Forgiveness is crucial for maintaining social relationships and mediating interpersonal conflict. Growing evidence has found that forgiveness benefits physical and psychological health, as well as wellbeing (e.g., Arias et al., 2023; Enright & Fitzgibbons, 2015; Lee & Enright, 2019; Toussaint et al., 2015; Wade et al., 2014). Specific to mental health, forgiveness helps individuals relieve stress (Liu et al., 2021) and has served as a protective factor for depression among adolescent girls (Kuzubova et al., 2021).

Traditionally, forgiveness was regarded as a religious or philosophical concept (Kim et al., 2022a, b), being discussed in various traditions like Christianity, Judaism, and Islam (Enright et al., 1992). It became a research theme in psychology in the late 1990s (Gao et al., 2022). Based on theorisation, forgiveness has been used in different forms in contemporary psychotherapy and counselling. Due to the distinctive moral component (Enright et al., 1998), forgiveness has recently been adopted in the legal research field; for example, forgiveness has been considered as a relevant process for therapeutic jurisprudence (Fariña et al., 2020; Fariña & Oyhamburu, 2020, 2021) and restorative justice (Fariña & Oyhamburu, 2021; Peterson et al., 2017).

A unified definition of forgiveness does not yet exist (Belicki et al., 2020). According to the American Psychological Association Dictionary of Psychology, forgiveness involves such a voluntary shift in one’s perceptions and behaviours that one would express positive emotions (e.g., compassion, generosity) rather than resentment (VandenBos, 2015). In addition, forgiveness is generally associated with decreased negative affect and altered cognitions, motivations, and behaviours toward the individual’s offender (Enright et al., 1989; Kaleta & Mróz, 2018; McCullough et al., 1998; Rye & Pargament, 2002; Tao et al., 2021). As forgiveness usually occurs in a social context, psychologists occasionally term it “interpersonal forgiveness” to highlight its social dynamics. Enright et al. (1998) further maintained that forgiveness is more than a psychological construct; it is a moral virtue within the Aristotelian tradition of virtue ethics, highlighting its volitional sense (Kim et al., 2022a, b).

In terms of measuring forgiveness, the Enright Forgiveness Inventory (EFI) developed by Subkoviak et al. (1995) is one of the most widely used interpersonal forgiveness measures across countries. The EFI has been used as an instrument to evaluate interpersonal forgiveness in therapy and educational interventions. It has been used as a research instrument among specific samples, such as victims of workplace harassment, sexual harassment, robbery, and family violence (Fariña et al., 2023). At the theoretical foundation of the EFI is the idea that interpersonal forgiveness occurs when a person considers abandoning both the right to resentment, negative cognitions, and negative behaviours towards the person who unjustly offended them (Enright et al., 1998). Simultaneously, the person begins to harbour compassion, mercy, and possibly love or positive emotions toward the offender. Thus, Enright et al. (2022) emphasised that forgiveness implies a reduction in negative affect, cognitions, and behaviours; and that forgiveness is associated with an increase in the presence of positive affect, cognitions, and behaviour. The multi-faceted nature of forgiveness indicates a multidimensional, rather than a unidimensional, model.

The EFI originally consisted of 60 items (Enright & Rique, 2004; Subkoviak et al., 1995) and was later reduced to 30 items (Enright et al., 2022). It measures attitudes around forgiveness via three components: affect, cognitions, and behaviours (Rodrigues et al., 2015). An individual can display different valences on each of the three components. For example, the offended person can simultaneously harbour positive and negative affect towards the offender (Enright et al., 1998; Fariña et al., 2023). As mentioned, the EFI (30-item version) can be theoretically structured into six factors: positive affect, positive cognition, positive behaviour, negative affect, negative cognition, and negative behaviour, with internal consistency ranging from good (α = 0.83) to excellent (α = 0.95) for the six scales across multiple countries (Enright et al., 2022). As a part of the scale development, the EFI has a one-item question, which measures the participant’s overall forgiveness tendency and is theoretically associated with all factors (i.e., subscales). While Spanish is one of the world’s most popular languages, the validation of the EFI among Spanish speakers had not been conducted.

Although there are several instruments for forgiveness in English, only two established scales have been used in Spanish to the authors’ knowledge. One is the Transgression-Related Interpersonal Motivations Inventory (McCullough et al., 2006), which cannot be explicitly considered a scale of forgiveness, but one for unforgiveness. Another is the Spanish version of the Heartland Forgiveness Scale (Thompson et al., 2005), which was translated and validated in Argentina by Casullo and Liporace (2005). As reviewed by Freedman (2022), the construct validity of the Heartland Forgiveness Scale is questionable, and this scale failed to capture the moral sense of forgiveness. Moreover, there are two Spanish scales related to forgiveness: La Escala del Perdón en la Relación de Pareja (Rosales-Sarabia, 2013; The Forgiveness Scale in the Couple Relationship) and La Escala de Actitud frente al Perdón (Martínez & Parra, 2019; The Attitude Towards Forgiveness Scale). These instruments do not possess strong theoretical foundations or cross-cultural validation and replication.

Such research appears to be dramatically insufficient in South America. Intrapsychically, forgiveness has appeared beneficial for an individual’s mental health and psychological wellbeing in South America (Rique et al., 2020). Interpersonally, forgiveness has been an important factor for sustaining peace in the Colombian post-conflict period (Cortés et al., 2016).

Overall, forgiveness studies with South American populations are needed, although validation research of its measurements is scare. As a part of an international project aiming to validate the EFI – 30-item version (EFI-30) in Spanish across Latin-American countries, this study examined the EFI-30’s scalability in Argentina. A validated EFI would be useful for forgiveness research and forgiveness therapies in this region. Specific research questions included: (1) whether the EFI-30 shows a good structural validation in Argentina as in other countries; and (2) whether its six subscales show strong associations with its convergent validator (i.e., its one-item validator).

Method

Procedure

Four tertiary education organisations with geographic differences were first selected in Buenos Aires, Argentina. The research team then randomly contacted teaching staff of the first and second years to invite their students to partake in our online survey. Participants were encouraged to resend the survey to their contacts such as family members and friends. Participation was voluntary and anonymous, with no monetary incentives. The survey was conducted in Spanish and administrated on Google Forms. During the translation, the team considered Argentinian linguistic expressions with the help of four professors in Argentina. To control the order effect of questionnaire design, we used two versions (referred to as versions A and B hereafter): one with a social desirability scale before the EFI, another with a social desirability after the EFI. Since both versions yielded similar results, we used the combined sample in the report. All participants were provided with informed consent. Prior to data collection, this study was reviewed and approved by the Research Ethics Committee of the Faculty of Education and Sports Sciences, University of Vigo, Spain (Ref. 11-250322).

Participants

All participants were originally from Argentina. As shown in Table 1, the sample (N = 472) was dominated by female (81.1%), well-educated participants. Nearly half (47.2%) the participants were married or cohabiting in a relationship.

Table 1 Demographic information of participants (N = 472)
Full size table

Measures

Enright Forgiveness Inventory (EFI-30)

The EFI-30 (Enright et al., 2022) was used, employing a Spanish adaptation (Fariña et al., 2023) tailored to Argentinian idiomatic expressions. Initially, an English-Spanish bilingual teacher translated the EFI-30 into Spanish. Subsequently, another bilingual teacher conducted a back-translation into English. Two additional professors in Argentina rigorously examined and compared the Spanish translation within the Argentinian context. The back-translation and cultural validation processes adhered to the guidelines outlined by Muñiz et al. (2013), ensuring linguistic and cultural relevance. The finalised version used can be accessed in the Supplemental materials.

The EFI is an established scale for the extent to which one forgives another person after a specific situation of offence. The EFI starts with introductory questions tapping information about (1) the degree of hurt felt after the offensive situation (from no hurt [1] to a great deal of hurt [5]), (2) who caused the hurt, and (3) how long ago the offence occurred. Participants were also asked to write a brief description of the offence. In the inventory section, all 30 questions were measured on a 6-point Likert scale (from strongly disagree [1] to strongly agree [6]).

The EFI’s 30 questions assessed six psychological “stances” of victims toward offenders: positive affect, negative affect, positive behaviour, negative behaviour, positive cognition, and negative cognition. Specifically:

To evaluate affect, it was indicated:

This set of items deals with your current feelings or emotions right now toward the person. Try to assess your actual feeling for the person on each item. For each item, please check the appropriate line that best describes your current feeling (warm, tender, unloving, repulsed, cold, dislike, caring, affection, friendly, disgust).

To evaluate behaviour, it was indicated:

This set of items deals with your current behaviour toward the person. Consider how you do act or would act toward the person in answering the questions. For each item, please check the appropriate line that best describes your current behaviour or probable behaviour (show friendship, avoid, ignore, neglect, not attend to him/her, lend him/her a hand, establish good relations with him/her, stay away, do a favour, aid him/her when in trouble).

To evaluate cognition, it was indicated:

This set of items deals with how you currently think about the person. Reflect on the kinds of thoughts that occupy your mind right now regarding the person. For each item, please check the appropriate line that best describes your current thinking (horrible, of good quality, dreadful, worthless, a good person, a bad person, wish him/her well, disapprove of him/her, think favourably of him/her, hope he/she succeeds).

In addition, the EFI-30 evaluates “pseudo-forgiveness” through five items, on a 6-point Likert scale (from not at all [1] to complete forgiveness [6]). These items are used as a convergent validator of the instrument. Participants were instructed to evaluate five questions with the prompt:

In thinking through the person and event you just rated, please consider the following final statements: There really was no problem now that I think about it; I was never bothered by what happened; the person was not wrong in what s/he did to me; my feelings were never hurt; what the person did was fair.

Participants (n = 23) with a score ≥ 20 were excluded from the study.

One-item forgiveness measure

Moreover, the one-item question (To what extent have you forgiven the person?) was measured on a 5-point Likert scale (from not at all [1] to complete forgiveness [6]), serving as the convergent validator.

Marlowe-Crowne Social Desirability Scale (MC-SDS)

The brief Spanish version of the 18-item version of the Marlowe-Crowne Social Desirability Scale (MC-SDS; Crowne & Marlowe, 1960; Gutiérrez et al., 2016) was used to assess social desirability. A binary scale (true [1], false [0]) was used for all items. Ten items were reverse-coded so that higher total scores reflect a higher tendency of adhering to socially accepted behaviours. The internal consistency reported by Gutiérrez et al. (2016) for Spanish adults was α = 0.78.

Demographic variables

Sex, gender, educational level, relationship status, and country of origin were assessed with study-created questions.

Data analysis

The primary purpose of this study was to validate the Spanish version of the EFI-30 scale among Argentina participants. Univariate tests on data distribution were first conducted. EFI items showed relatively normal distributions, with | skewness | < 1.52 and | kurtosis | < 1.49 (for criteria, see Enright et al., 2022). Consequently, confirmatory factor analysis (CFA) was performed to evaluate structural validity. Given the categorial nature of EFI items, the weighted least-square mean and variance-adjusted estimator was applied in factor analysis. Model fit was assessed according to the guidelines by Hu and Bentler (1999). Specifically, models are considered a “good fit” when comparative fit index (CFI) and Tucker-Lewis index (TLI) values ≥ 0.95, root mean square error of approximation (RMSEA) index ≤ 0.05 (≤ 0.08 adequate), and standardized root mean square residual (SRMR) value ≤ 0.08. Rather than focusing on one fit index, Hu and Bentler suggested a combinational rule of evaluating two indices (i.e., TLI and SRMR; RMSEA and SRMR; CFI and SRMR) as a way to mitigate both type I and II errors). To further validate the multidimensional structure of the EFI, a CFA was also conducted to compare the one-factor versus six-factor models with the fit indices above.

Per conventional practice, we also calculated Cronbach’s alpha (α) and McDonald’s omega (ω) for internal consistency. To assess convergent validity, consistent with Enright et al.’s (2022) study, we further conducted a Pearson’s correlation between EFI constructs and the single forgiveness item, as well as the social desirability results. Moreover, we used the Stata package “validscale” (Perrot et al., 2018) to generate further scalability metrics including Loevinger’s H coefficients (Mokken, 1971), which indicate the internal homogeneity among items with values above 0.50 reflecting strong scalability. Also, pairwise deletion was used in descriptive analyses, whereas full information maximum likelihood was used in factor analysis. IBM’s SPSS Statistics (Version 28) was used for data management and descriptive analysis. Mplus 8.2 was used for CFA. Stata 15.1 was used to perform the package “validscale” (Perrot et al., 2018). Significance levels of 0.05 and 0.01 are reported.

Results

Structural validation

Using CFA, the structural validity of the six-factor EFI-30 was established, as models showed an excellent “good fit” in line with the rule of combinational evaluation: RMSEA = 0.08 [90% CI: 0.075, 0.083], CFI/TLI = 0.97/0.96, SRMR = 0.04. The findings suggest a satisfactory structural validity of the six factors in EFI-30. As shown in Table 2, most standardized factor loadings were above 0.80. The lowest factor loading was for “unloving” (Λ = 0.56), belonging to subscale/factor negative affect. A CFA was also performed separately in versions A and B, both showing a good fit (Version A: RMSEA = 0.08 [90% CI: 0.075, 0.088], CFI/TLI = 0.96/0.96, SRMR = 0.05; Version B: RMSEA = 0.08 [90% CI: 0.069, 0.082], CFI/TLI = 0.97/0.97, SRMR = 0.04), suggesting a robust cross-sample validity of the scale. A CFA with the one-factor forgiveness item was also performed, resulting in a poor fit: RMSEA = 0.14 [90% CI: 0.132, 0.140], CFI/TLI = 0.89/0.88, SRMR = 0.07. This result suggests that the six-factor model outperformed the unidimensional structure.

Table 2 Standardized factor loadings of the Enright Forgiveness inventory – 30 (EFI-30)
Full size table

Subscale validation

Table 3 summarizes the reliability and validity of the EFI-30 subscales. All subscales had good internal consistency (i.e., α ≥ 0.85; ω ≥ 0.86). The composite reliability of the entire EFI-30 scale was excellent: α = 0.98, ω = 0.98. In addition, all Loevinger’s H values were over 0.50, indicating strong homogeneities in each subscale.

Table 3 Reliability and Validity of the Enright Forgiveness Inventory – 30 (EFI-30)
Full size table

In terms of convergent validity, all subscales were strongly associated with the one-item forgiveness measure (r > .45) in a consistent direction, with positive items showing positive correlations and negative items showing negative correlations (for detail, see Table 3). As for discriminant validity, we found no relationships between social desirability and positive cognition, positive behaviour, negative affect, negative cognition, and negative behaviour EFI subscales, but the positive affect subscale had a negative but small association (they only shared 1.21% of the variances) with social desirability: r(454) = − 0.11, p = .019.

Discussion

This study aimed to develop and validate the EFI-30 in the Argentine context. Our results confirmed the factorial structure of the original six factors, namely, positive affect, negative affect, positive behaviour, negative behaviour, positive cognition, and negative cognition, indicating a consistent construct validity with previous studies (Enright et al., 2022). In addition to the EFI-30 items’ construct validity, our findings also proved their discriminant and convergent validity in our Argentine scale. The internal consistency values of both subscales (α/ω > 0.80) and the entire scale (α/ω > 0.95) were deemed robust. We further tested the replicability of Enright et al.’s (2022) international study, confirming the validity of the EFI-30 in the Argentine context. Given the insufficient research and strong need for forgiveness, our validation study provides multiple future research directions.

As reviewed by Freedman (2022), the EFI-30 is based on a relatively complete theoretical conceptualisation compared to other self-report measures of forgiveness. Its moral component of forgiveness (i.e., seeing the offender as a human being) makes the EFI-30 a unique measurement (Enright et al., 1998). Using this view of forgiveness, therapies that are not based on individualistic self-help approaches could be developed (Freedman, 2022). For instance, forgiveness-based therapy has been applied to establish coparenting relationships after divorce (Kluwer et al., 2021), for mediation processes (Zivojinovic, 2017), as well as in restorative justice (Peterson et al., 2017; Rossner, 2019; Shapland, 2016) and therapeutic jurisprudence (Fariña et al., 2020; Fariña & Oyhamburu, 2020, 2021) efforts.

The strong scalability of the Argentine EFI-30 provides multiple avenues for future research. For instance, it would be beneficial to validate the EFI-30 among adolescents in Argentina. Studies on interpersonal issues, including inter-partner conflict, familial mediation, school mediation, parental coordination, and imprisonment, might also now use the EFI-30 to examine, for example, the relationships between physical and mental health, fulfilment, satisfaction, and the recidivism of offenders.

Students comprised most of the study sample, which is this study’s main limitation. Also, our sample was 80% female, which means males were underrepresented. Future studies should both include more participants from non-student samples and use geographically stratified sampling methods. Given our cross-sectional design, test-retest reliability is unable to be investigated. Although convergent validators were used in our study, the extent to which the EFI-10 is related to an external criterion (e.g., observed forgiving behaviours) was not examined; future studies would benefit from including criterion validators. Nevertheless, in response to Freedman’s (2022) call, our study demonstrating the strong ecological validity of the EFI-30 in Argentina significantly extends the prior research focus that is heavily based on Western, educated, industrialized, rich, and democratic (WEIRD) populations.