Content uploaded by Michael E Mccullough
Author content
All content in this area was uploaded by Michael E Mccullough
Content may be subject to copyright.
Writing About the Benefits of an Interpersonal Transgression
Facilitates Forgiveness
Michael E. McCullough, Lindsey M. Root, and Adam D. Cohen
University of Miami
The authors examined the effects of writing about the benefits of an interpersonal transgression on
forgiveness. Participants (N ⫽ 304) were randomly assigned to one of three 20-min writing tasks in
which they wrote about either (a) traumatic features of the most recent interpersonal transgression they
had suffered, (b) personal benefits resulting from the transgression, or (c) a control topic that was
unrelated to the transgression. Participants in the benefit-finding condition became more forgiving toward
their transgressors than did those in the other 2 conditions, who did not differ from each other. In part,
the benefit-finding condition appeared to facilitate forgiveness by encouraging participants to engage in
cognitive processing as they wrote their essays. Results suggest that benefit finding may be a unique and
useful addition to efforts to help people forgive interpersonal transgressions through structured inter-
ventions. The Transgression-Related Interpersonal Motivations Inventory—18-Item Version (TRIM–18)
is appended.
Keywords: forgiveness, benefit finding, meaning, LIWC, TRIM
Interpersonal transgressions are a class of interpersonal stressors in
which people perceive that another person has harmed them in a way
that they consider both painful and morally wrong. Interpersonal
transgressions can have negative effects on mental health. For exam-
ple, discovering that one’s spouse has been sexually unfaithful is
associated with a sixfold increase in the likelihood of major depres-
sive disorder (Cano & O’Leary, 2000), and the experience of humil-
iation is associated with a 70% increase in the risk of major depressive
disorder (Kendler, Hettema, Butera, Gardner, & Prescott, 2003).
Transgressions frequently elicit a desire to avoid the transgres-
sor, a desire to seek revenge against the transgressor, and a decline
in goodwill for the transgressor (McCullough et al., 1998; McCul-
lough, Worthington, & Rachal, 1997). Such motivational reactions
themselves can have negative interpersonal, psychological, and
health effects. For example, feeling avoidant and vengeful toward
one’s transgressor impedes the restoration of that relationship
(McCullough et al., 1998). In addition, people who tend to feel
vengeful or unforgiving after transgressions are prone to depres-
sive symptoms (Brown, 2003) and are more likely to be diagnosed
with major depression, generalized anxiety disorder, phobia, and
panic disorder (Kendler, Liu, et al., 2003). It is not surprising that
thoughts of revenge are among the strongest elicitors of anger
(DiGiuseppe & Froh, 2002), and entertaining one’s grudges and
thoughts of revenge lead to cardiovascular and sympathetic ner-
vous system arousal (Witvliet, Ludwig, & Vander Laan, 2001).
Therefore, helping people modify their responses to transgressions
may be useful for helping them improve their relationships as well
as their psychological and physical health.
Forgiveness of Transgressions: Is Benefit Finding an
Ingredient?
Following a transgression, most people will experience some mo-
tivation to seek revenge or to avoid the transgressor and some decline
in goodwill for the transgressor. However, some people will experi-
ence relatively quick returns to baseline in these motivations; others
will experience lingering negative motivations for days, weeks, or
even months. We have conceptualized forgiveness as prosocial
changes in these transgression-related interpersonal motivations, or
TRIMs (McCullough, Bellah, Kilpatrick, & Johnson, 2001; McCul-
lough, Fincham, & Tsang, 2003; McCullough & Hoyt, 2002; McCul-
lough et al., 1998, 1997): When people forgive, they become less
avoidant, less vengeful, and more benevolent toward the people who
have hurt them. Although some researchers acknowledge that for-
giveness can also be conceptualized as an active, deliberative process
(Worthington & Scherer, 2004), the unifying feature of forgiveness on
which most scholars seem to agree is that forgiveness is a change
process by which an individual becomes more positively disposed and
less negatively disposed toward an individual who has harmed him or
her at some point in the past (Baumeister, Exline, & Sommer, 1998;
Enright & Coyle, 1998; McCullough, Pargament, & Thoresen, 2000;
Worthington, 2005), irrespective of whether that process occurs ef-
fortfully or more passively.
How do people forgive? Theorists offer diverse answers. Attri-
butional theorists emphasize the role of responsibility attributions
(Weiner, 1995). Other work has emphasized the role of empathy
for the transgressor (McCullough, 2001; McCullough et al., 2003,
1998, 1997). Interdependence theorists emphasize the role of re-
lationship commitment (Finkel, Rusbult, Kumashiro, & Hannon,
2002). Clinical researchers often emphasize a series of steps that
people must complete (e.g., Enright & Coyle, 1998; Worthington,
2001). Clinical trials—including many published in this journal—
Michael E. McCullough, Lindsey M. Root, and Adam D. Cohen, De-
partment of Psychology, University of Miami.
This research was supported by National Institute of Mental Health
Grant R01 MH071258. We are grateful to Biing Shen and Maria Llabre for
helpful statistical advice.
Correspondence concerning this article should be addressed to Michael
E. McCullough, Department of Psychology, University of Miami, P.O.
Box 248185, Coral Gables, FL. E-mail: mikem@miami.edu
Journal of Consulting and Clinical Psychology Copyright 2006 by the American Psychological Association
2006, Vol. 74, No. 5, 887– 897 0022-006X/06/$12.00 DOI: 10.1037/0022-006X.74.5.887
887
indicate that interventions based on these approaches reduce a
wide range of psychological symptoms (Baskin & Enright, 2004;
Coyle & Enright, 1997; Freedman & Enright, 1996; Lin, Mack,
Enright, Krahn, & Baskin, 2004; Rye et al., 2005).
In the present paper, we consider a new possible ingredient in
the forgiveness process: benefit finding. Transgressions come with
costs to the victim (e.g., loss of trust, self-esteem, material re-
sources, physical or psychological well-being, etc.). However,
focusing on the benefits that one has gained (or might gain in the
future) from a transgression could help to negate some of the
transgression’s psychological costs and, by doing so, encourage
forgiveness. This proposition is consistent with other studies show-
ing that one’s attentional focus after negative life events (e.g., rumi-
nation vs. distraction, focus on the “hot” vs. “cool” features of the
event, or affirming key aspects of one’s self-image after the event) has
implications for emotion, well-being, and social behavior (e.g., Ay-
duk, Mischel, & Downey, 2002; Koole, Smeets, van Knippenberg, &
Dijksterhuis, 1999; Rusting & Nolen-Hoeksema, 1998).
Benefit finding is quite common even after traumatic transgressions
such as sexual abuse, rape, and mass shootings (Frazier & Burnett, 1994;
McMillen, 1999; McMillen, Zuravin, & Rideout, 1995). Common per-
ceived benefits include realizing one’s inner strengths, renewed spiritu-
ality, new appreciation for one’s life, improved interpersonal relation-
ships, developing new wisdom or new motivation to care for oneself, and
readjusting one’s priorities in life. Three studies suggest that benefit
finding might also encourage forgiveness.
In the first study suggesting this possibility, Zechmeister and
Romero (2002) had participants write autobiographical narratives
about incidents in which they were either the (a) victim or (b)
perpetrator of a transgression that was either (a) forgiven or (b) not
forgiven. Only 2% of the victims who did not forgive reported that
the transgression had positive consequences, whereas 21% of
victims who forgave reported positive consequences (effect size
r ⫽ .28). Only 6% of the unforgiven transgressors reported that the
transgression had positive consequences; 26% of the forgiven
transgressors reported positive consequences (effect size r ⫽ .26).
In the second relevant study, King and Miner (2000) instructed
participants to write essays about traumatic events they had expe-
rienced (including, but not limited to, interpersonal transgres-
sions). One group wrote only about traumatic aspects of the event,
whereas a second group wrote about benefits of the event. A third
group wrote about traumatic aspects of the event and benefits of
the event. A fourth group wrote about emotionally neutral mate-
rial. The essays of people who wrote about traumatic aspects of the
event were more negative and less positive in affective tone than
were those of people who wrote about benefits. In addition, people
who wrote about benefits were more likely to report that the trauma
had been resolved. They were also less bitter about the event than
were people who wrote about traumatic aspects of the event.
In the third relevant study, Romero (2004) assigned participants
(all of whom had suffered an interpersonal transgression) to one of
three conditions in which they wrote on three separate occasions:
(a) an emotionally neutral “daily events” condition as in King and
Miner (2000), (b) a condition in which they wrote about their
deepest thoughts and feelings regarding the transgression, or (c) a
condition in which they attempted to express empathy for the
transgressor and identify benefits to the self and/or the transgressor
that might come from forgiving. Of these three conditions, writing
about empathy and possible benefits of forgiveness was most
effective in promoting forgiveness. However, this intervention
involved both empathy and benefit finding (and a rather narrow
form of benefit finding at that), so the unique and complete effects
of benefit finding could not be examined. Nevertheless, these
studies provide sufficient impetus to examine the effects of benefit
finding on forgiveness more directly.
Mediators of the Benefit-Finding Effect: Language Use
and the Therapeutic Writing Paradigm
Benefit finding might promote forgiveness through several me-
diators. As mentioned above, it might increase the attention people
give to the personal benefits associated with the transgression or
reduce the attention they give to the personal costs of the trans-
gression. In such a fashion, victims might come to feel that their
transgressors owe them smaller “debts,” and as a result, their
reciprocity-based motivation to avoid the transgressor or seek
revenge may lessen (Gouldner, 1960).
The King and Miner (2000) study mentioned above (one of the
few experimental studies in the benefit-finding literature, whose
methods informed the present work) and the disclosive writing
paradigm on which it was based suggest other mediators as well.
For example, researchers have been examining the importance of
cognitive processing and the expression of affect (Pennebaker,
Mehl, & Niederhoffer, 2003) as mediators of the effects of disclo-
sive writing on measures of mental health and physical health.
Successful disclosive writing appears to be associated with high
levels of positive emotion expression (i.e., use of words such as
happy, pretty, and good), moderate levels of negative emotional
expression (i.e., use of words such as hate, worthless, and enemy),
and high levels of cognitive processing (i.e., use of causal words
such as because and hence and insight words such as think and
know; examples from Pennebaker, Francis, & Booth, 2001). In-
deed, King and Miner found that participants in their perceived
benefits group wrote essays that demonstrated greater positive
emotion and greater use of insight words than did participants in
the other conditions.
Theorists who work with process models of forgiveness (e.g.,
Enright & Coyle, 1998; Worthington, 1998) have also suggested
that recognition and expression of moderate amounts of the neg-
ative affect (e.g., hurt, fear, and anger) associated with an offense
may contribute to forgiveness. In support of this notion, Harber
and Wenberg (2005) found that the experience of negative emotion
(viz., anger) during an emotional disclosure task regarding an
interpersonal offense was correlated with the extent to which the
task restored subjective closeness toward the transgressor (which
is a well-established correlate of forgiveness). In addition, the
proposition that benefit finding may facilitate forgiveness by pro-
moting insight about the transgression is consistent with the work
of theorists (e.g., Enright & Coyle, 1998) who have proposed that
(a) identifying the meaning that a transgression has for oneself and
others and (b) realizing that one might have a new purpose in life
because of the transgression one has suffered are important aspects
of the forgiveness process. Taken together, these precedents sug-
gest that it would be worthwhile to evaluate whether benefit
finding facilitates forgiveness by influencing these emotional and
cognitive variables.
Present Study
In the present study, we evaluated the effects of a benefit-
finding writing task on forgiveness of a transgressor. To evaluate
888
MCCULLOUGH, ROOT, AND COHEN
the unique effects of benefit finding as distinct from writing about
the unpleasant aspects of the transgressions or from writing more
generally (as in King & Miner, 2000), we compared the benefit-
finding writing task with two other tasks: (a) a task in which
people wrote strictly about traumatic features of the transgression
and (b) a control task in which people wrote about a topic that was
unrelated to the transgression. We also set out to examine whether
the effects of these writing tasks were mediated by the types of
language that participants used in the essays that they wrote—
specifically, words that reflected thinking about the costs and
benefits associated with the transgression, expression of affect, and
cognitive processing (King & Miner, 2000; Pennebaker et al.,
2003).
Method
Participants
Participants were 304 undergraduates (213 women, 91 men; M age ⫽
19.31 years, SD ⫽ 2.81, range ⫽ 18 – 45; 156 White non-Hispanic, 72
Hispanic, 44 Black/African American, 32 from “other” ethnic group) who
were asked to think about the most recent time when someone with whom
they were in a relationship hurt or offended them. They received a small
amount of course credit in an introductory psychology course for
participating.
Procedures
Recruitment. Participants read a brief description of the study online. If
they chose to participate (and if they were able to recall a recent interper-
sonal transgression), they scheduled a 1-hr appointment in the first author’s
laboratory.
Completion of background questionnaire. When participants arrived at
the laboratory, a researcher explained that they would (a) complete a brief
questionnaire about the most recent occasion in their life when someone
hurt or offended them, (b) complete a 20-min writing task, and (c) com-
plete a final questionnaire about their current feelings toward the trans-
gressor. Then, participants completed the initial questionnaire that included
a communal strength measure (see Communal strength section below) and
items that elicited background information (e.g., the length of time since
the transgression occurred and a short description of the transgression).
Experimental writing conditions. Next, participants were randomly
assigned to one of three writing conditions, blocking on gender. For
participants assigned to the traumatic features writing condition (n ⫽ 101;
71 women and 30 men), the researcher read the following instructions
(participants were given a copy and asked to follow along):
In the questionnaire you just completed, you gave us some informa-
tion about a harmful thing that someone that you know did to you in
the past. For the next 20 minutes, we would like for you to write an
essay about that harmful thing they did to you. As you write, please
try to address the following points: (a) What actually happened to
you? What did this person do to you? (b) How did you feel about the
event right after it occurred to you? We would like to know especially
about the ways in which you felt angry, afraid, disgusted, or upset
after the event occurred. (c) How was your life negatively affected by
what this person did to you? In what ways is your life still negatively
affected by the negative thing that the person did to you? (d) What
sorts of negative emotions do you experience at this time in your life
when you think about the negative event that occurred to you? As you
write, really try to “let go” and experience your feelings about this
negative event. Try not to hold anything back. Be honest and candid
about this negative event, the negative feelings it created in you, and
its negative effects on your life.
Similar tasks tend to improve physical and mental health (Pennebaker,
1997; Smyth, 1998), though they are not associated with the reduction in
bitterness and the perception that a trauma has been resolved that results
from writing about benefits (King & Miner, 2000). These instructions were
similar to those that have been used in other studies of disclosive writing
(e.g., King & Miner, 2000); however, our intervention was focused on the
negative aspects of the transgression rather than on people’s thoughts and
feelings in general about the transgression (Pennebaker, 1997).
In the benefit-finding writing condition (n ⫽ 102; 72 women and 30
men), the researcher read these instructions:
In the questionnaire you just completed, you gave us some informa-
tion about a harmful thing that someone you know did to you in the
past. For the next 20 minutes, we would like for you to write an essay
related to that harmful thing they did to you. However, as you write,
we would like for you to write about positive aspects of the experi-
ence. In which ways did the thing that this person did to you lead to
positive consequences for you? Perhaps you became aware of per-
sonal strengths that you did not realize you had, perhaps a relationship
became better or stronger as a result, or perhaps you grew or became
a stronger or wiser person. Explore these issues as you write. In
particular, please try to address the following points: (a) In what ways
did the hurtful event that happened to you lead to positive outcomes
for you? That is, what personal benefits came out of this experience
for you? (b) In what ways has your life become better as a result of the
harmful thing that occurred to you? In what ways is your life or the
kind of person that you have become better today as a result of the
harmful thing that occurred to you? (c) Are there any other additional
benefits that you envision coming out of this experience for you—
perhaps some time in the future? As you write, really try to “let go”
and think deeply about possible benefits that you have gained from
this negative event, and possible benefits you might receive in the
future. Try not to hold anything back. Be as honest and candid as
possible about this event and its positive effects, or potential effects,
on your life.
In the control writing condition (n ⫽ 101; 70 women and 31 men), the
researcher read these instructions:
For the next 20 minutes, we would like for you to write an essay about
your plans for tomorrow. Please be very specific about these plans.
Imagine yourself waking up tomorrow morning. From that moment
on, what will you do? Please describe exactly what you plan to do, in
order, and describe the routes you will take to and from all of the
places you will go. If, after having written about your plans for
tomorrow, you still have time left before the 20 minutes are com-
pleted, we would like for you to write about your shoes. Beginning
with the pair of shoes that you are currently wearing, please describe
them in detail. If time still remains in the 20-minute writing period,
please also write about the other types of shoes that you own.
Control tasks like this one typically have no effects on health or well-
being (King & Miner, 2000).
Completion of the Transgression-Related Interpersonal Motivations
(TRIM) Inventory—18. After completing the writing task, participants
completed a measure of forgiveness called the TRIM–18 (McCullough et
al., 1998; see description below). Afterward, participants were debriefed,
thanked for participating, and dismissed.
Measures
TRIM–18. For a decade, forgiveness has been conceptualized as a
process of reducing one’s negative (viz., avoidance and revenge) motiva-
tions toward a transgressor and restoring one’s positive motivations re-
garding a transgressor (McCullough et al., 1997). To measure these moti-
vational changes, in the present study we used the TRIM–18 Inventory
(McCullough et al., 1998). The seven-item Avoidance subscale measures
889
SPECIAL SECTION: BENEFIT FINDING AND FORGIVENESS
motivation to avoid a transgressor (e.g., “I live as if he/she doesn’t exist,
isn’t around”). The five-item Revenge subscale measures motivation to
seek revenge (e.g., “I’ll make him/her pay”). Both have high internal
consistency (␣ ⱖ .85), moderate test–retest stability (e.g., 8-week test–
retest rs ⬇ .50), and evidence of construct validity (McCullough et al.,
2001, 1998). Items are rated on a 5-point Likert-type scale (1 ⫽ strongly
disagree to 5 ⫽ strongly agree). A recent addition is a six-item subscale for
measuring benevolence motivation (e.g., “Even though his/her actions hurt
me, I have goodwill for him/her”) that also has good reliability (McCul-
lough et al., 2003; McCullough & Hoyt, 2002). These 6 items are rated on
the same 5-point Likert-type scale as are the 12 avoidance and revenge
items.
All 18 items have never been factor analyzed jointly, so we submitted
them to a factor analysis with oblique rotation. Two factors with eigen-
values greater than one were extracted that explained 65.3% of the item
variance. The avoidance items loaded strongly and positively on the first
factor; the benevolence items loaded strongly and negatively on this factor
(explaining 53.1% of the total item variance). We named this factor
Avoidance versus Benevolence motivation (higher scores indicated higher
avoidance and lower benevolence). The five revenge items loaded strongly
and positively on the second factor (explaining 12.1% of the total item
variance). We named this second factor Revenge Motivation. The fact that
the avoidance and revenge items marked distinct but correlated factors is
consistent with previous research using the 12-item version of the TRIM
(McCullough et al., 1998). Items had relatively low cross-loadings (see
Appendix). We used the two factor scores (r ⫽ .45, p ⬍ .001) as dependent
variables.
Transgression Severity Rating Scale. When participants began the
experiment, they wrote short (typically between two-sentence and one-
paragraph) descriptions of the harms they had experienced. Later, six
research assistants read each of these 304 descriptions and then rated them
on four 7-point Likert-type scales (0 ⫽ not at all to 6 ⫽ extremely)to
indicate how “painful,” “serious,” “severe,” and “harmful” the offense
would be perceived by “the average person” who experienced it. A single
factor explained 77% of the variance in these four items, with item loadings
ranging from .80 to .93. We used the unweighted linear composite of the
four ratings as a measure of transgression severity (␣⫽.93), and then we
combined the six raters’ composite scores. The interrater reliability of this
six-rater composite, estimated as a generalizability coefficient per Hoyt and
Melby (1999), was
2
⫽ .69.
Communal strength. We measured the communal strength of partici-
pants’ relationships with their transgressors with Mills, Clark, Ford, and
Johnson’s (2004) communal strength measure, which includes 10 items
(e.g., “How much would you be willing to give up to benefit ____?”) that
participants completed on an 11-point Likert-type scale to indicate their
current degree of communal feeling toward their transgressor. In commu-
nal relationships, people are motivated by a desire to meet their partners’
needs (Mills et al., 2004). Communal relationships are exemplified by
strong marriages, family relationships, and friendships. Conversely, in
noncommunal relationships partners tend to behave according to the norm
of reciprocity (Clark, 1984). Work relationships and acquaintanceships are
typically noncommunal in nature. Mills et al. found that this measure had
high internal consistency reliability (␣⬎.85) and good evidence for
construct validity. In our sample, ␣⫽.94. We suspected that communal
strength would be correlated with forgiveness (McCullough et al., 1998),
so measuring it prior to random assignment allowed us to use it as a
covariate to increase the statistical power of our primary analyses.
Linguistic Inquiry and Word Count (LIWC; Pennebaker et al., 2001).
The LIWC is a software tool for characterizing the grammatical, linguistic,
and psychological features of text documents. We used the LIWC program
to check each word within each of our participants’ essays against LIWC’s
internal dictionary of over 2,000 words. Each word in each participant’s
essay was tagged for membership in more than 70 grammatical, linguistic,
and psychological categories. Then, after every word was tagged, the
relative prominence of each of these categories within each essay was
ascertained by dividing the number of words from a given category (e.g.,
the number of words that were related to negative emotion) that appeared
within the essay by the total number of words within that essay. As noted
previously, affect and cognitive words are particularly important mediators
of the beneficial effects of disclosive writing (Pennebaker et al., 2003), so
we focused here on three LIWC categories: (a) words indicating that the
participant experienced positive emotions while writing (e.g., happy,
pretty, and good; 261 words in the LIWC internal dictionary), (b) words
indicating that the participant experienced negative emotions while writing
(e.g., hate, worthless, and sad; 345 words in the LIWC internal dictionary),
and (c) words indicating that the participant engaged in cognitive processes
such as drawing cause-and-effect conclusions and developing insights
while writing (e.g., because, should, and maybe; 312 words in the LIWC
internal dictionary). The LIWC program’s codes are well validated against
human ratings (Pennebaker et al., 2001) and have been useful for under-
standing a variety of psychological processes (e.g., Cohn, Mehl, & Pen-
nebaker, 2004; Pennebaker et al., 2003; Pennebaker & Stone, 2003).
New LIWC categories for benefit-related and cost-related words. We
created two new LIWC categories to measure the extent to which partic-
ipants wrote about benefits and costs. After perusing the essays that our
first 50 participants wrote (roughly 16 from each of the three conditions, or
about 16% of all of the essays), the third author created a tentative list of
102 benefit-relevant words (e.g., learned, recovered, blessed, benefited,
stronger) and 127 cost-relevant words (e.g., devastation, violated, be-
trayed, unfair, broken). He then rated each of these words on two different
scales to indicate how well they reflected the constructs of benefits and
costs to the self, respectively, using two 5-point scales (1 ⫽ not at all,5⫽
extremely). The second author then completed the same rating task. The
interrater reliability of the two-rater composites of the benefit and cost
ratings were
2
⫽ .94 and .95, respectively (Hoyt & Melby, 1999). We
created a “benefit” dictionary that consisted of 72 words (or word stems)
whose mean benefit prototypicality ratings were greater than 4.0 and whose
mean cost prototypicality ratings were less than 2.0. Using an analogous
approach, we also created a “cost” dictionary consisting of 82 words and
word stems.
Other variables. We also examined whether the effects of the writing
tasks differed as a function of participants’ gender (0 ⫽ female, 1 ⫽ male)
or the recency of their transgressions. The number of days since partici-
pants’ transgressions had a log-normal distribution (range ⫽ 0 – 4,725
days, or approximately 13 years), so we applied a natural log transforma-
tion (after changing two scores of 0 days to .99 days) to render the
distribution approximately normal.
Results
Characteristics of Transgressions
Most participants’ transgressions were committed by girl-
friends/boyfriends (52%), friends of the same gender (21%), or
relatives (15%). A few reported being harmed by friends of the
other gender (7%), employers (1%), or “others” (4%). Participants
described several types of transgressions, including romantic infi-
delity (30%); insults by a friend or betrayals of a confidence
(20%); rejection, neglect, or insult by a family member (15%);
neglect or insult by a romantic partner or ex-romantic partner
(12%); termination of a love relationship (10%); rejection or
abandonment by a friend or prospective romantic partner (10%);
and insults by people other than family or friends (3%). The mean
score on the Transgression Severity Rating Scale was 4.01 (SD ⫽
0.78) on a scale ranging from 0 (not at all)to6(extremely).
Descriptive Statistics
Means, standard deviations, and correlations among major vari-
ables appear in Table 1.
890
MCCULLOUGH, ROOT, AND COHEN
What Types of Traumatic Features and Benefits Did
Participants Write About?
Participants in the traumatic features condition wrote essays that
were quite similar. Initially, most of the participants in this condition
focused on the transgression itself, including their thoughts (e.g.,
disbelief, confusion, attempts to rationalize or justify the event),
feelings (e.g., anger, pain, sadness, shock, humiliation), psychoso-
matic responses (e.g., dizziness, vomiting, sleep changes), and behav-
iors (e.g., aggression) in response to the transgression. Participants
then went on to describe negative consequences for the self (e.g.,
changes in self-esteem, body image, perceived value as a relationship
partner or friend) or relationships (e.g., negative effects on the rela-
tionship with the transgressor or on other relationships, loss of trust in
people). Thirteen percent of the participants in the traumatic features
condition spontaneously described benefits to the self or positively
reframed the event in some fashion.
1
Only a few participants in the benefit-finding condition de-
scribed the offense in much detail. They tended to begin by
describing the perceived benefits of the transgression. A list of the
reported benefits and their prevalences appear in Table 2. Only 2
participants (both of whose transgressions had occurred in the
previous 24 hr) were unable to think of any benefits. A 3rd
participant reported that he or she would have been better off
without suffering the transgression and could have learned the
same beneficial lesson without it.
Manipulation Check
Analyses of the LIWC data revealed that 2.63% (SD ⫽ 1.20%,
95% confidence interval [CI]: 2.45%–2.81%) of the words in the
essays of participants in the traumatic features condition were
cost-related words. This was a significantly higher percentage
(Cohen’s d ⫽ 0.69) than that for participants in the benefit-finding
condition (M ⫽ 1.87%, SD ⫽ 1.00%; 95% CI: 1.69%–2.05%),
which was significantly higher (Cohen’s d ⫽ 2.20) than that for
participants in the control condition (M ⫽ 0.37%, SD ⫽ 0.36%;
95% CI: 0.19%– 0.55%). Conversely, 4.56% (SD ⫽ 2.01%, 95%
CI: 4.29%– 4.84%) of the words in the essays of benefit-finding
participants were benefit-related words. This was a significantly
higher percentage (Cohen’s d ⫽ 1.60) than that for participants in
the traumatic features condition (M ⫽ 2.24%, SD ⫽ 1.18%; 95%
CI: 1.96%–2.52%), which was, in turn, significantly higher (Co-
hen’s d ⫽ 1.23) than that for participants in the control condition
(M ⫽ 1.10%, SD ⫽ 0.68%; 95% CI: 0.82%–1.37%). Therefore,
the writing conditions appeared to be successful manipulations of
benefit-related thinking and cost-related thinking.
Effects of Writing About the Transgression on
Forgiveness
A multivariate analysis of covariance with two dependent variables
(avoidance vs. benevolence motivation and revenge motivation) re-
vealed that the writing tasks did not interact with gender, recency of
the transgression, severity of the transgression, or communal strength,
so we did not consider these interactions further.
2
However, commu
-
nal strength and transgression severity accounted for unique variance
in the two dependent variables, so we retained them as covariates to
increase statistical power for the main analysis, which was a one-way
1
When we removed the data for these participants, the variance in the
outcomes that could be attributed to the differences among the three writing
conditions increased slightly but not substantially. However, it was statistically
more conservative to leave those participants in the analyses, so we did.
2
Prior to conducting these analyses, we used the expectation-
maximization routine in SPSS to estimate a small amount of missing data.
Expectation maximization is a maximum likelihood estimation procedure
that yields unbiased estimates of missing data when missingness can be
assumed to be at random. The first 60 participants were missing communal
strength scores because of an Institutional Review Board delay. Because
we randomly assigned participants to conditions, this cause of missingness
affected all three conditions equally. Thus, from the perspective of evalu-
ating differences among the three conditions, the missingness was at
random and was therefore ignorable. Ten people lacked complete data on
the TRIM Inventory, but their missingness came to no more than 3 items
per person out of a total of 18 items. Given the high intercorrelations
among the 18 items and the small amount of missingness overall, this low
amount of missingness was considered so trivial as to be ignorable in any
case. As would be expected, the means and standard deviations of the
observed and estimated values did not differ substantially. By estimating
the missing data, we restored our sample size to 304. For analyses involv-
ing the LIWC data, the sample size was 303: One participant wrote his or
her essay in Spanish, and we did not translate this essay into English;
therefore, it could not be coded with the LIWC software. We did not
estimate these missing LIWC data for this participant because, given the
low correlations of the LIWC data with the other variables in the data set
(see Table 1), estimation would have been poor.
Table 1
Means, Standard Deviations, and Correlations for Major Study Variables
Variable MSDRevenge
Transgression
Severity
Communal
Strength
% Positive
Emotion
% Negative
Emotion
% Cognitive
Processing
% Cost
Words
% Benefit
Words
1. Avoidance vs. benevolence 0 1.0 0.45*** 0.12* ⫺0.48*** ⫺0.02 ⫺0.07 ⫺0.15** ⫺0.05 ⫺0.02
2. Revenge 0 1.0 — 0.01 ⫺0.21*** ⫺0.13* ⫺0.12* ⫺0.20*** ⫺0.08 ⫺0.15*
3. Transgression severity 4.01 0.78 — 0.05 0.01 0.08 0.02 0.02 ⫺0.09
4. Communal strength 53.42 23.91 — ⫺0.10 0.05 ⫺0.01 0.02 ⫺0.13*
5. % positive emotion words 2.29 1.48 — 0.24*** 0.58*** 0.25*** 0.81***
6. % negative emotion words 1.94 1.49 — 0.58*** 0.81*** 0.17***
7. % cognitive processing words 6.73 3.45 — 0.56*** 0.57***
8. % cost words 1.62 1.31 — 0.24***
9. % benefit words 2.64 2.01 —
Note. Ns ⫽ 303 or 304.
* p ⬍ .05. ** p ⬍ .01. *** p ⬍ .001.
891
SPECIAL SECTION: BENEFIT FINDING AND FORGIVENESS
(writing task: traumatic features, benefit finding, control) multivariate
analysis of covariance with communal strength and transgression
severity as covariates. The two dependent variables were the avoid-
ance versus benevolence and revenge motivation factors from the
TRIM Inventory.
The writing tasks produced a significant multivariate effect on
the dependent variables, Wilks’s ⌳⫽.94, F(4, 596) ⫽ 4.87, p ⫽
.001 (two-tailed, as with all tests below), partial
2
⫽ .03. This
significant multivariate effect was explored with two univariate
analyses of covariance. There were significant differences among
the three conditions on both dependent measures. Differences
among the conditions accounted for small but significant propor-
tions of variance in the avoidance versus benevolence factor, F(2,
299) ⫽ 3.99, p ⫽ .02, partial
2
⫽ .03, and the revenge factor, F(2,
299) ⫽ 8.87, p ⬍ .001, partial
2
⫽ .06.
Pairwise comparisons on the adjusted means conducted with
Bonferroni’s correction revealed that the benefit-finding condition
(adjusted M ⫽⫺0.19, SD ⫽ 0.85, 95% CI: ⫺0.36, ⫺0.02) yielded
significantly ( p ⫽ .035) lower avoidance versus benevolence
scores (Cohen’s d ⫽⫺0.36) than did the control condition (ad-
justed M ⫽ 0.12, SD ⫽ 0.86, 95% CI: ⫺0.05, 0.28) and, with
marginal significance ( p ⫽ .059), lower avoidance versus benev-
olence scores (Cohen’s d ⫽⫺0.33) than did the traumatic features
condition (adjusted M ⫽ 0.09, SD ⫽ 0.85, 95% CI: ⫺0.08, 0.26).
The means of the traumatic features and control conditions did not
differ from each other ( p ⫽ 1.0; see Table 3).
Similarly, the benefit-finding condition (adjusted M ⫽⫺0.31,
SD ⫽ 0.95, 95% CI: ⫺0.49, ⫺0.12) yielded significantly ( p ⫽ .02)
lower revenge scores (Cohen’s d ⫽⫺0.38) than did the traumatic
features condition (adjusted M ⫽ 0.06, SD ⫽ 0.95, 95% CI ⫽
⫺0.13, 0.24). It also yielded significantly ( p ⬍ .001) lower re-
venge scores (Cohen’s d ⫽⫺0.58) than did the control condition
(adjusted M ⫽ 0.25, SD ⫽ 0.95, 95% CI: 0.06, 0.43). The means
of the traumatic features and control conditions did not differ from
each other ( p ⫽ .46). In sum, then, the benefit-finding condition
fostered more forgiveness than did the traumatic features condition
or the control condition, which did not differ from each other.
Mediational Analyses
Having discovered that the benefit-finding condition was more
effective than the other two conditions at promoting forgiveness (a
first condition for mediation), we evaluated whether the benefit-
finding condition’s superiority was mediated by its effects on
participants’ expression of benefit-related content, cost-related
content, negative emotion, positive emotion, and cognitive words
in their essays relative to the other two conditions. King and Miner
(2000) used a similar analytic strategy to compare the linguistic
features of essays from a benefit-finding writing condition, a
traumatic features condition, and an emotionally neutral control
group (see also Harber & Wenberg, 2005).
To test the second condition for mediation, we tested whether
the five putative mediators were influenced by the writing condi-
tions (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002).
To do so, we conducted five separate regressions in which the five
linguistic aspects of participants’ essays were used individually as
criterion variables. We predicted these criteria with four variables:
(a) communal strength, (b) transgression severity, (c) a dummy
Table 2
Benefits Reported in Benefit-Finding Essays
Benefit
Prevalence in
sample (%)
Grew stronger or discovered unknown strength 55
Wiser (i.e., slower to trust in relationships, less naı¨ve) 44
Allowed for new life experiences 29
Strengthened other relationships 26
Became better at communicating feelings 25
Ended a bad relationship 25
Increased confidence 23
Became more kind, more compassionate, less selfish 21
Became more aware of the feelings of others 20
Strengthened the relationship with the offender 19
Learned the importance of forgiveness 18
Learned what a romantic relationship requires 12
Not as concerned about others’ opinions/less need to please others 12
Realized the importance of other things and relationships, beyond romantic relationships 8
Learned what not to do in a romantic relationship 8
Learned about qualities to look for in friends 8
Learned the importance of dealing with anger and/or keeping a cool head 7
Learned to stand up for myself 6
Learned to value what I have 6
Importance of caring for myself 6
Learned what to look for in a romantic partner 5
Increased confidence in beliefs/increased faith 5
Other areas of life improved (e.g., athletics, artistic expression, schoolwork) 4
Importance of trusting instincts 4
Clarified sexual orientation 3
Learned the importance of thinking about the consequences of my actions 2
Became an advocate for others in similar situations 2
Provided purpose in life or a future goal 2
892
MCCULLOUGH, ROOT, AND COHEN
code representing the effect of the traumatic features condition
(traumatic features group coded 1, all others coded 0), and (d) a
dummy code representing the effect of the control condition (con-
trol group coded 1, all others coded 0). The advantage of this
coding scheme is that the unique effect of each dummy variable
represents the contrast of the group coded 1 from the group coded
0 on both dummy codes (in this case, the benefit-finding condition;
Cohen & Cohen, 1983).
In each of these five regressions, the dummy-coded variables
reflecting membership in the traumatic features condition and
membership in the control condition were both statistically signif-
icant ( ps ⬍ .001). The traumatic features condition encouraged the
use of (a) a lower percentage of benefit-related words (coefficient
[B] ⫽⫺2.30, SE ⫽ 0.20), positive emotion words (B ⫽⫺1.33,
SE ⫽ 0.15), and cognitive processing words (B ⫽⫺1.16, SE ⫽
0.27) and (b) a higher percentage of cost-related words (B ⫽ 0.76,
SE ⫽ 0.13) and negative emotion words (B ⫽ 1.14, SE ⫽ 0.14)
than did the benefit-finding condition (all Bs unstandardized; all
ps ⬍ .001). The control group encouraged the use of a lower
percentage of benefit-related words (B ⫽⫺3.43, SE ⫽ 0.20),
cost-related words (B ⫽⫺1.51, SE ⫽ 0.13), positive emotion
words (B ⫽⫺2.49, SE ⫽ 0.15), negative emotion words (B ⫽
⫺1.62, SE ⫽ 0.14), and cognitive processing words (B ⫽⫺6.64,
SE ⫽ 0.27) than did the benefit-finding condition. Table 3 shows
the adjusted means of the three conditions for these five mediators
and the results of pairwise comparisons conducted with Bonferro-
ni’s correction for multiple comparisons.
The third step in examining mediation was to examine whether
five hypothesized mediators were associated with the outcome
variables (MacKinnon et al., 2002) when both the mediators and
the independent variable were used to predict the outcome vari-
ables. As a preparatory step in evaluating this condition, we first
conducted 10 separate regression models. In these 10 models, we
regressed either the avoidance versus benevolence or the revenge
factor individually onto one of the five putative mediators, holding
communal strength and transgression severity constant (as in the
analyses in which we demonstrated that the benefit-finding con-
dition produced more change in the avoidance vs. benevolence and
revenge dependent variables than did the other two conditions).
Then, we evaluated whether the five putative mediators uniquely
predicted the criterion variables.
In the five regressions using avoidance versus benevolence
motivation as a criterion, in only one (the regression involving
cognitive processing) was a putative mediator uniquely associated
with the putative outcome after controlling for communal strength
and transgression severity: The use of cognitive processing words
(e.g., because, should, and maybe) was negatively associated (B ⫽
⫺0.05, SE ⫽ 0.01, p ⬍ .01) with avoidance versus benevolence
motivation.
In contrast, four of the five putative mediators had significant
unique associations with revenge motivation after controlling for
communal strength and transgression severity. Percentage of ben-
efit words (B ⫽⫺0.09, SE ⫽ 0.03, p ⬍ .01), percentage of
negative emotion words (B ⫽⫺0.07, SE ⫽ 0.04, p ⬍ .05),
percentage of positive emotion words (B ⫽⫺0.10, SE ⫽ 0.04, p ⬍
.01), and percentage of cognitive processing words (B ⫽⫺0.06,
SE ⫽ 0.02, p ⬍ .001) were all uniquely and negatively associated
with revenge motivation.
As the next part of this third step, we evaluated whether the
putative mediators maintained a significant association with the
dependent variables that they were hypothesized to mediate (in this
case, our avoidance vs. benevolence and revenge variables) when
the putative independent variable (in this case, the writing condi-
tions) and the putative mediator (the four linguistic features of
participants’ essays mentioned in the previous paragraph) pre-
dicted the outcomes simultaneously (MacKinnon et al., 2002).
Having found one putative mediator (percentage of cognitive
processing words) that was uniquely associated with avoidance
versus benevolence motivation and four putative mediators (per-
centage of benefit words, percentage of negative emotion words,
percentage of positive emotion words, and percentage of cognitive
processing) that were uniquely associated with revenge motiva-
tion, we therefore ran five additional regressions to test this con-
dition. In each of these regressions, one of the two dependent
variables was regressed on five variables: (a) one of the relevant
putative mediators, (b– c) the two dummy-coded variables repre-
senting the effects of the traumatic features and control conditions
(vs. the benefit-finding condition), (d) communal strength, and (e)
transgression severity.
In only one of these five regressions did any putative mediator
maintain a statistically significant association with the dependent
variable: Percentage of cognitive processing words had a statisti-
cally significant unique association with avoidance versus benev-
olence motivation (B ⫽⫺0.08, SE ⫽ 0.03, p ⬍ .01). In this
equation, the dummy-coded variables representing the contrast
between the traumatic features and the benefit-finding conditions
(B ⫽ 0.20, SE ⫽ 0.12, p ⬎ .10) and the contrast between the
control and the benefit-finding conditions (B ⫽ 0.19, SE ⫽ 0.21,
p ⬎ .35) were no longer significant.
The fourth step in testing mediation was to evaluate the statis-
tical significance of the mediated effect using Sobel’s (1982) test.
MacKinnon, Warsi, and Dwyer (1995) demonstrated that testing
the statistical significance of the product that results from multi-
plying the coefficient expressing the association of the putative
independent variable and the putative mediator (i.e., the path from
X to M) by the coefficient expressing the association of the
putative mediator and the putative dependent variable (i.e., the
path from M to Y when X is simultaneously used to predict Y)is
algebraically equivalent to testing the reduction in direct effect that
Table 3
Means for Dependent Variables and Putative Mediators by
Treatment Condition
Dependent variable / putative
mediator
Treatment condition M
Benefit
finding
Traumatic
features Control
Avoidance vs. benevolence motivation ⫺0.19
a
0.09
b
†
0.12
b
Revenge motivation ⫺0.31
a
0.06
b
0.25
b
% benefit-related words 4.54
a
2.24
b
1.11
b
% cost-related words 1.87
a
2.63
b
0.36
b
% negative emotion words 2.10
a
3.24
b
0.48
b
% positive emotion words 3.56
a
2.23
b
1.07
b
% cognitive processing words 9.32
a
8.17
b
2.68
b
Note. All means are adjusted for communal strength and transgression
severity. Means in a single row with different subscripts are significantly
different per pairwise comparisons conducted with Bonferroni’s correction
for multiple comparisons.
† p value for difference between benefit-finding condition and traumatic
features condition ⫽ .059.
893
SPECIAL SECTION: BENEFIT FINDING AND FORGIVENESS
results from simultaneously controlling the mediator (cf. Baron &
Kenny, 1986). Sobel tests for mediation demonstrated that cogni-
tive processing mediated the superior efficacy of the benefit-
finding condition relative to the traumatic features condition, Sobel
t ⫽ 2.41, p ⬍ .05, and its superior efficacy relative to the control
condition, Sobel t ⫽ 2.87, p ⬍ .01. Therefore, data supported the
idea that the benefit-finding condition was more effective than the
other two conditions in reducing avoidance versus benevolence
motivation in part because it was more effective in promoting
cognitive processing during the essay-writing task.
Cognitive processing did not predict unique variance in revenge
motivation in the regression equation in which it was added ( p ⫽
.26). None of the other putative mediators did either (all ps ⬎ .35).
Therefore, we concluded that the putative mediators in which we
were interested did not mediate the effects of the writing condi-
tions on revenge motivation.
Discussion
Benefit finding is surprisingly common in the aftermath of
adversity (McMillen, 1999) and is positively associated with a
variety of psychological outcomes (e.g., Affleck & Tennen, 1996).
Similarly, writing about traumatic events appears to foster physical
and psychological health (Pennebaker, 1997; Smyth, 1998). Com-
bining these two ideas, King and Miner (2000) examined the
health effects of writing about the perceived benefits of traumatic
experiences. They found that writing about benefits produced
some of the same psychological and health benefits as did more
standard forms of disclosive writing (e.g., Pennebaker, 1997), but
with additional payoffs: For example, when people wrote about the
perceived benefits of traumatic events, they felt less bitter about
the events and felt that the events had been more fully resolved
than did people who wrote about traumatic aspects of the events
they had suffered.
King and Miner’s (2000) findings, and those from two other
studies (Romero, 2004; Zechmeister & Romero, 2002), led us to
suspect that writing about the benefits of interpersonal transgres-
sions would help people forgive those transgressions. The present
results confirmed this suspicion. When our participants wrote
about the benefits or potential benefits of transgressions they had
recently suffered (a task that they found remarkably easy to com-
plete), they experienced reductions in avoidance versus benevo-
lence motivation and reductions in revenge motivation—the mo-
tivations underlying forgiveness (McCullough et al., 2003, 1998,
1997). The benefit-finding writing condition was more efficacious
than a writing condition that had participants consider the strictly
negative features of the transgressions they had suffered—a con-
dition that was, as mentioned above, considerably more negative
than Pennebaker’s (1997) standard writing task. It was also supe-
rior to an emotionally neutral control condition that has been used
elsewhere (e.g., King & Miner, 2000).
We hasten to note that the condition in which participants wrote
about traumatic features of the transgression was not significantly
more effective at encouraging forgiveness than was the control
condition. This suggests that the major benefits of disclosive
writing, at least in terms of forgiveness, came from some ingredi-
ent that is more characteristic of the benefit-finding condition than
of the traumatic features condition. (We return to this theme below
when we discuss mediation.) This is consistent with the work of
King and Miner (2000), who found that when people wrote about
the perceived benefits of traumatic events, they ultimately felt less
bitter about the event and felt that the traumatic event had been
more fully resolved than did people who wrote strictly about
traumatic aspects of the event. Therefore, on the basis of King and
Miner’s research and the present findings, we conclude that at least
some of the therapeutic benefits of disclosive writing about an
unpleasant or traumatic life event—which might go by names such
as “resolution,” “acceptance,” and “forgiveness,” are better ob-
tained with writing tasks that help writers to look for the personal
benefits to be gained from the negative experience than from tasks
in which writers simply vent their negative emotions or focus on
negative aspects of their experience.
The effect of the benefit-finding condition on forgiveness was
not large in magnitude: It yielded gains ranging from 0.33 to 0.58
SD units relative to the two other conditions. Cohen (1992) pro-
posed that treatment effects of this magnitude in the behavioral
sciences should be considered small to medium. Nevertheless,
these effect sizes were comparable to those of most other nonclini-
cal interventions (i.e., those lasting 1 or 2 hr) that have been used
to facilitate forgiveness in prior research (Baskin & Enright, 2004).
These effects would most likely have been larger if the writing task
had been repeated over the course of several days (Pennebaker,
1997), which is common in studies of disclosive writing. Given the
success of writing about the benefits of a transgression for even 20
min, it seems worthwhile to conduct further research to examine
how much more benefit might emerge from a more intensive
writing procedure carried out in several different sessions over
several days.
Mediators of the Effects of the Writing Conditions on
Forgiveness
We examined five possible mediators of the effects of the
writing conditions on forgiveness. To our surprise, neither the
percentage of benefit-related words nor the percentage of cost-
related words in people’s essays mediated the effects of the
benefit-finding condition on forgiveness. Perhaps this is because
our essay-based measure of benefit finding did not adequately
measure the benefit-finding construct. However, it may be that the
efficacy of the benefit-finding condition was not due to its ability
to help people focus on the benefits (vs. costs) of a transgression
but rather to its ability to facilitate another of the five mediators we
evaluated: cognitive processing. When we controlled for the ef-
fects of the writing conditions on the use of cognitive processing
words during the essay task, the salutary effects of the benefit-
finding condition on avoidance versus benevolence motivation
relative to the other two conditions disappeared. This suggests that
the benefit-finding writing task was superior to the other two tasks
at reducing avoidance motivation because it was so much better
than the other two conditions at encouraging cognitive processing
of the transgression—that is, at helping people to create essays
involving insight, cause-and-effect relations, and connections
among the various elements. Previous work has shown that suc-
cessful disclosive writing is associated with high levels of cogni-
tive processing (King & Miner, 2000; Pennebaker et al., 2003).
Pennebaker et al. (2003) observed that when people write about
negative life events, a high number of cognitive processing words
suggests that the writers have developed coherent stories or nar-
ratives surrounding the negative life events about which they are
writing. This is a particularly exciting insight because it suggests
894
MCCULLOUGH, ROOT, AND COHEN
that one of the key ingredients for coping with negative life events
is developing a coherent narrative for interpreting them—that is, a
way of making sense of why they happened and their implications
for one’s future. The proposition that benefit finding may facilitate
forgiveness by helping people find meaning in the transgressions
that they have suffered harkens back to the theoretical writing of
Enright and colleagues (e.g., Enright & Coyle, 1998), who pro-
posed that both (a) identifying the meaning that a transgression has
for oneself and others and (b) realizing that one might have a new
purpose in life because of the transgression one has suffered are
important elements in the forgiveness process. It appears that
benefit finding might help to create elements such as those that
Enright and his colleagues have proposed.
Of course, other mechanisms are probably also responsible for
some of the effects of the benefit-finding intervention. For exam-
ple, the differential efficacy of the three conditions on revenge
motivation was not mediated by cognitive processing or any of the
other variables that we explored using the LIWC software. This
means that many other candidate mediators, including increased
empathy for the transgressor (McCullough et al., 1997; Romero,
2004), reduced rumination about the transgression (McCullough et
al., 2001), and social connectedness (Romero, 2004) should be
explored in future research.
Limitations and Directions for Future Research
The present findings have a few limitations. First, the fact that
the dependent measures were self-reports limits our ability to
know whether the writing tasks created behaviorally relevant
changes in our participants’ representations of their transgressors.
Second, our results do not indicate how long the effects of the
benefit-finding writing task lasted. Did the writing conditions
create long-term changes in participants’ representations of their
transgressors, or did the reductions in avoidance and revenge
motivation that the benefit-finding condition created dissipate a
few moments after participants left the laboratory? Future studies
should extend the benefit-finding writing procedure over the
course of several days and measure forgiveness for several weeks
after the experiment is concluded. The use of growth modeling for
operationalizing forgiveness as a change process (McCullough et
al., 2003) and for studying the effects of forgiveness interventions
(McCullough & Root, 2005) would be particularly useful for this
purpose. Third, more work should be done to explore the media-
tors responsible for the links between benefit finding and forgive-
ness. Fourth, the applicability of these findings to nonstudent
populations and to populations who have undergone serious forms
of interpersonal trauma (e.g., sexual assaults or homicides of loved
ones) remains to be shown.
Fifth and finally, it is unclear that the effects of writing about
benefits on forgiveness would lead to changes in the psychological
and physical health outcomes that have been associated with
forgiveness in past research (Brown, 2003; Rye et al., 2005;
Witvliet et al., 2001). These are all useful and important directions
for future research.
Summary
Benefit finding is remarkably common in the face of adversity
and has been associated with a variety of positive psychological
outcomes. In this work, we demonstrated that writing about ben-
efits can facilitate forgiveness as well. People who write about the
benefits of transgressions they have encountered become less
avoidant, more benevolent, and less vengeful toward their trans-
gressors as a result. These results provide strong justification for
more research on the value of benefit finding for helping people
overcome the negative interpersonal, psychological, and health
effects of interpersonal transgressions. It also provides some en-
couragement for efforts to integrate benefit finding into ongoing
efforts to use forgiveness as a component of clinical interventions
for prevention and treatment.
References
Affleck, G., & Tennen, H. (1996). Construing benefits from adversity:
Adaptational significance and dispositional underpinnings. Journal of
Personality, 64, 899 –922.
Ayduk, O., Mischel, W., & Downey, G. (2002). Attentional mechanisms
linking rejection to hostile reactivity: The role of “hot” versus “cool”
focus. Psychological Science, 13, 443– 448.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator distinction
in social psychological research: Conceptual, strategic, and statistical
considerations. Journal of Personality and Social Psychology, 51, 1173–
1182.
Baskin, T. W., & Enright, R. D. (2004). Intervention studies on forgive-
ness: A meta-analysis. Journal of Counseling and Development, 82,
79 –90.
Baumeister, R. F., Exline, J. J., & Sommer, K. L. (1998). The victim role,
grudge theory, and two dimensions of forgiveness. In E. L. Worthington
(Ed.), Dimensions of forgiveness: Psychological research and theolog-
ical perspectives (pp. 79–104). Philadelphia, PA: Templeton Foundation
Press.
Brown, R. P. (2003). Measuring individual differences in the tendency to
forgive: Construct validity and links with depression. Personality and
Social Psychology Bulletin, 29, 759 –771.
Cano, A., & O’Leary, K. D. (2000). Infidelity and separations precipitate
major depressive episodes and symptoms of nonspecific depression and
anxiety. Journal of Consulting and Clinical Psychology, 68, 774 –781.
Clark, M. S. (1984). Record keeping in two types of relationships. Journal
of Personality and Social Psychology, 47, 549 –557.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation
analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Cohn, M. A., Mehl, M. R., & Pennebaker, J. W. (2004). Linguistic markers
of psychological change surrounding September 11, 2001. Psychologi-
cal Science, 15, 687– 693.
Coyle, C. T., & Enright, R. D. (1997). Forgiveness intervention with
postabortion men. Journal of Consulting and Clinical Psychology, 65,
1042–1046.
DiGiuseppe, R., & Froh, J. J. (2002). What cognitions predict state anger?
Journal of Rational–Emotive & Cognitive Behavior Therapy, 20, 133–
150.
Enright, R. D., & Coyle, C. T. (1998). Researching the process model of
forgiveness within psychological interventions. In E. L. Worthington
(Ed.), Dimensions of forgiveness: Psychological research and theolog-
ical perspectives (pp. 139 –161). Philadelphia, PA: Templeton Founda-
tion Press.
Finkel, E. J., Rusbult, C. E., Kumashiro, M., & Hannon, P. A. (2002).
Dealing with a betrayal in close relationships: Does commitment pro-
mote forgiveness? Journal of Personality and Social Psychology, 82,
956 –974.
Frazier, P. A., & Burnett, J. W. (1994). Immediate coping strategies among
rape victims. Journal of Counseling and Development, 72, 633– 639.
Freedman, S. R., & Enright, R. D. (1996). Forgiveness as an intervention
goal with incest survivors. Journal of Consulting and Clinical Psychol-
ogy, 64, 983–992.
895
SPECIAL SECTION: BENEFIT FINDING AND FORGIVENESS
Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement.
American Sociological Review, 25, 161–178.
Harber, K. D., & Wenberg, K. E. (2005). Emotional disclosure and close-
ness toward offenders. Personality and Social Psychology Bulletin, 31,
734 –746.
Hoyt, W. T., & Melby, J. N. (1999). Dependability of measurement in
counseling: An introduction to generalizability theory. Counseling Psy-
chologist, 43, 430 – 447.
Kendler, K. S., Hettema, J. M., Butera, F., Gardner, C. O., & Prescott,
C. A. (2003). Life event dimensions of loss, humiliation, entrapment,
and danger in the prediction of onsets of major depression and gener-
alized anxiety. Archives of General Psychiatry, 60, 789 –796.
Kendler, K. S., Liu, X., Gardner, C. O., McCullough, M. E., Larson, D. B.,
& Prescott, C. A. (2003). Dimensions of religiosity and their relationship
to lifetime psychiatric and substance use disorders. American Journal of
Psychiatry, 160, 496 –503.
King, L. A., & Miner, K. N. (2000). Writing about the perceived benefits
of traumatic events: Implications for physical health. Personality and
Social Psychology Bulletin, 26, 220 –230.
Koole, S. L., Smeets, K., van Knippenberg, A., & Dijksterhuis, A. (1999).
The cessation of rumination through self-affirmation. Journal of Per-
sonality and Social Psychology, 77, 111–125.
Lin, W. F., Mack, D., Enright, R. D., Krahn, D., & Baskin, T. W. (2004).
Effects of forgiveness therapy on anger, mood, and vulnerability to
substance use among inpatient substance-dependent clients. Journal of
Consulting and Clinical Psychology, 72, 1114 –1121.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., &
Sheets, V. (2002). A comparison of methods to test mediation and other
intervening variable effects. Psychological Methods, 7, 83–104.
MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study
of mediated effect measures. Multivariate Behavioral Research, 30,
41– 62.
McCullough, M. E. (2001). Forgiveness: Who does it and how do they do
it? Current Directions in Psychological Science, 10, 194 –197.
McCullough, M. E., Bellah, C. G., Kilpatrick, S. D., & Johnson, J. L.
(2001). Vengefulness: Relationships with forgiveness, rumination, well-
being, and the Big Five. Personality and Social Psychology Bulletin, 27,
601– 610.
McCullough, M. E., Fincham, F. D., & Tsang, J. (2003). Forgiveness,
forbearance, and time: The temporal unfolding of transgression-related
interpersonal motivations. Journal of Personality and Social Psychol-
ogy, 84, 540 –557.
McCullough, M. E., & Hoyt, W. T. (2002). Transgression-related motiva-
tional dispositions: Personality substrates of forgiveness and their links
to the Big Five. Personality and Social Psychology Bulletin, 28, 1556 –
1573.
McCullough, M. E., Pargament, K. I., & Thoresen, C. E. (2000). The
psychology of forgiveness: History, conceptual issues, and overview. In
M. E. McCullough, K. I. Pargament, & C. E. Thoresen (Eds.), Forgive-
ness: Theory, research, and practice (pp. 1–14). New York: Guilford.
McCullough, M. E., Rachal, K. C., Sandage, S. J., Worthington, E. L.,
Brown, S. W., & Hight, T. L. (1998). Interpersonal forgiving in close
relationships. II: Theoretical elaboration and measurement. Journal of
Personality and Social Psychology, 75, 1586 –1603.
McCullough, M. E., & Root, L. M. (2005). Forgiveness as change. In E. L.
Worthington (Ed.), Handbook of forgiveness (pp. 91–107). New York:
Routledge.
McCullough, M. E., Worthington, E. L., & Rachal, K. C. (1997). Inter-
personal forgiving in close relationships. Journal of Personality and
Social Psychology, 73, 321–336.
McMillen, J. C. (1999). Better for it: How people benefit from adversity.
Social Work, 44, 455– 468.
McMillen, J. C., Zuravin, S., & Rideout, G. B. (1995). Perceptions of
benefit from child sexual abuse. Journal of Consulting and Clinical
Psychology, 63, 1037–1043.
Mills, J., Clark, M. S., Ford, T. E., & Johnson, M. (2004). Measurement of
communal strength. Personal Relationships, 11, 213–230.
Pennebaker, J. W. (1997). Writing about emotional experiences as a
therapeutic process. Psychological Science, 8, 162–166.
Pennebaker, J. W., Francis, M. E., & Booth, R. J. (2001). Linguistic inquiry
and word count. Mahwah, NJ: Erlbaum.
Pennebaker, J. W., Mehl, M. R., & Niederhoffer, K. G. (2003). Psycho-
logical aspects of natural language use: Our words, our selves. Annual
Review of Psychology, 54, 547–577.
Pennebaker, J. W., & Stone, L. D. (2003). Language use over the life span.
Journal of Personality and Social Psychology, 85, 291–301.
Romero, C. (2004, August). Writing wrongs: An intervention to promote
forgiveness. Poster session presented at the annual meeting of the Amer-
ican Psychological Association, Honolulu, HI.
Rusting, C. L., & Nolen-Hoeksema, S. (1998). Regulating responses to
anger: Effects of rumination and distraction on angry mood. Journal of
Personality and Social Psychology, 74, 790 – 803.
Rye, M. S., Pargament, K. I., Pan, W., Yingling, D. W., Shogren, K. A., &
Ito, M. (2005). Can group interventions facilitate forgiveness of an
ex-spouse?: A randomized clinical trial. Journal of Consulting and
Clinical Psychology, 73, 880 – 892.
Smyth, J. M. (1998). Written emotional expression: Effect sizes, outcome
types, and moderating variables. Journal of Consulting and Clinical
Psychology, 66, 174 –184.
Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural
equations models. In S. Leinhart (Ed.), Sociological methodology 1982
(pp. 290 –312). San Francisco: Jossey-Bass.
Weiner, B. (1995). Judgments of responsibility: Foundations for a theory
of social conduct. New York: Guilford.
Witvliet, C. v. O., Ludwig, T. E., & Vander Laan, K. L. (2001). Granting
forgiveness or harboring grudges: Implications for emotion, physiology,
and health. Psychological Science, 12, 117–123.
Worthington, E. L. (1998). The pyramid model of forgiveness: Some
interdisciplinary speculations about unforgiveness and the promotion of
forgiveness. In E. L. Worthington (Ed.), Dimensions of forgiveness:
Psychological research and theological perspectives (pp. 107–137).
Philadelphia: Templeton Foundation Press.
Worthington, E. L. (2001). Five steps to forgiveness: The art and science
of forgiving. New York: Crown.
Worthington, E. L. (2005). More questions about forgiveness: Research
agenda for 2005–2015. In E. L. Worthington (Ed.), Handbook of for-
giveness (pp. 557–573). New York: Routledge.
Worthington, E. L., & Scherer, M. (2004). Forgiveness as an emotion-
focused coping strategy that can reduce health risks and promote health
resilience: Theory, review, and hypotheses. Psychology and Health, 19,
385– 405.
Zechmeister, J. S., & Romero, C. (2002). Victim and offender accounts of
interpersonal conflict: Autobiographical narratives of forgiveness and
unforgiveness. Journal of Personality and Social Psychology, 84, 675–
686.
896
MCCULLOUGH, ROOT, AND COHEN
Received May 31, 2005
Revision received May 16, 2006
Accepted May 19, 2006 䡲
Appendix
Transgression-Related Interpersonal Motivations Inventory—18-Item Version, With Factor Loadings
Item
Correlation with
Factor 1
Correlation with
Factor 2
I’ll make him/her pay. .32 .76
I am trying to keep as much distance between us as possible. .81 .37
Even though his/her actions hurt me, I have goodwill for him/her. ⫺.75 ⫺.59
I wish that something bad would happen to him/her. .46 .76
I am living as if he/she doesn’t exist, isn’t around. .80 .32
I want us to bury the hatchet and move forward with our relationship. ⫺.83 ⫺.30
I don’t trust him/her. .75 .35
Despite what he/she did, I want us to have a positive relationship again. ⫺.88 ⫺.38
I want him/her to get what he/she deserves. .38 .77
I am finding it difficult to act warmly toward him/her. .76 .36
I am avoiding him/her. .84 .42
Although he/she hurt me, I am putting the hurts aside so we could resume our relationship. ⫺.86 ⫺.33
I’m going to get even. .25 .77
I forgive him/her for what he/she did to me. ⫺.71 ⫺.46
I cut off the relationship with him/her. .85 .38
I have released my anger so I can work on restoring our relationship to health. ⫺.81 ⫺.33
I want to see him/her hurt and miserable. .46 .84
I withdraw from him/her. .85 .45
Note. Items were rated on a 5-point scale (1 ⫽ strongly disagree, 2 ⫽ disagree, 3 ⫽ neutral, 4 ⫽ agree, and 5 ⫽ strongly agree). Coefficients are loadings
from the structure matrix resulting from principal components analyses using oblimin rotation with Kaiser normalization. Loadings indicate the correlations
of the individual items with the two correlated factors (r ⫽ .45). Factor 1 accounted for 53.1% of total item variance, and Factor 2 accounted for an
additional 12.1% of total item variance.
897
SPECIAL SECTION: BENEFIT FINDING AND FORGIVENESS
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Consulting and Clinical Psychology
This content is subject to copyright. Terms and conditions apply.
