Journal of Counseling Psychology © 2015 American Psychological Association

2015, Vol. 62, No. 2, 280 –288 0022-0167/15/$12.00 http://dx.doi.org/10.1037/cou0000054

Making a Decision to Forgive

Don E. Davis Joshua N. Hook

Georgia State University University of North Texas

Daryl R. Van Tongeren Cirleen DeBlaere and Kenneth G. Rice

Hope College Georgia State University

Everett L. Worthington, Jr.

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Virginia Commonwealth University

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Prominent models and interventions designed to promote forgiveness have distinguished one’s decision
to forgive from achieving forgiveness as an end state, but because of a lack of a strong measure, there
is a weak research base on making a decision to forgive. Thus, in three studies, the authors developed
the Decision to Forgive Scale (DTFS) and examined evidence for its reliability and construct validity.
The article focused on distinguishing making a decision to forgive from achieved level of forgiveness.
Scores on the DTFS showed evidence of reliability, with Cronbach’s alpha coefficients ranging from .92
to .94, and a 1-week temporal stability coefficient of .68. Using several strategies, the authors demon-
strated that the DTFS is empirically distinct from the Transgression-Related Interpersonal Motivations
scale (TRIM; McCullough et al., 1998). Namely, a 3-factor confirmatory factor analysis that included the
DTFS and the 2 TRIM subscales showed excellent fit, suggesting these instruments assess 3 different
constructs. The DTFS was only moderately related to the TRIM subscales, was more strongly related to
stage of change than the TRIM, and predicted subsequent TRIM scores in a cross-lagged model. Finally,
although decisions to forgive generally suggested greater forgiveness, these constructs interacted to
predict existential distress. Namely, as decisional forgiveness increased, revenge was more strongly
related to existential distress. Overall, the DTFS shows considerable promise for further clinical and basic
research applications.

Keywords: forgiveness, measurement, positive psychology

Scientific research on forgiveness has burgeoned in recent years, Malcolm, 2008) and enrichment (Worthington et al., in press), or
with work linking forgiveness to mental and physical health and treatment of older adults (Allemand, Steiner, & Hill, 2013); (b)
identifying key correlates (for a review, see Fehr, Gelfand, & Nag., theory and research on forgiveness and appraisals of justice
2010), such as empathy (McCullough, Worthington, & Rachal, (Worthington, 2006) and on stress-related outcomes (Berry &
1997), benevolent attributions about the offender (Fincham, 2000), Worthington, 2001; Orcutt, 2006; Wade, Vogel, Liao, & Goldman,
and cultural influence (Hook, Worthington, & Utsey, 2009; 2008); and (c) theory and research on adapting of forgiveness to
Sandage, Hill, & Vang, 2003). Furthermore, there is a thriving client’s specific needs and culture (Davis et al., 2013; Hook et al.,
applied literature focused on forgiveness interventions (for a meta- 2009; Worthington et al., 2010).
analytic review, see Wade, Hoyt, Kidwell, & Worthington, 2014). One such idea is the distinction between making a decision to
Counseling psychologists have played prominent roles in this forgive and one’s actual level of forgiveness (Worthington,
work. For example, they have developed (a) forgiveness interven- Witvliet, Pietrini, & Miller, 2007). This distinction is included
tions for a variety of specific contexts, including group therapy
in much prominent theorizing and some research on promoting
(Wade et al., 2014), couples therapy (Greenberg, Warwar, &
forgiveness (Wade et al., 2014). For example, Worthington et
al. (2007) adduced research evidence to suggest that emotional
forgiveness was related more to physical and mental health than
This article was published Online First January 26, 2015. was a decision to forgive, but decisions were more related to
Don E. Davis, Department of Counseling and Psychological Services, relational and spiritual outcomes. However, despite the theo-
Georgia State University; Joshua N. Hook, University of North Texas; rizing and qualitative research support, the research base on
Daryl R. Van Tongeren, Hope College; Cirleen DeBlaere and Kenneth G. decisions to forgive has been hampered because prior attempts
Rice, Georgia State University; Everett L. Worthington, Jr., Virginia Com-
to measure the construct resulted in a scale that did not show
monwealth University.
Correspondence concerning this article should be addressed to Don E.
adequate divergent validity with other measures of forgiveness
Davis, Department of Counseling and Psychological Services, Georgia (Worthington, Hook, Utsey, Williams, & Neil, 2007). There-
State University, P.O. Box 3980, Atlanta, GA 30302-3980. E-mail: fore, the purpose of the present article is to develop a measure
ddavis88@gsu.edu of one’s decision to forgive and to conceptually and empirically

280
 DECISION TO FORGIVE 281

distinguish decisional forgiveness from forgiveness as an end mate decision to seek greater forgiveness, but she was now more
state. compassionate toward and patient with herself. Relieved of guilt,
From a stress-and-coping perspective (Exline, Worthington, she had greater energy to work toward forgiving while also setting
Hill, & McCullough, 2003; Worthington, 2006), offenses are in- appropriate boundaries and developing coping skills to engage in
terpersonal stressors that can lead to unforgiving emotions (e.g., gradual healing.
resentment, bitterness, anger) and motivations via appraisals of Although decisional forgiveness is considered an early step in
injustice, which are threatening. For example, consider Sarah who the prominent models of promoting forgiveness, the construct has
recently found out that her partner of 10 years has been involved a meager research base of only a handful of published studies
in a sexual affair for 6 months. Complicating her recovery, she (Holeman et al., 2011; Hook et al., 2013; Hook, Worthington,
feels pressure from her family and spiritual community to forgive. Utsey, Davis, & Burnette, 2012; Watkins et al., 2011). All used the
Feeling guilty because she believes that she should forgive, based same unpublished measure—The Decisional Forgiveness Scale
on her spiritual beliefs, she also feels afraid of being exploited (DFS; Worthington et al., 2007)—which has substantial limita-
again. If Sarah attended an intervention based on any of the leading tions. Most notably, that measure showed poor evidence of diver-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

models for promoting forgiveness, she would learn a more precise gent validity, being much too strongly related (i.e., rs as high as
This document is copyrighted by the American Psychological Association or one of its allied publishers.

definition of forgiveness. Forgiveness involves a reduction of .70; Hook, 2007) to the most widely used measure of forgiveness
negative (and perhaps an increase in positive) cognition, emotions, (i.e., the Transgression-Related Interpersonal Motivations Scale
and motivations toward an offender, and these changes often lead [TRIM]; McCullough et al., 1998). In addition, compared to the
to positive behavioral changes (Exline et al., 2003). Thus, forgive- TRIM, the DFS did not show a distinct pattern of relations with
ness is not excusing, justifying, exonerating, accepting, moving on, other constructs. Thus, our goal in the present article was to
seeking justice, forbearing, forgetting, or reconciliation. In addi- develop a measure of decisional forgiveness and differentiate it
tion, she would learn that, for severe betrayals for which hurt and conceptually and empirically from the TRIM.
resentment does not dissipate rapidly within a few weeks (Mc- One reason that the initial version of the DFS may have had
Cullough et al., 2003; McCullough, Luna, Berry, Tabak, & Bono, such poor evidence of divergent validity is that items were based
2010), forgiveness will likely require an effortful process that on a highly nuanced distinction between behavioral intentions and
involves two parts. forgiving motivations or emotions (Worthington et al., 2007).
The first part involves making a decision to forgive (DiBlasio, Indeed, items on the DFS (e.g., “I will try to help him or her if he
2000; Enright & Fitzgibbons, 2000; Exline et al., 2003). Making a or she needs help”; “I will try to get back at him or her” [reverse
decision to forgive involves committing to apply energy and effort scored]) appear conceptually very similar to items on the TRIM
to regulate negative emotions, thoughts, and behaviors until un- (e.g., “I’d find it difficult to act warmly toward him/her”; “I’m
forgiving emotions (e.g., hurt, resentment, bitterness) are substan- going to get even”). This fine-grained distinction between behav-
tially reduced. Indeed, for severe offenses, victims decide to for- ioral intentions and motivations did not hold up well empirically.
give without knowing how much energy, effort, or time will be In the current article, we focused on a more basic distinction: If
required, or if forgiveness is even possible. Decisions to forgive forgiveness is an endpoint that involves substantially reducing
may occur in discrete moments—like flipping a switch (e.g., negative (and in some situations increasing positive) thoughts,
making a clear and memorable decision to forgive)— especially if emotions, and motivations toward the offender, then deciding to
the decision is publicly acknowledged through a public ritual or forgive involves making a deliberate commitment to apply energy
communication (Hook et al., 2009). This step is important because and effort to move toward that endpoint. Indeed, consistent with
the remainder of the intervention assumes that the client is willing Prochaska and DiClemente’s (1982) Transtheoretical Model of
to work on forgiving the offense. Change, clients may enter treatment with a range of motivations to
The second part involves learning strategies to promote forgive- forgive: They may (a) feel vehemently opposed; (b) be considering
ness, based on the particular theoretical model. For example, in the the idea, but fear forgiveness will leave them vulnerable to exploi-
REACH Forgiveness model (Worthington, 2006), after deciding to tation; (c) have decided to forgive, but be exploring ways of
forgive, clients are taught that greater forgiveness is achieved increasing forgiveness; (d) have decided to forgive and be actively
through a process of emotional replacement, as the victim trans- employing coping strategies; or (e) feel that they have achieved
forms unforgiving emotions with positive, other-oriented emotions complete forgiveness. Forgiveness interventions usually seek to
such as empathy, sympathy, compassion, or love (Exline et al., allow clients to express this ambivalence and have their complex
2003). Thus, clients learn to alter their story about the offense in feeling validated before asking clients to move toward a decision
ways that reduce negative rumination and practice various activi- or commitment to working on the goal of forgiveness.
ties designed to promote other-oriented emotions, such as empathy This approach is informed and aligned with core themes of
or gratitude. counseling psychology such as strength-focused interventions
This distinction may be especially helpful for clients who feel (Fowers, 2005) and emphasis on common factors such as therapy
cultural pressure to forgive from their family or community (Davis alliance (e.g., Flückiger, Del Re, Wampold, Symonds, & Horvath,
et al., 2013; Hook & Worthington, 2008; Hook et al., 2009). In our 2012). For example, before moving into application of intervention
example, Sarah felt an obligation to forgive that will likely inter- activities, assessing the degree to which clients have made a
fere with healing. Although Sarah wants to forgive, it relieved decision to forgive helps the therapist or group leader negotiate
pressure when her counselor told her that forgiveness does not and clarify the tasks and goals of therapy to establish a strong
require repairing the relationship. The concept of decisional for- therapy alliance. Although forgiveness interventions have strong
giveness also helped her develop realistic expectations of herself. evidence of efficacy, there may be ways to enhance their potency
She accepted her current level of forgiveness. She made a legiti- by tailoring treatment to client’s degree of decisional forgiveness.
 282 DAVIS ET AL.

For example, clients with high levels of decisional forgiveness ined whether decision to forgive at Time 1 predicted unforgiveness
may not benefit from spending several hours completing activities at Time 2.
designed to reduce ambivalence and increase alignment with the Finally, although the DTFS should generally be negatively
therapy goal of forgiving. Likewise, clients with low decisional related to the TRIM subscales, we wanted to show that it is
forgiveness may not benefit from engaging in activities or inter- conceptually important when both are high (as might occur in the
ventions designed to teach skills to promote forgiveness if they beginning of an intervention when participants have made a deci-
have not made forgiveness a goal. Furthermore, promoting deci- sion to forgive, but they have not yet applied strategies to promote
sional forgiveness may relieve pressure to forgive in communities forgiveness). Accordingly, we recruited participants who had ex-
that strongly obligate forgiveness or reconciliation (Davis et al., perienced a recent betrayal by a religious/spiritual leader. We
2013; Hook et al., 2009), which may provide avenues for adapting picked this context because we believed it might reflect a partic-
interventions to the unique challenges that clients face based on the ular situation in which individuals would feel pressure to make a
way their community understands and practices forgiveness. These decision to forgive, yet still harbor unforgiving emotions toward
are exciting and promising possibilities, but they cannot be ex- the offender. We hypothesized not only would the DTFS be
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

plored until a measure to access decisional forgiveness has been negatively related to the TRIM, but these two scales would also
This document is copyrighted by the American Psychological Association or one of its allied publishers.

developed. interact with each other to predict existential distress. Specifically,

we expected that when scores were high on both the DTFS and the
TRIM, this would be associated with higher levels of existential
The Present Studies
distress than when TRIM scores were high, but DTFS scores were
The purpose of the present studies was to develop a brief yet low. Support for these five hypotheses would provide strong
psychometrically sound measure of making a decision to forgive. evidence for the distinction between decisional forgiveness and
Based on our definition, we created content-valid items designed forgiveness as an end state.
to assess one’s decision to forgive a specific offense. In Study 1,
we used factor analysis to establish the factor structure of the
Study 1
Decision to Forgive Scale (DTFS). We then sought to evaluate the
psychometric properties of the scale, including evaluating evi- In Study 1, we sought to develop the DTFS. We hypothesized
dence of reliability and construct validity. We were particularly that the DTFS would have a single-factor structure and would
focused on demonstrating that the DTFS was distinct conceptually show evidence of internal consistency. We also tested three hy-
and empirically from the TRIM (McCullough et al., 1998), given potheses focused on distinguishing the DTFS from the TRIM.
that this was the greatest limitation in prior attempts to assess the First, we hypothesized that a three-factor confirmatory factor anal-
construct. ysis with items from the DTFS and TRIM loading on separate
To that end, we examined the following five hypotheses. First, factors would show good fit. Second, we hypothesized that the DTFS
in Study 1, we used confirmatory factor analysis to test a model would be only moderately related with subscales of the TRIM. Third,
that loaded the decisional forgiveness items and TRIM items on we hypothesized that the DTFS would be more strongly related to
separate, correlated factors. We hypothesized that this model stage of change than the TRIM.
would show good fit, which would provide evidence that deci-
sional forgiveness is distinct from the two subscales of the TRIM
Method
(i.e., avoidance and revenge motivations).
Second, we examined the relation between the DTFS and the Participants and procedure. Participants were 432 under-
TRIM subscales. We hypothesized that the DTFS would be mod- graduates (298 women, 134 men) from a large, university in the
erately and positively related to the TRIM subscales (McCullough Southeastern United States. The mean age was 26.0 (SD ⫽ 7.2).
et al., 1998). Correlations of .60 or higher would cause us to The sample was diverse in terms of race/ethnicity (48.4% Black/
question whether we were measuring a sufficiently distinct con- African American; 31.3% White; 9.7% Asian/Pacific Islander;
struct. 5.3% Latino/a; and 4.6% Other; and 0.7% Did not respond). The
Third, we examined the correlations among the DTFS, the sample was predominately religious/spiritual (86.6%) and hetero-
TRIM, and a single-item measure of one’s stage of change regard- sexual (89.8%). Participants (all 18 years of age or older) com-
ing forgiveness (i.e., Prochaska & DiClemente, 1982). Namely, we pleted the survey in exchange for a small amount of course credit.
wanted to see that these two constructs were not just empirically After indicating consent online, participants completed demo-
distinct, but also show conceptually meaningful differences in their graphic questions. They then recalled a specific offense and com-
relations to other constructs. If the DTFS assesses the degree to pleted questionnaires assessing how they perceived the social
which one has decided to commit to a process of reducing negative context of the offense and how they coped with that offense.
emotions, thoughts, and motivations toward an offender, then it Measures
ought to be strongly related to this measure. Thus, we expected Decision to forgive. Informed by prior theory and research,
that the DTFS would be more strongly correlated to stage of the first author generated an initial list (N ⫽ 9) of content-valid
change than the TRIM. items based on our conceptualization of making a decision to
Fourth, in Study 2, we examined how DTFS scores were related forgive. Two of the coauthors (experts who had published fre-
to TRIM scores over time. Namely, decisional forgiveness is quently in the area of forgiveness research) reviewed this initial list
theorized to mobilize individuals to work toward the goal of and suggested revisions. The final item pool involved six items: “I
achieving greater forgiveness. We recruited participants who had have decided to forgive him or her”; “I made a commitment to
experienced a recent offense. In a cross-lagged model, we exam- forgive him or her”; “I have made up my mind to forgive him or
 DECISION TO FORGIVE 283

her”; “My choice is to forgive him or her”; “My choice is to factor loadings ranged from .57 to .96. The Cronbach’s alpha
release any negative feelings I have”; and “I have chosen not to coefficient was .93. Skew and kurtosis scores for the DTFS
intentionally harbor resentment toward him or her.” Participants were less than one (i.e., ⫺.91 and .22, respectively).
were asked to rate their agreement with items using 5-point ratings After evaluating the factor structure, next we tested three hy-
ranging from 1 (strongly disagree) to 5 (strongly agree). potheses related to construct validity of the DTFS. First, we
Forgiveness. The 12-item TRIM (McCullough et al., 1998) examined a three-factor model including items from the TRIM
was used to assess one’s experience of forgiveness. The TRIM has (i.e., five items for decision to forgive, five items for revenge
two subscales: Revenge (five items; e.g., “I’ll make him/her pay”) motivations, and seven items for avoidance motivations). Factors
and Avoidance (seven items; e.g., “I keep as much distance be- were modeled as correlated. This model also showed excellent fit,
tween us as possible”). Items are completed using 5-point ratings ␹2(116) ⫽ 258.71, p ⬍ .001, CFI ⫽ .97, RMSEA ⫽ .05 (95%
from (strongly disagree) to 5 (strongly agree). Higher scores (i.e., CI ⫽ .05, .06), SRMR ⫽ .04. Second, to examine our hypothesis
higher avoidance or revenge motivations) indicate greater unfor- that the DTFS would be moderately and negatively related to the
giveness (hence less forgiveness). The TRIM is the most widely TRIM subscales, we examined the factor correlations. As pre-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

used measure of forgiveness (Fehr et al., 2010). McCullough and dicted, the DTFS showed moderate relations with the revenge,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

colleagues (1998) reported Cronbach’s alpha coefficients for r ⫽ ⫺.39, p ⬍ .001 and avoidance, r ⫽ ⫺.47, p ⬍ .001 motiva-
scores on both subscales ranging from .85 to .93 and estimates of tions, suggesting that these measures assess related yet distinct
3-week temporal stability ranging from .44 to .65. Scores on the constructs.
scale showed evidence of construct validity, with subscales being Third, we examined whether the DTFS was more strongly
negatively correlated with other measures of forgiveness, relation- related to the stage of change item than the TRIM subscales. To
ship satisfaction, and commitment (McCullough et al., 1998). For test this prediction, we examined a model with the stage item
the present study, the Cronbach’s alpha coefficient was .90 for (manifest variable) regressed simultaneously on latent factors
revenge and .80 for avoidance. based on the DTFS and the TRIM, and we used the model test
Stage of change forgiveness. We adapted a single-item mea- command to compare the magnitude of these paths. The overall
sure that has been used regularly in motivational interviewing model showed excellent fit, ␹2(130) ⫽ 277.10, p ⬍ .001, CFI ⫽
research (e.g., Cook & Perri, 2004) to examine participants’ read- .97, RMSEA ⫽ .05 (95% CI ⫽ .04, .06), SRMR ⫽ .04. As
iness to work on forgiving an offense: “Please find the statement predicted, the stage of change item was more strongly related to
that best describes the way you feel right now about the person the DTFS (b ⫽ .81, p ⬍ .001) than it was to the Avoidance
who hurt you.” The item was rated using a 5-point rating from 0 (b ⫽ ⫺.45, p ⬍ .001; Wald test ⫽ 122.14, p ⬍ .001) or Revenge
(I am not considering forgiveness) to 4 (I have already forgiven). (b ⫽ ⫺.36, p ⬍ .001; Wald test ⫽ 110.27, p ⬍ .001) subscales.
We are aware of the limitations of singe items (e.g., difficulty Taken together, the results of Study 1 provided evidence for the
estimating reliability). However, given that no multiple item scales factor structure and estimated internal consistency of the DTFS.
existed to assess stage of change for forgiveness, we decided to Furthermore, in Study 1, we also provided initial evidence of the
begin by adapting this single item measure. Cook and Perri (2004) construct validity of the DTFS relative to the TRIM. These find-
found a .91 correlation between stages classifications based on a ings present initial evidence that making a decision to forgive is
single item and those based on multiple-item scales. different than the process of forgiving.

Results and Discussion Study 2

To refine the DTFS, we conducted a confirmatory factor In Study 2, DTFS to accumulate further evidence for the reli-
analysis. The covariance matrix was analyzed with MLR esti- ability of the scale, we examined the temporal stability of the
mation using MPLUS 7.11 (Muthén & Muthén, 2008). Missing DTFS. We examined additional evidence that decisional forgive-
data (less than 3% across items) was handled with full infor- ness is distinct from forgiveness as an end state. Prior work has
mation maximum likelihood. Items of the DTFS were used as demonstrated that TRIM scores tend to decline rapidly in the first
indicators of a single factor. Fit indices were used to evaluate few weeks after an offense (McCullough, Luna, Berry, Tabak, &
the fit of the model—the chi-square value, the comparative fit Bono, 2010). We hypothesized that decision to forgive would
index (CFI), and the root-mean-square-error-approximation predict TRIM scores one week later.
(RMSEA). As a rule of thumb, a CFI around .95, an RMSEA
equal or less than .06, and a standardized root mean squared
Method
residual (SRMR) equal to or less than .08 suggest good fit (Hu
& Bentler, 1999). The one-factor model showed adequate fit, Participants and procedure. Participants were 80 undergrad-
␹2(9) ⫽ 43.13, p ⬍ .001, CFI ⫽ .96, RMSEA ⫽ .09 (95% uate students (48 women, 32 men) from the same university as
confidence interval [CI] ⫽ .04, .13), SRMR ⫽ .04. Standard- Study 1. The mean age was 24.8 (SD ⫽ 7.14). The sample was
ized factor loadings ranged from .58 to .96. Given that the diverse in terms of race/ethnicity (40.0% Black/African American;
RMSEA was high, we dropped one poor item (i.e., I have 28.7% White; 12.5% Asian/Pacific Islander; 7.5% Latino/a; and
chosen not to intentionally harbor resentment toward him or 7.5% Other; and 3.8% did not respond; see Appendix for M and
her) that, based on modification indices, was too strongly SD across Studies 1–3). The sample was predominately religious/
related to another item. After dropping this item, the one-factor spiritual (89.6%) and heterosexual (92.5%). To be in the study,
model showed excellent fit, ␹2(5) ⫽ 9.14, p ⫽ .104, CFI ⫽ .99, participants had to identify an offense that had occurred to them
RMSEA ⫽ .04 (95% CI ⫽ .00, .09), SRMR ⫽ .01. Standardized within the past three months. They completed the survey in ex-
 284 DAVIS ET AL.

change for a small amount of course credit. They indicated consent Study 3
online and then completed a set of demographic questions. Next
they recalled a specific offense and completed the DTFS and the The primary goal of Study 3 was to examine whether the
TRIM. They were emailed a follow-up survey one week later that interaction between decisional forgiveness and forgiveness is con-
included the DTFS and the TRIM. ceptually important. Generally, making a decision to forgive and
Measures. Participants completed the DTFS (Cronbach’s al- forgiveness ought to align (i.e., negative correlation between the
pha coefficients were .93 at Time 1 and .92 at Time 2) and TRIM DTFS and TRIM subscales), but we wanted to see whether their
(Cronbach’s alpha coefficients for avoidance were .95 at Time 1 misalignment (as might occur early in an intervention when a
and .96 at Time 2; for revenge, they were .89 at Time 1 and .91 at client decides to forgive but has yet to employ strategies to work
Time 2). These measures were described in Study 1. toward this goal) would predict stress-related outcomes. Accord-
ingly, we recruited participants who had experienced a major
offense by a religious/spiritual leader within the last year. Specif-
Results and Discussion ically, we predicted that DTFS scores and TRIM subscales would
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

interact to predict existential distress, such that the highest level of

First, we hypothesized that the DTFS would show evidence of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

existential distress would be associated with higher scores on both

temporal stability. This hypothesis was supported. The 1-week
the DTFS (high decision) and the TRIM (still high unforgiveness).
test-retest correlation for the DTFS was .68 (p ⬍ .001). This
Namely, these are people who have made a strong and clear
suggests that around 46% of the variance in scores at Time 2 was
decision to forgive but have made little progress achieving for-
explained by scores at Time 1. For a state measure, which must
giveness. We expected these people to feel more distressed than
detect changes over time (see McCullough et al., 2010), the mag-
people with low forgiveness (i.e., high TRIM scores) but no
nitude of this correlation was consistent with our expectations.
intention to work on forgiving (i.e., low DTFS scores). This
Second, we hypothesized that decision to forgive would predict
subsequent forgiveness. We tested this hypothesis with a cross- pattern of findings would provide converging evidence that deci-
lagged path analysis on the manifest variables using MPLUS. sional forgiveness is distinct from one’s current level of forgive-
Missing data (less than 3%) was handled with full information ness.
maximum likelihood. This model simultaneously regressed Time 2
scores for the DTFS and the TRIM subscales on Time 1 scores, Method
controlling for correlations among DTFS and TRIM subscales at
Time 1, as well as at Time 2. Results are presented in Figure 1. As Participants. Participants were 335 undergraduates (206
predicted, simultaneously estimating other paths in the model, women, 129 men) from the same university as Studies 1 and 2. The
DTFS scores at Time 1 were negatively related to Revenge mean age was 24.8 (SD ⫽ 6.8). The sample was diverse in terms
(b ⫽ ⫺.39, p ⬍ .001) and Avoidance (b ⫽ ⫺.19, p ⫽ .029) scores of race/ethnicity (43.9% Black/African American; 25.1% White;
at Time 2. These results provide initial evidence that making a 13.1% Asian/Pacific Islander; 5.4% Latino/a; and 4.8% Other;
decision to forgive was associated with increased forgiveness over 1.5% did not respond). The sample was predominately religious/
time. spiritual (89.6%) and heterosexual (92.2%). To be in the study,
Taken together, the results of Study 2 provided additional evi- participants had to have experienced a major offense committed by
dence of the estimated reliability and construct validity of the a religious/leader. They received a small amount of credit for being
DTFS. Namely, the DTFS showed evidence for estimated tempo- in the study. They indicated consent online, completed demo-
ral stability. In regard to construct validity, decision to forgive graphic questions, recalled the offense by a religious leader and
predicted forgiveness scores a week later after controlling for prior completed several questionnaires regarding how they perceived
level of forgiveness. This is consistent with the idea that making a the social context of the offense and how they coped with that
decision to forgive may increase forgiveness and provides addi- offense, including the DTFS, the TRIM, and a measure of exis-
tional evidence for the distinction. tential distress.

ⴱ ⴱⴱ
Figure 1. Cross-lagged model in Study 2. T1 ⫽ Time 1; T2 ⫽ Time 2. p ⬍ .05. p ⬍ .01.
 DECISION TO FORGIVE 285

Measures. Participants completed the DTFS (␣ ⫽ .94) and was associated with greater existential distress (B ⫽ .12, p ⫽
TRIM (revenge ␣ ⫽ .81; avoidance ␣ ⫽ .90). These measures .023).
were described in Study 1. The second regression analysis examined avoidance as a mod-
Existential distress. Existential distress was assessed with the erator. The conditional main effects did not predict significant
7-item Negative Religious Coping subscale of the Religious Cop- variance in existential distress in Step 1 (R2 ⫽ .01, p ⫽ .141), and
ing Scale-Brief Version (RCOPE-B; Pargament, Koenig, Perez, the interaction term did not predict existential distress in Step 2
2000; Pargament et al., 1998). Items were rated from 0 (not at all) (⌬R2 ⫽ .00, p ⫽ .309).
to 5 (a great deal). Higher scores indicate greater struggle. An The results of Study 3 partially supported our hypothesis that
example item is, “Wondered whether God (or the Sacred) had decision to forgive would moderate the relation between forgive-
abandoned me.” The subscale demonstrated evidence of internal ness and existential distress. In addition, once again, decision to
consistency, with Cronbach’s alpha coefficients ranging from .69 forgive was only moderately related to the TRIM. Thus, this study
to .81 (Pargament et al., 1998). The RCOPE-B is one of the most provided additional evidence that the DTFS and the TRIM are
widely used measures of spiritual/existential struggle or alienation conceptually distinct constructs.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and has been robustly linked with physical and mental health
This document is copyrighted by the American Psychological Association or one of its allied publishers.

outcomes, including physiological markers of distress (Ai, Park,

General Discussion
Huang, Rodgers, & Tice, 2007; Ai, Seymour, Tice, Kronfol, &
Bolling, 2009; Pargament et al., 1998). For the present study, the Prominent models of promoting forgiveness have emphasized
Cronbach’s alpha coefficient was .91. the distinction between making a decision to forgive and forgive-
ness as an end state. Despite its central place in theory on forgive-
Results and Discussion ness, the research base on the construct is weak and in need of a
better measure. One attempt to measure the construct (e.g., Wor-
We hypothesized that the DTFS would be negatively related to thington et al., 2007) showed very poor evidence of discriminant
the TRIM subscales, but that DTFS scores would moderate the validity, being too strongly related to another measure of forgive-
relation between the TRIM subscales and existential distress, such ness. Therefore, we sought to address this need by developing the
that this relation between the TRIM subscales and existential DTFS.
distress would be stronger at higher DTFS scores. To test this Attempting to sharpen the definition (e.g., Exline et al., 2003),
hypothesis, we conducted a series of regression analyses with we conceptualized decision to forgive as a commitment to apply
centered conditional main effects entered in a first step and an effort and energy until unforgiveness (i.e., negative thoughts,
interaction term entered in a second step (Hayes, 2008). Missing emotions, and motivations toward an offender) is substantially
data at the item level (less than 3%) was handled by using the reduced. This definition is more in line with the pragmatic dis-
mean of available items on the same scale. Correlations among tinction many interventions include that is designed to help clients
variables are reported in Table 1. make a general decision to work on forgiving but also develop
The first regression analysis examined whether DTFS scores patience and perseverance for a potentially extended forgiveness
moderated the relation between revenge and existential distress. process. Accordingly, items were developed based on this refined
The results are presented in Table 2. The conditional main effects definition, and the resulting scale showed initial evidence of reli-
of revenge and decisional forgiveness accounted for 9% of the ability, temporal stability, and construct validity. Establishing ev-
variance in Step 1 (R2 ⫽ .09, p ⬍ .001), and the interaction term idence of construct validity is an ongoing process. Given problems
accounted for an additional 3% of the variance in Step 2 (⌬R2 ⫽ faced by Worthington et al. (2007), our primary focus in the
.03, p ⫽ .001). Regarding estimates at Step 2, revenge was a present studies was on evaluating whether the DTFS can be
positive predictor of existential distress (B ⫽ .38, p ⬍ .01), but the distinguished from the most widely used measure of forgiveness
conditional main effect of decision to forgive on existential dis- (i.e., the TRIM).
tress was not significant (B ⫽ .02, p ⫽ .638). The interaction term We used five strategies to evaluate this question. First, we
was significant (B ⫽ .14, p ⫽ .001). To interpret this effect, we examined a three-factor confirmatory factor analysis (with DTFS
conducted simple slopes analyses (see Figure 2), which revealed and TRIM items) and found excellent fit. Second, we examined
that at low levels of decision to forgive (i.e., ⫺1 SD), the effect of correlations between the DTFS and the TRIM, which showed only
revenge on existential distress was not significant (B ⫽ ⫺.08, p ⫽ moderate relations. Third, we compared the relation between the
.146). However, at high decision to forgive (i.e., ⫹1 SD), revenge DTFS and a criterion variable (i.e., stage of change) and found as
predicted that the DTFS was more strongly related to stage of
Table 1 change than was the TRIM. Fourth, we examined whether DTFS
Correlations of Variables in Study 3 scores predicted subsequent forgiveness in a cross-lagged model,
which it did. Fifth, although DTFS scores were related to greater
Variable 1 2 3 4 forgiveness, we compared participants who had high decisions to
ⴱ ⴱ
forgive with those with lower decisions to forgive. As we pre-
1. Decision to forgive 1.00 ⫺.46 ⫺.49 ⫺0.09
2. Revenge 1.00 .38ⴱ .30ⴱ
dicted, misalignment between decisions to forgive and forgiveness
3. Avoidance 1.00 0.10 were associated with greater existential distress, which shows
4. Existential distress 1.00 these two constructs do not necessarily align. Taken together, these
M 3.85 1.75 2.95 1.59 converging results provide a strong empirical evidence for the
SD 1.1 0.77 1.07 0.77 distinction between a decision to forgive and one’s level of for-
ⴱ
p ⬍ .01. giveness.
 286 DAVIS ET AL.

Table 2
Hierarchical Regression Analyses in Study 3

Step Variables B SE ␤ t p ⌬R2

Existential distress on revenge and DTFS

1 (Constant) 1.59 0.04 39.47 .000 .09ⴱ
Revenge 0.33 0.06 0.33 5.53 .000
Decision to forgive 0.04 0.04 0.06 1.01 .313
2 (Constant) 1.65 0.04 38.28 .000 .03ⴱ
Revenge 0.38 0.06 0.38 6.32 .000
Decision to forgive 0.02 0.04 0.03 0.47 .638
Decision to Forgive ⫻ Revenge 0.14 0.04 0.19 3.25 .001

Existential distress on avoidance and DTFS

1 (Constant) 1.59 0.04 37.84 .000 0.01
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Avoidance 0.05 0.05 0.07 1.12 .265

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Decision to forgive ⫺0.04 0.04 ⫺0.06 ⫺0.88 .378

2 (Constant) 1.62 0.05 34.52 .000 0.00
Avoidance 0.04 0.05 0.06 0.90 .370
Decision to forgive ⫺0.05 0.05 ⫺0.08 ⫺1.17 .242
Decision to Forgive ⫻ Avoidance 0.04 0.04 0.06 1.02 .309
Note. DTFS ⫽ Decision to Forgive Scale; TRIM ⫽ Transgression-Related Interpersonal Motivations Scale.
ⴱ
p ⬍ .01.

Our study had several limitations. First, all studies sampled to forgive among a broader array of constructs (e.g., motivations or
college students. Our samples were diverse in terms of race/ intentions associated with forgiveness).
ethnicity, but not in age or educational background. It will be Third, our studies only used self-report measures. The definition
important to establish norms for the DTFS based on samples more of forgiveness implies an intrapsychic change of cognitions, emo-
diverse in a variety of variables. In addition, the DTFS should be tions, motivations, and behaviors toward an offender. Although
normed with diverse treatment-seeking populations. self-report measures of forgiveness are useful, it may be important
Second, four of the five items on the DTFS included the word to supplement self-report measures of forgiveness with physiolog-
forgiveness, which may in some samples lead these items to be too ical measures of stress or behavioral indicators of forgiveness
strongly correlated with each other based on this common word. (Dorn, Hook, Davis, Van Tongeren, & Worthington, 2014).
The one item that did not explicitly mention forgiveness refers to Fourth, with the exception of Study 2, our studies used cross-
reduction of negative emotions, so the scale does not explicitly sectional designs. Even Study 2 was limited by the use of only two
reference increases in positive cognition, emotions, or motivations. time-points. Researchers might include the DTFS in future exper-
The scale may correlate more strongly with negative constructs imental or intervention studies. We hypothesize that successful
associated with forgiveness, such as anger or rumination, than it interventions promote rapid shifts in decisional forgiveness, which
does positive constructs such as empathy or conciliatory behaviors. might lead to subsequent increases in forgiveness over time. Fur-
The present study was focused on distinguishing the DTFS from thermore, we expect that clients who never make a decision to
the TRIM, but future work might work to situate making a decision forgive will be much less likely to benefit from forgiveness inter-
ventions. Being able to accurately identify these clients may help
researchers focus specific intervention strategies for clients who do
not want to forgive. Indeed, we are aware of no interventions that
specifically focus on clients who are highly opposed to the idea of
forgiveness, but it is very important to understand more about this
group. In some cases, perhaps these clients are reticent to forgive
for good reasons, and current interventions may have minimal
content focused on their particular reservations regarding forgive-
ness.

Conclusion
Although applied work on forgiveness has proliferated over the
last 20 years, there is still a need for more research on how to adapt
interventions to clients’ specific needs. The prominent strategies of
promoting forgiveness have focused on promoting a decision to
forgive, but very little is known about this construct. Advancing
research on different parts of the forgiveness process has important
Figure 2. Graph of interaction of commitment to forgive with revenge implications for how to promote forgiveness in clients who are
scores on spiritual struggle scores. ambivalent about whether they even want to forgive.
 DECISION TO FORGIVE 287

References Hook, J. N. (2007). Forgiveness, individualism, and collectivism. Unpublished

master’s thesis. Virginia Commonwealth University, Richmond, VA.
Ai, A. L., Park, C. L., Huang, B., Rodgers, W., & Tice, T. N. (2007). Hook, J. N., & Worthington, E. L., Jr. (2008). Decisional and emotional
Psychosocial mediation of religious coping styles: A study of short-term forgiveness: Implications for Christian spirituality and counseling.
psychological distress following cardiac surgery. Personality and Social Transformation, 2, 4 –5.
Psychology Bulletin, 33, 867– 882. http://dx.doi.org/10.1177/ Hook, J. N., Worthington, E. L., Jr., Davis, D. E., Watkins, D., Hui, E.,
0146167207301008 Luo, W., . . . Reyna, S. H. (2013). A China-New Zealand comparison of
Ai, A. L., Seymour, E. M., Tice, T. N., Kronfol, Z., & Bolling, S. F. (2009). forgiveness. Asian Journal of Social Psychology, 16, 286 –291. http://
Spiritual struggle related to plasma interleukin-6 prior to cardiac surgery. dx.doi.org/10.1111/ajsp.12033
Psychology of Religion and Spirituality, 1, 112–128. http://dx.doi.org/ Hook, J. N., Worthington, E. L., Jr., & Utsey, S. O. (2009). Collectivism,
10.1037/a0015775 forgiveness, and social harmony. The Counseling Psychologist, 37,
Allemand, M., Steiner, M., & Hill, P. L. (2013). Effects of a forgiveness 821– 847. http://dx.doi.org/10.1177/0011000008326546
intervention for older adults. Journal of Counseling Psychology, 60, Hook, J. N., Worthington, E. L., Jr., Utsey, S. O., Davis, D. E., & Burnette,
279 –286. http://dx.doi.org/10.1037/a0031839 J. L. (2012). Collectivistic self-construal and forgiveness. Counseling
Berry, J. W., & Worthington, E. L., Jr. (2001). Forgivingness, relationship
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and Values, 57, 109 –124. http://dx.doi.org/10.1002/j.2161-007X.2012

quality, stress while imagining relationship events, and physical and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

.00012.x
mental health. Journal of Counseling Psychology, 48, 447– 455. http://
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in
dx.doi.org/10.1037/0022-0167.48.4.447
covariance structure analysis: Conventional criteria versus new alterna-
Cook, C. L., & Perri, M., III. (2004). Single-item vs multiple-item mea-
tives. Structural Equation Modeling, 6, 1–55. http://dx.doi.org/10.1080/
sures of stage of change in compliance with prescribed medications.
10705519909540118
Psychological Reports, 94, 115–124. http://dx.doi.org/10.2466/pr0.94.1
McCullough, M. E., Fincham, F. D., & Tsang, J. A. (2003). Forgiveness,
.115-124
forbearance, and time: The temporal unfolding of transgression-related
Davis, D. E., Worthington, E. L., Jr., Hook, J. N., & Hill, P. C. (2013).
interpersonal motivations. Journal of Personality and Social Psychol-
Research on religion/spirituality and forgiveness: A meta-analytic re-
ogy, 84, 540 –557. http://dx.doi.org/10.1037/0022-3514.84.3.540
view. Psychology of Religion and Spirituality, 5, 233–241. http://dx.doi
McCullough, M. E., Luna, L. R., Berry, J. W., Tabak, B. A., & Bono, G.
.org/10.1037/a0033637
(2010). On the form and function of forgiving: Modeling the time-
DiBlasio, F. A. (2000). Decision-based forgiveness treatment in cases of
forgiveness relationship and testing the valuable relationships hypothe-
marital infidelity. Psychotherapy: Theory, Research, Practice, Training,
sis. Emotion, 10, 358 –376. http://dx.doi.org/10.1037/a0019349
37, 149 –158. http://dx.doi.org/10.1037/h0087834
McCullough, M. E., Rachal, K. C., Sandage, S. J., Worthington, E. L., Jr.,
Dorn, K., Hook, J. N., Davis, D. E., Van Tongeren, D. R., & Worthington,
Brown, S. W., & Hight, T. L. (1998). Interpersonal forgiving in close
E. L., Jr. (2014). Behavioral methods of assessing forgiveness. The
relationships: II. Theoretical elaboration and measurement. Journal of
Journal of Positive Psychology, 9, 75– 80. http://dx.doi.org/10.1080/
Personality and Social Psychology, 75, 1586 –1603. http://dx.doi.org/
17439760.2013.844267
10.1037/0022-3514.75.6.1586
Enright, R. D., & Fitzgibbons, R. P. (2000). Helping clients forgive.
Washington, DC: American Psychological Association. http://dx.doi McCullough, M. E., Worthington, E. L., Jr., & Rachal, K. C. (1997).
.org/10.1037/10381-000 Interpersonal forgiving in close relationships. Journal of Personality and
Exline, J. J., Worthington, E. L., Jr., Hill, P., & McCullough, M. E. (2003). Social Psychology, 73, 321–336. http://dx.doi.org/10.1037/0022-3514
Forgiveness and justice: A research agenda for social and personality .73.2.321
psychology. Personality and Social Psychology Review, 7, 337–348. Muthén, B., & Muthén, L. (2008). Mplus user’s guide. Los Angeles, CA:
http://dx.doi.org/10.1207/S15327957PSPR0704_06 Muthén and Muthén.
Fehr, R., Gelfand, M. J., & Nag, M. (2010). The road to forgiveness: A Orcutt, H. K. (2006). The prospective relationship of interpersonal forgive-
meta-analytic synthesis of its situational and dispositional correlates. ness and psychological distress symptoms among college women. Jour-
Psychological Bulletin, 136, 894 –914. http://dx.doi.org/10.1037/ nal of Counseling Psychology, 53, 350 –361. http://dx.doi.org/10.1037/
a0019993 0022-0167.53.3.350
Fincham, F. D. (2000). The kiss of the porcupines: From attributing Pargament, K. I., Koenig, H. G., & Perez, L. M. (2000). The many methods
responsibility to forgiving. Personal Relationships, 7, 1–23. http://dx.doi of religious coping: Development and initial validation of the RCOPE.
.org/10.1111/j.1475-6811.2000.tb00001.x Journal of Clinical Psychology, 56, 519 –543. http://dx.doi.org/10.1002/
Flückiger, C., Del Re, A. C., Wampold, B. E., Symonds, D., & Horvath, (SICI)1097-4679(200004)56:4⬍519::AID-JCLP6⬎3.0.CO;2-1
A. O. (2012). How central is the alliance in psychotherapy? A multilevel Pargament, K. I., Smith, B., Koenig, H. G., & Perez, L. (1998). Patterns of
longitudinal meta-analysis. Journal of Counseling Psychology, 59, 10 – positive and negative religious coping with major life stressors. Journal
17. http://dx.doi.org/10.1037/a0025749 for the Scientific Study of Religion, 37, 710 –724. http://dx.doi.org/
Fowers, B. J. (2005). Virtue and psychology: Pursuing excellence in 10.2307/1388152
ordinary practices. Washington, DC: American Psychological Associa- Prochaska, J. O., & DiClemente, C. C. (1982). Transtheoretical therapy:
tion. http://dx.doi.org/10.1037/11219-000 Toward a more integrative model of change. Psychotherapy: Theory,
Greenberg, L. J., Warwar, S. H., & Malcolm, W. M. (2008). Differential effects of Research, & Practice, 19, 276 –288. http://dx.doi.org/10.1037/h0088437
emotion-focused therapy and psychoeducation in facilitating forgiveness and Sandage, S. J., Hill, P. C., & Vang, H. (2003). Toward a multicultural
letting go of emotional injuries. Journal of Counseling Psychology, 55, 185– positive psychology: Indigenous forgiveness and Hmong culture. The
196. http://dx.doi.org/10.1037/0022-0167.55.2.185 Counseling Psychologist, 31, 564 –592. http://dx.doi.org/10.1177/
Hayes, A. F. (2008). Introduction to mediation, moderation, and condi- 0011000003256350
tional process analysis: A regression-based approach. New York, NY: Wade, N. G., Hoyt, W. T., Kidwell, J. E., & Worthington, E. L., Jr. (2014).
Guilford Press. Efficacy of psychotherapeutic interventions to promote forgiveness: A
Holeman, V. T., Dean, J. B., DeShea, L., & Duba, J. D. (2011). The meta-analysis. Journal of Consulting and Clinical Psychology, 82, 154 –
multidimensional nature of the quest construct forgiveness, spiritual 170. http://dx.doi.org/10.1037/a0035268
perception, and differentiation of self. Journal of Psychology and The- Wade, N. G., Vogel, D. L., Liao, K. Y.-H., & Goldman, D. B. (2008). Measuring
ology, 39, 31– 43. state-specific rumination: Development of the Rumination about an
 288 DAVIS ET AL.

Interpersonal Offense Scale. Journal of Counseling Psychology, 55, Worthington, E. L., Jr., Berry, J. W., Hook, J. N., Davis, D. E., Scherer, M.,
419 – 426. http://dx.doi.org/10.1037/0022-0167.55.3.419 Griffin, B. J., . . . Campana, K. L. (in press). Forgiveness-reconciliation
Watkins, D. A., Hui, E. K., Luo, W., Regmi, M., Worthington, E. L., Jr., and communication-conflict-resolution interventions versus rested con-
Hook, J. N., & Davis, D. E. (2011). Forgiveness and interpersonal trols in early married couples. Journal of Counseling Psychology.
relationships: A Nepalese investigation. The Journal of Social Psychol- Worthington, E. L., Jr., Hook, J. N., Utsey, S. O., Williams, J. K., & Neil, R. L.
ogy, 151, 150 –161. http://dx.doi.org/10.1080/00224540903368541 (2007, October 5). Decisional and emotional forgiveness. Paper presented at
Worthington, E. L., Hunter, J. L., Sharp, C. B., Hook, J. N., Van Tongeren, the International Positive Psychology Summit, Washington, DC.
D. R., Davis, D. E., . . . Monforte-Milton, M. (2010). A psychoeduca- Worthington, E. L., Jr., Witvliet, C. V. O., Pietrini, P., & Miller, A. J.
tional intervention to promote forgiveness in Christians in the Philip- (2007). Forgiveness, health, and well-being: A review of evidence for
pines. Journal of Mental Health Counseling, 32, 75–93. emotional versus decisional forgiveness, dispositional forgivingness,
Worthington, E. L., Jr. (2006). Forgiveness and reconciliation: Theory and and reduced unforgiveness. Journal of Behavioral Medicine, 30,
application. New York, NY: Brunner-Routledge. 291–302. http://dx.doi.org/10.1007/s10865-007-9105-8
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Appendix
Norms Across Studies 1–3 by Race

Black Asian Latino/a White

Variable n M SD n M SD n M SD n M SD

Decision to forgive 381 19.71a 4.98 96 17.16b 5.96 47 19.81ab 4.94 240 18.73a 5.99
Note. Different subscripts indicate that means were different from each other using a Bonferonni post hoc test at p ⬍ .05.

Received September 8, 2014

Revision received November 3, 2014
Accepted November 3, 2014 䡲