ANGER IN THE WORKPLACE: EFFECTS OF GENDER AND FREQUENCY
IN CONTEXT ON SOCIAL AND JOB-RELATED OUTCOMES
A DISSERTATION
SUBMITTED TO THE DEPARTMENT OF SOCIOLOGY
AND THE COMMITTEE ON GRADUATE STUDIES
OF STANFORD UNIVERSITY
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
Kristen Backor
December 2009
http://creativecommons.org/licenses/by-nc/3.0/us/
This dissertation is online at: http://purl.stanford.edu/gx987bn1645
© 2010 by Kristen Brooke Backor. All Rights Reserved.
Re-distributed by Stanford University under license with the author.
This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.
ii
I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.
Cecilia Ridgeway, Primary Adviser
I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.
Karen Cook
I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.
Norman Nie
Approved for the Stanford University Committee on Graduate Studies.
Patricia J. Gumport, Vice Provost Graduate Education
This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file inUniversity Archives.
iii
iv
Abstract
This dissertation explored expression of anger in the workplace through two
experiments that varied in terms of the gender of the person expressing the
anger, the frequency of the anger, and the object of the anger. Subjects reviewed
resumes and performance reviews that they were told belonged to two
consultants competing for the same promotion. Consultants who were said to
express anger frequently were less likely to be promoted than neutral
consultants. They also suffered social consequences, being rated less pleasant
and less likely to be asked for personal advice, among other things. Consultants
who expressed anger a single time suffered social consequences (albeit less
severe than those experienced by the frequently angry individuals) compared to
neutral consultants, but their promotional outcomes were less affected. When
respondents were given the option to promote both the neutral consultant and
the angry consultant, most subjects chose to do so even in the frequently angry
condition. These results from Study 1 were based on an un-directed expression
of anger; in Study 2, respondents were told that the anger was directed at a
subordinate. A single instance of directed anger was more damaging to social
and promotional outcomes than a single instance of un-directed anger. In
contrast to much previous research on anger in the workplace, few gender
effects were observed in either study, and the gender effects that were observed
tended to favor females. Possible explanations for the lack of gender differences
are discussed.
v
Acknowledgments
I first want to thank my Ph.D advisor, Cecilia Ridgeway. She encouraged
me to pursue my interests, even when they were unconventional, and was
always supportive. She is what every mentor, teacher, and advisor should aspire
to be.
I am also indebted to the other members of my reading committee, Karen
Cook and Norman Nie. Both have followed my various research interests
throughout my time at Stanford and provided advice and direction when I needed
it. Paula England and Elizabeth Mullen, members of my defense committee,
gave valuable feedback on the experimental procedures and suggested areas in
the literature that I would not have thought to explore myself.
I would like to express sincere gratitude to everyone who helped me sort
through the red tape and administrative issues that go along with getting a Ph.D,
especially Emily Borom, Lucia Gouet, Suzi Weersing, and Sue Martin. I am
forever indebted to them for their tremendous help with everything, from
providing moral support to helping arrange my oral defense.
Special thanks go to my friends and colleagues in the program, especially
Brandy Aven, Curtiss Cobb III, and Alicia Simmons. I could not have done this
without them.
I could use any number of clichés to thank my partner, Yaki Tsaig, but
they would still be inadequate. He is the best.
vi
I also wish to thank my parents, Ben and Sandra Backor, and the rest of
my family. Their support and encouragement has been invaluable throughout my
life, and the years spent getting this Ph.D were no exception.
Finally, I would like to dedicate this dissertation to my daughter Noa, who
respectfully waited to be born until I completed my first draft. I hope she will grow
up questioning social rules (and everything else).
vii
Table of Contents
Abstract ………………………………………………………………………….. iv Acknowledgements ……………………………………………………………. v Table of Contents ………………………………………………………………. vii List of Tables ……………………………………………………………………. ix List of Illustrations ...…………………………………………………………... xii Chapter 1: Introduction ………………………………………………………... 1 Research Contributions and Goals ……………………………………. 2 Dissertation Outline ……………………………………………………… 4 Chapter 2: Theoretical Background ………………………………………… 7 Expectations States and Status Characteristics ……………………... 7 Emotional Responses – Function and Norms ………………….......... 9 Emotions in Status Hierarchies ......................................................... 10 Status and Adherence to Feeling Rules ............................................ 12 Gender, Leadership, and Social Role Expectations .......................... 14 Gender and Emotional Expression in the Workplace ........................ 18 Anger as a Basic (and Social) Emotion ............................................. 19 Anger and Status ............................................................................... 21 Effects of Gender on Expression of Anger ........................................ 23 Effects of Anger Expression by Frequency – Hypotheses ................ 24 Chapter 3: Experimental Methodology and Design ................................ 34 Role of Context and Format in Presentation of Emotional Expression ...................................................................................... 34 Experimental Design ......................................................................... 36 Participant Recruitment ..................................................................... 38 In-Person Procedures ........................................................................ 40 Online Procedures ............................................................................. 42 Manipulation ...................................................................................... 43 Questionnaire .................................................................................... 45 Debriefing .......................................................................................... 46 Chapter 4: Frequency of Anger, Gender, and Promotional Decisions 51 Who Gets Promoted? ........................................................................ 51 Logistic Regressions Analyses of Promotion .................................... 53 Mediational Analyses – Considering Consultant Traits (via Principal Component Analysis) ....................................................... 55 Discussion ......................................................................................... 60
viii
Chapter 5: Frequency of Anger, Gender, and Social Outcomes ........... 79 Measuring Sociability ......................................................................... 79 OLS Regression Analysis of Factors Influencing Sociability ............. 80 Significance Tests Comparing the Sociability of Angry and Neutral Consultants ..................................................................................... 81 Discussion ......................................................................................... 84 Chapter 6: Study 2 – Directed Anger and Workplace Outcomes ........... 97 Theoretical Justification ..................................................................... 97 Experimental Design and Hypotheses .............................................. 100 Preliminary Analyses ......................................................................... 102 Discussion ......................................................................................... 105 Chapter 7: Conclusions and Future Directions ....................................... 117 Review of Major Findings and Contributions ..................................... 117 Directions for Future Research .......................................................... 121 Appendix A: Experimental Materials – In-Person ................................... 124 Resume S .......................................................................................... 124 Resume C .......................................................................................... 126 Performance Review T ...................................................................... 128 Performance Review A ...................................................................... 130 Set-Up Script ..................................................................................... 132 Debriefing Script ................................................................................ 134 Foothill Consent Form ....................................................................... 137 Stanford Consent Form ..................................................................... 139 Questionnaire .................................................................................... 141 Appendix B: Experimental Materials – Online, Including Study 2 ......... 149 Resumes and Performance Reviews (Study 1 Online and Study 2) 149 Set-Up Script (Study 1 Online and Study 2) ...................................... 149 Online Questionnaire Formatting ....................................................... 151 Consent Form – Online ..................................................................... 152 Appendix C: Univariate Statistics and Randomization Checks ............. 160 Appendix D: Principal Component Analyses of Questionnaire Variables ............................................................................... 176 List of References ...................................................................................... 183
ix
List of Tables
Table 2.1: Alternative Hypotheses Regarding Social and Job-Related Effects of Frequency of Anger ..................................................... 33 Table 3.1: Breakdown of Subjects by Gender and Experiment Format ......................................................................................... 48 Table 3.2: Study 1 Participants, by Condition .............................................. 49 Table 3.3: Study 2 Participants, by Condition .............................................. 50 Table 4.1: Chi-Square and z-Test Analyses of Likelihood of Promotion, by Frequency of Anger Within Gender ............................................. 68 Table 4.2: Chi-Square and z-Test Analyses of Likelihood of Promotion, by Frequency of Anger Across Gender ............................................ 69 Table 4.3: Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss, Dummy Independent Variables ................................................... 70 Table 4.4: Logistic Regression Analysis of Whether Respondents Would Promote the Angry Consultant (Over the Neutral Consultant), Interacted Independent Variables ............................................... 71 Table 4.5: Logistic Regression Analysis of Whether Respondents Would Prefer to Promote Both Consultants if Possible), Interacted Independent Variables ................................................................ 72 Table 4.6: Factor Loadings for Principal Component Analysis Using Manipulation Factors ................................................................... 73 Table 4.7: Factor Loadings for Principal Component Analysis Using Difference Factors ....................................................................... 74 Table 4.8: Eigenvalues and Percent Explained for Each Component (Created with Principal Component Analysis) ............................. 75 Table 4.9: Standardized Coefficients for OLS Regression Analyses of the “Suitability for Promotion” Component (Created with PCA using Manipulation Variables) ............................................................... 76 Table 4.10: Mediational Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss, Using Manipulation Components ............................. 77
x
Table 4.11: Mediational Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss, Using Difference Components ................................ 78 Table 5.1: Eigenvalues and Percent Explained for “Sociability” Components ............................................................................... 88 Table 5.2: Variables with Large Loadings on the “Sociability” Factor .......................................................................................... 89 Table 5.3: Standardized Beta Coefficients for OLS Regression Analyses of “Sociability” Factors (Created with PCA) ................................ 90 Table 5.4: Means and Standard Deviations for Trait Variables Used in RMANOVAs ................................................................................ 91 Table 5.5: RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Traits (F values) ....................................... 93 Table 5.6: Means and Standard Deviations for Behavioral Variables Used in RMANOVAs ............................................................................ 94 Table 5.7: RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Likelihood that Respondent Would... (F values) ........................................................................................ 96 Table 6.1: Chi-Square and z-Test Analyses of Likelihood of Promotion for Directed Anger, by Gender ......................................................... 108 Table 6.2: Chi-Square and z-Test Analyses of Likelihood of Promotion, by Gender of Consultant and Directed Anger .................................. 109 Table 6.3: Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss in Single-Anger Conditions ............................................................. 110 Table 6.4: Means and Standard Deviations for Trait Variables Used in RMANOVAs ................................................................................ 111 Table 6.5: Means and Standard Deviations for Behavioral Variables Used in RMANOVAs ............................................................................. 113 Table 6.6: RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Traits, Based on Study 2 Subjects and Online Study 1 Subjects in the Single Anger Condition .............. 115
xi
Table 6.7: RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Likelihood that Respondent Would… , Based on Study 2 Subjects and Online Study 1 Subjects in the Single Anger Condition ............................................................... 116 Table C.1: Univariate Statistics for Study 1 .................................................. 162 Table C.2: Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Condition .................................................. 166 Table C.3: Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Materials ................................................... 169 Table C.4: Univariate Statistics for Study 2 .................................................. 171 Table C.5: Randomization Checks (Study 2) – Chi-Square Tests for Control Variables, by Gender of Consultants ............................. 174 Table D.1: Factor Loadings for Principal Component Analysis Using Manipulation Factors .................................................................. 180 Table D.2: Factor Loadings for Principal Component Analysis Using Difference Factors ...................................................................... 181 Table D.3: Eigenvalues and Percent Explained for Each Component (Created with Principal Component Analysis) ............................ 182
xii
List of Illustrations Figure B.1: Formatting of Initial Survey Questions Online ........................... 154 Figure B.2: Formatting of Relative Ranking Questions Online ..................... 155 Figure B.3: Formatting of Likert-Scale Survey Questions Online ................. 156 Figure B.4: Screenshot of Online Consent Format ...................................... 157 Figure B.5: Screenshot of Debriefing Questions .......................................... 158 Figure B.6: Screenshot of Debriefing Text ................................................... 159 Figure D.1: Scree Plot of Eigenvalues for Components Based on Principal Component Analysis of Manipulation Variables ........................ 178 Figure D.2: Scree Plot of Eigenvalues for Components Based on Principal Component Analysis of Difference Variables ............................ 179
1
Chapter 1: Introduction and Overview
According to Goffman (1956), an individual participates in an interaction
not as a person, but in terms of his or her status. The status of interactants
informs the rules of conduct that govern the interaction, including rules
surrounding emotional behavior. Status permits or constrains the expression of
certain emotions, and the expression of those emotions can in turn affect the
conclusions observers draw about the expresser's status. The relationship
between emotional expression and status is key in the workplace, where relative
status is often clear to all involved parties and violations of social norms can have
a variety of implications, affecting job prospects, salary and benefits, and social
interaction.
When considering emotional expression and status in the workplace,
anger is of particular interest, as it is a powerful emotion that can be an effective
workplace tool (especially for individuals in leadership roles) and is often clearly
linked to status. For example, high status individuals are expected to experience
anger more often than low status individuals (Ridgeway & Johnson, 1990). The
relationship between anger and gender, a status characteristic that is often
salient in the workplace and for which women are generally considered to be
lower status than men, is further complicated by social norms since anger is
more normative for males than females (Tiedens, 2001). This can create
difficulties for women in the workplace who try to utilize anger to achieve goals.
The negative consequences of anger expression on the part of low status
individuals and females have been the subject of much research. However, such
2
studies often fail to explore anger expression in a systematic and realistic way. In
this dissertation, I investigate how gender and expression of anger in the
workplace affect occupational and social outcomes. I contribute to previous
research in this area by incorporating different frequencies of anger, identifying
the anger as occurring within a general context or directed at a specific person,
and using actual workplace materials (resumes and performance reviews) to
introduce the anger. I conducted two experimental studies that required
participants to review these workplace materials and provide information about
how they perceived the applicants in question and how they would compensate
or treat those applicants in both social and workplace settings.
Research Contributions and Goals
This research differs from previous research on anger in the workplace in
two important ways. First, the majority of previous studies have focused on a
single incident of anger, which does not allow for exploration of the impact
frequency of anger has on evaluations. In an actual workplace environment,
there are likely to be repeated interactions with the same individuals over time,
providing multiple opportunities for anger expression. An individual who has
become angry only once across these interactions is likely to be viewed
differently than an individual who is angry more often. When presented alongside
other information, being angry once (or otherwise infrequently) may appear to be
an isolated incident and thus seem more excusable than frequent expressions of
anger. In addition, if an employee is presented as otherwise competent, this
competence may mitigate the effects of anger or make the anger appear
3
justifiable. Based on previous research, it is not clear whether a single instance
of anger expression has negative consequences when presented in the context
of information about the employee’s performance, skills, and general demeanor.
Second, previous studies have often presented anger in ways that may
reduce the generalizability of the findings by confining expression of anger to a
certain situation, like a job interview or a meeting with a client. Presenting anger
in the context of such situations invokes the emotion norms specific to these
situations, and these may differ from actual workplace emotion norms. For
example, an individual who expresses or discusses anger in a job interview
(when that individual should be trying to present an extremely positive view of
himself or herself) likely gives a different impression than an individual who
expresses anger in an actual job environment. Such contextual effects may be
exaggerated by the limited amount of information given about the employee in
previous experiments, particularly with regard to his or her typical demeanor, but
also in terms of typical work performance.
Thus, I sought to build on past work in two specific ways. First, I
considered the effects of repeated expressions of anger versus a single incident.
In doing so, I investigated how the social and job-related outcomes of expressing
anger vary depending on how often that anger is expressed. This work aims to
lay a foundation upon which future scholars may be able to identify a threshold
effect for anger, such that infrequent anger may be inconsequential or even
beneficial, but these effects may be lost at some point as others’ perceptions of
the individual change. Second, I evaluated anger in a more realistic work context,
4
presenting the expression of anger in the context of resumes and performance
reviews that provide information about the individual and his or her capabilities.
Contextual effects may be particularly strong for women due to a number of
factors, including subjects’ willingness to devalue the performance of females
(Brescoll & Uhlmann, 2008; Heilman, 2001), so findings based on a more
realistic experimental design may differ from those reported in previous research.
This research contributes to current approaches to studying anger in the
workplace, evaluates the consequences of varying levels of anger expression,
and provides insight into how researchers’ decisions about the way in which
emotional responses are presented in an experimental context might influence
their results. The experimental procedure used in these studies can be modified
in several ways to continue to explore these issues in the future.
Dissertation Outline (Including Chapter Descriptions)
In Chapter 2, I present the theoretical background that supports my
research, drawing on sociological and psychological literature. I first discuss the
emotional implications of expectation states theory and status characteristics
theory, especially in the workplace. As emotions are linked with status, an
individual’s perceived status influences the emotions they are expected (or even
allowed) to experience and express. I then discuss how these expectations may
differ for males and females, drawing on gendered beliefs about social roles to
address behavioral and emotional expectations. I address the implications of this
dynamic and consider how different levels of anger expression might play out in
an actual workplace context. Chapter 2 concludes with a series of hypotheses.
5
Chapter 3 presents the rationale and design for the experiments I
conducted to examine how expression of anger in the workplace affects
perceptions of males and females and, in turn, how those perceptions influence
social and job-related outcomes. I discuss the importance of context and
consider the methods used by previous studies of anger in the workplace. I then
introduce the materials used in this study. I explain why I chose to use
performance reviews and resumes as my experimental stimuli and discuss the
process I used to design and develop the materials; I also address my choice of
a specific occupation, consulting. The remainder of Chapter 3 focuses on the
experimental procedures and manipulation, including a discussion of the different
types of measures I employed to capture social and job-related outcomes.
In Chapter 4, I examine the effects of frequency of anger expression and
gender on promotional outcomes. I use logistic regression analyses and principal
component analysis to consider explicit promotional decisions as well as factors
related to each consultant’s perceived suitability for promotion. I then discuss the
results of these analyses in light of my hypotheses.
In Chapter 5, I examine the effects of frequency of anger expression and
gender on social measures, focusing on traits and social behaviors. I again use
principal component analysis to determine what factors influence perceived
sociability. I also use analysis of variance to make within- and between-subjects
comparisons on traits like pleasantness as well as behaviors like being friends
with the consultant. I then discuss the results of these analyses in terms of my
hypotheses.
6
Chapter 6 presents the rationale and methods for a follow-up study (Study
2), which was conducted based on preliminary findings from the initial experiment
(Study 1). In Study 2, the anger described in the manipulation was specified to be
directed at a subordinate. The results of this study and their relationship to the
results of Study 1 are briefly discussed.
In Chapter 7, I draw overall conclusions based on my results and present
some directions for future research.
7
Chapter 2: Theoretical Background
In this chapter, I draw on theories of status, emotion, and gender in order
to develop and explore questions about the relationships between these
concepts, focusing on how gender affects perceptions of expression of anger in
the workplace.
Expectation States and Status Characteristics
Expectation states theory (EST) and status characteristics theory (SCT)
provide the initial framework for considering interaction in this study. In these
theories, interaction is considered in terms of status using status characteristics,
or attributes on which individuals differ and for which cultural beliefs define one
state as more desirable than another. These status characteristics become
activated in situations in which a status characteristic differentiates the
participants and/or is perceived to be relevant to the shared task (Correll &
Ridgeway, 2003). Specific status characteristics are associated with competence
in a specific area; thus, an individual with a degree in mathematics would be
assumed to be more competent in situations requiring mathematical ability than
an individual with a degree in sociology. Diffuse status characteristics are
characteristics like gender or race with differentially evaluated states that are
associated with enhanced performance generally, as well as on specific tasks
(Berger, Wagner, & Zelditch, 1985).1 With regard to gender, the focus of this
study, males are generally considered to be more competent and higher status
1 Diffuse status characteristics can also be linked to specific status characteristics. Thus, using gender as an example, females are assumed to be more competent in sewing, so this perceived skill may be activated as a specific status characteristic in discussions relating to dress-making.
8
than females (Berger, Rosenholtz, & Zelditch, 1980; Ridgeway & Diekema,
1992).
Status characteristics theory expands on EST by attempting to explain
how status beliefs based on an individual’s personal characteristics generate
expectations about behavior, called performance expectations; these
performance expectations are then self-fulfilling because those with high status
are given more opportunities to speak and evaluated more highly than low status
individuals, independent of their actual performance (Ridgeway, 1987; Ridgeway,
Berger, & Smith, 1985). High status individuals are thought to be more
competent and intelligent than those of lower status, and these attributions occur
even when participants know that status has been randomly assigned (Tiedens,
Ellsworth, & Mesquita, 2000). Individuals with higher status in a given situation
(typically determined by the combination of diffuse and specific status
characteristics) also behave differently; they speak more often, make more direct
eye contact, and interrupt more often (Berger, Cohen, & Zelditch, 1972; Dovidio,
Brown, Heltman, Ellyson, & Keating, 1988; Tiedens et al., 2000).
The implications of status characteristics and the accompanying
performance expectations are obviously far-reaching and have been the subject
of sociological research in a variety of contexts and situations, including the
workplace. Gender is an important diffuse status characteristic in this setting,
particularly because women are considered less competent than men. This belief
influences the expectations people hold about women's performance in the
9
workplace, which in turn affects how women are treated and responded to in the
workplace, independent of their structural position.
Emotional Responses – Function and Norms
In this dissertation, I focus on how the diffuse status characteristic of
gender interacts with expression of emotion in the workplace. Status
characteristics theory has been explicitly linked with emotional responses –
specifically, with what emotions individuals can express (Ridgeway & Johnson,
1990). Generally, emotions “serve as social signals and can
provoke...interpersonal processes” (Hareli & Rafaeli, 2007, p. 8). In functioning
as social signals, emotions provide valuable information about the status of
interactants. This status information combined with the emotional responses
themselves helps people in dyads or groups communicate and form social bonds
(Frijda, 1994; Izard, 1972). In addition, the emotions of a single individual can
shape the thoughts, beliefs, and emotions of others.
In Durkheim’s (1961, 1984) view, interpretation of emotion is
accomplished through collective interaction, which can result in the development
of social norms; these norms, in turn, are then used to guide and control behavior
in interaction. Of particular interest here are Hochschild’s (1983) feeling rules,
which suggest that emotional expression is governed to some extent by a system
of guidelines that delineate the extent, direction, and duration of appropriate
feelings for any given situation. In addition to feeling rules, which address an
individual’s internal emotional state, there are display rules that provide
information about when it is appropriate (or even necessary) to exhibit certain
10
emotions (Ekman & Friesen, 1969). These display rules can include explicit
orders about appropriate facial expressions, such as flight attendants being
instructed to smile, as well as learned habits about what emotions are
appropriate for whom (e.g., crying is less acceptable from men than from women,
even in situations in which feeling rules call for sadness) (Hochschild, 1983).
The existence of feeling rules suggests that there are “correct” and
“incorrect” emotions for particular situations (sadness at funerals is an oft-
mentioned example of the former). Individuals who do not experience the
appropriate emotion may use various strategies in an attempt to reconcile their
feelings with the situation (Gross, 2002; Hochschild, 1979, 1983), such as
attempting to evoke the "required" feeling or making superficial displays of the
correct emotion. Individuals who fail to respond appropriately face being judged
by others for their inappropriate emotions (Thoits, 1985).
Emotions in Status Hierarchies
As noted above, Ridgeway and Johnson (1990) have linked emotional
responses to status and status characteristics. Emotions are “embedded in and
influenced by the social context” (Jakobs, Fischer & Manstead, 1997, p. 104),
and the relationship between emotions and status is complex, affecting every
stage of emotion from assessment of physiological cues to production of facial
expressions. Status-linked emotions help stabilize and reinforce social
hierarchies. In fact, emotional responses have a circular relationship with
hierarchical position (Tiedens et al., 2000). As one part of this
relationship, individuals are expected to exhibit emotions in line with their social
11
position and in accordance with the corresponding emotional display rules. Thus,
the emotions individuals are expected to feel vary according to context and
individual status (Heilman, Block, Martell, & Simon, 1989; Hochschild, 1983;
Tiedens, 2001). High status individuals are expected to feel anger and pride,
while low status individuals are thought to feel sadness and guilt more often
(Ridgeway & Johnson, 1990). These feelings are expected to occur based on the
success or failure of a particular endeavor; more specifically, high-status
individuals are expected to feel responsible and proud with success, while with
failure, high-status individuals are expected to feel angry and low-status
individuals are expected to feel guilt. The emotions individuals do in fact
experience tend to be in line with these expectations (Tiedens et al., 2000).
Beliefs about how emotions link with status do not just influence behavior
and expectations about behavior based on status; as another part of the circular
relationship, characteristics are inferred on the basis of the emotions an
individual expresses, and emotional displays can lead people to confer status or
make assumptions about an individual’s social status (Rafaeli & Sutton, 1987).
When status was left ambiguous, angry characters in a vignette were thought to
be high status, while sad and guilty characters were seen as low status (Tiedens
et al., 2000). Thus, in some cases, emotional displays can themselves provide
direct evidence of status. In addition, people expressing anger are seen as
dominant, strong, competent, and smart, but less warm, friendly, and nice; the
opposite holds true for people expressing sadness (Carli, LaFleur, & Loeber,
12
1995; Tiedens, 2001).2 This relationship can be linked directly to status as well,
since, as previously noted, status can lead observers to attribute such
characteristics to individuals (Tiedens, 2001). Further, employees draw
conclusions about the abilities of their boss based on emotional displays (Lewis,
2000). These inferred characteristics have significant implications for public
expression of emotions, especially when status differences are involved.
Finally, and unsurprisingly given the relationships described above,
particular emotions have different consequences depending on the social status
of the person expressing them (Lewis, 2000). For example, expression of certain
emotions by a high status individual may reaffirm or inspire confidence in a that
individual, as when individuals had more positive feelings about a politician’s
capabilities when the politician was shown expressing anger and making threats
than when the same politician was seen expressing fear and evasiveness
(Tiedens, 2001).
Status and Adherence to Feeling Rules
Beyond influencing emotional expectations and assumptions, status may
also influence adherence to the feeling rules and/or emotional display rules
discussed by Hochschild (1983). The consequences of violating social norms
such as feeling rules differ significantly for high or low status individuals.
Expectancy Violation Theory (EVT) suggests that positively valenced expectancy
violations produce more favorable outcomes than expectancy confirmation; thus,
it follows that negatively valenced expectancy violations would be more likely to
2 The connection between this relationship and the work regarding the competence/warmth dimensions of stereotypes (e.g., Glick & Fiske, 2001) is clear.
13
produce unfavorable outcomes than expectancy confirmation (Sheldon, Thomas-
Hunt, & Proell, 2006). However, EVT further suggests that the status of the actor
violating expectations affects how these violations are valenced. Actor status
(often represented by expected competence within task situations) moderates
responses to violations: violations by high status individuals are considered more
acceptable than violations committed by low status individuals, even for deviant
behaviors (Burgoon & Hale, 1988). Norm violations can increase the amount of
attention paid to violators, eliciting different behavioral responses (Ridgeway,
1981); when a norm violation is associated with a high status individual, this
increased attention also brings attention to their high status, which may in fact
increase their ability to exert influence (Sheldon et al., 2006).
Thus, high status individuals may be allowed more flexibility in emotional
displays, as the consequences of violating social norms can differ considerably
for high versus low status individuals. Emotional expression, particularly for
negative emotions, is constrained by the status hierarchy; low status individuals
are more likely to feel bound by rules and expectations, while high status
individuals may have more leeway to express what they feel (Hess, Adams, &
Kleck, 2005; LaFrance & Hecht, 1999; Lucas & Lovaglia, 1998; Ridgeway &
Johnson, 1990; Sloan, 2004; Tiedens, 2001). Given that norm violations can
work in favor of high status individuals, violations of feeling rules by high status
individuals may possibly be perceived as acceptable or increase their influence
by highlighting their high status (Sheldon et al., 2006). For example, some
14
viewers felt that Joe Biden’s brief display of sadness during a 2008 Vice
Presidential debate humanized him.3
Gender, Leadership, and Social Role Expectations
Beliefs about women's competencies and characteristics, including feeling
rules and emotional expectations, can make the workplace in general, and
leadership in particular, a greater challenge for women than for men.4 Social role
theory suggests that women are stereotyped as more communal, while men are
seen as more agentic (Eagly, 2005). As a general tendency, people are expected
to engage in activities consistent with these culturally defined gender roles (Eagly
& Karau, 2002). Thus, women are generally expected to manifest feminine
(communal) values through affectionate, helpful, and kind behavior, even in the
workplace (Eagly, 2005).
However, these values are often in conflict with stereotypical leadership
qualities and effective status-earning behavior. Because people doubt that
women possess appropriate competencies and tend to resent any overturning of
the expected hierarchical relation between the sexes (Rudman & Fairchild,
2004), they may react negatively to female leadership, especially if female
leaders behave authoritatively (Eagly et al., 1992). Many of the tools available to
male leaders, like competitive assertiveness, are less effective for female leaders
3 A writer for the Huffington Post, a news website, noted that “Many have called Biden's tears political gold, but few have questioned their authenticity” (Blake, 2008, paragraph 2). 4 When gender-based expectations for behavior, including emotional display rules, carry over into the workplace, this is known as gender- or sex-role spillover (Eagly, Karau, & Makhijani, 1995).
15
(Yoder, 2001), likely because violating stereotypes can result in negative
personality evaluations (Algoe, Buswell, & DeLamater, 2000).
In fact, violations of norms in general have significant implications for
women, especially given the claims of EVT. Carli et al. (1995) suggest that
women who violate status norms are more likely to be disliked than men who do
so, and perceptions of likeability are more important for women than for men as a
result of gender expectations and status beliefs. Supporting the increased weight
of norm violations for women, Eagly (2005) suggests that female leaders may be
less tolerant of rule-breaking and ethical violations than men, and Ridgeway,
Backor, Li, Tinkler, and Erickson (2009) found that females were less willing to
take advantage when put into a high-status position. Thus, while women have
become more similar to men with regard to career aspirations, self-reports of
assertiveness and dominance (Twenge, 1997), and value placed on particular
job attributes (Konrad, Ritchie, Lieb, & Corrigall, 2000), climbing the career
ladder continues to present a greater challenge for women than for men.
Not only is leadership itself gendered by social role expectations, it occurs
in a gendered context (Yoder, 2001). When women do achieve leadership
positions, even legitimately, conflicts between their social role and the leadership
role persist; gender functions as a sort of “background identity” (Ridgeway, 2001,
p. 644) that colors perceptions of male and females in the workplace. Eagly and
Karau's (2002) role congruity theory predicts that prejudice against female
leaders varies with the amount of incongruity between the leadership role and the
feminine gender role. Thus, women must essentially choose between two types
16
of bias: descriptive, in which there is a lack of fit between the feminine role and
the leader role (i.e., the woman continues to act in a feminine way, thus failing to
fulfill leadership characteristics), and prescriptive, in which the woman adopts
more masculine characteristics and is punished for behaving outside her gender
role (Eagly, 1987; Rudman & Glick, 1999; Johnson, Murphy, Zewdie, & Reichard,
2008).5 According to Eagly (2005), someone whose stereotypical attributes are
associated with a positive performance will be preferred over someone whose
attributes are thought to hinder performance, even when there are no actual
observable differences between the individuals.
In fact, the definition of a leader or manager is often viewed as
synonymous with masculine traits (Johnson et al., 2008), even globally (Schein,
2001), and many people continue to report a preference for a male boss (Eagly,
Karau, & Makhijani, 1995; Rudman & Kilianski, 2000). Descriptions of successful
managers tend to focus on masculine characteristics (Algoe, Buswell, &
DeLamater, 2000; Heilman et al., 1989); even dressing in a more masculine style
can result in better job-related outcomes for females (Forsythe, 2006). Therefore,
it is unsurprising that masculine characteristics are important predictors of
leadership (Rudman & Kilianski, 2000). However, women who exhibit masculine
characteristics like assertiveness in a work environment face negative
consequences (Yoder, 2001), such as the firing of Ann Hopkins in the Price
Waterhouse v. Hopkins case (as discussed by Eagly, 2005). Furthermore,
5 Prescriptions include desirable as well as undesirable behaviors (Heilman, Wallen, Fuchs, & Tamkins, 2004).
17
women in leadership roles in areas thought to be masculine perform less well
than women in other roles (Eagly, 2005).
Previous research has found that there are ways for women to increase
their leadership effectiveness, including behaviors like avoiding dominant speech
and showing group orientation (Ridgeway, 1982; Yoder, 2001). In addition,
although men are believed to be more agentic and goal-directed, women’s
nurturing role is thought to favor superior interpersonal skills and ability to
communicate nonverbally (Eagly & Karau, 2002; Yoder, 2001). Generally, female
leaders are thought to be less hierarchical, more cooperative, and more likely to
focus on enhancing others' self-worth than men (Eagly & Johnson, 1990). Eagly
(2005) notes that females may be more likely to exhibit a transformational
leadership style than males, meaning they are more likely to aim to establish
themselves as a role model by gaining the trust and confidence of followers.
They also display more contingent reward behaviors than male leaders (Garcia-
Retamero & Lopez-Zafra, 2006).
These elements of a so-called "feminine" leadership style may be able to
minimize role conflict for female leaders. Essentially, women must act as
competent leaders while “reassuring others that they conform at least partially to
expectations concerning appropriate female behavior” (Eagly, 2005, p. 469).
When women do not temper the agentic behaviors necessary to assert
leadership with softer, female behaviors, they are often overlooked when hiring
and promotional decisions are made, and they may have difficulty obtaining the
support of their employees (Eagly, 2005; Rudman & Glick, 1999).
18
Gender and Emotional Expression in the Workplace
Given the need to present themselves as nurturing and goal-oriented,
women’s leadership behaviors may be further constrained by feeling rules
associated with high status positions. Women are believed (and in some cases
expected) to experience and express embarrassment, fear, happiness, guilt,
sympathy, sadness and love more often than men, while the emotions of anger
and pride are ascribed to men more often than women (Algoe et al., 2000; Plant,
Kling, & Smith, 2004). In the workplace, status characteristics theory would seem
to suggest that women would feel less positive emotion than men even in
leadership positions (as they would be evaluated less highly than males)
(Ridgeway & Johnson, 1990). Further, status-compatible emotions theory
proposes that high status group members should experience more positive
emotions than low-status group members, and research seems to support this
claim, as high status individuals do report more positive emotions (Lucas &
Lovaglia, 1998).
However, women actually smile more than men when interacting on
different types of tasks (Dovidio et al., 1988) and report more positive emotion
(Lucas & Lovaglia, 1998). Ridgeway and Johnson (1990) suggest that this is
because women are socialized to express more positive emotions than men in
work situations. Females may be more likely to be judged on the basis of
inappropriate emotions (Thoits, 1985), as emotional responses, and a pleasant
demeanor in particular, are often an important part of female gender roles and
women are thought to be more emotional than men (Plant et al., 2004).
19
Overall, women may feel that a failure to smile or behave warmly will be
met with social disapproval, whether due to their lower status (on the diffuse
status characteristic of gender) or feminine sex role expectations, while men feel
more socially permitted to display emotions like anger as a sign of dominance
(Hess et al., 2005). Female leaders use more positive emotion with subordinates
(presumably in keeping with a "feminine" leadership style); they also have higher
rates of agreement and make fewer counterarguments (Johnson, 1993; Johnson,
Clay-Warner, & Funk, 1996). Lewis (2000) notes that reactions to emotional
responses vary depending on whether the expressed emotion is “gender
endorsed” (p. 225). These gender-linked expectations in the workplace can affect
hiring decisions, promotions, job performance evaluations, and even job
performance itself.
Anger as a Basic (and Social) Emotion
This dissertation focuses specifically on anger in the workplace. Since
anger, a masculine emotion, is often associated with power, competence, and
strong leadership, the potential consequences of violating gendered emotion
norms regarding anger in the workplace are significant (Hess et al., 2005;
Tiedens, 2001). Anger has been the subject of much emotion research; it
represents one of the so-called “basic” emotions (Ekman & Friesen, 1975) and
has been blamed in the past for societal problems like crime (Linden et al.,
1997). In fact, Stearns and Stearns conclude that anger has been and remains
the “central emotional enemy” for contemporary Americans (1986, p. 211).
Though anger is considered a basic emotion, it is clearly influenced by
20
social factors. As noted previously, emotion rules dictate, to some extent, who
can express anger (and to/at whom); these same emotion rules indicate that
anger is more acceptable in some situations than in others (for example, an
angry outburst directed at an umpire by a coach during a baseball game is
obviously more widely accepted than would be a similar outburst directed at
young children by an elementary school teacher). Even the body’s physical
response to anger is subject to social and contextual influence; participants in
Schachter and Singer’s well-known (1962) study interpreted arousal as a result
of injected adrenaline differently (as anger or joy) depending on the social
context within which that arousal was experienced (in their case, the social
context was based on the behavior of a confederate).6
Emotions like anger can affect not only those directly involved in an
exchange, but also those witnessing the expression of emotions, further situating
anger within a social context (Hareli & Rafaeli, 2007). Someone observing an
incident (e.g., an angry outburst) may make dispositional inferences about the
actor (Semin & Manstead, 1981). Performance disruptions like those resulting
from angry outburst have consequences in terms of the individual’s personality,
the specific interaction, and the social structure (Goffman, 1959). This threefold
effect is particularly evident in the workplace, where impression management is
crucial and the social structure is relatively well-defined. Fisher and Chon (1989),
in discussing the specific application of Durkheim (1951) to anger, note that
6 Thoits (1989) points out that there have been some questions raised regarding this study, but she notes that the applicable point here (that social context can influence interpretations of physiological responses) has been supported by other researchers.
21
anger is aroused when an individual’s desires are frustrated, and society helps to
regulate angry responses through norms.
Numerous studies have observed that anger is viewed and experienced
differently based on gender, which is unsurprising given the relationship between
status and emotional expression. The social nature of anger and its regulation
means that societal norms and beliefs about males and females, including their
traits and the appropriateness of expressing particular emotions, often come into
play. Since anger, a masculine emotion, is often associated with power,
competence, and strong leadership, the potential consequences of violating
gendered emotion norms are particularly significant in the workplace (Hess et al.,
2005; Tiedens, 2001). Such norms may affect social interactions with and
perceptions of the angry individual as well as result in more concrete workplace
consequences (such as affecting promotional or hiring decisions).
Anger and Status
Anger’s associations with leadership qualities have made it a subject of
particular interest in status and workplace research (Sloan, 2004). Although it is
generally thought that individuals in the workplace should maintain an affectively
neutral tone (with the exception of emotions appropriate to specific events),
anger is in fact a relatively common workplace emotion (Fitness, 2000). In
addition, expression and experience of anger in the workplace has been
associated with both positive and negative consequences (Tiedens, 2001).
Anger can be used to achieve status or demonstrate worth and is often
used strategically to achieve certain ends (Tiedens [2001] provides several
22
examples of strategic use of anger, such as bill collectors trying to get people to
pay and parents controlling children; see also Ridgeway & Johnson, 1990 and
Sloan, 2004). Angry displays do not fulfill all social goals (they can, for example,
result in perceptions of coldness, especially for women [Wiley & Eskilson, 1985]),
but they are especially effective in attaining status by demonstrating
competence. Perceptions of competence have been shown to mediate the
relationship between emotional expression and status conferral; in other words,
anger encourages status indirectly by creating a perception of competence
(Tiedens, 2001). Likeability has not been found to have such associations with
competence (Tiedens, 2001); thus, it follows that people would give more power
and status to someone who expresses anger than someone who expresses
sadness, which is associated with warmth and likeability.
Anger is not always effective and can in fact be a risky method for
attaining status, especially when the expression of anger is in violation of an
activated stereotype. Jones and Pittman (1982) suggest that intimidation
strategies like anger are only successful for people who already have
power/status or when power/status is ambiguous. This was somewhat supported
by Tiedens (2001), who found that while angry job applicants received higher
salary offers, sad individuals were more likely to be offered the job in the first
place. She concludes that anger may be most useful once one is already part of
a group or organization; when trying to gain access, one might need to express
emotions that signal submission, like sadness.
23
Effects of Gender on Expression of Anger
The risks of employing anger may be particularly significant for women;
Tiedens (2001) notes that “anger is not considered typical or normative for
women and…norm-breaking often inhibits status attainment” (p. 92), concluding
that angry displays may be less successful for women trying to achieve status.
Men who express anger in a professional setting are more likely to be hired and
given status, power, and independence than men who express sadness (Brescoll
& Uhlmann, 2008; Tiedens, 2001). However, professional women who express
anger are ascribed lower status than angry men, regardless of the woman's
actual workplace status (CEO vs. assistant trainee) (Brescoll & Uhlmann,
2008). Angry males have also been found to receive higher compensation than
sad males or angry females, while sad females received higher salaries than
angry females (Brescoll & Uhlmann, 2008). Another study using sensitive/strong
rather than sad/angry found a similar pattern in which strong males were rated
more highly than sensitive males or strong females (Johnson et al., 2008).
Because anger is often associated with masculinity, expression of anger
may result in lower ratings of leader effectiveness for females (Eagly et al.,
1992). As the sensitive/strong study suggests, even assertiveness can have
negative consequences for women (Johnson et al., 2008; Yoder, 2001); male
applicants are more likely to be offered a job when behaving assertively, but
female applicants are more likely to be successful in obtaining a job offer when
using a rational, unemotional approach (Buttner & McEnally, 1996). Additionally,
being seen as “tough” is advantageous for males, but tough females are
24
perceived as “difficult to get along with” (Pierce, 1995, p. 17). Pierce further notes
that women are trapped in a double bind with regard to aggression, and women
who do act like men are punished. Overall, women are evaluated as less
effective when exhibiting more masculine styles (Eagly, 2005; Eagly et al., 1995);
when they do perform masculine jobs successfully, their success tends to be
attributed not to ability, but rather luck or hard work (Lyness & Judiesch, 1999).
Expression of anger may also be seen as an indication of lack of
emotional control, which Goleman (1998) found to be consistently related to
leader ineffectiveness. This is particularly likely for women; Brescoll and
Uhlmann (2008) found that when no explicit external justification was given for a
woman's anger, subjects seemed to assume an internal explanation, that the
woman is “an angry and out-of-control person” (p. 272). This did not hold true for
men. Furthermore, internal or external attribution of anger mediated the effects of
expressing anger on status: When an angry woman provided an external
explanation for her anger, she did not suffer the same loss in perceived status
and competence. Brescoll and Uhlmann (2008) suggest that counterstereotypical
actions provoke negative actions if there is a justification for derogating the
counterstereotypical individual; the researchers note that the workplace, in
particular, provides ample opportunity for observers to find cause for derogation.
Effects of Anger Expression by Frequency – Hypotheses
The above literature has somewhat unclear implications for anger when
considered in the full context of the workplace, yet there has been a lack of
research that can begin to resolve these conflicting implications. Previous work
25
has suggested that even a single instance of anger generally has negative
implications for females. However, presenting anger in the context of information
about the employee’s background and performance may produce results that
differ from much of this previous work. In the absence of empirical results that
can inform this specific approach, I present a series of alternative hypotheses on
the potential effects of my anger manipulations. I first discuss how a single
instance of anger might affect social and job-related outcomes for males and
females, then turn to the potential consequences of frequent anger expression.
Finally, I consider how the object of an angry outburst (unspecified vs. a
subordinate) may influence perceptions of the angry individual.
For women in particular, an isolated instance of anger in the workplace
may not convey the professionalism or authoritativeness associated with anger in
the workplace for men. In fact, a single expression of anger may be sufficient to
call to mind the stereotypical emotional female, who may be perceived as
inappropriate in a workplace setting. The violation of female gender roles, which
prescribe kindness and nurturing, might result in negative social outcomes (such
that individuals do not want to spend time with the angry person) as well as
negative job-related outcomes. Thus, based on previous research, one might
propose the following hypotheses for females:
H1Af: A female who has expressed anger in the workplace once
will be less likely to be promoted than a female who has not
expressed anger in the workplace.
26
H2Af: A female who has expressed anger in the workplace once
will receive lower scores in terms of social trait evaluations and
behavior patterns than a female who has not expressed anger in
the workplace.
The consequences would likely be less severe for males, as some
previous work has found expressing anger to be advantageous for males and
anger does not violate male gender norms. However, the generally neutral tone
of the workplace may nevertheless result in a preference for a neutral employee
rather than one who has expressed anger even a single time. An angry man may
also suffer negative social consequences due to perceived personality problems
based on expression of anger in a neutral setting. This results in the following
two hypotheses:
H1Am: A male who has expressed anger in the workplace once will
be less likely to be promoted than a male who has not expressed
anger in the workplace.
H2Am: A male who has expressed anger in the workplace once will
receive lower scores on in terms of social trait evaluations and
behavior patterns than a male who has not expressed anger in the
workplace.
27
However, it is possible that a single emotional outburst may also serve
some positive function. In an actual workplace setting, isolated expressions of
anger may demonstrate an ability to assert oneself and appear competent,
reducing fears of appearing too nice or being a pushover (in line with female
gender stereotypes). Anger may in fact have a curvilinear relationship with
workplace effectiveness (as Ames and Flynn [2007] find for assertiveness) such
that below a certain threshold, individuals seem ineffective, while above a certain
threshold, they seem antagonistic and even out-of-control. In fact, making it clear
that the single expression of anger is out-of-character may create the impression
that the individual is generally well-behaved (thus making the anger appear more
justified). Tiedens (2001) notes that expression of anger by someone who rarely
exhibits such an emotion may be more impactful than other expressions of anger
(although she does not speculate as to the direction of this effect).
By situating the anger in the context of positive information about the
employee and her capabilities, the effects of anger may be more positive than
those typically observed in anger research. The above speculation suggests that
male employees may similarly benefit from expression of anger; in addition,
anger does not violate male gender norms and has been found to have positive
effects in some previous studies. This would lead to the following set of
alternative hypotheses:
H1Bf: A female who has expressed anger in the workplace once
will be more likely to be promoted than a female who has not
expressed anger in the workplace.
28
H1Bm: A male who has expressed anger in the workplace once will
be more likely to be promoted than a male who has not expressed
anger in the workplace.
It is possible, though arguably less likely, that a rare instance of anger
may have positive social consequences (particularly for males, who may be
presenting an image that matches stereotypes of what males should be like).
Anger may make individuals seem more realistic (rather than just a description
on paper) or suggest a stronger personality or willingness to stand up for oneself
that would appeal to some individuals. If this is the case, then the following
hypotheses would hold true:
H2Bf: A female who has expressed anger in the workplace once
will receive higher scores in terms of social trait evaluations and
behavior patterns than a female who has not expressed anger in
the workplace.
H2Bm: A male who has expressed anger in the workplace once will
receive higher scores in terms of social trait evaluations and
behavior patterns than a male who has not expressed anger in the
workplace.
29
To further explore the nuances involved in expressing anger in the
workplace, I also consider more frequent expressions of anger, essentially
presenting it as a personality trait. Females may be punished for repeated
behavior that violates their gender role. Frequent anger expression may also
make them appear out-of-control. Anger is more acceptable for males than for
females; however, multiple expressions of anger may appear excessive even for
men and suggest that the male is unable to manage his emotions. The frequency
of these reactions may reduce their power (Tiedens, 2001) or suggest that they
are often struggling against threats to their status (Kemper, 1991), which may
result in perceptions of incompetence.
H3Af: A female who has expressed anger in the workplace multiple
times will be less likely to be promoted than a female who has not
expressed anger in the workplace.
H3Am: A male who has expressed anger in the workplace multiple
times will be less likely to be promoted than a male who has not
expressed anger in the workplace.
30
Alternatively, an angry personality may demonstrate that a particular
woman is more in keeping with stereotypes of leaders, subscribing to a more
masculine leadership style. This perceived masculine style may activate
perceptions of the female as competent and cold (a perception consistent with
what Cuddy, Fiske, and Glick [2004] find for female professionals). Thus,
performance reviews that mention frequent anger could be interpreted as
depicting a particular kind of woman, one who exhibits anger more frequently and
thus may be higher status and/or a better leader. Frequent expression of anger
on the part of males may be seen as a demonstration of power and/or status,
which would also result in positive job-related outcomes. Further, since anger is
in line with male gender norms, it may be permissible even if it occurs frequently.
H3Bf: A female who has expressed anger in the workplace multiple
times will be more likely to be promoted than a female who has not
expressed anger in the workplace.
H3Bm: A male who has expressed anger in the workplace multiple
times will be more likely to be promoted than a male who has not
expressed anger in the workplace.
There is no previous research that would suggest positive social outcomes
for repeated angry outbursts. Thus, I present no alternatives to the two
hypotheses regarding social outcomes for frequently angry individuals:
31
H4f: A female who has expressed anger in the workplace multiple
times will receive lower scores in terms of social trait evaluations
and behavior patterns than a female who has not expressed anger
in the workplace.
H4m: A male who has expressed anger in the workplace multiple
times will receive lower scores in terms of social trait evaluations
and behavior patterns than a female who has not expressed anger
in the workplace.
The above hypotheses are also presented in a simplified form in Table
2.1.
Given that anger is more consistent with social and emotional norms for
males than for females, I also propose the following general hypotheses with
respect to gender:
H5: When compared to a non-angry employee of the same gender,
females who express anger in the workplace will be less likely to be
promoted than males who express anger in the workplace.
32
H6: When compared to a non-angry employee of the same gender,
females who express anger in the workplace will have lower scores
in terms of social trait evaluations and behavior patterns than males
who express anger in the workplace.
Study 1 was designed to investigate the above hypotheses. Since Study 2
was designed to clarify the results of Study 1, I describe the rationale and
hypotheses and present the results for Study 2 in Chapter 6, following the
discussion of Study 1’s results in Chapters 4 and 5.
33
Table 2.1. Alternative Hypotheses Regarding Social and Job-Related Effects of Frequency of Anger*
Likelihood of Promotion
Scores on Social Measures
Single (One-Time) Expression of Anger
– H1A(m/f)
+ H1B(m/f)
– H2A(m/f)
+ H2B(m/f)
Frequent Expression of Anger – H3A(m/f)
+ H3B(m/f)
– H4(m/f)
Notes: + indicates a positive outcome (i.e., increased likelihood of promotion, higher scores on social measures) - indicates a negative outcome (i.e., decreased likelihood of promotion, lower scores on social measures) * In all cases, predictions are relative to a neutral consultant.
34
Chapter 3: Experimental Methodology and Design
Role of Context and Format in Presentation of Emotional Expression
The limited conception of anger used in much of past research introduces
several issues. Some studies have used job interview situations to explore
expression of anger (e.g., Tiedens, 2001). However, individuals being
interviewed for a new job should be attempting to make a good first impression,
and the feeling rules for a job interview situation would likely indicate a neutral or
generally positive tone. Indeed, common interview tips include things like “smile”
and ”set a positive tone”, indicating that common knowledge discourages
negative feelings, including anger (Koritko, n.d.).
Overall, interviewees should be attempting to control and/or downplay
emotional responses. Failure to do so could suggest a lack of emotional
intelligence (i.e., the interviewee does not know how to behave in certain social
situations). Interviewees are explicitly encouraged to avoid openly discussing
previous anger incidents because it can indicate that they sometimes lose control
(see, e.g., Stout, n.d.). Thus, even simply mentioning the experience of anger in
a job interview (as in Brescoll & Uhlmann, 2008 and Tiedens, 2001) may be
enough to trigger feeling rules, causing statements of feeling anger to reflect
poorly on the applicant.
This may be especially true for females, who are perceived to be more
emotional (Brody & Hall, 1992). Stereotypes regarding female emotional
responses are widely known (Hess et al., 2005), so the consequences of
activating such stereotypes would likely be apparent to many women. Since the
35
consequences of expressing anger are more severe for females (Rudman &
Glick, 1999), violating these norms in an interview environment would likely have
more significant consequences.
Past research has also often relied on actors to communicate emotions in
a video format (see, for example, Brescoll & Uhlmann, 2008, who use the set of
videotapes created by Tiedens, 2001). Emotions can be inferred based on facial
expressions, voice intonation, and behavior, among other factors (Horstmann,
2003; Tiedens, 2001), and contextual elements such as the situation in which a
facial expression is produced or the social position of the emoter provide
additional information that helps shape ideas about what the person means to
express (Algoe et al., 2000). Notably, gender may influence the emotion that
people perceive; previous work has found that male faces are interpreted as
showing more anger and less fear than women's, even when the actual facial
expression is the same (Algoe et al., 2000), and participants seem to associate
sad faces more often with females than with males (Magee & Tiedens, 2006;
Plant et al., 2004).
Other studies (see, e.g., Algoe et al., 2000; Tiedens et al., 2000) have
used vignettes to present information about emotional responses. The limited
amount of information available in these vignettes affects the conclusions
respondents can draw about individuals. The subject has little knowledge about
the individual other than his/her position and emotional response to a single
situation. Keeping the vignettes simple is necessary for this method of research,
but places significantly more weight on the expression of anger than one might
36
see in an actual workplace setting, where co-workers have repeated interactions
and personal knowledge about each other that can help mediate the effects of
anger. For example, increased availability of information may reduce sex bias in
promotional decisions as compared to hiring decisions (Lyness & Judiesch,
1999). Similarly, Petersen and Saporta (2004) observe that the lack of
information available to employers hiring for entry-level positions makes it more
difficult to overcome discrimination.
Experimental Design
This research aimed to explore anger in terms of gender in a particular
social context, the workplace. The anger in this study was said to be expressed
in a public setting (a meeting) and was introduced via a document that is created
for the purpose of being viewed by others (a performance review). Thus, the
experiments discussed herein were intended to assess how anger, a social
emotion, presented within a social context, affected perceptions of and rewards
granted to males and females.
In these experiments, I provided information about the employees via two
common workplace documents, resumes and performance reviews. Resumes
have been used to explore discrimination in previous studies (see, e.g., Correll,
Benard, & Paik, 2007) and provide a brief introduction to an individual’s
capabilities. Performance evaluations have also been used in past studies (e.g.,
Buttner & McEnally, 1996) and give concrete indicators of job performance. Such
reviews are conducted throughout the tenure of most individuals working in
professional settings (Georgeson & Harris, 1998). The performance evaluations
37
used in this study provided textual feedback and assigned numeric values to the
employees’ performance, which makes it more difficult for respondents to distort
the performance of the employees, particularly females (Heilman, 2001).
Resumes and performance reviews offer a valuable way to present expression of
anger in the context of other information about the individual and introduce the
anger in a way that is consistent with how information about employees is
actually conveyed in a workplace setting.
Consulting served as the occupational context for these studies.
Consulting has been used in previous studies involving gender and occupation
(see, e.g., Cuddy et al., 2004) and was pre-tested as part of a battery of sixteen
occupations and found to be gender neutral.7 Using a gender-neutral occupation
is important due to interaction effects between the perceived gendered nature of
an occupation and expectations for male and female employees regarding, for
example, promotional decisions (Garcia-Retamero & Lopez-Zafra, 2006); in
addition, Eagly et al.’s (1995) meta-analysis demonstrated differences in
leadership effectiveness for males and females when jobs were described with
masculine and feminine characteristics. The employees in this research were
presented as individuals who “have each been with the company for two years as
Consultants and are up for promotion for the same position, Project Manager”.8
Finally, this dissertation examined two frequencies of anger expression
(single and multiple), as the typical study’s use of a single instance limits the
7 Ratings of other occupations in the battery were consistent with those of previous occupational rating studies (e.g., Glick, Wilk, & Perreault, 1995; White, Kruczek, Brown, & White, 1989). 8 These titles and this time frame are, in fact, consistent with the actual promotion process for well-known consulting firms like the Boston Consulting Group (BCG).
38
conclusions that can be drawn about anger expression in the workplace. As
noted previously, individuals typically have repeated interactions with others over
time. The average job tenure in the United States is four years (Bureau of Labor
Statistics, 2008); thus, even if expression of anger were a rare occurrence, one
would expect some individuals to express it more often than others over this time
period. Multiple levels of anger expression may signal vastly different things
about the employees to those who encounter them. A single instance may
appear more justified and have positive results, or may be enough evidence to
suggest the individual has more far-reaching emotional issues. Multiple
expressions of anger may be more likely to suggest emotional issues; however,
they may also be interpreted as evidence of a powerful individual. The only way
to begin unpacking the actual implications of different frequencies of anger
expression is an explicit examination of these frequencies. The wording of the
manipulations is addressed in the discussion of the experimental procedures.
Participant Recruitment
Participants for Study 1 were recruited from Foothill Community College
and Stanford University, while participants from Study 2 were drawn from Foothill
College only. Foothill Community College students were recruited through the
Research Experience Program (REP) in which students enrolled in certain
classes are required to complete experiments in order to receive course credit;
these students receive credit toward this program for participation. Stanford
University students included students who had previously signed up to participate
in experiments through the Stanford Center for Social Research. These
39
individuals were recruited using emails sent through the online experiment
recruitment website.
Most of the research subjects participated in the study in-person; however,
some online subjects were included for experimental efficacy after preliminary
tests suggested that the two procedures (online and in-person) did not differ in
terms of results (see Table 3.1 for a breakdown of the participants by gender and
experiment format for each study and Tables 3.2 and 3.3 for a breakdown by
condition for Studies 1 and 2, respectively). Study 1 was conducted both online
and in-person, while Study 2 was conducted online only; Stanford University
students participated in person only and thus were not included in Study 2. All
online students were drawn from a separate subject pool of online students only
and were run using a slightly modified experimental procedure (described below).
The online participants students tended to be older than the in-person students,
with an average age of 27 for online participants versus 21 for in-person
participants.9 The protocols for the in-person and online versions of Study 1 and
the online-only Study 2 were reviewed and approved by the Stanford Institutional
Review Board.
Study 1 consisted of four conditions resulting from a 2 x 2 design: gender
of consultants (male/female) x expression of anger (single instance/multiple
instances). There were a total of 143 participants (60.8% female participants;
23.1% Stanford students; 29.4% online participants; again, see Table 3.1 for a
9 Whether a particular respondent participated online or in-person was controlled for in all regression analyses, as was whether the student was from Stanford or Foothill. These variables were rarely significant, and the significant differences that were observed are discussed in the context of the overall results.
40
full breakdown of participants in terms of gender and experiment format [online or
in-person] by study and school). Table 3.2 shows the breakdown of these
subjects by condition.
Study 2 consisted of two conditions resulting from a 2 x 1 design; gender
of consultants varied, but all Study 2 subjects evaluated a single instance of
directed anger (this is discussed further in Chapter 6). There were a total of 39
participants (61.5% female participants; 0% Stanford students; 100% online
participants). Table 3.3 shows the breakdown of these subjects by condition.
Subjects from Stanford University and Foothill College did not differ
significantly in terms of ethnicity, age, or gender. Stanford University students
were less likely to major in medical- or nursing-related fields (due to the high
number of nursing students at Foothill College) and more likely to be
science/engineering majors. Randomization checks showed very few differences
by condition or materials in terms of demographic factors like ethnicity, age, and
gender of subject (see Appendix C). These demographic factors were controlled
for in the analyses and had few effects.
In-Person Procedures10
For in-person subjects, upon arrival, respondents were seated in a room
and read a brief introduction in which they were told that the study was intended
“to find out how resumes and performance reviews affect decisions about
10 For examples of the actual materials and questionnaires used in the in-person study, please see Appendix A. For materials and screenshots of the questionnaire formatting used in the online studies, see Appendix B.
41
employees’ job title and salary”.11 Respondents were then presented with two
resumes and performance reviews (labeled A and B) with the contact information
blacked out; these were identified as “actual resumes and performance reviews
from two women (men) working at a consulting firm in the Bay Area”.12
As in previous research using such workplace documents (Correll et al.,
2007), the resumes and reviews subjects were given were not actually resumes
and performance reviews from real people. To develop resumes for the
experiment, several resumes were collected from Careerbuilder.com; the two
resumes used in this experiment were actually each composites of over 20
resumes. Company names were changed so that real companies were not
represented. Performance review forms were developed based on actual
performance reviews taken from real companies, and much of the language on
the performance reviews was taken from a document describing common
phrases used in performance reviews. The performance review documents were
evaluated by former consultants, and the language and information was found to
be believable.13 All subjects assessed either two males or two females to avoid
11 For the text of this script and the other materials used (e.g., the debriefing script), see Appendices A and B. 12 There was little suspicion on the part of either Foothill or Stanford students regarding this backstory (10 subjects [7% of the total subjects for Study 1] were dropped from the study, but only 3 of these [2% of total Study 1 subjects] were dropped due to suspicion). The other 7 subjects were dropped because they expressed confusion about the gender of the participants (e.g., writing “he/she” in their notes) or because they missed the manipulation entirely (i.e., they did not read the performance reviews thoroughly enough). Four of the dropped subjects in Study 1 were online participants, and six were in-person participants. 13 The resumes and performance reviews were also pre-tested in the subject pool with no manipulation (pre-test subjects were given the two performance reviews and two resumes with no mention of anger and asked to evaluate them) to ensure equivalence and found to be equivalent in terms of all measures.
42
triggering suspicion of gender comparisons and to provide a more sensitive
evaluation of the effects of anger.
Participants were randomly assigned to assess either male or female
consultants and either a single instance of anger or a frequent occurrence of
anger. Resumes and performance reviews were randomly assigned as A or B, so
all potential combinations of resumes and performance reviews were possible for
each subject. Finally, either Consultant A or Consultant B was randomly
assigned to be the angry consultant.14 After the brief introduction described
above (see Appendix A), participants were given both resumes and performance
reviews simultaneously and given ten minutes to review the materials and make
notes. After ten minutes, participants were asked if they need more time; those
who declined were given the questionnaire (described in further detail below and
presented in Appendix A). Those who requested more time were given an
additional five minutes to review the documents, then given the questionnaire.
Online Procedures
Every effort was made to make the online study procedure as similar to
the in-person procedure as possible. As with the in-person version of the study,
respondents signed up for a specific time slot. They then emailed the researcher
during their set time slot and were emailed the study information (which was
essentially identical to the script used in person) and asked to “sign” an online
14 All random assignments were generated using www.random.org. Again, see Appendix C for the results of randomization checks conducted at the conclusion of the experiments.
43
consent form (see Appendix B for the text of the emails and the online consent
form).
After the researcher verified the participant’s consent, respondents were
emailed the same two resumes and performance reviews used in the in-person
study (identified as two males or two females) in PDF format. Again, all
resume/performance review/gender assignments were random. The text of the
email sent with the documents instructed participants to take ten to fifteen
minutes to review the materials and make notes in Microsoft Word or a similar
program. Participants were instructed to mail these notes to the researcher when
they were finished in order to receive the link to the final survey. Participants who
exceeded the allotted time received email reminders.
Manipulation15
The manipulations used in the online and in-person versions of Study 1
were identical, as the documents used were exactly the same. For all subjects,
one consultant was neutral: his (her) performance review made no mention of
anger, stating in the “Additional comments” section that “Team members report
that he (she) is generally easy to work with.” The other consultant’s performance
review mentioned expression of anger, occurring either a single time or
frequently. For one-time angry consultants, the “Additional comments” section
noted: “Team members report that he (she) is generally easy to work with, but
recently became angry when his (her) idea was challenged in a meeting. This
15 This section briefly discusses the construction of the manipulation; for the actual performance reviews used to present this information, see Appendix A.
44
seems to have been a one-time occurrence.” For frequently angry consultants,
this manipulation read: “Team members report that he (she) is generally easy to
work with, but often becomes angry when his (her) ideas are challenged in
meetings. This seems to be a frequent occurrence.” This depiction of the angry
incident(s) (ideas being challenged) could be classified as any of a number of
antecedents found to be common in angry incidents in the workplace, including
“Frustration”, “Conflict over work procedures”, or “Perceived threat” (Glomb,
2002), and being disrespected or challenged is a common reason for anger in
the workplace (Sloan, 2004).16
To restrict attention to the particular anger expression mentioned, the
consultants were said to be “generally easy to work with”, and all consultants
received a score of 3.5 out of 5 on “Interpersonal Skills” in the categorical section
of the performance reviews. In addition, to avoid the additional social implications
of confronting or becoming angry at a subordinate or superior (or a male or a
female), the object of the anger was left vague; instead, the anger was said to be
in response to “being challenged”.17 This may also reduce the low-status
implications of the anger-causing incidents, as low-status individuals may be
more likely to experience anger due to poor treatment by others; focusing the
challenge on the ideas rather than the individuals themselves may reduce this
perception. The final sentence (“This seems to have been...”) was meant to
reinforce the single or multiple nature of the anger expression.
16 Previous research has found no gender differences in terms of frequency or type of issues causing anger (Gianakos, 2002). 17 Anger directed at a subordinate was considered in Study 2 (see Chapter 6).
45
Questionnaire
After reviewing the documents, participants were presented with a
questionnaire that included measures intended to capture both job-related and
social outcomes and opinions of participants. The questionnaires for the online
and in-person versions of Study 1 were the same, with minor formatting
differences; the online version was administered using Opinio software. For the
text of the questions included in the survey (both the online and in-person
versions), see Appendix A. For screenshots of the online version, see Appendix
B.
Job-Related Outcomes
Following Tiedens (2001), I used status conferral actions (actions that
provide status or legitimacy), such as deciding whether or not to promote an
individual (a choice between the two consultants) to assess status in a job-
related context. I also assessed how much status, power, and independence
respondents believed the applicants should have using an 11-point scale,
following Tiedens (2001) and Brescoll (personal communication, November 17,
2008). On scales used by Ridgeway et al. (2009), respondents also rated the
applicants on a 7-point scale for dimensions including
competence/incompetence, knowledgeable/unknowledgeable, respected/not
respected, and high status/low status.
On a 5-point scale, respondents reported how similar they felt each
consultant was to the company’s existing project managers (the promotion
position) and how risky they felt it would be to promote each individual.
46
Respondents further reported how likely they would be to ask each employee for
advice about work matters (5-point scale) and noted how many employees they
would place under each consultant. I also included dummy questions asking
respondents to choose which consultant they would promote or want to work
under (i.e., have as their boss). For these questions, respondents were forced to
choose between the consultants. Additional job-related questions and the exact
wording of the measures mentioned here can be found in the questionnaire in
Appendix A.
Social Outcomes
The 7-point rating scale format drawn from Ridgeway et al. (2009) was
also used to capture social measures, asking respondents to rank consultants
along scales including pleasant/unpleasant and likeable/not likeable. I also asked
respondents how likely they would be to be friends with each consultant, enjoy
socializing with each consultant or to ask each consultant for advice about
personal matters (5-point scales). Again, additional questions and the exact
wording of the measures mentioned here can be found in the questionnaire in
Appendix A.
Debriefing
After completing the questionnaire, in-person participants were
interviewed and asked a short series of additional questions about the
consultants and their choice regarding who should be promoted. Respondents
were asked whether they would have preferred to promote both consultants if
such an option were given to them; this was coded as an indicator of their
47
relative opinion of the two consultants and their actual willingness to promote the
angry consultant (when not forced to choose between the two consultants).
Finally, subjects were debriefed about the purpose of the experiment and given
an additional opportunity to ask questions or provide feedback, following the
debriefing procedure used by Ridgeway et al. (2009).
Debriefing of the online participants was conducted using the series of
questions from the in-person version in online form, with text boxes for the
respondents to ask questions or acknowledge that they understood the study
(see Appendix B for screenshots of this debriefing procedure). The researcher
also remained available via email for at least one hour after the study in order to
answer any questions.
48
Table 3.1. Breakdown of Subjects by Gender and Experiment Format N (%) Study 1 Total 142 Female 87 (60.8%) Online 42 (29.4%) Foothill 110 (76.9%) Female 69 (62.7%) Online 42 (38.3%) Female 28 (66.7%) Stanford 33 (23.1%) Female 18 (54.5%) Online 0 (0.0%) Study 2 Total 39 Foothill 39 (100.0%) Female 24 (61.5%) Online 39 (100.0%)
49
Table 3.2. Study 1 Participants, by Condition Male Consultants Female Consultants One-Time Anger N = 38
• 4 Stanford students (10.3%)
• 14 online students (35.9%)
• 26 female respondents (66.7%)
N = 40 • 7 Stanford students
(17.5%) • 14 online students
(35%) • 22 female
respondents (55%) Frequent Anger N = 31
• 7 Stanford students (22.6%)
• 8 online students (25.8%)
• 21 female respondents (67.7%)
N = 33 • 15 Stanford students
(45.5%) • 6 online students
(18.2%) • 18 female
respondents (54.5%)
50
Table 3.3. Study 2 Participants, by Condition Male Consultants Female Consultants One-Time (Directed) Anger
N = 20 • 13 female respondents
(68.4%)
N = 19 • 11 female respondents
(55%) Note: Study 2 included only online Foothill College respondents and had only one condition (a
single instance of directed anger).
51
Chapter 4: Frequency of Anger, Gender, and Promotional Decisions
Who Gets Promoted?
This study obtained information about promotion in two ways. First,
respondents were asked which consultant they would choose to promote in the
questionnaire they received at the end of the experiment by circling either
“Consultant A” or “Consultant B”. Second, respondents were asked directly
during the debriefing if they would promote both consultants “if it were possible to
do so”.18 These two different approaches to the promotional decisions resulted in
very different likelihoods of promoting the angry consultant.
In the first instance (when respondents were forced to choose only one
consultant), just over half of the respondents chose to promote a female who got
angry once over a neutral consultant, compared to the 30% of respondents who
chose to promote a female who got angry frequently over a neutral consultant
(the difference between these groups was statistically significant [z-score = 1.91,
p < .05]; see Table 4.1). The results for male consultants were similar; 44% of
respondents chose to promote males who were angry once over neutral
consultants, while about a quarter of respondents chose to promote a frequently
angry male over a neutral consultant (the difference between these groups was
borderline significant [z-score = 1.54, p < .1]; again, see Table 4.1) Thus, the
means for a forced choice between the two consultants suggest support for
Hypotheses 3Am and 3Af, which predicted negative promotional outcomes for
18 The open-ended responses to this question were coded by the researcher. Most individuals answered a clear “yes” or “no”. When this was not the case, the researcher placed assigned a value based on the verbal response; for example, “probably” was coded as “yes”, while “maybe” or conditional responses (e.g., “if...”) were coded as “no”.
52
frequently angry consultants, and suggest essentially no effect for individuals
who become angry once (i.e., no support for either H1A or H1B).
However, it seems that most respondents were impressed with the
consultants’ qualifications and preferred to promote both individuals if possible.
When provided with the opportunity to promote both consultants, the vast
majority of respondents chose to do so in the one-time anger condition (90% and
96% for females and males, respectively). Respondents were often willing to
overlook even frequent anger when not forced to make a choice between the two
consultants. In fact, 71% of respondents chose to promote both males when one
had expressed anger frequently. For female consultants, 79% of respondents
were willing to promote both consultants in the frequent anger condition. In other
words, in both cases, over half of the respondents who did not choose to
promote the frequently angry consultant were nonetheless willing to do so if it did
not have to be at the expense of the neutral consultant. The difference in
likelihood of choosing to promote both consultants was highly significant across
condition for males (z = 2.73, p < .01), but only borderline significant for females
(z = 1.33, p < .1; see Table 4.1).
z-tests comparing males and females for each condition (e.g., frequently
angry males to frequently angry females) showed no significant effects (all p-
values > .1, see Table 4.2). Thus, Hypothesis 5, which suggested that angry
females would be less likely to be promoted than angry males, is not supported
in simple z-tests.
53
Logistic Regression Analyses of Promotion
Using logistic regression models to consider the log-likelihood of
promoting the angry consultant or both consultants allowed me to control for
additional factors that may affect promotional decisions. The logistic regression
analyses presented in Table 4.3 indicated the importance of frequent anger
compared to a one-time occurrence of anger when making promotional
decisions, providing further support for Hypotheses 3Am and 3Af; an additional
logistic regression shown in this table regarding choosing one consultant as (the
respondent’s) boss also indicated the importance of frequent anger.
As shown in the column labeled “To Promote the Angry Consultant”, when
participants were forced to choose one consultant over the other, comparing a
neutral consultant to a frequently angry consultant rather than a one-time angry
consultant significantly decreased the log-odds of choosing to promote the angry
consultant by 0.86 (p < .05). None of the control variables were significant in this
model. The gender of the consultant also failed to have an effect; angry females
were no more or less likely than angry males to be promoted over non-angry
counterparts. Thus, even when other factors are controlled for, Hypothesis 5
continues to receive no support.
The second model shown in Table 4.3 predicts the log-likelihood of
choosing to promote both consultants. Again, frequency of expressed anger was
an important factor; comparing the neutral consultant to a frequently angry
consultant rather than a one-time angry consultant significantly decreased the
log-odds of choosing to promote the angry consultant by 1.55 (p < .05).
54
Individuals who responded online were more likely to choose to promote both
consultants (b = 1.71, p < .1).19
The results for a similar question, which consultant individuals would
prefer to have as a boss, also reflected the importance of frequency of anger. In
this case, individuals were asked to consider their own working situation rather
than that of others. Comparing the neutral consultant to a frequently angry
consultant rather than a one-time angry consultant significantly decreased the
log-odds of choosing the angry consultant as a boss by 1.04 (p < .05).
The lack of gender effects observed in these analyses fails to provide
support for Hypothesis 5, but considering a series of logistic regression models
with interacted manipulations (consultant gender x frequency of anger) provided
further insight into the relationships between gender, expression of anger, and
promotional decisions. Table 4.4, which considers the forced-choice promotional
decision, indicates that respondents who evaluated frequently angry consultants
(male or female) were significantly less likely than those who evaluated females
who were angry once to be promoted over a neutral consultant. Being angry
frequently decreased the log-odds of promotion by 1.26 (p < .05) and .97 (p < .1)
for males and females, respectively, compared to females who were angry once.
There were no significant differences between any other manipulation groups; in
particular, males who became angry a single time were not more or less likely to
19 This is likely due to the difference in modality for these respondents, who selected a response from a drop-down menu rather than being asked directly. Online effects were also observed in the mediational logistic regressions.
55
be selected over a neutral consultant than males who or females were angry
frequently.
When considering these interactions in terms of the decision to promote
both consultants (shown in Table 4.5), respondents who considered frequently
angry males were least likely to promote both individuals. Comparing a frequently
angry male consultant to a neutral male consultant decreased the log-odds of
choosing to promote both consultants by 1.76 and 2.08 (p-values < .05)
compared to considering a male or female consultant, respectively, who was
angry once to a neutral same-sex counterpart.
Mediational Analyses – Considering Consultant Traits (via Principal Component Analysis)
As this was an exploratory study, I included several measures from a
number of previous studies (e.g., Tiedens, 2001 and Ridgeway et al., 2009) in
addition to introducing measures of my own. To determine how these measures
correlated with each other and in order to create meaningful scales, I used
principal component analysis (PCA) to create variables based on (primarily)
ordinal measures, using first the responses associated with the angry individual
(both angry frequently and angry a single time), then the differences between
responses for the angry individual (again, both frequently and a single time) and
his or her neutral same-sex counterpart. For this dissertation, I chose to focus
primarily on the angry individual and how he or she is perceived as different from
the neutral individual. The ratings for the neutral consultant are given in relation
to those of the angry individual, and this is captured in the PCAs of the difference
56
variables, so I chose not to conduct analyses using only the ratings of the neutral
consultants.
Examination of the scree plots for both PCAs suggested three similar
components in each case (the scree plots are shown in Appendix D, which
provides a more in-depth discussion of the PCAs). The factor loadings for the
variables formed based on ratings of the angry individual and the differences
between the angry individual and the neutral individual are presented in Tables
4.6 and 4.7, respectively. These factor scores were calculated in SPSS using the
results of regression analyses. For both the difference components and the
components based on ratings of the angry individual, the three components
explain over 40% of the variance in the included variables. The Eigenvalues and
percent variance explained for each component are shown in Table 4.8.
I refer to the three components as “Sociability”, “Knowledgeability”, and
“Suitability for Promotion”. “Sociability” can be described as a group of variables
related to the personality of the individual(s), such as rating them as “Pleasant”
and “Likeable”. This component is discussed in greater detail in Chapter 5, which
considers the social effects of anger expression. In this chapter, the
“Knowledgeability” component, which is focused on intelligence, is discussed
briefly due to its association with competence. However, I focus primarily on the
component that I refer to as “Suitability for Promotion”, which is heavily
influenced by variables that essentially assess the risks and benefits of
promoting the individual(s). Variables with large loadings on this component
include being worried about the individual’s qualifications, feeling it is risky to
57
promote the individual, and believing the individual to be respected (see Tables
4.6 and 4.7). As discussed below, these variables together have a significant
influence on how likely respondents are to promote one individual over another.
I conducted an OLS regression (shown in Table 4.9) to determine which
variables influenced this perception of the consultant as a good candidate for
promotion and again found strong support for differentiation between frequent
expression of anger and a single occurrence of anger; individuals who expressed
anger frequently were seen as significantly less suitable for promotion than
individuals who expressed anger a single time (β = -0.33, p < .01). Gender also
had an effect, such that females were perceived to be significantly more suitable
for promotion than males ((β = 0.20, p < .05).20
I used all three components (“Knowledgeability”, “Sociability”, and
“Suitability for Promotion”) to conduct mediational analyses as well. In conducting
these mediational analyses, I hoped to gain insight into how different traits and
beliefs about the consultants influenced respondents’ choices about promoting or
being subordinate to the consultants. In other words, I expected promotional
decisions to be affected by perceptions of the consultants. If expression of anger
causes the angry individual to be viewed differently in terms of intelligence,
personality, or suitability for the job, meditational analyses including these factors
can be used to illustrate more clearly how these perceptions are affected. To find
out how expression of anger made the angry individuals less appealing
candidates, I conducted two sets of mediational analyses, the first using the
20An OLS regression of the factors influencing the difference “Suitability for Promotion” component was not significant (F-statistic < 1, not shown).
58
components created from variables related to the angry consultant and the
second using the components created from the difference between the angry
consultant and the neutral consultant on these same variables.
The results for the components using ratings of the angry consultant are
shown in Table 4.7. As might be expected given the nature of the variable, the
angry consultant’s perceived “Suitability for Promotion” does appear to mediate
the effects of frequency of anger on the promotion of the angry consultant,
increasing the log-odds by 0.58 (p < .05); while frequency of anger was
significant in a logistic regression without this variable (see Tables 4.3 and 4.4), it
was no longer significant in the mediated regression (Table 4.7).
“Suitability for Promotion” did not mediate the effects of frequency of anger
on promotion of both consultants. In this case, the individuals’ likelihood of
promoting the consultant they did not choose to promote in the questionnaire
(typically the angry consultant) was influenced primarily by how often the angry
consultant is angry (p < .05); this variable decreased the log-odds of promoting
both consultants by 1.64.21 The lack of significance for “Suitability for Promotion”
suggests that issues such as whether the consultant is respected or whether the
respondent is worried about the consultant’s qualifications did not factor in to the
decision to promote both consultants.
Finally, while “Suitability for Promotion” had a borderline significant (b = -
0.49, p < .1) effect on choosing the angry consultant as one’s boss, it did not
mediate the effects of frequency of anger, which decreased the log-odds of
21 Online respondents were significantly more likely to choose to promote the angry consultant and to promote both consultants; as noted previously, this is likely an artifact of data collection.
59
selecting the angry consultant as the boss by 1.24 (p < .05) as compared to 1.04
(p < .05) in the original model. In addition, the gender of the consultant became
significant when the mediational components were included in the logistic
regression. Thus, when “Sociability”, “Knowledgeability”, and “Suitability for
Promotion” were controlled for, the log-odds of selecting angry females as one’s
boss over neutral females were increased by 1.28 compared to the odds of
selecting angry males over neutral males (p < .05).22
Factors created from PCA using the differences in ratings between the
neutral and angry consultants mediated the relationship between the
independent variables (gender and frequency of anger of the consultant) and
promotional decisions more successfully (Table 4.8). All three of the difference
factors (which are based on the value for the angry consultant minus the value
for the neutral consultant) significantly increased the log-odds (bs range from
0.99 to 3.06, all p-values < .01) of promoting the angry consultant or preferring
him/her as boss. “Suitability for Promotion” had a particularly strong effect on
promotion of the angry consultant, increasing the log-odds by 3.06. Frequency of
anger had no effect on choosing the angry consultant to be promoted or serve as
one’s boss, but gender of the consultant was significant in selecting a boss when
the mediational components were included. The log-odds of selecting angry
22It is worth noting that the LR-Chi2 is only significant for promotion of the angry consultant over the neutral consultant; the equations for promoting both consultants or selecting the angry consultant as the boss are not significant. Thus, while the effects noted above are significant, they are very subtle.
60
females as the boss over neutral females were increased by 1.39 compared to
the odds of selecting angry males over neutral males (p < .05).
Like the manipulation factors, the difference factors failed to mediate the
relationship between frequency of anger and the decision to promote both
consultants. Instead, frequency of anger remained significant for respondents
making this decision; having an angry consultant who was angry frequently
(rather than a single time) decreased a respondent’s log-odds of choosing to
promote both consultants by 1.56 (p < .1).23
Discussion
The results of these analyses paint a somewhat complicated picture,
particularly with regard to gender. Frequent expression of anger was clearly
damaging to these candidates, though perhaps not as much as one might
expect. When comparing an angry consultant to a neutral consultant, about half
of respondents chose to promote the male or female who expressed anger a
single time. A quarter to almost a third of those evaluating an individual who was
angry frequently chose to promote him or her over a neutral consultant, indicating
that when frequent expression of anger occurred, a non-trivial number of
respondents still valued the angry individual over the neutral individual.
Nonetheless, these individuals were promoted less often than the individuals who
were angry once, and when considering ordinal questionnaire variables
23 A few control variables also became significant when the mediational variables were introduced; Stanford students and science/engineering students were slightly significantly more likely to promote the angry consultant (p < .1), and white students were less likely to promote both consultants (p < .05). These variables are not significant in any other analyses.
61
(organized into components using PCA), individuals who expressed anger
frequently were perceived to be less suitable for promotion than individuals who
expressed anger a single time. Thus, overall, the results of these analyses
support Hypotheses 3Am and 3Af, which predicted negative promotional
outcomes for frequently angry consultants, and allow Hypotheses 3Bm and 3Bf,
which predicted positive promotional outcomes, to be rejected.
However, it seems that most respondents were impressed with the
consultants’ qualifications and were willing to overlook even frequent anger when
not forced to make a choice between the two consultants, as over 70% in the
frequent anger condition chose to promote both consultants. Open-ended
comments from the respondents such as “Kind of a hothead, but gets the job
done” (Female, Frequently Angry Consultant) suggest that even frequent anger
did not necessarily reduce an individual’s perceived competence or intelligence.
In fact, in mediational analyses using components created from ratings of the
angry individual, “Knowledgeability” of the angry consultant had no effect. In
other words, the effects of anger were not apparent in assessments of the angry
consultant’s intelligence. Given that anger has been shown to be associated with
competence (which loaded more highly on this factor than either of the other two)
and intelligence (see Chapter 2), this is not unexpected; frequency of anger did
not significantly affect scores on this “Knowledgeability” component for either the
manipulation or difference variables.
Mediational analysis did suggest that “Suitability for Promotion” of the
angry consultant mediated the effects of frequency of anger on promotion of the
62
angry consultant; while frequency of anger was significant in a logistic regression
without this variable (see Tables 4.3 and 4.4), it was no longer significant in the
mediated regression (Table 4.7). It seems that those who were angry frequently
were perceived as less “promotable” than those who were angry a single time,
and this was conveyed through ratings of the angry consultant on such variables
as being worried about the individual’s qualifications. The effects of frequent
anger appeared to be most evident in the respondent’s judgment about the risks
of promoting such an individual rather than beliefs about that individual’s own
capabilities (or even behavior, as shown by the lack of significance of
“Sociability”).
Regardless of the mechanism, the negative effects of frequent anger
compared to a single occurrence of anger when respondents were making
promotional decisions or choosing someone to have as a boss remain.
Frequency of anger was particularly important when deciding whether both
consultants should be promoted; far fewer respondents agreed to promote the
frequently angry consultant alongside the neutral consultant. In fact, among
those who did choose to promote both consultants in the frequently angry
condition, their comments often reflected ambivalence or an awareness of the
need to address the frequent anger expression. For example, one individual
noted, “I would but giving less responsibility to (Female, Frequently Angry
Consultant)”; others proposed having the two consultants work as a team.
In addition, in mediational analyses, none of the considered components,
even the difference-based components, mediated the effects of frequency of
63
anger on promotion of both consultants. In this case, the individuals’ likelihood of
promoting the consultant they did not choose to promote in the questionnaire
(typically the angry consultant) was influenced primarily by how often the angry
consultant was angry. Some respondents seemed to be simply unwilling to
promote a frequently angry consultant, and the factors influencing this decision
were not captured by these analyses. These respondents may have some
personality trait (such as a strong desire to avoid conflict) that precludes
promoting both consultants. Some comments were indicative of such resistance;
for example, when asked about promoting both consultants, one respondent
simply said, “No, because (Male, Frequently Angry Consultant) gets angry when
his ideas are challenged”. In an open-ended response about the promotional
decision, another respondent who declined to promote both consultants
observed that “ultimately, though (Female, Frequently Angry Consultant) seemed
just as hard-working or more, (she) seemed not to work as well w/ others”.
Taken together, however, these findings suggest few deleterious effects
for individuals who become angry once (i.e., no support for either H1A or H1B);
these individuals were promoted over a neutral consultant about half of the time.
There is likely a threshold for tolerating anger; infrequent anger may be
inconsequential or even beneficial (as discussed below), but these effects may
be lost as others’ perceptions of the individual change. A curvilinear relationship
(as observed between workplace effectiveness and assertiveness by Ames and
Flynn [2007]) is also possible, as almost 50% of individuals chose to promote the
consultant who was angry once over the neutral consultant. Open-ended
64
comments from the respondents suggest that the expression of anger was, at
least in some cases, perceived positively (such as indicative of determination or
devotion). More detailed comparisons (e.g., a neutral individual to a sad
individual, or an individual who expresses anger once but is generally well-liked
to an individual who does not express anger but whose likeability is not
mentioned) would help flesh out this issue.
Logistic regressions and z-test analyses found no effect for the gender of
the consultant; angry females were no more or less likely than angry males to be
promoted over non-angry counterparts. Thus, Hypothesis 5, which suggested
that angry females would be less likely to be promoted than angry males, was
not supported. In fact, a closer look at the findings suggests that angry females
sometimes received a slight advantage over a neutral consultant compared with
angry males, particularly when those females were only angry once.
First, when respondents were given the choice to promote both
consultants, males who were angry once were significantly more likely to be
promoted than males who were angry frequently. Females who were angry once
were only slightly significantly (p = .09) more likely to be promoted than females
who were angry frequently, indicating that the consequences of repeated anger
may be more severe for males than females since most people chose to promote
both consultants when they could. Females who were angry once differed
significantly from females and males who were angry multiple times in terms of
being promoted over a neutral consultant, while males who were angry once did
not. This suggests an additional (albeit very subtle) benefit for females who were
65
angry once; though the difference between them and males who were angry
once did not achieve significance, females who were angry once were able to
distinguish themselves from neutral consultants more than male or female
frequently angry individuals.
Gender also had an effect on the “Suitability for Promotion” component,
such that females were actually perceived to be more suitable for promotion than
males. The gender of the consultant also became significant in determining
which consultant was preferred as boss when the mediational components were
included in the logistic regression; when “Sociability”, “Knowledgeability”, and
“Suitability for Promotion” were controlled for, angry females were more likely to
be selected as the boss over neutral females than angry males over neutral
males. Thus, it is clear that angry males enjoyed no advantages over angry
females compared to a non-angry same-sex counterpart, and in fact were
sometimes disadvantaged.
These gendered results raise some questions when presented alongside
numerous other studies (see Chapter 2 for a literature review) that have found
negative effects for females expressing anger in the workplace. As noted in
Chapter 3, the way in which this study introduced anger expression differed
significantly from previous experiments, and respondents’ comments suggest
that presenting the anger in context did, in fact, change respondents’
perceptions. In particular, many respondents noted that anger expression on the
part of females indicated “passion” or similar strong feelings (e.g., “the incident
when [Female, One-Time Angry Consultant] got angry seems only to indicate her
66
passion and individual vision”). Thus, consistent with stereotypes and even
female social roles, respondents may have viewed females as more emotional
than males; however, when presented with the female’s overall positive work
performance and general likeability, respondents interpreted anger as evidence
that she enjoyed her work and felt strongly about being in the workplace.24 As
females are sometimes viewed as less ambitious and work-focused (Stewart &
Moore, Jr., 1992), expression of anger may thus have worked to their advantage.
Some of these findings, such as angry females being perceived as more
suitable for promotion or being more likely to be chosen as the boss than angry
males, may be due to perceptions of angry females as less threatening than
angry males. People expect to like female employees better than males
(Heilman, Wallen, Fuchs, & Tamkins, 2004; Rudman & Kilianski, 2000), and it
may be that expression of anger in a workplace context, particularly a single
instance of anger, is not sufficient to overcome this expectation. When the
female’s overall competence and likeability is established by the resume and
performance review, female stereotypes may be somewhat beneficial.
In addition, particularly with frequent expression of anger, anger may be
interpreted differently for males and females; repeated experience of anger
(which could be viewed as repeated challenges of one’s status) may be seen as
indicative of low status, particularly for males (Sloan, 2004). In fact, respondents
24 This adherence to stereotypical beliefs might include more wide-ranging beliefs as well that can then be “overlooked” given the positive associations with the individual’s anger. For example, the respondent quoted in the previous sentence observed in the same open-ended comment that “[Female, One-Time Angry Consultant] might be technically less adept than [Neutral Counterpart]” before going on to choose to promote that individual based on the passion she exhibited.
67
may expect women (and their ideas) to be challenged due to their lower status;
thus, as noted above, women who respond strongly to these challenges may
seem passionate and involved. On the other hand, strong, established males
should be less likely to face open challenges or hostility. Those males who do
experience such behavior may be seen as having some problem (e.g., appearing
weak, having bad ideas) that encourages others to treat them with less respect.
According to status characteristics theory, higher performance expectations for
males generally result in them being given more opportunities to speak and being
evaluated more highly (Ridgeway, 1987; Ridgeway, Berger, & Smith, 1985);
however, when these opportunities backfire, the respondent may wonder why.
Open-ended comments suggest that the behavior of these males raised
questions in the mind of some respondents; one respondent noted, “I find it off
(sic) that [Male, Frequently Angry Consultant] was angry about others
challenging his ideas as if he were already the boss”.
In the next chapter, I consider social measures more explicitly and further
analyze gender differences between males and females in terms of their traits
and behaviors.
68
Table 4.1. Chi-Square and z-Test Analyses of Likelihood of Promotion, by Frequency of Anger Within Gender
Promote Angry Consultant Over
Neutral Consultant
Promote Both Consultants if
Possible Female/One-Time Occurrence (N = 40)
N = 21 .53
N = 36 .90
Female/Frequent Occurrence (N = 33)
N = 10 .30
N = 26 .79
χ2 = 3.65* z = 1.91**
χ2 = 1.78 z = 1.33*
Male/One-Time Occurrence (N = 39)
N = 17 .44
N = 37 .95
Male/Frequent Occurrence (N = 31)
N = 8 .26
N = 22 .71
χ2 = 2.38 z = 1.54*
χ2 = 7.45*** z = 2.73***
*p ≤ .1 **p < .05 ***p<.01
69
Table 4.2. Chi-Square and z-Test Analyses of Likelihood of Promotion, by Frequency of Anger Across Gender
Promote Angry Consultant Over
Neutral Consultant
Promote Both Consultants if
Possible Female/One-Time Occurrence (N = 40)
N = 21 .53
N = 36 .90
Male/One-Time Occurrence (N = 39)
N = 17 .44
N = 37 .95
χ2 = 0.63 z = -0.79
χ2 = 0.67 z = 0.82
Female/Frequent Occurrence (N = 33)
N = 10 .30
N = 26 .79
Male/Frequent Occurrence (N = 31)
N = 8 .26
N = 22 .71
χ2 = 0.16 z = -0.40
χ2 = 0.52 z = -0.72
*p ≤ .1 **p < .05 ***p<.01
70
Table 4.3. Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss, Dummy Independent Variables To Promote
the Angry Consultant
To Promote Both
Consultants (if Possible)
To Have the Angry
Consultant as a Boss
Independent Variables Female Employee 0.23 0.35 0.72 Frequent Occurrence of Anger -0.86** -1.55** -1.04** Control Variables Age 22 Years or Younger† - - - 23 Years or Older 0.12 -1.15 -0.22 Female Respondent 0.30 -0.93 0.00 Race White 0.67 -0.55 0.00 Asian 0.28 0.39 0.15 Other Ethnicity† - - - Major Business -0.08 0.66 -0.51 Social Science/Humanities -0.15 -0.29 -0.20 Medical/Nursing -1.21 0.39 -1.09 Science/Engineering 0.58 0.89 -0.20 Undecided† - - - Stanford Participant 0.04 -0.41 -0.98 Online Participant 0.82 1.71* 0.49 Intercept -0.99 3.14*** -0.75 N 136 136 136 LR-Chi2 16.52 20.64* 15.66 Degrees of Freedom 12 12 12 Notes: † Reference Category
*p ≤ .1 **p < .05 ***p<.01
71
Table 4.4. Logistic Regression Analysis of Whether Respondents Would Promote the Angry Consultant (Over the Neutral Consultant), Interacted Independent Variables
1 2 3 4 Independent Variables Female Employee, One-Time Occurrence#
1.26** 0.55 0.97*
Female Employee, Frequent Occurrence# -1.26** -0.71 -0.28
Male Employee, One-Time Occurrence # -0.55 0.71 0.42
Male Employee, Frequent Occurrence# -0.97* 0.28 -0.42
Control Variables Age 22 Years or Younger† - - - - 23 Years or Older 0.10 0.10 0.10 0.10 Female Respondent 0.33 0.33 0.33 0.33 Race White 0.77 0.77 0.77 0.77 Asian 0.34 0.34 0.34 0.34 Other Ethnicity† - - - - Major Business -0.03 -0.03 -0.03 -0.03 Social Science/Humanities -0.10 -0.10 -0.10 -0.10 Medical/Nursing -1.21 -1.21 -1.21 -1.21 Science/Engineering 0.69 0.69 0.69 0.69 Undecided† - - - - Stanford Participant 0.09 0.09 0.09 0.09 Online Participant 0.83 0.83 0.83 0.83 Intercept -0.74 -2.00** -1.29* -1.71** N 136 136 136 136 LR-Chi2 17.60 17.60 17.60 17.60 Degrees of Freedom 13 13 13 13 Notes: # To present a clear picture of how all four categories relate to each other, the regression
was run four times, using each manipulation as a reference category once. The empty cell in each column illustrates which manipulation served as a reference category for that model. Coefficients of control variables and LR-Chi2 values remain the same across the models. † These reference categories were used across all four iterations of the model.
*p ≤ .1 **p < .05 ***p<.01
72
Table 4.5. Logistic Regression Analysis of Whether Respondents Would Prefer to Promote Both Consultants (if Possible), Interacted Independent Variables
1 2 3 4 Independent Variables Female Employee, One-Time Occurrence# 1.02 -0.32 1.76**
Female Employee, Frequent Occurrence# -1.02 -1.34 0.74
Male Employee, One-Time Occurrence# 0.32 1.34 2.08**
Male Employee, Frequent Occurrence# -1.76** -0.74 -2.08**
Control Variables Age 22 Years or Younger† - - - - 23 Years or Older -1.21 -1.21 -1.21 -1.21 Female Respondent -0.93 -0.93 -0.93 -0.93 Race White -0.69 -0.69 -0.69 -0.69 Asian 0.34 0.34 0.34 0.34 Other Ethnicity† - - - - Major Business 0.65 0.65 0.65 0.65 Social Science/Humanities -0.34 -0.34 -0.34 -0.34 Medical/Nursing 0.42 0.42 0.42 0.42 Science/Engineering 0.76 0.76 0.76 0.76 Undecided† - - - - Stanford Participant -0.48 -0.48 -0.48 -0.48 Online Participant 1.76* 1.76* 1.76* 1.76* Intercept 3.30*** 2.28** 3.62*** -2.60*** N 136 136 136 136 LR-Chi2 21.42* 21.42* 21.42* 21.42* Degrees of Freedom 13 13 13 13 Notes: # To present a clear picture of how all four categories relate to each other, the regression
was run four times, using each manipulation as a reference category once. The empty cell in each column illustrates which manipulation served as a reference category for that model. Coefficients of control variables and LR-Chi2 values remain the same across the models. † These reference categories were used across all four iterations of the model.
*p ≤ .1 **p < .05 ***p<.01
73
Table 4.6. Factor Loadings for Principal Component Analysis Using Manipulation Factors
Sociability Component
Suitability for Promotion
Component
Knowledgeability Component
Power Should Have -0.07 0.19 0.17 Independence Should Have
0.06 0.02 0.31
Status Should Have 0.09 0.12 0.06 Competent -0.08 0.15 0.36 Respected 0.16 0.61 0.01 Pleasant 0.69 0.20 -0.11 Powerful 0.09 0.17 0.56 Likeable 0.58 0.42 -0.10 Knowledgeable 0.05 0.21 0.74 High Status 0.17 0.14 0.34 Leader 0.29 -0.15 0.57 Cooperative 0.43 0.52 0.06 Intelligent -0.15 0.20 0.77 Trustworthy 0.24 0.40 0.24 Confident in Qualifications 0.21 0.29 0.31 Confident of Success 0.24 0.39 0.14 Worried About Qualifications
0.13 -0.60 -0.35
Others Believe Qualified 0.12 0.34 0.01 Similar to Existing Managers
0.06 -0.03 0.03
Likely to Enjoy Socializing With
0.80 0.03 -0.02
Likely to Be Friends With 0.79 0.00 0.25 Likely to Ask for Work Advice
0.44 0.38 0.12
Likely to Ask for Personal Advice
0.66 -0.10 -0.09
Likely to Enjoy as Co- Worker
0.69 0.39 0.03
Risky to Promote -0.13 -0.80 -0.19 Employees Assigned To# 0.21 -0.08 0.05 Notes: Based on PCA of all ordinal variables and one continuous variable (marked with #) for the angry consultant. N=125
74
Table 4.7. Factor Loadings for Principal Component Analysis Using Difference
Factors Sociability
Component Suitability for
Promotion Component
Knowledgeability Component
Power Should Have 0.13 0.68 0.31 Independence Should Have
0.14 0.51 0.58
Status Should Have 0.12 0.76 0.27 Competent 0.05 0.08 0.62 Respected 0.26 0.37 0.01 Pleasant 0.73 0.13 0.07 Powerful -0.11 0.27 0.19 Likeable 0.68 0.28 0.00 Knowledgeable 0.00 0.17 0.71 High Status 0.12 0.38 0.29 Leader -0.05 0.26 0.28 Cooperative 0.65 0.21 0.27 Intelligent -0.07 0.07 0.67 Trustworthy 0.60 0.20 0.23 Confident in Qualifications 0.13 0.63 0.18 Confident of Success 0.13 0.72 0.19 Worried About Qualifications
-0.11 -0.45 -0.29
Others Believe Qualified 0.09 0.67 0.04 Similar to Existing Managers
0.03 0.66 -0.07
Likely to Enjoy Socializing With
0.71 0.22 0.00
Likely to Be Friends With 0.75 0.04 0.02 Likely to Ask for Work Advice
0.28 0.55 0.09
Likely to Ask for Personal Advice
0.74 -0.19 -0.17
Likely to Enjoy as Co- Worker
0.77 0.30 -0.05
Risky to Promote -0.29 -0.68 0.02 Employees Assigned To# 0.19 0.16 0.04 Notes: Based on PCA of the difference between the angry consultant and the neutral consultant using all ordinal variables and one continuous variable (marked with #). N = 123
75
Table 4.8. Eigenvalues and Percent Explained for Each Component (Created with Principal Component Analysis) Initial
Eigenvalues
Percent of Variance Explained
Manipulation Knowledgeability 1.80 5.99% Sociability 7.77 25.90% Suitability for Promotion 3.38 11.28% Difference (Angry – Neutral) Knowledgeability 1.48 5.68% Sociability 3.79 14.59% Suitability for Promotion 8.25 31.75% Notes: Based on PCAs of the ratings on all ordinal variables and the values for one continuous variable for the angry consultant (Manipulation) and for the angry consultant minus the neutral consultant (Difference).
76
Table 4.9. Standardized Coefficients for OLS Regression Analyses of the “Suitability for Promotion” Component (Created with PCA using Manipulation Variables) Independent Variables Frequent Occurrence of Anger -0.33*** Female Employee 0.20** Control Variables Age 22 Years or Younger† - 23 Years or Older -0.12 Female Respondent 0.03 Race White -0.14 Asian -0.06 Other Ethnicity† - Major Business 0.07 Social Science/Humanities -0.04 Medical/Nursing -0.06 Science/Engineering -0.15 Undecided† - Stanford Participant -0.05 Online Participant -0.07 N 118 R2 .19 F-statistic 2.00** Degrees of Freedom 12 Notes: † Reference category
*p ≤ .1 **p < .05 ***p<.01
77
Table 4.10. Mediational Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss, Using Manipulation Components To Promote
the Angry Consultant
To Promote Both
Consultants (if Possible)
To Have the Angry
Consultant as a Boss
Independent Variables Female Employee 0.33 0.68 1.28** Frequent Occurrence of Anger -0.67 -1.64** -1.24** Mediational Variables (Manipulation) Knowledgeability Factor 0.03 0.09 0.15 Sociability Factor 0.30 0.30 0.25 Suitability for Promotion Factor 0.58** 0.07 -0.49* Control Variables Age 22 Years or Younger† - - - 23 Years or Older 0.54 -1.47 -0.68 Female Respondent 0.43 -0.93 -0.05 Race White 0.78 -1.13 -0.54 Asian -0.32 -0.55 -0.22 Other Ethnicity† - - - Major Business -0.23 0.78 -0.08 Social Science/Humanities -0.11 -0.24 -0.04 Medical/Nursing -1.08 0.56 -0.81 Science/Engineering 1.01 0.52 0.05 Undecided† - - - Stanford Participant 0.52 -0.29 -1.02 Online Participant 1.30** 1.85* 0.74 Intercept -1.67** 3.57*** -0.87 N 118 118 118 LR-Chi2 27.34** 21.86 20.57 Degrees of Freedom 15 15 15 Notes: † Reference category
*p ≤ .1 **p < .05 ***p<.01
78
Table 4.11. Mediational Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss, Using Difference Components
To Promote the Angry Consultant
To Promote Both
Consultants (if Possible)
To Have the Angry
Consultant as a Boss
Independent Variables Female Employee 0.27 0.35 1.39** Frequent Occurrence of Anger -0.92 -1.56* -0.66 Mediational Variables (Difference) Knowledgeability Factor 0.94*** 0.30 0.99*** Sociability Factor 1.55*** 0.32 1.21*** Suitability for Promotion Factor 3.06*** 0.33 1.78*** Control Variables Age 22 Years or Younger† - - - 23 Years or Older 0.84 -1.16 -0.83 Female Respondent 0.62 -0.89 0.21 Race White 0.59 -1.94** -0.65 Asian -0.86 -1.28 0.05 Other Ethnicity† - - - Major Business 0.34 0.21 0.46 Social Science/Humanities 0.31 -0.65 0.71 Medical/Nursing -1.09 0.18 0.30 Science/Engineering 1.76* 0.25 0.43 Undecided† - - - Stanford Participant 1.81* -0.48 -0.56 Online Participant 1.65* 1.93* 0.37 Intercept -2.90*** 4.75*** -2.64** N 116 116 116 LR-Chi2 73.12*** 26.49** 42.00*** Degrees of Freedom 15 15 15 Notes: † Reference category
*p ≤ .1 **p < .05 ***p<.01
79
Chapter 5: Frequency of Anger, Gender, and Social Measures
Measuring Sociability
Capturing the perceived sociability of consultants is somewhat more
complex than considering relatively clear-cut issues like which consultant should
be promoted. This study assessed social measures primarily through ordinal
measures of the consultants. Respondents were asked to rate the consultants on
various scales such as “pleasant-unpleasant” and “likeable-not likeable”, as well
as to indicate the likelihood of several behaviors or experiences (e.g., how likely
the respondent would be to enjoy socializing with each consultant).25
Principal Component Analysis (PCA) suggested that responses to several
social-related questions did in fact cluster together as a single component; this is
discussed in greater detail below, along with the results of OLS regressions using
this “Sociability” component. As noted in Chapter 4, I conducted two different
principal component analyses to create meaningful scales using primarily ordinal
variables, based on first the variables associated with the angry individual, then
the differences between the angry individual and his or her neutral same-sex
counterpart. The ratings for the neutral consultant are given in relation to those of
the angry individual, which is captured in the PCAs of the difference variables;
thus, I did not conduct analyses using only the ratings of the neutral consultants.
Scree plots for each PCA suggested three components, which were similar
across the two PCAs. Two of these, “Knowledgeability” and “Suitability for
25 The questionnaire administered to respondents can be found in Appendix A.
80
Promotion”, are discussed in Chapter 4 (Appendix D includes further discussion
of the PCAs and the resulting components).
In this chapter, I focus on the “Sociability” components. The amount of
variation explained by the “Sociability” components for both the manipulation and
difference PCAs was substantial (25.90% and 14.59%, respectively; see Table
5.1 for Eigenvalues). Factors with large loadings on this variable were also
instructive; Table 5.2 lists all of the variables with loadings larger than .6 (positive
or negative) for the manipulation and difference “Sociability” components (see
Appendix D or Tables 4.6 and 4.7 for the complete factor loadings for these and
other components). The factors shown in this table clearly seem to represent a
particularly social dimension; in fact, a work-focused question involving social
interaction (“Likely to Ask for Work-Related Advice”) was excluded, while its more
social counterpart (“Likely to Ask for Personal Advice”) loaded highly.
These “Sociability” components were used in the analysis in two ways.
First, OLS regressions were used to consider what variables impacted the social
perception of the consultants (as measured by the “Sociability” components).
Second, those variables that loaded highly on the “Sociability” components were
compared within-subjects to determine significant differences between the angry
consultant and his or her neutral same-sex counterpart.
OLS Regression Analysis of Factors Influencing Sociability
I conducted OLS regressions using the “Sociability” components to
determine which variables influenced a socially positive perception of the angry
consultant (or the difference between the angry consultant and a neutral
81
counterpart) (see Table 5.3). The OLS regression using the manipulation
component found that the gender of the consultants and whether the angry
consultant was angry frequently or a single time did not affect perceived
sociability. Asians and Stanford University students were significantly more likely
to perceive the angry consultant as more sociable (βs = 0.24 and -0.27,
respectively; p < .05).
In contrast, an OLS regression predicting the “Sociability” component
based on the difference between the angry consultant and the neutral consultant
found a significant negative relationship between frequency of anger and
sociability. Individuals who expressed anger frequently were perceived as
significantly more different from their neutral counterparts in terms of social
measures than individuals who expressed anger a single time. Thus, while
frequent anger did not affect ratings of the angry consultants themselves, it did
affect the perception of those angry consultants relative to a neutral consultant.
Gender had no effect in either regression, so Hypothesis 6 (that females
would receive significantly lower scores on sociability measures than males) was
not supported.
Significance Tests Comparing the Sociability of the Angry and Neutral Consultants
The ratings of the consultants were analyzed using analysis of variance
(ANOVA) with ratings of the neutral vs. angry consultant as a within-subject
factor and gender of consultant and frequency of anger as between-subjects
factors. As noted above, I restricted the analyses to those variables that loaded
highly on the “Sociability” component(s) from the PCAs (shown in Table 5.2).
82
First, I considered trait ratings (7-point scales) for pleasantness, likeability,
cooperativeness, and trustworthiness, using repeated measures ANOVA
(RMANOVA) to compare ratings of the angry consultant to ratings of the neutral
consultant (see Table 5.4 for the means and standard deviations of these
variables and Table 5.5 for the results of the RMANOVA). In all cases, the
gender of the consultants had no effect on trait ratings, nor did the interaction
between gender and frequency of anger. However, significant between-subjects
differences according to the manipulation (frequent anger vs. single occurrence
of anger) were observed for all four traits, indicating overall differences between
the two conditions (Fs > 4, p < .05). Within-subjects differences were also
apparent; respondents did rate the two consultants significantly differently in
terms of all four traits considered, particularly pleasantness, likeability, and
cooperativeness (Fs > 90, p < .001).
The interactions between the four traits and frequency of anger were also
significant (Fs > 17 for all but trustworthiness [F = 4.70], p < .05), indicating larger
differences between the neutral consultant and the angry consultant among
respondents who assessed a frequently angry consultant rather than a
consultant who was angry once. No main or interaction effects (either between-
or within-subjects) were found when considering the gender of the consultant in
relation to these traits.
Next, I considered differences in how respondents perceived certain
potential encounters with the consultant (5-point scales; see Table 5.6 for the
means and standard deviations of these variables and Table 5.7 for the results of
83
the RMANOVA). These questions asked how likely the respondent would be to
enjoy socializing with, be friends with, ask personal advice from, and enjoy
working with each consultant. In other words, rather than focusing on traits, these
questions focused on behaviors and workplace encounters.
As with the trait ratings, no main between-subjects effect was observed for
gender of the consultants; however, the interaction term for gender and
frequency of anger did result in a significant difference across subjects in terms
of likelihood of enjoying having the consultant as a co-worker (F = 4.40, p < .05).
Expressing anger had less impact on the likelihood of enjoying having the
consultant as a co-worker (i.e., the ratings for the angry consultant and the
neutral consultant are more similar) when female consultants were evaluated
rather than males, particularly in the frequent anger condition. However, unlike
the trait ratings, frequency of anger generally had no significant overall effect on
these behavioral measures. Again, there was one exception: likelihood of asking
the consultant for personal advice (F = 6.78. p < .05). Respondents who
considered frequently angry and neutral consultants were significantly less likely
to ask the consultants for advice in relation to respondents evaluating neutral
consultants alongside consultants who were angry once.
Large within-subjects effects were observed. Respondents expressed
significantly different likelihoods for the two consultants for all four behaviors
considered (Fs range from 38.74 to 100.37, all p-values < .01); the interactions
between the behaviors and frequency of anger were also significant (Fs range
from 7.25 to 21.47, all p-values < .01), with larger differences between the neutral
84
consultant and the angry consultant among respondents who assessed a
frequently angry consultant rather than a consultant who was angry once.
Likelihood of being friends with the consultant and enjoying having the consultant
as a co-worker interacted significantly with the gender of the consultant (Fs =
3.98 and 11.70, respectively; p < .05), and likelihood of being friends with the
consultant (F = 6.03, p < .05) and asking the consultant for personal advice (F =
3.77, p < .1) interacted with both frequency of anger and gender of the
consultant.
Overall, these analyses provide support for Hypotheses 2Af and 2Am,
which suggested that females and males who expressed anger once would
receive lower scores on social measures than their non-angry same sex
counterparts. These analyses also support Hypothesis 4, which predicted that
females (f) and males (m) who expressed anger frequently would receive lower
scores on social measures than their non-angry same sex counterparts.26 These
analyses are further evidence of the lack of support for Hypothesis 6, which
suggested that females who express anger in the workplace would have less
positive social outcomes than males who express anger in the workplace.
Discussion
Although assessing social measures can be difficult, PCA indicated that
responses to a series of ordinal variables did form a consistent pattern in terms
of rating the sociability of consultants. These variables included both ratings of
26 As noted in Chapter 2, Hypothesis 4 was not presented in terms of alternative hypotheses, as there was no plausible explanation for frequent expressions of anger making someone seem more sociable (more likeable, for example) than a non-angry individual.
85
traits, like “pleasant-unpleasant” and “likeable-not likeable”, and behavioral
measures (e.g., how likely the respondent would be to enjoy socializing with each
consultant).
The results of analyses using these social measures indicate significant
detrimental effects for angry consultants in relation to neutral consultants, as
angry consultants received significantly lower scores on all social measures than
their neutral counterparts.27 Angry consultants were perceived as significantly
less pleasant, likeable, cooperative, and trustworthy than neutral consultants.
With regard to behaviors, respondents felt they were less likely to enjoy
socializing with the angry consultant or having him/her as a co-worker and were
less likely to ask the angry consultant for personal advice or be friends with
him/her than with the neutral consultant. Overall, it is clear that expressing anger
in the workplace has negative social consequences. While anger may be a
relatively typical workplace emotion (Fitness, 2000), it appears that it can
damage social relationships with co-workers, even when the angry outburst
occurs only once and is witnessed rather than personally experienced.
However, it is worth noting that the differences between the angry
consultant and the neutral consultant tended to be smaller when the consultant
was only angry once. In other words, these negative perceptions and
expectations for the angry consultant were particularly important for consultants
27 The between-subjects effects observed for all four traits and likelihood of asking the consultant for personal advice are probably due primarily to the nature of the scales; respondents positioned the consultants relative to one another, so if the frequently angry consultant received a particularly low rating, the rating of the neutral consultant would be lower than it might have been without the “anchor” of the frequently angry consultant. The larger scales (7-points vs. 5-points) explain why this occurred more often for traits than for behaviors.
86
who are angry frequently as compared to those who are angry a single time.
Significant interactions between the different traits and frequency of anger
indicated that individuals who were angry frequently were considered less
pleasant, likeable, cooperative, and trustworthy compared to a neutral consultant
than individuals who were angry once. Frequently angry individuals also suffered
more than one-time angry individuals relative to neutral consultants in terms of
behavioral measures regarding socializing, working with, making friends with,
and asking for personal advice. Overall, individuals who expressed anger
frequently were perceived as significantly more different from their neutral
counterparts in terms of sociability than individuals who expressed anger a single
time.
Thus, while anger is seen as negative overall, people seem to be more
forgiving of (or more willing to overlook) a single outburst than frequent
expression of anger; the results with regard to promotion discussed in Chapter 4
support this conclusion. Open-ended comments showed a willingness on the part
of some respondents to excuse the anger (noting that it happened only once) or
even see it as a positive; for example, one respondent commented in the
debriefing that the anger demonstrated that “(Female, Angry Consultant) knows
her position and knows what she’s doing. She’s more confident” (Single Anger
Condition).
Gender had no effect (either directly or through interactions) on any of the
trait ratings. In contrast with what previous research or social roles might
suggest, female consultants were not considered more pleasant, likeable,
87
cooperative or trustworthy in general, and angry female consultants were not
rated differently than angry male consultants on these measures. Gender also
did not affect the perceived likelihood of enjoying socializing with the
consultant(s).
Significant interactions indicated that the relationships between gender,
frequency of anger, and the other three behaviors were more complicated.
Expressing anger had less impact on the likelihood of being friends with the
consultant (i.e., the ratings for the two consultants are more similar) when female
consultants were evaluated rather than males, particularly in the frequent anger
condition. Similar (albeit less significant) relationships were observed with regard
to asking the consultant for personal advice and enjoying having the consultant
as a co-worker.
These interactions indicate that other factors may be at work when
respondents decide whether to respond positively to these questions; for
example, perhaps positive feelings about female co-workers are based less on
behavior and more on the respondent’s personal feelings about women in the
workplace or women in general. Social desirability (Edwards, 1957) may also
play a role; that is, respondents may be less willing to “punish” angry female
consultants or rate them differently for fear of appearing to favor (or disfavor)
assertive women. Nonetheless, these effects are generally small and do not
suggest any real explicit advantages (or disadvantages) for women in the
workplace.
88
Table 5.1. Eigenvalues and Percent Explained for “Sociability” Components Initial
Eigenvalues
Percent of Variance Explained
Sociability Component Manipulation 7.77 25.90% Difference (Angry – Neutral) 3.79 14.59% Note: Based on PCAs of the ratings on all ordinal variables and the values for one continuous variable for the angry consultant (Manipulation) and for the angry consultant minus the neutral consultant (Difference).
89
Table 5.2. Variables with Large Loadings on the “Sociability” Factor Variable Factor Loading
(Manipulation) Factor Loading
(Difference) Pleasant 0.69 0.73 Likeable 0.58 0.68 Cooperative 0.43 0.65 Trustworthy 0.24 0.60 Likely to Enjoy Socializing With 0.80 0.71 Likely to Be Friends With 0.79 0.75 Likely to Ask for Personal Advice
0.66 0.74
Likely to Enjoy as Co-Worker 0.69 0.77
90
Table 5.3. Standardized Beta Coefficients for OLS Regression Analyses of “Sociability” Factors (Created with PCA) Sociability
Manipulation Factor
Sociability Difference
Factor Independent Variables Frequent Occurrence of Anger -0.13 -0.33*** Female Employee 0.02 0.16 Control Variables Age 22 Years or Younger† - - 23 Years or Older -0.05 -0.08 Female Respondent -0.08 -0.09 Race White 0.03 0.08 Asian 0.24** 0.21** Other Ethnicity† - - Major Business -0.08 -0.13 Social Science/Humanities 0.01 -0.01 Medical/Nursing -0.09 -0.10 Science/Engineering 0.12 -0.08 Undecided† - - Stanford Participant -0.27** -0.07 Online Participant 0.09 0.12 N 118 116 R2 .17 .22 F-statistic 1.72* 2.40*** Degrees of Freedom 12 12 Notes: † Reference category
*p ≤ .1 **p < .05 ***p<.01
91
Table 5.4. Means and Standard Deviations for Trait Variables Used in RMANOVAs Mean (SD) Pleasant Angry Consultant 4.50 (1.67) Angry Once 4.96 (1.50) Angry Frequently 3.92 (1.70) Female Consultant 4.34 (1.55) Angry Once 4.95 (1.40) Angry Frequently 3.64 (1.43) Male Consultant 4.65 (1.78) Angry Once 4.97 (1.61) Angry Frequently 4.23 (1.92) Neutral Consultant 5.92 (1.18) Relative to Angry Once 5.86 (1.20) Relative to Angry Frequently 5.98 (1.17) Female Consultant 5.93 (1.17) Relative to Angry Once 5.98 (1.10) Relative to Angry Frequently 5.88 (1.27) Male Consultant 5.90 (1.20) Relative to Angry Once 5.74 (1.29) Relative to Angry Frequently 6.10 (1.06) Likeable Angry Consultant 4.70 (1.59) Angry Once 5.21 (1.25) Angry Frequently 4.08 (1.74) Female Consultant 4.69 (1.48) Angry Once 5.18 (1.19) Angry Frequently 4.12 (1.60) Male Consultant 4.70 (1.71) Angry Once 5.23 (1.33) Angry Frequently 4.03 (1.91) Neutral Consultant 5.91 (1.17) Relative to Angry Once 5.86 (1.22) Relative to Angry Frequently 5.97 (1.10) Female Consultant 5.78 (1.22) Relative to Angry Once 5.85 (1.20) Relative to Angry Frequently 5.70 (1.26) Male Consultant 6.04 (1.10) Relative to Angry Once 5.87 (1.26) Relative to Angry Frequently 6.26 (0.82)
92
Table 5.4 (continued). Means and Standard Deviations for Trait Variables Used in RMANOVAs
Cooperative Angry Consultant 4.60 (1.69) Angry Once 5.14 (1.38) Angry Frequently 3.92 (1.81) Female Consultant 4.56 (1.54) Angry Once 5.08 (1.27) Angry Frequently 3.94 (1.62) Male Consultant 4.64 (1.85) Angry Once 5.21 (1.51) Angry Frequently 3.90 (2.02) Neutral Consultant 6.04 (1.19) Relative to Angry Once 5.90 (1.29) Relative to Angry Frequently 6.22 (1.04) Female Consultant 5.88 (1.21) Relative to Angry Once 5.68 (1.38) Relative to Angry Frequently 6.12 (0.93) Male Consultant 6.22 (1.15) Relative to Angry Once 6.13 (1.15) Relative to Angry Frequently 6.33 (1.15) Trustworthy Angry Consultant 5.51 (1.41) Angry Once 5.80 (1.21) Angry Frequently 5.14 (1.55) Female Consultant 5.55 (1.40) Angry Once 5.88 (1.14) Angry Frequently 5.15 (1.60) Male Consultant 5.46 (1.42) Angry Once 5.72 (1.30) Angry Frequently 5.13 (1.53) Neutral Consultant 5.88 (1.32) Relative to Angry Once 5.94 (1.33) Relative to Angry Frequently 5.79 (1.32) Female Consultant 5.90 (1.27) Relative to Angry Once 6.00 (1.22) Relative to Angry Frequently 5.79 (1.34) Male Consultant 5.86 (1.39) Relative to Angry Once 5.90 (1.45) Relative to Angry Frequently 5.80 (1.32) Note: All variables are 7-point scales, with higher scores indicating a higher rating on the trait.
93
Table 5.5. RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Traits (F values) Pleasant Likeable Cooperative Trustworthy Factor Between subjects Female Employee (FE) 0.64 0.55 1.03 0.11 Frequency of Anger (FA) 5.48** 7.47*** 5.76** 4.14** FE X FA 1.79 0.29 .30 0.10 Within subjects Trait (T) 112.10*** 94.99*** 109.73*** 12.17*** T X FE 1.28 1.43 0.96 0.03 T X FA 17.24*** 22.63*** 27.85*** 4.70** T X FE X FA 0.05 1.66 0.02 0.00 *p ≤ .1 **p < .05 ***p<.01
94
Table 5.6. Means and Standard Deviations for Behavioral Variables Used In RMANOVAs
Mean (SD) Enjoy Socializing With _______ Angry Consultant 3.05 (1.04) Angry Once 3.22 (0.95) Angry Frequently 2.84 (1.12) Female Consultant 3.04 (0.98) Angry Once 3.20 (0.88) Angry Frequently 2.85 (1.06) Male Consultant 3.06 (1.11) Angry Once 3.23 (1.04) Angry Frequently 2.84 (1.19) Neutral Consultant 3.76 (0.89) Relative to Angry Once 3.61 (0.95) Relative to Angry Frequently 3.94 (0.77) Female Consultant 3.71 (0.92) Relative to Angry Once 3.62 (1.03) Relative to Angry Frequently 3.82 (0.77) Male Consultant 3.80 (0.86) Relative to Angry Once 3.59 (0.88) Relative to Angry Frequently 4.06 (0.77) Be Friends With ________ Angry Consultant 2.90 (0.88) Angry Once 2.94 (0.85) Angry Frequently 2.86 (0.92) Female Consultant 3.00 (0.82) Angry Once 3.00 (0.85) Angry Frequently 3.00 (0.79) Male Consultant 2.80 (0.94) Angry Once 2.87 (0.86) Angry Frequently 2.71 (1.04) Neutral Consultant 3.43 (0.95) Relative to Angry Once 3.30 (1.02) Relative to Angry Frequently 3.59 (0.85) Female Consultant 3.32 (0.91) Relative to Angry Once 3.30 (1.07) Relative to Angry Frequently 3.33 (0.69) Male Consultant 3.56 (0.99) Relative to Angry Once 3.31 (0.98) Relative to Angry Frequently 3.87 (0.92)
95
Table 5.6 (continued). Means and Standard Deviations for Behavioral Variables Used In RMANOVAs
Ask ______ for Personal Advice Angry Consultant 2.15 (1.09) Angry Once 2.06 (1.10) Angry Frequently 2.25 (1.07) Female Consultant 2.16 (1.11) Angry Once 2.13 (1.07) Angry Frequently 2.21 (1.17) Male Consultant 2.13 (1.08) Angry Once 2.00 (1.15) Angry Frequently 2.29 (0.97) Neutral Consultant 2.63 (1.23) Relative to Angry Once 2.32 (1.14) Relative to Angry Frequently 3.02 (1.24) Female Consultant 2.63 (1.18) Relative to Angry Once 2.50 (1.18) Relative to Angry Frequently 2.79 (1.19) Male Consultant 2.63 (1.29) Relative to Angry Once 2.13 (1.08) Relative to Angry Frequently 3.23 (1.26) Enjoy ______ as Co-Worker Angry Consultant 3.09 (0.96) Angry Once 3.34 (0.78) Angry Frequently 2.78 (1.06) Female Consultant 3.14 (0.95) Angry Once 3.48 (0.75) Angry Frequently 2.73 (1.01) Male Consultant 3.04 (0.97) Angry Once 3.21 (0.80) Angry Frequently 2.84 (1.13) Neutral Consultant 3.87 (0.74) Relative to Angry Once 3.78 (0.67) Relative to Angry Frequently 3.98 (0.83) Female Consultant 3.77 (0.75) Relative to Angry Once 3.80 (0.61) Relative to Angry Frequently 3.72 (0.91) Male Consultant 3.99 (0.73) Relative to Angry Once 3.77 (0.74) Relative to Angry Frequently 4.26 (0.63) Note: All variables are 5-point scales.
96
Table 5.7. RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Likelihood that Respondent Would... (F values)
Enjoy Socializing
With (Consultant)
Be Friends With
(Consultant)
Ask (Consultant) for Personal
Advice
Enjoy (Consultant)
as Co-Worker
Factor Between subjects Female Employee (FE) 0.19 0.05 0.01 0.58 Frequency of Anger (FA) 0.02 0.64 6.78** 2.41 FE X FA 0.20 0.46 2.30 4.40** Within subjects Behavior (B) 68.50*** 62.65*** 38.74*** 100.37*** B X FE 0.28 11.70*** 0.20 3.98** B X FA 15.37*** 7.25*** 10.01*** 21.47*** B X FE X FA 0.80 6.03** 3.77* 0.30 *p ≤ .1 **p < .05 ***p<.01
97
Chapter 6: Directed Anger and Workplace Outcomes (Study 2)
Theoretical Justification
The results of Study 1 discussed in Chapters 4 and 5 indicated that the
relationship between gender and anger in the workplace is complex. In particular,
as indicated by both quantitative analyses and qualitative comments, women
expressing anger in the workplace do not seem to be perceived as unilaterally
negatively as previous studies would suggest. This study helped demonstrate the
importance of context by providing more information about each individual’s work
history and performance than have previous studies. However, as part of this
new approach, information about the expression of anger was kept deliberately
vague. Thus, it remains to be seen how further information about expression of
anger might affect the results of Study 1, particularly those relating to gender.
In Study 2, I took the opportunity to begin exploring the object of the anger
as another dimension of workplace anger. In Study 1, females expressing anger
were not significantly different from angry males in terms of their likelihood of
receiving a promotion and in fact were sometimes slightly advantaged (see
Chapter 4). Females were also not generally more likely than males to
experience negative social outcomes when compared to a same-sex non-angry
counterpart (although some interaction effects were observed, see Chapter 5).
However, the object of the anger in Study 1 was left ambiguous. Responding
angrily to being challenged, particularly in the absence of status effects
introduced through workplace relationships (like superior/subordinate), is
arguably one of the most legitimate workplace behaviors, and being challenged
98
is a common source of workplace anger (Glomb, 2002; Sloan, 2004). In fact,
failure to stand up for oneself or appear dedicated may have negative
consequences, as evidenced by concerns expressed by respondents in open-
ended comments in Study 1. For example, in explaining his impression of the
neutral consultant, one respondent noted, “I’d choose (Male, Neutral Consultant),
but he seems more laid back than the other one and some ambition/emotion is
needed...” (Single Anger Condition, Study 1).
As women have more difficulty establishing legitimate leadership in the
workplace (see Chapter 2 for a more detailed approach to this issue and others
discussed here), expressing anger directly at another individual may be more
damaging to their job and social prospects. Women may be seen as having the
right to express anger generally, such as in defending their ideas, but addressing
that anger to someone else introduces a (further) element of status by placing
the woman in a more superior position, or at least makes her appear as though
she believes herself to be superior. As seen in Ridgeway et al. (2009), women
are aware that acting on their status by treating others differently can be
problematic.
In addition, while women are expected to be more pleasant and friendly
(see, e.g., Ridgeway & Johnson, 1990), expression of anger in general,
particularly when it occurs only once, may be less likely to damage this
perception; indeed, Study 1 found no difference between males and females in
terms of traits like cooperativeness and likeability, even when these individuals
expressed anger (see Chapter 5). Standing up for one’s ideas, especially when it
99
is not at others’ expense, may not indicate an unpleasant person. Furthermore, in
Study 1, the way in which the anger was expressed was not described, so the
individuals could conceivably have exhibited the anger in a more socially
acceptable way (e.g., sitting quietly for the rest of the meeting or scowling for a
few seconds) rather than displaying anger verbally or physically.
Introducing an object of anger (a subordinate) may alter how the angry
outburst is perceived. Women may be seen as less pleasant and likeable when
their anger is explicitly directed at another person. Even if this other person is a
subordinate, female leaders are expected to be more approachable and use
behaviors like avoiding dominant speech and making an effort to seem group-
oriented (Ridgeway, 1982; Yoder, 2001).
Anger is more socially acceptable for males and can be used more
effectively in the workplace (see, e.g., Hess et al., 2005; Brescoll & Uhlmann,
2008; Tiedens, 2001); thus, while males may be free to express anger at
subordinates, females may be constrained by social role expectations requiring
them to be friendly (Eagly & Karau, 2002). Additionally, as noted previously,
females face more severe consequences when violating social norms in general
(e.g., Carli et al., 1995). Study 1 demonstrated that anger, while a common
workplace emotion, still resulted in negative social and, in some cases,
promotional, outcomes (see Chapters 4 and 5). Directing the anger at another
person further increases the social risks and norm violations associated with the
anger, potentially making it more problematic for women. Thus, although no
effect was observed for anger expression when the object of that anger is left
100
ambiguous, specifying that the anger was directed at a subordinate may change
how the angry individual is perceived.
Experimental Design and Hypotheses
Study 2 was conducted as a preliminary exploration of these additional
considerations, and only a single instance of anger was explored. This was done
for several reasons. First, Study 2 was conducted as an exploratory study, meant
to build on the results of Study 1. In fact, Study 2 was conducted concurrently
with Study 1 and used the same materials with one variation (described below),
allowing for comparability between studies. Second, the results of Study 1
indicated that individuals who became angry multiple times suffered harsher
consequences than those who were angry once; thus, it is highly likely that this
would just be magnified by presenting the anger as directed at a particular
individual. Finally, repeatedly becoming angry at a particular individual can
introduce a number of other factors, such as that individual’s competence, and
seems likely to suggest poor social skills on the part of the angry person or a
personality clash between two particular individuals rather than the mere hot-
headedness respondents seemed to assume in Study 1.
The participants in Study 2 were drawn from the pool of online Foothill
College students (those who participated in Study 1 were excluded), and the
study was run concurrently with Study 1.28 Study 2 included a total of 39
participants, 20 who evaluated female consultants and 19 who evaluated male
28 The first Study 2 subject was run after the results from approximately 50 Study 1 subjects suggested a lack of gender effects. Almost 100 additional Study 1 subjects were run following those initial findings, with Study 2 being conducted simultaneously with the first half of those subjects.
101
consultants (see Chapter 3 for a discussion of the recruitment process and Table
3.3 for a breakdown of subjects by condition). Like Study 1, Study 2 participants
assessed neutral consultants alongside consultants who exhibited anger, but in
Study 2, the anger was said to be directed at another person. Thus, the
procedure for Study 2 was identical to the online version of the single anger
manipulation in Study 1 (described in Chapter 3; sample materials and
screenshots can be found in Appendix B) with the exception of the description of
the angry incident in the manipulated performance review. Rather than noting
that the consultant “recently became angry when his (her) idea was challenged in
a meeting”, the manipulated performance review for Study 2 stated that the
individual “recently got angry at someone who works for him (her)”. In both Study
1 and Study 2, the performance review then concluded with “This seems to have
been a one-time occurrence.”
Although the results of Study 1 did not show gender effects (i.e.,
Hypotheses 5 and 6 were not supported), I expect introducing an object of anger
to have more negative consequences for women than for men. Thus, I present
the following hypotheses, based on Hypotheses 5 and 6 (additions to
Hypotheses 5 and 6 are italicized and underlined):
H7: When compared to a non-angry employee of the same gender,
females who express anger in the workplace directed at a
subordinate will be less likely to be promoted than males who
express anger in the workplace.
102
H8: When compared to a non-angry employee of the same gender,
females who express anger in the workplace directed at a
subordinate will receive lower scores in terms of social trait
evaluations and behavior patterns than males who express anger in
the workplace.
Preliminary Analyses
Like Study 1, Study 2 obtained information about promotion by asking
respondents which consultant they would choose to promote in the
questionnaire, then asking during the debriefing if they would promote both
consultants “if it were possible to do so” (see Chapter 3 for more details). These
two different approaches to the promotional decisions resulted in very different
likelihoods of promoting the angry consultant in the directed anger condition (see
Figure 6.1 and Table 6.1).
When considering only the binary choice between the consultants, 35% of
respondents in the directed anger condition chose to promote a female who
directed a single instance of anger at a subordinate over a non-angry female,
while 37% of respondents chose to promote a similarly angry male consultant
over a neutral male consultant; the difference was not significant (see Table 6.1),
providing no support for Hypothesis 7, which stated that female consultants who
expressed directed anger would be less likely to be promoted over a neutral
consultant than male consultants who expressed directed anger. These figures
103
can also be compared to the online respondents who assessed one-time angry
consultants in Study 1; 64% of these respondents chose to promote a female
who got angry once over a neutral consultant (the difference between these
groups was statistically significant [z = 1.68, p < .05], see Table 6.2). When
assessing male consultants, 57% of online respondents chose to promote males
who were angry once over neutral consultants in Study 1; the difference between
this and the 37% observed for males in Study 2 was not significant.29
As in Study 1, most respondents seemed to be willing to promote both
individuals if possible. When provided with the opportunity to promote both
consultants, the majority of respondents (85% and 74% for females and males,
respectively; see Table 6.2) in the directed anger condition chose to do so,
compared to 86% for both females and males in the single anger condition for
online Study 1 respondents. There were no significant differences across groups.
I also conducted a series of logistic regression models to consider the log-
likelihood of promoting the angry consultant or both consultants (see Table
6.3).30 These models included Study 1 respondents in the single-anger condition
as well as Study 2 (directed anger) respondents, with the anger consultant in
both cases being compared to a neutral consultant.31 Including both of these
conditions in the model while controlling for additional factors allows for
consideration of the effects of directed anger versus un-directed anger. 29 Since I could not control for online vs. in-person in these analyses and all Study 2 subjects were run online, I restricted the analyses to only online subjects from Study 1 for comparability. 30 All logistic regression analyses were conducted including some subjects from Study 1, as the sample size for Study 2 was too small to allow for these analyses if used alone. However, the results are consistent with z-tests using only Study 2 respondents (p < .01, not shown). 31 Since the studies were run concurrently on the same subject pool, they can essentially be viewed as two conditions of the same experiment.
104
Comparing a consultant who exhibited directed anger (rather than un-directed
anger) to a neutral consultant significantly (p < .05) reduced the log-odds of
promoting the angry consultant (b = -1.18), promoting both consultants (b = -
1.59) or choosing to have the angry consultant as a boss (b = -1.30).32 Like the
basic z-tests, these models fail to support Hypothesis 7, as no gender differences
were observed. Older individuals were less likely to promote both individuals than
younger individuals; this was not observed in any other regression analyses (b =
-1.64, p < .05).
To assess differences on social measures, I again conducted analyses of
variance (ANOVA) with ratings of the neutral vs. angry consultant as a within-
subject factor and gender of consultant and directed anger as between-subjects
factors.33 For comparability, I used Study 2 subjects and online subjects in the
single anger condition of Study 1.34 As in Study 1, these analyses assessed four
traits (pleasantness, likeability, cooperativeness, and trustworthiness) and four
behaviors (likelihood of enjoy socializing with, being friends with, asking personal
advice from, and enjoying working with). Tables 6.4 and 6.5 show the means and
standard deviations for these variables.
Three of the four traits (pleasantness, likeability, and cooperativeness)
showed significant (p < .01) within-subjects differences, with the angry consultant
being perceived more negatively (see Table 6.6). No within-subjects interactions 32 These regressions were also run with all subjects, controlling for frequency of anger as well. Both directed anger and frequency of anger were significant in this case. 33 I also ran these analyses using only Study 2 subjects, omitting directed anger as a between-subjects factor (not shown). No significant differences were found between those results and the ones reported here; primarily within-subjects effects were observed. 34 Frequently angry consultants and in-person participants were excluded to minimize the number of factors that could influence the results.
105
were significant. Trustworthiness, the only trait that did not show a significant
within-subjects difference, did differ significantly between-subjects, with
individuals in the directed anger condition being perceived as significantly more
trustworthy (F = 5.79, p < .05).
All four behaviors (enjoy socializing or working with the consultants, being
friends with the consultants, and asking the consultants for personal advice)
showed highly significant within-subjects differences (Fs range from 11.92 to
18.00, p < .01; see Table 6.7), with the angry consultant being perceived more
negatively. No within-subjects interactions were significant. The only between-
subjects differences were found for likelihood of being friends with the
consultants. Respondents in the directed anger condition reported an increased
likelihood of being friends (F = 6.17, p < .05), and a slightly significant interaction
between the directed anger condition and the gender of the consultants was
observed (F = 3.74, p < .1). Taken together, the results for analysis of social
measures allow me to reject Hypothesis 8, which stated that female consultants
who expressed directed anger would receive lower scores on social measures
(relative to neutral consultants) than male consultants.
Discussion
Based on these analyses, even directed anger does not appear to be
sufficient to introduce unfavorable gender effects for females, for the most part;
Hypothesis 7 receives little support. However, one observable gender difference
deserves further exploration; while females who expressed directed anger did not
differ from males who did so, they did differ from females whose anger was not
106
directed at a subordinate. Females who directed anger at a subordinate were
significantly less likely to be promoted over an individual who did not express
anger than females who became angry a single time but did were not said to
direct that anger at another person. However, when given the opportunity to
promote both the angry and non-angry individuals, most respondents still chose
to do so, and no statistically significant difference was observed. Nonetheless,
when other factors are not controlled for, the consequences of directing anger
appear to be slightly more severe for females.
However, this relationship disappeared in logistic regression analyses that
control for other factors. These logistic regression analyses do suggest that
directed anger was viewed more negatively than un-directed anger when
promotional decisions were being made, and comments from respondents
reflected the importance of how others are treated; for example, respondents
noted that “You can’t respect someone who treats you disrespectfully” (Male,
Angry Consultant) and “(Male, Neutral Consultant) is good at getting his job done
without having to belittle others” (Directed Anger Condition).
With regard to social measures, Hypothesis 8 receives no support, as no
negative social effects for females were observed. As in Study 1, within-subjects
effects showed that respondents viewed angry consultants as less pleasant,
likeable, and cooperative than non-angry consultants. The one exception was
trustworthiness, for which respondents did not rate the two consultants
differently; instead, consultants in the directed anger condition were seen as
more trustworthy than consultants in the un-directed anger condition. This may
107
be because respondents were significantly more likely to view the consultants in
Study 2 (the directed anger condition) as leaders than consultants who did not
direct their anger at a subordinate (F = 6.81, p < .01, not shown). Leadership can
be associated with trustworthiness (Brower et al., 2000), particularly when the
leader is effective (as might be indicated in this case by the overall positive tone
of the performance reviews), but it is not typically associated with the other social
traits assessed, like pleasantness.
Respondents were also less likely to believe they would enjoy socializing
or working with the angry consultant, less likely to expect to be friends with the
angry consultant, and less likely to ask the angry consultant for personal advice.
Respondents in the directed anger condition were more likely to believe they
would be friends with the consultants and slightly more likely to believe they
would be friends with females. This finding may again have been associated with
perceived differences in leadership; the perceived leadership abilities of these
consultants may have made the idea of their friendship more appealing.
108
Table 6.1. Chi-Square and z-Test Analyses of Likelihood of Promotion for Directed Anger, by Gender Promote Angry
Consultant Over Neutral Consultant
Promote Both Consultants if
Possible Female/Directed Anger (N = 20)
N = 7 .35
N = 17 .85
Male/Directed Anger (N = 19)
N = 7 .37
N = 14 .74
χ2 = 0.01 z = 0.12
χ2 = 0.77 z = -0.87
*p ≤ .1 **p < .05 ***p<.01
109
Table 6.2. Chi-Square and z-Test Analyses of Likelihood of Promotion, by Gender of Consultant and Directed Anger Promote Angry
Consultant Over Neutral Consultant
Promote Both Consultants if
Possible Female/One-Time Un-Directed Occurrence (Online) (N = 14)
N = 9 .64
N = 12 .86
Female/Directed Occurrence (N = 20)
N = 7 .35
N = 17 .85
χ2 = 2.84* z =1.68**
χ2 = 0.00 z = 0.06
Male/One-Time Un-Directed Occurrence (Online) (N = 14)
N = 8 .57
N = 12 .86
Male/Directed Occurrence (N = 19)
N = 7 .37
N = 14 .74
χ2 = 1.34 z = 1.16
χ2 = 0.70 z = 0.84
*p ≤ .1 **p < .05 ***p<.01
110
Table 6.3. Logistic Regression Analyses of Respondents’ Likelihood of Choosing to Promote or Have the Consultant(s) as a Boss in Single-Anger Conditions+ To Promote
the Angry Consultant
To Promote Both
Consultants (if Possible)
To Have the Angry
Consultant as a Boss
Independent Variables Female Employee 0.60 0.45 0.40 Directed Anger -1.18** -1.59* -1.30** Control Variables Age 22 Years or Younger† - - - 23 Years or Older 0.00 -1.64** 0.03 Female Respondent 0.23 0.09 0.41 Race White 0.51 0.79 0.89 Asian 0.24 1.39 0.88 Other Ethnicity† - - - Major Business 0.89 -0.18 -0.43 Social Science/Humanities 1.03 0.09 0.30 Medical/Nursing 0.39 0.50 0.17 Science/Engineering 0.92 0.97 -0.01 Undecided† - - - Stanford Participant& -0.26 - -0.68 Online Participant 0.69 -0.40 0.36 Intercept -1.78** 2.47** -1.93** N 107 96 107 LR-Chi2 10.48 12.90 10.88 Degrees of Freedom 12 11 12 Notes: † Reference Category
& Stanford was dropped from the “Promote Both” regression as there was no differentiation on this variable. + Regressions including interaction terms were also run, but those interactions were not significant, so the results are not reported here. *p ≤ .1 **p < .05 ***p<.01
111
Table 6.4. Means and Standard Deviations for Trait Variables Used in RMANOVAs Mean (SD) Pleasant Angry Consultant 5.48 (1.54) Un-Directed Anger (Once) 5.61 (1.29) Directed Anger 5.38 (1.71) Female Consultant 5.56 (1.28) Un-Directed Anger (Once) 5.50 (1.16) Directed Anger 5.60 (1.39) Male Consultant 5.39 (1.78) Un-Directed Anger (Once) 5.71 (1.44) Directed Anger 5.16 (2.01) Neutral Consultant 6.07 (1.49) Relative to Un-Directed Anger (Once) 6.00 (1.12) Relative to Directed Anger 6.13 (1.72) Female Consultant 6.09 (1.60) Relative to Un-Directed Anger (Once) 6.00 (1.24) Relative to Angry Frequently 6.15 (1.84) Male Consultant 6.06 (1.39) Relative to Un-Directed Anger (Once) 6.00 (1.04) Relative to Angry Frequently 6.11 (1.63) Likeable Angry Consultant 5.70 (1.38) Un-Directed Anger (Once) 5.68 (1.25) Directed Anger 5.72 (1.49) Female Consultant 5.62 (1.18) Un-Directed Anger (Once) 5.57 (1.02) Directed Anger 5.65 (1.31) Male Consultant 5.79 (1.58) Un-Directed Anger (Once) 5.79 (1.48) Directed Anger 5.79 (1.69) Neutral Consultant 6.36 (1.03) Relative to Un-Directed Anger (Once) 6.25 (0.75) Relative to Directed Anger 6.44 (1.19) Female Consultant 6.29 (1.14) Relative to Un-Directed Anger (Once) 6.14 (0.86) Relative to Angry Frequently 6.40 (1.31) Male Consultant 6.42 (0.90) Relative to Un-Directed Anger (Once) 6.36 (0.63) Relative to Angry Frequently 6.47 (1.07)
112
Table 6.4 (continued). Means and Standard Deviations for Trait Variables Used in RMANOVAs
Cooperative Angry Consultant 5.48 (1.48) Un-Directed Anger (Once) 5.43 (1.64) Directed Anger 5.51 (1.37) Female Consultant 5.65 (1.12) Un-Directed Anger (Once) 5.50 (1.09) Directed Anger 5.75 (1.16) Male Consultant 5.30 (1.78) Un-Directed Anger (Once) 5.36 (2.10) Directed Anger 5.26 (1.56) Neutral Consultant 6.06 (1.23) Relative to Un-Directed Anger (Once) 5.86 (1.43) Relative to Directed Anger 6.21 (1.06) Female Consultant 6.09 (1.31) Relative to Un-Directed Anger (Once) 5.50 (1.70) Relative to Angry Frequently 6.50 (0.76) Male Consultant 6.03 (1.16) Relative to Un-Directed Anger (Once) 6.21 (1.05) Relative to Angry Frequently 5.89 (1.24) Trustworthy Angry Consultant 6.16 (1.37) Un-Directed Anger (Once) 5.79 (1.50) Directed Anger 6.44 (1.21) Female Consultant 6.32 (1.17) Un-Directed Anger (Once) 6.00 (1.18) Directed Anger 6.55 (1.15) Male Consultant 6.00 (1.54) Un-Directed Anger (Once) 5.57 (1.79) Directed Anger 6.32 (1.29) Neutral Consultant 6.25 (1.34) Relative to Un-Directed Anger (Once) 5.75 (1.58) Relative to Directed Anger 6.62 (1.02) Female Consultant 6.32 (1.22) Relative to Un-Directed Anger (Once) 5.93 (1.21) Relative to Angry Frequently 6.60 (1.19) Male Consultant 6.18 (1.47) Relative to Un-Directed Anger (Once) 5.57 (1.91) Relative to Angry Frequently 6.63 (0.83) Note: All variables are 7-point scales, with higher scores indicating a higher rating on the trait.
113
Table 6.5. Means and Standard Deviations for Behavioral Variables Used in RMANOVAs Mean (SD) Enjoy Socializing With _________ Angry Consultant 3.36 (0.88) Un-Directed Anger (Once) 3.18 (0.86) Directed Anger 3.49 (0.88) Female Consultant 3.38 (0.85) Un-Directed Anger (Once) 3.07 (0.73) Directed Anger 3.60 (0.88) Male Consultant 3.33 (0.92) Un-Directed Anger (Once) 3.29 (0.99) Directed Anger 3.37 (0.90) Neutral Consultant 3.90 (0.97) Relative to Un-Directed Anger (Once) 3.75 (0.93) Relative to Directed Anger 4.00 (1.00) Female Consultant 4.00 (1.07) Relative to Un-Directed Anger (Once) 3.71 (1.07) Relative to Angry Frequently 4.20 (1.06) Male Consultant 3.79 (0.86) Relative to Un-Directed Anger (Once) 3.79 (0.80) Relative to Angry Frequently 3.79 (0.86) Be Friends With _________ Angry Consultant 3.13 (1.00) Un-Directed Anger (Once) 2.82 (0.90) Directed Anger 3.36 (1.01) Female Consultant 3.21 (1.04) Un-Directed Anger (Once) 2.71 (0.99) Directed Anger 3.55 (0.94) Male Consultant 3.06 (0.97) Un-Directed Anger (Once) 2.93 (0.83) Directed Anger 3.16 (1.07) Neutral Consultant 3.51 (1.08) Relative to Un-Directed Anger (Once) 3.18 (1.16) Relative to Directed Anger 3.74 (0.97) Female Consultant 3.65 (1.12) Relative to Un-Directed Anger (Once) 3.00 (1.18) Relative to Angry Frequently 4.10 (0.85) Male Consultant 3.36 (1.03) Relative to Un-Directed Anger (Once) 3.36 (1.15) Relative to Angry Frequently 3.37 (0.96)
114
Table 6.5 (continued). Means and Standard Deviations for Behavioral Variables Used in RMANOVAs
Ask _________ for Personal Advice Angry Consultant 2.18 (1.15) Un-Directed Anger (Once) 2.04 (1.17) Directed Anger 2.28 (1.15) Female Consultant 2.21 (1.07) Un-Directed Anger (Once) 2.14 (1.17) Directed Anger 2.25 (1.02) Male Consultant 2.15 (1.25) Un-Directed Anger (Once) 1.93 (1.21) Directed Anger 2.32 (1.29) Neutral Consultant 2.49 (1.33) Relative to Un-Directed Anger (Once) 2.36 (1.31) Relative to Directed Anger 2.59 (1.35) Female Consultant 2.59 (1.33) Relative to Un-Directed Anger (Once) 2.57 (1.45) Relative to Angry Frequently 2.60 (1.27) Male Consultant 2.39 (1.34) Relative to Un-Directed Anger (Once) 2.14 (1.17) Relative to Angry Frequently 2.58 (1.46) Enjoy _________ as a Co-Worker Angry Consultant 3.49 (0.84) Un-Directed Anger (Once) 3.39 (0.79) Directed Anger 3.56 (0.88) Female Consultant 3.56 (0.74) Un-Directed Anger (Once) 3.43 (0.85) Directed Anger 3.65 (0.67) Male Consultant 3.42 (0.94) Un-Directed Anger (Once) 3.36 (0.74) Directed Anger 3.47 (1.07) Neutral Consultant 4.03 (0.70) Relative to Un-Directed Anger (Once) 4.04 (0.58) Relative to Directed Anger 4.03 (0.78) Female Consultant 4.18 (0.52) Relative to Un-Directed Anger (Once) 4.07 (0.47) Relative to Angry Frequently 4.25 (0.55) Male Consultant 3.88 (0.82) Relative to Un-Directed Anger (Once) 4.00 (0.68) Relative to Angry Frequently 3.79 (0.92) Note: All variables are 5-point scales.
115
Table 6.6. RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Traits, Based on Study 2 Subjects and Online Study 1 Subjects in the Single Anger Condition Pleasant Likeable Cooperative Trustworthy Factor Between subjects Female Employee (FE) 0.04 0.36 0.19 0.62 Directed Anger (DA) 0.02 0.18 0.50 5.79** FE X DA 0.26 0.04 1.97 0.22 Within subjects Social Measure (SM) 11.92*** 18.00*** 12.15*** 0.65 SM X FE 0.08 0.01 1.32 0.86 SM X DA 1.16 0.23 0.67 1.44 SM X FE X DA 0.86 0.01 2.31 0.29 *p ≤ .1 **p < .05 ***p<.01
116
Table 6.7. RMANOVAs Comparing the Angry Individual and the Neutral Individual in Terms of Likelihood that Respondent Would... , Based on Study 2 Subjects and Online Study 1 Subjects in the Single Anger Condition Enjoy
Socializing With
(Consultant)
Be Friends With
(Consultant)
Ask (Consultant) for Personal
Advice
Enjoy (Consultant)
as Co-Worker
Factor Between subjects Female Employee (FE) 0.21 0.40 0.26 1.50 Directed Anger (DA) 2.01 6.17** 0.66 0.23 FE X DA 1.43 3.74* 0.34 0.60 Within subjects Social Measure (SM) 19.34*** 19.35*** 8.80*** 26.28*** SM X FE 0.43 0.36 0.51 0.44 SM X DA 0.06 0.01 0.01 0.74 SM X FE X DA 0.01 1.95 0.09 0.44 *p ≤ .1 **p < .05 ***p<.01
117
Chapter 7: Conclusions and Future Directions
Review of Major Findings and Contributions
This study found that while anger in the workplace may be relatively
common, it was still damaging to perceptions of employees, particularly with
regard to social measures. However, open-ended negative comments about a
consultant who expressed anger once sometimes included that information as a
caveat or even a positive. Frequent expression of anger was seen as more
problematic than a single angry outburst that was not directed at another
individual.
Overall, particularly with regard to promotional decisions, the negative
effects of anger expression were not as pronounced as might be expected. About
fifty percent of respondents chose to promote an individual who was angry only a
single time over an individual whose performance review did not mention anger
when that anger was not directed at another person; in other words, an individual
who was angry once had basically an even chance of being promoted compared
to a neutral individual. The majority of respondents were also willing to promote
even someone who was angry frequently if they did not have to do so at the
expense of someone who was not said to be angry. Social measures like
pleasantness and likelihood of being friends were more affected by anger than
promotional decisions, and even a single expression of anger significantly
reduced scores on such measures. Furthermore, anger that was directed at a
subordinate was viewed more negatively than non-directed anger. Overall, the
results of this work suggest that while anger may affect social perceptions of an
118
individual, it is not always detrimental in a workplace environment. This is not
entirely unexpected given anger’s associations with valued workplace attributes
like competence (Tiedens, 2001).
Surprisingly, Study 1 failed to produce the clear gender differences seen
in previous work; in fact, males were sometimes punished more for expressing
anger than females, and females were seen as more suitable for promotion than
males. Study 2, a preliminary attempt to produce such differences by introducing
an object of anger, also showed few disadvantages for females who expressed
anger, even when that anger was directed at a particular individual (specifically, a
subordinate). Simple z-tests suggested that females who directed anger at a
particular individual were less likely to be promoted over a neutral consultant
than females whose anger was un-directed, but this effect was not present in
logistic regression analyses that controlled for other factors, and no direct gender
effects were observed for the social measures. Overall, expression of anger in
Study 2 continued to be detrimental in terms of promotional and social outcomes.
This study is one of the first to find that anger in the workplace may have
no detrimental effects for females and may even be beneficial. While there is a
significant body of literature addressing conditions in which anger is damaging to
females (see Chapter 2 for some examples), few studies have considered factors
that may help mediate the effects of angry outbursts in the workplace. In actual
workplace environments, individuals are likely to have significantly more
knowledge about the person being assessed than is presented in most
experimental designs (including this one). As this additional information is likely
119
to have significant effects on how individuals are perceived, experimental studies
should make an effort to replicate the context provided by a workplace
environment as closely as possible.
Study 2, which manipulated the object of the anger in attempt to produce
the gender differences not found in Study 1, also found minimal gender effects.
Future studies should continue to explore the potential features of both
experiments that may have contributed to the lack of consistent gender
differences. Some possible approaches to this issue are described below as part
of the directions for future research.
It is likely that the context provided by the experimental materials played a
role in minimizing the negative effects of anger for women. By using realistic
documents based on actual workplace materials to introduce anger, these
experiments situated anger within a larger context of overall workplace
performance. Participants clearly took this additional information regarding the
individuals’ capabilities and overall sociability (i.e., being “generally easy to work
with”) into account when making decisions and rating the consultants, as
evidenced by their open-ended comments and notes taken during the
experiment. Thus, this research suggests a new experimental procedure for
assessing the effects of emotions in the workplace that more closely
approximates actual workplace conditions.
The potential for contextual effects indicates that researchers should keep
in mind the potential effects of how anger is presented when designing
experimental studies, particularly those involving emotions. Researchers,
120
especially those focused on the workplace context, should consider using a
variety of methods and situations to introduce their manipulations and attempt to
mimic a real-world setting as closely as possible before drawing conclusions
based on laboratory findings. This study demonstrated one way that contextual
information could be introduced (resumes and performance reviews), but
additional documents (e.g., cover letters) could be used as well.
This research also demonstrated the value of using different levels and
objects of anger expression. Respondents clearly distinguished between those
who expressed anger frequently and those who expressed anger a single time,
with the consequences of frequent anger being significantly more severe in terms
of both social measures and promotional decisions. Few studies have compared
different levels of anger expression, yet this represents a fruitful area for further
research. Given that most individuals remain in a single workplace for a few
years (Bureau of Labor Statistics, 2008) and anger is experienced relatively
frequently in the workplace (Fitness, 2000), determining when expression of
anger becomes problematic is useful at both the individual and organizational
levels.
Respondents also distinguished between cases in which the object of
anger was unspecified and cases in which the anger was directed at a
subordinate. Thus, framing anger may provide a way to mediate negative effects;
for example, managers who make a concerted effort to present their anger more
generally, such as expressing anger at a situation rather than the individual who
caused that situation, may be viewed more positively than individuals who direct
121
their anger at a particular person. This method may be slightly more effective for
females than males, based on the significant effect observed for females when
comparing consultants expressing directed anger to those expressing un-
directed anger; however, the lack of significance for this factor in analyses
controlling for other factors suggests that more research is needed before
concrete conclusions can be drawn.
Directions for Future Research
This work suggests several directions for future research, particularly in
terms of focusing on gender differences. Future research could continue to
provide details about expression of anger in an attempt to determine those
factors that might replicate the negative effects for females found in other
studies. For example, the nature of the angry outburst (e.g., scowling vs.
throwing something) could have different effects for males and females. In
addition, as noted above, context plays an important role in expression of anger;
perhaps anger outside the workplace, or even within the workplace but outside of
a meeting (such as in the breakroom), would have different effects. It would also
be interesting to consider others’ responses to anger (on the part of the angry
individual or others in the workplace); expression of anger can result in a variety
of responses, from reconciling with the angry individual to withdrawing from the
situation (Kuppens, Van Mechelen, & Meulders 2004), and these responses
provide further information about the angry individual that may inform how
subjects react. Since some effects of expression of anger were mediated by
concerns about the risks of promoting the angry individual and being worried
122
about his/her qualifications as well as by worries about working with that
individual, explicit information about how others respond to the anger may
contribute to respondents’ decision-making process.
Individuals in this study were generally perceived as very successful and
competent, as indicated by respondent comments and questionnaire responses.
There may be an interaction between gender and success with regard to angry
outburst; perhaps the established success of the consultants made anger seem
more legitimate (or at least more excusable) for both males and females.
Heilman et al. (2004) note that individuals will use cognitive distortion to maintain
their image of women as less successful than men unless the women’s success
is made so explicit that such distortion becomes difficult; similar distortion could
take place with regard to justifying angry outbursts. In this case, manipulating the
materials to make the consultants less successful might leave females more
open to judgments based on their negative behaviors.
The results of these studies should be viewed as applicable to a specific
context and situation. Participants were given very limited information about
expression of anger, and these findings were not intended to contribute to the
literature on types of anger expression (e.g., anger-in vs. anger-out, as discussed
by Funkenstein, King, & Drolette, 1954).35 These studies also did not aim to
make direct comparisons between two angry individuals (male or female or
expressing different levels of anger). Finally, it should be made clear that the 35 Briefly, anger-in refers to reactions in which overt expression of anger is suppressed, while anger-out encompasses open expression of anger. This typology was later expanded to address different types of expressed anger (O’ Connor, Archer, & Wu, 2001).
123
anger in this case was situated within a single social context, the workplace; the
results of these studies should not be extrapolated to expression of anger in
other environments, like the home. Certain workplace situations, particularly
those involving customer-facing employees such as in retail or service industries,
would also represent a social context that differs significantly from the scenario
presented in this study. Furthermore, this study used promotional decisions,
which are less likely to be affected by gender biases than hiring (Lyness &
Judiesch, 1999). Future studies might use resumes and, for example, a letter of
recommendation, to introduce anger in a realistic way with regard to hiring
decisions in order to see if this holds true when more context is provided.
124
Appendix A: Experimental Materials – In-Person36
Resume S Longer Name Blacked Out Address and other kkfkkfkfkk.k Info Here andmmmmmm Blacked out........... Pppppppppppp,ppppp EXPERIENCE YTS Global Business Services – Cambridge, MA Global Business Advisor (11/03-10/06)
o Supported development and management, planning and execution, financial reporting, and global team management/coordination.
o Drove global practice communication and managed leadership calls, including developing agendas, ensuring executive participation and interaction, taking meeting minutes, and managing follow ups.
o Oversaw the development and deployment of Knowledge Management, Marketing and Thought Leadership activities, and made recommendations on behalf of the Global Industry Leader.
o Actively managed the pipeline of opportunities within the practice area. o Supervised planning processes to ensure investments support business
area plans and growth and profit targets. CCL Communications – Boston, MA Independent Consultant/Contractor (4/01-8/03)
o Accountable for new business development, executing marketing promotions, public relations, special events, complete accounting responsibilities and recruitment from third-party vendors.
o Led the development of fully integrated, strategic marketing plans, including market assessment and analysis, objectives and key strategies, innovative approaches and new ideas and tactical evaluation to benchmark ROI.
o Managed budgets to maximize every dollar and forecast to meet financial objectives. Created and produced advertisements, brochures, and other literature as required. Demonstrated success in the development of long-term business partnerships.
o Client List: HMC-La Agency, Glen, Smith & Glen Development, Plise Development, Martz Agency
36 The letters assigned to the resumes (“S” and “C”) indicate the location of the contact information on the respective resumes (sides vs. center). In the format given to participants, both resumes fit on a single page. Different fonts were used to help make the resumes look different. The letters assigned to the performance reviews (“T” and “A”) represent the first letter of the first word in the review. Resumes and performance reviews were randomly assigned as “A” or “B” for each participant.
125
EBaven, Inc. – Boston, MA Internet Marketing Initiative Manager (1/99-3/01)
o Led client and project management efforts to develop and maintain patient education program and women’s health portal to be provided to managed care organizations (MCOs) for patient use.
o Successfully sold programs into 10 (of 11) leading MCOs; collaborated with MCO customers to build and integrate online patient education program into existing managed care websites; work resulted in the attainment of formulary status of 2 blockbuster drugs with a large HMO.
o Led team in performing in-depth research of pharmaceutical industry to identify key industry trends and current and future issues, and to fully understand marketplace.
o Analyzed industry’s historical performance and future projections to determine firm’s sales opportunities.
o Outlined specific opportunities for companies to leverage online strategies. Work resulted in a $200K project for firm, and was used successfully on several client engagements.
Shaw’s Supermarkets – W. Bridgewater, MA Manager, Customer Segmentation (7/97-11/98)
o Supervised loyalty program database containing data on 4.5MM+ cards and 2.5MM households; used data to define optimal target audience for targeted marketing programs to drive customer loyalty and retention.
o Developed, analyzed, and summarized information on customer purchase trends for senior management.
o Managed Technical Managers to ensure appropriate matching of skills to workload and assignments while fostering professional development.
EDUCATION University of Massachusetts – Boston, MA (9/94-5/98) BSc. Business Administration
o Interned with local public relations firm
126
Resume C MARGARET WILSON
This will be inked out with black marker so it looks like [email protected]
EXPERIENCE Global Communications Services 2003-2006 Communications and Operations Manager, Houston, TX • Facilitated the running of a $4B global business by helping to drive business
development and strategy, managing communications, and managing/planning the management calendar.
• Identified areas of improvement within the global management organization and processes.
• Collaborated with global senior leadership team and other key parties to ensure the business ran smoothly and efficiently while leadership maintained a focus on clients.
• Drove all sector-level communications, including planning staff and Partner calls; developing comprehensive communications plans; and building meeting agendas and ensuring participation.
• Maintained client campaign plan, provided follow up on client meetings, developed a rigorous client agenda, and prepared for quarterly client meetings.
The Trevi Group 2002-2003 Sales and Marketing Director, Dallas, TX • Areas of responsibility included marketing promotions, trade shows and the
company’s entire public relations function, including media relations, executive media relations and event relations.
• Represented company in collaborations with consultants, strategic partners. vendors and clients.
• Developed, planned, and executed marketing initiatives directed at attaining growth, retention and brand recognition.
• Responsible for all marketing and business development activities and functions including advertising, marketing communications/collateral development, direct marketing, campaign/program development, tracking, market analysis, research, reporting, branding, events and community outreach/PR.
• Developed materials for external web site. Developed and pitched story ideas and speaking opportunities.
Yellowdot Media Group 2001-2002 Advertising Sales Manager, Dallas, TX • Responsible for key account sales within the largest family-owned media
company in Dallas. Publications included an alternative weekly publication, a
127
weekly business publication, a monthly lifestyles magazine, a weekly tourist magazine, a golf magazine and a national image magazine.
• Called on local and national advertisers directly in order to plan and implement advertising, marketing and promotional programs.
• Determined and met client’s specific advertising needs; planned and conducted sales presentations with key clientele.
• Consulted clients with design and layout of ads to maximize sales effectiveness and exposure.
Ball Communications 1998-2000 Account Executive, Denton, TX • Designed and expedited large-scale multi-media marketing and promotion
programs in major markets; organized all aspects of writing, graphic arts and production for printed materials along with radio and TV broadcast.
• Managed new and existing accounts. Developed advertising/marketing plans, promotions, strategies and budgets. Reviewed billing process, media plans and reconciliations.
• Determined and met specific client needs; conducted presentations and worked extensively with designers in the use of graphic designs for project promotions.
• Client List: Texas Power, Danielson Financial Group, Desert Radiologist, Sam’s Town Hotel & Gambling Hall, Gaming Entertainment & Media (GEM), Cox Communications, Canyon Gate Country Club, Bear’s Best Golf Course
EDUCATION UNIVERSITY OF TEXAS AT AUSTIN May 1998 Bachelor of Science, Industrial Management, Austin, TX OTHER • Trained in Psychology of Marketing
128
Performance Review T YEAR: 2008 Name: Black out this name COMPANY: Black this out Consulting Job Title: Consultant 1. Performance Assessment a) Assessment of performance against key business targets as defined in Jan 2008 Took on new responsibilities in 2008. Moved from back-up to primary contact with client. This change in responsibility included taking on new accounting procedures. Provided information on client sales and revenues for business development.
b) Assessment of performance against personal development targets as defined in Jan 2008 Attended in-house training on current best practices. Attended 2008 Strategic Outsourcing Conference.
2. Objectives (Specific Client Projects) a) Project 1 – Outsourcing Opportunity (Client 1148) Managed client relations effectively Came in on-time and on budget Percent of Project Completed: 95 out of 100%
b) Project 2 – Increasing Sales (Client 674) Increased sales by 12% Maintained good clients communications Generated reports to increase client value Percent of Project Completed: 90 out of 100%
129
3. Competencies a) CORE: Customer Focus 4.0 out of 5 Works well with customers to understand and meet their needs. b) CORE: Execution and Results 5.0 out of 5 Both projects on schedule. Consistently follows through on details of work assigned. c) CORE: Interpersonal Skills 3.5 out of 5 Generally works well with and assists others.
d) CORE: Job/Technical Skills 4.5 out of 5 Able to apply previous skills to new projects. Creates clear, well-developed presentations. 4. Additional Comments Asks to learn new duties. Makes significant contributions to the job and the organization.
4. Additional Comments included one of the following as the final line: - Team members report that she (he) is generally easy to work with, but often gets angry at co-workers when challenged. - Team members report that she (he) is generally easy to work with, but recently got angry at a co-worker in a meeting. - Team members report that she (he) is generally easy to work with.
130
Performance Review A YEAR: 2008 Name: Black out this longername COMPANY: Black this out Consulting Job Title: Consultant 1. Performance Assessment a) Assessment of performance against key business targets as defined in Jan 2008 Assigned greater responsibility this year. Developed client reports for potential acquisitions. Made lead communication contact for client, which included additional reporting and management duties.
b) Assessment of performance against personal development targets as defined in Jan 2008 Attended 17th Annual Mergers and Acquisitions workshop at financing conference. Took company seminar on best practices.
2. Objectives (Specific Client Projects) a) Project 1 – Increasing Market Share (Client 252)
• Produced deliverables on-time • Met client expectations • Managed budget
Percent of Project Completed: 93 out of 100% b) Project 2 – Potential Acquisition (Client 1583)
• Communicated objectives clearly to client • Investigated revenues of potential company • Made recommendation that saved company 7% on budget
Percent of Project Completed: 92 out of 100%
131
3. Competencies a) CORE: Customer Focus 5.0 out of 5 Able to assess and respond to customer requirements. Comfortable with customers. b) CORE: Execution and Results 4.5 out of 5 Plans and organizes time very efficiently and effectively. Projects completed within expected timeframe. c) CORE: Interpersonal Skills 3.5 out of 5 Develops and maintains good working relationships.
d) CORE: Job/Technical Skills 4.0 out of 5 Very comfortable with Excel and Powerpoint. Makes good use of existing skillset. 4. Additional Comments Seeks out new tasks and responsibilities and looks for ways to enhance the organization.
4. Additional Comments included one of the following as the final line: - Team members report that she (he) is generally easy to work with, but often gets angry at co-workers when challenged. - Team members report that she (he) is generally easy to work with, but recently got angry at a co-worker in a meeting. - Team members report that she (he) is generally easy to work with.
132
Set-Up Script
“Thanks for coming to participate in this study. I am going to tell you a
little bit about today’s study and then get you started. Let me apologize for
reading to you, but this is the way we make sure that the details of the study
remain the same for all participants.
This is a study about promotional decisions in the workplace. Questions
have been asked about how employers choose whether to promote one
employee over another. Our primary goal in this study is to find out how resumes
and performance reviews affect decisions about employees’ job title and salary.
Today you will review the resumes and performance reviews of two
individuals, Consultant A and Consultant B. These are actual resumes and
performance reviews from two women (men) working at a consulting firm in the
Bay Area. These individuals have each been with the company for two years as
Consultants and are up for promotion for the same position, Project Manager.
You will carefully consider each person’s resume and her (his) performance
review from the previous year, then answer some questions about the individuals
and how you think they should be compensated.
Before we begin, I need to have you read and sign the consent form. Both
copies are the same. After you read it, please sign the top one and give it to me.
The second copy is for you to keep. [Collect the consent form.] Thank you.
When you are ready, you will need to read all of the provided information
for each employee carefully and thoroughly. You will be given ten minutes to
review the documents and make notes. After ten minutes, I will return and ask if
133
you are ready to respond to some questions about the employees. If you are
ready, you will be given a questionnaire. If you need more time, you will be given
another five minutes to review the documents, then provided with the
questionnaires.”
134
Debriefing Script
“At this point, I would like to take a few minutes to tell you more about the
study in which you just participated. But first, I am curious about what you
thought of this study.
What did you think of the employees’ resumes today?
What did you think of the employee performance reviews?
Which employee did you feel deserved the promotion? If it were possible
to promote both employees, would you prefer to do so?
In the questionnaire we asked you how you arrived at your promotional
decision -- is there anything else about this that you want to add?
Do you have any other comments or questions?
In this study, we are trying to learn more about how the opinions of others
affect hiring and promotional decisions. Specifically, we are curious about
emotions in the workplace and how their effects differ for males and females.
To answer these questions, we had you review two resumes and two
performance reviews. We expected these candidates to appear very similar to
you. The resumes did not belong to actual individuals, but were in fact
135
composites of several real resumes. The performance reviews were based on
actual performance reviews from real companies, but the information was
constructed to make the two individuals seem to be equivalent. By choosing
individuals with similar backgrounds and capabilities, we hope to encourage you
to consider other factors in your decisions. Specifically, we are interested in how
emotional behavior affects perceptions of employees in the workplace.
In this study, we hope to learn more about people’s perceptions and
beliefs about males’ and females’ emotional expression in the workplace. We
know from other sociological studies that women tend to be punished for
expressing anger in the workplace, while men are sometimes rewarded. We will
look to see if different types of anger expression have different effects on
promotional decisions for males and females. We will also see if people feel
differently about males and females and have different preferences about
spending time with individuals socially based on performance reviews that
mention anger. These behaviors would tell us something about the effects of
expressing anger in the workplace and how they differ for men versus women.
Do you have any questions about how we conducted this study?
Finally, it is possible that some of your friends may participate in this
study. We would greatly appreciate that you not tell them about the purpose of
our study. If they know about the study in advance, they may behave differently
during the study.
136
Thank you very much for helping us with this study. Let me remind you
that you still have the right to withdraw your data without penalty. Please feel
free to call us later if you have any further questions. We can be reached
anytime at [redacted], or via email at [redacted]”
137
Foothill Consent Form37
Nonmedical Human Participants Consent Form
STUDY TITLE: Effect of Performance Reviews on Hiring Decisions - REP (Research Experience Program) Protocol Director: Kristen Backor DESCRIPTION: You are invited to participate in a research study on how performance reviews and resume information influence job decisions. You will be asked to review this information and fill out some related questionnaires. RISKS AND BENEFITS: There are no risks associated with this study. In addition to course credit, you may benefit by learning about social research and how it is performed. We cannot and do not guarantee or promise that you will receive any benefits from this study. If you have any questions regarding this study, please contact Protocol Director Kristen Backor at [redacted] or via email at [redacted]. TIME INVOLVEMENT: The session will take about 45 minutes. PAYMENTS: You will receive 1 hour toward your course requirement through the REP. SUBJECT'S RIGHTS: If you have read this form and have decided to participate in this project, please understand your participation is voluntary and you have the right to withdraw your consent or discontinue participation at any time without penalty. Your instructor will provide you with an alternative assignment. You have the right to refuse to answer particular questions. Your individual privacy will be maintained in all published and written data resulting from the study. CONTACT INFORMATION: *Questions, Concerns, or Complaints: If you have any questions, concerns or complaints about this research study, its procedures, risks and benefits, you should ask the Protocol Director, Kristen Backor at [redacted]. *Independent Contact: If you are not satisfied with how this study is being conducted, or if you have any concerns, complaints, or general questions about the research or your rights as a participant, please contact the Stanford
37 In their original format, consent forms fit on a single page.
138
Institutional Review Board (IRB) to speak to someone independent of the research team at (650)-723-2480 or toll free at 1-866-680-2906. You can also write to the Stanford IRB, Stanford University, Stanford, CA 94305-5401. The extra copy of this consent form is for you to keep. SIGNATURE _____________________________ DATE ____________ Protocol Approval Date: __12/19/08__________________ Protocol Expiration Date: __12/18/09______________________
139
Stanford Consent Form
Nonmedical Human Participants Consent Form
STUDY TITLE: Effect of Performance Reviews on Hiring Decisions Protocol Director: Kristen Backor DESCRIPTION: You are invited to participate in a research study on how performance reviews and resume information influence job decisions. You will be asked to review this information and fill out some related questionnaires. RISKS AND BENEFITS: There are no risks associated with this study. In addition to course credit, you may benefit by learning about social research and how it is performed. We cannot and do not guarantee or promise that you will receive any benefits from this study. If you have any questions regarding this study, please contact Protocol Director Kristen Backor at [redacted] or via email at [redacted]. TIME INVOLVEMENT: The session will take about 45 minutes. PAYMENTS: You will receive $10 in exchange for your participation in this study. SUBJECT'S RIGHTS: If you have read this form and have decided to participate in this project, please understand your participation is voluntary and you have the right to withdraw your consent or discontinue participation at any time without penalty. Your instructor will provide you with an alternative assignment. You have the right to refuse to answer particular questions. Your individual privacy will be maintained in all published and written data resulting from the study. CONTACT INFORMATION: *Questions, Concerns, or Complaints: If you have any questions, concerns or complaints about this research study, its procedures, risks and benefits, you should ask the Protocol Director, Kristen Backor at [redacted]. *Independent Contact: If you are not satisfied with how this study is being conducted, or if you have any concerns, complaints, or general questions about the research or your rights as a participant, please contact the Stanford
140
Institutional Review Board (IRB) to speak to someone independent of the research team at (650)-723-2480 or toll free at 1-866-680-2906. You can also write to the Stanford IRB, Stanford University, Stanford, CA 94305-5401. The extra copy of this consent form is for you to keep. SIGNATURE _____________________________ DATE ____________ Protocol Approval Date: __12/19/08__________________ Protocol Expiration Date: __12/18/09______________________
141
Questionnaire (Male Version)38 Please indicate your opinion on the following statements:
1. How much power do you think each Consultant should be entitled to in his current job?
Consultant A 1 2 3 4 5 6 7 8 9 10 11
none
A great deal
Consultant
B 1 2 3 4 5 6 7 8 9 10 11
none
A great deal
2. How much independence do you think each Consultant should have in his
current job?
Consultant A 1 2 3 4 5 6 7 8 9 10 11
none
A great deal
Consultant
B 1 2 3 4 5 6 7 8 9 10 11
none
A great deal
38 The female version is identical, but uses “she” and “her” where applicable.
142
3. How much status do you think each Consultant should have in his current
job?
Consultant A 1 2 3 4 5 6 7 8 9 10 11
none
A great deal
Consultant
B 1 2 3 4 5 6 7 8 9 10 11
none
A great deal
4. What yearly salary should each Consultant receive: In his current position as a Consultant?
Consultant A $__________ Consultant B $__________
If he were promoted to Project Manager? Consultant A $__________ Consultant B $__________
143
For each of the following pairs of adjectives, place an A and a B on the scale where you would rank Consultant A and Consultant B. competent | | | | | | | | incompetent respected | | | | | | | | not respected unpleasant | | | | | | | | pleasant powerful | | | | | | | | powerless unlikable | | | | | | | | likable knowledgeable | | | | | | | | unknowledgeable low status | | | | | | | | high status leader | | | | | | | | follower cooperative | | | | | | | | uncooperative intelligent | | | | | | | | unintelligent untrustworthy | | | | | | | | trustworthy 5. How confident are you that each Consultant is qualified to be Project Manager?
Consultant A 1 2 3 4 5
Not at all confident
Slightly Somewhat Very Extremely confident
Consultant
B 1 2 3 4 5
Not at all confident
Slightly Somewhat Very Extremely confident
144
6. How confident are you that each Consultant would be successful as Project Manager?
Consultant A 1 2 3 4 5
Not at all confident
Slightly Somewhat Very Extremely confident
Consultant
B 1 2 3 4 5
Not at all confident
Slightly Somewhat Very Extremely confident
7. How worried would you be about each Consultant’s qualifications to be Project Manager?
Consultant A 1 2 3 4 5
Not at all worried
Slightly Somewhat Very Extremely worried
Consultant
B 1 2 3 4 5
Not at all worried
Slightly Somewhat Very Extremely worried
8. How qualified do you think other Consultants at the company will feel each Consultant is to be Project Manager?
Consultant A 1 2 3 4 5
Not at all qualified
Slightly Somewhat Very Extremely qualified
Consultant
B 1 2 3 4 5
Not at all qualified
Slightly Somewhat Very Extremely qualified
145
9. How similar do you think each Consultant is to the company’s existing Project Managers?
Consultant A 1 2 3 4 5
Not at all similar
Slightly Somewhat Very Extremely similar
Consultant
B 1 2 3 4 5
Not at all similar
Slightly Somewhat Very Extremely similar
10. How likely would you be to enjoy socializing with each Consultant?
Consultant A 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
Consultant
B 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
11. How likely is it that you would be friends with each Consultant?
Consultant A 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
Consultant
B 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
146
12. How likely would you be to ask each Consultant for advice about work matters?
Consultant A 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
Consultant
B 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
13. How likely would you be to ask each Consultant for advice about personal matters?
Consultant A 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
Consultant
B 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
14. How likely is it that you would enjoy having each Consultant as a co-worker?
Consultant A 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
Consultant
B 1 2 3 4 5
Not at all likely
Slightly Somewhat Very Extremely likely
147
15. How risky do you think it would be to promote each Consultant?
Consultant A 1 2 3 4 5
Not at all risky
Slightly Somewhat Very Extremely risky
Consultant
B 1 2 3 4 5
Not at all risky
Slightly Somewhat Very Extremely risky
16. How many employees would you assign to work under each Consultant? Consultant A _____ employees Consultant B _____ employees 17. If you had to choose one Consultant to serve as your boss, which
Consultant would you prefer to be your boss? ___ Consultant A ___ Consultant B 18. If you had to choose one Consultant to work with on a task, which
Consultant would you prefer to work with? ___ Consultant A ___ Consultant B 19. If you had to choose one Consultant to sit next to on an airplane, which
Consultant would you prefer to sit with? ___ Consultant A ___ Consultant B 20. If you were the boss, which Consultant would you promote to Project Manager? (Circle one).
Consultant A Consultant B Why?
148
What was your overall impression of Consultant A? What was your overall impression of Consultant B? Do you have any other comments or observations about the Consultants or this study?
149
Appendix B: Experimental Materials – Online, Including Study 2
Resumes and Performance Reviews (Study 1 Online and Study 2) The resumes and performance reviews used online for Study 1 were the
same as those used in the in-person procedures (shown in Appendix A). The
same resumes and performance reviews were also used in Study 2, with the
exception of the additional comments section of the performance reviews, which
read either “Team members report that she is generally easy to work with.”
(control) or “Team members report that she is generally easy to work with, but
recently got angry at someone who works for her. This seems to have been a
one-time occurrence.” (manipulation).
Set-Up Script (Study 1 Online and Study 2)
Text of first email to subject:
“Thanks for signing up to participate in this study. I am going to tell you a
little bit about today’s study and then get you started. Please be sure you have
sent me your REP ID so you can receive credit at the end of the study.
This is a study about promotional decisions in the workplace. Questions
have been asked about how employers choose whether to promote one
employee over another. Our primary goal in this study is to find out how resumes
and performance reviews affect decisions about employees’ job title and salary.
Today you will review the resumes and performance reviews of two individuals,
Consultant A and Consultant B. These are actual resumes and performance
reviews from two women working at a consulting firm in the Bay Area. These
employees have each been with the company for two years as Consultants and
150
are up for promotion for the same position, Project Manager. You will carefully
consider each individual’s resume and her performance review from the previous
year, then answer some questions about the employees and how you think they
should be compensated.
Before we begin, I need to have you read and sign the consent form. You
will find it here: (link to online consent form, shown below). The consent form will
not be linked to any of your other responses, and the data cannot be traced to
the consent form. Thus, your data will remain anonymous. Please go read and
sign the consent form, then send me an email letting me know you are ready to
receive the documents.”
Text of second email to subject:
“Attached, you will find two resumes and two performance reviews. Please
take ten to fifteen minutes to review the documents and make notes in Notepad
or Microsoft Word. When you have finished, please send me your notes, and I
will send you a link to a questionnaire. The study is complete once you have
completed the questionnaire. Please feel free to email me with any questions.”
Text of third email to subject:
“Great, thanks! After you take the survey, you'll receive credit.
IMPORTANT: Your REP ID should not be used when you take the survey. Use
ID number ___ as your ID. Here's the link to the final survey - (link to appropriate
[male or female] version of online survey)”
151
Online Questionnaire Formatting
The survey administered in the online studies (both Study 1 and Study 2)
included the same questions as the in-person version of the study. Figures B.1
through B.3 show examples of the formatting of questions in the online survey.
For the full text of the survey, see Appendix A.
152
Consent Form – Online39 Informed Consent 1. Nonmedical Human Participants Consent Form
STUDY TITLE: Effect of Advisor Reports on Hiring Decisions - REP (Research Experience Program) Protocol Director: Kristen Backor DESCRIPTION: You are invited to participate in a research study on how advisor reports and resume information influence job and social decisions. You will be asked to review this information and fill out some related questionnaires. RISKS AND BENEFITS: There are no risks associated with this study. You may benefit by learning about social research and how it is performed. We cannot and do not guarantee or promise that you will receive any benefits from this study. If you have any questions regarding this study, please contact Protocol Director Kristen Backor at [redacted] or via email at [redacted]. TIME INVOLVEMENT: The session will take about 30 minutes. PAYMENTS: You will receive 1 credit toward your course requirement through the REP. SUBJECT'S RIGHTS: If you have read this form and have decided to participate in this project, please understand your participation is voluntary and you have the right to withdraw your consent or discontinue participation at any time without penalty. Your instructor will provide you with an alternative assignment. You have the right to refuse to answer particular questions. Your individual privacy will be maintained in all published and written data resulting from the study. CONTACT INFORMATION: *Questions, Concerns, or Complaints: If you have any questions, concerns or complaints about this research study, its procedures, risks and benefits, you should ask the Protocol Director, Kristen Backor at [redacted]. *Independent Contact: If you are not satisfied with how this study is being conducted, or if you have any concerns, complaints, or general questions about the research or your rights as a participant, please contact the Stanford Institutional Review Board (IRB) to speak to someone independent of the research team at (650)-723-2480 or toll free at 1-866-680-2906. You can also write to the Stanford IRB, Stanford University, Stanford, CA 94305-5401. I have read this consent form and the risks and benefits of this study have been explained to me. I understand that my participation is voluntary and that I may choose to withdraw at any time. I understand that I will receive course credit for my participation and agree to participate in this research study. Please check the box below and type your name to indicate your consent. This survey will not be linked to your data, so your data will remain anonymous.
39 See Figure B.4 for a screenshot of the consent box referenced in this form.
153
research team at (650)-723-2480 or toll free at 1-866-680-2906. You can also write to the Stanford IRB, Stanford University, Stanford, CA 94305-5401. I have read this consent form and the risks and benefits of this study have been explained to me. I understand that my participation is voluntary and that I may choose to withdraw at any time. I understand that I will receive course credit for my participation and agree to participate in this research study. Please check the box below and type your name to indicate your consent. This survey will not be linked to your data, so your data will remain anonymous. Approval date 3/17/09 Expiration date 12/18/09
I have read this consent form and the risks and benefits of this study have been explained to me. I understand that my participation is voluntary and that I may choose to withdraw at any time. I understand that I will receive course credit for my participation. Please check the box below and type your name to indicate your consent. This survey will not be linked to your data, so your data will remain anonymous.
154
Figure B.1. Formatting of Initial Survey Questions Online
155
Figure B.2. Formatting of Relative Ranking Questions Online
156
Figure B.3. Formatting of Likert-Scale Survey Questions Online
157
Figure B.4. Screenshot of Online Consent Format
158
Figure B.5. Screenshot of Debriefing Questions
159
Figure B.6. Screenshot of Debriefing Text
160
Appendix C: Univariate Statistics and Randomization Checks
Tables C.1 and C.4 show the minimum, maximum, means, and, where
applicable, standard deviations for all variables used in the analyses for Studies
1 and 2, respectively. These univariate statistics are based on only those
subjects who were included in the final analysis. Table C.1 includes the
component variables created based on the PCA (described in Appendix D and
used in the analyses in Chapters 4 and 5).
I conducted chi-square tests to determine whether conditions were
randomly distributed across control variables in Study 1 (see Table C.2).
Significant differences were found with regard to white/non-white (though not any
of the other race dummies or race in general, as tested by ANOVA [not shown])
and science/engineering majors compared to other majors (though not any of the
other major dummies or major in general, as tested by ANOVA [not shown]).
These variables were controlled for in the analyses and were not significant
except in two of the mediational analyses using the difference components. A
significant difference was also observed for Stanford University/Foothill College.
This occurred as a result of randomization in the data collection process; as the
data collection phase of the experiment ended in Fall 2009, only female/frequent
occurrence trials remained. Data collection in the fall at Stanford University
began slightly earlier than collection at Foothill College, so these trials took place
primarily at Stanford University. Again, this variable was controlled for in the
analyses and significant only in a single mediational analysis using the difference
components.
161
I also conducted chi-square tests to determine whether the materials were
randomly distributed across control variables in Study 1 (see Table C.3); the
possible combinations of materials are denoted by letters identifying the resume
(C or S) and performance review (A or T) used in each case.40 A slightly
significant difference was found for science/engineering majors compared to
other majors. As noted previously, this variable was controlled for in the analyses
and was significant only in two mediational analyses using the difference
components. With the exception of the Stanford/Foothill difference explained
above, all differences by control variables described above appear to be artifacts
rather than actual failures of randomization.
Finally, I conducted chi-square tests to determine whether the gender of
the consultants in the directed anger condition was randomly distributed across
control variables in Study 2 (see Table C.5). No significant differences were
found. The Study 2 sample was too small to conduct these analyses effectively
for the materials, but there was no change in procedure for assigning these
materials from Study 1 and no differences were expected. All Study 2 tables omit
the dummy variables for Stanford students, online respondents, and frequent
expression of anger; since all Study 2 subjects were online Foothill college
students comparing a neutral consultant to a consultant who had been angry
once, there was no differentiation on these measures in Study 2.
40 To avoid confusion, A and B were not used except to identify the documents to subjects during the study. Instead, the letters used here identify the resumes and performance reviews based on their content. C and S refer to the two resumes according to a formatting difference: whether they have their contact information located in the center (C) or on the sides (S). A and T refer to the first lines of the performance reviews (“acquired” vs. “took on”, respectively).
162
Table C.1. Univariate Statistics for Study 1 Min Max Mean Std.
Dev. Independent Variables Frequent Occurrence of Anger 0 1 .45 Female Employee 0 1 .51 Breakdown by Condition (Gender and Frequency of Anger):
Female Employee, One-Time Occurrence 0 1 .28 Female Employee, Frequent Occurrence 0 1 .23 Male Employee, One-Time Occurrence 0 1 .27 Male Employee, Frequent Occurrence 0 1 .22 Control Variables Age 22 Years or Younger 0 1 .74 23 Years or Older 0 1 .26 Female Respondent 0 1 .61 Race White 0 1 .40 Asian 0 1 .28 Other Ethnicity 0 1 .32 Major Business 0 1 .11 Social Science/Humanities 0 1 .38 Medical/Nursing 0 1 .13 Science/Engineering 0 1 .20 Undecided 0 1 .19 Stanford Participant 0 1 .23 Online Participant 0 1 .29 Dependent Variables Binary Quantitative Measures Choose Angry Consultant as Boss 0 1 .24 Choose to Work with Angry Consultant 0 1 .30 Choose to Sit Next to Angry Consultant on Plane
0 1 .26
Choose to Promote Angry Consultant 0 1 .39
163
Table C.1 (continued). Univariate Statistics for Study 1 Ordinal Quantitative Measures Power Should Have (Neutral [N]) 4 11 8.34 1.47 Power Should Have (Angry [A]) 3 11 7.92 1.34 Power Should Have (Difference [A-N]) -5 5 -0.41 1.49 Independence Should Have (N) 3 11 8.29 1.55 Independence Should Have (A) 4 11 8.14 1.54 Independence Should Have (Diff. [A-N]) -7 6 -0.15 1.69 Status Should Have (N) 3 11 8.27 1.47 Status Should Have (A) 3 11 7.96 1.47 Status Should Have (Diff. [A-N]) -8 5 -0.31 1.50 Competent (N) 3 7 6.23 0.93 Competent (A) 1 7 6.05 1.08 Competent (Diff. [A-N]) -6 2 -0.17 1.17 Respected (N) 2 7 5.97 1.13 Respected (A) 1 7 5.32 1.32 Respected (Diff. [A-N]) -6 4 -0.65 1.49 Pleasant (N) 1 7 5.92 1.18 Pleasant (A) 1 7 4.50 1.67 Pleasant (Diff. [A-N]) -6 2 -1.42 1.74 Powerful (N) 1 7 5.32 1.22 Powerful (A) 1 7 5.34 1.23 Powerful (Diff. [A-N]) -3 3 0.01 1.27 Likeable (N) 1 7 5.91 1.17 Likeable (A) 1 7 4.70 1.59 Likeable (Diff. [A-N]) -6 3 -1.21 1.67 Knowledgeable (N) 1 7 5.90 1.17 Knowledgeable (A) 1 7 5.88 1.23 Knowledgeable (Diff. [A-N]) -3 2 -0.21 0.99 High Status (N) 3 7 5.62 1.09 High Status (A) 1 7 5.49 1.20 High Status (Diff. [A-N]) -6 2 -0.13 1.16 Leader (N) 1 7 5.64 1.35 Leader (A) 1 7 5.65 1.43 Leader (Diff. [A-N]) -6 6 0.01 1.65 Cooperative (N) 2 7 6.04 1.19 Cooperative (A) 1 7 4.60 1.69 Cooperative (Diff. [A-N]) -6 3 -1.44 1.88
164
Table C.1 (continued). Univariate Statistics for Study 1 Intelligent (N) 3 7 6.05 0.94 Intelligent (A) 1 7 6.05 1.02 Intelligent (Diff. [A-N]) -2 2 -0.01 0.87 Trustworthy (N) 1 7 5.88 1.32 Trustworthy (A) 1 7 5.51 1.41 Trustworthy (Diff. [A-N]) -6 4 -0.37 1.37 Confident in Qualifications (N) 1 5 4.07 0.78 Confident in Qualifications (A) 1 5 3.87 0.74 Confident in Qualifications (Diff. [A-N]) -4 2 -0.20 1.02 Confident of Success (N) 1 5 4.04 0.76 Confident of Success (A) 1 5 3.82 0.85 Confident of Success (Diff. [A-N]) -4 2 -0.22 1.03 Worried about Qualifications (N) 1 5 1.71 0.93 Worried about Qualifications (A) 1 5 1.97 0.93 Worried about Qualifications (Diff. [A-N]) -2 3 0.26 0.95 Others Believe Qualified (N) 2 5 4.02 0.72 Others Believe Qualified (A) 1 5 3.58 0.87 Others Believe Qualified (Diff. [A-N]) -4 2 -0.44 1.00 Similar to Existing Project Managers (N) 2 5 3.58 0.74 Similar to Existing Project Managers (A) 1 5 3.50 0.76 Similar to Existing Project Managers (Diff. [A-N])
-4 2 -0.09 0.95
Likely to Enjoy Socializing (N) 1 5 3.76 0.89 Likely to Enjoy Socializing (A) 1 5 3.05 1.04 Likely to Enjoy Socializing (Diff. [A-N]) -4 2 -0.71 1.12 Likely to Be Friends With (N) 1 5 3.43 0.95 Likely to Be Friends With (A) 1 5 2.90 0.88 Likely to Be Friends With (Diff. [A-N]) -3 2 -0.53 0.89 Likely to Ask for Work Advice (N) 1 5 4.08 0.77 Likely to Ask for Work Advice (A) 1 5 3.64 1.09 Likely to Ask for Work Advice (Diff. [A-N]) -4 2 -0.44 1.11 Likely to Ask for Personal Advice (N) 1 5 2.63 1.23 Likely to Ask for Personal Advice (A) 1 5 2.15 1.09 Likely to Ask for Personal Advice (Diff. [A- N]) -4 4 -0.48 1.01
165
Table C.1 (continued). Univariate Statistics for Study 1 Likely to Enjoy as Co-Worker (N) 1 5 3.87 0.75 Likely to Enjoy as Co-Worker (A) 1 5 3.09 0.96 Likely to Enjoy as Co-Worker (Diff. [A-N]) -4 1 -0.78 1.06 Risky to Promote (N) 1 5 1.92 0.94 Risky to Promote (A) 1 5 2.31 0.94 Risky to Promote (Diff. [A-N]) -3 4 0.38 1.09 Employees Assigned To (N) 1 160 16.22 24.31 Employees Assigned To (A) 0 100 13.61 18.96 Employees Assigned To (Diff. [A-N]) -110 40 -2.61 14.92 Factor Variables Manipulation Sociability Factor -2.39 2.30 0 1 Suitability for Promotion Factor -3.22 2.75 0 1 Knowledgeability Factor -3.59 3.43 0 1 Difference Sociability Factor -3.32 2.46 0 1 Suitability for Promotion Factor -4.38 2.79 0 1 Knowledgeability Factor -4.07 3.60 0 1 Note: Ns range from 123 (for factor variables) to 143.
166
Table C.2. Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Condition Female/
One-Time Occurrence
Female/ Frequent
Occurrence
Male/ One-Time
Occurrence
Male/ Frequent
Occurrence Age 22 Years or Younger
26 (65.0%)
26 (78.8%)
29 (76.3%)
24 (77.4%)
23 Years or Older 14 (35.0%)
7 (21.2%)
9 (23.7%)
7 (22.6%)
χ2 = 2.37
Respondent Gender Female 22
(55.0%) 18
(54.5%) 26
(68.4%) 21
(67.7%) Male 18
(45.0%) 15
(45.5%) 12
(31.6%) 10
(32.3%) χ2 = 2.66
Race White 19
(50.0%) 17
(51.5%) 14
(38.9%) 5
(16.7%) Non-White 19
(50.0%) 16
(48.5%) 22
(61.1%) 25
(83.3%) χ2 = 10.22* Asian 6
(15.8%) 8
(24.2%) 12
(33.3%) 12
(40.0%) Non-Asian 32
(84.2%) 25
(75.8%) 24
(66.7%) 18
(60.0%) χ2 = 5.72 Other Ethnicity 13
(34.2%) 8
(24.2%) 10
(27.8%) 13
(43.3%) White/Asian 25
(65.8%) 25
(75.8%) 26
(72.2%) 17
(56.7%) χ2 = 3.06
167
Table C.2 (continued). Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Condition Major Business 6
(15.0%) 3
(9.1%) 4
(10.8%) 2
(6.5%) Non-Business 34
(85.0%) 30
(90.9%) 33
(89.2%) 29
(93.5%) χ2 = 1.46 Social Sciences 11
(27.5%) 11
(33.3%) 17
(45.9%) 14
(45.2%) Non-Social Sciences 29
(72.5%) 22
(66.7%) 20
(54.1%) 17
(54.8%) χ2 = 3.85 Medical/Nursing 6
(15.0%) 1
(3.0%) 6
(16.2%) 5
(16.1%) Non-Medical/Nursing 34
(85.0%) 32
(97.0%) 31
(83.8%) 26
(83.9%) χ2 = 3.70 Science/Engineering 6
(15.0%) 14
(42.4%) 4
(10.8%) 4
(10.8%) Non-Science/Engineering 34
(85.0%) 19
(57.6%) 33
(89.2%) 27
(87.1%) χ2 =14.00** Undecided 11
(27.5%) 4
(12.1%) 6
(16.2%) 6
(19.4%) All Majors 29
(72.5%) 29
(87.9%) 31
(83.8%) 25
(80.6%) χ2 =3.06
School Stanford University 7
(17.5%) 15
(45.5%) 4
(10.3%) 7
(22.6%) Foothill College 33
(82.5%) 18
(54.5%) 35
(89.7%) 24
(77.4%) χ2 = 13.63**
168
Table C.2 (continued). Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Condition Location Online 14
(35.0%) 6
(18.2%) 14
(35.9%) 8
(25.8%) In-Person 26
(65.0%) 27
(81.8%) 25
(64.1%) 23
(74.2%) χ2 = 3.59 Notes: Significance values are based on LR-χ2 tests when all cells have at least 5 participants in
them and on Fisher’s exact tests when they do not. *p ≤ .05 **p < .01 ***p<.001
169
Table C.3. Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Materials 1 (CT) 2 (CA) 3 (ST) 4 (SA) Age 22 Years or Younger
29 (69.0%)
23 (82.1%)
25 (73.5%)
26 (72.2%)
23 Years or Older 13 (31.0%)
5 (17.9%)
9 (26.5%)
10 (27.8%)
χ2 = 1.53
Respondent Gender Female 28
(66.7%) 15
(53.6%) 23
(67.6%) 20
(55.6%) Male 14
(33.3%) 13
(46.4%) 11
(32.4%) 16
(44.4%) χ2 = 2.30
Race White 14
(35.0%) 12
(44.4%) 14
(42.4%) 15
(42.9%) Non-White 26
(65.0%) 15
(55.6%) 19
(57.6%) 20
(57.1%) χ2 = 0.83 Asian 15
(37.5%) 4
(14.8%) 8
(24.2%) 10
(28.6%) Non-Asian 25
(62.5%) 23
(85.2%) 25
(75.8%) 25
(71.4%) χ2 = 4.39 Other Ethnicity 11
(27.5%) 11
(40.7%) 11
(33.3%) 10
(28.6%) White/Asian 29
(72.5%) 16
(59.3%) 22
(66.7%) 25
(71.4%) χ2 = 1.54
Major Business 7
(16.7%) 3
(10.7%) 3
(8.8%) 2
(5.7%) Non-Business 35
(83.3%) 25
(89.3%) 31
(91.2%) 33
(94.3%) χ2 = 2.58
170
Table C.3 (continued). Randomization Checks (Study 1) – Chi-Square Tests for Control Variables, by Materials Social Sciences 14
(33.3%) 6
(21.4%) 16
(47.1%) 15
(42.9%) Non-Social Sciences 28
(66.7%) 22
(78.6%) 18
(52.9%) 20
(57.1%) χ2 = 5.16 Medical/Nursing 3
(7.1%) 5
(17.9%) 3
(8.8%) 7
(20.0%) Non-Medical/Nursing 39
(92.9%) 23
(82.1%) 31
(91.2%) 28
(80.0%) χ2 = 3.91 Science/Engineering 6
(14.3%) 12
(42.9%) 5
(14.7%) 5
(14.3%) Non-Science/Engineering
36 (85.7%)
16 (57.1%)
29 (85.3%)
30 (85.7%)
χ2 =11.25* Undecided 12
(28.6%) 2
(7.1%) 7
(20.6%) 6
(17.1%) All Majors 30
(71.4%) 26
(92.9%) 27
(79.4%) 29
(82.9%) χ2 =5.09
School Stanford University 8
(19.0%) 8
(28.6%) 6
(17.6%) 11
(29.7%) Foothill College 34
(81.0%) 20
(71.4%) 28
(82.4%) 26
(70.3%) χ2 = 2.32
Location Online 15
(35.7%) 6
(21.4%) 9
(26.5%) 10
(27.0%) In-Person 27
(64.3%) 22
(78.6%) 25
(73.5%) 27
(73.0%) χ2 = 1.87 Notes: Significance values are based on LR-χ2 tests when all cells have at least 5 participants in
them and on Fisher’s exact tests when they do not. The four columns represent the possible combinations of two resumes (C and S) and two performance reviews (A and T). *p ≤ .05 **p < .01 ***p<.001
171
Table C.4. Univariate Statistics for Study 2 Min Max Mean Std.
Dev. Independent Variables Female Employee 0 1 .51 Control Variables Age 22 Years or Younger 0 1 .68 23 Years or Older 0 1 .32 Female Respondent 0 1 .62 Race White 0 1 .46 Asian 0 1 .29 Other Ethnicity 0 1 .26 Major Business 0 1 .13 Social Science/Humanities 0 1 .31 Medical/Nursing 0 1 .28 Science/Engineering 0 1 .26 Undecided 0 1 .03 Dependent Variables Binary Quantitative Measures Choose Angry Consultant as Boss 0 1 .21 Choose to Work with Angry Consultant 0 1 .41 Choose to Sit Next to Angry Consultant on Plane 0 1 .36 Choose to Promote Angry Consultant 0 1 .36 Ordinal Quantitative Measures Power Should Have (Neutral [N]) 5 11 8.49 1.48 Power Should Have (Angry [A]) 3 11 8.10 1.76 Power Should Have (Difference [A-N]) -5 2 -0.38 1.37 Independence Should Have (N) 6 11 8.54 1.29 Independence Should Have (A) 4 11 8.14 1.54 Independence Should Have (Diff. [A-N]) -3 2 -0.15 1.14 Status Should Have (N) 6 11 8.46 1.25 Status Should Have (A) 3 11 7.96 1.47 Status Should Have (Diff. [A-N]) -3 3 -0.18 1.12
172
Table C.4 (continued). Univariate Statistics for Study 2 Competent (N) 2 7 6.21 1.42 Competent (A) 1 7 6.03 1.55 Competent (Diff. [A-N]) -2 1 -0.18 0.56 Respected (N) 2 7 6.26 1.27 Respected (A) 1 7 5.62 1.53 Respected (Diff. [A-N]) -5 1 -0.64 1.04 Pleasant (N) 1 7 6.13 1.17 Pleasant (A) 1 7 5.38 1.71 Pleasant (Diff. [A-N]) -6 1 -0.74 1.37 Powerful (N) 3 7 5.72 0.97 Powerful (A) 4 7 5.74 0.94 Powerful (Diff. [A-N]) -2 4 0.03 1.09 Likeable (N) 2 7 6.44 1.18 Likeable (A) 1 7 5.72 1.49 Likeable (Diff. [A-N]) -6 1 -0.72 1.26 Knowledgeable (N) 2 7 6.21 1.06 Knowledgeable (A) 1 7 6.13 1.15 Knowledgeable (Diff. [A-N]) -2 1 -0.08 0.77 High Status (N) 2 7 6.15 1.11 High Status (A) 2 7 6.08 1.13 High Status (Diff. [A-N]) -2 1 -0.08 0.91 Leader (N) 3 7 6.08 0.98 Leader (A) 5 7 6.26 0.68 Leader (Diff. [A-N]) -2 4 0.18 1.14 Cooperative (N) 3 7 6.21 1.06 Cooperative (A) 2 7 5.51 1.37 Cooperative (Diff. [A-N]) -5 1 -0.69 1.20 Intelligent (N) 5 7 6.30 0.74 Intelligent (A) 4 7 6.32 0.71 Intelligent (Diff. [A-N]) -1 1 0.03 0.37 Trustworthy (N) 2 7 6.62 1.02 Trustworthy (A) 2 7 6.44 1.21 Trustworthy (Diff. [A-N]) -4 1 -0.18 0.72 Confident in Qualifications (N) 3 5 4.36 0.71 Confident in Qualifications (A) 2 5 4.03 0.78 Confident in Qualifications (Diff. [A-N]) -3 1 -0.33 1.03
173
Table C.4 (continued). Univariate Statistics for Study 2 Confident of Success (N) 3 5 4.38 0.67 Confident of Success (A) 2 5 4.05 0.73 Confident of Success (Diff. [A-N]) -3 1 -0.34 0.91 Worried about Qualifications (N) 1 5 1.72 1.00 Worried about Qualifications (A) 1 5 1.92 1.01 Worried about Qualifications (Diff. [A-N]) -1 3 0.21 0.77 Others Believe Qualified (N) 2 5 4.08 0.81 Others Believe Qualified (A) 2 5 3.62 0.91 Others Believe Qualified (Diff. [A-N]) -3 1 -0.46 0.79 Similar to Existing Project Managers (N) 2 5 3.69 0.80 Similar to Existing Project Managers (A) 2 5 3.46 0.79 Similar to Existing Project Managers (Diff. [A-N]) -3 1 -0.23 0.78 Likely to Enjoy Socializing (N) 2 5 4.00 1.00 Likely to Enjoy Socializing (A) 1 5 3.49 0.88 Likely to Enjoy Socializing (Diff. [A-N]) -4 2 -0.51 1.05 Likely to Be Friends With (N) 2 5 3.74 0.97 Likely to Be Friends With (A) 1 5 3.36 1.01 Likely to Be Friends With (Diff. [A-N]) -4 1 -0.38 0.96 Likely to Ask for Work Advice (N) 1 5 3.95 0.96 Likely to Ask for Work Advice (A) 1 5 3.74 0.97 Likely to Ask for Work Advice (Diff. [A-N]) -3 2 -0.21 1.04 Likely to Ask for Personal Advice (N) 1 5 2.59 1.35 Likely to Ask for Personal Advice (A) 1 5 2.28 1.15 Likely to Ask for Personal Advice (Diff. [A-N]) -3 1 -0.31 0.95 Likely to Enjoy as Co-Worker (N) 2 5 4.03 0.78 Likely to Enjoy as Co-Worker (A) 1 5 3.56 0.88 Likely to Enjoy as Co-Worker (Diff. [A-N]) -4 1 -0.46 0.85 Risky to Promote (N) 1 5 1.64 0.93 Risky to Promote (A) 1 5 2.33 1.03 Risky to Promote (Diff. [A-N]) -2 4 0.69 1.03 Employees Assigned To (N) 2 100 18.69 19.62 Employees Assigned To (A) 2 75 14.44 15.81 Employees Assigned To (Diff. [A-N]) -50 25 -4.15 11.39 Note: Ns range from 37 to 39.
174
Table C.5. Randomization Checks (Study 2) – Chi-Square Tests for Control Variables, by Gender of Consultants Male
Consultants Female
Consultants Age 22 Years or Younger 12
(63.2%) 14
(73.7%) 23 Years or Older 7
(36.8%) 5
(26.3%) χ2 = 0.49
Respondent Gender Female 13
(68.4%) 11
(55.0%) Male 6
(31.6%) 9
(45.0%) χ2 = 0.74
Race White 9
(52.9%) 11
(61.1%) Non-White 8
(47.1%) 7
(38.9%) χ2 = 0.70 Asian 4
(23.5%) 6
(33.3%) Non-Asian 13
(76.5%) 12
(66.7%) χ2 = 0.41 Other Ethnicity 4
(23.5%) 5
(27.8%) White/Asian 13
(76.5%) 13
(72.2%) χ2 = 0.08
Major Business 2
(10.5%) 3
(15.0%) Non-Business 17
(89.5%) 17
(85.0%) χ2 = 0.17
175
Table C.5 (continued). Randomization Checks (Study 2) – Chi-Square Tests for Control Variables, by Gender of Consultants Social Sciences 6
(31.6%) 6
(30.0%) Non-Social Sciences 13
(68.4%) 14
(70.0%) χ2 = 0.01 Medical/Nursing 5
(26.3%) 6
(30.0%) Non-Medical/Nursing 14
(73.7%) 14
(70.0%) χ2 = 0.07 Science/Engineering 5
(26.3%) 5
(25.0%) Non-Science/Engineering 14
(73.7%) 15
(75.0%) χ2 =0.01 Undecided 1
(5.3%) 0
(0.0%) All Majors 18
(94.7%) 20
(100.0%) χ2 =1.08 Notes: Significance values are based on LR-χ2 tests when all cells have at least 5 participants
in them and on Fisher’s exact tests when they do not. *p ≤ .05 **p < .01 ***p<.001
176
Appendix D: Principal Component Analyses of Questionnaire Variables
As noted in Chapters 4 and 5, I used Principal Component Analysis (PCA)
to create factors based on the relationships between responses to the
questionnaire given at the end of the study. PCA allows me to organize the large
set of variables generated from my questionnaire before conducting further
analysis on these factors. I conducted two sets of PCA on the ordinal measures
and the number of employees assigned to the consultant(s): the first used ratings
of the angry consultant, and the other used the difference in ratings between the
two consultants. I did not conduct PCA on the ratings of the neutral consultant
only, because I chose to focus primarily on the angry individual and how he or
she is perceived as different from the neutral individual. The structure of the
questions means that the ratings for the neutral consultant are given in relation to
those of the angry individual, which is captured in the PCAs of the difference
variables. In both cases (angry consultant ratings and differences in ratings
between the two consultants), the scree plots (see Figures D.1 and D.2)
suggested three components: a clearly job-related component, an intelligence
component, and a sociability component. As described below, the three
components were similar across both PCAs.
The factor loadings for the components created from the manipulation
variables alone are shown in Table D.1.41 These factor scores are based on the
results of regression analyses. The “Sociability” component loads highly on
41 These factor loadings and those shown in Table D.2 are based on varimax rotation, the most commonly used PCA rotation; this rotation reduces the complexity of the analysis by increasing the size of large loadings and decreasing the size of small loadings for each component.
177
factors such as “Pleasant”, “Likely to Be Friends With”, and “Likely to Enjoy
Socializing With”. The “Suitability for Promotion” component loads highly on
factors like “Respected” and “Risky to Promote” (negative). Finally, the
“Knowledgeability” component loads highly on factors such as “Knowledgeable”
and “Intelligent”. Together, these components explain 43.17% of the variability in
the data (see Table D.3 for the eigenvalues and variance explained for the
individual components); “Knowledgeability”, the weakest component, accounts
for less than six percent of this.
The factor loadings for the components created from the differences
between ratings of the angry individual and ratings of the neutral individual are
shown in Table D.2. While the loadings differ slightly from those created with the
manipulation variables only, the three components are essentially the same.
Again, there is a “Sociability” component loads highly on factors such as
“Pleasant”, “Likely to Be Friends With”, and “Likely to Enjoy Socializing With”.
The “Knowledgeability” component loads highly on factors such as
“Knowledgeable” and “Intelligent”, and, in this case, “Competent”. The “Suitability
for Promotion” component differs the most from its manipulation-rated
counterpart; it loads highly on factors like “Status Should Have”, “Confident of
Success” and again “Risky to Promote” (negative). Together, these components
explain 52.02% of the variability in the data (see Table D.3 for the components’
individual eigenvalues and variance explained). “Knowledgeability” is again the
weakest component, representing less than six percent of the variance
explained.
178
Figure D.1. Scree Plot of Eigenvalues for Components Based on Principal Component Analysis of Manipulation Variables
179
Figure D.2. Scree Plot of Eigenvalues for Components Based on Principal Component Analysis of Difference Variables
180
Table D.1. Factor Loadings for Principal Component Analysis Using Manipulation Factors Sociability
Component Suitability for
Promotion Component
Knowledgeability Component
Power Should Have -0.07 0.19 0.17 Independence Should Have
0.06 0.02 0.31
Status Should Have 0.09 0.12 0.06 Competent -0.08 0.15 0.36 Respected 0.16 0.61 0.01 Pleasant 0.69 0.20 -0.11 Powerful 0.09 0.17 0.56 Likeable 0.58 0.42 -0.10 Knowledgeable 0.05 0.21 0.74 High Status 0.17 0.14 0.34 Leader 0.29 -0.15 0.57 Cooperative 0.43 0.52 0.06 Intelligent -0.15 0.20 0.77 Trustworthy 0.24 0.40 0.24 Confident in Qualifications 0.21 0.29 0.31 Confident of Success 0.24 0.39 0.14 Worried About Qualifications
0.13 -0.60 -0.35
Others Believe Qualified 0.12 0.34 0.01 Similar to Existing Managers
0.06 -0.03 0.03
Likely to Enjoy Socializing With
0.80 0.03 -0.02
Likely to Be Friends With 0.79 0.00 0.25 Likely to Ask for Work Advice
0.44 0.38 0.12
Likely to Ask for Personal Advice
0.66 -0.10 -0.09
Likely to Enjoy as Co- Worker
0.69 0.39 0.03
Risky to Promote -0.13 -0.80 -0.19 Employees Assigned To# 0.21 -0.08 0.05 Notes: Based on PCA of all ordinal variables and one continuous variable (marked with #) for the angry consultant. N=125
181
Table D.2. Factor Loadings for Principal Component Analysis Using Difference Factors Sociability
Component Suitability for
Promotion Component
Knowledgeability Component
Power Should Have 0.13 0.68 0.31 Independence Should Have
0.14 0.51 0.58
Status Should Have 0.12 0.76 0.27 Competent 0.05 0.08 0.62 Respected 0.26 0.37 0.01 Pleasant 0.73 0.13 0.07 Powerful -0.11 0.27 0.19 Likeable 0.68 0.28 0.00 Knowledgeable 0.00 0.17 0.71 High Status 0.12 0.38 0.29 Leader -0.05 0.26 0.28 Cooperative 0.65 0.21 0.27 Intelligent -0.07 0.07 0.67 Trustworthy 0.60 0.20 0.23 Confident in Qualifications 0.13 0.63 0.18 Confident of Success 0.13 0.72 0.19 Worried About Qualifications
-0.11 -0.45 -0.29
Others Believe Qualified 0.09 0.67 0.04 Similar to Existing Managers
0.03 0.66 -0.07
Likely to Enjoy Socializing With
0.71 0.22 0.00
Likely to Be Friends With 0.75 0.04 0.02 Likely to Ask for Work Advice
0.28 0.55 0.09
Likely to Ask for Personal Advice
0.74 -0.19 -0.17
Likely to Enjoy as Co- Worker
0.77 0.30 -0.05
Risky to Promote -0.29 -0.68 0.02 Employees Assigned To# 0.19 0.16 0.04 Notes: Based on PCA of the difference between the angry consultant and the neutral consultant using all ordinal variables and one continuous variable (marked with #). N = 123
182
Table D.3. Eigenvalues and Percent Explained for Each Component (Created with Principal Component Analysis) Initial
Eigenvalues
Percent of Variance Explained
Manipulation Knowledgeability 1.80 5.99% Sociability 7.77 25.90% Suitability for Promotion 3.38 11.28% Difference (Angry – Neutral) Knowledgeability 1.48 5.68% Sociability 3.79 14.59% Suitability for Promotion 8.25 31.75% Note: Based on PCAs of the ratings on all ordinal variables and the values for one continuous variable for the angry consultant (Manipulation) and for the angry consultant minus the neutral consultant (Difference).
183
List of References Algoe, S. B., Buswell, B. N., & DeLamater, J. D. (2000). Gender and job status
as contextual cues for the interpretation of facial expression of emotion. Sex Roles, 42 (3/4), 183-208.
Ames, D. R., & Flynn, F. J. (2007). What breaks a leader: The curvilinear relation
between assertiveness and leadership. Journal of Personality and Social Psychology, 92 (2), 307-324.
Berger, J., Cohen, B. P., & Zelditch, M. J. (1972). Status characteristics and
social interaction. American Sociological Review, 37 (3), 241-255. Berger, J., Rosenholtz, S. J., & Zelditch, M. J. (1980). Status organizing process.
Annual Review of Sociology, 6, 477-508. Berger, J., Wagner, D., & Zelditch, M. J. (1985). Introduction: Expectations states
theory. In J. Berger, & M. J. Zelditch (Eds.), Status, rewards, and influence (pp. 1-72). San Francisco: Jossey-Bass.
Blake, C. (2008, October 22). Joe Biden's Tears Ease a Father's Fears.
Retrieved April 15, 2009, from the Huffington Post: http://www.huffingtonpost.com/chris-blake/joe-bidens-tears-ease-a-f_b_137003.html
Brescoll, V. L., & Uhlmann, E. L. (2008). Can an angry woman get ahead? Status
conferral, gender, and expression of emotion in the workplace. Psychological Science, 19 (3), 268-275.
Brody, L., & Hall, J. (1992). Gender and emotion. In M. Lewis, & J. Haviland
(Eds.), Handbook of Emotions (pp. 447-460). New York: Guilford Press. Brower, H.H., Schoorman, F.D., & Hwee, H.T. (2000). A model of relational
leadership: The integration of trust and leader-member exchange. Leadership Quarterly, 11 (2), 227-250.
Bureau of Labor Statistics. (2008, September 26). Employee Tenure in 2008.
Retrieved April 20, 2009, from the Bureau of Labor Statistics: http://www.bls.gov/news.release/tenure.nr0.htm
Burgoon, J., & Hale, J. (1988). Nonverbal expectancy violations: Model
elaboration and application to immediacy behaviors. Communication Monographs, 55, 58-79.
184
Buttner, E. H., & McEnally, M. (1996). The interactive effect of influence tactic, applicant gender, and type of job on hiring recommendations. Sex Roles, 34 (7/8), 581-591.
Carli, L. L., LaFleur, S. J., & Loeber, C. C. (1995). Nonverbal behavior, gender,
and influence. Journal of Personality and Social Psychology, 68 (6), 1030- 1041.
Cuddy, A.J.C., Fiske, S.T., & Glick, P. (2004). When professionals become
mothers, warmth doesn’t cut the ice. Journal of Social Issues, 60 (4), 701-718.
Correll, S. J., Benard, S., & Paik, I. (2007). Getting a job: Is there a motherhood
penalty? American Journal of Sociology, 112 (5), 1297-1338. Correll, S. J., & Ridgeway, C. L. (2003). Expectation States Theory. In J.
DeLamater (Ed.), Handbook of Social Psychology (pp. 29-52). New York, New York: Kluwer Academic/Plenum Publishers.
Dovidio, J. F., Brown, C. E., Heltman, K., Ellyson, S. L., & Keating, C. F. (1988).
Power displays between women and men in discussions of gender-linked tasks: A multichannel study. Journal of Personality and Social Psychology, 55 (4), 580-587.
Durkheim, E. (1951). Suicide: A study in sociology. New York: Free Press. Durkheim, E. (1961). The elementary forms of the religious life. New York:
Collier. Durkheim, E. (1984). The division of labor in society. New York: Free Press. Eagly, A. H. (1987). Sex differences in social behavior: A social-role
interpretation. Hillsdale, NJ: Erlbaum. Eagly, A. H. (2005). Achieving relational authenticity in leadership: Does gender
matter? The Leadership Quarterly, 16, 459-474. Eagly, A., & Johnson, B. (1990). Gender and leadership style: A meta-analysis.
Psychological Bulletin, 108, 233-256. Eagly, A.H., & Karau, S.J. (2002). Role congruity theory of prejudice toward
female leaders. Psychological Review, 109, 573-598. Eagly, A. H., Karau, S. J., & Makhijani, M. G. (1995). Gender and the
effectiveness of leaders: A meta-analysis. Psychological Bulletin, 117 (1), 125-145.
185
Eagly, A., Makhijani, M., & Klonsky, G. (1992). Gender and evaluation of leaders:
A meta-analysis. Psychological Bulletin, 111 (1), 3-22. Edwards, A.L. (1957). The social desirability variable in personality assessment
and research. New York: Dryden Press. Ekman, P., & Friesen, W. (1969). The repertoire of nonverbal behavior:
Categories, usage, and coding. Semiotica, 1, 49-98. Ekman, P., & Friesen, W.V. (1975). Unmasking the Face. Cambridge,
MA: Malor Books. Fisher, G.A., & Chon, K.K. (1989). Durkheim and the social construction of
emotions. Social Psychology Quarterly, 52 (1), 1-9. Fitness, J. (2000). Anger in the workplace: An emotion script approach to anger
episodes between workers and their superiors, co-workers and subordinates. Journal of Organizational Behavior, 21 (2), 147-162.
Forsythe, S. M. (2006). Effect of applicant's clothing on interviewer's decision to
hire. Journal of Applied Social Psychology, 20 (19), 1579-1595. Frijda, N. (1994). Varieties of affect: Emotions and episodes, moods, and
sentiments. In P. Ekman, & R. Davidson (Eds.), The Nature Emotion: Fundamental Questions (pp. 59-67). Oxford, UK: Oxford University Press.
Funkenstein, D.H., King, S.H., & Drolette, M. (1954). The direction of anger
during a laboratory stress-inducing situation. Psychosomatic Medicine, 16, 404-413.
Garcia-Retamero, R., & Lopez-Zafra, E. (2006). Prejudice against women in
male-congenial environments: Perceptions of gender role congruity in leadership. Sex Roles, 55, 51-61.
Georgeson, J. C., & Harris, M. J. (1998). Why's my boss always holding me
down? A meta-analysis of power effects on performance evaluations. Personality and Social Psychology Review, 2 (3), 184-195.
Gianakos, I. (2002). Issues of anger in the workplace: Do gender and gender role
matter? Career Development Quarterly, 51 (2), 155-171. Glick, P., & Fiske, S. T. (2001). Ambivalent Stereotypes as Legitimizing
Ideologies. In J. T. Jost, & B. Major (Eds.), The Psychology of Legitimacy: Emerging Perspectives on Ideology, Justice, and Intergroup Relations. New York: Cambridge University Press.
186
Glick, P., Wilk, K., & Perreault, M. (1995). Images of Occupations: Components
of Gender and Status in Occupational Stereotypes. Sex Roles, 32, 565-582.
Glomb, T.M. (2002). Workplace anger and aggression: Informing conceptual
models with data from specific encounters. Journal of Occupational Health Psychology, 7(1), 20-36.
Goffman, E. (1956). The nature of deference and demeanor. American
Anthropologist, 58 (3), 473-502. Goffman, E. (1959). The presentation of self in everyday life. New York: Anchor. Goleman, D. (1998). What makes a leader. Harvard Business Review, Nov-Dec,
93-102. Gross, J. J. (2002). Emotion regulation: Affective, cognitive, and social
consequences. Psychophysiology, 39, 281-291. Hareli, S., & Rafaeli, A. (2008). Emotion cycles: On the social influence of
emotion in organizations. Research in Organizational Behavior, 28, 35-59. Heilman, M. E. (2001). Description and prescription: How gender stereotypes
prevent women's ascent up the organizational ladder. Journal of Social Issues, 57 (4), 657-674.
Heilman, M. E., Wallen, A. S., Fuchs, D., & Tamkins, M. M. (2004). Penalties for
success: Reactions to women who succeed at male gender-typed tasks. Journal of Applied Psychology, 89 (3), 416-427.
Heilman, M.E., Block, C., Martell, R., & Simon, M. (1989). Has anything
changed? Current characterizations of men, women, and managers. Journal of Applied Psychology, 74, 935-942.
Hess, U., Adams, R. B., & Kleck, R. E. (2005). Who may frown and who should
smile? Dominance, affiliation, and the display of happiness and anger. Cognition and Emotion, 19 (4), 515-536.
Hochschild, A. R. (1979). Emotion Work, Feeling Rules, and Social Structure.
American Journal of Sociology, 85 (3), 551-575. Hochschild, A. R. (1983). The Managed Heart: The Commercialization of Human
Feeling. Berkeley: University of California Press.
187
Horstmann, G. (2003). What do facial expressions convey: Feeling states,
behavioral intentions, or action requests? Emotion, 3 (2), 150-166. Izard, C.E. (1972). Patterns of Emotions. New York: Academic Press. Jakobs, E., Fischer, A.H., & Manstead, A.S.R. (1997). Emotional experience as a
function of social context: The role of the other. Journal of Nonverbal Behavior, 21 (2), 103-130.
Johnson, C. (1993). Gender and formal authority. Social Psychology Quarterly,
56 (3), 193-210. Johnson, C., Clay-Warner, J., & Funk, S. J. (1996). Effects of authority structures
and gender on interaction in same-sex task groups. Social Psychology Quarterly, 59 (3), 221-236.
Johnson, S. K., Murphy, S. E., Zewdie, S., & Reichard, R. J. (2008). The strong,
sensitive type: Effects of gender stereotypes and leadership prototypes on the evaluation of male and female leaders. Organizational Behavior and Human Decision Processes, 106, 39-60.
Jones, E., & Pittman, T. (1982). Toward a general theory of strategic self-
presentation. In J. Suls (Ed.), Psychological perspectives on the self (pp. 231-262). Hillsdale, NJ: Erlbaum.
Kemper, T. D. (1991). Predicting Emotions from Social Relations. Social
Psychology Quarterly, 54 (4), 330-342. Konrad, A., J.E. Ritchie, J., Lieb, P., & Corrigall, E. (2000). Sex differences and
similarities in job attribute preferences: A meta-analysis. Psychological Bulletin, 126, 593-641.
Koritko, B. (n.d.). Interview tips: Set a positive tone. Retrieved 4 28, 2009, from
Employment 360: http://www.employment360.com/interview-tips-positive-tone.html
Kuppens, P., Van Mechelen, I., & Meulders, M. (2004). Every cloud has a silver
lining: Interpersonal and individual differences determinants of anger-related behaviors. Personality and Social Psychology Bulletin, 30 (12), 1550-1564.
LaFrance, M., & Hecht, M. (1999). Option or obligation to smile: The effects of
power and gender on facial expression. In P. Phillipot, R. Feldman, & E. Coats (Eds.), The Social Context of Nonverbal Behavior (pp. 45-70). Cambridge, UK: Cambridge University Press.
188
Lewis, K. (2000). When leaders display emotion: How followers respond to
negative emotional expression of male and female leaders. Journal of Organizational Behavior, 21, 221-234.
Linden, W., Leung, D., Chawla, A., Stossel, C., Rutledge, T., & Tance, S.A.
(1997). Social determinants of experienced anger. Journal of Behavioral Medicine, 20 (5), 415-432.
Lucas, J. W., & Lovaglia, M. J. (1998). Leadership status, gender, group size,
and emotion in face-to-face groups. Sociological Perspectives, 41 (3), 617-637.
Lyness, K. S., & Judiesch, M. K. (1999). Are women more likely to be hired or
promoted into management positions? Journal of Vocational Behavior, 54, 158-173.
Magee, J. C., & Tiedens, L. Z. (2006). Emotional ties that bind: The roles of
valence and consistency of group emotion in inferences of coheisveness and common fate. Personality and Social Psychology Bulletin, 32 (12), 1703-1715.
O’Connor, D.B., Archer, J., & Wu, F.W.C. (2001). Measuring aggression: Self-
reports, partner reports, and responses to provoking scenarios. Aggressive Behavior, 27, 79-101.
Petersen, T. and Saporta, I. (2004.) The opportunity structure for discrimination. American Journal of Sociology, 109, 852-901. Pierce, J. L. (1995). Gender Trials: Emotional Lives in Contemporary Law Firms.
Berkeley: University of California Press. Plant, E. A., Kling, K. C., & Smith, G. L. (2004). The influence of gender and
social role on the interpretation of facial expressions. Sex Roles, 51 (3/4), 187-196.
Rafaeli, A., & Sutton, R. (1987). Expression of emotion as part of the work role.
Academy of Management Review, 12, 23-37. Ridgeway, C. L. (1981). Nonconformity, competence, and influence within
groups: A test of two theories. American Sociological Review, 46, 333- 347.
Ridgeway, C. L. (1982). Status in groups: The importance of motivation.
American Sociological Review, 47, 76-88.
189
Ridgeway, C. L. (1987). Nonverbal behavior, dominance, and the basis of status in task groups. American Sociological Review, 52, 683-694.
Ridgeway, C. L. (2001). Gender, status and leadership. Journal of Social Issues,
57 (4), 637-655. Ridgeway, C. L., Backor, K., Li, Y. E., Tinkler, J., & Erickson, K. G. (2009). How
easily does a social difference become a status distinction? Gender matters. American Sociological Review, 74 (1), 44-62.
Ridgeway, C. L., Berger, J., & Smith, L. (1985). Nonverbal cues and status: An
expectation states approach. American Journal of Sociology, 90, 955-978. Ridgeway, C. L., & Diekema, D. (1992). Are gender differences status
differences? In C. L. Ridgeway (Ed.), Gender, Interaction, and Inequality (pp. 157-180). New York: Springer-Verlag.
Ridgeway, C. L., & Johnson, C. (1990). What is the relationship between
socioemotional behavior and status in task groups? American Journal of Sociology, 95, 1189-1212.
Rudman, L.A., & Fairchild, K. (2004). Reactions to counterstereotypic behavior:
The role of backlash in cultural stereotype maintenance. Journal of Personality and Social Psychology, 87, 157-176.
Rudman, L.A., & Glick, P. (1999). Feminized management and backlash toward
agentic women: The hidden costs to women of a kindler, gentler image of middle managers. Journal of Personality and Social Psychology, 77, 1004-1010.
Rudman, L. A., & Kilianski, S. E. (2000). Implicit and explicit attitudes toward
female authority. Personality and Social Psychology Bulletin, 26 (11), 1315-1328.
Schachter, S., & Singer, J. (1962). Cognitive, social and psychological
determinants of emotional state. Psychological Review, 69, 378-399. Schein, V. E. (2001). A global look at psychological barriers to women's progress
in management. Journal of Social Issues, 57 (4), 675-688. Semin, G.R., & Manstead, A.S.R. (1981). The beholder beheld: A study of social
emotionality. European Journal of Social Psychology, 11, 253-265.
190
Sheldon, O. J., Thomas-Hunt, M. C., & Proell, C. A. (2006). When timeliness
matters: The effect of status on reactions to perceived time delay within distributed collaboration. Journal of Applied Psychology, 91 (6), 1385-1395.
Sloan, M.M. (2004). The effects of occupational characteristics on the experience
and expression of anger in the workplace. Work and Occupations, 31 (1), 38-72.
Stearns, C.Z., & Stearns, P.N. (1986). Anger: The struggle for emotional control
in America’s history. Chicago: University of Chicago Press. Stewart, P.A., & Moore, J.C. Jr. Wage disparities and performance expectations.
Social Psychology Quarterly, 55 (1), 78-85. Stout, K. (n.d.). Job interview answer: When was the last time you were angry?
What happened? Retrieved April 28, 2009, from About.com: http://jobsearch.about.com/od/interviewquestionsanswers/qt/angry.htm
Thoits, P. A. (1985). Self-labelling processes in mental illness: The role of
emotional deviance. American Journal of Sociology, 91, 221-249. Thoits, P.A. (1989). The sociology of emotions. Annual Review of Sociology, 15,
317-342. Tiedens, L. Z. (2001). Anger and advancement versus sadness and subjugation:
The effect of negative emotion expressions on social status conferral. Journal of Personality and Social Psychology, 80 (1), 86-94.
Tiedens, L. Z., Ellsworth, P. C., & Mesquita, B. (2000). Sentimental stereotypes:
Emotional expectations for high- and low-status group members. Personality and Social Psychology Bulletin, 26, 560-575.
Twenge, J. M. (1997). Attitudes toward women, 1970-1995: A meta-analysis.
Psychology of Women Quarterly, 21, 35-51. White, M. J., Kruczek, T. A., Brown, M. T., & White, G. B. (1989). Occupational
Sex Stereotypes among College Students. Journal of Vocational Behavior, 34, 289-298.
Wiley, M., & Eskilson, A. (1985). Speech style, gender stereotypes, and
corporate success: What if women talk more like men? Sex Roles, 12, 993-1007.
191
Yoder, J. D. (2001). Making Leadership Work More Effectively for Women. Journal of Social Issues, 57 (4), 815-828.