When Are Apologies Effective? - Northeastern University1871/fulltext.pdf · An apology, as defined...
Transcript of When Are Apologies Effective? - Northeastern University1871/fulltext.pdf · An apology, as defined...
1
When Are Apologies Effective?
An Investigation of the Components that Increase an Apology’s Efficacy
A dissertation presented
by
Krista M. Hill
To
The Department of Psychology
In partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in the field of
Psychology
Northeastern University
Boston, Massachusetts
May 10, 2013
2
WHEN ARE APOLOGIES EFFECTIVE?
AN INVESTIGATION OF THE COMPONENTS THAT INCREASE AN APOLOGY’S
EFFICACY
by
Krista M. Hill
ABSTRACT OF DISSERTATION
Submitted in partial fulfillment of the requirements
For the degree of Doctor of Philosophy in Psychology
In the College of Science of
Northeastern University
May 10, 2013
3
ABSTRACT
Although apologies are a staple of civil society, it is unclear whether they are effective
and if effective, what components are involved in the perfect apology. The term “components”
refers to general categories of actions (both verbal and nonverbal) that may be present in an
apology. Two studies were conducted to examine (1) whether apologies are effective in eliciting
positive outcomes for an apologizer and (2) potential apology components that may obtain
positive outcomes for an apologizer.
For Study 1, six meta-analyses of previously published studies, examined the relation
between apologies and offended parties’ (1) forgiveness, (2) attributions of positive qualities to
the apologizer, (3) positive emotions toward the apologizer, (4) positive legal outcomes for the
apologizer, (5) intentions to purchase goods from the apologizer, and (6) overall positive
reactions and behaviors toward the apologizer (i.e., combining across all outcomes). High-
inference coding was used to determine which theory-driven components contribute most to the
effectiveness of apologies. Analyses revealed a significant influence of apologizing on
forgiveness (k = 79, r = .32, random effects Z = 8.16 p < .001), positive attributions of the
apologizer (k = 60, r = .24, random effects Z = 6.69, p < .001), positive emotions toward the
apologizer (k = 43, r = .33, random effects Z = 9.41, p < .001), legal sentencing (k = 11, r = .13,
random effects Z = 3.49, p < .001), and purchase intentions (k = 10, r = .23, random effects Z =
2.85, p < .01). Combining across all outcomes apologizing was effective (k = 144, r = .27,
random effects Z = 10.72, p < .001). All distributions of effect sizes were significantly
heterogeneous. Significant moderators included the apology components of remorse, offers of
compensation, and an acknowledgment of violated rules and norms.
4
The aim of Study 2 was to examine the relationship between apology components and
judge-rated outcomes. Participants apologized for a transgression they committed on video.
Trained coders then rated the apologies for apology components. Finally, videos were watched
by new participants (i.e., judges) who rated the apologies on various outcomes. This was the first
time this paradigm was used to study apologies. The apology components included both
apologizer-rated emotions and coder-rated verbal (i.e., remorse, acknowledgment of violated
rules and norms, and compensation) and expressive behavior (i.e., guilt and shame). The judge-
rated outcomes included empathy, sympathy, dispositional attributions, forgiveness, trust, and
sincerity. Analyses revealed that coder-rated remorse, guilt, and shame were significant
predictors of judge-rated empathy, sympathy, forgiveness, trust, and sincerity. Similarly,
apologizer-rated self-conscious emotions also predicted these outcomes. These relationships
remained significant even when controlling for judge-rated severity of the transgression.
5
Acknowledgments
First and foremost I would like to thank my family for their endless encouragement and
support. Thank you for being my biggest fans throughout this process and throughout my life.
Words cannot express how much your support and love have meant to me. From a very young
age you instilled in me the importance of hard work, which made this degree possible.
I would also like to extend my deepest thanks to the faculty that have helped me
tremendously during this process. To my advisor, Randy Colvin, I can never thank you enough
for taking a chance on me as an undergraduate so many years ago. Thank you for your guidance,
support, and believing in my abilities. I hope we continue our friendship and collaborate for
years to come. Thank you, Judith Hall and Nancy Kim, for your helpful comments on the many
stages of my dissertation.
To my research brother and sister, Sun Park and Stefanie Tignor, thank you for your
professional advice, encouragement, and bringing joy to this experience. I loved working with
the both of you and hope to continue to do so.
Thank you Allison Seitchik and Sarah Gunnery. Not only I am leaving graduate school
with a degree, but also with two best friends. I feel so fortunate to have gone through this process
with such wonderful and intelligent women and I promise many “networking” meetings in the
future.
To Martha Caffrey, Amy DiBattista, Mollie Ruben, Jolie Baumann, Leah Dickens,
Danielle Blanch Hartigan, and Susan Andrzejewski, you are the reasons why I enjoyed coming
to the office everyday. Thank you for the coffee breaks and laughs.
6
Thank you to my dear friends, specifically Jessie Canor and Corinne McHugh, for always
being there to listen, especially during these last few years. You never stopped encouraging me
to work hard and always brought a smile to my face with your daily emails.
I would also like to thank the following undergraduate assistants who contributed to the
completion of this project: Katie McEnaney, Lisa Bartucca, Gina Strumolo, Nicole Colley,
Colleen Trinh, and Kelly Kolkmeyer.
And finally I dedicate this dissertation to my wonderful husband-to-be, Matt. I can never
thank you enough for your love and encouragement. Thank you for listening when I needed
someone to listen, for opening your arms when I needed support, and for making me laugh when
I needed to smile. These past five years have brought so many wonderful things for us and I
look forward to our post-PhD life together.
7
TABLE OF CONTENTS
Abstract 2
Acknowledgments 5
Table of Contents 7
List of Tables 8
List of Figures 9
Chapter 1: Introduction 11
I. What Is An Apology and Why Do People Apologize? 13
II. Are Apologies Effective in Producing Positive Outcomes? 15
III. Which Components Make Apologies More Effective? 17
IV. Why Do Apologies and The Components Work? 20
Chapter 2: Study 1- Meta-Analysis 23
I. Methods 24
II. Results 28
III. Discussion 35
Chapter 3: Study 2 40
I. Methods 40
II. Results 46
III. Discussion 54
Chapter 4: General Discussion 58
I. Limitations and Future directions 60
II. Conclusion 62
References 63
Footnotes 81
Tables 82
Figures 107
Appendix A 109
Appendix B 111
8
List of Tables
Table 1: Characteristics of Studies Included in the Meta-Analysis 82
Table 2: Results for Meta-Analyses Examining Efficacy of Apologies 92
Table 3: Correlations Among Apology Components and Effect Size 93
Table 4: Interaction Analysis for Remorse, Acknowledgment of Violated Rules 94
and Norms, and Compensation for High Social Distance Relationships
Table 5: Inter-Rater Reliability Indices and Inter-Correlations For Apology 95
Components
Table 6: Phase 1 Apologizer Self-Rated Emotion Descriptive Statistics and 96
Factor Loadings
Table 7: Apologizer Self-Rated Emotion Factor Descriptive Statistics and 98
Inter-correlations
Table 8: Apologizer Self Rated Emotion Factors Correlated with Apology and 99
Transgression Specific Questions
Table 9: Correlations Among Judge-Rated Items 100
Table 10: Coder-Rated Items Correlated with Apologizer and Judge-rated Items 101
Table 11: Apologizer-rated Items Correlated with Judge-rated Items 103
Table 12: Apologizer and Coder-Rated Components Correlated with 104
Judge-Rated Outcomes Controlling for Judge-Rated Severity
Table 13: Apologizer-rated Emotions Predicting Judge Reactions 105
Table 14: Acknowledgment, Remorse, and Compensation Predicting Judge 106
Reactions
9
Table 15: Interaction Analysis for Coder-Rated Remorse, Coder-Rated 107
Guilt-Shame Aggregate, and Apologizer-Rated Self-conscious
emotions
Table 16: Coder Rated Remorse, Coder Rated Guilt-Shame Aggregate, and 108
Apologizer Self Conscious Emotions Predicting Judge Reactions
10
List of Figures
Figure 1: Apology-Outcome Process Model 109
Figure 2: Mediation Analysis Diagram 110
11
Chapter 1: Introduction
Apologies are a core feature of human relationships. Politicians, CEOs, public figures,
criminals, romantic partners, and friends all use apologies to repair relationships and protect their
reputations (Lazare, 2005). Although apologies are a staple of civil society, it is unclear when
they are effective and which components are involved in the perfect apology. Some studies
suggest that apologies are beneficial (e.g., Darby & Schlenker, 1982; Eaton, Struthers, &
Santelli, 2006; Exline & Baumeister, 2000) and lead to positive outcomes such as forgiveness
(e.g., Fehr, Gelfand, & Nag, 2010), positive feelings toward the transgressor (e.g., DeCremer,
van Dijk, & Pillutla, 2010), and reduced sentencing in legal cases (e.g., Robbennolt, 2003). Other
studies suggest that apologies can have detrimental effects, especially if not done well (e.g.,
DeCremer et al., 2010; Eaton, Struthers, Shomrony, & Santelli, 2007; Zechmeister, Garcia,
Romero, Vas, 2004).
Researchers who find the latter suggest that an apology’s effectiveness is highly
dependent on whether certain components of the apology are present or absent. The term
“components” refers to three general categories of actions (both verbal and nonverbal) that may
be present in an apology. These three categories are (1) an expression of empathy for the
offended party/an expression of remorse, (2) an acknowledgment of violated rules and norms,
and (3) an offer of compensation. Researchers have found that when one or more of these
components is (are) manipulated, an apology’s effectiveness is likely to change (e.g., Scher &
Darley, 1997). This begs the question: which components produce the most positive outcomes?1
To date, there is no agreed upon empirical answer. For my dissertation two studies were
conducted to answer this question. In Study 1 six meta-analyses were conducted that evaluated
the empirical literature on the effectiveness of apologies and the components that moderate the
12
relationship between apology and offended party outcomes. One meta-analysis was completed
for each of five outcomes and the sixth meta-analysis examined all outcomes together. These
outcomes include forgiveness, positive perceptions of the transgressor, positive emotions toward
the transgressor, positive legal outcomes for the transgressor, and offended parties’ purchase
intentions (i.e., willingness to purchase from a transgressor in the future). In Study 2 judges
observed and evaluated real life apologies (i.e., apologies for transgressions given by those who
transgressed) on several dimensions to examine the components that make an apology more and
less effective.
Figure 1 displays the model used to guide the presentation of the literature as well as the
proposed studies. The model outlines the apology process starting at the transgression. The
offended party and others who experienced the transgression react to it. They might become
upset, angry, or terminate their relationship with the transgressor. Next, the transgressor either
admits to him or herself that a transgression was committed and acknowledges its consequences
on others, or the transgressor might be unaware of the transgression because its consequences
were not observed or defensive processes kept it from the transgressor’s conscious awareness. If
the transgressor acknowledges the effect of his or her transgression on others, the transgressor
will have a cognitive-emotional response (Figure 1, a). The transgressor might experience
feelings of empathy and guilt, or be motivated to manage his or her public reputation. In both
cases, the transgressor will provide an apology. One or more components (Figure 1, b) of the
apology may elicit emotional or cognitive responses from the offended party (Figure 1, c). These
internal responses may be followed by behavioral responses (Figure 1, d) from the offended
party. In this dissertation, I specifically focus on the processes occurring during and after the
transgressor’s apology (i.e., Figure 1, a).
13
What is an Apology and Why do People Apologize?
An apology, as defined by Lazare (2005) is “an encounter between two parties in which
one party, the offender, acknowledges responsibility for an offense or grievance and expresses
regret or remorse to a second party, the aggrieved” (p. 23). Apologies may be interpersonal,
intergroup, or a mix of the two. In the latter case, an organization might apologize to an
individual (e.g., a CEO apologizes to an individual) or a person might apologize to a group (e.g.,
person apologizes to church congregation).
Researchers have identified several motives for why people apologize. These motives fall
into two categories, internal regulation and social reputation maintenance (e.g., Lazare, 2005;
McCullough, Worthington, & Rachal, 1997; Zechmeister et al., 2004). Internal regulation
motives are characterized by internal distress such as empathic concern for another, or emotions
such as guilt and embarrassment (Baumeister, Stillwell, & Heatherton, 1994; Hareli &
Eisikovits, 2006; Tangney & Dearing, 2003). For internal regulation motives, transgressors seek
to restore their dignity and self-esteem, and restore the image of the harmed. For example, after
cheating on a spouse, a person may feel guilt for what he or she has done and feel empathy for
the spouse who was hurt. Transgressors who are motivated to maintain their social reputation
attempt to influence how others perceive and behave toward them. For example, an athlete who
is caught using performance-enhancing drugs may apologize to restore his status as a beloved
celebrity as opposed to relieving internal distress. Below both motives are discussed in greater
detail.
As seen in Figure 1 both internal regulation and social reputation maintenance motives
are (1) the transgressor’s emotional and cognitive reactions to how they perceive the reactions of
14
others and (2) reasons for why people apologize. Below I discuss both the internal and external
motives in further detail.
Internal regulation. Research suggests that when people become aware their behavior
has offended someone, they will experience various emotional states such as guilt and empathy
for the offended party (Baumeister et al., 1994; Tangney, Stuewig, & Mashek, 2007). Guilt is
defined as regret about behavior that violated an ethical or social code. It is considered a self-
conscious emotion because it occurs in response to self-reflection (Tangney & Dearing, 2002).
Subjectively, guilt is an unpleasant emotional state that has two affective components--empathic
concern for the suffering of the offended party (e.g., Hoffman, 1982) and anxiety about the
transgression and its implications for the relationship between the transgressor, the offended
party, and observers (see Baumeister & Tice, 1990).
Empathy is, in part, a “shared emotional response between an observer and a stimulus
person” (Feshbach, 1975). Definitions of empathy typically include cognitive and affective
components. The cognitive component emphasizes understanding another person’s affective or
cognitive status (e.g., Borke, 1971; Hogan, 1969). For example, a man experiencing cognitive
empathy toward his romantic partner may understand that she is feeling sad. The affective
component focuses on the matching of another’s affective state (e.g., you feel sad so I feel sad).
It may also include feeling an emotion that is similar, but different from the emotion the other
person is feeling (e.g., I feel upset when you feel depressed).
Guilt and empathy may prompt apologizing for two reasons. First, guilt prompts the
transgressor to focus on the transgression and its implications, including how the transgression
affected the offended party. Focusing on the offended party leads to awareness of the internal
state of the offended party (i.e., empathy) and the damaged status of the relationship between the
15
transgressor and the offended party (Eisenberg, 1986; Hoffman, 1982; Zahn-Waxler &
Robinson, 1995). Once the transgressor understands the offended party’s feelings about the
transgression and the relationship, the transgressor may be motivated to help the offended party
feel better and mend the relationship. Second, guilt and empathy may motivate apologies
because the transgressor may wish to relieve the personal distress that accompanies these states
(Eisenberg & Miller, 1987; Lazare, 2005).
Social reputation maintenance. Transgressors often offer apologies with the goal of
repairing their post-transgression reputations (Hareli, Shomrat, & Biger, 2005). In these self-
serving situations, transgressors apologize in an attempt to return to their pre-transgression status
without regard for how the offended party is feeling. For example, an athlete who is caught using
performance-enhancing drugs may apologize to restore his celebrity status as opposed to
relieving the shame felt by being a poor role model for young fans. In this case, apologizing acts
as a tool for impression management (Schlenker, 1985; Tedeschi & Riess, 1981).
Impression management refers to “the process by which individuals attempt to control the
impressions others form of them” (p. 34, Leary & Kowalski, 1990). People’s reputations are
important because they affect how others will behave toward them. A good reputation is
associated with social approval, successful interpersonal relationships, social support from
others, and power. People whose reputations diminish may lose these rewards. Therefore, people
often feel compelled to fix their damaged reputations by apologizing for their bad deeds (Leary
& Kowalski, 1990, Schlenker, 1980).
Are Apologies Effective in Producing Positive Outcomes?
Although peoples’ motivations for apologizing range from well-intentioned to self-
centered, the majority of research indicates that apologies are associated with positive outcomes,
16
such as forgiveness. In Fehr, Gelfand, and Nag’s (2010) meta-analysis on the situational and
dispositional correlates of forgiveness, they synthesized the effects of 23 studies on the
relationship between apology and forgiveness and found an overall positive effect size of r = .42.
Similarly, apologies have been found to be associated with the offended party’s positive
emotions for, and positive perceptions of, the transgressor (DeCremer, van Dijk, & Pillutla,
2010; Xie & Peng, 2009). For example, apologies are associated with ratings of transgressor (i.e.,
apologizer) trustworthiness (DeCremer et al., 2010), liking (Bono, 2005), and satisfaction by the
offended party (Howley, 2009). Further, apologies often alter an offended party’s behavior,
leading to less retaliation (e.g., Mullet, Riviere, & Sastre, 2007), increases in purchasing
behavior after a corporate apology (Liao, 2007), and positive intentions such as investing in a
company, recommending a company, and requesting more information about a company after
the company apologized (Lyon & Cameron, 2004).
However, the correlation between the presence of an apology and positive outcomes (e.g.,
forgiveness) is not perfect. Apologies vary greatly and may contain components (i.e., categories
of verbal and nonverbal actions) that influence their effectiveness (e.g., Fehr & Gelfand, 2009).
This raises the question: How do components influence the effectiveness of an apology? The
answer is not yet clear. This may be due to the fact that in most apology studies participants are
simply assigned to ‘‘apology present” versus ‘‘no apology present” conditions (e.g., Frantz &
Bennigson, 2005; McCullough, Rachal, Sandage, Worthington, Brown, & Hight, 1998; Ohbuchi,
Kameda, & Agarie, 1989; Struthers, Eaton, Santelli, Uchiyama, & Shirvani, 2008; Tomlinson,
Dineen, & Lewicki, 2004) and the implications of specific components are not systematically
evaluated (for exceptions see Darby & Schlenker, 1982; Fehr & Gelfand, 2009; Santelli,
Struthers, & Eaton, 2009; Schmitt, Gollwitzer, Forster, & Montada, 2004). This is not to say that
17
components are not present in the apology stimuli provided to participants. In fact, in most
apology studies these components are present, but are not empirically examined. For example, in
a study by Giacalone and Payne (1995), participants read the apology, "I am very sorry and
embarrassed by what I did. I understand that revealing such personal information is wrong, and
that I made a mistake. I can only promise that I will not do it again. I am prepared to make
restitution to the company and to the employees affected as a result of my actions." Within this
apology several components are present, such as acknowledgment of a violated rule and
compensation. Because these components are present in many studies, they can be analyzed in a
meta-analysis. Meta-analysts can code each apology for the components and then examine how
these components moderate the impact of the apology on positive outcomes across many studies.
I use this method in the current meta-analyses to provide insight into the composition of an
effective apology.
Which Components Make Apologies More Effective?
Across such fields as law, sociology, and psychiatry, scholars have proposed components
that may increase an apology's effectiveness. This discussion is mostly theoretical with the
exception of some recent empirical studies (Boyd, 2011; Cunningham, 2004; Darby &
Schlenker, 1982; Fehr & Gelfand, 2009; Goffman, 1967; Govier & Verwoerd, 2002; Lazare,
2004; Scher & Darley, 1997; Schlenker, 1980; Schmitt et al., 2004; Tavuchis, 1991; Tedeschi &
Nesler, 1993; Tedeschi & Riess, 1981). Throughout the literature there has been a focus on three
apology components: (1) expressions of empathy/remorse, (2) acknowledgments of violated
rules/norms, and (3) offers of compensation. These components can be seen in Figure 1, b. As
the figure shows, an apology only occurs if the transgressor feels guilt, empathy, or is motivated
to repair his or her reputation (Figure 1, a). Depending on the motivation behind the apology, the
18
apology may contain different components. For example, a transgressor motivated purely by
empathy may say and do different things than a transgressor who is purely motivated to fix his or
her reputation.
Expressions of empathy/remorse. An apology that includes an expression of empathy
occurs when a transgressor recognizes how an offended party is feeling and communicates this to
the offended party or when the transgressor simultaneously feels what the offended party is
feeling and expresses this outwardly. Transgressors may demonstrate empathy in several ways.
From the cognitive perspective, transgressors may indicate verbally that they understand the
offended party’s feelings. For example, “I know that you are upset and are feeling embarrassed.”
From the affective perspective, transgressors may respond with their own affective feelings
through expressions of sadness or shame. Transgressor empathy may even be communicated
through warmth or compassion toward the offended party (Fehr et al., 2010). Transgressors
experiencing empathy often experience remorse once they become cognizant of how they made
the offended party feel. Expressing this remorse may also be helpful in eliciting forgiveness (e.g.,
Boyd, 2011; Darby & Schlenker, 1982; Schmitt et al., 2004).
Apologizers who express empathy and remorse tend to experience positive outcomes. For
example, in Schmitt et al.’s (2004) study participants read a vignette in which they took the
perspective of an individual whose friend borrowed and lost a bike that belonged to him or her.
In one condition the friend provided an apology that included an expression of remorse/empathy.
In this condition the friend said, “I feel really sorry for what I have done. I know how you feel
now” (p. 469). Relative to the no empathy condition, the empathy vignette was associated with
interpersonal trust, reduced anger, positive character attributions of the transgressor, and positive
emotions toward the transgressor.
19
Acknowledgments of violated rules/norms. The literature indicates transgressors’
apologies are often successful if they include an admission of wrongdoing (e.g., Scher & Darley,
1997). This component involves stating what the transgression was and expressing that the
transgression involved unacceptable behavior. For example a transgressor may say “I failed to
show up on time and have disappointed those who rely on me. My behavior is unacceptable.”
This component may also involve an internal attribution (Boyd, 2011). In this case transgressors
blame themselves instead of an external source. In his public apology for actions resulting in the
loss of billions of dollars, Gerald Levin, CEO of Time Warner during the Time Warner-AOL
merger, made an internal attribution stating that “I presided over the worst deal of the century,
apparently. And I guess it's time for those who were involved in companies to stand up and say,
you know what, I'm solely responsible for it. I was the CEO, I was in charge. I'm really very
sorry about the pain and suffering and loss that was caused” (Levin, 2010). By taking
responsibility for their actions, transgressors show their awareness of violating social norms and
their desire to behave differently in the future (Scher & Darley, 1997). Apologizers who make
internal attributions may also include a promise for change. This may be a statement such as, “I
will seek therapy so I never behave this way again” or “I will avoid situations in which I am
tempted to transgress.” This strategy is effective because (1) offended parties’ feelings are
validated and they are shown that what they believed to be wrong is viewed similarly by others
and (2) offended parties believe that similar wrongdoing by the transgressor will not occur in the
future (Lazare, 2005).
Offers of compensation. Apologies are often effective when transgressors offer to
compensate victims for their pain and suffering. For example, a CEO may offer monetary
compensation to consumers for a problem caused by his or her company. Compensation does not
20
always need to be an offer of a tangible good. Restoring the offended party’s reputation or
showing respect to the offended party in front of others can also be considered compensatory
(Boyd, 2011).
Several studies have shown the importance of compensation in an apology (Conlon &
Murray, 1996; Scher & Darley, 1997; Schmitt et al., 2004). For example, Conlon and Murray
(1996) asked business students if they had been compensated by a company after complaining
about one of their products. Results indicated that compensation was associated with satisfaction
and the likelihood of doing business with the company in the future.
Nonverbal behaviors. Despite little empirical evidence, researchers have suggested that
nonverbal behaviors may be associated with the effectiveness of an apology because they serve
as cues to the internal states of transgressors (Darby & Schlenker, 1982; Scher & Darley, 1997).
Recipients of apologies can use cues to judge the transgressor’s sincerity. For example, a
transgressor may state that he is feeling remorseful. However, the apology may not be effective
unless “remorseful” facial expressions accompany the apology.
Why Do Apologies and the Components Work?
The literature that attempts to answer the question of why apologies work has largely
focused on the apology-forgiveness relationship. In this literature offended parties are thought to
be motivated to protect themselves and thus seek revenge and estrangement from the
transgressor. Therefore, any reparative actions taken by the transgressor must a) decrease
retaliation motivation; b) decrease the offended party’s desire for estrangement; and c) motivate
conciliation and goodwill toward the transgressor (McCullough, Worthington, & Rachal, 1997).
Keeping this in mind, researchers have attempted to identify why apologies are associated with
forgiveness. The proposed explanations are the offended party’s emotional and cognitive
21
reactions to the transgressor’s apology and can be seen in Figure 1, c. As seen in the figure, each
component in Figure 1, b is associated with a different offended party reaction in Figure 1, c. For
example the component of compensation is associated with the offended party believing that
equity has been restored to the relationship. The figure also demonstrates that the offended
party’s emotional and cognitive reactions influence the offended party’s behavioral reactions
(see Figure 1, d). These reactions are the outcomes of apology such as forgiveness, legal
outcomes, and purchase intentions. Below I discuss the proposed explanations shown in Figure
1, c.
Apologies induce empathy. Several researchers have proposed that the offended party’s
empathy for the transgressor affects the apology-forgiveness relationship (e.g., McCullough et
al., 1997). As mentioned above, empathy has the ability to motivate individuals to help others.
This was discussed as a reason for why people may apologize, but it might also explain why
victims forgive transgressors. In this case, the offended party feels empathy for the transgressor.
The expression of an apology, especially if the transgressor is expressing guilt, sadness,
loneliness, or other negative emotions, may induce empathy in the offended party (Baumeister et
al., 1994). In other words, transgressor empathy may beget empathy in the offended party. If the
offended party experiences empathy for the transgressor, the offended party may exhibit
prosocial behavior toward him or her, including forgiveness, relationship repair, and a reduced
motivation for retaliation (Batson, 1990; Batson, 1991; Baumeister et. al, 1994; Eisenberg &
Miller, 1987; McCullough et al., 2007).
Apologies affect attributions. Apologies can enhance victims’ perceptions of the
transgressor. Apologizing may convey the message that the transgressor is not a bad person and
that the misdeed was a chance event. As a result, an apology often overrides people’s tendency to
22
make trait attributions (Weiner, Graham, Peter, & Zmuidinas, 1991). Furthermore, if the
offended party believes the transgressor is fundamentally good and is unlikely to repeat the
transgression, the offended party will forgive the transgressor. Thus, an important component of
successful apologies may be the transgressor’s acknowledgment of violated rules or norms.
Apologies restore equity. Equity theory states that transgressions create relationship
inequality because negative outcomes are more frequent or intense for the offended party than
the transgressor (Walster, Berscheid, & Walster, 1973). In order for forgiveness to occur,
equality must be restored. An apology can restore equality through compensation strategies. For
example, if a transgressor spoke poorly about an offended party to others, the transgressor may
compensate the offended party by promising to talk positively about the offended party in the
future or acting respectfully toward the offended party in front of others (Ohbuchi et al., 1989).
Apologies may also help to restore equality because when transgressors apologize they
tend to put themselves in a subservient role, coming to the offended party “on bended knee”, and
asking for forgiveness. In this case, transgressors offer power to the offended party as a social
offering and place the offended party in the dominant role (Exline, Deshea, & Holeman, 2007).
If offended parties are satisfied with this power and the return to equality then they may become
more forgiving (Fagenson & Cooper, 1987).
23
Chapter 2: Study 1
The goal of Study 1 was to examine the relationship between apologies and offended
party outcomes, some of which were interpersonal, while others were behavioral (e.g.,
purchasing intentions). Six meta-analyses of previously published studies were completed on the
relationship between apology and (a) forgiveness, (b) positive perceptions of the transgressor, (c)
positive emotions toward the transgressor, (d) positive legal outcomes for the transgressor, (e)
purchase intentions, and (f) overall positive outcomes (i.e., combining across all outcomes).
These categories were chosen after reviewing the literature and making logical categories out of
the various dependent variables used. The inclusion criteria for forgiveness included dependent
variables that were associated with decreased retaliation motivation, decreased desire for
estrangement by the offended party, and motivation for conciliation and goodwill toward the
transgressor. Positive perceptions of the transgressor included any positive attributions made by
the judges about the transgressor after the apology. Positive emotions toward the transgressor
included any positive affective feelings toward the transgressor such as liking, less anger, and
satisfaction. Positive legal outcomes included legal settlements as well as more lenient
punishment decisions in a legal setting. Finally, purchase intentions included any dependent
variable assessing an offended parties’ willingness to purchase from a transgressor. For this
dependent variable, the transgressors were exclusively companies. See Appendix A for a list of
all outcomes.
There were two goals for the meta-analyses: (1) to collect and summarize the existing
published literature to draw conclusions and identify research gaps; and (2) to calculate mean
effect sizes for the relationship between apologies and forgiveness, positive perceptions of the
transgressor, positive emotions toward the transgressor, positive legal outcomes for the
24
transgressor, and purchase intentions. Even when apologies produce positive outcomes for the
apologizer, questions remain as to when they are most effective (i.e., what are the components
involved in an effective apology). Therefore, I coded potential moderators of apology
effectiveness in each meta-analysis.
Methods
Search Method
The following procedures were used to locate studies: (1) PsycINFO search from earliest
possible year through September 2011 using a list of terms that included apology, repentance,
confession, regret, penance, service-recovery, crisis management, and remorse, (2) PsycINFO
search of names of key authors known to conduct apology or forgiveness research, (3) search of
bibliographies of relevant sources.
Inclusion Criteria
The criteria for study inclusion included the following: (1) Participants were from a
typical population of adults (i.e., not clinically diagnosed), 2) Participants were at least 18 years
old, (3) The apology in the study must have had a verbal acknowledgment of a wrong doing (i.e.,
emotion displays alone were not counted as apologies), (4) The study was published in an
English-language article or book, (5) The apology was manipulated within a vignette, recalled
from a previous experience, or occurred during a laboratory interaction, and (6) The judges of
the apology (i.e., the participants making the ratings of the apology outcomes) could be the
receivers or onlookers of the apology. Transgressions often affect more than the direct victim,
therefore it was important to understand the reactions of those indirectly affected. The final two
criteria were: (7) Both dyadic and intergroup apologies were included, (8) The apology must
have occurred after the transgression. Offended parties may react differently to ex ante apologies
25
because they have yet to experience the cognitive and emotional outcomes of the transgression,
therefore it was important to keep the different apologies separate.
Variables Coded from Each Study
The following information was recorded from each study: (1) date of publication, (2)
sample size, (3) country the study was conducted in, (4) stimulus materials (vignette; recall; or
lab interaction), and (5) social distance (i.e., interpersonal connectedness of the relationship
dichotomously coded as “low” or “high”) between apologizer and offended party. These
variables were coded by two independent raters.
For stimulus materials, a vignette study was defined as a study in which participants
reported on various outcomes (e.g., forgiveness, emotions) after reading a hypothetical
transgression in which an apology occurred or did not occur. A recall study was defined as a
study in which offended parties reported on various outcomes, such as forgiveness, in response
to an actual transgression that occurred outside of the lab and indicated whether they received an
apology. A laboratory interaction methodology was defined as a study in which participants
experienced a transgression in the laboratory and were subsequently given or not given an
apology. Participants then reported their reactions to the situation.
Because forgiveness is more likely to occur in close relationships (McCullough, Kurzban,
& Tabak, 2013) it was important to code for social distance. Social distance was dichotomously
coded as “low” or “high.” A relationship was “low” on social distance when the apologizer and
offended party were members of the same ingroup (e.g., romantic partners, friends, coworkers,
and participants within the same study). A relationship was “high” on social distance when their
ingroups differed (e.g., boss-employee, company-customer, doctor-patient). On the social
distance continuum, low social distance implies greater interpersonal connectedness and
26
relatively egalitarian relationships. In contrast, high social distance implies little or no
interpersonal connectedness or power imbalance between members.
Variables Constructed From Scripts of Apologies
The apology components were the primary moderators of interest. Three researcher
assistants were trained as coders and directed to read each apology (if provided by authors) and
coded for the presence of remorse (the observable signal of empathy), acknowledgments of
violated rules/norms, and compensation, all on a 1(not at all present)-5 (very present) scale.
Coders were informed an expression of remorse involves a verbal expression of negative affect.
For example, “I feel terrible.” An acknowledgment of violated rules and norms was defined as
recognizing the transgression committed violated rules and norms that interpersonal behavior is
bound by. Finally, to identify compensation coders were asked, “does the apologizer offer to fix
what they did (i.e., do they focus on restoration of equity through exchange)?” For example, in
offering compensation an apologizer may say, “Let me make it up to you” or “I want to bring
you out to make up for this.”
The three trained coders also rated the severity of each transgression on a 1(not at all
severe)-9(very severe) scale. Inter-rater reliability was then calculated. The Cronbach’s alphas
for the moderators were .85, .78, .96, and .94 respectively. With high reliability among ratings I
then aggregated the scores by averaging across the three coders resulting in a score on each
moderator for each apology.
Statistical Methodology
The meta-analyses focused on effect sizes from between-subjects designs, where
participants who received an apology were compared to control groups who did not receive an
apology. The Pearson correlation, r, was used as the effect size indicator, which represents the
27
strength of association between the presence of an apology and a positive outcome. A positive r
indicated an apology was associated with a positive outcome in the original study. When r was
not present, it was calculated using the statistics provided. For example, in many cases the
original authors reported means and standard deviations in both apology and control samples. A
basic two-sample t-test was calculated using these means and standard deviations and the t-
statistic was then converted to r using standard formulas. When a partial r or standardized
regression coefficient was present and the Pearson r was not, the former was used. When a result
was reported as nonsignificant with no data, r = .00 was assigned.
Several meta-analyses were conducted, one for each dependent variable. These dependent
variables included (1) forgiveness, (2) positive perceptions of the transgressor, (3) positive
emotions toward the transgressor, (4) positive legal outcomes for the transgressor, and (5)
purchase intentions. Additionally, a “combined outcomes” meta-analysis was conducted across
all dependent variables (i.e., forgiveness, positive perceptions, positive emotions, positive legal
outcomes, and purchase intentions). For the combined outcomes meta-analysis there were some
cases in which effect sizes had to be averaged if multiple dependent variables were assessed in
the same sample. For example, if participants in one study reported on more than one dependent
variable such as forgiveness and legal outcomes, the effect sizes for these were combined. The
moderator analyses were conducted using the combined meta-analysis.
Both fixed and random effects models were calculated. A fixed effects model assumes
that there is one true effect size, which underlies all observed effects and that differences among
these observed effects are due to sampling error. A fixed effects analysis can be generalized to
new subjects that are run though the same stimuli and procedures. Fixed effects models are often
used in fields with little methodological variance. In a random effects model the true effect is
28
assumed to vary from study to study because of both subject-level sampling error and study-level
sampling error. In a random-effects meta-analysis it is assumed that the true effects are normally
distributed. Random effects analyses can generalize to new people in similar conditions as well
as to new, conceptually similar conditions. The fixed effects model produces smaller confidence
intervals and more significant results, while the random effects model is more conservative. For
a more in depth discussion of fixed and random effects models see Lipsey and Wilson (2001).
The heterogeneity statistic was computed to assess whether each effect size was an
estimate of the common population; it determined whether the variability among the effect sizes
was greater than the variability due to sampling error. Determining the effect sizes to be
heterogeneous allows for the testing of moderators to examine if any of them account for the
variability in the effect sizes (Rosenthal & Rubin, 1982). Finally, the “file drawer N” to estimate
the number of unretrieved study effect sizes that average to zero needed to produce a
nonsignificant effect.
The goals of the moderation analyses were to (1) assess the impact of various
components (i.e., remorse, acknowledgment of violated rules and norms, and compensation) on
the apology-outcome relationship and (2) assess the impact of study methodology on the
magnitude of the effect sizes. Analyses were facilitated by the Comprehensive Meta-Analysis
Software program (Bornstein et al., 2005).
Results
Study Characteristics/Summary of the Existing Literature
A list of the studies and effect sizes included in the meta-analyses can be seen in Table 1.
The meta-analyses were based on a sample of 34,399 participants. The average sample size was
165 participants (SD = 162). The sample sizes of individual studies ranged from 22 to 1652
29
participants. The majority of studies used samples drawn from college student populations. A
large amount of the studies were conducted in the United States and Canada while others were
conducted in Australia (k = 11), The Netherlands (k = 6), France (k = 4), Singapore (k = 3),
Lebanon (k = 2), China (k = 1), Taiwan (k = 4), Kuwait (k = 1), Germany (k = 1), and Greece (k
= 1). The majority of studies were vignette studies (58%), while only 16% were recall and 26%
were live interaction. The live interactions always used confederates who either interacted in
person with the participants or through a computer. In these studies the confederates transgressed
against the participants and then either apologized using a scripted apology or did not apologize.
I obtained apology scripts from 42% of the studies. Scripts were not used in recall studies
or live interaction studies. In recall studies, participants recalled a time in which they were the
victim of a transgression and reported whether or not the transgressor apologized. Thus, no
standard apology was provided. In several other studies, researchers stated that a transgressor
apologized to a victim but provided no additional information about the apology.
For studies that utilized scripts, 40% involved high social distance relationships (i.e., the
apologizer and offended party belonged to different in-groups; k = 48), 44% included low
social distance relationships (i.e., the apologizer and offended party are from the same in-group;
k = 53), and 16% (k = 19) could not be categorized due to insufficient information.2
Effects of Apology
Forgiveness. Combining 79 independent effect sizes that investigated the effect of
apology on forgiveness (see Table 2), the random effect size was r = .32 (Z = 8.16, p < .001)
and the fixed, weighted mean effect size (computed by weighting each effect size by the inverse
of its variance) was r = .41 (Z = 38.58, p < .001). The significant combined Z indicated the
presence of an apology has a significant effect on forgiveness. Fail-safe N analysis indicated it
30
would take 25,193 additional findings that average to r = .00 for this fixed combined Z to no
longer be significant. The effect sizes ranged from r = -.29 to r = .97 and were significantly
heterogeneous.
Positive perceptions of the transgressor. Combining 60 independent effect sizes that
investigated the effect of apology on positive perceptions (e.g., trustworthy, honest) of the
transgressor (see Table 2), the random effect size was r = .24 (Z = 6.69, p < .001) and the
fixed, weighted mean effect size was r = .19 (Z = 17.80, p < .001). The significant combined Z
indicated the presence of an apology has a significant effect on positive perceptions of the
transgressor. Fail-safe N analysis indicated it would take 5,647 findings that average to r = .00
for this combined Z to no longer be significant. The effect sizes ranged from r = -.51 to r = .85
and were significantly heterogeneous.
Positive emotions toward the transgressor. Combining 43 independent effect sizes that
investigated the effect of apology on positive emotions (e.g., satisfaction, liking) toward the
transgressor (see Table 2), the random effect size was r = .33 (Z = 9.41, p < .001) and the
fixed, weighted mean effect size was r = .39 (Z = 37.74, p < .001). The significant combined Z
indicated the presence of an apology has a significant effect on positive emotions toward the
transgressor. Fail-safe N analysis indicated it would take 98 findings that average to r = .00 for
this combined Z to no longer be significant. The effect sizes ranged from r = .00 to r = .71 and
were significantly heterogeneous.
Legal sentencing. Combining 11 independent effect sizes that investigated the effect of
apology on legal sentencing (see Table 2), the random effect size was r = .13 (Z = 3.49, p <
.001) and the fixed, weighted mean effect size was r = .11 (Z = 6.30, p < .001). The significant
combined Z indicated the presence of an apology is associated with less harsh legal sentencing.
31
Fail-safe N analysis indicated it would take 98 findings that average to r = .00 for this combined
Z to no longer be significant. The effect sizes ranged from r = -.05 to r = .37 and were
significantly heterogeneous.
Purchase intentions. Combining 10 independent effect sizes that investigated the effect
of apology on purchase intentions from the transgressor (see Table 2), the random effect size was
r = .23 (Z = 2.85, p < .001) and the fixed, weighted mean effect size was r = .24 (Z = 10.82,
p < .001). The significant combined Z indicated the presence of an apology has a significant
effect on purchase intentions. Fail-safe N analysis indicated it would take 222 findings that
average to r = .00 for this combined Z to no longer be significant. The effect sizes ranged from
r = -.27 to r = .57 and were significantly heterogeneous.
Combined outcomes. Combining 144 independent effect sizes from 94 articles that
assessed the effects of apology on positive outcomes (i.e., forgiveness, positive emotions,
positive perceptions, legal sentencing, and purchase intentions; see Table 2), the random effects
mean effect size was r = .27 (Z = 10.72, p < .000) and the fixed, weighted mean effect size
(i.e., the effect size weighted by sample size) was r = .22 (Z = 46.83, p < .000) 3
. The
significant combined Z indicated that the presence of an apology has a significant effect on
positive outcomes. Fail-safe N analysis indicated it would take 24,177 findings that average r =
.00 for the combined Z to no longer be significant. The effect sizes ranged from r = -.51 to r =
.97 and were significantly heterogeneous.
Key Moderators of an Apology’s Effectiveness
Stimulus materials. Seventy-seven studies employed a vignette paradigm, twenty-nine
studies were recall studies, and thirty-eight studies were live interactions (with a confederate in
person or on a computer). I ran a fixed effects contrast to examine the moderating effect of
32
methodology. Across all studies, stimulus material moderated the relationship between apology
and positive outcome (Q(2) = 164.03, p = .00; vignette use r = .28, recall methodology r = .38
, interaction studies r = .15). Similarly, a one way ANOVA (the random effects analogue to the
fixed effects moderator analysis) also revealed a significant effect of stimulus materials, F(2,141)
= 3.067, p = .05. Vignette use (r = .30) and recall methodologies (r = .35) had larger effect
sizes than live interaction studies (r = .18). A Tukey test revealed that recall and live interaction
effect sizes significantly differed, p = .06.
Apology components. To begin, I first examined the relationships between our coded
apology component variables. It is important to note that a random-effects model was used for
these analyses and all subsequent analyses. That is, the effects reported are weighted equally and
not by sample size. Furthermore, these analyses are completed across all dependent variables.
Table 3 displays these results. As can be seen, severity was significantly correlated with remorse
and compensation. Therefore apologies for more severe transgressions included more remorse
and offers of compensation. The three apology components (acknowledgment of violated rules
and norms, remorse, and compensation) were all significantly positively correlated suggesting
that these components often appear simultaneously in apologies.
Next, I correlated each of the coded components with the independent effect sizes found
in our studies. As seen in Table 3, studies with apologies that contained remorse or compensation
had larger effect sizes, while there was no association between acknowledgment of violated rules
and norms and severity with effect sizes.
To better understand these relationships I entered the components as predictors of effect
size in a multiple regression. For ease of discussion these results are presented in absolute terms
(e.g., remorse is present or not present), but it is important to note that the components were
33
coded on a continuous scale and should be interpreted as such. The model was significant, R2 =
.129, F(4, 66) = 2.44, p = .05. Remorse was a significant predictor of effect size when
controlling for compensation, acknowledgment of violated rules and norms, and severity (β =
.27, p = .04)4. Compensation (β = .25, p = .05) and acknowledgment of violated rules and
norms (β = -.24, p = .05) also significantly predicted effect size, while controlling for the
other moderators. Severity was not a significant predictor (β = -.11, p = .38).Therefore studies
containing apologies that include remorse or compensation tend to obtain higher effect sizes than
those without these components, while studies with apologies that include acknowledgments of
violated rules and norms tend to obtain smaller, or even negative effect sizes than those without
this component.
Social distance. Although the construct of social distance represents a continuum, the
methods used and details provided limit how social distance can be characterized in apology
studies. As a result, the social distance between transgressor and victim was assigned one of two
values: low or high social distance. To evaluate social distance, I conducted a multiple regression
analysis in which social distance (i.e., low social distance = 0, high social distance = 1),
remorse, acknowledgment of violated rules and norms, and compensation predicted effect size.
This model was significant, R2 = .18, F(4,59) = 4.53, p = .003. Furthermore, social distance
was significant (β = -.11, p = .08) suggesting a stronger relationship between apology and
offended party outcomes for low social distance relationships.
To examine whether the components moderated effect size differentially for the social
distance groups, the data was split into two separate files: one for high social distance studies (k
= 48) and one for low social distance studies (k = 53). Next, I entered acknowledgment of
violated rules and norms, remorse, and compensation into a multiple regression model predicting
34
effect size for high social distance studies and then for low social distance studies. For high
social distance, acknowledgment of violated rules and norms (β = -.40, p = .03), remorse (β
= .34, p = .09), and compensation (β = .41, p = .02) were all significant predictors. For low
social distance, remorse was a significant predictor (β = .48, p = .02), while acknowledgment
of violated rules and norms (β = -.26, p = .21) and compensation were not (β = -.01, p =
.97).
Interaction and additive effects. Given that remorse, acknowledgment of violated rules
and norms, and compensation were all significant moderators of the apology-outcome
relationship in high social distance relationships it was important to test for interaction and
additive effects. To do this all three variables were centered by subtracting the mean of all data
points from each individual data point. I then created four interaction variables by multiplying
(1) remorse with acknowledgment of violated rules and norms, (2) remorse with compensation,
(3) acknowledgment of violated rules and norms with compensation, and (4) remorse,
acknowledgment of violated rules and norms, and compensation. The centered variables of
interest and the interaction variables were entered as predictors of effect size in high social
distance relationships. The model was significant (R2 = .30, F(7, 23) = 3.410). See Table 4 for
results. As seen in the table, there were no significant interaction effects. Furthermore, this model
was no more predictive of effect size than the three component model in which remorse,
acknowledgment of violated rules and norms, and compensation predicted effect size (R2 = .29,
F(3, 27) = 5.081, p = .006, p = .012; Δ R2 = .15, p = .176). Thus, for the sake of parsimony
the three component model was used to examine the additive effects of the components. As
mentioned above, all three components were significant predictors of effect size when entered
35
simultaneously, thus suggesting an additive effect. In other words, the regression of effect size
on one predictor is constant over all values of the other predictors.
Discussion
Results from the six meta-analyses demonstrate that apologies were effective in
producing positive outcomes even when tested with the more conservative random effects
model, which permits wider generalization to new studies than the fixed effects model does
(Lipsey & Wilson, 2001). These positive outcomes include forgiveness of the apologizer,
positive perceptions of the apologizer, positive emotions toward the apologizer, positive legal
outcomes for the apologizer, and offended parties’ positive purchase intentions.
Studies utilizing vignette or recall methodologies were more effective than studies using
live interactions, suggesting that apologies may not be as effective in more naturalistic settings.
Confounding variables found in live-interaction studies may affect the efficacy of apologies. For
example, nonverbal behaviors may be important factors in live-interaction studies. If nonverbal
behaviors do not correspond to the verbal statements an apologizer is making (e.g., an apologizer
expresses remorse while smiling) then apologies may be less effective. This is an important
avenue for researchers to pursue in the future.
Remorse, compensation, and acknowledgment of violated rules and norms were all
significant moderators. Remorse and compensation enhanced, whereas acknowledgment of
violated rules and norms diminished, apology effectiveness. However, in transgressor-offended
party relationships characterized by low social distance a different pattern emerged. In these
relationships, only remorse predicted effect size. Compensation had no effect. The rarity of
compensation in low social distance relationship apologies may explain this. In fact, when
examining the frequency of compensation I found forty-two out of the fifty-three apologies that
36
occurred in low social-distance relationships were given a score of “1” (i.e., not at all present).
Thus, suggesting that compensation may not be used in emotion-based close relationships.
Similarly, an acknowledgment of violated rules and norms may be ineffective because it
makes the transgression salient again, causing offended parties to reflect on the source of their
emotional discomfort. After a transgression occurs, offended parties often make dispositional
attributions (as opposed to situational attributions) when evaluating the transgressor (Ross,
1977), believing that only a malevolent person would commit such a hurtful act (Weiner,
Graham, Peter, & Zmuidinas, 1991). If an apology can override the tendency to make negative
dispositional attributions it should be more effective. An acknowledgment of violated rules and
norms appears not to override this tendency and instead may reinforce the offended parties’
beliefs that the transgression was reflective of the apologizer’s character. By reminding an
offended party about the transgression and detailing why it was morally wrong, it becomes easier
for an offended party to assume the apologizer is the cause of the transgression (Kim, Ferrin,
Cooper, & Dirks, 2004).
Gaps in the Existing Literature
Besides recall studies, which were unable to report apology components, there was not
one study in which real apologies were examined. Thus, our examination of apology components
may not generalize to real life situations because real apologies may not contain these
components. For example, in interpersonal apologies it may be fairly rare for an apologizer to
offer compensation to a close other. This component may occur more often in public apologies
given by corporate executives. For example, when ice storms trapped Jetblue passengers on
planes for hours, the CEO apologized and offered passengers monetary compensation that could
help restore their finances after the cancelled flight.
37
The absence of naturally occurring apologies in the literature inherently suggests that
nonverbal behaviors and emotions that typically accompany real apologies (e.g., frowning,
smiling, crying, posture) are unexamined. However, theorists believe nonverbal behaviors are
associated with apology efficacy because these behaviors serve as cues to the internal states of
transgressors (Darby & Schlenker, 1982; Scher & Darley, 1997). Offended parties can therefore
use nonverbal cues to judge the transgressor’s sincerity. For example, an apologizer may state
that he is feeling remorseful, but his apology may be ineffective unless “remorseful” expressions
(e.g., a bowed head and crying; Scher & Darley, 1997; Weisman, 2004) accompany it. In the
future, researchers should examine the expressive behaviors that accompany an apology and how
they affect outcomes.
Only eight studies were conducted in eastern culture countries. Therefore the current
results can only be generalized to western culture apologies. The components considered
effective may differ in eastern cultures. This is especially true considering those in eastern
cultures apologize quite differently than those from western cultures (Lazare, 2004). This
difference most likely stems from the individualistic mentality of the west and the collectivist
mentality of the east (Schwartz & Bilsky, 1987). In countries that tend to be individualistic such
as the USA and Canada, there is an emphasis on personal achievement at the expense of the
group’s goals. In countries that tend to be collectivist such as China and Singapore, there is an
emphasis on family and work group goals above individual needs or desires. Thus, collectivist
culture apologies are more likely than individualistic culture apologies to focus on restoring the
relationship with the offended party, rather than on relieving an internal state, such as guilt.
Subsequently, offended parties in collectivist cultures may prefer this type of apology. Interested
researchers may wish to examine this possibility.
38
Limitations of the Present Meta-Analyses
Although meta-analysis is a powerful tool for understanding data across multiple studies,
there are inherent limitations. Much of the variance in effect sizes cannot be fully explained by
the moderator analyses. There may be additional moderators that were not coded or could not be
coded from the information provided by the authors. For example, characteristics of the
apologizer such as age, attractiveness, and gender may influence the efficacy of apologies, but
were not available for the current meta-analyses.
Publication bias is also a concern especially because this meta-analysis did not include
dissertations or unpublished manuscripts. However, the large fail-safe N indicators suggest that it
would take many studies with null results to produce a nonsignificant combined effect size for
apology effectiveness.
Practical Implications
The take-home message of the current meta-analyses is apologies work. Apologies
produce positive outcomes such as forgiveness, positive emotions and perceptions, legal
sentencing, and purchase intentions in favor of the apologizer in many contexts including
business-customer (e.g., Bolkan & Daly, 2009), doctor-patient (e.g., Bornstein et al., 2002),
colleague (Braaten et al., 1993), and romantic relationships (e.g., Gunderson & Ferrari, 2008).
However, an apology can backfire, especially if not done correctly. In low social distance
relationships, apologizers need only include an expression of their remorse. Neither
compensation nor an acknowledgment of violated rules or norms predict offended party
outcomes.
Those in high social distance relationships wishing to provide an effective apology might
need to express their remorse and include an offer of compensation. Interestingly, an apologizer
39
in a high social-distance relationship should sparsely acknowledge violated rules and norms.
Reminding the offended party of what has been done and why it was wrong is detrimental and
instead may offend the recipient of the apology. Instead the apologizer may wish to focus on
expressing their regret or compensating the offended party. Future research should explore
additional components to better understand what a “perfect” apology includes.
40
Chapter 3: Study 2
The results reported in the present chapter extend the results reported in previous
chapters. As seen in the meta-analyses, not one study involved evaluating real apologies. The
research method used to study apologies typically involves a vignette that describes a person’s
transgression. Participants are told that an apology was, or was not, given. Then, participants
report whether they would forgive the transgressor based on this apology. However, this method
does not accurately represent how apologies are perceived and how transgressions are forgiven
in real life. Real apologies involve emotion and expressive behaviors (e.g., facial expressions,
vocal tone) that are absent in vignette studies.
Furthermore, the meta-analyses were unable to answer the question of why the apology
components (i.e., remorse, acknowledgment of violated rules and norms, and compensation) are
associated with offended party outcomes such as forgiveness. There were two main goals of
Study 2: (1) to examine the relationship between apology verbal components, expressive
behavior, apologizer self-reported emotions, and offended party-rated emotional, cognitive, and
behavioral reactions using real apologies given by participants who committed transgressions;
(2) to test the mediating role of offended party-rated emotional and cognitive reactions (Figure
2,c) in the apology component (Figure 2,b) and offended party-rated behavioral reactions (Figure
2,d) relationship. From here on, participants who present apologies will be called apologizers
and participants who evaluate apologies will be referred to as judges.
Methods
Study 2 investigates real apologies given by participants who committed the
transgression for which they are apologizing. Apologizers were video recorded while
apologizing for a transgression they committed. Trained coders rated the apologies on various
41
dimensions identified in the literature as important for an effective apology (e.g., remorse,
compensation). Finally judges observed the video recorded apologies and evaluated their
sincerity, and whether the apologizer deserved forgiveness and could be trusted.
The relationships between the rated components (e.g., remorse, offers of compensation,
empathy) and judges’ perceptions of the apology (i.e., forgiveness, trust, and sincerity) were
examined.
Phase 1: Creating Apology Stimuli
Participants
50 apologizers were recruited and paid $10 for participating in a study that they were told
investigated forgiveness and life events. The apologizers were primarily senior undergraduate
students from Northeastern University. Judges ranged in age from 18 to 25. The sample was
racially diverse. Thirteen videos were removed from our stimuli set for various reasons. Three
participants requested their videos not be seen by future participants, two participants reported
extremely low levels of sincerity during the task (i.e., they did not take the task seriously), and
the remaining eight were randomly chosen to be eliminated for the judgment-phase time
purposes. Therefore 37 stimuli videos were used. Twenty of these videos had female apologizers,
while seventeen had male apologizers.
Measures
Each apologizer completed several questionnaires post-apology. Apologizers reported on
their apology experience as well as the transgression they chose to apologize for.
Directly following the apology, apologizers rated their emotional state on a 5-point Likert
scale. Twenty-two emotions were rated; although, the emotions of primary interest were “guilt,”
“shame,” “embarrassment,” “pity,” and “empathy for the offended party.” The other emotions
42
assessed were “amusement,” “anger,” “compassion, ” “contempt,” “contentment,” “disgust,”
“excitement,” “fear,” “happiness,” “hope,” “inspiration,” “jealousy,” “love,” “relaxation,”
“sadness,” “surprise,” and “worry.”
In addition to reporting their emotions after the apology, apologizers were asked to
provide more information about their experience while giving the apology as well as the
transgression that led to the apology. Participants were asked, “How sincere was the apology that
you just provided?”, “How upset was the person who was hurt by you after they found out about
your behavior?”, and “How reasonable was the person that you hurt's reaction to your
behavior?”All questions were responded to using a 7-point scale. A score of one represented low
levels of the construct being assessed (e.g., not at all sincere, not at all upset, not at all
reasonable, not at all bad/severe), while a seven represented a very high level of the construct
(e.g., very sincere, very upset, very reasonable, very bad/severe). Participants also rated on a 5-
point scale, “How bad/severe do you think the behavior that you apologized for was?” A score of
one indicated behaviors were not at all bad/severe, while a five indicated the behavior was very
bad/severe.
Procedure
Each participant was run individually in the lab. Each participant completed an informed
consent and was given a sheet of paper that read, “Please take the next few minutes to think of a
time in which you committed a transgression (i.e., you did something wrong) that emotionally
hurt another person that you are close with. This should be a transgression for which you have
not yet apologized. Once you think of a transgression please spend the next two minutes writing
about the transgression. You may wish to write about how you felt, to write about how the
43
offended party of the transgression felt, to write about how you feel about it now, or to even
write about the transgression itself.”
The researcher then left the room while the apologizer wrote about the transgression.
After two minutes the researcher returned to the room and let the apologizer know that he or she
would be apologizing for this transgression. Before the apology was given, the apologizers were
told that only the researchers would be allowed to see their video tape. Furthermore, they could
request that their video be destroyed if they felt uncomfortable with others watching it.
Apologizers were told that their video was private before they apologized in order to reduce
impression management during the apology. The apologizer was then asked to imagine that the
intended recipient of the apology was present in the room. The apologizer then offered an
apology to a laptop camera approximately 2 feet away.
The researcher left the room while the apologizers apologized. The apologizers were told
to retrieve the researcher once they had completed the apology. This method allowed apologizers
to choose the amount of time they wanted for their apology. After completing the video recorded
apologies, apologizers completed several questionnaires about their feelings while giving the
apology and about the transgression.
After completing all questionnaires apologizers were given a second consent form that
asked for permission to show their videotaped apologies to prospective participants. Apologizers
were told that they would be paid either way (i.e., whether they said yes or no) and were not
pressured in any way to respond yes. To better understand what these stimuli consisted of, see
Appendix B for complete transcripts of two apologies
Coded Behavior
44
Three trained research assistants independently coded each videotaped apology. Coders
viewed each apology a minimum of two times and read a transcript of the apology before making
any ratings. They then rated the apology and apologizer on verbal and expressive behavior
components.
Verbal components. Similar to the coding for the meta-analyses, the verbal components
of an apology that may influence its effectiveness were coded. The same three research assistant
coders rated the extent to which each apology contained an (a) expression of remorse (the verbal
cue of empathy), (b) acknowledgment of violated rules/norms, and (c) offer of compensation on
a 0-7 scale. A score of zero was given if the component was not present at all, while a 7
represented a very high level of the component. Table 5 provides the reliabilities of each item.
The Cronbach’s alpha for each item ranged from .54-.86, suggesting inter-rater reliability, thus,
the three ratings were averaged to create a composite rating of each component for each
apologizer.
Expressive behaviors. Two expressive behaviors were coded that may be cues to
apologizers’ internal states and their sincerity. Coders rated to what extent the apologizer
appeared to be experiencing 1) guilt and 2) shame. Coders were asked to focus on the nonverbal
behaviors of the apologizers (e.g., facial expressions, tone of voice, etc.) while making their
judgments. However, coders watched the apology videos with sound so it is not possible to know
whether or not coders made their judgments based on nonverbal behavior (as they were told),
verbal behavior, or a mixture of both. Coders demonstrated high inter-rater reliability
(Cronbach’s alphas reported in Table 5) so I collapsed the scores across coders resulting in a
rating of each component for each apologizer.
Phase 2: Judgments of the Apologies
45
Participants
Eighty-nine judges were recruited for Introduction to Psychology class credit. The judges
were primarily white undergraduate students from Northeastern University. Forty percent of the
judges were female. Judges ranged in age from 18 to 27.
Measures
Judges’ ratings of the transgressions. Before watching each apology, judges read a short
description of the transgression. For example, before watching one of the apologizers, judges
read, “This apologizer told his friend he did not want to be friends with him anymore.” After
reading the transgression description judges responded to the following three questions, “Have
you ever committed a similar transgression?”, “How severe was the transgression that this person
committed?”, and “Has anyone ever committed this transgression against you?”
Judges ’ Ratings of the Apologies. After watching an apology, judges answered the
following questions: “If you were the recipient of this apology, to what extent would you forgive
the person?”, “How sincere was this person's apology?”, and “To what extent would you trust
this person?” These questions were answered on a 7 point scale and indicate the effectiveness of
each apology. A score of one was given for low levels of the construct being measured (e.g.,
definitely would not forgive, not at all sincere, definitely would not trust), while a score of seven
indicated high levels (i.e., definitely would forgiven, very sincere, definitely would trust). Judges
also answered the following questions about forgiving the transgressor: “To what extent can you
feel/understand what this person is feeling?” (this question gets at empathy for the offended
party), and “To what extent do you feel badly for this person (feel sympathy for this person)?”
Furthermore, judges were asked the question, “The cause of this behavior was...” with a rating
scale from 1 (reflects an aspect of the situation) to 7 (reflects an aspect of the transgressor). From
46
now on this variable is referred to as judge-rated attributions. Means were calculated across
judges to create a single score on each of the judge-rated variables for each apologizer5.
Procedure
After being consented, judges were seated in front of a computer monitor displaying a
movie player. Up to five judges were run at the same time in the lab. Judges were asked not to
speak with one another. Judges watched all apologies collected from apologizers in Phase 1.
Prior to watching each videotaped apology, judges read a short description of the transgression
and answered three questions about it. After watching each video, judges evaluated the apology
on several dimensions.
Results
I will present the results section as follows: First I will present the descriptive statistics of
all study variables and the relationship among them. Next, I will examine the apologizer and
coder-rated components associated with judge-rated outcomes by conducting correlation and
multiple regression analyses. I will examine the interactive and additive effects of the
components in the multiple regression analyses. Finally, I will conduct analyses to examine
whether judge-rated emotional and cognitive reactions mediate the relationship between
apologizer and coder-rated components and judge-rated behavioral reactions.
Descriptive Statistics for Apologizer Variables
Table 6 displays the means of apologizer self-reported emotions. Apologizers reported
high levels of guilt, compassion, shame, sadness, and empathy and low levels of jealousy,
surprise, disgust, anger, and excitement. All 22 items were subjected to a principal-factors
analysis followed by a varimax rotation for purposes of data reduction. The resulting
eigenvalues, the percent of variance accounted for by each factor, and the scree plot were
47
examined. There was a clear break in the scree plot between the fourth and fifth factor. To
determine the most appropriate factor solution I examined a four, five, and six factor solution.
The five-factor solution was most interpretable. The factor loadings for each item are shown in
Table 6. I labeled these 5 factors as follows: self-consciousness, apology-reactions, insincerity,
empathy, and contentment. Self-consciousness was made up of guilt, shame, sadness, love,
worry, fear, and embarrassment. Apology-reactions consists of excitement, anger, surprise,
relaxation, inspiration, and happiness. Jealousy, contempt, disgust, and amusement are all part of
the factor labeled insincerity. Empathy consists of pity, empathy, and compassion and
contentment is made up of contentment and hope.
The self-consciousness items are self-conscious emotions that may occur upon reflection
of the transgression. The emotions within the apology-reactions factor would not normally
correspond with one another because of their mix of valence and arousal. However, in the
context of an apology, the emotions may be related because transgressors were likely to
experience them after completing their apology. Apologizers may feel relief, excitement, or even
anger post-apology. The insincerity factor consists of emotions that may occur in a fake or
forced apology. The empathy factor is made up of emotions that are associated with identifying
how another is feeling. The contentment factor contains two emotions associated with being
content.
The means and intercorrelations of the five factors are presented in Table 7. The data
reveal that the factors of self-consciousness and empathy are significantly correlated.
Considering that empathy often leads to the experience of guilt and shame (Eisenberg, 1986;
Hoffman, 1982; Zahn-Waxler & Robinson, 1995), this relationship is understandable. The data
also reveal that empathy is associated with feelings of contentment. The apology-reactions factor
48
is associated with insincerity, empathy, and contentment. As mentioned earlier, this factor
contains emotions that vary greatly in valence and arousal, which may account for these
relationships.
Table 8 displays correlations between the five emotion factors and the apology and
transgression-specific question variables. Only one relationship was significant among these
analyses: the relationship between self-consciousness and apology sincerity. Although not a
strong correlation, this relationship suggests that those experiencing self-conscious emotions
may provide sincere apologies.
Descriptive Statistics for Coder Variables
The descriptive statistics for the coder variables are presented in Table 5. The mean for
compensation was low relative to the other coded variables. Compensation received a score of
zero for 28 of the 37 apologies suggesting that compensation was rarely offered.
Remorse was positively correlated with acknowledgment of violated rules and norms,
guilt, and shame. Compensation was not associated with any of the self-conscious emotions,
while acknowledgment of violated rules and norms was positively correlated with guilt and
shame.
Descriptive Statistics for Judge Variables
The descriptive statistics and relationships among the judge variables are presented in
Table 9. Judges reported less empathy and sympathy for apologizers when they (i.e., the judges)
had committed a similar transgression. Judges also reported less empathy for the apologizer if
they had been the recipient of a similar transgression in the past (i.e., if someone committed this
transgression against the judges).
49
The three judge-rated outcomes (i.e., forgiveness, trust, and sincerity) were all positively
correlated. Furthermore, judges reported high levels of these three variables when they also
reported experiencing empathy and sympathy and attributed the transgression to external causes.
Interestingly, severity of the transgression was only related to the forgiveness variable. Judges
were less willing to forgive when the transgression was seen as severe.
Relationships Between Apologizer, Coder, and Judge Variables
I next examined the relationships between apologizer, coder, and judge variables (see
Tables 10 and 11). First, I correlated apologizer emotion (Figure 1,a) with coder ratings of verbal
and expressive behavior (Figure 1,b). Results revealed that apologizers who reported high levels
of self-conscious emotions were rated by coders as expressing guilt. Thus, coders may have been
able to detect apologizers’ internal experience.
Next I examined the relationship between apologizer emotion (Figure 1,a) and judges’
emotional and cognitive reactions (i.e., empathy, sympathy, attributions; Figure 1,c).
Apologizers’ self-conscious emotions were significantly correlated with judge sympathy and
empathy, suggesting judges recognize the apologizers’ negative emotions and subsequently feel
sympathy for them. I next correlated apologizers’ emotions (Figure 1,a) with judges’ behavioral
reactions (e.g., forgiveness, sincerity, and trust; Figure 1,d). Self-conscious emotions were
positively correlated with forgiveness, trust, and ratings of sincerity.
Finally, I correlated coders’ ratings (Figure 1,b) with judges’ emotional and cognitive
reactions (Figure 1,c) and judges’ behavior reactions (Figure 1,d). Coder-rated remorse,
acknowledgment of violated rules and norms, guilt, and shame were all positively correlated
with judge-rated empathy, sympathy, forgiveness, sincerity, and trust. Thus, judges have positive
reactions to apologizers when apologizers include remorse and an acknowledgment of violated
50
rules and norms in their apologies as well as when apologizers appear to be experiencing guilt
and shame.
As shown in Table 9, judge-rated severity is negatively associated with judge-rated
forgiveness. This relationship has also been shown in the literature (Fincham, Jackson, & Beach,
2005). Therefore, it was important to test whether the apology components predicted the judge-
rated outcomes while controlling for judge-rated severity of the transgression. I ran several
partial correlations, correlating the significant predictors (i.e., self-conscious emotion factor,
guilt, shame, remorse, and acknowledgment of violated rules and norms) with the judge-rated
outcomes while controlling for judge-rated severity. See Table 12 for results. Guilt, shame,
remorse, acknowledgment of violated rules and norms, and the self-conscious emotion factor
were all significant predictors of empathy, sympathy, forgiveness, trust, and sincerity. Thus, the
apology components predicted judge-rated outcomes at all levels of transgression severity.
Apologizer and Coder Variables Predicting Judge Ratings
Thus far, transgressors’ emotional reactions (Figure 1,a) and apology components (Figure
1,b) predict the role-playing offended parties’ emotional and cognitive reactions (Figure 1,c) and
behavioral reactions (Figure 1,d). These analyses, however, do not test (1) the individual
contributions of the apologizer-rated emotions (Figure 1,a) and coder-rated components (Figure
1,b) in the prediction of the judge-rated outcomes (Figure 1,c and Figure 1,d), (2) the interactions
among the verbal (i.e., remorse, acknowledgment of violated rules and norms, and
compensation) and expressive behavior (i.e., guilt, shame, self-conscious emotions) components;
(3) the additive effects of the various components (i.e., does including more than one component
improve prediction of judge-rated outcomes); or (4) why the apology components are associated
with the judge-rated outcomes. In order to answer these questions I first used multiple regression
51
to assess the individual contributions each of the components makes in the prediction of judge-
rated outcomes. These analyses were conducted separately for apologizer-rated emotions and
coder-rated verbal components. Next, to assess the interactive and additive effects between the
components, coder and apologizer-rated variables and their interaction terms were entered as
predictors of judge-rated emotional, cognitive, and behavioral reactions. Finally, analyses were
conducted to examine whether judge-rated emotional reactions (Figure 1,c) mediated the
relationship between coder-rated verbal and expressive behavior apology components (Figure
1,b), apologizer-rated emotions (Figure 1,a), and judge-rated behavioral reactions (Figure 1,d).
Apologizer-rated emotions predicting judge reactions. To examine the individual
contributions that apologizer-rated emotions (Figure 1,a) have on judges’ emotional, cognitive,
and behavioral reactions (Figure 1,c and Figure 1,d), the five apologizer-rated emotion factors
were entered as predictors of the judge-rated outcomes. Table 13 displays the results of this
analysis. The self-conscious emotion factor was the sole significant predictor of empathy,
sympathy, trust, and sincerity. The self-consciousness factor did not significantly predict
forgiveness or attributions.
Coder-rated verbal components predicting judge reactions. To test the predictive
ability of the coder-rated verbal components I entered acknowledgment of violated rules and
norms, remorse, and compensation as predictors of judges’ emotional and cognitive reactions
(Figure 1,c) and judges’ behavioral reactions (Figure 1,d). To control for severity of the
transgression, judge-rated severity was also entered as a predictor. The individual contributions
of each component (i.e., acknowledgment, remorse, compensation, and severity) were examined
for each judge-rated outcome (Figure 1,c and Figure 1,d). See Table 14 for the results. Remorse
was a significant predictor for all judge-rated outcomes except for attributions. Acknowledgment
52
of violated rules and norms and compensation did not predict any of the judge-rated outcomes.
Severity was a positive predictor of attributions (i.e., severity was associated with the tendency
to make a dispositional attribution) and a negative predictor of forgiveness and trust.
Interaction and additive effects of the verbal and expressive behavior components.
Thus far the analyses reveal that coder-rated remorse predicts judge-rated empathy, sympathy,
forgiveness, trust, and ratings of apologizer sincerity. Furthermore, coder-rated guilt and shame
and apologizer-rated self-conscious emotions predict judge-rated outcomes. What is still
unknown is whether there are interaction or additive effects between these components. Because
guilt and shame were highly correlated, I aggregated them to create a coder-rated guilt-shame
aggregate variable for these analyses. Six interaction analyses were completed, one for each of
the judge-rated outcomes: empathy, sympathy, attributions, forgiveness, trust, and sincerity.
To begin, I centered the variables of interest: coder-rated remorse, guilt, and shame as
well as apologizer-rated self-conscious emotions factor. I then created four interaction variables
by multiplying (1) remorse with self-conscious emotions (2) remorse with the guilt-shame
aggregate, (3) self-conscious emotions and the guilt-shame aggregate, and (4) remorse, self-
conscious emotions, and the guilt-shame aggregate. For each analysis the centered variables and
the interaction variables were entered as predictors to predict the judge-rated outcome (i.e.,
empathy, sympathy, attribution, forgiveness, trust, or sincerity). See Table 15 for interaction
results. As the table shows, there were no significant interaction effects. Thus, to examine
additive effects (i.e., whether the components are significant predictors of the dependent
variables when controlling for each other) I simplified the model by including remorse, self-
conscious emotions, and the guilt-shame aggregate as predictors of empathy, sympathy,
attributions, forgiveness, trust, and sincerity. See Table 16 for the results. As the table reveals,
53
there were additive effects of remorse and self-conscious emotions in the prediction of empathy,
sympathy, and sincerity. There were no additive effects (i.e., the betas were nonsignificant) in
the prediction of trust and forgiveness.
The mediating role of judges’ cognitive and affective responses. The final question I
sought to answer was why the components are effective. With coder-rated verbal (i.e., remorse)
and expressive behavior components (i.e., coder-rated guilt, coder-rated shame, and apologizer-
rated self-conscious emotions) as well as judge-rated emotional and cognitive responses (i.e.,
empathy and sympathy) significantly predicting judge-rated behavioral reactions (i.e.,
forgiveness, trust, and sincerity), I conducted a mediation analysis to test whether the judge-rated
emotional and cognitive responses mediated the relationship between the coder-rated verbal and
expressive behavior components and the judge-rated behavioral reactions (see Figure 2). For
these analyses, coder-rated remorse, guilt, shame, and apologizer self-conscious emotions were
aggregated to create a components variable. Judge-rated empathy and sympathy were aggregated
to create a judge-rated empathic response variable. Judge-rated forgiveness, trust, and sincerity
were aggregated to create a judge-rated outcomes variable.
To examine the statistical significance of the effect of the components variable (i.e.,
Figure 2,a; the mediated variables) on the judge-rated outcomes variable (i.e., Figure 2,c;
forgiveness, trust, and sincerity) via the mediators (i.e., Figure 2, b; judge-rated empathic
response) I followed procedures suggested by Kenny, Kashy, and Bolger (1998). As shown in
Figure 2, judge-rated empathic response mediated the effect of components on judge-rated
outcomes, Sobel Z = 4.299, p = .000. Thus, the apology components predicted judge-rated
forgiveness, trust, and sincerity by virtue of their ability to elicit judge empathic response (i.e.,
judge-rated empathy and sympathy). However, these results can also be interpreted in an
54
alternative manner. The path between judge empathic response and judge-rated outcomes is
quite strong and may be the reason why the mediation analysis was significant. In fact, if judge
empathic response and judge-rated outcomes are switched in the mediation model (i.e., judge
empathic response becomes the dependent variable and judge-rated outcomes becomes the
mediator), the mediation analysis is also significant, Sobel Z = 4.022 , p < .001. Another way to
interpret this data is that judge empathic response is an ingredient of the judge outcomes. For
example, empathy is an element of forgiveness, rather than a cause.
Discussion
The aim of the present study was to examine the relationship between apology
components (i.e., verbal and expressive behavior actions displayed by an apologizer) and judge-
rated outcomes. The apology components included both apologizer-rated emotions and coder-
rated verbal (i.e., remorse, acknowledgment of violated rules and norms, and compensation) and
expressive behavior (i.e., guilt and shame) components. The judge-rated outcomes included
empathy, sympathy, dispositional attributions, forgiveness, trust, and sincerity.
Descriptive statistics revealed apologizers experienced high levels of self-conscious
emotions. Therefore, the methodology used was able to evoke the emotional experience typically
felt by apologizers (Lazare, 2004). It also was revealed that apologizers rarely made verbal offers
of compensation. This may be due to the nature of the relationship between the apologizer and
recipients. Apologizers were asked to apologize to a “close other.” As previously speculated,
compensation may not be used in emotion-based close relationships whereas it might be a
common strategy used by individuals apologizing to outgroups or exchange-based relationships.
Correlation and multiple regression analyses revealed that coder-rated remorse, guilt, and
shame were significant predictors of judge-rated empathy, sympathy, forgiveness, trust, and
55
sincerity. Similarly, apologizer-rated self-conscious emotions also predicted these outcomes.
Thus, apologizers who felt and expressed their self-conscious emotions were more likely to
receive positive responses from offended parties. The relationships between judge-rated
empathy, sympathy, forgiveness, trust, and sincerity with coder-rated remorse, guilt, and shame,
and apologizer-rated self-conscious emotions remained significant after controlling for judge-
rated severity of the transgression. Thus, these components predict judge-rated outcomes
regardless of transgression severity level.
Analyses revealed an additive effect for coder-rated remorse and apologizer-rated self-
conscious emotions. These additive effects suggest that apologies produce the most effective
judge outcomes when apologizers feel bad about the transgression (i.e., reports self-conscious
emotions) and express negative emotions externally through verbal behavior (i.e., coder-rated
remorse). This finding has an important implication for apologizers suggesting that the best
apologies are ones in which apologizers genuinely feel sorry for what they have done.
The final question addressed by Study 2 was why verbal and expressive behavior apology
components elicit judge-rated outcomes. To answer this question analyses were conducted to test
whether judge-rated empathy and sympathy mediated the relationship between the apology
components (i.e., coder-rated remorse, guilt, shame, and apologizer self-conscious emotions) and
judge-rated outcomes (i.e., forgiveness, trust, and sincerity). Although the mediation model was
significant, the mediator (i.e., empathy and sympathy) and judge-rated outcome (i.e.,
forgiveness, trust, and sincerity) path was extremely high. The strong relationship between judge
empathic response (i.e., empathy and sympathy) and judge-rated outcomes (forgiveness, trust,
and sincerity) suggests these variables may be so highly correlated that they are interchangeable
rather than causal. McCullough et al. (1997) reported a similar finding when they examined the
56
relationship between empathy and forgiveness. In their study, items from an empathy scale and a
forgiveness scale were factor analyzed. The factor analysis revealed two factors. However, three
of the eight items from the empathy scale loaded on both factors suggesting that empathy and
forgiveness share similar features.
Limitations. Apologies were observed by judges instead of the actual individuals who
were offended. This approach may not be ideal, but it is similar to the analogue patient
methodology used in doctor-patient communication research (Blanch-Hartigan, Hall, Krupat, &
Irish, 2012; Schmid Mast, Hall, & Roter, 2008). In this research, analogue patients and doctors
are believed to provide results that are informative about actual patient-doctor communication. In
fact, research has demonstrated that subjective ratings provided by analogues correspond with
ratings provided by real patients (Blanch-Hartigan, Hall, Krupat, & Irish, 2012). Therefore the
judgments made by the judges in the current study may closely reflect how the intended apology
recipients would respond.
Using this methodology, insight may also be gained into the effects apologies have on
people indirectly affected by a transgression. A transgression typically offends people in addition
to the intended recipient. The transgressor’s apology may influence these unintended victims.
While at a holiday celebration, a spouse’s hurtful comments to her husband might be offensive to
family members within hearing distance. If these same bystanders hear the woman apologize to
her husband, their vicarious feelings of being offended may be relieved.
The effects of transgression type were not thoroughly investigated in the current study.
Although transgression severity was measured and statistically controlled in several analyses,
this variable only provided insight into how the intensity of a transgression influences judge
outcomes and not how the type of transgression affects judge outcomes. Transgressions receiving
57
similar ratings of severity may in fact be quite different. For example, in the current study one
apologizer reported cheating on a significant other while another reported breaking a sibling’s
iPod. Both transgressions were rated high on severity, but most likely for different reasons. The
cheating transgression most likely included emotional consequences whereas the iPod
transgression was probably associated with financial consequences. In these two examples, an
effective apology may have required different apology components to elicit victims’ forgiveness.
Future studies may wish to examine this possibility.
58
Chapter 4: General Discussion
According to theorists and researchers, apologies represent an effective tool for
communicating regret and remorse. Previous research has shown that apologies are associated
with positive outcomes for transgressors and offended parties (e.g., Fehr, Gelfand, & Nag, 2010;
DeCremer, van Dijk, & Pillutla, 2010; Robbennolt, 2003). However, apologies can backfire if
not done properly (e.g., DeCremer et al., 2010; Eaton, Struthers, Shomrony, & Santelli, 2007;
Zechmeister, Garcia, Romero, Vas, 2004). The present research investigated the components that
influence an apology’s effectiveness.
Across a meta-analysis and lab study, remorse strongly predicted offended parties’
emotional and cognitive reactions. In the meta-analysis, remorse predicted effect size for both
high and low social-distance relationships. In the lab study, remorse predicted offended party
empathy, sympathy, forgiveness, trust, and sincerity. Furthermore, the outward expression of
remorse (i.e., coder-rated guilt, shame, and apologizer rated self-conscious emotions) similarly
predicted these same outcomes.
A mediation analysis in which the apology components predicted judge outcomes with
judge empathic response as the mediator, was unable to explain why the components work.
Although the proposed mediators were statistically significant, these effects were best explained
by the relationship between the mediators (i.e., empathy and sympathy) and the judge-rated
outcomes (i.e., forgiveness, trust, and sincerity). The strong relationship between the mediators
and the judge-rated outcomes may suggest two things: (1) the measurement of the constructs was
not sufficiently precise or (2) the two constructs may be one in the same, rather than causal. If
the former is true then researchers may wish to use more detailed scales or behavioral measures
of these constructs. If that latter is true, alternative explanations for why the components elicit
59
judge-rated outcomes should be explored. One potential framework to use is McCullough et al.’s
(2013) theory on forgiveness systems. According to this framework, forgiveness can occur when
apologizers signal to the offended parties that they (i.e., the apologizers) are valuable (i.e., they
are willing to share fitness-relevant resources) and safe (i.e., will not harm the offended party
again in the future).
Remorse most likely indicates safety. By expressing remorse, apologizers signal to
offended parties that they realize the benefits gained by committing the transgression did not
outweigh the consequences (i.e., negative affect, loss of friendship, etc.). Thus, apologizers
implicitly, perhaps explicitly, promise that they will not commit the same transgression again.
Apologies that contain an acknowledgment of violated rules and norms might also signal
to victims the relative “safety” of the transgressor. After experiencing a transgression, offended
parties typically search for an explanation for why it occurred. This might result in assigning
blame to the transgressor (i.e., the offended party will attribute the cause of the behavior).
Typically, offended parties make dispositional attributions, believing the transgressor was the
cause of the transgression and the transgression was not due to some situational force (e.g.,
Weiner, Graham, Peter, & Zmuidinas, 1991). An acknowledgment of violated rules and norms
can cement this inference by reminding offended parties of the transgression and its negative
impact (Kim, Ferrin, Cooper, & Dirks, 2004). Supporting this argument, acknowledgment of
violated rules and norms negatively predicted effect size for high social-distance relationships in
the meta-analysis. This finding was not observed, however, in low social-distance relationships
or in the lab study (which involves low social distance relationships). In the latter two cases
acknowledgment of violated rules and norms did not predict effect size or judge-rated outcomes.
60
In close relationships, attributions have been well-established over time and an apology that
acknowledges rules and norms were violated will not change those attributions.
Finally, using McCullough et al.’s (2013) forgiveness systems model, compensation may
be effective because it is a signal of value. When apologizers make an offer of compensation,
they express their willingness to share resources and indicate the value in maintaining a
relationship with them. In the meta-analyses compensation was a significant predictor of effect
size in high social-distance relationships, but not for low social-distance relationships.
Furthermore, it was not significant in the lab study. The type of compensation coded for in the
current studies may not occur in low social-distance relationships because these relationships are
not exchange-based. In these relationships it may be the case that compensation comes in the
form of emotional support or services. Alternatively it may simply not occur and value is
signaled through other channels. Future research may wish to examine alternative components
that may elicit value perceptions.
Limitations and Future directions
The current studies examine the effectiveness of three apology components (i.e., remorse,
an acknowledgment of violated rules and norms, and compensation). However, there may be
additional components that influence an apology’s effectiveness. Using McCullough et al.’s
(2013) forgiveness framework, components that may signal to offended parties that the
apologizer is valuable and safe are predicted to be effective in eliciting positive outcomes from
the offended party such as forgiveness. One potential component is excuse making. Excuse
making is often examined outside the context of apology or in comparison to an apology (e.g.,
Conlon & Murray, 1996), but apologies may also include excuse or external blame. By holding a
61
force outside of the apologizer culpable, the apologizer is signaling that the transgression was not
the fault of the apologizer and is unlikely to happen again.
Another interesting line of work involves linguistic analysis of apologies. For example,
using Pennebaker, Booth, and Francis’ (2007) Linquistic Inquiry and Word Count (LIWC)
analysis researchers can calculate the degree to which apologizers use different categories of
words such as negative emotions, self-references, causal words and determine how effective they
are in producing positive offended party outcomes. Similarly, nonverbal behavior researchers
may be interested in using the Facial Action Coding System (FACS; Ekman, Friesen, Hager,
2002) to identify the apologizers’ expressions associated with offended party outcomes.
Another potential avenue of research is to examine the interaction between verbal and
expressive behavior components. The lab study did not find significant interactions between
verbal and expressive behavior components, but this may be because apologizers rarely exhibited
low levels of guilt, shame, and self-conscious emotions when exhibiting high levels of verbal
remorse. It would be interesting to create videos in which actors, who are not experiencing true
self-conscious emotions, read scripts of apologies given by truly remorseful apologizers. Judges
could then watch both videos and researchers could examine whether judges forgive the truly
remorseful apologizers more than the actors.
Finally, researchers interested in apologies may wish to further explore the interaction
between judges’ traits and apology components. Fehr and Gelfand (2010) found individual
differences in what offended parties must hear in an apology in order for it to be effective.
Specifically they found that forgiveness was more likely when the apology components matched
the offended parties’ self-construal. Therefore, it can be predicted that individual differences in
62
judges will affect the type of apology components they wish to hear in order to forgive. This may
be expanded to cultural and gender differences.
Conclusion
This dissertation examined the components that render an apology more effective.
Identifying effective components may help apologizers who wish to mend relationships. Three
theoretically driven components were examined: remorse, an acknowledgment of violated rules
and norms, and compensation. Based on the findings of the current research studies, apologizers
in high social distance relationships who wish to mend their relationship with an offended party
may wish to include a verbal statement of remorse, an offer of compensation, and should not
make an acknowledgment of a violated rule or norm. Making this acknowledgment may remind
offended parties of what the apologizer has done and makes it difficult for them to forgive.
Apologizers in low social distance relationships seeking forgiveness should include a verbal
expression of remorse in their apology. Furthermore, and more importantly, apologizers should
genuinely feel remorseful for what they have done. This remorse may then express itself through
expressive behaviors and act as a cue to sincerity for the offended party.
63
References
*Ahmed, R. A., Azar, F., & Mullet, E. (2007). Interpersonal forgiveness among Kuwaiti
adolescents and adults. Conflict Management and Peace Science, 24, 159-170.
*Allan, A., Allan, M. M., Kaminer, D., & Stein, D. J. (2006). Exploration of the association
between apology and forgiveness amongst victims of human rights violations. Behavioral
Sciences and the Law, 24, 87-102.
*Anderson, J. C., Linden, W., & Habra, M. E. (2006). Influence of apologies and trait hostility
on recovery from anger. Journal of Behavioral Medicine, 29(4), 347-358.
*Azar, F., Mullet, E., & Vinsonneau, G. (1999). The propensity to forgive: Findings from
Lebanon. Journal of Peace Research, 36(2), 169-181.
*Azar, F., & Mullet, E. (2001). Interpersonal forgiveness among Lebanese: A six-community
study. International Journal of Group Tensions, 30(2), 161-181.
*Bachman, G. F., & Guerrero, L. K. (2006). Forgiveness, apology, and communicative responses
to hurtful events. Communication Reports, 19(1), 45-56.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal Of
Personality And Social Psychology, 51(6), 1173-1182. doi:10.1037/0022-3514.51.6.1173
Bateman, T. S., & Crant, J. (1993). The proactive component of organizational behavior: A
measure and correlates. Journal Of Organizational Behavior, 14(2), 103-118.
Batson, C.D. (1990). How social an animal? The human capacity for caring. American
Psychologist, 45 (3), 336-346. doi: 10.1037/0003-066X.45.3.336
Batson, C. (1991). The altruism question: toward a social-psychological answer. Hillsdale, NJ:
Erlbaum.
64
Baumeister, R. F., Stillwell, A. M., & Heatherton, T. F. (1994). Guilt: An interpersonal
approach. Psychological Bulletin, 115(2), 243-267. doi:10.1037/0033-
2909.115.2.243
Baumeister, R. F., & Tice, D. M. (1990). Anxiety and social exclusion. Journal Of Social
And Clinical Psychology, 9(2), 165-195. doi:10.1521/jscp.1990.9.2.165
*Bhandari, M. S., & Polonsky, M. J. (2006). Reducing Consumer Switching Intentions
Following Service Failure: Do Empowerment and Apology Help? Paper Presented to the
Annual Conference of the Australian and New Zealand Marketing Academy, Brisbane,
Australia, 4-6 December, 2006.
Blanch-Hartigan, D., Hall, J.A., Irish, J., Krupat, E. (2012). Can naive viewers put themselves in
the patients’ shoes? Reliability and validity of the analogue patient methodology. Medical
Care. doi: 10.1097/MLR.0b013e31822945cc
*Bolkan, S., & Daly, J. A. (2009). Organizational responses to consumer complaints: An
examination of effective remediation tactics. Journal of Applied Communication
Research, 37(1), 21-39.
*Bono, G. (2005). Commonplace forgiveness: From healthy relationships to healthy society.
Humboldt Journal of Social Relations, 29(2), 82-111.
*Bono, G., McCullough, M. E., & Root, L. M. (2008). Forgiveness, feeling connected to others,
and well-being: Two longitudinal studies. Personality and Social Psychology Bulletin,
34(2), 182-195.
Borke, H. (1971). Interpersonal perception of young children: Egocentrism or empathy?.
Developmental Psychology, 5(2), 263-269. doi:10.1037/h0031267
65
Bornstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2005). Comprehensive meta-analysis
(2nd ed.). Englewood, NJ: Biostat.
*Bornstein, B. H., Rung, L. M., & Miller, M. K. (2002). The effects of defendant remorse on
mock juror decisions in a malpractice case. Behavioral Sciences and the Law, 20, 393-
409.
Boyd, D.P. (2011) Art and artifice in public apologies. Journal of Business Ethics, 104 (3), 299-
309. DOI: 10.1007/s10551-011-0915-9
*Braaten, D. O., Cody, M. J., & DeTienne, K. B. (1993). Account episodes in organizations:
Remedial work and impression management. Management Communications Quarterly,
6(3), 219-250.
*Bradford, J. L., & Garrett, D. E. (1995). The effectiveness of corporate communicative
responses to accusations of unethical behavior. Journal of Business Ethics, 14, 875-892.
*Brown, R. P., & Phillips, A. (2005). Letting bygones be bygones: Further evidence for the
validity of the Tendency to Forgive scale. Personality and Individual Differences, 38,
627-638.
*Brown, R. P., Wohl, M. J. A., & Exline, J. J. (2008). Taking up offenses: Secondhand
forgiveness and group identification. Personality and Social Psychology Bulletin, 34(10),
1406-1419.
Conlon, D. E., & Murray, N. M. (1996). Customer perceptions of corporate responses to
product complaints: The role of explanations. Academy Of Management Journal, 39(4),
1040-1056. doi:10.2307/256723
* Coombs W. T., Fediuk T. A., & Holladay S. J. (2007). Further explorations of post-crisis
communication and stakeholder anger: The negative communication dynamic model.
66
Paper presented at the 9th International Public Relations Research Conference, Miami,
FL, 8-11 March.
*Coombs, W. T., & Holladay, S. J. (2008). Comparing apology to equivalent crisis response
strategies: Clarifying apology’s role and value in crisis communication. Public Relations
Review, 34, 252-257.
Cunningham, M. (2004). Apologies in Irish politics: A commentary and critique. British
History, 18(4), 80-92. doi: 10.1080/13619460412331296919
Darby, B. W., & Schlenker, B. R. (1982). Children's reactions to apologies. Journal of
Personality And Social Psychology, 43(4), 742-753. doi:10.1037/0022-3514.43.4.742
*Dardis, F., & Haigh, M. M. (2009). Prescribing versus describing: Testing image restoration
strategies in a crisis situation. Corporate Communications: An International Journal,
14(1), 101-118.
*Dawar, N., & Pillutla, M. M. (2000). Impact of product-harm crises on brand equity: The
moderating role of consumer expectations. Journal of Marketing Research, 37, 215-226.
*Day, M. V., & Ross, M. (2011). The value of remorse: How drivers’ responses to police predict
fines for speeding. Law and Human Behavior, 35, 221-234.
*De Cremer, D., & Schouten, B. C. (2008). When apologies for injustice matter: The role of
respect. European Psychologist, 13(4), 239-247.
*DeCremer, van Dijk, & Pillutla (2010). Explaining unfair offers in ultimatum games and
their effects on trust: An experimental approach. Business Ethics Quarterly, 20(1), 107-
126.
67
*Dean, D. H. (2004). Consumer reaction to negative publicity: Effects of corporate reputation,
response, and responsibility for a crisis event. Journal of Business Communication, 41(2),
192-211.
*Dean, D. H. (2005). After the unethical ad: A comparison of advertiser response strategies.
Business and Society Review, 110(4), 433-458.
*Dirks, K. T., Kim, P. H., Ferrin, D. L., & Cooper, C. D. (2011). Understanding the effects of
substantive responses on trust following a transgression. Organizational Behavior and
Human Decision Processes, 114, 87-103.
*Dunn, D., & Cody, M. J. (2000). Account credibility and public image: Excuses, justifications,
denials, and sexual harassment. Communication Monographs, 67(4), 372-391.
*Eaton, J., & Struthers, C. W. (2006). The reduction of psychological aggression across varied
interpersonal contexts through repentance and forgiveness. Aggressive Behavior, 32, 195-
206.
*Eaton, J., Struthers, C., & Santelli, A. G. (2006). The Mediating Role of Perceptual
Validation in the Repentance-Forgiveness Process. Personality And Social
Psychology Bulletin, 32(10), 1389-1401. doi:10.1177/0146167206291005
*Eaton, J., Struthers, C., Shomrony, A., & Santelli, A. G. (2007). When apologies fail: The
moderating effect of implicit and explicit self-esteem on apology and forgiveness.
Self And Identity, 6(2-3), 209-222. doi:10.1080/15298860601118819
Eisenberg, N. (1986). Altruistic emotion, cognition, and behavior. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Eisenberg, N., & Miller, P. A. (1987). The relation of empathy to prosocial and related
behaviors. Psychological Bulletin, 101(1), 91-119. doi:10.1037/0033-2909.101.1.91
68
Ekman, P., Friesen, W. V., & Hager, J. C. (2002). Facial action coding system. Salt Lake City: A
Human Face.
Exline, J., & Baumeister, R. F. (2000). Expressing forgiveness and repentance: Benefits and
barriers. In M. E. McCullough, K. I. Pargament, C. E. Thoresen (Eds.), Forgiveness:
Theory, research, and practice (pp. 133-155). New York, NY US: Guilford Press.
*Exline, J. J., Baumeister, R. F., Bushman, B. J., Campbell, W. K., & Finkel, E. J. (2004). Too
proud to let go: Narcissistic entitlement as a barrier to forgiveness. Journal of Personality
and Social Psychology, 87(6), 894-912.
*Exline, J. J., Baumeister, R. F., Zell, A. L., Kraft, A. J., & Witvliet, C. V. O. (2008). Not so
innocent: Does seeing one’s own capability for wrongdoing predict forgiveness? Journal
of Personality and Social Psychology, 94(3), 495-515.
*Exline, J. J., Deshea, L., & Holeman, V. T. (2007). Is apology worth the risk? Predictors,
outcomes, and ways to avoid regret. Journal of Social and Clinical Psychology, 26(4),
479-504.
Fagenson, E. A., & Cooper, J. (1987). When push comes to power: A test of power restoration
theory’s explanation for aggressive conflict escalation. Basic and Applied Social
Psychology, 8, 273–293.
Fehr, R., & Gelfand, M. J. (2009). When apologies work: How matching apology components to
victims' self-construals facilitates forgiveness. Organizational Behavior and Human
Decision Processes.doi: 10.1016/j.obhdp.2010.04.002
Fehr, R., Gelfand, M. J., & Nag, M. (2010). The road to forgiveness: A meta-analytic
synthesis of its situational and dispositional correlates. Psychological Bulletin, 136(5),
894-914. doi:10.1037/a0019993
69
*Ferrin, D. L., Kim, P. H., Cooper, C. D., & Dirks, K. T. (2007). Silence speaks volumes: The
effectiveness of reticence in comparison to apology and denial for responding to
integrity- and competence-based trust violations. Journal of Applied Psychology, 92(4),
893-908.
Feshbach, N. D. (1975). Empathy in children: Some theoretical and empirical considerations.
The Counseling Psychologist, 5(2), 25-30. doi:10.1177/001100007500500207
Fincham, F.D., Jackson, H., & Beach, S.R.H. (2005). Transgression severity and forgiveness:
Different moderators for objective and subjective severity. Journal of Social and Clinical
Psychology.
Frantz, C., & Bennigson, C. (2005). Better late than early: The influence of timing on
apology effectiveness. Journal Of Experimental Social Psychology, 41(2), 201-207.
doi:10.1016/j.jesp.2004.07.007
*Giacolone, R.A., & Payne, S.L. (1995). Evaluations of employee rule violations: The impact of
impression management effects in historical context. Journal of Business Ethics, 6, 477-
487. doi: 10.1007/BF00872088
*Giner-Sorolla, R., Castano, E., Espinosa, P., & Brown, R. (2008). Shame expressions reduce
the recipient’s insult from outgroup reparations. Journal of Experimental Social
Psychology, 44, 519-526.
*Girard, M., & Mullet, E. (1997). Forgiveness in adolescents, young, middle-aged, and older
adults. Journal of Adult Development, 4(4), 209-220.
*Goei, R., Roberto, A., Meyer, G., & Carlyle, K. (2007). The effects of favor and apology on
compliance. Communication Research, 34(6), 575-595.
70
Goffman, E. (1967). Interactional ritual: Essays on Face-to-Face-Behavior. New York, NY:
Anchor.
*Gold, G. J., & Weiner, B. (2000). Remorse, confession, group identity, and expectancies about
repeating a transgression. Basic and Applied Social Psychology, 22(4), 291-300.
*Goodwin, C., & Ross, I. (1992). Consumer responses to service failures: Influence of
procedural and interactional fairness perceptions. Journal of Business Research, 25, 149-
163.
Gottman, J. (1994). What predicts divorce?. Hillsdale, NJ: Lawrence Erlbaum Associates.
Govier, T. & Verwoerd, Wilhelm. (2002). The promise and pitfalls of apology. Journal of Social
Philosophy, 33(1), 67-82. doi: 10.1111/1467-9833.00124.
*Green, J. D., Burnette, J. L., & Davis, J. L. (2008). Third party forgiveness: (Not) forgiving
your close other’s betrayer. Personality and Social Psychology Bulletin, 34(3), 407-418.
*Gunderson, P. R., & Ferrari, J. R. (2008). Forgiveness of sexual cheating in romantic
relationships: Effects of discovery method, frequency of offense, and presence of
apology. North American Journal of Psychology, 10(1), 1-14.
*Haigh, M. M., & Brubaker, P. (2010). Examining how image restoration strategy impacts
perceptions of corporate social responsibility, organization-public relationships, and
source credibility. Corporate Communications: An International Journal, 15(4), 453-468.
Hareli, S., & Eisikovits, Z. (2006). The role of communicating social emotions accompanying
apologies in forgiveness. Motivation And Emotion, 30(3), 189-197.
doi:10.1007/s11031-006-9025-x
71
Hareli, S., Shomrat, N., & Biger, N. (2005). The role of emotions in employees' explanations for
failure in the workplace. Journal Of Managerial Psychology, 20(8), 663-680.
doi:10.1108/02683940510631435
*Harrell, W. A. (1980). Retaliatory aggression by high and low Machiavellians against
remorseful and non-remorseful wrongdoers. Social Behavior and Personality, 8(2), 217-
220.
*Hodgins, H. S., & Liebeskind, E. (2003). Apology versus defense: Antecedents and
consequences. Journal of Experimental Social Psychology, 39, 297-316.
Hoffman, M. L. (1982). Development of prosocial motivation: Empathy and guilt. In N.
Eisenberg (Ed)., The development of prosocial behavior (pp. 218-231). New York:
Academic Press.
Hogan, R. (1969). Development of an empathy scale. Journal Of Consulting And Clinical
Psychology, 33(3), 307-316. doi:10.1037/h0027580
*Howley, M.J. (2009). The use of apology in health care. Journal of Medical Marketing, 9(4),
279-289. doi: 10.1057/jmm.2009.30
*Huang, Y. (2006). Crisis situations, communication strategies, and media coverage: A
multicase study revisiting the communicative response model. Communication Research,
33, 180-205.
Jehle, A., Miller, M. K., & Kemmelmeier, M. (2009). The influence of accounts and
remorse on mock jurors’ judgments of offenders. Law And Human Behavior, 33(5), 393-
*Kelley, D. L., & Waldron, V. R. (2005). An investigation of forgiveness-seeking
communication and relational outcomes. Communication Quarterly, 53(3), 339-358.
72
*Kim, H. K., & Yang, S-U. (2009). Cognitive processing of crisis communication: Effects of
CSR and crisis response strategies on stakeholder perceptions of a racial crisis dynamics.
Public Relations Journal, 3(1), 1-39.
*Kim, P. H., Ferrin, D. L., Cooper, C. D., & Dirks, K. T. (2004). Removing the shadow of
suspicion: The effects of apology versus denial for repairing competence-versus integrity-
based trust violations. Journal of Applied Psychology, 89(1), 104-118.
*Kleinke, C. L., Wallis, R., & Stalder, K. (1992). Evaluation of a rapist as a function of
expressed intent and remorse. The Journal of Social Psychology, 132(4), 525-537.
Lazare, A. On Apology. (2005). Oxford, England: Oxford University Press Inc.
Leary, M. R., & Kowalski, R. M. (1990). Impression management: A literature review and two-
component model. Psychological Bulletin, 107(1), 34-47. doi:10.1037/0033-
2909.107.1.34
*Leonard, D. J., Mackie, D. M., & Smith, E. R. (2011). Emotional responses to intergroup
apology mediate intergroup forgiveness and retribution. Journal of Experimental Social
Psychology, 47(6), 1198–1206.
Levin, G. (2010). 10 Years after AOL-Time Warner, Gerald Levin says he's sorry'.
Retrieved from http://www.businessinsider.com/10-years-after-aol-time-warner-gerald-
levin-says-hes-sorry-2010-1.
*Liao, H. (2007). Do it right this time: The role of employee service recovery performance
in customer-perceived justice and customer loyalty after service failures. Journal of
Applied Psychology, 92(2), 475-489. doi:10.1037/0021-9010.92.2.475
Lipsey, M.W., & Wilson, D.B. (2001). Practical Meta-Analysis. Thousand Oaks, CA: Sage
Publications, Inc.
73
*Luzombe, L. O., & Dean, K. E. (2009). Moderating and intensifying factors influencing
forgiveness by priests and lay people. Pastoral Psychology, 57, 263-274.
*Lyon, L., & Cameron, G. T. (2004). A Relational Approach Examining the Interplay of Prior
Reputation and Immediate Response to a Crisis. Journal Of Public Relations Research,
16(3), 213-241. doi:10.1207/s1532754xjprr1603_1
McCullough, M. E., Kurzban, R., & Tabak, B. A. (2013). Cognitive systems for revenge and
forgiveness. Behavioral and Brain Sciences, 1(1), 1-15.
*McCullough, M. E., Luna, L. R., Berry, J. W., Tabak, B. A., & Bono, G. (2010). On the form
and function of forgiving: Modeling the time-forgiveness relationship and testing the
valuable relationships hypothesis. Emotion, 10(3), 358-376.
*McCullough, M. E., Rachal, K., Sandage, S. J., Worthington, E. R., Brown, S., & Hight,
T. L. (1998). Interpersonal forgiving in close relationships: II. Theoretical elaboration
and measurement. Journal Of Personality And Social Psychology, 75(6), 1586-1603.
doi:10.1037/0022-3514.75.6.1586
*McCullough, M. E., Worthington, E. r., & Rachal, K. C. (1997). Interpersonal forgiving in
close relationships. Journal Of Personality And Social Psychology, 73(2), 321-336.
doi:10.1037/0022-3514.73.2.321
*Morse, C. R., & Metts, S. (2011). Situational and communicative predictors of forgiveness
following a relational transgression. Western Journal of Communication, 75(3), 239-258.
*Mullet, E., Rivière, S., & Sastre, M. (2007). Cognitive processes involved in blame and blame-
like judgments and in forgiveness and forgiveness-like judgments. The American Journal
Of Psychology, 120(1), 25-46.
74
*Niedermeier, K. E., Horowitz, I. A., & Kerr, N. L. (2001). Exceptions to the rule: The effects of
remorse, status, and gender on decision making. Journal of Applied Social Psychology,
31(3), 604-623.
*Ohbuchi, K., Kameda, M., & Agarie, N. (1989). Apology as aggression control: Its role in
mediating appraisal of and response to harm. Journal of Personality and Social
Psychology, 56(2), 219-227. doi:10.1037/0022-3514.56.2.219
*Orleans, J. F., & Gurtman, M. B. (1984). Effects of physical attractiveness and remorse on
evaluations of transgressions. Academic Psychology Bulletin, 6, 49-56.
*Pace, K. M., Fediuk, T. A., & Botero, I. C. (2010). The acceptance of responsibility and
expressions of regret in organizational apologies after a transgression. Corporate
Communications: An International Journal, 15(4), 410-427.
Paulhus, D. L., Robins, R.W., Trzesniewski K.H., and Tracy, J.L. Two replicable suppressor
situations in personality research. Multivariate Behavior Research, 39, 303–328.
Pennebaker, J. W., Booth, R. J., & Francis, M. E. (2007). LIWC2007: Linguistic inquiry and
word count. Austin, Texas: liwc. net.
*Philpot, C. R., & Hornsey, M. J. (2008). What happens when groups say sorry: The effect of
intergroup apologies on their recipients. Personality and Social Psychology Bulletin,
34(4), 474-487.
*Philpot, C. R., & Hornsey, M. J. (2011). Memory for intergroup apologies and its relationship
with forgiveness. European Journal of Social Psychology, 41, 96-106.
*Pipes, R. B., & Alessi, M. (1999). Remorse and previously punished offense in assignment of
punishment and estimated likelihood of a repeated offense. Psychological Reports, 85,
246-248.
75
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects
in simple mediation models. Behavior Research Methods, Instruments, & Computers,
36(4), 717-731.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and
comparing indirect effects in multiple mediator models. Behavior research methods,
40(3), 879-891.
*Proeve, M. J., & Howells, K. (2006). Effects of remorse and shame and criminal justice
experience on judgements about a sex offender. Psychology, Crime, & Law, 12(2), 145-
161.
*Risen, J. L., & Gilovich, T. (2007). Target and observer differences in the acceptance of
questionable apologies. Journal of Personality and Social Psychology, 92(3), 418-433.
Robbennolt, J.K. (2003). Apologies and legal settlement: An empirical examination.
Michigan Law Review, 102, 460-516.
*Robbennolt, J. K. (2006). Apologies and settlement levers. Journal of Empirical Legal Studies,
3(2), 333-373.
*Robbennolt, J. K. (2008) Attorneys, Apologies, and Settlement Negotiation. Harvard
Negotiation Law Review, 13, 349 – 397.
Rosenthal, R., & Rubin, D.B. (1982). A simple general purpose display of magnitude of
experimental effect. Journal of Educational Psychology, 74, 166–169.
Ross, L. D. (1977). The intuitive psychologist and his shortcomings. Distortions in the attribution
process. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 10, pp.
173-220). New York: Academic Press.
76
*Rumsey, M. G. (1976). Effects of defendant background and remorse on sentencing judgments.
Journal of Applied Social Psychology, 6(1), 64-68.
*Santelli, A. G., Struthers, C., & Eaton, J. (2009). Fit to forgive: Exploring the interaction
between regulatory focus, repentance, and forgiveness. Journal of Personality and Social
Psychology, 96(2), 381-394. doi:10.1037/a0012882
Scher, S. J., & Darley, J. M. (1997). How effective are the things people say to apologize?
Effects of the realization of the apology speech act. Journal Of Psycholinguistic
Research, 26(1), 127-140. doi:10.1023/A:1025068306386
Schlenker, B.R. (1980). Impression management: The self-concept, social identity, and
interpersonal relations. Monterey, CA: Brooks/Cole Publishing Company.
Schlenker, B. R. (1985). Identity and self-identification. In B. R.Schlenker ( Ed.) , The self and
social life (pp. 65– 99). New York: McGraw-Hill.
Schwartz, S. H., & Bilsky, W. (1987). Toward a universal psychological structure of human
values. Journal of Personality and Social Psychology, 53(3), 550-562.
doi:10.1037/0022-3514.53.3.550
Mast, M. S., Hall, J. A., & Roter, D. L. (2008). Caring and dominance affect participants’
perceptions and behaviors during a virtual medical visit. Journal of general internal
medicine, 23(5), 523-527.
*Schmitt, M., Gollwitzer, M., Förster, N., & Montada, L. (2004). Effects of objective and
subjective account components on forgiving. The Journal of Social Psychology, 144(5),
465-485. doi:10.3200/SOCP.144.5.465-486
*Schweitzer, M. E., Hershey, J. C., & Bradlow, E. T. (2006). Promises and lies: Restoring
violated trust. Organizational Behavior and Human Decision Processes, 101, 1-19.
77
*Sigal, J., Hsu, L., Foodim, S., & Betman, J. (1988). Factors affecting perceptions of political
candidates accused of sexual and financial misconduct. Political Psychology, 9(2), 273-
280.
*Struthers, C., Eaton, J., Santelli, A. G., Uchiyama, M., & Shirvani, N. (2008). The effects
of attributions of intent and apology on forgiveness: When saying sorry may not help the
story. Journal of Experimental Social Psychology, 44(4), 983-992.
doi:10.1016/j.jesp.2008.02.006
*Struthers, C. W., Eaton, J., Mendoza, R., Santelli, A. G., & Shirvani, N. (2010).
Interrelationship among injured parties’ attributions of responsibility, appraisal of
appropriateness to forgive the transgressor, forgiveness, and repentance. Journal of
Applied Social Psychology, 40(4), 970-1002.
*Tabak, B. A., McCullough, M. E., Luna, L. R., Bono, G., & Berry, J. W. (2012). Conciliatory
gestures facilitate forgiveness and feelings of friendship by making transgressors appear
more agreeable. Journal of Personality, 80(2), 503-536.
Tangney, J., & Dearing, R. L. (2002). Shame and guilt. New York, NY US: Guilford Press.
Tangney, J., Stuewig, J., & Mashek, D. J. (2007). Moral Emotions and Moral Behavior. Annual
Review Of Psychology, 58345-372. doi:10.1146/annurev.psych.56.091103.070145
Tavuchis, N. (1991). Mea culpa: A sociology of apology and reconciliation. Stanford
University Press.
*Taylor, C., & Kleinke, C. L. (1992). Effects of severity of accident, history of drunk driving,
intent, and remorse on judgments of a drunk driver. Journal of Applied Social
Psychology, 22(21), 1641-1655.
78
Tedeschi, J. T., & Nesler, M. S. (1993). Grievances: Development and reactions. In R. B. Felson,
J. T. Tedeschi (Eds.) , Aggression and violence: Social interactionist perspectives (pp.
13-45). Washington, DC US: American Psychological Association. doi:10.1037/10123-
001
Tedeschi, J. T., & Riess, M. ( 1981). Identities, the phenomenal self, and laboratory research. In
J. T.Tedeschi ( Ed.) , Impression management theory and social psychological
research (pp. 3– 21). New York: Academic Press.
*Thomas, R. L., & Millar, M. G. (2008). The impact of failing to give an apology and the need-
for-cognition on anger. Current Psychology, 27, 126-134.
*Tomlinson, E. C., Dineen, B. R., & Lewicki, R. J. (2004). The Road to Reconciliation:
Antecedents of Victim Willingness to Reconcile Following a Broken Promise.
Journal of Management, 30(2), 165-187. doi:10.1016/j.jm.2003.01.003
*Tucker, S., Turner, N., Barling, J., Reid, E. M., & Elving, C. (2006). Apologies and
transformational leadership. Journal of Business Ethics, 63(2), 195-207.
*Van Dijke, M., & De Cremer, D. (2011). When social accounts promote acceptance of unfair
ultimatum offers: The role of the victim’s stress responses to uncertainty and power
position. Journal of Economic Psychology, 32, 468-479.
*Van Laer, T., & De Ruyter, K. (2010). In stories we trust: How narrative apologies provide
cover for competitive vulnerability after integrity-violating blog posts. International
Journal of Research in Marketing, 27, 164-174.
*Vassilikopoulou, A., Siomkos, G., Chatzipanagiotou, K., & Triantafillidou, A. (2009). Hotels
on fire: Investigating consumers’ responses and perceptions. International Journal of
Contemporary Hospitality Management, 21(7), 791-815.
79
*Vinsonneau, G., & Mullet, E. (2001). Willingness to forgive among young adolescents: A
comparison between two groups of different cultural origins living in France.
International Journal of Group Tensions, 30(3), 267-278.
Walster, E., Berscheid, E., & Walster, G. W. (1973). New directions in equity research. Journal
of Personality and Social Psychology, 25, 151–176.
*Weiner, B., Graham, S., Peter, O., & Zmuidinas, M. (1991). Public confession and forgiveness.
Journal of Personality, 59, 281-312. doi: 10.1111/j.1467-6494.1991.tb00777.x
Weisman, R. (1999). Detecting remorse and its absence in the criminal justice system. Studies in
Law, Politics, and Society, 19, 121-135.
*Wenzel, M., & Okimoto, T. G. (2010). How acts of forgiveness restore a sense of justice:
Addressing status/power and value concerns raised by transgressions. European Journal
of Social Psychology, 40, 401-417.
*Whited, M. C., Wheat, A. L., & Larkin, K. T. (2010). The influence of forgiveness and apology
on cardiovascular reactivity and recovery in response to mental stress. Journal of
Behavioral Medicine, 33, 293-304.
*Wirtz, J., & Mattila, A. S. (2004). Consumer responses to compensation, speed of recovery and
apology after a service failure. International Journal of Service Industry Management,
15(2), 150-166.
*Wooten, D. B. (2009). Say the right thing: Apologies, reputability, and punishment. Journal of
Consumer Psychology, 19, 225-235.
*Xie, Y., & Peng, S. (2009). How to repair customer trust after negative publicity: The roles of
competence, integrity, benevolence, and forgiveness. Psychology & Marketing, 26(7),
572-589. doi:10.1002/mar.20289
80
*Younger, J. W., Piferi, R. L., Jobe, R. L., & Lawler, K. A. (2004). Dimensions of forgiveness:
The views of laypersons. Journal of Social and Personal Relationships, 21(6), 837-855.
Zahn-Waxler, C., & Robinson, J. (1995). Empathy and guilt: Early origins of feelings of
responsibility. In J. Tangney, K. W. Fischer (Eds.), Self-conscious emotions: The
psychology of shame, guilt, embarrassment, and pride (pp. 143-173). New York, NY US:
Guilford Press.
Zhao, X., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and truths
about mediation analysis. Journal of Consumer Research, 37(2), 197-206.
*Zechmeister, J. S., Garcia, S., Romero, C., & Vas, S. N. (2004). Don't Apologize Unless You
Mean It: A Laboratory Investigation of Forgiveness And Retaliation. Journal Of Social
And Clinical Psychology, 23(4), 532-564. doi:10.1521/jscp.23.4.532.40309
81
Footnotes
1 This dissertation examines how apologies elicit positive outcomes for apologizers (e.g.,
being forgiven by an offended party). However, positive outcomes for apologizers are not always
favorable for offended parties. Forgiving an insincere apologizer who will transgress again is not
adaptive for offended parties. Therefore, it is important to state that the results of these studies
are not meant to be used in a manipulative way. Instead the results should be used to inform
good-intentioned apologizers.
2The majority of the 16% involved jury-defendant relationships. Although these
relationships could be classified as high social distance they differ from others in this category
because the defendant (i.e., an individual) is apologizing to a group, while the other relationships
in this category involve a group (or a representative of a group) apologizing. Thus, we excluded
defendant-jury relationships in social-distance analyses.
3Five additional studies were included in the overall meta-analysis, but did not fall into
our five categories. These studies examined the relationship between apologies and voting (Sigal,
1988) and apologies and hiring intentions (Ferrin, 2007; Kim, 2004).
4The standardized β coefficients are presented for all analyses.
5Means were also calculated separately for female and male judges. Independent t-tests
comparing the male and female means for the six judge-rated variables revealed no differences.
82
Table 1
Characteristics of Studies Included in the Meta-Analysis
First author Year N Effect size (r) Dependent variable
Ahmend 2007 517 .40 Forgiveness
Allan 2006 134 .10 Positive Perceptions
Anderson 2006 184 .00 Positive Emotions
Azar 2001 96 .73 Forgiveness
Azar 1999 48 .66 Forgiveness
Bachman 2006 263 .40 Forgiveness
Bhandari 2012 80 .21 Purchase Intentions
80 .57 Purchase Intentions
Bolkan 2009 134 .13 Positive Emotions
134 .18 Purchase Intentions
Bono 2005 32 .38 Positive Emotions
32 .00 Positive Emotions
241 .35 Positive Emotions
241 .11 Forgiveness
149 .33 Positive Emotions
149 .25 Forgiveness
Bono 2011 115 .13 Forgiveness
112 .17 Forgiveness
Bornstein 2002 173 .04 Legal Sentencing
173 .26 Positive Perceptions
83
128 .18 Legal Sentencing
Braaten 1993 51 .43 Forgiveness
Bradford 1995 85 .30 Positive Emotions
Brown 2005 200 .37 Forgiveness
Brown 2008 80 .58 Forgiveness
80 .63 Positive Emotions
Coombs 2007 167 .07 Positive Perceptions
167 .14 Positive Emotion
Coombs 2008 83 .38 Positive Perceptions
83 .00 Positive Emotions
Dardis 2008 189 .05 Positive Perceptions
Dawar 2000 90 .42 Positive Perceptions
164 .30 Positive Perceptions
Day 2010 518 .16 Legal Sentencing
512 .18 Legal Sentencing
Dean 2005 107 .36 Positive Perceptions
291 .05 Positive Perceptions
DeCremer 2008 128 .81 Positive Perceptions
119 .33 Positive Emotions
119 .36 Positive Emotions
DeCremer 2010 36 .27 Positive Perceptions
Dijke 2011 95 .08 Forgiveness
128 .13 Forgiveness
84
Dirks 2011 143 .43 Positive Perceptions
60 .26 Positive Perceptions
Dunn 2000 709 .02 Positive Perceptions
Eaton 2006 107 .54 Forgiveness
107 .34 Positive Emotions
Eaton 2006 67 .22 Forgiveness
67 .27 Positive Perceptions
62 .15 Forgiveness
62 -.08 Positive Perceptions
Eaton 2007 79 -.29 Forgiveness
Exline 2004 276 .42 Forgiveness
Exline 2007 86 .32 Forgiveness
Exline 2008 218 .53 Forgiveness
218 .35 Positive Emotions
Ferrin 2007 67 -.12 Positive Perceptions
67 -.03 Other/Hiring Decision
160 .02 Positive Perceptions
160 -.05 Other/Hiring Decision
Giacalone 1995 87 .03 Forgiveness
Giner-Sorolla 2007 271 .32 Positive Emotions
Girard 1997 236 .58 Forgiveness
Goei 2007 64 .05 Positive Emotions
64 -.01 Forgiveness
85
186 .03 Positive Emotions
186 .11 Forgiveness
Gold 2000 112 .39 Positive Perceptions
112 .65 Positive Emotions
112 .29 Forgiveness
Goodwin 1992 285 .00 Positive Perceptions
285 .00 Positive Emotions
Green 2011 65 .35 Forgiveness
65 .27 Forgiveness
Gunderson 2008 196 .67 Forgiveness
Haigh 2010 170 .11 Positive Attributions
Harrell 1980 48 .58 Forgiveness
Hodgins 2003 64 .48 Forgiveness
Howley 2009 1652 .58 Positive Emotions
Huang 2006 233 .06 Positive Perceptions
456 -.01 Positive Perceptions
350 .12 Positive Perceptions
177 .43 Positive Perceptions
Huang 2010 278 .26 Positive Perceptions
Kelley 2005 186 .17 Forgiveness
Kim 2009 207 .24 Forgiveness
Kim 2004 200 -.01 Positive Perceptions
200 -.09 Other/Hiring Decision
86
116 .02 Positive Perceptions
116 .03 Other/Hiring Decision
Kleinke 2001 98 .29 Positive Perceptions
184 .25 Positive Perceptions
184 .13 Legal Sentencing
Leonard 2011 95 .25 Positive Emotions
95 .15 Positive Perceptions
95 .22 Forgiveness
Liao 2007 658 .37 Purchase Intentions
658 .53 Positive Emotions
395 .07 Purchase Intentions
395 .12 Positive Emotions
Luzombe 2009 98 .15 Forgiveness
Lyon 2004 80 .37 Purchase Intentions
80 .46 Positive Perceptions
McCullough 2010 125 .09 Forgiveness
372 .11 Forgiveness
McCullough 1998 187 .56 Positive Emotions
187 .42 Forgiveness
McCullough 1997 239 .45 Forgiveness
Morse 2011 360 .24 Forgiveness
Mullet 2007 224 .97 Forgiveness
258 .74 Forgiveness
87
Niedermeier 2001 362 -.05 Legal Sentencing
264 .27 Legal Sentencing
Ohbuchi 1989 58 .38 Positive Perceptions
58 .25 Positive Emotions
58 .38 Forgiveness
80 .85 Positive Perceptions
80 .71 Positive Emotions
80 .36 Forgiveness
Orleans 1984 60 .24 Forgiveness
60 .37 Positive Perceptions
Pace 2010 264 .20 Positive Perceptions
264 .37 Positive Emotions
Philpot 2008 60 .03 Positive Emotions
60 .12 Forgiveness
60 .40 Positive Emotions
73 .15 Forgiveness
73 .58 Positive Emotions
214 .07 Forgiveness
214 .54 Positive Emotions
79 .18 Forgiveness
79 .10 Positive Emotions
79 .17 Forgiveness
79 .09 Positive Emotions
88
79 .35 Forgiveness
79 .47 Positive Emotions
79 .54 Forgiveness
79 .36 Positive Emotions
Philpot 2010 116 .24 Forgiveness
76 .14 Forgiveness
Pipes 1999 149 .17 Forgiveness
Proeve 2006 77 .02 Legal Sentencing
Risen 2007 43 .22 Positive Perceptions
42 .20 Positive Perceptions
41 .05 Positive Perceptions
37 -.51 Positive Perceptions
24 .44 Positive Perceptions
25 .44 Positive Perceptions
22 .51 Positive Perceptions
23 .00 Positive Perceptions
Robbennolt 2006 556 .20 Forgiveness
556 .09 Legal Sentencing
Robbennolt 2008 190 .00 Forgiveness
190 -.15 Legal Sentencing
Rumsey 1976 45 .37 Legal Sentencing
Santelli 2009 28 .55 Forgiveness
28 .17 Forgiveness
89
28 .26 Forgiveness
28 .27 Forgiveness
Schmitt 2010 480 -.15 Positive Perceptions
480 .41 Positive Emotions
Schweitzer 2006 132 .11 Positive Emotions
Sigal 1988 73 -.26 Other/Voting
Struthers 2010 168 .46 Forgiveness
250 .13 Forgiveness
250 .43 Positive Emotions
90 .17 Forgiveness
Struthers 2008 177 .50 Positive Perceptions
177 .39 Forgiveness
Tabak 2012 100 .19 Positive Perceptions
100 .84 Forgiveness
Taylor 1992 320 .00 Legal Sentencing
Thomas 2008 60 .35 Positive Emotions
Tomlinson 2004 90 .07 Forgiveness
Tucker 2006 94 .47 Positive Perceptions
50 .30 Positive Perceptions
224 .27 Positive Perceptions
van Laer 2010 76 -.39 Positive Perceptions
76 -.27 Purchase Intentions
76 .47 Positive Perceptions
90
76 .39 Purchase Intentions
145 .03 Positive Perceptions
Vassilikopoulous 2008 240 .39 Positive Perceptions
240 .39 Purchase Intentions
Vinsonneau 2001 203 .82 Forgiveness
Weiner 1991 125 .45 Positive Perceptions
125 .29 Forgiveness
87 .20 Forgiveness
Wenzel 2010 88 .18 Forgiveness
88 .06 Positive Perceptions
Whited 2010 79 .18 Positive Emotions
79 .00 Forgiveness
Wirtz 2003 187 .47 Positive Emotions
Wooten 2009 200 .23 Forgiveness
200 .20 Positive Perceptions
162 .26 Forgiveness
162 .16 Positive Perceptions
216 .28 Forgiveness
Xie Peng 2009 220 .57 Positive Perceptions
220 .53 Forgiveness
Younger 2004 103 .17 Forgiveness
Zechmeister 2004 56 .20 Forgiveness
56 -.24 Forgiveness
91
113 -.25 Positive Perceptions
N = 94 papers
92
Table 2
Results for Meta-Analyses Examining Efficacy of Apologies
Mean r
Outcome k w uw Range Z fixed Z random Heterogeneity File drawer
Combined 144 .22 .27 -.51-.97 46.83*** 10.72*** 2300.71*** 24177
Forgiveness 79 .41 .32 -.29-.97 38.576*** 8.155*** 1427.81*** 25193
Positive Attributes 60 .19 .24 -.51-.85 17.796*** 6.689*** 599.474*** 5647
Positive Emotions 43 .39 .33 .00-.71 37.741*** 9.414*** 422.347*** 98
Legal Sentencing 11 .11 .13 -.05-.37 6.297*** 3.493*** 29.897** 98
Purchase Intentions 10 .24 .23 -.27-.57 10.815*** 2.848** 97.965*** 222
Note. k effect sizes= the total number of samples utilized to estimate a given effect; Mean r= the unweighted mean of correlations for a
given correlate; Weighted mean r= the weighted mean of correlations for a given correlate; Z-combined= test of whether the mean r is
greater than zero; Heterogeneity=whether effects vary more than expected by chance; Fail-safe k= the number of studies needed to
shift the demonstrated effect to include zero.
93
Table 3
Correlations Among Apology Components and Effect Size
Component Severity Remorse Ack Compensation Effect Size
Severity --------- .35** -.06 .30** .11
Remorse --------- --------- .33** .28* .24*
Ack --------- --------- --------- .32** -.08
Compensation --------- --------- --------- --------- .21+
Note. Ack = acknowledgment of violated rules and norms
k = 77
+p < .10, * p < .05, **p < .01
94
Table 4
Interaction Analysis for Remorse, Acknowledgment of Violated Rules and Norms, and
Compensation for High Social Distance Relationships
Apology component as predictor of effect size β
Remorse .66*
Acknowledgment of Violated Rules and Norms -.52
Compensation .03
Remorse x Acknowledgment of Violated Rules and Norms .37
Remorse x Compensation .04
Acknowledgment of Violated Rules and Norms x Compensation -.36
Remorse x Acknowledgment of Violated Rules and Norms x Compensation .38
Note. R2 = .30, F(7, 23) = 3.410, p = .012
k = 37
+p < .10, *p < .05, **p < .01
95
Table 5
Inter-Rater Reliability Indices and Inter-Correlations for Apology Components
Component Alpha Mean SD 1 2 3 4 5
1. Remorse .77 2.76 1.94 --- .05 .56** .63** .70**
2. Compensation .86 0.58 1.42 --- --- -.00 -.09 -.18
3. Acknowledgment .54 4.65 1.39 --- --- --- .65** .57**
4. Guilt .65 3.29 1.71 --- --- --- --- .66**
5. Shame .67 1.47 1.50 --- --- --- --- ---
Note. n = 37
*p < .05, **p < .01
96
Table 6
Phase 1 Apologizer Self-Rated Emotion Descriptive Statistics and Factor Loadings
Factor
Emotion Mean SD 1 2 3 4 5
Shame 3.30 1.02 .80 .00 -.19 .12 -.01
Guilt 4.00 1.08 .78 .05 -.17 .28 .15
Love 2.54 1.41 .73 .11 .21 .01 .05
Sadness 2.92 1.12 .72 .05 .11 .32 .06
Worry 2.51 1.24 .67 .39 .33 .51 -.08
Fear 2.03 1.19 .59 .46 .21 .15 -.25
Embarrassment 2.89 1.31 .57 .13 .21 .34 .02
Excitement 1.51 0.96 .20 .85 .42 .14 .08
Anger 1.49 0.73 .08 .73 .32 .42 -.19
Surprise 1.38 0.68 .09 .72 .57 .32 .08
Relaxation 1.64 0.99 .12 .71 .15 .17 .36
Inspiration 1.78 1.03 .25 .63 .39 .23 .38
Happiness 1.54 0.93 -.11 .61 .26 -.04 .52
Jealousy 1.27 0.73 .04 .47 .85 -.02 .17
Contempt 1.78 0.92 -.02 .42 .73 .39 .00
Disgust 1.49 0.77 .39 .15 .72 .20 -.02
Amusement 2.08 1.26 -.27 .33 .54 -.01 .26
Pity 1.89 0.97 .11 .28 .17 .83 -.11
Empathy 2.73 1.31 .44 .10 .04 .79 .23
97
Compassion 3.32 1.16 .24 .18 .21 .69 .48
Contentment 2.16 1.04 -.07 .14 .07 -.05 .81
Hope 2.59 1.36 .26 .16 .21 .40 .81
Note. Factor 1 = self-consciousness; Factor 2 = apology-reactions; Factor 3 = insincerity;
Factor 4 = empathy; Factor 5 = contentment.
n = 37
98
Table 7
Apologizer Self-Rated Emotion Factor Descriptive Statistics and Inter-correlations
Factor
Emotion factor Mean SD 1 2 3 4 5
1. Self-consciousness 2.88 0.84 --- .37 .12 .39* .15
2. Apology-Reactions 1.55 0.66 --- --- .54** .34* .38*
3. Insincerity 1.66 0.66 --- --- --- .22 .29
4. Empathy 2.65 0.94 --- --- --- --- .46**
5. Contentment 2.38 1.06 --- --- --- --- ---
Note. n = 37
*p < .05, **p < .01
99
Table 8
Apologizer Self-Rated Emotion Factors Correlated with Apology and Transgression-Specific
Questions
Emotion factor
Question Self-Consciousness Apology-Reactions Insincerity Empathy Contentment
Sincere .28+ .22 -.05 .05 -.21
Upset -.23 .14 .19 .00 -.02
Reasonable .28 -.15 .01 -.06 -.11
Severe .20 -.09 -.23 .05 -.22
Note. Sincere refers to the question, “How sincere was the apology that you just provided?”
Upset refers to “How upset was the person who was hurt by you after they found out about your
behavior?” Reasonable refers to “How reasonable was the person that you hurt's reaction to your
behavior?” Severe refers to “How bad/severe do you think the behavior that you apologized for
was?”
n = 37
+p < .10*p < .05, **p < .01
100
Table 9
Correlations among Judge-rated Items
Item Mean SD 1 2 3 4 5 6 7 8 9
1. Commit 1.56 .25 ---- .77** -.23 -.17 -.26 .20 -.37* -.30+ .08
2. Against 1.62 .21 ---- ---- -.14 -.17 -.18 .11 -.28+ -.24 .23
3. Forgive 4.61 .94 ---- ---- ---- .91** .96** -.32+ .87** .86** -.65**
4. Sincere 4.38 1.14 ---- ---- ---- ---- .89** -.00 .94** .93** -.55**
5. Trust 4.14 .90 ---- ---- ---- ---- ---- -.25 .90** .91** -.71**
6. Severe 3.87 .80 ---- ---- ---- ---- ---- ---- -.01 .08 .31
7. Empathy 4.11 .75 ---- ---- ---- ---- ---- ---- ---- .96** -.52**
8. Sympathy 3.37 .93 ---- ---- ---- ---- ---- ---- ---- ---- -.60**
9. Attributions 4.61 .79 ---- ---- ---- ---- ---- ---- ---- ---- ----
Note. n=37
+p<.10, *p<.05, **p<.01
101
Table 10
Coder-rated Items Correlated with Apologizer and Judge-rated Items
Coder-rated items
Remorse Compensation Acknowledge Guilt Shame
Apologizer-rated
1. Self-consciousness .23 .11 .16 .29+ .22
2. Apology-reactions -.06 .02 -.06 .18 .08
3. Insincerity .02 -.16 .12 .11 .10
4. Empathy -.01 .01 .15 .18 .00
5. Contentment -.07 .27 .10 -.01 -.19
Judge-rated
1. Commit -.09 .30+ -.07 .06 -.21
2. Against -.02 .15 .05 .25 -.03
3. Forgive .49** .21 .48** .43** .38*
4. Sincere .62** .17 .51** .54** .46**
5. Trust .46** .19 .43** .42* .35*
6. Severe .15 .05 -.04 .09 .03
102
7. Empathy .53** .16 .44** .48** .39*
8. Sympathy .54** .20 .40* .48** .37*
9. Attributions -.16 -.26 -.13 .02 .04
Note. Commit= Have you ever committed a similar transgression?; Against= Has anyone ever committed this transgression against
you?
n=37
+p<.10, *p<.05, **p<.01
103
Table 11
Apologizer-rated Items Correlated with Judge-rated Items
Apologizer-rated Items
Judge-rated Items Self-Consciousness Apology-reactions Insincerity Empathy Contentment
1. Commit -.26 -.10 .09 -.05 -.12
2. Against -.32 -.03 -.04 -.19 -.19
3. Forgive .33* .05 -.07 .17 .04
4. Sincere .40* -.04 -.13 .19 .07
5. Trust .36* .01 -.11 .09 -.01
6. Severe .13 -.26 -.27 -.11 -.02
7. Empathy .40* -.08 -.26 .03 -.02
8. Sympathy .49** -.07 -.24 .06 -.03
9. Attributions -.15 .12 .13 -.07 .08
Note. n=37
+p<.10, *p<.05, **p<.01
104
Table 12
Apologizer and Coder-Rated Components Correlated with Judge-Rated Outcomes Controlling for Judge-Rated Severity
Component Empathy Sympathy Forgiveness Trust Sincerity
Guilt .39* .37* .41* .37* .46**
Shame .48** .48** .49** .45** .55**
Remorse .54** .54** .57** .52** .63**
Acknowledgment .44** .41* .49** .43** .51**
Compensation .16 .19 .24 .21 .17
Self-Conscious Factor .40* .48** .40* .41* .41*
Note. n=37
+p<.10, *p<.05, **p<.01
105
Table 13
Apologizer-rated Emotions Predicting Judge Reactions
Apologizer-rated emotions betas
Judge reactions Self-consciousness Apology-reactions Insincerity Empathy Content
Empathy .48** -.04 -.29 -.11 .06
Sympathy .57** -.06 -.26 -.10 .02
Attributions -.16 .13 .07 -.10 .08
Forgiveness .31 .01 -.12 .08 -.01
Trust .40* -.01 -.14 -.02 -.02
Sincerity .42* -.11 -.16 .07 .06
Note. n = 37; The significant models included empathy (R2 = .14, F(5,31) = 2.206, p = .079)
and sympathy (R2 = .23, F(5,31) = 3.16, p = .020). The models for attributions (R
2 = .-.09,
F(5,31) = .402, p = .844), forgiveness (R2 = .-.02, F(5,31) = .887, p = .501), trust (R
2 = .022,
F(5,31) = 1.159, p = .351), and sincerity (R2 = .081, F(5,31) = 1.63, p = .181) were
nonsignificant.
+p < .10, *p < .05, **p < .01
106
Table 14
Acknowledgment, Remorse, and Compensation Predicting Judge Reactions
Coder-rated verbal component betas
Judge reactions Acknowledgment Remorse Compensation Severity
Empathy .20 .42* .14 -.08
Sympathy .16 .44* .17 .01
Attributions -.00 -.20 -.27 .36*
Forgiveness .24 .41* .21 -.39**
Trust .20 .39* .19 -.31*
Sincerity .23 .50** .15 -.08
Note. n = 37; This model was significant for all judge-rated outcomes: empathy (R2 = .25,
F(4,32) = 3.98, p = .010), sympathy (R2 = .25, F(4,32) = 4.03, p = .009), attributions (R
2 = .11,
F(4,32) = 2.16, p = .096), forgiveness (R2 = .41, F(4,32) = 7.35, p = .000), trust (R
2 = .30,
F(4,32) = 4.85, p = .004), and sincerity (R2 = .39, F(4,32) = 6.64, p = .001).
+p < .10, *p < .05, **p < .01
107
Table 15
Interaction Analysis for Coder-Rated Remorse, Coder-Rated Guilt-Shame Aggregate, and Apologizer-Rated Self-conscious emotions
Judge reactions Remorse SC GS-Agg Rem Rem SC All R2
x x x
SC GS-Agg GS-Agg
Empathy .40+ .25 .12 -.18 -.08 .10 -.04 .30
Sympathy .42* .44* .07 -.16 -.01 .05 -.14 .35
Attributions -.34 -.28 .27 .14 -.14 .11 .18 -.09
Forgiveness .32 .27 .21 -.12 .03 -.13 -.11 .16
Trust .30 .38+ .18 -.15 .03 -.08 -.19 .15
Sincerity .46* .27 .18 -.22 .08 -.03 -.05 .37
Note. Remorse=coder-rated remorse; SC= apologizer-rated self-conscious emotions; GS-Agg=coder-rated guilt-shame aggregate;
Rem+SC= interaction between coder-rated remorse and apologizer self-conscious emotions; Rem+GS-Agg=interaction between
coder-rated remorse and coder guilt-shame aggregate; SC+ GS-Agg=interaction between apologizer self-conscious emotions and
coder-rated guilt-shame aggregate; All=interaction between coder-rated remorse, coder-rated guilt-shame aggregate, and apologizer
self-conscious emotions; Adjusted R2 reported.
n=37
+p<.10, *p<.05, **p<.01
108
Table 16
Coder Rated Remorse, Coder Rated Guilt-Shame Aggregate, and Apologizer Self Conscious
Emotions Predicting Judge Reactions
Betas(β)
Judge reactions Remorse Guilt-Shame SC R2
Empathy .37+ .13 .28+ .31
Sympathy .41* .07 .38* .38
Attributions -.37 .34 -.15 .01
Forgiveness .34 .14 .22 .23
Trust .32 .12 .26 .23
Sincerity .46* .14 .26+ .42
Note. Remorse = Coder rated remorse; Guilt-Shame = coder rated guilt-shame aggregate; SC =
Apologizer rated self-conscious emotions; Adjusted R2 reported.
+p < .10, *p < .05, **p < .01
109
110
Figure 2
Mediation Analysis Diagram
Note. Judge empathic response as a mediator of judge-rated outcomes. Coefficients in
parentheses indicate zero-order correlations. Coefficients not in parentheses represent parameter
estimates for a recursive path model including both predictors. Double asterisks (**) indicate
parameter estimates or correlations that differ from zero at p < .01.
111
Appendix A
List of Dependent Variables and Their Categorizations in Study 1
1. Forgiveness
a. Forgiveness
b. Less punishment (outside of a legal context)
c. Less aggression
d. Less revenge/less retaliation
e. Positive ratings of future relationship closeness/friendship
f. Trust repair
g. Cooperation
h. Reconciliation
i. Less avoidance
j. Benevolence
k. Compliance
2. Positive Attributions
a. Honesty
b. Trustworthiness
c. Positive press
d. Competence
e. Integrity
f. Positive regard
g. Positive leadership (i.e., rating the apologizer as a good leader)
h. Positive character ratings
i. Positive word-of-mouth
j. Morality
k. Socially responsible
l. Credibility
m. Ethical
n. Favorable
3. Positive Emotions
a. Positive emotions/positive emotional reaction
b. Satisfaction
c. Compassion
d. Empathy
e. Concern
f. Less begrudging emotions
g. Liking
h. Less anger
i. Sympathy
j. Gratitude
k. Indebtedness
l. Less unpleasant affect
112
m. Trust
n. Less hostility
o. Respect
4. Legal Sentencing
a. Less punishment
b. Shorter prison sentence
c. Less harsh legal settlement
d. Less likely to be found guilty
5. Purchase Intentions
a. Purchase intentions
b. Intention to switch hotels (reverse coded)
c. Repurchase intentions
d. Intent to invest in company
e. Intention to switch doctors (reverse coded)
6. Other
a. Hiring decisions
b. Voting intentions
113
Appendix B
Transcripts and Coder Ratings of Three Apologies from Study 2
Participant 26:
“I'm sorry that the last time we spoke we were not very kind to each other. I wish I had taken
more time to really figure out the best way to say that you've changed. You've changed in a way
that you, you let your ego get in the way. And you could not relate to the problems that are
going on in my life. For a whole year you, you've never, you've never asked me how I was
handling my problems at home. You never did. And I tried to tell you that you were just not the
same, but I blew up on you. And that, I was having a bad time I had just, it was, I was too proud,
I don't know. I couldn't tell you what was going on without actually screaming it to your face,
and you should have known. But, I'm sorry that it had to end this way, that our last words
together will never be that we are just gonna agree to disagree. I'll, I'll always love you, but right
now I don't think that we're the same people, and as much as it hurts me that we aren't speaking
anymore, I think it's for the best in which we need to just take time to do ourselves.”
**This apology received the following coder-rated scores: Remorse M=1.67; Acknowledgment
of violated rules and norms M=2.33; Compensation M=0
Participant #32:
“Hi (name removed to protect privacy). Umm I want to apologize for all the times that I was not
there for you growing up. I know that being your older brother and being the second, sibling in
our family I didn't treat you as well as I should have. I know that growing up I always wanted to
be like Andrew because he was the oldest one and I thought he was super cool and just the best
person that I could think of to want to be. And I know growing up you were just doing what I
was doing. Umm looking up to me, hoping for some sort of role model, some sort of someone to
lean on. Umm, and I was so focused on trying to be like (name removed to protect privacy), I
sort of pushed you to the side and didn't really give you the, I guess uhh the love and the role
model guidance that you needed or wanted growing up. And, you know, you are an amazing
person. Now and today you were umm pre-med, you're doing everything that you've always
wanted to do, even without my help. Umm, but I know that if I was there for you more, umm
maybe you would have had a lot more confidence in yourself, a lot more strength, not that you're
not a strong person, and I just want to say I'm sorry for not being there for you growing up. You
are a beautiful, amazing woman, and I am so proud of being your brother, and I know that now
that we're a little bit older I am trying even harder to create and strengthen our relationship
because I feel like you've had ten years of me being not the best big brother to you, and I want
you to know that I will always be there for you. And, yeah, so, I love you, bye.”
**This apology received the following coder-rated scores: Remorse M=4.67; Acknowledgment
of violated rules and norms M=6.00; Compensation M=0
Participant #34
“I'm really sorry for hooking up with someone else on Valentine’s day, and, even though it
wasn't my fault and he went in for it and I was not into it and shoved him away, and like it wasn't
mutual, we shouldn't have been that close to each other in the first place, and I'm sorry that it
ever happened because I didn't enjoy it and it shouldn't have happened.”
114
** This apology received the following coder-rated scores: Remorse M=1.67; Acknowledgment
of violated rules and norms M=4.33; Compensation M=0