiveness, and advice taking: Why the powerful don’t listen. Organizational Behavior and Human Decision Processes, 117(1), 53-65. van der Vegt, G., Emans, B., & van de Vliert, E. (1998). Motivating effects of task and outcome interdependence in work teams. Group and Organization Management, 23(2), 124-143.

Van Kleef, G.A., De Dreu, C.K.W., & Manstead, A.S.R. (2004). The interpersonal effects of emotions in negotiations: A motivated information processing approach. Journal of Personality and Social Psychology, 87(4), 510-528.

Vescio, T.K., Snyder, M., & Butz, D. (2003). Power in stereotypically masculine domains: A social influence strategy x stereotype match model. Journal of Personality and Social Psychology, 85, 1062-1078.

Von Grünau, M. & Anston, C. (1995). The detection of gaze: A stare in the crown effect. Perception, 24, 1297-1313.

Watson, D., Clark, L.A., & Tellegen, A. (1988). Development and Validation of Brief Measures of Positive and Negative Affect: The PANAS Scales. Journal of Personality and Social Psychology, 54(6), 1063-1070.

Waytz, A., Chou, E.Y., Magee, J.C., & Galinsky, A.D. (2015). Not so lonely at the top: The relationship between power and loneliness. Organizational Behavior and Human Decision Processes, 130, 69-78.

Weber, M. (1947). The theory of social and economic organization (A.M. Henderson & T. Parsons, Trans.). New York: Oxford University Press.

Werbel, J.D., & Henriques, P.L. (2009). Different views of trust and relational leadership: supervisor and subordinate perspectives. Journal of Managerial Psychology, 24(8), 780-796.

Yarbus, A.L. (1967). Eye movements and vision. New York: Plenum.

Yukl, G., & Falbe, C.M. (1990). Influence tactics and objectives in upward, downward, and lateral influence attempts. Journal of Applied Psychology, 75(2), 132-140.

Appendix A: Calculation of the Emotional Contagion Measure

Emotional contagion is defined as one person’s mood moving toward another’s

(Hatfield et al., 1994). Prior studies on emotional contagion have measured contagion as the difference between a subject’s initial (time 1) and ending (time 2) emotions (Barsade,

2002; Hsee et al., 1990; Spoor & Kelly, 2009). However, these measures may be unsuitable for the current study. Prior studies (Barsade, 2002; Hsee et al., 1990) have used a constant target (either recorded person or confederate highly trained to display particular affect), and thus the target’s affect remains consistent across subjects.

However, the current study uses other subjects as the target and thus, it is necessary to account for the target’s initial mood in order to see whether the subject’s mood moves toward the target’s. Indeed, although the manipulation check indicated a successful initial mood manipulation, there was still some variability in initial mood within mood condition, whereas there was no variation in target’s initial mood in prior studies.

In the context of the current study, emotional contagion is defined as the extent to which a subject’s initial mood (PAt1) changes between the initial time and the ending time (PAt2), and this move is towards the partner’s initial mood (Partner PAt1). Thus, simply accounting for change between time 1 and time 2 is insufficient to calculate

78 emotional contagion, as it may change but move away from the partner’s mood.

Accordingly, the equation for PA contagion is the following:

PAt1− PartnerPAt1 − PAt2 − PartnerPAt1

Please refer to Table 18 for examples of emotional contagion scores. As can be seen with Subject1,€ the subject’s ending PA (PAt2) moves away from the partner’s initial

PA (Partner PAt1) and receives a PA Contagion score of -0.4. As can be seen with

Subject 2, the subject’s ending PA moves away from the initial PA (PAt1) towards the partner’s initial PA (Partner PAt1), and thus receives a PA Contagion score of 0.7. In comparison, Subject 3’s ending PA moves away from the initial PA (PAt1) towards the partner’s initial PA (Partner PAt1), but to a much lesser magnitude that Subject 2, and thus receives a lower PA Contagion score of 0.2.

It is important to note that in these analyses, I also control for the change that occurs in the partner by creating a measure of their change from the beginning to the end.

This measure is their end score minus their beginning score, which shows their change over the interaction. A positive value for this would indicate that the partner increased in

PA or NA, while a negative value would indicate a decrease in PA or NA.

Appendix B: Tables and Figures Appendix B: Tables and Figures

1 2 3 4 5 6 7 8 9 10 11 12 13 1. Gender (T1) -- 2. Tenure (T1) -0.14 -- 3. Length of Relationship (T1) -0.10 0.59** -- 4. Formal power (T3) 0.25* 0.04 -0.03 -- 5. Interaction Frequency (T3) 0.01 0.13 0.21† -0.17 -- 6. Power (T1) -0.08 0.24* 0.10 0.39** 0.00 -- 7. Rel Ident (T1) -0.02 0.10 0.14 -0.08 0.24* -0.07 -- 8. Rel Ident (T2) -0.05 0.28** 0.22* -0.05 0.25* 0.02 0.81** -- 9. Rel Ident (T3) 0.01 0.19† 0.24* -0.04 0.23* 0.05 0.78** 0.73** -- 10. Benev (T1) 0.07 0.23* 0.21† -0.04 0.30** -0.04 0.76** 0.74** 0.63** --

80 11. Benev (T2) 0.03 0.23* 0.25* -0.10 0.31** 0.00 0.77** 0.79** 0.65** 0.87** --

12. Benev (T3) 0.05 0.25* 0.18† -0.02 0.30** 0.07 0.73** 0.74** 0.68** 0.82** 0.87** -- 13. Integrity (T1) 0.02 0.31** 0.26** -0.01 0.14 0.14 0.72** 0.72** 0.68** 0.82** 0.78** 0.74** -- 14. Integrity (T2) -0.08 0.29** 0.18 -0.04 0.24* 0.18† 0.70** 0.79** 0.70** 0.72** 0.78** 0.80** 0.84** 15. Integrity (T3) 0.01 0.25* 0.23* 0.02 0.20† 0.18 0.74** 0.72** 0.73** 0.71** 0.77** 0.84** 0.85** 16. Ability (T1) -0.02 0.26* 0.26* -0.10 0.24* 0.01 0.63** 0.59** 0.59** 0.70** 0.61** 0.57** 0.80** 17. Ability (T2) -0.08 0.23* 0.24* -0.16 0.34** -0.03 0.53** 0.59** 0.54** 0.63** 0.65** 0.64** 0.71** 18. Ability (T3) -0.03 0.25* 0.23* -0.10 0.37** 0.11 0.59** 0.58** 0.61** 0.61** 0.62** 0.68** 0.73** 19. Monitoring (T3) -0.03 -0.09 -0.05 -0.12 -0.01 -0.17 0.36** 0.25* 0.26* 0.16 0.22* 0.18 0.12 20. General Attention (T2) 0.05 0.06 0.13 -0.16 0.18 -0.29** 0.64** 0.62** 0.59** 0.57** 0.62** 0.56** 0.44** 21. General Attention (T3) 0.01 0.05 0.10 -0.21† 0.16 -0.38** 0.66** 0.55** 0.50** 0.54** 0.58** 0.58** 0.42** 22. Attention to Emotions (T2) 0.16 0.14 0.23* -0.17 0.30** -0.31** 0.49** 0.56** 0.45** 0.42** 0.51** 0.44** 0.32** 23. Attention to Emotions (T3) 0.11 0.22 0.25* -0.11 0.28* -0.20† 0.54** 0.51** 0.51** 0.51** 0.59** 0.58** 0.41** Mean 1.48 8.42 7.32 1.35 5.26 2.98 3.40 3.32 3.24 3.55 3.59 3.55 3.86 0.92 Standard Dev 0.53 7.13 6.96 0.47 1.54 1.11 0.98 0.97 0.97 0.88 0.90 0.89

Table 1: Means, standard deviations, and intercorrelations for variables in Study 1. (Continued)

Table 1, continued Table 1, continued

14 15 16 17 18 19 20 21 22 23 1. Gender (T1) 2. Tenure (T1) 3. Length of Relationship (T1) 4. Formal power (T3) 5. Interaction Frequency (T3) 6. Power (T1) 7. Rel Ident (T1) 8. Rel Ident (T2) 9. Rel Ident (T3) 10. Benev (T1) 81 11. Benev (T2)

12. Benev (T3) 13. Integrity (T1) 14. Integrity (T2) -- 15. Integrity (T3) 0.88** -- 16. Ability (T1) 0.68** 0.69** -- 17. Ability (T2) 0.76** 0.72** 0.84** -- 18. Ability (T3) 0.76** 0.81** 0.84** 0.88** -- 19. Monitoring (T3) 0.06 0.09 0.08 0.04 -0.02 -- 20. General Attention (T2) 0.42** 0.41** 0.46** 0.43** 0.38** 0.55** -- 21. General Attention (T3) 0.38** 0.43** 0.41** 0.43** 0.39** 0.58** 0.77** -- 22. Attention to Emotions (T2) 0.33** 0.35** 0.36** 0.34** 0.32** 0.40** 0.76** 0.61** -- 23. Attention to Emotions (T3) 0.44** 0.47** 0.37** 0.38** 0.40** 0.40** 0.62** 0.70** 0.70** -- Mean 3.90 3.88 4.01 4.10 4.12 2.57 3.28 3.12 3.67 3.57 Standard Dev 0.82 0.91 0.88 0.85 0.90 1.03 1.00 0.96 0.72 0.73

DV: General Attention DV: Attention to Emotions Independent Variables Intercept 1.94** 2.04** Gender (T1) 0.04 0.20 Tenure (T1) 0.02 0.02† Length of Relationship (T1) 0.01 0.01 Formal Power Relationship (T3) -0.02 0.01 Interaction Frequency (T3) 0.09 0.11* Monitoring (T3) 0.50** 0.28** Power (T1) -0.27** -0.12† F = 9.54** F = 5.98** N = 84 N = 84 Table 2: Results for regression analyses testing the main effect of power on attention.

DV: General Attention (T3) DV: Attention to Emotions (T3)

Independent Variables Model 1 Model 2 Model 1 Model 2

Intercept 1.01* 1.22† 1.42** 0.91

Gender (T1) -0.05 -0.04 0.14 0.14

Tenure (T1) 0.00 0.00 0.01 0.01

Length of Relationship (T1) 0.01 0.01 0.01 0.01

Formal Power Relationship (T3) -0.01 -0.01 0.01 0.00

Interaction Frequency (T3) 0.00 0.00 0.05 0.05

Monitoring (T3) 0.40** 0.40** 0.21** 0.21**

Power (T1) -0.30** -0.38† -0.14* 0.07

Benevolence (T3) 0.56** 0.50** 0.38** 0.53**

Power*Benevolence (T3) 0.02 -0.06

F = 20.59** F = 18.12** F = 10.40** F = 9.39**

N = 84 N = 84 N = 84 N = 84 Table 3: Results for regression analyses testing the interaction between power and benevolence predicting attention

DV: General Attention (T3) DV: Attention to Emotions (T3)

Independent Variables Model 1 Model 2 Model 1 Model 2

Intercept 0.93† 0.92 1.35** 0.87

Gender (T1) -0.01 -0.01 0.16 0.17

Tenure (T1) 0.01 0.01 0.02 0.02

Length of Relationship (T1) 0.00 0.00 0.01 0.01

Formal Power Relationship (T3) 0.00 0.00 0.02 -0.01

Interaction Frequency (T3) 0.05 0.05 0.08† 0.08†

Monitoring (T3) 0.44** 0.44** 0.24** 0.24**

Power (T1) -0.33** -0.33 -0.17** 0.04

Integrity (T3) 0.44** 0.45* 0.31** 0.44**

Power*Integrity (T3) 0.00 -0.05

F = 15.47** F = 13.56** F = 8.72** F = 7.85**

N = 84 N = 84 N = 84 N = 84 Table 4: Results for regression analyses testing the interaction between power and integrity predicting attention.

DV: General Attention (T3)

Independent Variables Model 1 Model 2 Model 3

Intercept 0.68 0.40 -2.30 Gender (T1) 0.02 0.02 0.04 Tenure (T1) 0.01 0.01 0.01 Length of Relationship (T1) 0.00 0.00 -0.01 Formal Power Relationship (T1) 0.06 0.06 0.11 Interaction Frequency (T3) 0.00 0.01 0.02 Monitoring (T3) 0.50** 0.50** 0.51** Power (T1) -0.31** -0.31** 0.63 Ability (T3) 0.45** 0.62 2.71* Ability2 -0.02 -0.36* Power*Ability -0.74 Power*Ability2 0.12† F = 14.78** F = 13.00** F = 11.54** N = 84 N = 84 N = 84 Table 5: Results for analyses testing the curvilinear moderation between power and ability predicting general attention.

DV: Attention to Emotions (T3)

Independent Variables Model 1 Model 2 Model 3

Intercept 1.30** 0.64 -2.34

Gender (T1) 0.19 0.18 0.19

Tenure (T1) 0.02 0.01 0.02†

Length of Relationship (T1) 0.01 0.01 0.00 Formal Power Relationship (T1) 0.06 0.06 0.07

Interaction Frequency (T3) 0.06 0.07 0.08

Monitoring (T3) 0.28** 0.28** 0.29**

Power (T1) -0.15* -0.16* 1.07

Ability (T3) 0.27** 0.68 2.73**

Ability2 -0.06 -0.37*

Power*Ability -0.81†

Power*Ability2 0.12*

F = 7.37** F = 6.64** F = 5.98** N = 84 N = 84 N = 84 Table 6: Results for analyses testing the curvilinear moderation between power and ability predicting attention to emotions.

DV: General Attention (T3) DV: Attention to Emotions (T3)

Independent Variables Model 1 Model 2 Model 1 Model 2

Intercept 1.44** 1.20† 1.69** 0.70

Gender (T1) 0.01 0.01 0.17 0.19

Tenure (T1) 0.01 0.01 0.02† 0.02†

Length of Relationship (T1) 0.00 0.00 0.01 0.01

Formal Power Relationship (T3) -0.01 -0.03 0.01 -0.05

Interaction Frequency (T3) 0.04 0.04 0.08† 0.08†

Monitoring (T3) 0.40** 0.39** 0.21** 0.18**

Power (T1) -0.29** -0.20 -0.14* 0.25

Relational Identification (T3) 0.37** 0.45* 0.26** 0.60**

Power*Relational Identification -0.03 -0.12**

F = 13.08** F = 11.53** F = 7.86** F = 7.87**

N = 84 N = 84 N = 84 N = 84 Table 7: Results for regression analyses testing the interaction between power and relational identification predicting attention.

1 2 3 4 5 6 7 8 9 10 11 1. Gender -- 2. Knew Partner 0.04 -- 3. Susceptibility to EC 0.26** -0.30** -- 4. Engagement -0.25** -0.01 -0.07 -- 5. Interdependence -0.11 0.05 -0.03 0.05 -- 6. Task Interdependence -0.01 0.08 0.08 0.06 0.23* -- 7. Power Condition 0.09 -0.04 -0.04 0.09 0.02 0.12 -- 8. Similarity 0.02 0.07 0.13 0.11 0.38** 0.29** 0.08 -- 9. Ability -0.02 -0.12 0.22* 0.14 0.24** 0.36** 0.05 0.46** -- 10. Benevolence -0.30** -0.20* 0.19* 0.30** 0.19* 0.18* -0.08 0.25** 0.58** -- 88

11. Integrity -0.09 0.12 0.03 0.26** 0.31** 0.22* -0.07 0.49** 0.50** 0.49** -- 12. Partner Change in PA -0.12 -0.12 -0.04 -0.08 0.07 0.09 -0.04 0.00 0.08 -0.09 -0.11 13. Partner Change in NA 0.07 0.10 -0.01 0.04 0.00 -0.12 0.02 -0.06 -0.17† -0.07 -0.07 14. PA Contagion 0.08 -0.04 -0.05 -0.01 -0.06 -0.02 -0.15 0.02 0.02 0.00 -0.02 15. NA Contagion -0.07 0.10 0.21* -0.01 -0.02 -0.13 -0.03 -0.08 0.02 0.05 -0.06

16. Attention Percentage 0.04 -0.06 0.00 -0.25** -0.11 -0.09 -0.45** -0.05 0.02 -0.05 -0.04 17. Average Gaze Length 0.07 -0.01 -0.11 -0.24** -0.11 0.06 -0.39** -0.03 -0.04 -0.16† -0.06

18. Self-Report Attention -0.06 0.02 0.04 -0.05 -0.07 0.04 0.02 0.03 0.10 0.25** 0.11

Mean 1.50 0.04 3.45 4.92 5.71 3.34 0.49 4.18 4.72 4.45 5.08

Standard Deviation 0.50 0.20 0.62 0.99 0.89 0.80 0.50 1.17 0.94 0.88 0.70 Table 8: Means, standard deviations, and intercorrelations for variables in Study 2. Note N = 121. (Continued)

Table 8, continued

12 13 14 15 16 17 18 1. Gender 2. Knew Partner 3. Susceptibility to EC 4. Engagement 5. Interdependence 6. Task Interdependence 7. Power Condition 8. Similarity 9. Ability 10. Benevolence 89 11. Integrity

12. Partner Change in PA -- 13. Partner Change in NA -0.46** -- 14. PA Contagion -0.15 0.15† -- 15. Self/Coded NA Contagion -0.01 0.18* 0.05 -- 16. Attention Percentage 0.02 0.00 -0.02 0.12 -- 17. Average Gaze Length 0.08 -0.01 0.01 -0.04 0.79** --

18. Self-Report Attention -0.09 -0.08 -0.08 0.03 0.01 -0.01 --

Mean 0.45 -0.01 0.01 0.21 0.21 3.08 28.91

Standard Deviation 0.60 0.44 0.57 0.56 0.12 1.54 13.47

DV: PA Contagion DV: NA Contagion

Independent Variables

Intercept 0.74 -0.69

Gender 0.13 -0.21†

Knew Partner -0.32 0.58*

Susceptibility to EC -0.12 0.30**

Outcome Interdep -0.02 0.00

Task Interdep 0.03 -0.11†

Engagement 0.01 -0.02

Partner’s Change in Emotions -0.16 0.20

Power -0.20† 0.03

F = 1.08 F = 2.51**

N = 119 N = 119 Table 9: Results of analyses testing the direct effect of power on emotional contagion.

DV: Self- DV: Average DV: Attention Reported Gaze Length Percentage Attention

Independent Variables Intercept 33.82** 7.41** 0.55** Gender -2.91 0.36 0.01 Knew Partner 3.11 -0.83 -0.07 Susceptibility to EC 1.66 -0.55* -0.02 Outcome Interdep -1.34 -0.20 -0.01 Task Interdep 0.75 0.33* 0.00 Engagement -1.01 -0.30* -0.02* Power 1.14 -1.26** -0.11** F = 0.38** F = 5.69** F = 5.81** N = 121 N = 121 N = 121 Table 10: Results of analyses testing the indirect effect of power on the three measures of attention. .

DV: PA Contagion DV: NA Contagion

Independent Variables Intercept 0.94† 0.87 0.80 -1.32** -1.06* -1.03* Gender 0.14 0.14 0.12 -0.21* -0.19† -0.19† Knew Partner -0.36 -0.33 -0.30 0.57* 0.52† 0.51† Susceptibility to EC -0.13 -0.13 -0.11 0.30** 0.29** 0.28** Partner’s Change -0.16† -0.15 -0.16† 0.22* 0.23* 0.23* Power -0.27* -0.23* -0.20† 0.10 0.01 0.00 Attention Percentage -0.64 0.84† Average Gaze Length -0.03 0.01

Self-Reported Attention 0.00 0.00

F = 1.73 F = 1.52 F = 1.56 F = 3.42** F = 2.77* F = 2.76* N = 119 N = 119 N = 119 N = 119 N = 119 N = 119 Table 11: Results of analyses testing the indirect effect of attention on emotional contagion.

DV: Self-Reported DV: Average Gaze DV: Attention

Attention Length Percentage Independent Variable Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Intercept 20.90 26.33 7.98** 8.71** 0.55** 0.58** Gender -0.46 -0.63 0.25 0.22 0.01 0.01 Knew Partner 6.36 6.31 -0.98 -0.98 -0.07 -0.07 Susceptibility to EC -0.03 0.32 -0.47* -0.42† -0.02 -0.01 Outcome Interdep -2.04 -1.90 -0.17 -0.15 -0.01 -0.01 Task Interdep -0.03 -0.12 0.37* 0.36* 0.00 0.00 Engagement -2.14 -2.23 -0.25† -0.26† -0.02* -0.02* Power 2.06 -9.83 -1.31** -2.91* -0.11** -0.17 Benevolence 5.25** 3.85† -0.23 -0.42† 0.00 -0.01 Power*Benevolence 2.69 0.36 0.01 F = 1.70 F = 1.61 F = 5.26** F = 4.87** F = 5.05** F = 4.49** N = 121 N = 121 N = 121 N = 121 N = 121 N = 121 Table 12: Results of analyses examining the interaction between power and benevolence predicting the three measures of attention.

DV: Self-Reported DV: Average Gaze DV: Attention

Attention Length Percentage Independent Variable Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Intercept 27.57† 39.39* 7.47** 7.05** 0.54** 0.45** Gender -2.85 -2.72 0.36 0.35 0.01 0.01 Knew Partner 1.74 0.81 -0.82 -0.79 -0.07 -0.06 Susceptibility to EC 1.37 2.03 -0.54* -0.57* -0.02 -0.02 Outcome Interdep -2.00 -2.32 -0.19 -0.18 -0.01 -0.01 Task Interdep 0.34 0.17 0.34* 0.34* 0.00 0.00 Engagement -1.58 -1.68 -0.29* -0.29* -0.02* -0.02* Power 1.63 -23.66 -1.27** -0.35 -0.11** 0.07 Integrity 3.21 1.17 -0.03 0.04 0.01 0.02 Power*Integrity 5.00 -0.18 -0.04 F = 0.66 F = 0.78 F = 4.94** F = 4.38** F = 5.06** F = 4.67** N = 121 N = 121 N = 121 N = 121 N = 121 N = 121 Table 13: Results of analyses examining the interaction between power and integrity predicting the three measures of attention.

DV: Self-Reported Attention

Independent Variables Model 1 Model 2 Model 3

Intercept 31.30† 62.30* 53.84 Gender -2.90 -3.14 -3.06 Knew Partner 3.90 -0.73 -1.13 Susceptibility to EC 1.18 1.44 1.81 Outcome Interdep -1.65 -1.25 -1.42 Task Interdep 0.11 0.17 0.29

Engagement -1.22 -1.49 -1.36

Power 1.16 0.96 12.35

Ability 1.76 -10.70 -6.07

Ability2 1.34 0.71

Power*Ability -7.27

Power*Ability2 0.99 F = 0.50** F = 0.71** F = 0.64**

N = 121 N = 121 N = 121 Table 14: Results of analyses examining the curvilinear interaction between power and ability predicting self-reported attention.

DV: Average Gaze Length Independent Variables Model 1 Model 2 Model 3 Intercept 7.39** 8.64** 11.05** Gender 0.36 0.35 0.27 Knew Partner -0.83 -1.01 -0.87 Susceptibility to EC -0.55* -0.54* -0.56* Outcome Interdep -0.20 -0.18 -0.13

Task Interdep 0.33† 0.33† 0.32†

Engagement -0.30* -0.31* -0.35* Power -1.26** -1.27** -5.29 Ability 0.01 -0.49 -1.64 Ability2 0.05 0.19

Power*Ability 1.90

Power*Ability2 -0.21

F = 4.93** F = 4.40** F = 3.71** N = 121 N = 121 N = 121 Table 15: Results of analyses examining the curvilinear interaction between power and ability predicting average gaze length.

DV: Attention Percentage

Independent Variables Model 1 Model 2 Model 3

Intercept 0.52** 0.51* 0.66* Gender 0.01 0.01 0.01 Knew Partner -0.06 -0.06 -0.05 Susceptibility to EC -0.02 -0.02 -0.03 Outcome Interdep -0.01 -0.01 -0.01 Task Interdep 0.00 0.00 -0.01 Engagement -0.03* -0.03* -0.03*

Power -0.11** -0.11** -0.31

Ability 0.02 0.02 -0.06 Ability2 0.00 0.01

Power*Ability 0.12

Power*Ability2 -0.02 F = 5.32** F = 4.69** F = 4.14**

N = 121 N = 121 N = 121 Table 16: Results of analyses examining the curvilinear interaction between power and ability predicting attention percentage.

DV: Self-Reported DV: Average Gaze DV: Attention

Attention Length Percentage Independent Variable Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Intercept 34.70* 35.19* 7.54** 7.76** 0.55** 0.57** Gender -2.98 -3.01 0.35 0.33 0.01 0.01 Knew Partner 2.83 2.89 -0.87 -0.85 -0.07 -0.07 Susceptibility to EC 1.50 1.61 -0.57* -0.52* -0.02 -0.02 Outcome Interdep -1.61 -1.63 -0.24 -0.24 -0.01 -0.01 Task Interdep 0.58 0.61 0.31† 0.32† 0.00 0.00 Engagement -1.08 -1.07 -0.31* -0.30* -0.02* -0.02* Power 1.09 -0.84 -1.27** -2.14* -0.11** -0.15† Similarity 0.61 0.38 0.09 -0.02 0.01 0.00 Power*Similarity 0.46 0.21 0.01 F = 0.36** F = 0.32 F = 5.02** F = 4.55** F = 5.12** F = 4.55** N = 121 N = 121 N = 121 N = 121 N = 121 N = 121 Table 17: Results of analyses examining the interaction between power and similarity predicting the three measures of attention.

PAt1 PAt2 Partner PA |PAt1- |PAt2- PA T1 PartnerPAt1| PartnerPAt1| Contagion Score

Subject1 1.7 2.1 1.4 0.3 0.7 -0.4

Subject2 2.9 2.2 1.8 1.1 0.4 0.7

Subject3 1.4 1.6 3.9 2.5 2.3 0.2

Table 18: Examples of emotional contagion scores.

Relational

Identification

Power Attention Emotional

Contagion

Trust

Figure 1: Proposed theoretical model.

100

Low Relational Identification High Relational Identification Attention Paid to Target Target Attention to Paid

Low Power High Power Actor's Power

Figure 2: Proposed interaction of Hypothesis 5.

101

3 Low Benevolence/Integrity

2 High Benevolence/ Integrity Attention Paid to Target Target Attention to Paid

0 Low Power High Power Actor's Power

Figure 3: Proposed interaction of Hypotheses 3a and 3b.

102

Figure 4: Plot showing the curvilinear interaction between power and ability predicting attention to emotions.

103

4.3

4.1

3.9

3.7

3.5 Low Relational Identification (T3) 3.3 High Relational 3.1 Identification (T3)

Attention to Emotions (T3) 2.9

2.7

2.5 Low Power High Power

Figure 5: Plot showing the moderated effect of relational identification on the power- attention relationship.

104