ChapterPDF Available

Measures of Forgiveness


Abstract and Figures

We review 14 measures of forgiveness, including measures of state forgiveness, forgivingness as a disposition or trait, self-reported state self-forgiveness, and trait self-forgivingness. The state forgiveness measures include self-report measures, chemical measures of state unforgiveness and forgiveness, measures of peripheral physiology (i.e., blood pressure, heart rate, skin conductance, and electrical activity of facial muscles) to supplement self-reports – unforgiveness, measures of brain activity associated with forgiveness, and behavioral indices of forgiveness. We also review a ten-item scale and a five-scenario scale of dispositional forgivingness. Self-forgiveness may also be assessed at state (one measure) and trait levels (one measure). These measures of forgiveness and their various targets (self or other) and different levels (state or trait) are generally strong – with considerable evidence supporting estimated reliabilities and construct validities.
No caption available
Content may be subject to copyright.
Provided for non-commercial research and educational use only.
Not for reproduction, distribution or commercial use.
This chapter was originally published in the book Measures of Personality and Social Psychological
Constructs, published by Elsevier, and the attached copy is provided by Elsevier for the author’s benefit
and for the benefit of the author’s institution, for non-commercial research and educational use including
without limitation use in instruction at your institution, sending it to specific colleagues who know you,
and providing a copy to your institution’s administrator.
All other uses, reproduction and distribution, including without limitation commercial reprints, selling
or licensing copies or access, or posting on open internet sites, your personal or institution’s website or
repository, are prohibited. For exceptions, permission may be sought for such use through Elsevier’s
permissions site at:
From Everett L. Worthington, Caroline Lavelock, Charlotte vanOyen Witvliet, Mark S. Rye, Jo-
Ann Tsang and Loren Toussaint, Measures of Forgiveness: Self-Report, Physiological, Chemical,
and Behavioral Indicators. In: Gregory J. Boyle, Donald H. Saklofske and Gerald Matthews, editors,
Measures of Personality and Social Psychological Constructs. Oxford: Academic Press, 2014, pp. 474-
ISBN: 978-0-12-386915-9
Copyright © 2014 Elsevier Inc.
Academic Press.
Measures of Forgiveness: Self-Report,
Physiological, Chemical, and
Behavioral Indicators
Everett L. Worthington, Jr.
, Caroline Lavelock
, Charlotte vanOyen Witvliet
Mark S. Rye
, Jo-Ann Tsang
and Loren Toussaint
Virginia Commonwealth University, Richmond, VA, USA;
Hope College, MI, USA;
Skidmore College, Saratoga
Springs, NY, USA;
Baylor University, Waco, TX, USA;
Luther College, Decorah, IA, USA
Forgiveness is one way that people can respond to transgressions. Because people frequently misunderstand,
hurt, and offend each other, forgiveness is important in understanding romantic relations, families, work organi-
zations, crimes, and political and economic interactions. It is vital to know whether people who have been
offended hold grudges that are filled with resentment and vengeful motives, seek to avoid their offenders, cut off
relationships, or seek to restore more positive motives and emotions toward the offender and to forgive.
The most common models of forgiveness include the stress-and-coping model of forgiveness (
Strelan & Covic,
2006; Worthington, 2006), the process model of forgiveness (Enright & Fitzgibbons, 2000
), and the evolutionary
model of forgiveness (
McCullough, 2008). We will rely most heavily on the stress-and-coping model, but the
models are complementary. A transgression is a moral and relational injustice that violates people’s physical,
psychological, relational, and/or spiritual boun daries. Unforgiveness is an emotional and motivational state
toward an offender that includes grudges, revenges, and other negative responses. If the injustice is not dealt
with, or if rumination is prolonged, then unforgiveness can occur. A transgression is not only an injustice, but is
also a stressor that prompts stress responses. If we understand this offence using stress-and-coping theory
Lazarus, 1999), as has been proposed by Strelan and Covic (2006) and Worthington (2006), we might expect that
people who are transgressed against engage in appraisals. First, threat appraisal is first engaged, and if the trans-
gression is seen as threatening, a secondary appraisal of coping adequacy is made.
Worthington (2006; see also
Exline, Worthington, Hill, & McCullough, 2003) suggested that a third appraisal is made regarding injustice. The
injustice gap is subjectively the net injustice currently experienced given an accounting of subsequent events that
exacerbate or mitigate it. For instance, an offender’s denial of responsibility and refusal to show remorse exacer-
bates the sense of injustice, but an offender’s abject apology and offer of restitution mitigates the sense of
People who experience chronic stress or frequent intermittent stressors experience physical, psychological,
interpersonal, and spiritual effects. The physical, emotional, and psychological responses are considered stress
responses (
Lazarus, 1999). These tend to trigger attempts to cope with the stress, the appraisals, and the trans-
gression, and to modify the stressful situation to mitigate its effects and protect from future harm. Forgiveness is
one of many possib le responses (
Strelan & Covic, 2006; Worthington, 2006). Others include seeking justice, reap-
praising the transgression as less harmful, reattributing the motives of the transgressor to be less pernicious, or
accepting the transgression and moving on with life to name a few.
Measures of Personality and Social Psychological Constructs.
DOI: © 2015 Elsevier Inc. All rights reserved.
Authors personal copy
Forgiveness is a response that holds an offender responsible for an offence while replacing negative thoughts,
emotions, and behaviors toward the offender with prosocial responses. Forgiveness is facilitated by empathy and
compassion for the humanity of the offender, but forgiveness also results in less negative or even net positive
emotions and motivations toward the offender. The nature of the process and outcomes are related to man y fac-
tors, including transgress ion characteristics, relational context of the offence, and whether one’s forgiveness is a
decisional commitment or a change of heart (i.e., motivation and emotion). Forgiveness in response to a particu-
lar offenc e is characterized as state forgiveness, while cross situational consistency in forgiving is characterized
as a forgivi ng character trait or disposition.
The adequacy of coping affects people’s lives. This is especially true when coping responses are frequently
needed, such as when transgressions are experienced frequently or if rumination keeps re-introducing events.
Many effects of forgiveness have been documented. The sequelae can involve improved physical health
Worthington, Witvliet, Pietrini, & Miller, 2007b), mental health (see Toussaint & Webb, 2005), relationships
Rusbult, Hannon, Stocker, & Finkel, 2005), and spiritual well-being (Davis, Worthington, Hook, & Hill, 2013).
Psychologically, forgiveness involves ‘an intra-individu al prosocial change toward a perceived transgressor
that is situated within a specific i nterpersonal context’ (
McCullough, Pargament, a nd Thoresen 2000a,b p. 9).
This broad definition captures much that is agreed upon by researchers. Forgiveness happens within an indi-
vidual but in interpersonal context. Thus, a host of individual variables might affect forgiveness, such as
attachment styles, attributional styles, personal beliefs and values, and personality. Additionally, the context
within which the offence occurred includes the offender and his or her experi ences, the transgression, the rela-
tionship between offender and victim, behavioral interactions and communica tions around the transgression,
the spiritual or religious stance of the victim (Davis, Hook, & Worthington, 2008
), and other situational condi-
From this complex set of variables we can make the distinction between self-forgiveness, and forgiveness of
others (see
Toussaint & Webb, 2005; Webb, Robinson, & Brower, 2011). Forgiveness of oneself typically looks at
forgiveness from the point of view of the offender and thus is different from forgiving anoth er person. For exam-
ple, when a person does wrong and considers self-forgiveness, then regret and remorse over wrongdoing, guilt,
contrition, and relational repair steps are paramount (Fisher & Exline, 2006
). In contrast, forgiving a transgressor
is likely to be most concerned more with resentment, bitterness, hostility, hatred, anger, and fear (
Enright &
Fitzgibbons, 2000
Forgiveness is distinct from other related experiences that occur in the presence of transgressions. For example,
Worthington (2005) reviewed 20 research teams’ definitions of forgiveness. He summarized that all teams stated
that forgiveness is not condoning, approving of, excusing, or justifying wrongdoing, but require s that the forgiver
affirm that a moral wrong was committed. Forgiveness is also not reconciling a relationship, which takes two
people; forgiving can be done even if an offender is absent or dead. Forgiveness is not saying, ‘I forgive you,’
which can sound sin cere but be a set up for back-stabbing. Nor is not saying, ‘I forgive you’ the same as not for-
giving; a victim who has forgiven might not communicate this to the offender so that the offender is manipulated
into feeling guilty (see
Baumeister, Exline, & Sommer, 1998). Or the forgiver might simply lack the opportunity
to express forgiveness or simply be reticent.
The extent to which forgiveness invo lves positive feelings toward the offender may depend upon the nature of
the relationship.
Worthington (2005) observed that typically researchers who study forgiveness in relationships
(a) among strangers (e.g., a criminal perpetrator); (b) with people in non-valued relationships (e.g., a hostile boss,
or former abusive landl ord); or (c) in cases that the relationship is doomed (e.g., ex-partners or estranged spouses
who have cut off contact) treat forgiveness as simply eliminating negative, other-diminishing responses toward
the offender. Thus, complete forgiveness is feeling no negative emotions, having no vengeful or avoidant
motives, engaging in few angry, vengeful, anxious, or depressive ruminations about the transgress ion, and hav-
ing made a firm decision about one’s behavioral intentions that do not include vengeance. However, in relation-
ships that are valued and continuing such as hurt feelings by a valued marriage partner, family member,
friend, or work colleague whom one cannot avoid the participants are not usually content to solely eliminate
negativity. Rather, they seek to reach a positive emotional balance (
Paleari, Regalia, & Fincham, 2009 ).
Theorists have emphasized different aspects of the prosocial changes involved in forgiveness. Various theoreti-
cians emphasize affect, cognition, and behavior (
Enright & Fitzgibbons, 2000); motivations (McCullough, Exline, &
Baumeister 1998a
); decisions (DiBlasio, 1998); emotions (Worthington, 2006) and physiological reactivity patterns
Witvliet, Ludwig, & Vander Laan, 2001); and patterns of brain activation (Farrow et al., 2001).
Finally, McCullough and colleagues have suggested that forgiveness requires measurement at multiple assess-
ment points (see
McCullough, Fincha m, & Tsang, 2003). They have argued that two people who scored the same
Authors personal copy
on a forgiveness instrument might have begun at different places. Person A might have become less forgiving
over time, but Person B might have become more forgiving over time. Also, attitudes and moods can fluctuate
following a transgression casting doubt on the validity of a single measurement. Thus, they suggested that trend
forgiveness might describe the best fit of a line through repeated assessments, and temporal forgiveness might assess
daily variations from the best-fit curve.
This chapter focuse s on measures of forgiveness that can be grouped into eight categories on the basis of their
content. We review 16 measures across those eight categories.
Forgiveness of a Human Offender for a Transgression
1. Enright Forgiveness Inventory (
Enright & Rique, 2004)
2. Transgression-Related Inventory of Motivations (
McCullough, Rachal, Sandage, Worthington,
Brown, & Hight, 1998b
3. Rye Forgiveness Scale (
Rye et al., 2001)
4. Decisional Forgiveness Scale (DFS) and Emotional Forgiveness Scale (EFS) (
Worthington, Hook, Utsey,
Williams, & Neil, 2007
Chemical Measures
5. Cortisol
6. Oxytocin
Peripheral Physiological Measures
7. Heart Rate
8. Blood Pressure
9. Heart Rate Variability
10. Facial Electromyographic Measures
Central Physiological Measures
11. fMRI
Behavioral Measures
12. Resource Distribution Measure of State Forgiveness (
Carlisle et al., 2012)
13. Defection in a Prisoner’s Dilemma as a Behavioral Measure of Forgiveness (
Exline, Baume ister, Bushman,
Campbell, & Finkel, 2004
, Study 5)
Disposition of Being Forgiven Consistently across Transgression Events , Time, and Relationship s
14. Trait Forgivingness Scale (Berry, Worthington, O’Connor, Parrott, & Wade, 2005)
Measures of State Self-Forgiveness
15. State Self-Forgiveness Scale (SSFS;
Wohl, DeShea, & Wahkinney, 2008)
Measures of Trait Self-Forgi vingness
16. Heartland Forgiveness Scale (HFS; Thompson et al., 2005)
The measurement of forgiveness is well-developed for research purposes. Give n that forgiveness definitions
emphasize different aspects of forgiveness, measures have been developed that capture the aspects that are
highlighted in the competing definitions. Thus, researchers can focus on theoretically relevant aspects of forgive-
ness (or unforgiveness). We suggest that forgiveness is an internal experience that involves affect, cognition,
Authors personal copy
behavioral intentions, unforgiving motives, benevolent motives, positive aspects and negative aspects. One can
find measures with good psychometric properties that adequately assess eac h component of forgiveness.
In this chapte r, we review the psychometric adequacy of five self- report instruments of different aspects of for-
giving transgressions. The most commonly used measures include the Transgression-Related Interpersonal
Motivations (TRIM; McCullough et al., 1998b) and the Enright Forgiveness Scale (
Enright & Rique, 2004). The
Rye Forgiveness Scale and Decisi onal and Emotional Forgiveness Scale s have been used far less but are important
for researchers who differentiate those theoretically relevant components of the experience of forgiving.
Some chemical markers of forgiveness have also been investigated. Cortisol and oxytocin are potential chemical
measures of forgiveness, but other chemical markers have also been assessed (see
Temoshok & Wald, 2005). Most of
the chemical markers are based on a stress-and-coping theory of forgiveness (
Strelan & Covic, 2006; Worthington,
2006). However, the measurement of oxytocin is also derived from evolutionary (McCullough, 2008
) and relationship
Paleari et al., 2009) theories. We focus on cortisol and oxytocin.
Both unforgiveness and forgiveness are manifested in the body. Thus, physiological indices of stress and relief
from stress are markers that can suggest when forgiveness has or has not occurred. The problem with physiologi-
cal indices is usually their lack of specificity. If unforgiveness is conceptualized as a stress response, the sympa-
thetic nervous system is involved in experiencing transgressions and in remembering or ruminating about them.
If people forgive, the stress response is ameliorated. However, a stress response could also be ameliorated
through distraction, psychological defenses, and even successful revenge. Central nervous system responding
might also be patterned when one experiences transgressions and forgi ves. We review research by
Farrow et al.
as they try to disentangle the central nervous system patterns associated with ho lding grudges, experienc-
ing empathy, and making judgments of forgivability.
Forgiveness occurs in interpersonal contexts. People act on their experiences of unforgiveness or forgiveness.
Some experimental paradigms allow us to infer forgiveness. Behavioral manifestations of internal experiences of
forgiveness are often put forth as a way to troubleshoot soc ial desirability concerns of self-report measures, but
they are still open to falsification by a prevaricating victim. Nevertheless, we review several laboratory methods
for assessing forgiveness-consistent behavior. These include a resource distribution task as a measure of forgive-
ness (
Carlisle et al., 2012) and defection within a Prisoner’s Dilemma game (Exline et al., 2004).
In addition, we review two measures of trait forgi vingness. One measure involves responses to 10 face-valid
items (
Berry et al., 2005). A second (Thompson et al., 2005) includes three subscales dispositional forgiveness
of others, of self, and of situations.
Whereas forgiveness of others occurs when one has been hurt or offended, forgiveness of self typically has
more to do with being an offender of others in ways that results in guilt, shame, remorse, and lowered self-
esteem. Self-forgiveness might also occur after self- condemnation for failing to live up to one’s own standards.
Because interest in self-forgiveness lagged behind interest in forgiving others (
Fehr, Gelfand, & Nag, 2010),
so did development of theories (see
Fisher & Exline, 2006; Hall & Fincham, 2005 ) and measures. We review
one measure of state forgiveness of self (Wohl et al., 2008
) and one measure of trait self-forgivingness
Thompson et al., 2005).
Enright Forgiveness Inventory (EFI)
(Enright & Rique, 2004; Subkoviak, Enright, Wu, & Gassin, 1995).
The EFI assesses the degree to which an individual forgives an offender who has hurt him or her deeply
and unfairly.
The EFI was one of the first measures of forgiveness developed. The EFI was constructed to assess positive
and negative affect, behavio r, and cognition toward to offender as a consequence of a particular offence, though
it can be used without reference to the particular offence to assess forgiveness of the offender.
Enright’s team generated 150 items for the EFI 25 for each of six key definitional domains including: absence
of negative affect (e.g. hostility), presence of positive affect (e.g., warmth), absence of negative cognition
Authors personal copy
(e.g., disapproval), presence of positive cognition (e.g., favorability), absence of negative behavior (e.g., avoid-
ance), and presence of positive behavior (e.g., considerateness). They administered items to 197 college students
and 197 of their same-sex parent. Of the 150 items , 60 were retained (10 on each domain). The final EFI is scored
on affect, belief, and behavior 20-item subscales (each combining 10 positive and 10 negative items) and a
60-item total sco re. Example items from the EFI include: (a) affect: ‘I feel warm toward him/her’; (b) behavior:
‘Regarding the person I do or would show friendship’; and (c) cognition: ‘I think he or she is horrible.’
Additional example items and details of the EFI are available in other sources (see
Enright & Fitzgibbons, 2000).
The 60 items each correlated abov e .65 with its subscale score and below .17 with social desirability. Items are
rated on a 6-point response scale ranging from 1 5 strongly disagree to 6 5 strongly agree. High values indicate
more forgiveness. Subscale scores range from 20 to 60 and the total score ranges from 60 to 360.
Additional items on the EFI assess hurt severity, source of the hurt (e.g., child, spouse, etc.), time since the
hurt, and whether the offender is still alive. Five items are used to assess ‘pseudo-forgiveness,’ thought of as a
disingenuous form of forgiveness. A single, face-valid item assesses the respondent’s self-report of offered for-
giveness (1 5 not at all ;55 complete forgiveness). The EFI begins with a written description of the offence and all
items reference that specific event. Thus, the EFI is an event- or person-specific assessment. The EFI is used for
young adolescents, young adults, and adults. A fifth-grade reading level is required.
The original sample consisted of 197 Midwestern college students and 197 same-sex parents (
Subkoviak et al.,
). Mean college student and parent a ge were 22.1 (SD 5 4.7) and 49.6 (SD 5 9.6), respectively. Sexes were
equally represented. Same-s ex parents were used to control variation in education, social class, and values that
were anticipate d to be shared among parents and children.
Internal Consistency
Cronbach alpha coefficients, representing hundreds of individuals are typically greater than .95 (
Enright &
Rique, 2004; Subkoviak et al., 1995).
Temporal stability estimates over four weeks range from .67 to .91 using 36 college students (
et al., 1995
The positive affect, negative affec t, and EFI total scores were correlated with a state-measure of anxiety in both
college-students (r
2.40) and same-sex parents (r
2.50; Subkoviak et al., 1995). EFI-anxiety correlations
increased for respondents reporting deep hurt by their offender.
Enright and Rique (2004) reported that the
EFS positive ly correlated highly with the Wade Forgiveness Scale (
Brown, Gorsuch, Rosik, & Ridley, 2001;
Wade, 1990).
Divergent/Discrim inant
The EFI was not correlated with depression in healthy samples (
Subkoviak et al., 1995; no exact correlation
reported). The relationship between religion and the EFI was unexpectedly low (r [392] 5 .09; cf. a meta-analysis
of forgiveness with both spirituality and religion;
Davis, Worthington, Hook, & Hill, 2013); investigators have
addressed several hypotheses for why this might be the case (e.g.,
McCullough & Worthington, 1999).
Construct/Factor Analytic
The factor structure of the EFI has been examined and, although the scale authors describe this as a confirma-
tory factor analysis, it is clearly exp loratory in nature (
Enright & Rique, 2004). Exploratory factor analysis (EFA)
was conducted on 394 college stu dents and their same-sex parents using principal axis factoring. Three factors
were set for extraction with oblique rotation (exact type unsp ecified). The resulting solution favored a one-factor
interpretation with almost 60 percent of the item variance being explained by the first factor (a combination of
affect, behavior, and cognition items) and little more than 10 percent being explained by additional factors
(neither of which had item s clearly loadi ng on those factors but not also loading high on factor 1). An EFA with
Authors personal copy
Belgium participants similarly suggested a one-factor solution (Orathinkal, Vansteenweg en, Enright, &
Stroobants, 2007
). We conclude that there is strong support for the EFI’s assessment of forgiveness but less
support for subscales.
Enright and Rique (2004) suggest that numerous intervention studies have shown that the EFI is sensitive to
clinical change in forgiveness in patient samples where hurts have been deep and psychologically compromising
(for a meta-analysis, see (
Wade, Hoyt, Kidwell, & Worthington, 2014). Studies using the EFI with emotionally dis-
turbed patients provide strong evidence for its usefulness in clinical settings. Neither EFI subscale nor total score
is correlated (r
, .02) with the Marlowe-Crowne Social Desirability Scale (Enright & Rique, 2004; Subkoviak
et al., 1995).
The EFI can be purchased from Mind Garden at: (Retrieved
January 9, 20 14).
Results and Comments
The EFI is credited with offering researchers and practitioners one of the first tools for assessment of interper-
sonal forgiveness, and is one of the two most widely used measures of forgiveness . It is available in six languages
from at least six countries: Austria, Brazil, Israel, Korea, Norway, and Taiwan. The EFI manual contains norms
for these countries and for the United States. The manual and website provide conflicting information about the
construct validity of the translated versions of the EFI.
The EFI has proven especially useful for clinical applications and is sensitive to clinical change in forgiveness. It
must be considered the premier instrument to use in clinical contexts for assessment leading to diagnosis and prog-
nosis with disorders involving unforgiveness and anger, or disorders that involve large amounts of rumination.
The EFI’s length may be prohibitive for some uses. One abbreviated version of the EFI has been published
with subscales for affect, behavior, and cognition with eight, six, and eight items, and Cronbach alphas for the
subscales are .91, .91, and .89, respectively (
McLernon, Cairn s, Hewstone, & Smith, 2004). The total 22-item abbre-
viated scale alpha coefficient is .95. Subscale intercorrelations exceed .70 (
McLernon et al., 2004 ). The abbreviated
EFI may offer researchers an efficient and effective means of assessing interpersonal forgiveness.
Transgression Related Interpersonal Motivations Inventory (TRIM)
(McCullough et al., 1998b)
The TRIM inventory measures unforgiving motivations toward the offender after a particular identified trans-
gression. The motives are avoidance (TRIM -A) of the person, revenge (TRIM-R), and (added later) benevolent
(TRIM-B) motives.
The original source of the TRIM was Susan Wade Brown’s doctoral dissertation at Fuller Seminary
Wade, 1990). She originally created 600 items winnowed them to 83 items in nine subscales resulting in he Wade
Forgiveness Scale (WFS;
Wade, 1990). Using 5-point anchors, responses ranged from 1 5 Strongly disagree to
5 5 Strongly agree for each item. The WFS was subsequently published (
Brown et al., 2001) with all 83 items
grouped into 11 subscales four that reflected cognition (Revenge, Freedom from Obsession, Affirmation, and
Victimization), one that reflected affect (Positive versus Negative Feelings), and six that reflected behaviors
(Avoidance, Moveme nt Against, Toward God, Conciliation, Holding a Grudge, and Personal Benefit). Prior to
the publication of the WFS, Brown’s scale had been trimmed to two subscales (Revenge and Avoidance) with
five and seven items, respectively (McCullough et al., 1998a), and that abbreviated version was called the TRIM.
Because the TRIM-12 measures unforgiving motivations, its use has limitations. For example, there are many
ways to reduce unforgiveness besides forgiving (
Worthington, 2006) such as acceptance, seeing justice done, or
even getting succe ssful revenge. Thus, in some cases, it is not conceptually advisable to equate lower scores on
the TRIM-12 (reduced unforgiveness) with actually forgiving (see
Wade & Worthington, 2003). In studies
Authors personal copy
intervention research to promote forgiveness, we might assume that reduced unforgiveness implies forgiveness.
But in studies that do not directly target forgiveness, it is important either to supplement the TRIM-12 with a
measure of forgiveness or use a measure of positivity that suggests that forgiveness might have occurred.
McCullough and Hoyt (2002) used seven items taken from the WFS that indicated benevolent mo tives toward
the offender to provide more assurance that the TRIM was assessing forgiveness. (These items were listed in
Note 1 in McCullough & Hoyt; we might designate them TRIM-B7.)
McCullough et al. (2003) later constructed a
measure of benevolent motives toward the offender from five (TRIM-B5) of the seven items in McCullough
and Hoyt.
Finally, McCullough et al. (2010) used Rasch modeling within IRT to scale the TRIM-R, TRIM-A, and a six-
item benevolence scale (the TRIM-B6) to make up the TRIM-18. They treated this combined measure as a mea-
sure of forgiveness indicating reduced unforgiving motives and simultaneous increased benevolence motives by
reverse-scoring the TRIM-R and TRIM-A items.
McCulloughl et al. (1998b) established initial psychometric data on the TRIM with university students
(N 5 239). Students responded to the 18 items on Wade’s original Revenge and Avoidance subscales. The individ-
ual subscales might be more appropriate for different samples, and might also be used independently. For exam-
ple, marri ed couples often have very low scores on TRIM-R and moderate scores on TRIM-A after experiencing a
transgression. For divorced couples, TRIM-R scores tend to be higher, but because they are separated and in
many cases remarried or living at great distances from each other, their TRIM-A scores are low. They must exert
little energy to avoid someone who is virtually out of their life already.
Internal Consistency
McCullough et al. (1998b) reported that for 5-item TRIM-R the Cronbach alpha coefficient was .90; for 7-item
TRIM-A α 5 .86.
McCullough and Hoyt (2002) found for TRIM-B7 α 5 .85. McCullough et al. (2003) found that, for
the TRIM-B5 α 5 .91 to .93. McCullough et al. (2010) used IRT and thus they did not calculate the alpha coefficient.
The Person separation reliability, which is interpreted analogously to Cronbach’s alpha, was .92; the item separation
reliability was 1.0, indicating evenly spaced items in terms of difficulty of endorsement. The TRIM-B5
McCullough et al., 2003) had Cronbach alpha coefficients from .91 to .93 in various samples. The TRIM-18 was sub-
jected to item response analysis (McCullough et al., 2010). The fit of the 18 items to the Rasch model, using an
unweighted least-squares fit statistics, was between .73 and 1.55; the expected value is 1.0, scores greater than 1.5 are
considered to indicate a less than optimal fit, and scores of 2.0 are considered unacceptable. Thus, for the sample of
372 undergraduates, the TRIM-18 measured forgiveness well according to IRT (McCullough et al., 2010).
In McCullough et al. (1998a), TRIM-12 scores over thr ee weeks were correlated r 5 .86 and r 5 .79, for TRIM-R
and TRIM-A, respectively, and over eight weeks r 5 .53, r 5 .44, for TRIM-R and TRIM-A, respectively, and over
nine weeks r 5 .64, r 5 .65, for TRIM-R and TRIM-A, respectively. For the TRIM-B5 (
McCullough et al., 2003),
temporal stability correlations weekly for four weeks ranged from .87 (one week) to .52 (four weeks), which is
appropriate for a state measure of forgiveness for a particular transgression in college students.
TRIM-R and TRIM-A were correlated (
McCullough et al., 1998b; Study 3) with a linear composite of the
Dyadic Adjustment Scale (Spani er, 1979) and the Commitment Inventory (Constraint and Dedication; Stanley &
Markman, 1992) at .37 to .62. In
McCullough et al. (1998b, Study 4), TRIM-R was correlated with negative affec-
tivity (r 5 .32), and self-deception (r 5 .30). TRIM subscales were positively correlated with single-item measures
of forgiveness at r 52.67 and r 52.41 for TRIM-R and TRIM-A, respectively. There was evidence supporting
construct validity.
Divergent/Discrim inant
McCullough et al. (1998b, Study 4), TRIM-R was correlated negatively with impression management
(r 52.17), but TRIM-A was not correlated with negative affectivity, self-deception, or impression management.
Authors personal copy
Construct/Factor Analytic
Item responses were subjected to a principal components analysis (PCA) using Varimax rotation, and two
components were detected with five of the original 10 items on TRIM- R and seven of the original eight on
TRIM-A. The mean for TRIM-R was 8.68 (SD 5 4.46; alpha 5 .90; range 5 5 to 25), and the mean for TRIM-A was
18.14 (SD 5 8.46; alpha 5 .86; range 5 7 to 35). The structure of the model was tested using two confirmatory fac-
tor analyses (CFAs). A two-factor model (Comparative Fit Index, CFI, was .94; χ
/df 5 3.8) was superior to a
one-factor model (CFI 5 .76; χ
/df 5 12.1). The factors were correlated at .67, suggesting that the items might be
summed to give a measure of unforgiveness (TRIM-12).
McCullough and Hoyt (2002), Study 1, TRIM-R was predicted by the Big Five, as measured by John,
Donahue, and Kentle’s (1991)
Big Five Inventory (BFI) (R
5 .18; standardized β for agreeableness was 2.36). The
TRIM-A was also predicted by the Big Five (R
5 .18; standardized β for neuroticism was .36), as was the TRIM-
B7 (R
5 .15; standar dized β for neuroticism was 2.22 and for agreeableness was .19). In Study 2 of McCullough
and Hoyt (2002)
, for self-ratings the results were similar to Study 1. TRIM-R was predicted by the Big Five
5 .33; standardized β for agreeableness was 2.51). The TRIM-A was also predicted by the Big Five (R
5 .30;
standardized β for neuroticism was .24 and agreeableness was 2.41). Similarly, TRIM-B7 was predicted by the
Big Five (R
5 .40; standardized β for neuroticis m was 2.32 and for ag reeableness was .48).
McCullough, M.E., Rachal, K.C., Sandage, S.J., Worthington, E.L., Jr., Brown, S.W., & Hight, T.L. (1998).
Interpersonal forgiveness in close relationships II: Theoretical elaboration and measurement. Journal of Personality
and Social Psychology, 75, 15861603.
McCullough, M.E., & Hoyt, W.T. (2002). Transgression-related mo tivational dispositions: Personality substrates
of forgiveness and their links to the Big Five. Personality and Social Psychology Bulletin, 28, 15561573.
McCullough, M.E., Fincham, F.D., & Tsang, J-A. (2003). Forgiveness, forbearance, and time: The temporal
unfolding of transgression-related interpersonal motivations. Journal of Personality and Social Psychology, 84,
Results and Comments
The versatile TRIM is the most widely used forgiveness self-report instrument. Entering ‘Transgression Related
Interpersonal’ into PsycINFO (accessed October 26, 2012) and searching ‘Any Field’ resulted in 148 entries, which
is over 50 percent more often than the next most used instrument (the EFI), which resulted in 104 entries. Studies
suggest that the TRIM is psychometrically well supported, can be administered in less than 10 minutes, requires
no cost to experimenters, and is generative of research findings relevant to theory. Its psychometric properties
also make it potentially usable with clinical samples. Its scores typically have high estimated reliabilities, and the
evidence supporting its construct and criterion-related validity is strong and consistent. It is particularly useful in
that the TRIM-R and TRIM-A may be used to assess forgiveness and unforgiveness, whereas some instruments
only permit an assessment of forgiveness; yet, by including the items that measure benevolence motives, investi-
gators can also measure forgiveness using the TRIM-18. It does not have as much evidence supporting its use
with clinical populations as does the EFI, but it has been used extensively in psyc hoeducational interventions.
Directions: For the following questions, please indi-
cate what you imagine your current thoughts and feel-
ings would be about the person who wounded you. Use
the following scale to indicate your agreement or dis-
agreement with each of the statements.
1 5 strongly disagree
2 5 mildly disagree
3 5 agree and disagree equally
4 5 mildly agree
5 5 strongly agree
1. ____ I’ll make him or her pay.
2. ____ I wish that something bad would happen to
3. ____ I want him-her to get what he/she deserves.
4. ____ I’m going to get even.
Authors personal copy
5. ____ I want to see him/her hurt and
miserabl e.
6. ____ I’d keep as much distance between us as
7. ____ I’d live as if he/she doesn’t exist, isn’t
8. ____ I wouldn’t trust him/her.
9. ____ I’d find it difficult to act warmly toward
10. ____ I’d avoid him/her.
11. ____ I’d cut off the relationship with him/her.
12. ____ I’d withdraw from him/her.
13. ___ Even though his/her actions hurt me, I still
have goodwill for him/her.
14. ___ I want us to bury the hatchet and move
forward with our relationship.
15. ___ Despite what he/she did, I want us to have a
positive relationship again.
16. ___ I have given up my hurt and resentment.* 1
17. ___ Although he/she hurt me, I put the hurts
aside so we could resume our relationship.
18. ___ I forgive him/her for what he/she did to me.*
19. I have released my anger so I could work on
restoring our relationship to health.
Items 1 through 5 are Revenge.
Items 6 through 12 are Avoidance.
Items 13 through 19 are Benevolence.
*omitted from 13-19 to make the TRIM-B5.
1 omitted to make the TRIM-B6.
Reproduced with permission.
Rye Forgiveness Scale (RFS)
(Rye et al., 2001).
The RFS is a self-report measure of the extent to which one has forgiven an offender.
The original version of the RFS (
Rye, 1998), consisted of 16 items designed to measure the extent to which
respondents had forgiven a romantic partner. Items assessed cognitive, emotional, and behavioral responses to a
romantic partner’s wrongful actions. The scale was subseque ntly revised (
Rye et al., 2001) to be applicable to for-
giveness of any offender. Following a principal components analysis (PCA) with Varimax rotation using 328 uni-
versity students, one item was dropped because its loading was less than .40, leaving 15 items in the revised
scale. The PCA revealed two subscales; the Absence of Ne gative subsca le (Forgiveness AN) consists of 10 items
and measures the extent to which one has overcome negativ e thoughts, feelings, and behaviors toward an
offender (e.g., ‘I have been able to let go of my anger toward the person who wronged me’). The Presence of
Positive subscale (Forgiveness PP) consists of five items and measures positive thoughts, feelings, and behaviors
toward an offender (e.g., ‘I have compassion for the person who wronged me’) . Possible responses for items on
each subscale range from 1 (Strongly Disagree)to5(Strongly Agree). Scores can range from 10 to 50 for
Forgiveness (AN), 5 to 25 for Forgiveness (PP), and 15 to 75 for RFS total.
Rye et al. (2001) examined the psychometric properties of the RFS using 328 students enroll ed at a Midwestern
Catholic university. Means and standard deviations were M 5 36.6 (SD 5 7.8) for Forgiveness AN, M 5 16.7
(SD 5 4.4) for Forgiveness PP.
Internal Consistency
Rye et al. (2001) found the Cronbach alpha coefficients for the RFS to be α 5 .86 (Forgiveness AN), α 5 .85
(Forgiveness PP), and α 5 .87 (for RFS total) (cf.
Boyle, 1991).
Rye et al. (2001) found that the RFS scores over two weeks were correlated r 5 .76 (Forgiveness AN), r 5 .76
(Forgiveness PP), r 5 .80 (RFS total).
Authors personal copy
RFS subscales were positively correlated with the EFI (Forgiveness AN, r 5 .52; Forgiveness PP, r 5 .75;
Rye et al., 2001) and RFS total was positive ly correlated with an observer measure of forgiveness (r 5 .32;
Rye et al., 2005).
Divergent/Discrim inant
The RFS has been negatively related to mental health measures such as state anger (Forgiveness AN, r 52.41;
Forgiveness PP, r 52.13), trait anger (Forgiveness AN, r 52.34; Forgiv eness PP, r 52.21).
Rye et al. (2001)
found the RFS to be weakly related to the Marlowe-Crowne Social Desirability Scale (Crowne & Marlow, 1960;
Forgiveness AN, r 5 .16; Forgiveness PP, r 5 .22).
Construct/Factor Analytic
A principal components analysis (PCA) with varimax rotation (N 5 328), resulted in extraction of two orthogo-
nal compo nents labeled: the Absence of Negative subscale (Forgiveness AN) and The Presence of Positive sub-
scale (Forgiveness PP), as mentioned above (
Rye et al., 2001). Although a three component solution was also
feasible, the third component was dropped because it loaded substantially ($.40) on only two of the RFS items.
Forgiveness interventions have resulted in increased scores on the RFS (
Rye et al., 2012; Rye & Pargament,
2002; Rye et al., 2005 ). RFS scores also have been positively predictive of mental health measures such as existen-
tial well-being (Forgiveness AN, r 5 .40; Forgiveness PP, r 5 .21), and religious well-being (Forgiveness AN,
r 5 .20; Forgiveness PP, r 5 .30).
Rye, M.S., Loiacono, D.M., Folck, C.D., Olszewski B.T., Heim, T.A., & Madia, B.P. (2001). Evaluation of the psy-
chometric properties of two forgiveness scales. Current Psyc hology, 20, 260277.
Results and Comments
The RFS subscales enable researchers to examine whether aspects of forgi veness (i.e., Absence of Negative,
Presence of Positive) relate differentially to constructs of interest. Similar to the TRIM, researchers can separ ate
the reduction of negative thoughts, feelings, and behaviors, which could occur in many ways that do not involve
forgiving (
Wade & Worthington, 2003) from the presence of positive thoughts, feelings, and behaviors, which
probably arise from forgiving. The RFS subscales can be combined to create a total RFS sco re. Evidence support-
ing the validity of the RFS scores comes from relationships with scores on other measures of forgiveness (self
and observer report), scores on measures of mental health variables, and also from the finding that scores
increase following forgiv eness interventions. The RFS has been translated into several languages including
Portuguese (
Neto, Ferreira, & Pinto, 2007), Chinese (Wang, 2008), Kinyarwana (translator: Carly Raby), Kannada
(translator: Shanmukh Kamble), and Korean (translator: Dong Hwan Lim). A variety of studies have shown that
the RFS has strong psychometric properties with diverse populations, cultures, and langu ages.
Think of how you have responded to the person
who has wronged or mistreated you. Indicate the degree to
which you agree or disagree with the following statements.
12 345
Strongly Disagree Neutral Agree Strongly
disagree agree
____1. I can’t stop thinking about how I was wronged
by this person. (R)
____2. I wish for good things to happen to the person
who wronged me.
____3. I spend time thinking about ways to get back
at the person who wronged me. (R)
____4. I feel resentful toward the person who
wronged me. (R)
____5. I avoid certain people and/or places because
they remind me of the person who wronged
me. (R)
____6. I pray for the person who wronged me.
Authors personal copy
____7. If I encountered the person who wronged me
I would feel at peace.
____8. This person’s wrongful actions have kept me
from enjoying life. (R)
____9. I have been able to let go of my anger toward
the person who wronged me.
____10. I become depressed when I think of how I was
mistreated by this person. (R)
____11. I think that many of the emotional wounds
related to this person’s wrongful actions have
____12. I feel hatred whenever I think about the person
who wronged me. (R)
____13. I have compassion for the person who
wronged me.
____14. I think my life is ruined because of this person’s
wrongful actions. (R)
____15. I hope the person who wronged me is treated
fairly by others in the future.
Scoring Instructions:
First, reverse-score items marked ‘R’. To calculate RFS
total, add the values for all items. To create the subscales,
add items for each subscale as indicated below:
Absence of Negative subscale items: 1, 3, 4, 5, 8, 9,
10, 11, 12, 14
Presence of Positive subscale items: 2, 6, 7, 13, 15
Reproduced with permission.
Decisional Forgiveness Scale (DFS) and Emotional Forgiveness Scale (EFS)
Worthington et al., 2007a).
The DFS and EFS are brief self-report measures of the extent to which one has made a decision to forgive an
offender and behave differently toward the person (i.e., the DFS) and has experienced emotional replacement of
negative emotions with positive emotions toward the offender (i.e., the EFS).
The original versions of the DFS and EFS were developed as part of a master’s thesis (
Hook, 2007), and results
from that thesis and other independen t studies were reported as part of the Positive Psyc hology Summit
Worthington et al., 2007a ). A sample of N 5 679 undergraduate s was split into samples of n 5 400 (sample 1A),
n 5 179 (sample 1B), and n 5 100 (sample 1C), and a new sample of N 5 298 (sample 2) was then collected. Each
item was rated from 1 5 Strongly Disagreeto55 Strongly Agree . On sample 1A, 15 decisional forgiveness and 37
emotional forgiveness items were subjected to PCAs with oblique rotations. Two components were found for the
DFS (Prosocial Intentions and Inhibition of Harmful Intentions; r 5 .40), and two components were found for the
EFS (Presence of Positive Emotion and Reduction of Negative Emotions; r 5 .50). Because the 4-item subscales for
the DFS and the two 4-item subscales for the EFS were correlated, the authors recommended typically using DFS
and EFS as two scales and not attending to their subscales unless specific hypotheses were examined.
Worthington et al. (2007a) examined the psyc hometric properties of the DFS and the EFS using the thesis
samples (
Hook, 2007), plus independent samples of students who participated in (a) a scenario study (N 5 100);
(b) an ego depletion task (N 5 100); and (c) an implicit measures test (N 5 62). Later, a dissertation adduced
evidence of the relationship of DFS and EFS scores to diastolic and systolic blood pressure, mean arterial pres-
sure, heart rate, and salivary cortisol in 60 adult women (
McCrocklin, 2009).
Internal Consistency
Cronbach alpha coefficients for the DFS ranged from .80 to .83, and for the EFS ranged from .69 to .83
Worthington et al., 2007a).
Both DFS (r 5 .73) and EFS scores (r 5 .73) were relatively stable over three weeks (
Worthington et al., 2007a).
Authors personal copy
Worthington et al. (2007a) reported a positive correlation of the DFS with the TRIM-B7 (r 5 .68); The DFS was
correlated with several dispositional measures of forgiveness like the Rye Forgiveness Likelihood Scale (r 5 .44)
and Trait Forgivingness Scale (r 5 .46). Worthington, Hook, Utsey, Williams, and Neil also reported positive cor-
relations of the EFS with the TRIM-B7 (r 5 .75), empathy for the offender (r 5 .54), dispositional measures of for-
giveness like the Rye Forgiveness Likelihood Scale (r 5 .67) and Trait Forgivingness Scale (r 5 .36).
Divergent/Discrim inant
Worthington, Hook, Utsey, Williams, and Neil also reported correlations of the DFS with the TRIM-A
(r 52.63), TRIM-R (r 52.61), however, no relationships were found between DFS scores and empathy for the
offender or rumination. The EFS correlated negatively with rumination (r 52.29). Worthington, Hook, Utsey,
Williams, and Neil also reported negative correlations of the EFS with the TRIM- A (r 52.73), and TRIM-R
(r 52.44).
Construct/Factor Analytic
All items were initially analyzed using principal components analysis (PCA) constrained to two factors
(N 5 399;
Worthington et al., 2007a). Items on the DFS and EFS were subjected to separate PCAs using Promax
(oblique) rotation. A simple structure was found both for DFS and for EFS. CFAs were carried out using
sample 1B (
Worthington et al., 2007a) with acceptable fit statistics on each scale. For the DFS, χ
/df was 2.5,
comparative fit index (CFI) 5 .98; normed fit index (NFI) 5 .97; goodness of fit index (GFI) 5 .96; root mean
squared error of appro ximation (RMSEA) 5 .08). For the EFS, χ
/df 5 3, CFI 5 .97; NFI 5 .96; GFI 5 .95;
RMSEA 5 .08.
Predictive validity evidence has been adduced using an implicit measures test (Worthington et al., 2007a).
In sample 1C of Worthington et al. (2007a), participants were randomly assigned to identify a relationship in
which either they held a grudge, had granted decisional forgiveness but had not experienced complete emotional
forgiveness, or had experienced both. Scores on the DFS and EFS were consonant with the condition, and reaction
times were correlated with scores on the DFS. Furthermore, the reaction times for the congruent conditions
(M 5 687 ms, SD 5 104 ms) were significantly faster than for the incongruent conditions (M 5 822 ms,
SD 5 186 ms, t
5 5.47, p , .001). EFS scores (but not DFS scores) were found to predict physiological responses
McCrocklin, 2009). Harper et al. (in press) fo und that EFS and DFS both responded to six-hour workbook-based
forgiveness interventions.
Worthington, E.L., Jr., Hook, J.N., Utsey, S.O., Williams, J.K., & Neil, R L. (2007, October). Decisional and emo-
tional forgiveness . Paper presented at the Positive Psychology Summit, Washington, DC, October 5, 2007.
Note: To obtain a copy of this paper, contact Everett Worthington at the following email address:
Hook, J.N. (2007). Forgiveness, individualism, and collectivism. Unpublished master’s thesis, Virginia
Commonwealth University, Richmond.
McCrocklin, C. (2009). Cardiovascular reactivity to forgiveness in females. Unpublished doctoral dissertation,
Virginia Commonwealth University, Richmond.
Results and Comments
Studies suggest that the DFS measures a decision to forgive and might be related to what people often
mean when they say they have f orgiven an offender; they often mean t hey intend to act more positively
toward the perso n. Emo tional forgiveness is informed by a t heory of replacing negative unforgiving emotions
with more positive ones toward the offender (fo r evidence supporting the theory, see
Worthington, 2006).
Emotional forgiveness is related to what people m ean when they say they ‘feel like’ they have forgiven an
offender. The DFS and EFS have been translated into Korean and used with samples from North Korea
Park, 2012)andSouthKorea(Chong, 2010). More evidence exists for scores on either of the DFS and EFS
than for the two subscales.
Authors personal copy
Directions: Think of your current intentions toward the person who hurt you. Indicate the degree to which you
agree or disagree with the following statements.
1. I intend to try to hurt him or her in the same way
he or she hurt me.
2. II will not try to help him or her if he or she
needs something.
3. If I see him or her, I will act friendly. SD D N A SA
4. I will try to get back at him or her. SD D N A SA
5. I will try to act toward him or her in the same way
I did before he or she hurt me.
6. If there is an opportunity to get back at him or her,
I will take it.
7. I will not talk with him or her. SD D N A SA
8. I will not seek revenge upon him or her. SD D N A SA
Scoring Instructions: Items 1, 2, 4, 6, and 7 are reverse scored. Subscales are Prosocial Intentions (items 2, 3, 5, 7)
and Inhibition of Harmful Intentions (items 1, 4, 6, 8).
Reproduced with permission.
Directions: Think of your current emotions toward the person who hurt you. Indicate the degree to which you
agree or disagree with the following statements.
1. I care about him or her. SD D N A SA
2. I no longer feel upset when I think of him or her. SD D N A SA
3. I’m bitter about what he or she did to me. SD D N A SA
4. I feel sympathy toward him or her. SD D N A SA
5. I’m mad about what happened. SD D N A SA
6. I like him or her. SD D N A SA
7. I resent what he or she did to me. SD D N A SA
8. I feel love toward him or her. SD D N A SA
Scoring Instructions: Items 3, 5, and 7 are reverse scored. Subscales are Presence of Positive Emotion (items 1, 4,
6, 8) and Reduction of Negative Emotions (items 2, 3, 5, 7).
Reproduced with permission.
Authors personal copy
Chemical Measures of Forgiveness
Cortisol and oxytocin are potential chemical measures of forgiveness (
Berry & Worthington, 2001). Aso, Tabak,
McCullough, Szeto, Mendez, and McCabe (2011, p. 115)
stated that, ‘Elevated mean peripheral oxyto cin reactivity
(but not baseline levels of oxytocin or cortisol reactivity) was associated with increased post-conflict anxiety and
decreased levels of forgiveness.’
Cortisol has been associated with trait forgivingness (
Berry & Worthington, 2001) and can be measured using
blood, salivary, or urine samples analyzed by enzyme or radioimmunoassays (EIAs/RIAs; Nicolas, 2008).
One study determining the association between cortisol and perceived agreeableness of a transgressor used
0.2 μg/mL as the minimal detectable level, an intra-assay coefficient of variation (CV) of 5.1% , and an inter-assay
CV of 4.0% (
Tabak & McCullough, 2011). Cortisol takes time (at least 15 minutes) to reach a peak after experienc-
ing or thinking about a stressful episode that stimulates unforgiveness.
Oxytocin, a peptide associated with relational attentiveness (
Kemp & Gaustella, 2011), has been shown to ele-
vate if forgiveness is low (
Tabak et al., 2011). Oxytocin and cortisol were measured with periodic blood samples,
and .38 mL of Aprotinin reagent was added to each sample. Oxytocin was extracted using solid phase chroma-
tography and analyzed using RIA.
Berry and Worthingto n (2001) measured salivary cortisol using a sample of 39 students at Virginia
Commonwealth University. Means and standard deviations for baseline and post-imagery salivary cortisol were
(in μg/mL) M 5 .19 (SD 5 .11) and M 5 .16 (SD 5 .08). Both
Tabak and McCullough (2011) and Tabak et al. (2011)
sampled 39 students at the University of Miami. Cortisol means and standard deviations for the first and final
time points were (in pg/tube): M 5 18.0 (SD 5 7.61) and M 5 18.99 (SD 5 9.17). Oxytocin means and standard
deviations across for the first and final time points were (in pg/mL): M 5 1.61 (SD 5 2.78) and M 5 1.48
(SD 5 2.60).
Salivary assays are a reliable measure of momentary cortisol (
Walker, Riad-Fahmy, & Read, 1978) and are
highly correlated with blood cortisol measurement (
Nicolson, 2008). RIA and EIA are the best available measures
of oxytocin, but may not be highly reliable (
Tabak et al., 2011).
Internal Consistency
Estimates for consistency of assays for these chemical measures of forgiveness have not been established.
Temporal stability for cortisol and oxytocin are low, perhaps due to habi tuation over time points and to sensi-
tivity of the measures to many environmental and cognitive events (
Young, Abelson, & Lightman, 2004).
Temporal stab ility has yet to be determined for cortisol and oxytocin as measures of forgiveness.
Despite lower levels being associated with higher forgiveness and related constructs (
Berry & Worthington,
2001; Tabak & McCullough, 2011), salivary cortisol is best seen as a measure of transient response to stressors. To
the extent that unforgiveness is a stress response and alternative causes for the stress response can be ruled out,
cortisol might have evidence supporting its construct validity. Berry and Worthington (2001)
compared indivi-
duals who were currently in a satisfying romantic relationship relative to those who were in either divorced, sep-
arated, or conflictual relationships. The two groups of individuals also different in theory-consistent ways on two
measures of trait forgivingness, trait anger, salivary cortisol reactivity and baseline levels. Salivary cortisol base-
line and reactivity measures were correlated with trait forgivingness ( r 52.35 and r 52.36, respectively) and
with trait anger (r 52.20, and r 52.42, respectively).
Past research suggests that oxytocin is inversely related to forgiveness and positively associated with relational
stress (
Tabak et al., 2011), suggesting that its release may serve as a response to relational distress. Oxytocin has
Authors personal copy
frequently been described as a bonding peptide. Recent research, however, interprets it more as attentiveness to
relational cues (
Kemp & Gaustella, 2011).
RIAs and EIAs to measure cortisol and oxyto cin can be accessed through such companies as Diagnostic
Products Corporation and Siemens Medical Solution Diagnostics (
Results and Comments
Chemical measures have been shown to be psychometrically sound measures of stress and distr ess in their
own right, but are not necessarily accepted measures of forgiveness. These measures are objective, observable,
and transcend lang uage, giving insight into the physical pro cesses that might be associated with forgiveness, but
they require more than a handful of studies, each of which testing only a few participants, to be considered psy-
chometrically supported measures of forgiveness.
Peripheral Physiological Measures
(e.g., Lawler et al., 2003; Witvliet, Knoll, Hinman, & DeYoung, 2010).
Investigations of both state and trait forgiveness have incorporated a range of physiological measures, includ-
ing autonomic nervous system indicators and facial electromyography to assess expressions at different muscle
regions. Physiological measures may be considered indirect measures of how forgiveness-related responses are
embodied. Forgiveness has often been found to down-regulate negative and aroused emotion, with concomitant
physiological changes.
Autonomic indicators of unforgiveness or forgiveness have included heart rate and blood pressure.
Assessments of the sympathetic branch of the autonomic nervous system have measured electrodermal (i.e., skin
conductance) ind icators. Measurement of the parasympathetic nervous system (PSNS) has focused on heart
rate variability (HRV). Facial electromyographic (EMG ) measures have focused especially on mu scle activity at
the corrugator (above the brow), the orbicularis oculi (under the eye), the zygomatic (cheek), and frontalis
(forehead) regions.
The first experimental assessment of fo rgiveness-related responses incorporating measures of emotion and
peripheral indicators was conducted by
Witvliet et al. (2001). Participants were asked to think of a person they
blamed for offending them. Next, they systematically engaged four different responses to that offender that incor-
porated either forgiving or unforgiving imagery, in counterbalanced orders. Four measures of autonomic arousal
were collected: including: (1) facial EMG; (2) skin conductance; (3) mean arterial blood pressure; and (4) heart
rate. These were measured at multiple trials per condition. Each trial had a pre-trial baseline, imagery, and recov-
ery phase. During the unforgiving imagery, heart rate, blood pressure, and skin conductance, and EMG activity
showed more reactivity than the forgiving imagery conditions.
Recently, PSNS activity has been assessed thr ough measuring HRV. It was assessed with the high frequency
component derived through spectral analysis. In two experimental paradigms, compassionate reap praisal
prompted forgiveness and kept HRV at baseline levels, whereas offence rumination prompted significant reduc-
tions in PSNS activity (
Witvliet et al., 2010; Witvliet, DeYoung, Hofelich, & DeYoung, 2011).
In another program of research, Lawler-Row and colleagues have used peripheral physiological measures to
study forgiveness.
Lawler et a l. (2003) employed an experimental stress-reactivity paradigm. Participants were
asked to recall a time when they were deeply hurt by someone close to them. Participants were interviewed
about the hurt and recounted in detail what had happened. They were asked how long ago the offence occurred,
what feelings they had at the time, whether they expressed these feelings, and why they thought the offender
acted in the way he or she did. Peripheral measures included blood pressure, heart rate, facial EMG activity, and
electrodermal activity. Lawler et al. found that the most consistent peripheral correlates of forgiveness were
related to cardiovascular function; however, the exact magnitude of the effects and the connections to forgiveness
showed variability across cardiovascular outcomes. Blood pressure seemed most consistently related to forgiving,
as compared to heart rate and other peripheral measures.
Authors personal copy
Witvliet et al. (2001) used 72 introductory psychology students (36 male, 36 female) and Lawler et al. (2003)
used 108 college students (44 male, 64 female) in their init ial investigations providing construct validity data
regarding peripheral physiological measures as indexes for unforgiveness.
Smith and Uchino (2008) highlighted that whereas physiological measures are touted as being more direct assess-
ments, the interpretation of peripheral physiological assessments is not unambiguous. Sources of error variance in
physiological measurements include sociodemographic characteristics, nutrition, health behaviors (e.g., smoking,
alcohol consumption). Also, artifacts may occur during data collection due to equipment or software error, electrode
placement, or human movement, posture, and speech. Measurement guidelines for calculations of HRV are pro-
vided by the
Task Force (1996). Gerin, Goyal, Mostofsky, and Shimbo (2008) enumerate problems of reliability in the
assessing blood pressure: blood pressure changes with heart beats and interbeat periods and blood pressure is
highly variable and is influenced by equipment, software, environmental, psychosocial, and physical condition fac-
tors. Reliable assessment of peripheral physiology requires attention to the possible sources of unwanted error vari-
ance, repeated assessments, and careful measurement techniques. Nevertheless, the reliability of peripheral
physiology assessments is easily estimated and quite high (
Smith & Uchino, 2008). In Witvliet’s general method of
investigating peripheral physiological measures, she uses a within subjects design; Lawler uses between subjects
designs. While Witvliet has reported no actual correlations on the estimated reliability of measurements, the use of
between subjects methods suggests that high testretest reliability exists over the period of two to three minutes.
Organizations such as the Association for the Advancement of Medical Instrumentation have developed guide-
lines and standards for physiological assessment equipment (see also
Fridlund & Cacioppo, 1986). Construct validity
of physiological assessments rests on correlations of averages with self-report instruments that indicate that forgive-
ness has occurred, including movement of a joystick to measure degree of forgiveness and associated emotions.
Witvliet et al. (2001) provided convergent construct validity in the form of an experiment using within-
subjects design. She instructed participants to reflect on a grudge and either (1) rehearse hurtful memories;
(2) nurse the grudge; (3) think empathically about why the offender might have inflicted the hurt; or (4) imagine
forgiving the offender. The two unforgiving imaginal tasks resulted in more diastolic and systolic blood pressure
(combined as mean arterial pressure), higher heart rate, higher skin conductance, and more muscular tension in
the corrugator (i.e., brow) relative to the empathy and forgiveness im agery conditions. These were related to self-
ratings of sadness, anger, and unforgiveness.
Lawler, K.A., Younger, J.W., Piferi, R.L., Billington, E., Jobe, R., Edmondson, K., & Jones, W.H. (2003). A
change of heart: Car diovascular correlates of forgiveness in response to interpersonal conflict. Journal of Behavioral
Medicine, 26, 373 393.
Witvliet, C.V.O., DeYoung, N.J., Hofelich, A.J., and DeYoung, P.A. (2011). Compassionate reappraisal and
emotion suppression as alternatives to offence-focused rumination: Implications for forgiveness and psychophysi-
ological well-being. Journal of Positive Psychology, 6, 286299.
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, 121, 117123.
Note: Equipment for peripheral physiological assessment is ava ilable from sources such as Biopac, Coulbourn,
Finapres, Ohmeda, and Omron, and others.
Results and Comments
Each of the peripheral physiological assessments discussed above should be evaluated for their reliability and
validity independent of their use as measures of forgiveness. It is vital to consider in inte rdisciplinary frame-
works how these physiological measures are related to the ways that people embody forgiveness and unforgive-
ness responses. At present, it would be prudent to conclude that various measures of peripheral physiological
activity offer the potential to provide insight into how state-forgiveness and trait-forgivi ngness processes are
embodied and objectified. All software packages that come with equipment should be cross-checked carefully for
the accuracy of calculations, particularly for HRV (see
Witvliet et al., 2010, 2011, Methods and notes).
Authors personal copy
Forgiveness research has embraced functional magnetic resonance imaging (fMRI) as a central physiological
measure (
Farrow et al., 2001, 2008; Pietrini et al., 2004).
In an early study,
Farrow et al. (2001) employed fMRI to assess which brain structures were active wh en mak -
ing judgments about what one might forgive (or not forgive), empathize with, and make judgments about in
social situations.
Farrow et al. (2001) found that forgivability and empathy judgments were associated with left
fronto-temporal activity, unlike fairness judgments. Subsequently,
Farrow and Woodr uff (2005) measured brain
activity in patients with post-traumatic stress disorder and with schizo phrenia. The regions of the brain assessed
were areas involved in making forgivability and empathy judgments, as determined in
Farrow et al. (2001).
The patients attended forgiveness-focused cognitive behavioral therapy (10 weekly 1-hour sessions) and had
higher forgivability and empathy judgments post-test vs. pre-test in 13 patients with post-traumatic stress disor-
der and 14 patients with schizophrenia.
Pietrini et al. (2004) used fMRIs to assess participants who imagined a hurtful event and were
then instructed to either forgive or focus on the hurt. In the hurt-focus condition, females showed more activity
in the a nterior cingulate cortex (which was associated with state forgiveness and the capacity to be forgivi ng)
than did mal es.
Pietrini et al. (2004) used positron emission tomography (PET) to measure cerebral bloodflow during anger
and aggression. Harmon-Jones (2004) used electroencephalograp hic (EEG) measures to show that responses
of sympathy opposed anger arousal, decreasing anger-related left-frontal cortical activity and orbitofrontal
Farrow et al. (2001) studied 10 participants (7 male, 3 female). Farrow and Woodruff (2005, Farrow et al.,
Farrow et al., 2008)
studied 27 participants (13 with post-traumatic stress disorder and 14 with schizophrenia).
Pietrini et al. (2004) studied 10 healthy participants (5 males, 5 females).
The reliability of functional brain imaging techniques is difficult to evaluate. Diverse study designs and experi-
mental tasks, head motion, software and hardware variety, and lack of standardized procedures contribute to dif-
ficulty in establishing the reliability of functional imaging (
Seixas & Lima, 2011). Though no standards currently
exist for acceptable levels of fMRI reliability, a recent review shows that most studies have reliability estim ates
(i.e., intra-class correlation coefficients) ranging between .33 and .66 (
Bennett & Miller, 2010).
Farrow et al. (2001), participants were presented with three scenarios, which they read carefully and then
imagined for 16 seconds. They then made five discriminations (7 seconds each) related to each scenario.
Scenarios required judgments of basic social reasoning (control condition), empathic judgments (i.e., can you be
empathic toward the offending party), and forgivability judgments (i.e., is the offence forgivable). Although no
computations of estimated reliability were presented, some measure of reliability was achieved by using five sim-
ilar judgments instead of a single judgment in each imagined condition.
Critics have questioned the validity of fMRI techniques (
Logothetis, 2008; Vul, Harris, Winkielman, & Pashler,
2009), and some have even questioned wh at specific type of neurological activity is reflected by blood oxygen-
ation levels ( Attwell & Iadecola, 2002
). Interpreting areas that ‘light up’ under during an experience might not be
causal to the experience. Rather, directly causal pathways might be inhibitory pathway s with the areas that ‘light
up’ being results, not causes, of the menta l activity that presumably is being monitored. Despite limitations in
reliability and validity, functional brain imaging tools remain the best currently available assessments of in vivo
brain activity (
Bennett & Miller, 2010; Logoth etis, 2008).
In the Farrow et al. (2001) study, brain areas activated by social reaso ning and empathy were found to be simi-
lar to previous research studies on those cognitive tasks. Forgivability judgments activated many of the same
areas as empathy judgments, but the posterior cingulate activation was unique to the forgivability judg ments.
Authors personal copy
The posterior cingulate and its limbic connections are often activated by attentional tasks. Thus, this experiment
provides evidence of basic convergent validity for the imaging method to detect empathy judgments and sug-
gests discriminant validity evidence for a difference between judging whether an offence is worthy of empathy
or is forgi vable.
Equipment for central physiological assessment is typically available at large research centers and medical
schools. Individual researchers with technical expertise, funding, space, and support who are interested in fMRI
assessment could consult the Radiological Society of North America. All tasks were designed and presented
using the program Psyscope version 1.2.5 PPC [13] on an Apple Macinto sh G3 computer. The computer output
was displayed via an Epson EMP7300 projector on a screen visible to the subject in the scanner.
Results and Comments
Central physiological assessments should be evaluated for their reliability and validity independently, and as
indicators of the processes that might be involved in forgiveness and unforgiveness. Similar to peripheral mea-
sures, it is valuable to consider from interdisciplinary perspectives how these physiological measures are related
to the ways people embody granting and receiving forgiveness and unforgiveness responses. It is wise not to
over-interpret findings of physiological indices or to interpret them without considering self-reports or behavioral
indications that forgiveness has occurred in tandem with the physiological indicators.
Resource Distribution Measure of State Forgiveness
(Carlisle et al., 2012).
The resource distribution paradigm was designed to function as a behavioral indicator of state forgiveness,
thereby providing triangulat ing data or, in some cases where self-reports are unavailable, an indirect indication
of forgiveness. Behavioral measures might avoid common method variance from a packet of self-report question-
naires and might address some limitations of self-report forgiveness measures.
Laboratory-based behavioral measures of forgiveness are embedded within a particular method for inducing a
transgression and manipulating forgive ness-related variables. In the
Carlisle et al. (2012) experiment, participants
believe that they are exchanging resour ces with another participant in another room. The resources are raffle tick-
ets for a $50 prize. In Round 1, the partner takes 8 of 10 raffle tickets. In some conditions in Round 2, the partner
provides restitution by giving the particip ant 9 of 10 tickets; in other conditions the 9 of 10 tick ets come from the
experimenter. Also in Round 2, apology was manipulated by a note from partner to participant. In Round 3, the
participant can distribute raffle tickets between self and partner, constituting the behavioral measure of forgive-
ness. In a control condition, the Round 1 distribution of eight of ten tickets to the partner is sa id to occur by ran-
dom assignment .
This behavioral measure can be paired with self-rep ort measures. Particip ants rated the motivations underly-
ing their distribution in Round 3. One motive was ‘to express forgiveness.’ Subsequent unpublished research has
also included an item after each round asking participants how ‘forgiving’ they feel toward their partner, embed-
ded with a number of other emotion items. If not manipulating restitu tion and apology in Round 2, the proce-
dure can be shortened so that the offence occurs in Round 1, and the behavioral measure of forgiveness in
Round 2.
In the original sample of 136 university students (98 female), the mean number of tickets participants gave
their partners was 4.07 (SD 5 2.56). Participants who received restitution from their partner distributed more tick-
ets (M 5 5.63, SD 5 1.56) than those who did not receive restitution (M 5 4.18, SD 5 2.21). Participants receiving
Authors personal copy
an apology did not distribute more than those not receiving an apology. Participants who did not experience an
offence distrib uted a mean of 5 tickets (SD 5 1.11).
The reliabili ty of behavioral manipulations at instigating desired effects is difficult to estimate. Different exper-
imental situations, contexts, and other interacting variables make replication meaningless.
Carlisle et al. (2012) found that the behavioral measure was correlated with the self-report measure of the
desire to express forgiveness through the distribution (r 5 .40) and self-reported emp athy (r 5 .34).
Divergent/Discrim inant
Carlisle et al. (2012) found that the behavioral measure was negatively correlated with the self-reported moti-
vation to get pay-back for the partner’s previous distribution (a measure of unforgiveness; r 52.58). Also , the
behavioral measure of forgiveness showed no correlation with the motivation to maintain justice (r 52.11) or to
be fair (r 5 .27).
Carlisle, R.D., Tsang, J., Ahmad, N.Y., Worthington, E.L., Jr., Witvliet, C.V.O., & Wade, N.G. (2012). Do actions
speak louder than words? Differential effects of restitution and apology on behavioral and self-reported forgive-
ness. Journal of Positive Psychology, 7, 294305.
Results and Comments
This covert behav ioral measure of forgiveness can help address the social desirability and common method
variance co nfounds inherent in many self-report measures of forgiveness. Participants are likely not aware that
the researcher is assessing forgiveness. Giving up hard-won assets is more costly than is a generous forgiving
self-report, making self-p resentation contamination less likely. However, this measure is meant to complement
rather than replace self-report measures of forgiveness.
Defection in a Priso ner’s Dilemma as a Behavioral Measure of Forgiveness
(Exline et al., 2004, Study 5).
Behavioral forgiveness was assessed with multiple measures, including defection in a 10-trial prisoner’s
dilemma game, affective tone in a written response to a moderately antagonistic note, and money allocated to
a partner.
To create a moderately severe transgression, a computer program posing as another participant defected on
trials 1, 7, and 10 of a 10-trial prisoner’s dilemma. For all other rounds, the computer used a tit-for-tat strategy.
After Round 5, the computer sent a writte n message, ‘Is that the best you’ve got?’ Beh avioral forgivenes s was
interpreted to be the number of times the participant defected, the affective tone in a written response to the mes-
sage, and the amount of money the participant allocated to their partner out of $2. Self-reported forgiveness was
also assessed using ratings of positive and negative emotions felt toward the partner. Exline et al. used these
behavioral measu res to study how narcissistic entitlement re lated to forgivi ng behavior.
Exline et al. (2004) recruited 120 undergraduate students (59 women) for their study. The average amount of
money allocated to the partner was $1.53 (SD 5 $0.58).
Authors personal copy
Negative attitudes toward the partner were correlated with less money allocated (r 52.20, p , .05) and more
post-message defections (r 5 .20, p , .05).
Divergent/Discrim inant
Defections were negatively correlated with positive attitudes toward the partner (r 52.17, p , .10).
Exline, J.J., Baumeister, R.F., Bushman, B.J., Campbell, W.K., & Finkel, E.J. (2004). Too proud to let go:
Narcissistic entitlement as a barrier to forgiveness. Journal of Personality and Social Psychology, 87, 894-912.
DOI: 10.1037/0022-3514.87.6.894.
Results and Comments
Exline et al. found that narcissistic entitlement was associated with distributing less money to the partner and
with a mo re negative affective tone in a response to the partner’s hostile message. There was no relationship
between entitlement and defection after the message. This may mean that defection in a prisoner’s dilemma is a
less optimal behavioral measure of forgiveness compared to money distribution, or simply that prisoner’s
dilemma defection does not tap into a dimension of forgiveness that is related to narcissistic entitlement.
Resource distribution, this time in the form of money, and affective tone in written communication, seemed
to measure forgiveness, showing similar patterns of results to previous studies utilizing self-report measures
Exline et al., 2004, Studies 14). Weiner, Graham, Peter, and Zmuidinas (1991) used resource sharing in children
who shared letters in a word-puzzle solving task.
Santelli, Struthers, and Eaton (2009, Study 2) combined
a resource distribution measure with a social distance measure in studying 70 undergraduates (43 women).
A transgress ion was induced in the laboratory, and participant behaviors were observed to assess forgiveness.
Participants were to ld that they were working online with another participant. The experimenter instructed both
participants to avoid dragging the cursor into a covered area on the computer screen. The partner failed to
adhere to instructions, freezing the computer and causing the participant to have to return later to complete
the experiment. For the first behavioral measure of forgiveness, participants were given 10 raffle tickets for a
$50 prize and allowed to anonymously divide the tickets between themselves and their partner. Second, the parti -
cipants were led to a row of seven chairs with a jacket draped over one of the chairs. The jacket ostensibly
belonged to their partner. The experimenter recorded how far from the offender the participant chose to sit (typi-
cally from two to five chairs away). Researchers found similar results using resource distribution, social distance,
and self-reports.
Overall, both the resource distribution and social distance forms of behavioral forgiveness are mo re covert and
costly indications of forgiveness than self-reports. Behavioral measures of forgiveness tend to be weak.
Sometimes treatment and control conditions did not differ, and differences were small when they did occur.
Nevertheless, we suggest devising laboratory manipulations that provide strong, valid, and reliable measures of
forgiveness are well worth the effort.
Trait Forgivingness Scale (TFS)
(Berry et al., 2005).
The TFS is a brief self-report measure of interpersonal dispositional forgiveness.
Forgivingness is defined as a disposition of being forgiving across time and situations. The TFS consists of
ratings of four statements that describe willingness to forgive in different situations and over time, a summary
statement of being a forgiving person, and five statements (reverse scored) that describe generally or frequently
Authors personal copy
feeling bitterness, resentment, unwillingness to forgive, and desire for revenge. High scores indicate a disposition
to forgive. Each item is rated on a 5-point Likert-type response scale from 1 5 Strongly Disagree to
5 5 Strongly Agree.
Berry and Worthington (2001) reported an initial study in which a 15-item scale assessing trait forgivingness
was correlated with other measures of forgivingness. In a pilot study to
Berry et al. (2005), the authors reduced
the length of the TFS, using Rasch modeling within item response theory (IRT), to its final version of ten items.
The TFS had good person separation reliability, item separation reliability, and mean square weighted fit statis-
tics. The assumptions of IRT suggest that items are differentially harder to endo rse along a scale measured in
logit units. The obje ctive of IRT is to sc ale items from easy to endorse to hard to endorse in almost equal steps
with clearly defined items (i.e., each item has a small standard error of measurement). Item characteristic curves
are then fit to an ideal IRT model. Once this is done, it is assumed that all items measure the same latent con-
struct (though not equally well, as assumed by classical measurement theory, CMT), and CMT can be used to
further support the estimated internal consi stency (albeit wi th somewhat lower alphas if computed than
when items are constructed under CMT alone) of the of scale scores, consistency of the factor structure (using
EFAs), and fit to the measurement model (using CFAs).
Berry et al. (2005) examined the psychometric properties of the TFS using four samples of 179, 233, 80, and 66
undergraduate students. The first two samples were from a Mid-Atlantic state university, the third from a private
university in the Pacific Northw est, and the fourth from a large state uni versity in the San Francisco Bay area.
In the fourth study, partners of students also participated.
Internal Consistency
Berry et al. (2005) reported all IRT and classical measurement statistics across four studies. Cronbach alpha
coefficients were found to range from .74 to .80 across the four stu dies.
TFS scores were stable over eight weeks, r 5 .78. The correlations between item difficulty estimate s across the
eight weeks was r 5 .98 (
Berry et al., 2005). Berry, Worthington, Parrott, O’Connor, and Wade (2001) employed
four samples and analyzed the combined sample, adducing evidence in support of temporal stability of the TFS
(8 weeks, r 5 .69; correlations between item difficulty estimates across the eight weeks was r 5 .99), and construct
validity of the TFS against similar instruments with which the TFS was correlated.
In two studies,
Berry et al. (2005) reported positive co rrelations (.50 and .48) between scores on the TFS and
the Tran sgression Narrative Test of Forgivingness (TNTF;
Berry et al., 2001) (the TNTF involves five scenarios
relevant for college students; IRT was employed in its construction). Across three studies, scores on the TFS
were positively correlated with the Big Five Inventory: agreeableness (mean r 5 .56), and conscientiousness (mean
r 5 .25), Interpersonal Reactivity Inventory: empathic concern (mean r 5 .30), and perspective taking (mean r 5 .29).
In romantic couples, self-ratings and partner ratings were positively correlated (r 5 .35) (
Berry et al., 2005).
Divergent/Discrim inant
Berry et al. (2005) reported that across three studies, scores on the TFS were negatively correlated with the
Trait Anger Scale (mean r 52.47), Big Five Inventory: neuroticism (mean r 52.38), Fear Questionnaire (mean
r 52.36), Dissipation Rumination Scale (2 .65), Aggression Questionnaire: anger (2 .40), hostility (2 .35), verbal
aggression (2 .10), and physical aggression (2 .17). Also, negligible correlations were found with the Big Five
Inventory: extraversion (.09), and openness (.08) (
Berry et al., 2005).
Construct/Factor Analytic
Berry et al. (2005) used item response theory to provide evidence for structural validity rather than exploratory
or confirmato ry factor analytic methods.
Authors personal copy
Berry, J.W., Worthington, E.L., Jr., O’Con nor, L., Parrott, L. III, & Wade, N.G. (2005). Forgivingness , vengeful
rumination, and affective traits. Journal of Personality, 73, 183225.
Results and Comments
Studies suggest that the TFS measures trait forgivingness in a rapid self-report format. The TFS has been used
across a variety of populations. PsycINFO (accessed December 6, 2013) lists 117 citations of the article, suggesting
widespread usage. Trait forgivingness has been related to many other personality characteristics (see
Fehr et al.,
) and mental health (see Griffin, Worthington, Wade, Hoyt, and Davis in press) and physical health indices
Worthington et al., 2007b).
Directions: Indicate the degree to which you agree or
disagree with each statement below by using the follow-
ing scale:
5 5 Strongly Agree
4 5 Mildly Agree
3 5 Agree and Disagree Equally
2 5 Mildly Disagree
1 5 Strongly Disagree
____1. People close to me probably think I hold a
grudge too long.
____2. I can forgive a friend for almost anything.
____3. If someone treats me badly, I treat him or
her the same.
____4. I try to forgive others even when they don’t
feel guilty for what they did.
____5. I can usually forgive and forget an insult.
____6. I feel bitter about many of my relationships.
____7. Even after I forgive someone, things often
come back to me that I resent.
____8. There are some things for which I could
never forgive even a loved one.
____9. I have always forgiven those who have
hurt me.
____10. I am a forgiving person.
Scoring Instructions: Reverse score the following
items: 1, 3, 6, 7, and 8. Add the scores on each item.
The range is from 10 to 50. High Scores will indicate
more trait forgivingness.
Reproduced with permission.
State Self-Forgiveness Scale (SSFS)
(Wohl et al., 2008).
The SSFS is the first self-rep ort measure of one’s current feelings of self-forgiveness regarding a particular
wrongdoing. Whereas trait self-forgivingness had been measured for years (see
Thompson et al., 2005), states of
self-forgiveness had been inferred by assessing reduced regret and remorse or by excusing one’s own transgres-
sions (see
Fisher & Exline, 2006).
The original version of the SSFS was composed of 30 rationally-derived items but was revised to contain 18
items. An unrestricted EFA found that responses grouped along a Self-Forgiveness Feelings and Action s (SFFA)
subscale and a Self-Forgiveness Behaviors (SFB) subscale. The first eight items (i.e., ‘As I consider what I did that
was wrong, I punish myself’) load onto the SFFA; the follow ing nine (i.e., ‘As I consider what I did that was
wrong, I believe I am worthy of love’) load onto the SFB. The final item provides a validity check for the measure
at large (‘As I consider wha t I did that was wrong, I have forgive n myself’). Responses to items on each subscale
range from 1 5 not at all to 4 5 completely (
Wohl et al., 2008).
Authors personal copy
Wohl et al. (2008) constructed the SSFS and examined its psychometric proper ties using 113 students at the
University of Oklahoma (Study 1) and to 60 undergraduates at Carleton University (Study 2). Means and stan-
dard deviations for the subscales were as follows: for the SFFA, M 5 3.34 (SD 5 .84); for the SFB, M 5 4.08
(SD 5 1.06).
Internal Consistency
Cronbach alpha coefficients fo r the SFFA were reported as α 5 .74; for the SFB, α 5 .89 (
Wohl et al., 2008).
SSFS estimates of temporal stability have yet to be determined.
Participants who reported higher self-forgiven ess on the final item also scored higher on the two subscales of
the SSFS Self-Forgiveness Feelings and Actions and Self-Forgiveness Beliefs. Evidence for the convergent valid-
ity of the SSFS can also be seen in its correlations with depression (SFFA, r 52.42; SSBA, r 52.39) and self-
blame (SSFA, r 52.38; SSB, r 52.36).
Divergent/Discrim inant
SSFS scores were not correlated with tendency to forgive as measured by
Brown’s (2003) four-item Tendency
to Forgive Scale (SFFA, r 5 .03; SSBA, r 5 .14) . Neither subscale of the SSFS was correlated significantly with guilt
(SFFA, r 52.08; SSBA, r 52.05), state self-esteem (SFFA, r 5 .04; SSBA, r 5 .16), or life satisfaction (SFFA, r 5 .23;
SSBA, r 5 .20;
Wohl et al., 2008).
Construct/Factor Analytic
Items scores were subjected to exploratory factor analysis using principal axis factoring with oblimin rotation.
Two correlated factors emerged. Following that , item response theory applied to each subscale separately
resulted in SFFA having an item reliability of .99 and SFB having an item reliability of .95.
Wohl, M.J.A., DeShea, L., & Wahkinney, L. (2008). Looking within: Measuring state self-forgiveness and its
relationship to psychological well-being. Canadian Journal of Behavioural Science, 40,110.
Results and Comments
Wohl et al. (2008) tentatively suggested that the SSFS is a psychometrically sound measure of self-forgiveness
of a specific offence and that it has evidence supporting its estimated reliability and construct validity. The SSFS
subscales provide a closer look at two hypothesized processes of self-for giveness, and each provides pred ictive
value to the achievement of self-forgivenes s (
Wohl et al., 2008). However, the SSFS remains largely unstudied,
with only a handful of studies using it is a measure of self-forgiveness in a peer-reviewed journal (e.g.,
Root, Yadavalli, Martin, & Fisher, 2011
). The SSFS is available in English, and though the items are listed in Wohl
et al. (2008)
, users must request permission to use it. The SFSS requires much more psychometric support before
it can be confidently used as a measure of state self-forgiveness. It is the best measure available, but much more
work is needed if researchers are to distinguish self-forgiveness from constructs such as self-condoning, self-
acceptance, and pseudo-self-forgiveness or letting oneself off the hook.
Heartland Forgiveness Scale (HFS)
(Thompson et al., 2005).
Authors personal copy
The HFS measures dispositional forgivingness of self, others, and situations. The focus of attention here is the
dispositional forgivingness of self subscale. (There is some controversy about whether a construct like disposi-
tional forgivingness of situations is meaningful (for a critique, see
Enright & Rique, 2004). Dispositional forgiving-
ness of others is less psychometrically sound (probabl y due to its fewer items) than some other measures of
dispositional forgivingness (e.g., TFS, see abov e).
The pilot version of the HFS consisted of 90 items (30 items for each type of forgive ness, with equal positi vely
and negatively worded items). Following an EFA, the scale was reduced to three subscales with six items each
(total HFS scale 5 18 items). The first subscale, HFS Self, measures self-forgiveness (e.g., ‘With time, I am under-
standing of myself for mistakes I’ve made’). The second subscale, HFS Other, measures interpersonal forgiveness
(e.g., ‘If others mistreat me, I continue to think badly of them’). The third subscale, HFS Situation, measures for-
giveness of situations (e.g., ‘When things go wrong for reasons that can’t be controlled, I get stuck in negative
thoughts about it’). Rating poss ibilities range from 1 5 Almost Always False of Me to 7 5 Almost Always True of Me.
Scores on each subscale range from 6 to 42; total HFS scores range from 18 to 126. We emphasize HFS Self below.
Thompson et al. (2005) reported six studies related to the development of the HFS. All samples were from a
large Midwestern city (5 of 6 studies from the university students there) . Sam ple sizes were from Studies 1 to 6,
respectively, 499 (Study 1), 1111 (Study 2), 504 (Study 3), 123 (Study 4; city dwellers in romanti c relationships), 55
(Study 5), and 230 (Study 6). Means on the HFS Self were provided for Study 2, M 5 30.99, SD 5 6.17; for Study 3,
M 5 31.25, SD 5 5.67; and for Study 4, M 5 31.89, SD 5 5.75. Means for the HFS Other ranged from 30.01 to 30.41
(SDs from 6.0 to 6.5). Means for HFS Situation ranged from 30.25 to 32.11 (SDs from 5.8 to 6.5).
Internal Consistency
The range of Cronbach alpha coefficients, reported for studies 2-4, are for HFS Self, α 5 .72 to .76; for HFS
Other, α 5 .78 to .81; for HFS Situation, α 5 .77 to .82 (
Thompson et al., 2005).
Three-week temporal stability estimates (Study 3) were for HFS Self, r 5 .72; for HFS Other, r 5 .73; for HFS
Situation, r 5 .77. The nine-month temporal stability estimates (Study 4) were for HFS Self, r 5 .69; for HFS Other,
r 5 .69; for HFS Situation, r 5 .68 (
Thompson et al., 2005).
HFS Self scores correlated with the
Mauger et al. (1992) Forgiveness of Self scale, r 5 .61, and with the
Multidimensional Forgiveness Inventory (MFI)-Self (
Tangney, Boone, Fee, & Reinsmith, 1999), r 5 .33. Scores on
the HFS Other subscale correlated with the Mauger et al. Forgiveness of Others scale, r 5 .53, p , .001, and with
the MFI-Other scale, r 5 .47. Scores on the HFS Situation subscale were correlated with Mauger et al.’s total scale,
r 5 .51, and the MFI-total, r 5 .42. The HFS subscale scores were related to mental health variables such as and
satisfaction with life (for HFS Self, r 5 .39; for HFS Other, r 5 .31; for HFS Situation, r 5 .39).
Divergent/Discrim inant
The HFS subscale scores correlated negatively with mental health variabl es suc h as depression (for HFS Self,
r 52.44; for HFS Other, r 52.27; for HFS Situation, r 52.40), and trait anger (for HFS Self, r 52.32; for HFS
Other, r 52.51; for HFS Situation, r 52.43).
The HFS was also related to the MarloweCrowne Social Desirability Scale (for HFS Self, r 5 .27; for HFS
Other, r 5 .34; for HFS Situation, r 5 .30; for HFS total, r 5 .38).
Thompson, L.Y., Snyder, C.R., Hoffman, L., Michael, S.T., Rasmussen, H.N., Billings, L.S., Heinze, L., Neufeld,
J. E., Shorey, H.S., Roberts, J.C ., & Roberts, D.E. (2005). Dispositional forgiveness of self, others, and situations.
Authors personal copy
Journal of Personality, 73, 313-359. DOI: 10.1111/j.14676494.2005.00311.x.
(Retrieved Janua ry 9, 2014).
Results and Comments
The HFS allows assessment of three forms of dispositional forgivingness (self-forgiving ness, other-forgiving-
ness, forgivingness of situations). The self-forgivingness subscale, which is the primary focus here, is composed
of six items five of which entail self-condemnation and one of which alludes to understanding oneself.
Thus, the scale is more appropriate as a measure of self-condemnation and does not speak as strongly to how
one rids oneself of self-condemnation. Some have suggested that many people do so by letting oneself off of the
hook or pseudo-self-forgiveness (see
Hall & Fincham, 2005). Particularly noteworthy are the nine-month temporal
stability estimates and the finding that the HFS predicted mental health and relationship quality. The scale has
been translated into Greek (translator: Despina Moraitou), Japanese (
Osanai & Furukawa, 2005), Thai (translator:
Ruck Chunhakan), and Turkish (translator: Asli Bugay) (for details, see:
(Retrieved Janua ry 9, 2014).
The notion that one can forgive situation s (e.g., natural disasters) is unique to the HFS, and because they are
non-agentic, some forgiveness re searchers do not embrace that construct. The HFS Self and HFS Other are good
brief measures of two types of dispositional forgivingness. Although
Thompson et al. (2005) reported total scores,
we suggest that theoretical reasons would argue against this.
In evaluating the status of measurement of constructs related to forgiving, we suggest new directions in two
areas. First, method ological advances from the field of assessment need to be applied to forgiveness constru cts.
Second, theoretical advances in understanding forgiveness need to be considered.
Methodological Advances in Assessment
IRT has been underused in constructing self-report measures of forgiveness. Only
Berry et al. (2001), Berry
et al. (2005)
, and McCullough et al. (2010) have employed IRT in this regard. The great advantage of IRT is its
promise for creating brief measures that have strong psychometric properties. Computer assessment has also
been underused. In today’s world, so many people are internet active that developing good computer-based
assessments seems a next step. Many researchers and clinicians simply load the existing measures onto the com-
puter and either email scales, use local computers in-office or in-lab for computer assessment,