Predicting Relapse for Catholic Clergy Sex Offenders

Abstract

The Static-99 is an instrument commonly used to measure the likelihood of recidivism among sex offenders. The current study explores whether the Static-99 is an effective predictor of relapse among Catholic clergy who have had sexual contact with minors. Static-99 scores were compiled for 337 treated clergy who had offended against minors, including 21 who were known to have relapsed after treatment. Clergy were followed up for 5 to 25 years posttreatment (M = 16.05; SD = 5.12) after their completion of treatment. Post release, they were closely supervised, with explicit rules limiting their contacts with minors and church officials monitoring their compliance with posttreatment plans. Descriptive information on the victims of clergy sex offenders is provided. Although all clergy offenders had the same score on 4 of the 10 Static-99 items, Static-99 total scores still significantly predicted relapse with a moderate to large effect size (area under the curve [AUC] = .672; Cohen’s d = .808). Predictive accuracy of each item is also reported. Issues concerning the use of the Static-99 with this population are discussed.

Keywords

Static-99 recidivism Catholic clergy hebephilia pedophilia abuse

There is a growing body of literature about the characteristics of sex offenders and the factors that seem associated with recidivism (e.g., Hanson & Morton-Bourgon, 2005). Research on the profile of Catholic clergy who have molested children is less extensive but suggests that there are important differences between the clergy offender population and the overall sex offender population. The Static-99 is an instrument that has been used effectively to predict relapse among sex offenders (Hanson & Morton-Bourgon, 2009); as Catholic clergy show some differences from the general population of sex offenders, it seems important to know whether the Static-99 can be used effectively with this subgroup.

An important psychological profile of Roman Catholic clergy sex offenders has begun to emerge in the empirical literature over the past two decades. These studies make distinctions between clergy sex offenders and nonclergy sex offenders. Data suggest priest pedophiles and hebephiles are (a) likely to be older at first reported offense; (b) better educated; (c) more sexually conflicted; (d) likely to show less antisocial and criminal behavior disorders; (e) likely to have fewer victims overall but have a greater number of male pubescent victims; (f) less likely to report psychopathology; (g) less likely to exhibit hypersexuality; and (h) likely to have fewer additional paraphilic behaviors (Cartor, Cimbolic, & Tallon, 2008; Farrell, 2009; Haywood, Kravitz, Grossman, Wasyliw, & Hardy, 1996; John Jay College of Criminal Justice, 2004, 2006; Kafka, 2004; Langevin, Curnoe, & Bain, 2000; Loftus & Camargo, 1993; Meyer, Gray, & Calculator, 2008; Shupe, 2007).

The less severe psychopathology and offense histories among clergy offenders may be explained in part by the unique training, socialization, and screening experiences these men undergo in preparation for ordination. Factors such as a stable prosocial history, mental stability, and religious orientation should lessen vulnerabilities toward a criminal history or antisocial personality disorder.

According to a study conducted by the John Jay College of Criminal Justice (2004), the first reported incident of abuse for the majority of priests who have been accused of sexually abusing minors occurred an average of 11 years postordination. An analysis of the number of allegations per priest revealed that 56% of the 4,382 accused priests in the John Jay Study sample had one allegation made against them; 27% had two to three allegations; 14% had four to nine; and 3% had 10 or more allegations made against them. These statistics show evidence of a chronic offender effect, wherein a small percentage of clerics (3.5%) is responsible for a substantial proportion of all sexual offenses against minors (26%). The high incidence of male victims abused by clergy, the frequency of single victim allegations, and the typically late onset of deviant sexual behavior suggest additional etiological influences may need to be considered in future studies seeking to understand sexual offending cycles among the clergy (Cimbolic & Cartor, 2006; Dempsey, 1992; John Jay College of Criminal Justice, 2004, 2006; Kafka, 2004; Robinson, Greer, Estadt, & Thompson, 1994).

Similar to nonclergy sex offenders, clergy sex offenders who have completed sex offender treatment are less likely to reoffend than untreated clergy sex offenders (Hanson, Bourgon, Helmus, & Hodgson, 2009; Hanson & Morton-Bourgon, 2004; John Jay College of Criminal Justice, 2004, 2006; Lösel & Schmucker, 2005). Meta-analytic studies report a 10.9% sexual recidivism rate for treated sex offenders compared with a 19.2% recidivism rate for untreated sex offenders (Hanson et al., 2009). Montana and Thompson (2005) report a 7% recidivism rate for treated clergy sex offenders.

Surveys have found that Static-99 is the most common actuarial instrument used to predict recidivism among adult sex offenders (Jackson & Hess, 2007; McGrath, Cumming, Burchard, Zeoli, & Ellerby, 2010,). Research has demonstrated high levels of interrater reliability (Looman, 2006) and moderate predictive accuracy (Hanson & Morton-Bourgon, 2009). Many studies have supported the usefulness of the Static-99 in predicting sexual recidivism (Doren, 2004), although research studies typically involve men who have been incarcerated. Studies have been done on various populations including Canadian, American, British, Belgian, Dutch, and Swedish offenders (de Vogel, de Ruiter, van Beek, & Mead, 2004; Ducro & Pham, 2006; Hanson & Thornton, 2000; Sjöstedt & Långström, 2001). To date, the authors know of no published studies regarding the use of the Static-99 with clergy sex offenders. The present study was designed to understand whether the Static-99 is useful as a predictor of recidivism among Catholic clergy.

Method

Participants

The offenders were 337 male, Catholic priests and brothers who participated in a residential treatment program at St. Luke Institute between 1985 and 2005. Participants were between 5 and 25 years posttreatment, with a mean of 16.05 years and a standard deviation of 5.12. All participants had received a diagnosis of either Pedophilia (sexual involvement with prepubescent children) or Paraphilia—Not Otherwise Specified (sexual involvement with adolescents). For those diagnosed as hebephiles (both heterosexual hebephiles and homosexual hebephiles), victims were pubescent and below the age of 18 years. There were no differences depending on the victim gender. The diagnoses were made on the basis of the Diagnostic and Statistical Manual of mental disorders (4th ed. [DSM-IV]; American Psychiatric Association, 1994) criteria. Diagnoses were made by licensed psychologists and psychiatrists following an extensive assessment program that included multiple interviews, collateral data, and psychological testing. Participants were included in the study only if they met criteria and had at least one sexual contact with a minor. Some small number of clients were sent to a permanent care facility immediately after treatment because their diocese or community could not provide posttreatment monitoring; these clients were excluded from the study. The reason they were excluded from this study is that the level of monitoring at these facilities precluded contact with any minors and thus effectively removed any risk for relapse.

Participants were divided into Nonrelapser and Relapser categories. Of the clients treated, 21 have been documented to have relapsed following treatment. Relapse was defined broadly and included any self-report or report from others about any posttreatment behavior that we divided into the following three categories: (a) sexual contact with minors (sexual contact was defined as any touch on a sexual area of the body. Minors were defined as anyone below the age of 18 years); (b) use of child pornography; and (c) behavior that when interrupted seemed about to lead to sexual contact. An example of this third category would be a discharged patient who was convicted of violating the terms of his probation by stalking a former victim and went to prison for 15 months. Of the 21 relapsers, 12 fell into the first category, 2 into the second, and 8 into the third (with one relapser falling into both second and third categories).

The data from the Static-99 score sheets reveals that of the 337 participants, 298 had male victims; 39 participants, therefore, had only female victims. Our available data do not include how many of the 298 participants with male victims also had female victims. To analyze data regarding sex of victims and number of sexual contacts, we were able to extract more specific data for 91 participants; these data are presented in the Results section.

Because the participants were priests and brothers, we were able to obtain relapse information from multiple sources including reports from Bishops and religious superiors, newspaper reports, and reports from groups who monitor and publish data on Catholic clergy misconduct. The availability of these data from multiple sources leads us to believe that the accuracy of the relapse data is at least comparable, if not more complete, compared with studies that use rearrest data.

Program Description

Prior to participating in the treatment program, participants underwent a 4-to-5-day comprehensive evaluation that consisted of clinical, psychosocial, and spiritual interviews, psychological and neuropsychological testing, and a physical examination. The residential treatment program typically involved a 6-month length of stay. It included individual therapy, group therapies, spiritual guidance, psychoeducational groups (including relapse prevention), and other treatment modalities. The treatment program has always included a combination of cognitive-behavioral, psychoeducational, and psychodynamic approaches. There have been minor changes in emphasis over the years, but the treatment approach has generally remained intact.

Instrument

The Static-99 is an actuarial instrument that uses static (unchangeable) factors that correlate with sexual reconviction in adult males. It was developed by Hanson and Thornton (2000) and provides a baseline level of risk for sexual offender recidivism. The Static-99 includes 10 items that involve the following factual historical information: age of perpetrator, history of having lived with a lover for at least 2 years, convictions for index nonsexual violence, convictions for prior nonsexual violence, prior sexual offenses, prior sentencing dates (excluding index), sexual convictions for noncontact sexual offenses, a history of contact with unrelated victims, a history of contact with stranger victims, and a history of contact with male victims. Scores on individual items are added together to determine a total Static-99 score, which can range from 0 to 12. Total scores of 0 and 1 are designated to be in the Low Risk category, scores of 2 and 3 are in the Medium-Low Risk category, scores of 4 and 5 are in the Medium-High Risk category, and scores of 6 or higher are in the High Risk category.

In reviewing the Static-99 items, it was noted that there were two ways in which a priest or brother could have been involved in an intimate committed relationship for 2 years. One possibility is that this type of relationship could have occurred preordination, the other postordination. Both would lower Static-99 scores, however, the latter case would represent a lifestyle that was inconsistent with stated vows. Nevertheless, after consultation with one of the developers of the instrument (R. K. Hanson, personal communication, February 14, 2003), it was decided this item should remain unchanged in the scoring of Catholic clergy Static-99s. We understand that an intimate, committed relationship is incompatible with celibacy. Nevertheless, Hanson’s point is that lack of an intimate, committed relationship has been shown to increase the risk of relapse in the general population and must be included in the Static-99 for clergy. In addition, it is important to note that the Static-99 coding rules (Harris, Phenix, Hanson, & Thornton, 2003) were explicitly followed in the scoring of all items, including the items that are referred to in prior sex offenses and sentencing dates.

Overview of Analyses

Actuarial risk scales provide information on two properties of relapse risk. The first property, commonly referred to as relative risk, looks at the differences between the relapsers and the nonrelapsers. The other property, absolute risk, examines the actual relapse rates associated with particular scores. In other words, relative predictive accuracy examines whether higher risk offenders are more likely to relapse than lower risk offenders (regardless of the base rate for relapse), whereas absolute predictive accuracy assesses how well the predicted recidivism rates from the scale apply to a new sample. Different statistics are required to evaluate the adequacy of a risk-assessment scale for these separate tasks.

To evaluate relative predictive accuracy for the Static-99, we used the area under the receiver operating characteristic curve (ROC AUC), which is one of the most commonly used and recommended effect size statistics for recidivism prediction (Rice & Harris, 2005). AUC values are typically preferred to other measures of predictive accuracy (e.g., correlations) because they are not affected by the base rate of relapse (Humphreys & Swets, 1991; Rice & Harris, 2005). They are, however, influenced by the variance in risk scores (Hanson, 2008; Humphreys & Swets, 1991). In other words, AUCs tend to be smaller when there is a smaller range of Static-99 scores, which is the case in the current study.

AUC analyses plot the false positive rate by the true positive rate for each possible cut-off score on the risk scale, creating a curve (Swets, Dawes, & Monahan, 2000). The AUC can vary between 0 and 1, where an AUC value of .5 indicates the level of prediction that would be expected by chance. An AUC value less than .5 indicates negative predictive accuracy (e.g., those with lower scores would be more likely to relapse). AUC values between .5 and 1 indicate positive predictive accuracy, with numbers closer to 1 showing stronger predictive accuracy. Because .5 indicates chance level, then a confidence interval that does not include .5 demonstrates predictive accuracy significantly greater (or lesser) than chance. Another way to interpret the AUC is that it is the probability that a randomly selected relapser will have a higher Static-99 score than a randomly selected nonrelapser (assuming that AUC > .5 and the confidence interval does not include .5).

To assess whether the relative predictive accuracy of Static-99 with clergy sex offenders was significantly different than other validations using correctional samples of sex offenders, cumulative meta-analysis was used (Hanson & Broom, 2005). The predictive accuracy of Static-99 was compared with the meta-analytic findings of Hanson and Morton-Bourgon (2009). Given that Hanson and Morton-Bourgon’s (2009) meta-analysis used Cohen’s d (i.e., standardized mean difference scores) as the effect size for relative predictive accuracy, Cohen’s d was calculated for the current study as well. Cohen’s d measures the difference in Static-99 scores between relapsers and nonrelapsers, relative to how much relapsers normally differ from each other and how much nonrelapsers normally differ from each other. Typically, d’s of .20, .50, and .80 are considered small, moderate, and large, respectively (Cohen, 1988).

To examine whether the individual Static-99 items were significantly associated with relapse, Pearson’s chi-square significance test was used. To examine absolute predictive accuracy (i.e., are the recidivism estimates from Static-99 applicable to clergy offenders), we compared the observed number of relapsers with the predicted recidivism rate from the original and updated Static-99 norms using the E/O index (Gail & Pfeiffer, 2005; Rockhill, Byrne, Rosner, Louie, & Colditz, 2003). In this study, the E/O index was defined as the ratio of the predicted number of relapsers (E) divided by the observed number of relapsers (O) (Method M₀ from Viallon, Ragusa, Chavel-Chapelon, & Bénichou, 2009). Following Rockhill et al. (2003), the 95% confidence intervals for the E/O index were calculated using the Poisson variance for the logarithm of the observed number of cases (O):

95 % CI (E / O) = (E / O) \exp (\pm 1.96 \sqrt{1 / O})

The E/O index is both a significance test and a measure of effect size. If the predicted number of relapsers perfectly matches the observed number of relapsers, the E/O index will be 1. E/O values less than 1 mean that Static-99 underestimated relapse rates, and E/O values greater than 1 mean that Static-99 overestimated relapse. Ninety-five percent confidence intervals that do not include 1.0 indicate significant differences between the observed and predicted relapse rates.

Results

Victim Information

Table 1 presents descriptive information on the victims, including the number of contacts (i.e., incidents), the number of victims (broken down by male and female), and the youngest and oldest victim age. As noted above, information about the number of victims and number of victim contacts was available for only 91 clergy offenders, and, in many cases, the number of contacts or victims was approximated (e.g., 500). Most variables had an extreme positive skew (e.g., an offender with approximately 1,300 victim contacts), which severely distorts the mean. The median, therefore, provides a better summary. The number of victim contacts ranged between 1 and 1,300, with a median of 20, and the number of child victims ranged between 1 and 500, with a median of 4. The number of male victims ranged between 0 and 500 (median = 4) with 81% of offenders with victim information (n = 74) having exclusively male victims. The number of female victims ranged between 0 and 10 (median = 0), with 10% of the offenders having exclusively female victims (n = 9). Eight offenders (9%) had both male and female victims. The youngest victim age ranged between 5 and 17, whereas the oldest victim age ranged between 8 and 17.

Table I.

Summary of Victim Information (n = 91 clergy)

	Minimum	Maximum	Median	M	SD
Number of contacts	1	1,300	20	56.7	149.3
Number of child victims	1	500	4	15.4	54.0
Number of male victims	0	500	4	14.7	54.0
Number of female victims	0	10	0	0.4	1.2
Youngest victim age	5	17	13	12.8	2.7
Oldest victim age	8	17	17	15.9	1.9

Independent samples t tests were used to compare victim information for clergy offenders with solely male victims (n = 74) to those with solely female victims (n = 9). Offenders with victims of both genders were excluded.¹ Given the skewed data, equal variances were not assumed, which reduces the degrees of freedom. Offenders with male victims had significantly more total victims (M = 17.2, SD = 59.6) than offenders with female victims (M = 2.7, SD = 2.8; t[75] = −2.075, p = .041). In addition, the oldest victim age was significantly higher for offenders with male victims (M = 16.0, SD = 1.4) compared to those with female victims (M = 13.2, SD = 3.4; t[8] = −2.352, p = .045). Last, the youngest victim age appeared lower for offenders with female victims (M = 10.6, SD = 4.3) compared to those with male victims (M = 13.3, SD = 2.1), though this difference only approached significance (t[8] = −1.864, p = .097). These results indicate that offenders with exclusively male victims tend to have more victims and target older victims.

Relative Predictive Accuracy

Table 2 presents the distribution of Static-99 total scores for the sample as well as the relapse rates per score. The mean score was 3.1 (SD = 0.6, ranging from 1 to 6). Notably, the 0.6 standard deviation is substantially lower than the 2.0 standard deviation from the development samples (Hanson & Thornton, 2000), which demonstrates the restricted range of Static scores among clergy offenders. Interestingly, however, the mean Static-99 score was almost identical to the mean of the development samples (M = 3.2). Static-99 demonstrated statistically significant predictive accuracy. Specifically, a randomly selected relapser would have a 67% chance of having a higher Static-99 score than a randomly selected nonrelapser (AUC = .672, p = .008, 95% CI of [.552, .793]; Cohen’s d = .808, 95% CI of [.362, 1.253]).

Table 2.

Static-99 Score Distribution and Relapse Rates

Static-99 Score	N	n relapse (%)
1	3	0 (0.0%)
2	34	0 (0.0%)
3	243	12 (4.9%)
4	51	8 (15.7%)
5	5	0 (0.0%)
6	1	1 (100.0%)
Total	337	21 (6.2%)

Note: Static-99 scores can range between 0-12. Only scores with at least one nonzero cell frequency are presented in the table.

The predictive accuracy of Static-99 was compared with Hanson and Morton-Bourgon’s (2009) meta-analysis of 63 Static-99 studies (N = 20,010). They found an average weighted Cohen’s d of .67 (95% CI of [.62, .72]), with a Q statistic of 129.85 (Q measures the variability in predictive accuracy across studies). Although the Cohen’s d of .808 in the current study was higher than the meta-analytic average, this difference was not statistically significant (this is assessed by the change in variability across studies, Q_Δ = 0.37, df = 1, p = .543). Integrating the current effect size into the meta-analysis yields the same weighted average for Static-99 (d = .67, 95% CI of [.62, .72]), with a slightly different Q statistic (Q = 130.22, k = 64, N = 20,347).

Table 3 displays the scores on individual items, broken down by relapse status. There was no variability on four items. Specifically, all participants had less than four prior sentencing dates and had no prior convictions (or the equivalent of a conviction) for nonsexual violence, none of the participants were less than 25 years old, and none had lived with a lover for at least 2 years. For these four items, it was not possible to examine their relationship to relapse. In addition, several items had limited variability. Only one participant had a conviction (or its equivalent) for nonsexual violence as part of the index sex offence, and only two participants had convictions for noncontact sexual offences. Of the six items with at least some variability, three were significantly associated with recidivism (prior sex offences, noncontact sex offences, and index nonsexual violence). In addition, the association between relapse and having male victims approached significance (p = .087).

Table 3.

Frequencies and Predictive Accuracy for each Static-99 Item (N = 337)

Static-99 item	Frequency	%	Nonrelapse	Relapse	χ²	df	p
Prior sex offences					15.86	2	<.001
0	291	86.4	275	16
1	45	13.4	41	4
2	1	0.3	0	1
3	0	0.0	0	0
Prior sentencing dates					—	—	—
0	337	100.0	316	21
1	0	0.0	0	0
Noncontact sex offences					30.28	1	<.001
0	335	99.4	316	19
1	2	0.6	0	2
Index nonsexual violence					15.09	1	<.001
0	336	99.7	316	20
1	1	0.3	0	1
Prior nonsexual violence					—	—	—
0	337	100.0	316	21
1	0	0.0	0	0
Unrelated victims					0.34	1	.561
0	5	1.5	5	0
1	332	98.5	311	21
Stranger victims					0.78	1	.379
0	319	94.7	300	19
1	18	5.3	16	2
Male victims					2.93	1	.087
0	39	11.6	39	0
1	298	88.4	277	21
Young					—	—	—
0	337	100.0	316	21
1	0	0.0	0	0
Ever lived with lover					—	—	—
0	0	0.0	0	0
1	337	99.7	316	21

Absolute Predictive Accuracy

Table 4 presents the observed number of relapsers as well as the number that would have been predicted by the Static-99 recidivism norms and the E/O values. Although results are presented for each risk category, only the moderate-low and moderate-high risk categories had a sufficient number of relapsers for a meaningful analysis. In the current sample, the follow-up period ranged between 5 and 25 years. Given that the recidivism norms for Static-99 are estimated for specific follow-up periods, it would be ideal to use fixed follow-up periods for the current sample, but that information was not available. For the current sample, the average length of follow-up was 16 years.

Table 4.

Comparing Observed Relapse Rates to Predicted Number of Recidivists Based on Static-99 Norms

	Current study			Original norms—15 years			2008 norms—Complete sample 10 years			2008 norms—Routine CSC 10 years
Static-99 category	N	n relapse	%	Predicted n relapse	E/O	95% CI	Predicted n relapse	E/O	95% CI	Predicted n relapse	E/O	95% CI
Low	3	0	0.0	0.3	—	—	0.2	—	—	0.1	—	—
Mod-low	278	12	4.3	48.7	4.06	[2.30, 7.14]	33.0	2.75	[1.56, 4.84]	13.2	1.10	[0.63, 1.94]
Mod-high	56	8	14.3	20.9	2.62	[1.31, 5.23]	11.7	1.46	[0.73, 2.92]	5.4	0.67	[0.33, 1.34]
High	1	1	100.0	0.5	0.52	[0.07, 3.69]	0.4	0.40	[0.06, 2.82]	0.2	0.22	[0.03, 1.57]
Total	338	21	6.2	87.9	4.18	[2.73, 6.42]	64.2	3.06	[1.99, 4.69]	28.6	1.36	[0.89, 2.09]

Note: CSC = Correctional Service of Canada. E/O ratios could not be computed for the low-risk category because at least one relapser was required. E/O ratios in italics are significantly different than 1.

Extrapolating from the 15-year recidivism estimates for the original Static-99 development sample (k = 3, n = 1,086; Harris et al., 2003), these norms would have predicted about 88 relapsers in the current sample. This is more than 4 times the number of relapsers that were actually observed (n = 21), which represents a statistically significant overestimation of relapse (E/O = 4.2, 95% CI of [2.7, 6.4]). The original norms also significantly overestimated the number of relapsers in both the moderate-low and moderate-high risk categories. These results suggest that the original recidivism norms for Static-99 are not applicable to clergy sex offenders.

The developers of Static-99 have noted that the original norms tend to overestimate recidivism and have released updated norms (see Helmus, Hanson, & Thornton, 2009), which are available from www.static99.org. The 10-year recidivism estimates combined across all available samples (k = 10, n = 1,621), predicted approximately 3 times as many relapsers than what was observed (E/O = 3.1, 95% CI of [2.0, 4.7]), which was statistically significant. In addition, these norms significantly overestimated relapse for the moderate-low risk category.

The Static-99 samples in the updated norms overrepresent samples of high-risk offenders and, therefore, may not be representative of all sex offenders. Restricting the normative data to routine prison samples (i.e., relatively unselected) from Canada provides lower relapse estimates. The 10-year estimates from the routine Correctional Service of Canada (CSC) samples (k = 2, n = 342) predicted a total of 28.6 relapsers, which was close to the observed 21 relapsers. This was still an overestimate, but not significantly so (E/O = 1.4, 95% CI of [0.9, 2.1]). Taking into account that the length of follow-up in the current study was greater than 10 years for most participants and that relapse information was obtained from multiple sources, it is possible that routine CSC norms may still be a bit too high for clergy sex offenders. Nonetheless, they appear to provide estimates of relapse that are within roughly the same range.

Discussion

It is important to note how relapse was defined in the current study to understand the discussion of the data. As mentioned in the Method section, relapse was defined broadly to include sexual contact, use of child pornography, and behavior that was judged to be about to lead to sexual contact at the time the behavior was interrupted. This is a broader definition of relapse than is used in most studies using the Static-99, and thus the number of relapsers must be understood in this context.

The data indicate that Relapsers have higher Static-99 scores than Nonrelapsers and the predictive accuracy of Static-99 is comparable with previous meta-analytic results. The predictive accuracy was impressive given that several individual items were scored in a single direction for all participants (i.e., there was a restriction of range on items and total scores). All 337 participants were above age 25; this is because priests and brothers go through an extensive educational and spiritual formation process before they are permitted to function ministerially. All participants had three or less sentencing dates and no prior convictions for nonsexual violence; it would be extremely unlikely that a priest or brother could have more than three incidents of criminal activity followed by sentencing and still continue in ministry. Finally, none of the participants had a history of having an intimate relationship of at least 2 years’ duration. This, obviously, would be prohibited by their vocational choice.

Of note, in our sample, only two participants (both relapsers) had convictions for a noncontact sexual offense. The fact that both these participants were relapsers raises the question of the possible importance of their history of a noncontact sex offense. It is possible that having had a noncontact sex offense in addition to a child victim is evidence of a broader level of sexual dysfunction, perhaps indicative of hypersexuality. Although the data for this are, of course, only present in this small number of participants, it is a subject which may need further study.

Only one participant (a relapser) received a point for “Index nonsexual violence.” The relatively low incidence of violent behavioral histories is probably due to a self-selection process in the clergy population; it is rare for those who are inclined toward violence to pursue the priesthood. Notably, both these items significantly predicted relapse.

In addition, having prior sex offences significantly predicted relapse. Compared to nonrelapsers, relapsers had a higher incidence of having had previous inpatient treatments for sexual offending prior to the event that precipitated their referral for the treatment program used in this current study. This suggests that previous treatment “failures” predict future posttreatment relapse. According to Hanson (personal communication, February, 14, 2003), clergy offender residency at inpatient treatment programs should be understood as equivalent to convictions. Therefore, previous inpatient treatment failure is synonymous with previous convictions.

As shown in Table 2, of the 337 participants, 243 (72%) scored a “3” on the Static-99. Of those who scored a “3,” 12 (5%) were relapsers. The typical participant with a score of “3” received one point for having unrelated victims, one point for having had a male victim, and one point for a “No” score on the item about having had a 2-year committed relationship. As this group is by the rules of their vocation, unmarried (thus without a committed, intimate relationship), without children (thus reducing the possibility of incest), and most often have male victims, a score of “3” is very common. This is important to consider when working with this population. Scores lower or higher than 3 were more infrequent and seemingly more meaningful in regard to relapse risk. None of the 38 participants with a Static-99 score of 1 or 2 was a Relapser. This suggests there may be a subgroup of priest/brother child molesters identified by the Static-99 who are at very low risk for relapse. This finding may have an impact on the risk level and consequent monitoring needs for low-risk participants. Also of interest, there were 57 clergy in our sample with Static-99 scores of 4 (medium-high range) and above. Of these, nine (16%) were Relapsers.

The meaning of this finding must be understood carefully. It suggests that there is a subgroup of clergy identified on the Static-99, who are in the medium-high range or above on this instrument, who may be at significantly higher risk for relapse than those clergy who score lower on the Static-99. However, the low relapse base rate in this study has important implications. The likelihood of a particular individual with a score of 4 or more being a relapser is still quite low; 84% of our sample clergy with these scores were not relapsers. Thus, although the relationship between Static-99 scores and relapse is statistically significant, the likelihood of any particular person with a score of 4 or more relapsing is still low. The original recidivism estimates from Static-99 significantly overestimated recidivism and are, therefore, not appropriate to use with this population. The updated Static-99 recidivism estimates from routine Canadian prison samples provided more plausible estimates, although it is still possible that these numbers will overestimate relapse rates for clergy offenders.

Interpretation of the current findings should also consider similarities and differences between the current clergy sample and the samples used to derive the Static-99 recidivism estimates. Notably, the current sample consisted of clergy members whose sex offending was dealt with by their religious organization, whereas the Static-99 normative samples included offenders from the traditional criminal justice system. Aside from offending circumstances, it is unclear how these offender groups may differ. We suspect (supported by the item analyses) that the clergy offenders had much lower levels of antisociality than the normative samples, which would reduce their recidivism risk. In addition, it is important to note that all the participants in our study were given treatment, completed treatment, and had strict limitations on their contact with minors posttreatment and had church officials monitor their compliance with posttreatment plans. In contrast, the samples included in the original Static-99 recidivism estimates were largely untreated (Hanson & Thornton, 2000). Most offenders in the routine CSC samples included in the updated norms, however, were exposed to multiple treatment programs adhering to principles of effective correctional treatment (Helmus et al., 2009), which may partly explain the closer match to the clergy sample in terms of recidivism. Both normative samples would have been exposed to community supervision, though not to the same extent as the clergy sample. It is possible that this additional supervision would have reduced relapse rates among the clergy offenders, although this is unlikely to have a large effect given meta-analytic findings that community supervision does not notably decrease recidivism (Bonta, Rugge, Scott, Bourgon, & Yessine, 2008). In addition, it is possible that the closer supervision and multiple sources of recidivism information would have increased detection rates of new offences, increasing the observed recidivism rate. The posttreatment implications of these findings are complex. At the present time, all clergy child molesters who have been through our program had the above-mentioned restrictions placed on them posttreatment. The results of the current study (as well as the considerable research on the principles of effective correctional treatment espoused by Andrews & Bonta, 2010) suggest that given limited resources for monitoring, church officials might target their resource allocation toward clergy who have medium-high range or higher scores. However, it would be necessary that church authorities understand the level of relapse risk for any individual.

There are several directions for future research with this population. Clearly, the research findings would be more easily generalized if the data could include Catholic clergy offenders from other sources including those treated at other treatment centers. This would have the benefit of increasing the sample size in this study.

Although Static-99 significantly predicted relapse, the lack of variability on many of the items suggest that there are likely other important factors/dynamics to consider with this population. Our current research is focused on finding additional factors that may contribute to recidivism risk. Some preliminary analysis has suggested the possibility that higher scores on the Masculinity/Femininity scale on the MMPI-II and the severity of neuropsychological impairment may increase risk of recidivism. More research is also needed to examine dynamic risk factors among this population (e.g., sexual self-regulation, problem solving). In addition, given that scores on certain items showed no variability and several items showed very limited variability, further analysis of Static-99 items may result in a more efficient risk assessment instrument for Catholic clergy offenders. Minimally, however, the current research provides preliminary support for using the Static-99 with clergy sex offenders, particularly for relative risk decisions (e.g., clergy offenders with higher Static-99 scores are more likely to reoffend than those with lower Static-99 scores). The absolute recidivism estimates of Static-99, however, should be used with caution. Currently, the results suggest that the routine samples provide the best normative comparison but that they may still overestimate recidivism. Evaluators could consider the routine norms as an upper-bound estimate of relapse rates for this population.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

American Psychiatric Association. (1994). Diagnostic and Statistical Manual of mental disorders (4th ed.). Washington, DC: Author.

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). Newark, NJ: LexisNexus/Anderson.

Bonta

Rugge

Scott

T.-L.

Bourgon

Yessine

A. K.

(2008). Exploring the black box of community supervision. Journal of Offender Rehabilitation, 47, 248-270. doi:10.1080/10509670802134085

Cartor

Cimbolic

Tallon

(2008). Differentiating pedophilia from ephebophilia in cleric offenders. Sexual Addiction and Compulsivity, 15, 311-319. doi:10.1080/10720160802289280

Cimbolic

Cartor

(2006). Looking at ephebophilia through the lens of cleric sexual abuse. Sexual Addiction and Compulsivity, 13, 247-359. doi:10.1080/10720160601011257

Cohen

(1988). Statistical power analysis for the behavioural sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

de Vogel

de Ruiter

van Beek

Mead

(2004). Predictive validity of the SVR-20 and Static-99 in a Dutch sample of treated sex offenders. Law and Human Behavior, 28, 235-251. doi:10.1023/B:LAHU.0000029137.41974.eb

Dempsey

(Ed.). (1992). Report of Cardinal’s Commission on clergy sexual misconduct with minors. Report to the Cardinal Archbishop of the Chicago Roman Catholic Archdiocese. Archdiocese of Chicago: Origins-CNS Documentary Service, 22(1), 273-283.

Doren

D. M.

(2004). Stability of the interpretive risk percentages for the RRASOR and Static-99. Sexual Abuse: Journal of Research and Treatment, 16, 25-36. doi:10.1177/107906320401600102

10.

Ducro

Pham

(2006). Evaluation of the SORAG and the Static-99 on Belgian sex offenders committed to a forensic facility. Sexual Abuse: A Journal of Research and Treatment, 18, 15-26. doi:10.1177/107906320601800102

11.

Farrell

D. P.

(2009). Sexual abuse perpetrated by Roman Catholic priests & religious. Mental Health, Religion and Culture, 12(1), 39-53. doi:10.1080/13674670802116101

12.

Gail

M. H.

Pfeiffer

R. M.

(2005). On criteria for evaluating models of absolute risk. Biostatistics, 6, 227-239. doi:10.1093/biostatistics/kxi005

13.

Hanson

R. K.

(2008). What statistics should we use to report predictive accuracy? Crime Scene, 15(1), 15-17. Retrieved from http://www.cpa.ca/cpasite/userfiles/Documents/Criminal%20Justice/Crime%20Scene%202008-04.pdf

14.

Hanson

R. K.

Broom

(2005). The utility of cumulative meta-analysis: Application to programs for reducing sexual violence. Sexual Abuse: A Journal of Research and Treatment, 17, 357-373. doi:10.1007/s11194-005-8049-1

15.

Hanson

R. K.

Morton-Bourgon

K. E.

(2005). The characteristics of persistent sexual offenders: A meta-analysis of recidivism studies. Journal of Consulting and Clinical Psychology, 73, 1154-1163. doi:10.1037/0022-006X.73.6.1154

16.

Hanson

R. K.

Morton-Bourgon

K. E.

(2009). The accuracy of recidivism risk assessments for sexual offenders: A meta-analysis of 118 prediction studies. Psychological Assessment, 21, 1-21. doi:10.1037/a0014421

17.

Hanson

R. K.

Morton-Bourgon

K. E.

(2004). Predictors of sexual recidivism: An updated meta-analysis (Research Report No. 2004-02). Ottawa, Ontario: Public Safety and Emergency Preparedness Canada.

18.

Hanson

R. K.

Thornton

(2000). Improving risk assessments for sex offenders: A comparison of three actuarial scales. Law and Human Behavior, 24, 119-136. doi:10.1023/A:1005482921333

19.

Hanson

R. K.

Bourgon

Helmus

Hodgson

(2009). The principles of effective correctional treatment also apply to sexual offenders: A meta-analysis. Criminal Justice and Behavior, 36, 865-891. doi:10.1177/0093854809338545

20.

Harris

A. J. R.

Phenix

Hanson

R. K.

Thornton

(2003). Static-99 coding rules: Revised 2003. Ottawa, Ontario: Solicitor General Canada.

21.

Haywood

T. W.

Kravitz

H. M.

Grossman

L. S.

Wasyliw

O. E.

Hardy

D. W.

(1996). Psychological aspects of sexual functioning among cleric and noncleric alleged sex offenders. Child Abuse and Neglect, 20, 527-536.

22.

Helmus

Hanson

R. K.

Thornton

(2009). Reporting Static-99 in light of new research on recidivism norms. Forum, 21(1), 38-45.

23.

Humphreys

L. G.

Swets

J. A.

(1991). Comparison of predictive validities measured with biserial correlations and ROCs of signal detection theory. Journal of Applied Psychology, 76, 316-321. doi:10.1037/0021-9010.76.2.316

24.

Jackson

R. L.

Hess

D. T.

(2007). Evaluation for civil commitment of sex offenders: A survey of experts. Sexual Abuse: A Journal of Research and Treatment, 19, 409-448. doi:10.1007/s11194-007-9062-3

25.

John Jay College of Criminal Justice. (2004). The nature and scope of sexual abuse of minors by Catholic priests and deacons in the United States. Washington, DC: USCCB.

26.

John Jay College of Criminal Justice. (2006). The nature and scope of sexual abuse of minors by Catholic priests and deacons in the United States: Supplementary data analysis. Washington, DC: USCCB.

27.

Kafka

M. P.

(2004). Sexual molesters of adolescents, ephebephilia, and Catholic clergy: A review and analysis. In Hanson

R. K.

Friedmann

Lutz

(Eds.), Sexual abuse in the Catholic Church: Scientific and legal perspectives (pp. 51-59). Vatican City: Libreria Editrice Vaticana.

28.

Langevin

Curnoe

Bain

(2000). A study of clerics who commit sexual offenses: Are they different from other sex offenders? Child Abuse and Neglect, 24, 535-545. doi:10.1016/S0145-2134(00)00113-7

29.

Loftus

J. A.

Camargo

R. J.

(1993). Treating the clergy. Annals of Sex Research, 6, 287-303. doi:10.1007/BF00856865

30.

Looman

(2006). Comparison of two risk assessment instruments for sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 18, 193-206. doi:10.1177/107906320601800206

31.

Lösel

Schmucker

(2005). The effectiveness of treatment for sexual offenders: A comprehensive meta-analysis. Journal of Experimental Criminology, 1, 117-146. doi:10.1007/s11292-004-6466-7

32.

McGrath

R. J.

Cumming

G. F.

Burchard

B. L.

Zeoli

Ellerby

(2010). Current practices and emerging trends in sexual abuser management: The Safer Society 2009 North American survey. Brandon, VT: Safer Society Foundation Press.

33.

Meyer

Gray

J. J.

Calculator

L. K.

(2008). The effect of personality pathology on treatment outcome for Axis 1 disorders in dysfunctional clergy. Mental Health, Religion and Culture, 11, 239-246. doi:10.1080/13674670701276972

34.

Montana

Thompson

(2005, November). Risk assessment for priest sex offenders: Developing a risk assessment tool to measure relapse risk. Paper presented at the annual Research and Treatment Conference of the Association for the Treatment of Sexual Abusers, Salt Lake City, UT.

35.

Rice

M. E.

Harris

G. T.

(2005). Comparing effect sizes in follow-up studies: ROC area, Cohen’s d, and r. Law and Human Behavior, 29, 615-620. doi:10.1007/s10979-005-6832-7

36.

Robinson

E. A.

Greer

J. M.

Estadt

B. K.

Thompson

(1994). Psychometric discrimination of ephebophilic, pedophilic and sexually compulsive clergy. Unpublished manuscript, Saint Luke Institute, Suitland, MD.

37.

Rockhill

Byrne

Rosner

Louie

M. M.

Colditz

(2003). Breast cancer risk prediction with a log-incidence model: Evaluation of accuracy. Journal of Clinical Epidemiology, 56, 856-861. doi:10.1016/S0895-4356(03)00124-0

38.

Shupe

(2007). Spoils of the kingdom: Clergy misconduct and religious community. Chicago: University of Illinois Press.

39.

Sjöstedt

Långström

(2001). Actuarial assessment of sex offender recidivism risk: A cross-validation of the RRASOR and the Static-99 in Sweden. Law and Human Behaviour, 25, 629-645. doi:10.1023/A:1012758307983

40.

Swets

J. A.

Dawes

R. M.

Monahan

(2000). Psychological science can improve diagnostic decisions. Psychological Science in the Public Interest, 1, 1-26. doi:10.1111/1529-1006.001

41.

Viallon

Ragusa

Clavel-Chapelon

Bénichou

(2009). How to evaluate the calibration of a disease risk prediction tool. Statistics in Medicine, 28, 901-916. doi:10.1002/sim.3517