Dimensions underlying sixteen models of forgiveness and reconciliation.

Numerous models of interpersonal forgiveness and reconciliation have proliferated, while the number of constructs common to these models and how these constructs interrelate has gone untested. Sixteen path models of forgiveness and reconciliation were examined in 180 predominantly Christian college students. Results of factor analysis suggested five underlying constructs--Hurt and Anger, Receiving God's Forgiveness, Emotional Forgiveness, Empathy, and Reconciliation. Structural equations modeling supported a model whereby people simultaneously experience Hurt and Anger while Receiving God's Forgiveness, and which then relate to Emotional Forgiveness, then Empathy, then Reconciliation. These results suggest substantial overlap among models, empirically support a distinction between Emotional Forgiveness and Reconciliation, and suggest that Receiving God's Forgiveness is an important factor for religious people beginning reconciliation.

**********

Psychologists have given increased consideration to explaining the dynamics of forgiveness and reconciliation in the last decade (McCullough, Pargament, & Thoresen, 2000; Worthington, 1998a). Although a number of path models of forgiveness and reconciliation have been suggested to explain and aid individuals' attempts to for give, the efficacy of such models to actually promote forgiveness and reconciliation has gone untested (McCullough & Worthington, 1994; Sells & Hargrave, 1998). The first purpose of this study is to explore the underlying dimensions of sixteen models of forgiveness and reconciliation in order to determine the dimensions underlying the set of models. This will be done via common factor analysis on items created to represent the steps in each of the path models. The second purpose of the study is to empirically evaluate the relationships among the constructs when attempting to forgive or reconcile. This will be done via structural equations modeling using the obtained factors.

The Christian Basis of Forgiveness

The Christian basis of forgiveness lies in the redemptive life, death, and resurrection of Christ. As Meek and McMinn (1997) explain,

... forgiveness in the Christian Scriptures is much more than religious
ritual. It is a progression of healing where people are confronted with
the grace and mercy of God, despite their continual failure to deserve
it. They learn to proffer the same grace and mercy to others in full
awareness of their own fallibility. (p. 51)

Biblically, the Scriptures offer much in the way of a Christian ethic concerning forgiveness. When asked by Peter how many times one should forgive, Jesus responded by saying "up to seventy times seven" (Matt. 18: 22). In concluding the parable of the unmerciful servant, Jesus warned that, like the master in the parable, His Heavenly Father would also be unhappy if each of his disciples did not also forgive others their trespasses (Matt. 18:35). When instructing the disciples to pray, Jesus prayed by saying "Father forgive us our debts, as we forgive our debtors" (Matt. 6:12). Even while being crucified, Jesus prayed that God would have mercy on the people responsible, praying "Father forgive them, for they know not what they do" (Luke 23:34). Although much is offered Biblically in terms of an ethic concerning forgiveness, little is offered in the Bible in terms of how to actually complete the forgiveness process. Thus, the way in which people actually complete the process of forgiveness has been a matter of speculation from theorists from both the fields of psychology and theology.

Selection of Forgiveness Models

Theories about the nature of forgiveness have grown rapidly in recent years, as increasingly, theorists have posited models of forgiveness without attempts to identify overlap among models or to test competing theories about the process of forgiveness. The following criteria were used to select forgiveness models in this study: (a) the study was a published work from either the fields of psychology or theology, (b) the model must be referring explicitly to the process of forgiveness or reconciliation (or both), and (c) the model must have been available in the published research literature at the time this research began. (1) A number of therapeutic models of forgiveness designed to help therapists aid their clients in their attempts to forgive have been proposed since this research began (e.g., Enright, 2001; Enright & Fitzgibbons, 2000; Ferch, 1998; Gordon, Baucom, & Snyder, 2000; Malcolm & Greenberg, 2000) that also would have been appropriate for consideration. These models were not available when the process of data collection was begun.

The use of the second criterion, in which only models which containing a process were selected, also resulted in the exclusion of several models which would otherwise have been included. Of particular note are the landmark studies conducted by McCullough and his colleagues (McCullough et al., 1997; McCullough et al., 1998). Models from these two studies were not included because the models contained a series of correlated social-psychological variables (such as empathy and relational closeness), which were used to predict situational forgiveness, rather than a series of interdependent psychological or theological steps (e.g. feeling angry, then having empathy for an offender, etc.), which an individual was theorized to progress through.

The resulting literature search found sixteen models of forgiveness meeting the three criteria. These models may be classified generally as therapeutic, popular, or religious in nature (defined by explicit involvement of God or religious faith in the process) and are summarized in Table 1. Across the sixteen models, there are seven common constructs as well as a number of unique constructs that have been proposed. Common constructs across the models include (a) feeling hurt, (b) deciding to forgive (Donnelly, 1982; Enright and the Human Development Study Group, 1991, 1996; Fitzgibbons, 1986), (c) understanding the offender/empathy (Brandsma, 1982; Enright and the Human Development Study Group, 1991, 1996; Fitzgibbons, 1986; Smedes, 1996; Worthington, 1998b), (d) receiving God's forgiveness (Pingleton, 1997; Stanley, 1987), (e) giving up one's anger (Augsberger, 1988; Brandsma, 1982; Fitzgibbons, 1986; Smedes, 1984, 1996; Stanley, 1987), (f) forgiving oneself (Cunningham, 1985; Donnelly, 1982; Pingleton, 1997), and (g) reconciling with the other person (Enright and the Human Development Study Group, 1991; Hargrave, Sells, 1997; Pettitt, 1987; Smedes, 1984; Stanley, 1987).

Rationale for the Study

McCullough and Worthington (1994) provided an insightful analysis of the forgiveness literature, which has since exploded in terms of the amount of available literature from the time of their review. At that time, they concluded that while some consensus existed regarding the components of forgiveness, no research as yet addressed the validity of forgiveness models, including the number of or interrelationship among the constructs thought to underlie models of forgiveness. The current state of forgiveness literature in terms of addressing these concerns is much the same. Research has yet to empirically address the validity of various models of forgiveness. Thus, the rationale for the current study was as follows.

The proposed sixteen models of forgiveness and reconciliation have two elements. First, each of the models has an identified set of constructs. In this study, items were created to represent each of the steps of forgiveness proposed by the various models of forgiveness and then submitted to common factor analysis to identify an underlying set of constructs. As noted elsewhere, "the purpose of factor analysis is to identify the fewest possible constructs needed to reproduce the original data" (Gorsuch, 1997, p. 533). In this study, common factor analysis is used to identify the underlying constructs used in the sixteen models of forgiveness.

Once the underlying constructs are identified, the second element to the sixteen forgiveness models is the interrelationship of the constructs. Path analysis will then be performed using the obtained factors to determine the model that best fits the relationship among the factors in forgiveness and reconciliation.

METHOD

Measures

Due to the fact that none of the forgiveness path models had measures or items created with which to measure their model (with the exception of Hargrave and Sells [1997] model of forgiveness), items to measure each of the models were generated by the first author. Other measures of forgiveness have been created, but these measures do not represent the constructs represented in the models of forgiveness currently under investigation (e.g., The Enright Forgiveness Inventory [EFI]; Subkoviak, Enright, Wu, & Gassin, 1995). The items were generated by taking the proposed steps of forgiveness from each original author's publication and then placing the suffix "ing" after the core phrase of the step. Across the sixteen forgiveness models, a total of 94 individual steps were posited, thus, a total of 94 items were generated (one for each and every step in the model). For example, empathy in Worthington's (1998b) empathy-humility-commitment model of forgiveness was measured with the item "empathizing with the person who hurt me."

After the items were generated, the items were administered in paper and pencil format to a group of 5 doctoral students in an APA-accredited Ph.D. program in clinical psychology for review. During this review procedure, the doctoral students were instructed to read the item, and then the section in each original author's article or book to determine how well each item represented the proposed stage. The doctoral students then provided written feedback to the first author on the questionnaire sheet. On the questionnaires provided for feedback, the group of doctoral students either retained an item, or suggested an alternative version of the item based on their understanding of the original author's theory. An item was retained if 4 of the 5 doctoral students in the group retained it, and altered if 2 of the 5 doctoral students suggested altering it. In these cases, an individual item was altered by the senior author using the suggested alternative from the group that appeared to best fit the original author's theory. After this review process, the items were revised accordingly to fit each model.

A precedent for this procedure is found in the personality literature. Block (1995), for example, reviewed the process underlying the development of Cattell's 16PF. In his review, Block notes that Cattell developed the 16PF by reviewing a list of 4,504 personality traits that were proposed by Allport and Odbert (1936). This list of 4,504 personality traits was itself compiled from a larger set of 17,953 single word descriptor terms from the 1925 edition of Webster's New International Dictionary that were considered to be usable to distinguish individual behavior. Cattell then took the list of 4,504 personality traits, and eliminated all but 171 of the terms to adhere to his definition of a personality trait. Following that process, he applied the terms to factor analysis, and eventually compiled a list of 35 bipolar rating scales which he concluded overaly 12 primary factors of personality.

Participants

The participants in the current study consisted of one hundred eighty students (137 females, 43 males) taken from Christian (N = 109), and secular (N = 71) universities in Southern California. The ages ranged from 18 to 55, (M = 23.20, SD = 7.56). The following religious denominations were represented: 113 Protestant, 27 Catholic, 9 Christian Scientist, 2 Mormon, 5 Buddhist, 2 Eastern Orthodox, 2 Jehovah's Witness, 3 Jewish, 9 None, 2 Agnostic, 1 Atheist, 1 Bahai, and four participants who did not state their religious denomination. The sample consisted of 117 White, 29 Asian, 16 Hispanic, 1 African-American, and 13 Multi-Racial participants.

Regarding sample size for exploratory factor analysis, Arrindell and van der Ende (1985) demonstrated empirically that when an adequate number of variables with substantive loadings is analyzed, a stable factor solution is possible when the sample size is approximately 20 times the number of factors obtained from the analysis. Theoretically, the number of factors we expected based upon our literature review ranged from 3 to 9, so we judged our sample of 180 to be adequate in satisfying this criteria.

Procedure

The students were invited to participate during a class session, and then completed the materials either individually or in groups ranging from 2 to 34. The students were given the following instructions, adapted from previous forgiveness studies (Enright, 1999; McCullough et al., 1997): "We ask you now to think of one living person whom you have experienced as treating you unfairly and hurting you at some point in the past. For a moment, visualize in your mind the events and the interactions you may have had with the person who offended you. Try to visualize the person and recall what happened. Below is a set of questions about the person who hurt you. It is possible that you have experienced this type of hurt from more than one person, but please answer the questions that follow as you focus on a single person who hurt you." The students then completed the 87-item forgiveness questionnaire that was described earlier. As indicated in the Appendix, the forgiveness items in this questionnaire packet were rated on a 6-point Likert type scale (0 = Not very true of me, 5 = Very true of me).

Offense-Related Information

To assess both the level of hurt and the types of hurt that the participants experienced, three questions were asked of the participants. The first question asked, "how deeply were you hurt when the incident occurred?" Responses were scored from 1 (No hurt) to 5 (A great deal of hurt). The mean response to this item was 4.24, SD = .82, indicating that most participants responded to a very painful hurt when responding to the questionnaire. The second item asked, "who hurt you?" The most frequent responses to this category were (a) romantic partner or exromantic partner (50), (b) friend (40), (c) father or mother (27), and (d) other relatives (husband, brother, sister, etc., 15). This indicates that the offending party used by the participants to respond to the questionnaires was generally close to them. Finally, a third question asked the participants to briefly describe what happened. Examples of participants' responses to this question include "I was raped four years ago, and my husband believed that I wanted it to happen." Another participant wrote "I was suicidal in high school, and my mom blamed me for feeling that way." A third participant wrote "My boyfriend used me for one last fling."

RESULTS

Constructs Identification

To examine the underlying constructs of the sixteen forgiveness models, a common factor analysis was conducted using the principal axis method. In accordance with our literature review, we examined factor solutions that ranged from 3 to 9 factors. Rotation was by promax, which allows correlated (i.e. oblique) factors, based on the recommendation provided by Gorsuch (1997). In this procedure, the value of k for promax is started at 2 and is increased in steps of 1 until the hyperplane count shows no improvement from an increase in the value of k. A final criterion was applied: a rotated factor was retained if it had 3 or more items with loadings greater than or equal to .40 which did not cross-load as high or higher on another factor (Gorsuch, 1983, 1997). The extraction result was five factors with 3 or more items that loaded .40 or greater and were consistent with the other criterion noted above; thus, a five factor solution was retained.

Factor 1 had 24 of the 87 items load on it, and five items had loadings on the factor which were .80 or greater. The highest loading items on this factor consisted of (a) reaffirming the relationship, .86 (taken from Augsberger, 1988; step 5), (b) reestablishing my relationship with them, .83 (taken from Smedes, 1984, step 4), and (c) reconstructing the relationship, .82 (taken from Augsberger, 1988; step 3). The majority of the items on this factor referred to reconciling a relationship with the offender, thus this factor was labeled Reconciliation.

Factor 2 had twenty-three items load on it. Nine items loaded .60 or higher. The highest loading items on this factor consisted of (a) deciding to forgive, .70 (taken from Donnelly, 1982, step 2), (b) becoming willing to explore forgiveness as an option, .70 (taken from Enright and the Human Development Study Group, 1996, step 10), (c) experiencing an emotional release, .69 (taken from Enright and the Human Development Study Group, 1996, step 20), and (d) experiencing emotional forgiveness, .69, (taken from Fitzgibbons, 1986, step 6). The items on this factor appeared to tap an aspect of emotional forgiveness separate from behavioral reconciliation, thus this factor was labeled Emotional Forgiveness.

Factor three had five items load on it. Three items loaded .87 or higher; these were (a) feeling loved by Christ and wanting Him to be the center of my life, .92 (taken from Linn and Linn, stage 5), (b) recognizing that we have been totally forgiven by God, .89 (taken from Stanley, 1987; step 1), and (c) accepting that God forgives me, .87 (taken from Pingleton, 1997). These items appeared to reflect an aspect of receiving God's forgiveness, therefore, this factor was labeled Receiving God's Forgiveness.

Factor four had six items load on it. The two highest loading items on this factor consisted of (a) examining my own psychological defenses, .58 (taken from Enright and the Human Development Study Group, 1996, step 1), and (b) appreciating the reason for the person's actions, .57 (taken from Brandsma, 1982; step 2). These items appeared to tap an aspect of empathy, thus this factor was labeled Empathy.

Factor five had 17 items load on it, and eight items loaded on this factor .60 or greater. The highest loading items on this factor consisted of (a) feeling angry, .73 (taken from Linn & Linn, 1978, step 2; and Menninger, 1996, step 4), (b) feeling depressed, .70 (taken from Linn & Linn, 1978, step 4), and (c) experiencing negative emotional consequences, .69 (taken from Enright et al. 1991, step 1). The majority of the items loading on this factor appeared to tap aspects of feeling hurt and anger, thus this factor was labeled Hurt and Anger.

Scales

Scales to measure each of the factors were created based on the substantive item factor loadings. To create the scales, the five items with the highest item factor loadings were used, and then checked using coefficient alpha for reliability. Items used to create the scales, scale means, and standard deviations are presented in the Appendix. The correlations of the scales with their respective factors, and scale coefficient alphas are presented in Table 2. As can be seen from Table 2, all of the scales had coefficient alphas in the range of .82 to .93, with the exception of the Empathy scale, which had an alpha reliability estimate of .67. As can also be seen from Table 2, intercorrelations among the scales ranged in size from .02 to .56. The largest negative correlation among the scales was -.22. A few scales had intercorrelations that ranged from .54 to .56. In these cases, it is notable that these moderate correlations occurred among scales whose steps in the path to reconciliation were considered to be most proximate to each other (e.g., the .56 correlation between reconciliation and emotional forgiveness).

Path Analysis

To explore models of the relationships among the factors, path analysis was conducted with the covariances among the scales. Path analysis was done using only the scales created from each of the five factors since there were sixteen theoretical models with no empirical research to suggest one model over the other. Since no clear model of forgiveness and reconciliation existed in the literature that would serve directly for the factors that were identified, six path models were tested. Each path model used Hurt and Anger as the starting point (exogenous variable) and Reconciliation as the end point in the model (endogenous variable). These dimensions were chosen as endpoints because most authors of the original 16 models of forgiveness and reconciliation that were used for construction considered some aspect of being hurt or angry as a starting point and either reconciliation or forgiveness as an end point. Reconciliation was chosen as an end point because Reconciliation is a behavioral act that seems to require more effort on the part of the offended than forgiveness (Worthington, 2001).

After choosing Hurt and Anger and Reconciliation as endpoints, the path models explored several ways in which Receiving God's Forgiveness, Empathy, and Emotional Forgiveness fit into the path model. Since there was nothing in the literature to suggest placing Receiving God's Forgiveness, Empathy, or Emotional Forgiveness in any particular order, several different path models were tested in which the order was rearranged. No explorations were based on examinations of the residuals or other empirical data from this study to reduce possible capitalization on chance. A cutoff of .90 was used for fit indices, based on the recommendation provided by Bentler and Bonett (1980). The first path model tested in Figure 1 was a simple "straight through" model. This is similar to the models of forgiveness and reconciliation most often suggested by the literature, that is, a model where each factor occurred consecutively. The results of Model 1 are in Table 3, and indicate that it was not satisfactory. Having little success with the original path model, the second path model was briefer. In this path model, Hurt and Anger was related directly to both Empathy and Emotional Forgiveness, Receiving God's Forgiveness was related directly to both Empathy and Emotional Forgiveness, and both Empathy and Emotional For-giveness related directly to Reconciliation. The second path model results are diagrammed in Figure 2, and presented in Table 3. The results demonstrate that this minimal approach also did not fit. In this path model, the adjusted fit indices, which adjust for the complexity of the model, decreased significantly. The next three path models, presented in Figures 3, 4, and 5, tested several possible variations. None of these were as good as Model 6 in Table 3, and presented in Figure 6. The final path model was the one of the six which overall best met the SEM criteria; it had a low chi square/df (4.05) and root mean squared residual (.05), and high GFI (.97; AGFI = .88) and improvement over a null model (Bentler-Bonnett Normed fit of 93%, adjusted = 87%).

There is one result in the table of chi-squares that is unusual. Model 6 is Model 4 with one less path. In this case it can be recommended to subtract the chi-square with the greater number of parameters (Model 6) from the chi-square for the other model (Model 4) and use that as the chi square for the one parameter that is used in Model 4 but not in Model 6. When that is done with these data, the chi-square is negative. While a negative value cannot occur in generalized least squares analysis where the chi square is based on what is predicted, it can occur in path analysis where the chi square is based on what is not predicted. The reason, in this case, is that Model 6 requires the correlation be zero between Receiving God's Forgiveness and Empathy after Emotional Forgiveness is partialled out. That is not so, and so the error of prediction increases if the needed path is missing.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Separate analyses for Christian versus non-Christian students were considered. However, close inspection of the question included in the demographics, "how important is your religious faith to you?", revealed that too few of the total sample (N = 31) considered their faith to be unimportant to them, making such an analysis unfeasible.

DISCUSSION

The central purpose of this study was to determine the underlying dimensions across 16 models of forgiveness and reconciliation. These results provide initial empirical support for a path model in which Hurt and Anger and Receiving God's Forgiveness simultaneously predicted Emotional Forgiveness, which predicted Empathy, which predicted Reconciliation.

As noted earlier, some suggested models explicitly incorporated dimensions that were obtained here. The hurt and anger dimension, for example, is present in several models (e.g., Brandsma, 1982; Enright and the Human Development Study Group, 1991; Fitzgibbons, 1986; Menninger, 1996; Smedes, 1984). An empathy dimension is also present in several models (e.g., Brandsma, 1982; Enright and the Human Development Study Group, 1996; Worthington, 1998b), as well as a receiving God's forgiveness dimension (Pingleton, 1997; Stanley, 1987).

Across these models, it has sometimes been unclear whether behavioral reconciliation is a dimension separate from emotional forgiveness, as some forgiveness theorists have tended to use the terms interchangeably. Worthington and Drinkard (2000) presented one of the few arguments in favor of a theoretical distinction between behavioral reconciliation and emotional forgiveness. In discussing these differences, they cited a definition of forgiveness advanced by McCullough, Worthington, and Rachal (1997), in which forgiveness is considered to have occurred when an individual (a) releases anger, (b) ceases to withdraw or avoid the offending party, and (c) experiences a desire for reunion if possible to do so. Using this definition, they considered, as have others (e.g., Baumeister, Exline, & Sommer, 1998; Worthington, 2001), that forgiveness is thus an intrapersonal phenomenon distinct from reconciliation. Conversely, reconciliation was suggested to involve "interpersonal actions surrounding requesting, granting, and accepting forgiveness" (Worthington & Drinkard, 2000, p. 94). These results provide initial empirical support for this distinction between these two constructs.

It is interesting to note that both Emotional Forgiveness and Reconciliation were present in the path model with the best fit. Although forgiveness theorists have at times used the two terms inter-changeably, only recently have theorists begun to consider the underlying dynamics of each process separately. Worthington (2001), for example, proposes that each has a separate underlying process. The process of forgiveness, according to Worthington, consists of five steps (labeled the REACH model): first, recalling the hurt; second, empathizing; third, the altruistic gift of forgiveness; fourth, committing publicly to forgive; and fifth, holding onto forgiveness. Conversely, the process of reconciliation, according to Worthington, consists of four steps: first, deciding whether, how, and when to reconcile; second, discussing the transgressions; third, detoxifying the relationship of past hurts; and fourth, devoting each other to building up a mutually valuing relationship. In these two models, Worthington considers forgiveness to be a motivating factor that sometimes, though not always, leads one onto the process of reconciliation.

Although sometimes considered to consist of separate, individual dynamics, the presence of both Emotional Forgiveness and Reconciliation in the same model suggests that it is possible that the underlying dynamics for each construct may be more similar than different. Or, perhaps Emotional Forgiveness provided a bridge to Reconciliation in the model because many of the students in the sample had been able to reconcile with the people who had hurt them. When grouped together, 92 of the respondents (slightly over half) said that they had experienced a hurt from a parent, sibling, or other relative. Perhaps experiencing a hurt in the context of those relationships gave them greater motivation to reconcile. Or, perhaps the longstanding nature of the relationship (family) caused them to continue to maintain contact with the person, resulting in a reconciliation of some sort.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

Another finding that ran contrary to expectation was that Emotional Forgiveness was predicted by Hurt and Anger but Empathy was not. Typically, forgiveness theorists have suggested that the experience of empathy is a necessary pre-cursor to experiencing emotional forgiveness (e.g., Brandsma, 1982; Hargrave & Sells, 1997; Worthington, 1998, 2001). Although these data lacked temporal data to either support or refute this suggestion, our results do not provide support for this hypothesis (see Figure 1, Figure 3). However, it should also be noted that at least one other model (model 2) was quite close to the final accepted model in terms of goodness of fit. In this model, both Empathy and Emotional Forgiveness were simultaneously predicted by Hurt and Anger, and simultaneously predicted Reconciliation, although they were not correlated in this model. It may be that, if the underlying processes to Emotional Forgiveness and to Reconciliation are truly unique and separate, then the final model reflects a Reconciliation model. As stated earlier, the final model was limited due to the nature of the original underlying 16 models of forgiveness and reconciliation from which it was built.

A somewhat surprising finding concerned the obtained Receiving God's Forgiveness factor. Since the sample consisted of primarily Christian college students, it is impossible to know whether this factor would have emerged in factor analyses of the same data collected on a non-religious population. It would be a mistake to consider Receiving God's Forgiveness as simply a step in the process of forgiveness or a religious ritual. It is possible that the model in which Receiving God's Forgiveness was present but uncorrelated with Hurt and Anger reflected an ongoing awareness of being forgiven by God. This experience of being forgiven by God is quite separate from one in which a person turns to God for assistance because of difficulty forgiving someone else. Rather, it the experience of Christian forgiveness alluded to earlier that reflects a progression of healing where undeserving people are confronted with the grace and mercy of God, and then forgive others because of the great love that is inside them (Meek & McMinn, 1997).

Limitations and Directions for Future Research

The major limitation of the present study is the lack of either longitudinal or cross-sectional data with which to evaluate the nature of the models over time. In the absence of such data, it is impossible to determine how the process of forgiveness and reconciliation occurs over time. It is hoped however, that the present research provides a stimulus to begin such endeavors. This research was also limited by the nature of the models that were incorporated, and a number of additional models have been proposed even since this work began. These results must be viewed tentatively, until such time as they are replicated using confirmatory factor analysis or structural equations modeling to confirm the factor structure using more diverse samples (in more diverse age ranges, with non-religious people, and with clinical populations), and incorporating models of forgiveness that have been advanced since the time of data collection for this project.

Several cautions regarding the present findings should be noted. One such caution concerns over-interpreting the results of the path analyses that were conducted. A temptation regarding these results, which incorporated some suggested therapeutic models, is to take them and apply them to the practice of therapy. In this regard, it should be noted that these results are descriptive, attempting to describe how people forgive, rather than prescriptive, attempting to say how they should forgive. The study was also conducted on a sample of primarily Christian female college students. Since this study used a sample of undergraduate students rather than clients in therapy, it remains to be seen whether these results will generalize to the therapeutic process of forgiveness, particularly in therapeutic situations with religious clients. The five-factor scale that was created from the factor analysis is encouraged for use as a research stimulus for endeavors such as these. As mentioned previously, the range of reliability estimates of the five subscales ranged from .84 to .93, with the exception of the Empathy subscale, which had a coefficient alpha of .67.

A second caution concerns the use of the Empathy scale in further research. This factor had the fewest loading items, and had the lowest obtained coefficient alpha. Although the highest reliability estimate for the scale was obtained using all five items rather than deleting some of the items, three of the items appear to better reflect empathy than the other two items. Both "becoming aware of my own part in setting up the unpleasant situation (if applicable)," and "examining my own psychological defenses," appear to better reflect a cognitive aspect of empathy, such as introspection, than an emotional aspect. If these items measure some cognitive aspect of empathy, the empathy scale would not be inconsistent with other theoretical formulations regarding the nature of empathy. For example, Carl Rogers, in his formulation of empathy, concluded that empathy involved understanding not only the feelings, but also the thoughts, perspective, and viewpoints of another person (Rogers, 1961).

Conversely, the two items that appear to tap into some aspect of introspection may be measuring something other than empathy, regardless of their factor loadings (which were greater than .40, and did not cross-load higher on another factor). In any event, given the noticeably lower coefficient alpha for this scale, it should be interpreted and used cautiously.

Are the five factors obtained here replicable? Is turning to God for forgiveness a more important factor for promoting forgiveness and reconciliation with religious people? To what extent will the results of these path analyses carry over to clinical (non-student) populations and corresponding interventions? These issues promise to be important ones for psychologists and forgiveness scholars alike.

APPENDIX

Final Items Used For Each Of The Scales

Now please answer the following questions about your current attitude toward the person.

We do not want your rating of past attitudes, we want your attitudes right now.

0 1 2 3 4 5

Not Very True of Me Very True of Me

Reconciliation, A = .92, M = 14.08, SD = 7.81

Reconstructing the relationship.

Looking for the good in the other person.

Reestablishing my relationship with them.

Feeling compassion for the person who hurt me.

Reaffirming the relationship.

Emotional Forgiveness, A = .84, M = 17.37, SD = 5.78

Experiencing emotional forgiveness.

Experiencing an emotional release.

Feeling an internal, emotional release.

Deciding to forgive.

Becoming willing to explore forgiveness as an option.

Receiving God's Forgiveness, A = .93, M = 16.51, SD = 7.99

Accepting that God forgives me.

Recognizing that we have been totally forgiven by God.

Becoming aware of our commonality under God.

Feeling loved by Christ and wanting Him to be the center of my life.

Viewing them as tools in our understanding of the grace of God.

Empathy, A = .67, M = 12.28, SD = 5.80

Understanding that he or she was doing their best at the time.

Appreciating the reason for the person's actions.

Becoming aware of my own part in setting up the unpleasant situation (if applicable).

Feeling as if I had come from this person's background, I might have done the same thing.

Examining my own psychological defenses.

Appendix, continued

Hurt and Anger, A = .84, M = 12.48, SD = 6.56

Feeling hurt by them.

Experiencing negative emotional consequences.

Feeling angry.

Feeling depressed.

Becoming aware that I replay the event in my mind.

Note. A = alpha, M = mean, SD = standard deviation

Table 1

Fitzgibbons (1986)
  1). Analyze the origins of the hurt
  2). Reenact hurtful situations
  3). Decide to forgive
  4). Retrace own development and identify hurts in developmental stages
  5). Understand offender's motives
  6). Experience emotional forgiveness

Pettitt (1987)
  1). Set the stage (Commit to no longer suffer from past memories)
  2). 5 Necessary Mindsets
      A. Seek to know the unmet needs that underlie people's behavior
      B. Daily relaxation training to evoke calmness
      C. Become open to accept responsibility for one's own part in the
         hurtful situation
      D. Find a course of action fair to all
      E. Remember the good in the other person
  3). Change one's mind about what one wanted to happen
  4). Accept the past, and cancel expectations
  5). Restore the relationship

Enright and The Human Development Study Group (1991)
  1). Experience negative emotional consequences
  2). Experience a need for resolution
  3). Decide among strategies (justice vs. mercy)
  4). Find a forgiveness motive
  5). Decide to forgive
  6). Execute internal forgiveness strategies
  7). Experience a need for action
  8). Execute behavioral reconciliation
  9). Feel internal, emotional release

Enright and the Human Development Study Group (1996)
  1). Examination of psychological defenses
  2). Confrontation of anger
  3). Admittance of shame
  4). Cathexis
  5). Cognitive rehearsal of the offense
  6). Possible comparison of self with the offender
  7). Realization that self may be permanently changed by the hurt
  8). A possibly altered "just world" view
  9). Insight that old resolution strategies are not working
 10). Become willing to explore forgiveness as an option
 11). Commitment to forgive
 12). Viewing the offender in context
 13). Empathy for the offender
 14). Awareness of compassion for the offender
 15). Acceptance of the pain of the offense
 16). Finding meaning in the suffering and forgiveness process
 17). Realization of the need to be forgiven in the past
 18). Realization that self is not alone
 19). Realization of possible new purpose in life
 20). Internal emotional release

Hargrave & Sells (1997)
  1). Insight
  2). Understanding
  3). Giving the opportunity for compensation
  4). Overt act of forgiving

Worthington (1998b)
  1). Empathy
  2). Humility
  3). Commitment

Smedes (1984)
  1). Feeling hurt by the offender
  2). Hating the offender
  3). Remembering the pain of the offense, but wishing them well
  4). Reestablish a relationship with the offender

Smedes (1996)
  1). Rediscover the humanity of the offender
  2). Surrender one's right to get even
  3). Revise one's feelings toward the offender

Donnelly (1982)
  1). Acknowledge the hurt
  2). Decide to forgive
  3). Remember that forgiveness isn't easy
  4). Forgive oneself
  5). Consider the consequences of not forgiving

Brandsma (1982)
  1). Re-experience the hurt
  2). Appreciate the reason for the offender's actions
  3). Evaluate expectation that they should have acted differently
  4). Give up one's anger

Augsberger (1988)
  1). Restore the attitude of love
  2). Release the painful past
  3). Reconstruct the relationship
  4). Reopen the future
  5). Reaffirm the relationship

Menninger (1996)
  1). Claiming the hurt
  2). Guilt
  3). Victim
  4). Anger
  5). Wholeness

Linn and Linn (1978)
  1). Denial
  2). Anger
  3). Bargaining
  4). Depression
  5). Acceptance

Cunningham (1985)
  1). Judgement vs. denial stage
  2). Humility vs. humiliation stage
  3). Mutuality and negotiation stage
  4). Actual living out of forgiveness

Stanley (1987)
  1). Recognize that we have been totally forgiven by God
  2). Release the person from the debt that is owed
  3). Accept the offender as they are
  4). View the offender as a tool in our understanding of the grace of
      God
  5). Make reconciliation with the offender

Pingleton (1997)
  1). Receive forgiveness from God
  2). Forgive oneself
  3). Forgive the other person

Table 2 Correlations between factors, scales, and alpha coefficient
reliabilities for scales.

Factors                                     Scales

                                1.    2.    3.    4.    5.    A    Items

1. Reconciliation                .97   .56   .24   .49  -.03  .92  5
2. Emotional Forgiveness         .57   .90   .54   .45  -.22  .84  5
3. Receiving God's Forgiveness   .32   .54   .97   .40  -.12  .93  5
4. Empathy                       .44   .55   .49   .83   .06  .67  5
5. Hurt and Anger                .05   .02  -.12   .15   .93  .82  5

Note. Numbers in the diagonal = correlation between the factor and the
corresponding scale; A = alpha reliability estimate of the scale, items
= number of items in the scale. A correlation of .15 or greater is
significant in a two-tailed test (p < .05).

Table 3 Goodness of Fit Indices for Path Models

Model      [chi square]  [chi square]/df  RMSR   GFI(AGFI.)  BBFit(Adj)
(Desired)  (Low)         (Low)            (Low)  (High)      (High)

1          80.01         13.33            .21    .86(.88)    .68(.45)
2           8.12          8.12            .09    .98(.73)    .97(.70)
3          97.3          16.25            .20    .85(.63)    .60(.36)
4          26.1           8.70            .06    .95(.74)    .89(.67)
5          31.3          10.40            .07    .94(.68)    .87(.60)
6          16.2           4.05            .05    .97(.88)    .93(.87)

Note. N = 180. GFI = Goodness of Fit Index, BBFit = Bentler-Bonett Fit
Index.

(1) Astute readers of the forgiveness literature will not how problematic this last criterion is for this study, which had the goal of reducing a large number of forgiveness models into one basic model. The authors recognize, regrettably, that several models have been published with were not available at the time data collection on this project had begun

REFERENCES

Allport, G. W., & Odbert, H. S. (1936). Trait names: A psycho-lexical study. Psychological Monographs, 47(1, Whole No. 211).

American Psychological Association Ethical Principles of Psychologists and Code of Conduct. (1992). American Psychologist, Washington, D.C.

Arrindell, W. A., & van der Ende, J. (1985). An empirical test of the utility of the observation-to-variables ratio in factor and components analysis. Applied Psychological Measurement, 9, 165-178.

Augsberger, D. (1988). The freedom of forgiveness. Moody Press: Chicago, IL.

Baumeister, R. F., Exline, J. J., & Sommer, K. L. (1998). The victim role, grudge theory, and two dimensions of forgiveness. In E. L. Worthington, Jr. (Ed.), Dimensions of forgiveness: Psychological research and theological speculations (pp. 79-104). Philadelphia: The Templeton Foundation Press.

Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psycho-logical Bulletin, 88, 588-606.

Block, J. (1995). A contrarian view of the five-factor approach to personality. Psychological Bulletin, 117, 187-215.

Brandsma, J. M. (1982). Forgiveness: A dynamic, theological and therapeutic analysis. Pastoral Pychology, 31, 40-51.

Coyle, C. T., & Enright, R. D. (1997). Forgiveness intervention with postabortion men. Journal of Consulting and Clinical Psychology, 65, 1042-1046.

Cunningham, B. B. (1985). The will to forgive: A pastoral theological view of forgiving. The Journal of Pastoral Care, 39, 141-149.

Donnelly, D. (1982). Putting forgiveness into practice. Allen, TX: Argus Communications.

Enright, R. D. (1999). "Interpersonal forgiving in close relationships": Correction to McCullough et al. (1997). Journal of Personality and Social Psychology, 77, 218.

Enright, R. D. (2001). Forgiveness is a choice: a step-by-step process for resolving anger and restoring hope. American Psychological Association: Washington, D.C.

Enright, R. D., & Fitzgibbons, R. P. (2000). Helping clients forgive: an empirical guide for resolving anger and restoring hope. American Psychological Association: Washington, D.C.

Enright, R. D., & The Human Development Study Group (1991). The moral development of forgiveness. In W. M. Kurtines & J. L. Gertwitz (Eds.), The Handbook of Moral Behavior and Development (vol. 1, pp. 123-152). Hillsdale, NJ: Earlbaum.

Enright, R. D., & The Human Development Study Group (1996). Counseling within the forgiveness triad: On forgiving, receiving forgiveness, and self-forgiveness. Counseling and Values, 40, 107-126.

Ferch, S. R. (1998). Intentional forgiving as a counseling intervention. Journal of Counseling and Development, 76, 261-270.

Fitzgibbons, R. P. (1986). The cognitive and emotive uses of forgiveness in the treatment of anger. Psychotherapy, 23, 629-633.

Gordon, K. C., Baucom, D. H., & Snyder, D. K. (2000). The use of forgiveness in marital therapy. In M.E. McCullough, K. I. Pargament, & C. E. Thoresen (Eds.), Forgiveness: Theory, research, and practice (pp. 203-227). New York: Guilford Press.

Gorsuch, R. L. (1997). Exploratory factor analysis: Its role in item analysis. Journal of Personality Assessment, 68, 532-560.

Gorsuch, R. L. (1983). Factor analysis (2nd Ed). Hillsdale, NJ: Erlbaum.

Hargrave, T., & Sells, J. (1997). The development of a family forgiveness scale. Journal of Marital and Family Therapy, 23, 41-62.

Hebl, J., & Enright, R. D. (1993). Forgiveness as a psychotherapeutic goal with elderly females. Psychotherapy, 4, 658-667.

Linn, M., & Linn, D. (1978). Healing life's hurts through the five stages of forgiveness. New York: Paulist Press.

Malcolm, W. M., & Greenberg, L. S. (2000). In M. E. McCullough, K. I., Pargament, & C. E. Thoresen (Eds.), Forgiveness: Theory, research, and practice (pp. 179-202). New York: Guilford Press.

McCullough, M. E., Pargament, K. I., & Thoresen, C. E. (2000). Forgiveness: Theory, research, and practice. New York: Guilford Press.

McCullough, M. E., Rachal, K. C., Sandage, S. J., Worthington, E. L., Brown, S. W., & Hight, T. L. (1998). Interpersonal forgiving in close relationships II: Theoretical elaboration and measurement. Journal of Personality and Social Psychology, 75, 1586-1603.

McCullough, M. E., & Worthington, E. L., Jr. (1994). Encouraging clients to forgive people who have hurt them: Review, critique, and research prospectus. Journal of Psychology and Theology, 22, 3-20.

McCullough, M. E., Worthington, E. L., Jr. & Rachal, K. C. (1997). Interpersonal forgiving in close relationships. Journal of Personality and Social Psychology, 73, 321-336.

Meek, K. R., & McMinn, M. R. (1997). Forgiveness: More than a therapeutic technique. Journal of Psychology and Christianity, 16, 51-61.

Menninger, W. A. (1996). The process of forgiveness. New York: Continuum.

Pettitt, G. A. (1987). Forgiveness: A teachable skill for creating and maintaining health. New Zealand Medical Journal, 25, 180-182.

Pingleton, J. P. (1997). Why we don't forgive: A biblical and object relations theoretical model for understanding failure in the forgiveness process. Journal of Psychology and Theology, 25, 403-413.

Rogers, C. R. (1961). On becoming a person. Houghton Mifflin: New York.

Sells, J., & Hargrave, T. (1998). Forgiveness: A review of the theoretical and empirical literature. Journal of Family Therapy, 20, 21-36.

Smedes, L. (1984). Forgive and forget: Healing the hurts we don't deserve. San Francisco: Harper and Row.

Smedes, L. (1996). The art of forgiving. Nashville, TN: Moorings.

Stanley, C. (1987). Forgiveness. Nashville, TN: Olver-Nelson.

Subkoviak, M. J., Enright, R. D., Wu, C., & Gassin, E. A. (1995). Measuring interpersonal forgiveness in late adolescence and middle adulthood. Journal of Adolescence, 18, 641-655.

Worthington, E. L., Jr. (1998a). Dimensions of forgiveness: Psychological research and theological perspectives. Templeton Foundation Press: Philadelphia.

Worthington, E. L., Jr. (1998b). An empathy-humility-commitment model of forgiveness applied within family dyads. Journalof Family Therapy, 20, 59-76.

Worthington, E. L., Jr. (2001). Five steps to forgiveness: The art and science of forgiving. Crown: New York.

Worthington, E. L., & Drinkard, D. T. (2000). Promoting reconciliation through psychoeducational and therapeutic interventions Journal of Martal and Family Therapry, 26, 93-101.

DONALD F. WALKER and RICHARD L. GORSUCH

Graduate School of Psychology

Fuller Theological Seminary 

Results of this research are based largely on a Master's Thesis completed by Donald F. Walker under the direction of Richard L. Gorsuch. A portion of this research was presented at the 108th Annual Meeting of the American Psychological Association, Washington, D.C. Correspondence regarding this article should be addressed to Donald F. Walker, 570 N. Los Robles #11, Pasadena, CA, 91101. Email: [e-mail]

AUTHORS

WALKER, DONALD F. Address: 255 N. Madison Ave. #1, Pasadena, CA 91101. Title: Doctoral candidate, Graduate School of Psychology, Fuller Seminary. Degrees: BA, MA., Graduate School of Psychology, Fuller Seminary. Specializations: Use of religion and spirituality in psychotherapy, child and adolescent psychotherapy.

GORSUCH, RICHARD L. Address: 180 N. Oakland Ave, Pasadena, CA 91101. Degrees: BA, PhD., University of Illinois. Title: Professor of Psychology. Specializations: Psychology of religion, multivariate statistics.