“Church Sex Scandal,” “Bishop Reveals Church Secrets,” “Woman Accuses Pastor of . . .” Headlines such as these suggest a fairly constant stream of stories depicting the failures of religious and spiritual leaders. Unfortunately, stories about the vast majority of leaders who live lives of integrity and minister to the needs of billions of adherents worldwide are largely unreported. History is replete with stories of leaders who violate the trust of their followers. Leaders of religious and spiritual groups violate the faithful. Can fallen leaders change? Will congregants forgive or restore derailed pastors or other spiritual leaders to their calling? In this article, we report the psychometric properties of the Clergy Situational Restoration Inventory (CSRI) and the two Leadership Restoration Scales (LRS) developed to help assess attitudes of congregants toward clergy offenders.

Given the press coverage of church scandals in recent years, it seems reasonable to wonder about the actual scope of the problem. Thomas et al. (2008) summarized findings reporting that 12 % to 15 % of clergy committed sexual boundary violations. Sutton et al. (2007) found that between 29 % and 37 % of their participants (American Pentecostal) knew a pastor with a problem that affected ministry and that between 20 % and 27 % reported knowing a pastor who had a sexual problem while in the ministry. When clergy and other Christian leaders fail, their acts of transgression may have profound effects on a faith community, given their position of trust (e.g., Kline 2007; Kline et al. 2008). Congregants and those in an accountability relationship with the errant leader are faced with at least two significant tasks. As fellow Christians they are challenged to forgive. A second and more complex task is the challenge of restoration for those leaders who desire not only forgiveness but an opportunity to minister again (Sutton and Thomas 2005a).

Spirituality and religiosity have been defined in various ways since the early psychological theorizing by James (1902). In the past decade, psychological scientists have referred to religion as more formal than the more personal and subjective experience subsumed in the concept of spirituality (Hill and Pargament 2003). In samples of experts in death studies drawn from journal editorial board members and attendees at two conferences on spiritual studies, Mahoney and Graci (1999) found the following concepts were related to the construct of spirituality: charity, community (connection, relationship), compassion, forgiveness, hope, meaning, and morality. In this study, we considered a few aspects of the complex construct of spirituality that we thought might be related to restoration and thus offer evidence that we had adequately measured relevant dimensions of the construct. Gordon et al. (2008) reported a relationship between intrinsic/extrinsic religious orientation and forgiveness.

Forgiveness is a concept common in many religions (Rye et al. 2000). Forgiveness research continues to expand. Scientists have examined self-forgiveness (Fisher and Exline 2006; Hall and Fincham 2008), forgiveness between romantic partners (DiBlasio and Benda 2008; Gordon et al. 2005), and forgiveness between social groups. Of particular relevance to this study are findings indicating a positive correlation between religious or spiritual factors and forgiveness (McCullough and Worthington 1999; Tsang et al. 2005).

Restoration is usually a communal act whereby a group is tasked with the responsibility to review the offenses of a fallen leader and, following an assessment, agree with psychologically healthy clergy on a restoration plan (Sutton and Thomas 2005b; Thomas and Sutton 2008). However, any group could hardly be successful without the willingness of various responsible and offended parties to restore the leader. Although the research is minimal, some have found a link between spirituality, forgiveness, and restoration among American Pentecostal congregants (e.g., Sutton et al. 2007; Thomas et al. 2008). Also, Luzombe and Dean (2009) found that Catholic priests were more forgiving toward priests who committed a transgression than were Catholic college students. Those who scored higher on a measure of spirituality were more forgiving. Considering the various spirituality concepts that might be related to clergy restoration, we selected not only measures of forgiveness but also of compassion, as represented by the Brief Compassion Scale (Hwang et al. 2008), and intrinsic-extrinsic religiosity represented by the revised formulation of the Intrinsic-Extrinsic Religiosity Scale (Gorsuch and McPherson 1989).

Our purpose was to assess the psychometric properties of the CSRI and the LRS (Sutton and Thomas 2004). We hypothesized that two factors would capture perceptions of offense severity on the CSRI. Based on previous research with small samples, we expected the reliability values would be adequate for both measures. We hypothesized that we would find support for validity of the restoration construct operationally defined by the scenarios in the CSRI by finding significant positive correlations with other measures of leadership restoration (LRS items) focused on an identified leader who committed an offense. We also hypothesized that measures of forgiveness and compassion would significantly correlate with measures of restoration. Finally, we measured spirituality in order to better describe the participants and because we thought spirituality would be correlated with a willingness to restore errant clergy.

Method

Participants

A total of 210 participants completed the CSRI and other measures (153 women and 57 men). The majority of the sample was European American (85 %). Participants attended a Christian evangelical university located in the American Midwest. Religiously, they identified mostly with the Assemblies of God (66 %) and nondenominational Protestant groups (21 %).

Some questions were only relevant to participants who knew a member of the clergy who had committed an offense. This subsample included 120 women and 49 men. The majority of the subsample was European American (89 %). The ages of the participants ranged from 18 to 58, with a median age of 20. These participants were similar in religious identity to the larger sample. They identified most with the Assemblies of God (67 %), nondenominational (20 %), and Baptist (7 %).

Following review board approval, the participants were recruited through their campus email. Some were offered course credit for their participation.

Design and procedures

This study was designed to expand previous knowledge of the validity and reliability of the CSRI and LRS. Participants were asked if they knew of any church leader who had committed a serious offense. The first measure (CSRI) asked participants to rate 10 clergy offense vignettes, such as substance abuse and sexual misconduct, in terms of their willingness to restore a clergy person following the offense. Participants then completed the two Leadership Restoration Scales. We included the Trait Forgiveness Scale (TFS; Berry et al. 2005), 12 items from the Transgression-Related Interpersonal Motivations Inventory (TRIM; McCullough et al. 1998), the Santa Clara Brief Compassion Scale (Hwang et al. 2008), and the Intrinsic/Extrinsic Revised Religiosity Scale (IER; Gorsuch and McPherson 1989). Participants were then thanked for their participation and debriefed.

Measures

CSRI

The CSRI assesses attitudes toward restoration based on participant responses to 10 transgression scenarios in which a pastor violated a common social expectation (Sutton et al. 2007; Sutton and Thomas 2004). The scale uses Likert-type ratings that range from one (no restoration to ministry) to seven (full restoration to the position previously held). The transgression scenarios include problems of substance abuse, infidelity, and embezzlement. Because of the range of common yet hypothetical scenarios, we expected the CSRI to assess a disposition to restore. A full copy of the CSRI can be found in Appendix A.

LRS

The LRS consists of two subscales composed of 7-point Likert-type items. One set of three items (LRSF) focuses on attitudes toward both forgiving and restoring a specific member of the clergy known to the participant. A second set of six items (LRSR) asks participants to consider only restoration (without specified forgiveness) of the offending clergy member. See Appendix B for a copy of the LRS items.

Trait Forgiveness Scale

The 10-item TFS was developed by Berry et al. (2005) to measure dispositional forgiveness. Cronbach’s alpha for the four studies conducted by Berry et al. (2005) was between .74 and .80. In the present study, we found Cronbach’s alpha to be .82.

TRIM

The TRIM-12 was developed by McCullough et al. (1998) to measure the motivations underlying forgiveness. They developed a two-component system consisting of Revenge (TRIM-R) and Avoidance (TRIM-A). Internal consistency values (coefficient alpha) were TRIM-R = .90 and TRIM-A = .94. In this study, the Revenge subscale coefficient alpha = .82 and the Avoidance subscale coefficient alpha = .90.

Compassion

The 5-item Santa Clara Brief Compassion Scale was developed by Hwang et al. (2008), who reported alpha = .90. In this study, the alpha was .91.

Spirituality

The Intrinsic/Extrinsic Revised Religiosity Scale (IER; Gorsuch and McPherson 1989) consists of 14 items divided into three subscales to assess the Intrinsic (IER-IN; alpha = .83), Extrinsic Social (IER-ES; alpha = .58), and Extrinsic Personal (IER-EP; alpha = .57) dimensions of religiosity.

Results

Preliminary analysis

We examined the descriptive statistics and found that skew and kurtosis were within acceptable ranges (George and Mallery 2009) for all measures. See Table 1.

Table 1 Descriptive statistics for the measures
Full size table

Principal components analysis of the CSRI

We conducted a principal components analysis to identify the structure of the 10-item CSRI among the data obtained from the full sample of 210 participants. Using a varimax rotation, we found that two factors with eigenvalues above 1.0 accounted for 65 % of the variance. An analysis of the items loading on each factor suggested that participants discriminated on the basis of scenario severity. Four scenarios suggested more severe offenses than the other six. The more severe component included a second affair (item 3), an affair lasting more than one year (item 6), a worsening anger problem lasting 5 years (item 8), and physical abuse evident by bruises and described as a result of a severe spanking (item 10). For simplicity, we refer to the two components as Level 1 and Level 2 offenses, where Level 2 consists of the four items that appeared more severe than the other six items of Level 1. See Appendix A for a list of the 10 offenses in the CSRI.

Next, we conducted a principal components analysis on the data obtained from those participants (n = 169) who reported an experience with a clergy member who had committed a serious offense. Using a varimax rotation, we found the same two factors with eigenvalues above 1.0 accounted for 65 % of the variance. As before, an analysis of the items loading on each factor suggested that participants discriminated on the basis of scenario severity with the same scenarios loading on the two levels as was true for the full sample.

We examined the reliability values of the two subscales in both samples. In the large sample (N = 210), the coefficient alpha was .86 for the Level 1 offenses and .79 for Level 2 offenses. In the subsample of participants who reported knowing a clergy offender (n = 169), the coefficient alpha values were .87 for Level 1 offenses and .76 for Level 2 offenses.

Finally, we examined the relationship between the two components and found a positive correlation (r = .64, n = 169, p < .001, one-tailed). When we averaged the means for the two components and conducted a paired-samples t-test, we found participants were significantly less willing to restore pastors involved in Level 2 situations (M = 2.81, SD = 1.07) than those involved in level 1 situations (M = 4.32, SD = 1.26), t (168) = 19.34, p < .01.

Concurrent validity

We examined intercorrelations among the measures to evaluate the concurrent validity of the CSRI. Because the LRS items and the TRIM items refer to a specific person, we limited these analyses to the subsample of participants (n = 169) who knew a clergy person who had committed a serious offense. We expected the CSRI to positively correlate with measures of restoring a known leader (LRSR ratings) and with the measure of trait forgiveness (TFS) as well as compassion (SCBCS). In contrast, we expected negative correlations between the CSRI and measures of avoidance and revenge. We expected similar findings for the LRS ratings.

There were significant positive correlations between CSRI situations and a willingness to restore an actual leader (LRSR), but not when forgiveness of that leader was considered (LSRF). High trait forgiveness was positively correlated with a willingness to restore clergy who committed Level 1 CSRI offenses but not Level 2 offenses. The TRIM measure of avoidance was significantly negatively correlated with both Level 1 and Level 2 offenses in the CSRI scenarios as well as the restoration ratings of actual pastors represented in the two LRS measures. Although the TRIM revenge measure was positively correlated with TRIM avoidance, it was not significantly related to any of the restoration measures. Compassion was only related (negative correlation) to the restoration of leaders who committed an actual offense. Extrinsic, but not intrinsic, religiosity was significantly related to a willingness to restore clergy. See Table 2 for the intercorrelations.

Table 2 Intercorrelations and significance values: restoration scales and other measures
Full size table

Discussion

Our purpose was to provide additional data regarding the psychometric properties of two measures of clergy restoration that were used in previous research. We found adequate evidence that both the situational measure (CSRI) and the two sets of items assessing restoration attitudes toward an actual offender (LRSF and LRSR) met usual criteria for normality and an internal consistency measure of reliability. In addition, we found evidence for a two-component structure for the CSRI based on two levels of offense severity within the 10 scenarios. The two components are moderately correlated with each other and allow for discrimination between attitudes that favor restoring clergy who have committed various offenses.

We found some support for concurrent validity in that people who considered restoring a known clergy offender were also likely to produce higher ratings on restoring clergy based on offenses presented in hypothetical scenarios, which supports the value of the CSRI to assess general attitudes toward clergy restoration. In addition, we found evidence that respondents with high trait forgiveness scores were also more willing to restore clergy who committed less serious offenses, but trait forgiveness was not related to restoring clergy who committed the more serious Level 2 offenses. This discrimination between Level 1 and Level 2 offenses is consistent with the significant differences in average scores for Level 1 and Level 2 offenses and suggests the importance of clarifying the seriousness of the offenses when asking participants their attitudes toward restoring clergy offenders. The discrimination also suggests that the participants were attentive to key text variations in the scenarios.

As noted, trait forgiveness was significantly related to restoring clergy who committed less serious offenses as defined by the CSRI Level 1 scenarios. Although there was a trend favoring a positive correlation between trait forgiveness and Level 2 offenses, the correlation was not statistically significant. Participants were clearly finding it difficult to respond favorably toward clergy who committed the more serious offenses. Additional information about forgiveness motivations can be found in the TRIM data. Findings from the significant TRIM avoidance correlations provide additional evidence of concurrent validity. Participants who reported high avoidance scores were likely to report low restoration scores. This finding suggests the value of the avoidance dimension of unforgiveness-forgiveness as an important consideration in further research on the restoration of errant clergy. The lack of a revenge motivation related to restoration issues may suggest a general willingness to consider restoration issues without seeking revenge. We speculate that revenge may be such a personal response to an offense that revenge motivations are much more related to interpersonal offenses than to community-based offenses, whereas both trait forgiveness and avoidance motivations are important to forgiving and restoring errant leaders in the community context.

Compassion was negatively correlated with attitudes toward restoring an actual offender. A likely explanation for this relationship is the finding that in situations where there are identified perpetrators and victims, most people identify with and feel sympathy for victims rather than perpetrators (e.g., Pinker 2011). However, we did not specifically test that hypothesis and offer it as an opportunity for further research into factors affecting the restoration of errant clergy and other leaders in faith communities.

The inclusion of a measure of spirituality provided value in linking extrinsic religiosity to a general unwillingness to restore fallen leaders, whereas intrinsic spirituality was not significantly correlated with any of the restoration measures. One possibility for the extrinsic religiosity relationship to restoration might be the focus of both restoration and extrinsic religiosity on a community rather than the internal spirituality assessed by the intrinsic factor. As is, the findings provide statistically significant albeit weak support for the role of a particular aspect of spirituality as related to restoration.

Our findings add to the few published studies that include measures of clergy restoration. The scales in this study yielded adequate results for research purposes and lend support to the notions that a construct of restoration is distinct yet related to trait and situational forgiveness and that extrinsic religiosity may be a factor in attitudes toward restoration. We also find evidence that participants are careful to discriminate when thinking about restoration, as suggested by the different results related to the severity of offenses identified in the Level 1 vs. Level 2 scenarios. Although the findings support the few extant studies, they have limited generalizability because our sample was a relatively homogeneous group of evangelical Christian students attending a midwestern American university. Although some studies using similar measures included congregants, the range of participants remains small. Other limitations include the self-report methodology and the lack of longitudinal data to see if attitude change occurs over time. We hope the inclusion of these data and the scale items encourages further exploration of the psychometric properties in diverse samples or the development of additional scales that will better assess what factors affect congregants when they consider forgiving and restoring errant clergy and other Christian leaders.