ArticlePDF Available

Better Late Than Early: The Influence of Timing on Apology Effectiveness


Abstract and Figures

Two studies examined whether the timing of an apology influences its effectiveness. We hypothesized that victims who received apologies later in a conflict would feel more satisfied with the resolution of the conflict, primarily because they would have more opportunity for self-expression and would feel better understood. Undergraduates provided retrospective interpersonal conflict narratives (Study 1) and responded to a hypothetical scenario (Study 2) in which they were wronged. The results showed that later apologies were more effective than earlier ones, and that this effect was mediated by feeling heard and understood. The ramifications for creating a “ripeness” or readiness for conflict resolution are discussed.
Content may be subject to copyright.
Journal of Experimental Social Psychology 41 (2005) 201–207
0022-1031/$ - see front matter 2004 Elsevier Inc. All rights reserved.
Better late than early: The inXuence of timing on apology
Cynthia McPherson Frantza,¤,1, Courtney Bennigsonb
a Amherst College, United States
b Williams College, United States
Received 27 September 2001; revised 18 June 2004
Available online 18 November 2004
Two studies examined whether the timing of an apology inXuences its eVectiveness. We hypothesized that victims who received
apologies later in a conXict would feel more satisWed with the resolution of the conXict, primarily because they would have more
opportunity for self-expression and would feel better understood. Undergraduates provided retrospective interpersonal conXict nar-
ratives (Study 1) and responded to a hypothetical scenario (Study 2) in which they were wronged. The results showed that later apol-
ogies were more eVective than earlier ones, and that this eVect was mediated by feeling heard and understood. The ramiWcations for
creating a “ripeness” or readiness for conXict resolution are discussed.
2004 Elsevier Inc. All rights reserved.
Apologies are the world’s most basic and pervasive
conXict resolution technique. Among marriage partners,
friends, business associates, and nations, apologies
attempt to right wrongs and wipe the slate clean, thus
serving a crucial social lubrication role. In a sense, apol-
ogies are magical—they transform an act from some-
thing oVensive into something acceptable (GoVman,
1971). In his classic work, Relations in Public, GoVman
argued that through an apology “ƒ an individual splits
himself into two parts, the part that is guilty of an
oVense and the part that dissociates itself from the [tran-
gression] and aYrms a belief in the oVended rule.” (GoV-
man, 1971, p. 113). The oVender thus admits blame, but
at the same time shows that he or she may be worthy of a
second chance.
Research supports the notion that apologies achieve
this transformation and thereby reduce or resolve inter-
personal conXict. For example, Ohbuchi, Kameda, and
Agarie (1989) found that victims had better impressions
of an oVender, felt less angry, and were less aggressive
when they received an apology. And McCullough et al.
(1998) found that apologies were associated with greater
empathy for an oVender, less avoidance and revenge
among victims, and greater closeness between and
oVender and a victim. In the business domain, Goodwin
and Ross (1992) found that apologies from companies
enhanced consumers’ satisfaction and the perceived fair-
ness of responses to service failures.
But as we know from personal experience, all apolo-
gies are not created equal. What distinguishes a good
apology from a poor one? Advice on giving apologies
can be found in a wide range of sources, including eti-
quette manuals (Sugimoto, 1998), scholarly work on
therapeutic practice (e.g., Mitchell, 1989) and practical
guides on mediation and conXict resolution (e.g., Rubin,
Pruitt, & Kim, 1994; Schneider, 2000). Two common
issues that emerge from this work are timing and sincer-
ity. For example, Rubin et al. (1994, p. 165) noted that
The authors thank Molly Burnett, Sarah Zilzer, Dick Moreland,
and an anonymous reviewer for their helpful comments on earlier
drafts of this article.
*Corresponding author.
E-mail address: (C.M. Frantz).
1Present address: Severance Hall, Department of Psychology,
Oberlin College, Oberlin, OH 44074, United States.
202 C.M. Frantz, C. Bennigson / Journal of Experimental Social Psychology 41 (2005) 201–207
“ƒif improperly timed or meant insincerely, an apology
can arouse suspicion on the part of Other.” Similarly,
Mitchell (1989, p. 285) observed, “If an abbreviated
apology is given immediately following a transgression,
it may be perceived as superWcial and insincere.” Both of
these statements suggest that although what someone
says is important, how and when it is said will also inXu-
ence an apology’s eVectiveness.
Given the importance of apologies, it is not surprising
that so many people have tried to analyze what makes
an apology eVective. What is surprising, however, is how
little empirical work underlies these analyses. Most
research on apologies has simply demonstrated their
general eVectiveness. A few studies have examined char-
acteristics of an oVense (e.g., oVense severity or oVender
responsibility, Bennet & Earwaker, 1994; Itoi, Ohbuchi,
& Fukuno, 1996) that can inXuence an apology’s eVec-
tiveness. However, most oVenders cannot really alter the
facts of a transgression they have already committed,
and so they would probably be more interested in know-
ing what to say, and when and how to say it, once the
deed is done. Scher and Darley (1997) and Darby and
Schlenker (1989) carried out the only published studies
of what makes an apology more or less eVective. Both
studies showed that oVenders’ expressions of remorse
made apologies more eVective. Scher and Darley (1997)
also found that expressions of responsibility, promises of
forbearance, and oVers of reparation were helpful. Pre-
sumably, these variables all serve to make apologies
seem more sincere.
Meanwhile, the role of timing in apology eVectiveness
has been ignored. Our research seeks to Wll this gap.
Research on timing as a factor in the eVectiveness of
apologies might have practical beneWts, if it helped
oVenders to better heal the wounds they have caused.
But it also has potential theoretical implications: Apolo-
gies are a way for one person to produce emotional
change in another, a phenomenon that is fundamental to
many interpersonal processes, including psychotherapy,
conXict resolution, and social support. Understanding
this phenomenon may thus clarify how negative emo-
tions can be managed more productively.
What is the proper timing of an apology? To answer
this question, we drew from work on conXict resolution.
An apology is an attempt by one party (the oVender) to
de-escalate a conXict by creating a change in the victim
(e.g., more forgiveness, less anger). People who study
conXict resolution use the term “ripeness” to describe
someone’s readiness for conXict de-escalation (see Cole-
man, 1997; for a review and synthesis).
Thus, a victim’s ripeness, or readiness to receive an
apology, may be a key determinant of its eVectiveness.
An apology that is oVered too early during conXict may
pressure someone who is not ready for de-escalation.
Furthermore, once an apology has been oVered, the vic-
tim faces certain social constraints: We are supposed to
forgive and forget, whether we are ready to or not (Ben-
net & Dewberry, 1994). This tension between the exter-
nal social constraint and the victim’s internal state may
further limit the apology’s eVectiveness. As a colleague
of ours once said to someone who oVered an apology
prematurely, “Don’t apologize. I’m not done being mad
at you.” Ripeness takes time. Thus, one major hypothe-
sis of our research is that apologies are more eVective
when they come later in a conXict.
Coleman (1997) argued that many aVective, cognitive,
behavioral, and environmental factors inXuence whether
someone is ripe for conXict resolution. We chose to
examine the roles of voice and understanding in creating
such readiness. Voice and understanding are associated
with positive outcomes in a variety of contexts. For
example, Lind, Kanfer, and Earley (1990) found that
voice (the opportunity to express one’s views) makes
decisions seem more fair, even when someone’s views
have had no impact on the decision-making process.
Similarly, Frantz and JanoV-Bulman (1999) found that
when each party in an interpersonal conXict believes that
the other party understands his or her point of view,
both parties view one another more favorably.
GoVman’s observation about how apologies work
helps to explain why voice and understanding might be
important. As he noted, apologies split the oVender in
two, and the transgression seems more acceptable if the
victim believes in this split. Feeling heard and under-
stood increases a victim’s faith in the “good” part of the
oVender, by helping the victim to believe that the
oVender knows what he or she did wrong, and why it
was hurtful to the victim.
Thus, victims may be more ready to accept an apol-
ogy when they have a chance to express themselves and
feel understood Wrst. Apologies that are oVered too
quickly may not be eVective, in part because the victim
still feels unheard, and is not convinced that the
oVender knows what he or she did wrong, or why it was
hurtful, or how hurtful it was. We therefore hypothe-
sized that apologies that occur later in a conXict are
more eVective, primarily because victims are more likely
to have expressed themselves, and thus feel better
understood. These cognitive changes lead to a state of
For obvious ethical reasons, experimental research on
the role of apologies in meaningful interpersonal con-
Xicts is diYcult. Ohbuchi et al. (1989) actually committed
a transgression against their participants in a laboratory
experiment on apologies, but most other researchers
have used hypothetical scenarios and role-playing (e.g.,
Scher & Darley, 1997) or retrospective accounts of trans-
gressions from participants’ lives (e.g., McCullough et
al., 1998). Because each of these methodologies has its
own strengths and weaknesses, both personal narratives
and hypothetical scenarios were used in our research. In
Study 1 we focused on meaningful real-life conXicts that
C.M. Frantz, C. Bennigson / Journal of Experimental Social Psychology 41 (2005) 201–207 203
have high ecological validity and considerable emotional
force. Study 2 focused on a conXict described in a hypo-
thetical scenario. Its purpose was to replicate the main
Wndings of Study 1, and establish causality, through an
experimental design.
In both studies, we predicted that the later an apol-
ogy came in a conXict scene, the more eVective it would
be. We also predicted that later apologies would be
associated with greater feelings of voice and under-
standing, and that these eVects would mediate the rela-
tionship between apology timing and outcome
Study 1
Twenty-four volunteer undergraduates at Williams
College (7 males, 16 females) were recruited to partici-
pate in a study on “Apology and ConXict.” They
received a candy bar for their participation.
After providing their informed consent, participants
received a questionnaire that contained all the instruc-
tions. First, they were asked to describe on a blank page
“a recent conXict (within the last six months) you have
had with another individual ƒ in which you felt you
were wronged, and also in which the other gave you an
apology of some kind.” After writing this description,
participants reported how long ago the conXict
occurred. Next, they read a list of typical, everyday con-
Xict events (see Appendix A). This list was generated by
a diVerent sample of 15 undergraduates, who were
asked to name events that “typically happen in a con-
Xict.” Participants ranked these events in the order that
they occurred during the personal conXict that they
wrote about previously (putting a 0 next to events that
never occurred).
The list included several key events. The rank for the
item “Other apologized to me” served as our measure
of when an apology was received during each partici-
pant’s own conXict. Ranks varied from 1 to 19, with
higher numbers indicating that the apology occurred
later in the conXict. The items “I believed Other under-
stood my feelings and point of view,” “Other showed
non-verbally that s/he understood me,” “Other said s/
he understood my side,” “Other asked questions to
understand what I was saying,” and “I recounted my
grievances and stated my point of view” were used to
create an index of how much voice and understanding
each participant experienced. Each item was scored as
either occurring (1) or not (0). These scores were
summed to create an index, which varied from 0 to 5.
Higher scores indicated that the victim felt heard and
Respondents then went on to answer questions
about how they felt now about the conXict they
described. They rated how angry they remained, how
well they felt the conXict had been resolved, and how
much they had forgiven the oVender, using 10 point
scales (0 Dnot at all to 10 Dcompletely). Negative items
were reverse coded. These ratings were averaged into
an outcome measure of present satisfaction with the
conXict’s resolution (D.89), which served as our
dependent variable.
Finally, the participants were debriefed, thanked for
their help, and dismissed. No one seemed to have
guessed our hypothesis.
All the conXict descriptions were read and coded by
two people who were blind to our hypotheses. They
rated the severity of each transgression on a seven-point
scale (1 Dtrivial to 7Dextremely severe). Coders were
asked to focus on the initial transgression, not on how
the victim or oVender behaved afterwards. The ratings
from both coders were averaged together to create a
measure of conXict severity (D.71).
Preliminary analyses revealed no signiWcant gender
eVects of any sort, so that variable will not be discussed
further. Means and standard deviations for the other
variables are shown in Table 1. Table 1 also shows corre-
lations that are key to testing our main hypotheses.
Spearman’s rho, a non-parametric measure, was used to
calculate correlations involving apology timing, which
was an ordinal variable. We Wrst tested to see how con-
Xict severity related to apology timing, voice and under-
standing, and outcome satisfaction. ConXict severity
correlated negatively with outcome satisfaction, but did
not correlate with the other variables. This suggested
2Whether voice and feeling understood occur before or after an
apology is important. We assumed that they occurred before the apol-
ogy, and in fact this was true for all but one of our participants, who
was dropped from the analyses.
Study 1: Means, standard deviations, and correlations between inde-
pendent, dependent, and mediating variables
*p< .07.
** p< .05.
Measure (ND23) Mean SD Correlations
12 3 4
1. Outcome satisfaction 7.10 2.34 .38*.48** ¡.45**
2. Timing of apology 6.74 3.02 .69** .20
3. Voice/understanding 0.61 .36 .11
4. Severity of conXict 3.08 1.13
204 C.M. Frantz, C. Bennigson / Journal of Experimental Social Psychology 41 (2005) 201–207
that it was not necessary to include conXict severity as a
factor in tests of our hypotheses.
Our Wrst hypothesis, that later apologies would be
associated with more outcome satisfaction, was sup-
ported. Apology timing was positively correlated with
outcome satisfaction; when the apology came later in
the conXict, participants reported greater satisfaction.
Our second hypothesis, that voice and understanding
mediated this eVect, was also supported. As Baron and
Kenny (1986) noted, mediation occurs when a variable
(voice and understanding) accounts for the relation-
ship between the predictor (apology timing) and the
dependent variable (outcome satisfaction). Because
apology timing was an ordinal variable, we chose to
test for mediation using a series of spearman’s rho cor-
relations and partial correlations, rather than the mul-
tiple regression procedure recommended by Baron and
Kenny. Table 1 shows that apology timing correlated
signiWcantly with voice and understanding; later apolo-
gies were associated with higher levels of voice and
understanding. Voice and understanding correlated
signiWcantly with outcome satisfaction; higher levels of
voice and understanding were associated with higher
levels of outcome satisfaction. Finally, when voice and
understanding were held constant, the correlation
between outcome satisfaction and apology timing
disappeared, rD.06. Thus, the relationship between
apology timing and outcome satisfaction can be
explained by the mediating eVect of voice and under-
Study 2
The results of Study 1 support our hypotheses.
Because our results were correlational, however, the
causal relations among the variables that we measured
were unclear. For example, people who express them-
selves more often during a conXict may receive later
apologies because they have annoyed the oVender, or
because all their talking kept the oVender from apologiz-
ing earlier. If so, then voice aVects the timing of apolo-
gies, rather than the other way around.
Because of these concerns, we ran a second study
with a larger sample and an experimental methodology.
In Study 2, participants read a hypothetical scenario in
which they all experienced voice and understanding, but
received apologies at diVerent times. Some participants
received an apology before feeling understood, some
after feeling understood, and some never received an
apology. This allowed us to test whether receiving an
apology after voice and understanding indeed produces
greater outcome satisfaction. We also wanted to
compare those who received an apology to those who
did not, to see whether any apology is better than none
at all.
In a single mass-testing situation, 83 undergraduates
(35 males and 47 females) at Amherst College received
extra course credit for participating in a study on “reac-
tions to being wronged.” They were told when they were
recruited that the study involved responding to hypo-
thetical conXict scenarios.
After providing their informed consent, participants
were asked to read about a hypothetical situation and
urged to “try to imagine the described situations as
clearly as possible, and answer all questions based on
how YOU would feel if the same thing happened to
you.” All participants then read the same description of
an event:
One Friday night, you were waiting in your dorm room
for your friend Chris to get back from his baseball game.
The two of you had made plans earlier in the week to
rent a movie you had heard was really good, and later go
to a party on campus. Chris was supposed to show up
around 9 when he got back from the game with his team,
but sometimes games ran a little later. At around 8:30
your roommates all left to go to an oV-campus party and
asked you to join them. It sounded like fun to you, but
you knew you could not just break the plans you had
made with Chris. You were ready at 9:00 and ended up
waiting until 10:30, until it was clear that either some-
thing was wrong or he had forgotten. The next day, you
learned from a mutual friend that Chris had been out
partying from 8:30 on. You were angry, so you called
Chris to talk to him.
Participants were given a blank space in which to write
how they would feel, and asked to write “as if this had
actually happened to you.” Then, they rated seven emo-
tion words (angry, understanding, frustrated, forgiving,
satisWed, resentful, and irritated) to indicate how much
they would feel each emotion. These ratings, which were
made on seven-point scales (1DNot at all to 7 DVery),
were averaged together to create a single score that mea-
sured initial responses to the transgression (D.88).
Participants then read one of three diVerent descrip-
tions of the phone conversation that occurred the next
day with Chris. These conversations were identical
except for whether an apology occurred at the very
beginning of the conversation, or after the participant
had (hypothetically) voiced her concerns and received
acknowledgement from Chris that her feelings were
understood. In a third condition, participants received
no apology at all.
In the later apology condition, for example, the par-
ticipants read:
C.M. Frantz, C. Bennigson / Journal of Experimental Social Psychology 41 (2005) 201–207 205
When Chris answered, you said “What happened last
night? We had plans.” You talked for a while, you telling
about the party you eventually went to and Chris telling
about the party he went to. You told Chris that you had
waited for nearly 2 hours because you did not want to
ditch Chris if the game had gone longer than expected.
You told Chris that you had even skipped going to a
party with a bunch of your other friends so that you did
not break your plans with Chris. Besides, you added,
Chris had forgotten other plans you had made in the
past and that just made you more upset about the whole
situation. Chris told you that he could understand why
you were upset and that he himself would be upset in the
same situation. Chris said “Sorry,” and also explained
that he had played horribly that day and his mind was
still on the game when he got back to school. When he
returned, he just wanted to go out and party to get his
mind oV the game, and he completely forgot the plans he
and you had made earlier in the week. Then Chris said,
“I’ve got to run to class. Talk to you later” and hung up.
Participants were given a blank space in which to
write about their reactions to this conversation. Then
they rated the same emotions are before, allowing us to
create a single score measuring participants’ Wnal reac-
tions to the transgression (D.91). Participants were also
asked whether they had received an apology from Chris.
After completing the questionnaire, participants were
debriefed, thanked for their help, and dismissed. No one
seemed to have guessed the hypotheses.
Responses to our question about the apology were
analyzed to verify that participants correctly noticed and
remembered this information from the scenario. Twenty-
three participants, spread evenly across the three condi-
tions, answered the question incorrectly.3 The analyses
reported below do not include their data (although
including them did not substantially change the results).
Table 2 provides the means and standard deviations
for all three conditions at Time 1 and Time 2. To test
whether the timing of apologies aVected changes in reac-
tions to the transgression, responses were submitted to a
2 (gender) £3 (apology condition) £2 (time) mixed-
design ANOVA. There were only three signiWcant
eVects. First, there was a signiWcant gender eVect, F(1,
53) D5.44, p<.001. At both times, women were less posi-
tive (MD2.82, SD D1.04) than men (MD3.53,
SD D1.37), perhaps because the scenario involved a
male friend, which may have connoted being stood up
for a date.
There was also a main eVect of time, F(1, 53)D6.63,
p< .01. Participants were generally more positive at Time
2 (MD3.38), SD D1.64 than at Time 1 (MD2.97,
SD D1.24). However, this was qualiWed by an apology
condition by time interaction, F(2, 53)D6.03, p< .001.
Planned comparisons were conducted to determine
whether the interaction eVect supported our hypotheses.
Changes in reactions from Time 1 to Time 2 diVered sig-
niWcantly across all three conditions. Reactions
improved most in the later apology condition, followed
by the early apology condition, and then by the no apol-
ogy condition (where reactions actually worsened).
Improvement in the late apology condition was signiW-
cantly greater than improvement in the early apology
condition, t(52) D2.41, p<.05, or improvement in the no
apology condition, t(52) D4.50, p< .001. And improve-
ment in the early apology condition was signiWcantly
greater than improvement in the no apology condition,
t(54) D2.18, p< .05, where reactions actually became
slightly more negative. However, participants who
received an apology after voice and understanding were
the only group whose reactions changed signiWcantly
from Time 1 to Time 2, t(25) D4.44, p< .001.
In Study 1, apologies that occurred later in a conXict
were more eVective, in the sense that they were associated
with greater outcome satisfaction. There was also some
evidence that voice and understanding mediate the rela-
tionship between apology timing and apology eVectiveness.
In Study 2, we experimentally manipulated the timing
of apologies, and replicated the results from Study 1. We
found that a later apology, occurring after voice and
understanding, was more eVective than an early apology,
occurring before voice and understanding. This suggests
that the timing of an apology does matter, and that late
is better than early. Not receiving an apology at all, even
when there was voice and understanding, was worse
than receiving an early apology. Thus, even an early
apology is better than no apology—the words “I’m
sorry” have a power of their own, even when circum-
stances are less than ideal.
3Although a 28% manipulation check failure rate may seem high,
Oppenheimer and Davidenko (2002) have documented similar rates in
other large group testing sessions.
Study 2: Change in feelings from Time 1 to Time 2
Note. There were no signiWcant diVerences among conditions in feel-
ings at Time 1.
** p< .001.
Condition Time 1 Time 2 T1 vs. T2
NMean SD Mean SD t
Later apology 26 2.64 0.84 3.64 1.55 4.44**
Early apology 28 3.09 1.35 3.38 1.49 1.51
No apology 28 3.12 1.28 2.85 1.44 1.58
206 C.M. Frantz, C. Bennigson / Journal of Experimental Social Psychology 41 (2005) 201–207
Delayed apologies were thus more eVective in both
studies, probably because the victim had more time to
feel heard and understood. Our results were obtained
with two diVerent methodologies. The high ecological
validity of the real life conXicts used in Study 1 suggests
that the results apply to meaningful conXicts, and the
experimental design used in Study 2 establishes the
causal role of apology timing in outcome satisfaction.
Our research suggests that delaying an apology until
after the victim has a chance to feel heard and under-
stood may be the most eVective way to right wrongs.
Feeling heard and understood apparently fosters ripe-
ness, a readiness to de-escalate conXict.
These results are also consistent with a recent litera-
ture review on the adaptiveness of expressing negative
emotions (Kennedy-Moore & Watson, 2001). Work on
catharsis suggests that expressing anger often strength-
ens, rather than weakens, aggression, and anger (e.g.,
Bushman, 2002). However, other researchers (e.g.,
Pennebaker, 1997) have documented the beneWts of self-
expression. Kennedy-Moore and Watson (2001) resolve
this contradiction in the literature by arguing that
expressing negative feelings is adaptive when it leads to
some form of resolution involving the source of distress.
Giving voice to personal concerns, and eliciting under-
standing from an oVender, should create these condi-
tions. Communicating what bothered you is a way of
reassuring yourself that an oVender understands the rule
that was broken, and eliciting that understanding before
receiving an apology makes you feel more convinced
that the oVender will “do the right thing” in the future.
Before readers rush out and wait to apologize, however,
several cautions are in order. Our research is only a start-
ing point for investigating the factors that make apologies
eVective. More research is needed to be fully conWdent in
our results. For example, the time frames that we exam-
ined were quite limited (within the last 6 months for Study
1, and within 24h for Study 2.) The possible eVects of long-
delayed apologies are not yet known. Recent research sug-
gests that some apologies are too late to be eVective (Zilzer
& Frantz, 2002). Thus, there may be a U-shaped function
relating apology timing and outcome satisfaction—
extremely early and extremely late apologies are likely to
be ineVective. There are probably also occasions when an
apology is ineVective no matter when it occurs.
Another limitation of our research is the way in which
the hypothesized mediating variables were measured.
Voice and understanding were not measured separately.
This choice was inXuenced by practical considerations
rather than theoretical ones; voice and understanding
are clearly separate processes at the theoretical level. The
narrative methodology that we used in Study 1 did not
lend itself to such Wne-grained analysis, because current
feelings of being understood probably strongly inXuence
memories about expressing one’s viewpoint. In Study 2,
too many cells would have been created by systemati-
cally varying the presence and absence of both under-
standing and voice, along with the timing of the apology.
Future research should examine voice and understand-
ing separately. Voice without understanding may be less
beneWcial than voice with understanding, and may in
fact intensify feelings of anger.
We argued that ripeness leads to greater outcome sat-
isfaction, but ripeness is a latent variable that cannot be
directly measured. Many cognitive and emotional fac-
tors are involved in ripeness (Coleman, 1997), and these
can be measured. However, few of them were measured
here. For example, voice and understanding should
strengthen a victim’s belief that an oVender will not com-
mit the same oVense again. Future research should mea-
sure this and other cognitive factors that might be
inXuenced by the timing of an apology.
Emotions are more diYcult to measure, yet they
undoubtedly play an important role in ripeness too. For
example, anger is an emotion with a time course, so ripe-
ness (“I’m done being mad”) may depend in part on
anger subsiding. The methodologies used here did not
allow us to explore this issue. We doubt that people can
accurately remember the time course of their anger dur-
ing past events, and responses to scenarios involve a
hypothetical, rather than an actual time frame.
Many of these limitations could be overcome by
studying ongoing conXicts. This might allow for the
direct exploration of both a victim’s ability to voice his
or her concerns, and an oVender’s ability to convey
understanding of the victim’s viewpoint. The time course
of such emotions as anger could also be measured.
Although we left many questions unanswered, our
research provides an important Wrst step towards under-
standing the factors that aVect an apology’s eVectiveness.
Letting someone Wrst voice his or her concerns, and then
assuring the person that these concerns are understood,
may “ripen” the person for conXict de-escalation. More
broadly, our research suggests that a fundamental Wrst step
toward changing emotions may be to acknowledge their
existence. This is resonant with some psychological theo-
ries (cf. Carl Rogers’ idea of unconditional regard) as well
as many Eastern approaches to life (see Kabat-Zinn, 1994).
Appendix A. List of conXict events ordered by participants
Other interrupted me while I spoke.
I interrupted Other while s/he spoke.
Other asked questions in order to understand what I was
Other showed non-verbally that s/he understood me.
I felt Other owed me an apology.
Other said s/he understood my side.
Other threatened me.
I threatened other.
Other stated his/her point of view.
C.M. Frantz, C. Bennigson / Journal of Experimental Social Psychology 41 (2005) 201–207 207
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator vari-
able distinction in social psychological research: Conceptual, stra-
tegic, and statistical considerations. Journal of Personality and
Social Psychology, 51, 1173–1182.
Bennet, M., & Dewberry, C. (1994). “I’ve said I’m sorry, haven’t I?”: A
study of the identity implications and constraints that apologies
create for their recipients. Current Psychology: Developmental,
Learning, Personality, Social, 13, 10–20.
Bennet, M., & Earwaker, D. (1994). Victims’ responses to apologies:
The eVects of oVender responsibility and oVense severity. Journal of
Social Psychology, 134, 457–464.
Bushman, B. (2002). Does venting anger feed or extinguish the
Xame? Catharsis, rumination, distraction, anger, and aggressive
responding. Personality and Social Psychology Bulletin, 28, 724–
Coleman, P. (1997). RedeWning ripeness: A social-psychological per-
spective. Journal of Peace Psychology, 3, 81–103.
Darby, B. W., & Schlenker, B. R. (1989). Children’s reactions to trans-
gressions: EVects of the actor’s apology, reputation and remorse.
British Journal of Social Psychology, 28, 353–364.
Frantz, C. M., & JanoV-Bulman, R. (1999). Seeing both sides: Lessons
from conXict narratives. Poster presented at the Annual Meeting of
the American Psychological Association, Boston, MA.
GoVman, E. (1971). Relations in public. New York: Basic Books.
Goodwin, C., & Ross, I. (1992). Consumer responses to service failures:
InXuence of procedural and interactional fairness perceptions.
Journal of Business Research, 25, 149–163.
Itoi, R., Ohbuchi, K., & Fukuno, M. (1996). A cross-cultural study of pref-
erence of accounts: Relationship closeness, harm severity, and motives
of account making. Journal of Applied Social Psychology, 26, 913–934.
Kabat-Zinn, J. (1994). Wherever you go, there you are: Mindfulness
meditation in everyday life. New York: Hyperion.
Kennedy-Moore, E., & Watson, J. C. (2001). How and when does emo-
tional expression help?. Review of General Psychology, 5, 187–212.
Lind, E. A., Kanfer, R., & Earley, P. C. (1990). Voice, control, and proce-
dural justice: Instrumental and noninstrumental concerns in fairness
judgments. Journal of Personality and Social Psychology, 59, 952–959.
McCullough, M., Rachal, K., Steven, J., Worthington, E., Brown, S., &
Hight, T. (1998). Interpersonal forgiving in close relationships II:
Theoretical elaboration and measurement. Journal of Personality
and Social Psychology, 75, 1586–1603.
Mitchell, C. E. (1989). EVects of apology on marital and family rela-
tionships. Family Therapy, 16, 282–287.
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, 219–227.
Oppenheimer, D. M., & Davidenko, N., (2002). Instructional Manipu-
lation Checks: How to Maintain Statistical Power with up to 30%
Fewer Subjects. Paper presented at the meeting of the Association
for Psychological Science, New Orleans, LA.
Pennebaker, J. W. (1997). Writing about emotional experiences as a
therapeutic process. Psychological Science, 8, 162–166.
Rubin, J. Z., Pruitt, D. G., & Kim, S. H. (1994). Social conXict: Escala-
tion, stalemate, and settlement (2nd ed.). New York: McGraw-Hill.
Scher, S. J., & Darley, J. M. (1997). How eVective are the things people
say to apologize? EVects of the realization of the apology speech
act. Journal of Psycholinguistic Research, 26, 127–140.
Schneider, C. D. (2000). What it means to be sorry: The power of apol-
ogy in mediation. Mediation Quarterly, 17, 265–280.
Sugimoto, N. (1998). Norms of apology depicted in U.S. American and
Japanese literature on manners and etiquette. International Journal
of Intercultural Relations, 22, 251–276.
Zilzer, S. J., & Frantz, C. M., (2002). The inXuence of timing on apology
eVectiveness: Exploring the phenomenon of a “too late” apology.
Poster presented at the Annual Meeting of the American Psycho-
logical Society, New Orleans, LA.
I recounted my grievances and stated my point of view.
I believed Other understood my feelings and point of
I understood Other’s feelings and point of view.
Other apologized to me.
I apologized to Other.
Other yelled at me.
I yelled at Other.
I said I wanted an apology.
Other insulted me.
I insulted Other.
... Drawing on the literature on apology timing (Frantz and Bennigson 2005;Min et al. 2020), we argue that a sequential order of an employee's listening first and then apologizing, along with acknowledging customers' voiced concerns, will produce favorable customer response if a service failure occurs at the beginning of the service interaction; customers want to ensure that the same failure will not be repeated. The employee's careful listening before 1 We included this particular statement to create a situation where customers are likely to interact with the same service person in the near future. ...
... Additionally, it makes customers feel that the employee fully understood their issues (Frantz and Bennigson 2005), thus increasing the perception that they will be treated well (Lacey, Suh, and Morgan 2007). However, as documented in the active listening literature (Jones, Bodie, and Hughes 2019;Min, Lim, and Magnini 2015), customers are less likely to recognize the employee's perspective if the employee fails to verbally acknowledge their concerns. ...
... Several studies also report that the sequential order of offering an apology, and having customers' concerns understood, matters to the customers (Frantz and Bennigson 2005;Huan 2021;Hubbard et al. 2013). They reveal that when customers are motivated to voice their concerns, it will be more desirable for the employee to apologize after, rather than before, having them voiced and heard. ...
Full-text available
A service employee’s active listening plays a crucial role in restoring a damaged customer relationship. However, previous studies reveal little about how listening to customer complaints operates in recovering a service failure. The purpose of this research is to explore when and why the employee’s active listening has a positive influence on customer response. We define active listening as (1) listening to customers’ concerns before apologizing and (2) verbally acknowledging them. Using scenario-based experiments, we demonstrate that active listening improves customer satisfaction, which in turn increases tip size (Study 1). Moreover, we find that active listening fosters customers’ perceptions of preferential treatment, which lead to greater customer satisfaction (Study 2). Yet, such positive effects of active listening diminish when customers are unexpectedly offered a complimentary service such as a room upgrade. The implications for academic researchers and marketing managers are discussed.
... While on the one hand, delivering an apology as soon as possible had been shown to repair trust. On the other hand, later apologies were preferred to earlier ones because they promoted feelings of being heard and understood (Frantz & Bennigson, 2005;Tomlinson et al., 2004). Studies could therefore examine the differences in restoration intentions based on the temporal order of compensation and non-financial repair tactics. ...
... Studies could therefore examine the differences in restoration intentions based on the temporal order of compensation and non-financial repair tactics. It is known that, in addition to the restoration tactics, the timing and sincerity of the tactics are also important (Frantz & Bennigson, 2005;Haesevoets et al., 2016). ...
While the extant literature has examined causes for buyer-supplier relationship dissolution, the restoration of severed buyer-supplier relationships has been overlooked. Drawing on organizational justice theory, our research develops and tests a model of relationship restoration. We examine how the supplier's restoration tactics - acknowledgment, compensation, and operational transparency, influence the interactional, distributive, and procedural fairness perception, respectively, of the buyer, resulting in relationship restoration. The results are based on a 2 (Acknowledgment -Yes/No) × 2 (Compensation -Yes/No) × 2 (Operational Transparency -Yes/No) vignette-based study with 390 experienced practitioners. The analysis shows that compensating the buyer and providing transparent procedures for dealing with similar situations in the future, lead to higher distributive fairness and procedural fairness, respectively, resulting in restored relationships. Compensation makes up for past supplier malperformance, while operational transparency mitigates future concerns. We also find that restoration tactics based on interactional justice are less effective than those based on procedural and distributive justice. There is only marginal support for the indirect positive effect of acknowledgment on restoration intentions (p<.10). These results point to the importance of knowing how, precisely, to approach a buyer to initiate relationship restoration. Managers must understand and evaluate the specific needs of each buyer when proposing a compensatory design that appeals to the buyer. Additionally, establishing procedures that are appealing to all buyers can be a challenge for a supplier, due to the differing benefits to the supplier provided by each buyer.
... Due to the exploratory nature of this study, and the limited prior research regarding apologies, we decided to code written apologies for several exploratory components that have not been thoroughly studied, but may be particularly relevant for apologies within a parent-child context. These components include: (1) physical affection (e.g., "Can I give you a hug?"); (2) timing (Ebesu Hubbard et al., 2013;Frantz & Bennigson, 2005; e.g., "I'll give you some space, and we can talk more when you're ready"); (3) term of endearment or love (e.g., "I love you, sweetie"); (4) expression of gratitude (You et al., 2020; e.g., "Thank you for putting your bike away"); (5) request for forgiveness (Lewicki et al., 2016; e.g., "Will you forgive me?"); (8) use of a specific emotion word (e.g., "I was feeling frustrated"); and (9) invalidating/minimizing comments (Molinsky, 2016; e.g., "I'm sorry if I hurt your feelings"). We also collected information regarding the length of apology (in number of words). ...
Full-text available
Parenting is a stressful and difficult endeavor, and even the most skillful parents are apt to make mistakes that warrant an apology to their child. However, little is known regarding parents’ overall willingness to apologize to their child or the content of parental apologies when provided. Informed by research on apology outside the parenting context, the current exploratory study aimed to characterize maternal apology attitudes and behavior. A sample of mothers (N = 186) was recruited from the community, and they self-reported their proclivity to apologize to their child, as well as their proclivity to apologize generally. Mothers were also asked to model a skillful parental apology through an open-ended text response to a vignette, and these responses were coded for the presence of effective apology components. Mothers reported a high proclivity to apologize to their child, and this was associated with their willingness to apologize generally. However, the majority of mothers’ written apologies included few of the elements recommended for effective apologies. Similar to other literature on apology, this finding suggests that although mothers may be willing to apologize when needed, knowing and implementing an effective apology is challenging.
... Research suggests that how the transgressor responds to the transgression may have a big role to play. For example, the victim is likely to forgive the transgressor if s/he feels that s/he has been heard and understood (Frantz and Bennigson, 2005), and/or the transgressor appears remorseful, compassionate, and responsible (Darby and Schlenker 1989;Moon and Rhee, 2012;Scher and Darley 1997). ...
Purpose When a product fails out of negligence on the seller’s part, consumers can either retaliate against the seller, more so if a third party encourages them to do so, or forgive the seller should the seller express remorse. This paper aims to examine how the fit between the consumer’s promotion/prevention regulatory orientation and the promotion/prevention frame of a message of contrition (retaliation), such as an apology from a chief executive officer (CEO) (a class action suit threat by a lawyer), affects such forgiveness (retaliation) intentions in the form of product repurchase decisions. Design/methodology/approach In two laboratory experiments, this paper temporally induces a promotion or prevention orientation in the study participants and thereafter ask them to imagine experiencing a product failure and listening to (1) the CEO apologize for the harm (eliciting sympathy/encouraging repurchase); or (2) a lawyer inviting them to seek damages for the harm (eliciting anger/discouraging repurchase). This paper frames the messages from the CEO/lawyer such that they fit either with a promotion mindset or with a prevention mindset. Findings This paper finds that, following a message of apology, a frame-focus fit (compared to a frame-focus misfit) elicits sympathy and encourages repurchase universally across promotion and prevention-oriented consumers. However, following a message encouraging retaliation, the same fit elicits anger and discourages repurchase more among prevention-oriented than promotion-oriented consumers. Originality/value Although past research has investigated how regulatory fit affects forgiveness intentions, this paper fills three research gaps therein by (a) addressing both forgiveness and retaliation intentions, (b) deconstructing the fit-induced “just right feelings” by exploring their underlying emotions of sympathy and anger, and (c) showing that fit effects are not universal across promotion and prevention-oriented consumers. For practice, the results suggest that managers can lessen the fallout from product failures by putting consumers in a promotion mindset that strengthens the effect of a promotion-framed apology and inoculates them against all types of retaliatory messages.
... This is because the recipients of such emotion may react in an inappropriate way and as a result, both parties may not be able to reach settlement point. Several existing literatures (e.g, Lu Wang, Northcraft, & Van Kleef, 2012;Frantz, & Bennigson, 2005;Tiedens, 2001;Lewicki, & Stevenson, 1998;Allred et al, 1997) reported that expression of negative emotion such as anger may result in poorer negotiation outcomes. According to empirical evidence, negotiators who employ anger expression on the negotiation table may succeed for the first time, but they may not succeed when they display similar behaviour on the bargaining table. ...
Anger appears to be a neurocognitive adaptation designed to bargain for better treatment, and is primarily triggered by indications that another individual values the focal individual insufficiently. Once activated, anger orchestrates cognitive, physiological, and behavioral responses geared to incentivize the target individual to place more weight on the welfare of the focal individual. Here, we evaluate the hypothesis that anger works by matching in intensity the various outputs it controls to the magnitude of the current input—the precise degree to which the target appears to undervalue the focal individual. By magnitude-matching outputs to inputs, the anger system balances the competing demands of effectiveness and economy and avoids the dual errors of excessive diffidence and excessive belligerence in bargaining. To test this hypothesis, we measured the degree to which audiences devalue each of 39 negative traits in others, and how individuals would react, for each of those 39 traits, if someone slandered them as possessing those traits. We observed the hypothesized magnitude-matchings. The intensities of the anger feeling and of various motivations of anger (telling the offender to stop, insulting the offender, physically attacking the offender, stopping talking to the offender, and denying help to the offender) vary in proportion to: (i) one another, and (ii) the reputational cost that the slanderer imposes on the slandered (proxied by audience devaluation). These patterns of magnitude-matching were observed both within and between the United States and India. These quantitative findings echo laypeople's folk understanding of anger and suggest that there are cross-cultural regularities in the functional logic and content of anger.
Full-text available
El propósito de esta tesis es explicar el comportamiento del momento de la decisión estratégica desde una aproximación organizacional. Particularmente, desde la Integración Interfuncional presente en el proceso de desarrollo de nuevo producto. La metodología está centrada en la Simulación Basada en Agentes, especialmente en el software NetLogo. Se desarrollan experimentos virtuales identificando el efecto de la interacción de los agentes entre sí y con el entorno. El momento de la decisión, desde esta aproximación, responde a el surgimiento de la decisión de manera emergente a partir de la interacción entre agentes que deben concordar con otros, por medio de interacciones, que en algunos casos ven deconstruida su estrategia deliberada, por causa de la concurrencia de decisiones y acciones. Puntualmente, de los experimentos virtuales se obtiene que la presencia de información previa y la disposición a los encuentros de interacción entre departamentos en los procesos de integración interfuncional, tiene incidencia en la disminución de la cantidad de pasos requeridos en la construcción de trayectos (cuasi-trayectorias tecnológicas) para llegar a un acuerdo, pero no tiene incidencia la cantidad de interacciones necesarias. Como limitación se puede plantear que el desarrollo de la comprensión del momento de la decisión que se realiza dentro de este proceso de investigación, está concentrado en la aproximación organizacional sobre el desarrollo de nuevo producto. Resulta de interés contrastar los resultados con otras aproximaciones organizacionales. Como aporte original se destaca la mayor comprensión sobre la formación de las estrategias mediante el uso de la Simulación Basada en Agentes, en particular sobre el funcionamiento del momento de la decisión estratégica en el contexto organizacional.
Artificial intelligence (AI) failures are increasingly common as more and more companies race to implement AI solutions. The implementation of AI and its inevitable malfunctions are an unprecedented type of crisis for corporate communication professionals. This study reviews (1) 23 instances of AI failures, (2) subsequent corporate communication, and (3) resultant media coverage to investigate the various strategies employed to deal with AI failures. We also identify if these strategies lead to positive or negative responses and/or mitigation of the crisis. Results show that several response strategies included in extant crisis response frameworks can be effective in dealing with AI crises, whereas other strategies tend to be unsuccessful. Our analysis also points to the emergence of a crisis communication strategy that takes advantage of the uncertainty surrounding the accountability of AI to mitigate the crisis.
Full-text available
Emoticons are pictorial/textual depictions of facial expressions used in marketing communications. Little is known about how customers interpret positive or negative emoticons used by customer service employees in service failure contexts. We investigate the impact of emoticon type on customer satisfaction and re-purchasing intention, and examine the sequential mediating role of perceived sincerity and willingness to forgive. Results show that the use of a negative emoticon in a response leads to a higher level of customer satisfaction and re-purchasing intention than responses with a positive emoticon. We further demonstrate that customers perceive that the presence of a negative emoticon in a response is more sincere and generates a higher level of forgiveness than those responses that use positive emoticons, but only when the communal relationship is salient in the customer’s mind. Our findings offer important theoretical and practical implications in service failure contexts.
Full-text available
In this article, we attempt to distinguish between the properties of moderator and mediator variables at a number of levels. First, we seek to make theorists and researchers aware of the importance of not using the terms moderator and mediator interchangeably by carefully elaborating, both conceptually and strategically, the many ways in which moderators and mediators differ. We then go beyond this largely pedagogical function and delineate the conceptual and strategic implications of making use of such distinctions with regard to a wide range of phenomena, including control and stress, attitudes, and personality traits. We also provide a specific compendium of analytic procedures appropriate for making the most effective use of the moderator and mediator distinction, both separately and in terms of a broader causal system that includes both moderators and mediators. (46 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Full-text available
Does distraction or rumination work better to diffuse anger? Catharsis theory predicts that rumination works best, but empir- ical evidence is lacking. In this study, angered participants hit a punching bag and thought about the person who had angered them (rumination group) or thought about becoming physically fit (distraction group). After hitting the punching bag, they reported how angry they felt. Next, they were given the chance to administer loud blasts of noise to the person who had angered them. There also was a no punching bag control group. People in the rumination group felt angrier than did people in the distrac- tion or control groups. People in the rumination group were also most aggressive, followed respectively by people in the distraction and control groups. Rumination increased rather than decreased anger and aggression. Doing nothing at all was more effective than venting anger. These results directly contradict catharsis theory.
Full-text available
179 undergraduates took part in a study of the effects of instrumental and noninstrumental participation on distributive and procedural fairness judgments. In a goal-setting procedure, Ss were allowed voice before the goal was set, after the goal was set, or not at all. Ss received information relevant to the task, irrelevant information, or no information. Both pre- and postdecision voice led to higher fairness judgments than no voice, with predecision voice leading to higher fairness judgments than postdecision voice. Relevant information also increased perceived fairness. Mediation analyses showed that perceptions of control account for some, but not all, of the voice-based enhancement of procedural justice. Results show that both instrumental and noninstrumental concerns are involved in choice effects. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Full-text available
The paradox of distress expression is that expression of negative feelings is both a sign of distress and a possible means of coping with that distress. This article describes research illustrating the paradox of distress expression. It reviews evidence concerning 3 possible mechanisms by which expression might alleviate distress, focusing on the role of expression in (a) reducing distress about distress, (b) facilitating insight, and (c) affecting interpersonal relationships in a desired way. The authors conclude by highlighting the circumstances under which expression is most likely to be adaptive. Overall, the authors argue that expression of negative feelings is adaptive to the extent that it leads to some kind of resolution involving the source or significance of distress. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Because they impose constraints on those who receive them, apologies are seldom rejected (Bennett & Dewberry, in press). It was hypothesized that two variables, degree of offender responsibility and outcome severity, determine whether an apology is rejected. Subjects in Scotland role played the victim of a negative event in which offender responsibility and outcome severity were independently manipulated. The findings provided substantial support for the hypothesis.
For the past decade, an increasing number of studies have demonstrated that when individuals write about emotional experiences, significant physical and mental health improvements follow. The basic paradigm and findings are summarized along with some boundary conditions. Although a reduction in inhibition may contribute to the disclosure phenomenon, changes in basic cognitive and linguistic processes during writing predict better health. Implications for theory and treatment are discussed.
Ripeness is a construct of tremendous importance to both conflict resolution theory and practice. The conceptualizations of this construct in the literature, however, have varied. This article presents a social-psychological model for conceptualizing ripeness that integrates these various perspectives by viewing them as interactive components of a multimodal system that is contained by both change and resistance forces in a state of "unripeness." This model redefines ripeness at the individual level as a commitment to change the direction of the normative social processes of the relations toward deescalation. This lens offers a more specific, yet comprehensive understanding of the construct, as well as suggests avenues for intervention through the creation of ripeness. For illustration, the model is applied to an interpersonal example of conflict and its applicability to larger-system conflicts is discussed.
The importance of apology as acknowledgment of injury is familiar in some forms of mediation, including victim-offender mediation, but it has been much less understood in divorce mediation. The act of apology represents one of the core reparative opportunities in damaged relations. But it's not easy. This article describes the opportunity that apology presents, the difficulty we have in seizing that opportunity, and the role that third parties can play in inviting apology. It identifies (1) what is involved in a genuine apology, including three essential components; (2) the place of apology in mediation, including recognizing it as an acknowledgment of injury and identifying how to assist clients in offering an apology; and (3) the relation of apology to the adversarial system.