Differences in Sexual Interest in Children Between Men Who Commit Undetected Versus Detected Sexual Offenses

Abstract

This study examined whether men who committed undetected sexual offenses would show more evidence of sexual interest in children than their detected counterparts. It also considered whether denial or minimization of offending history explained observed differences between undetected and detected men. In an archival database of 2,236 men, 96 were undetected and the remainder were detected men who varied in the extent to which they admitted their offending: complete deniers, partial deniers, those who admitted their offense history, and those who disclosed additional unknown offenses. There were differences in self-reported sexual interest in children, but relatively few differences when sexual interest in children was assessed by phallometry or sexual victim history. There were no differences between undetected and detected men who admitted to additional victims. The results suggest that denial is likely more important in understanding group differences on sexual interest in children than detection status.

Keywords

pedophilia sexual interest victim selection SSPI-2 phallometry undetected offending dark figure of sexual crime denial

Victimization data from Canada and the United States suggest that over 80% of sexual offenses are not reported to police (Department of Justice, 2017; Perreault & Brennan, 2009). Data from men who admit to committing sexual offenses also suggest that many offenses are undetected by the criminal justice system (e.g., Abel et al., 1987; Asahina, 2010; Lisak & Miller, 2002). Despite research suggesting that many men who commit sexual offenses remain undetected, most research on sexual offending has been conducted on samples of detected individuals recruited in forensic clinics, correctional institutions, or while on probation or parole supervision.

The impact of detection status on our knowledge about sexual offending is unknown: it may be that our explanatory models, risk assessment tools, or treatments do not apply—or apply differently—to undetected men. Studies comparing undetected and detected offenders are needed. Ideally, we would compare undetected and detected men using the same measures conducted under similar circumstances, as opposed to less direct comparisons that introduce methodological or other confounds that might explain any differences (e.g., anonymous online recruitment of undetected offenders compared with clinical recruitment of detected offenders). Of particular interest are factors important in models of sexual offending and in established sexual offense risk assessment tools, mainly antisociality and atypical sexual interests (Hanson & Morton-Bourgon, 2005; Seto, 2018). Antisociality refers to general criminal tendencies, as indicated by measures of criminal history, antisocial personality traits, and general antisocial behavior. Atypical sexual interests refer to excessive sexual preoccupation or paraphilic interests that are empirically associated with sexual offending, including pedophilia (sexual interest in prepubescent children), non-consensual sexual sadism (suffering or humiliation of others), exhibitionism (exposing one’s genitals to an unsuspecting stranger), or voyeurism (surreptitiously watching an unsuspecting stranger undress or engaged in other usually private behavior). We may find, for example, that undetected offenders are higher in both antisociality and atypical sexual interests than detected offenders, which would mean the current evidence underestimates the importance of these factors.

The generalizability of findings from detected to undetected men who have sexually offended has been the focus of a few studies (e.g., Kuhle et al., 2017; Neutze et al., 2011, 2012). Overall, those who were undetected were younger, had a greater number of victims, and committed more non-sexual offenses than those who were detected (Asahina, 2010; Lisak & Miller, 2002). Neutze and colleagues (2012) compared men who were undetected and detected for sexual offending, all of whom identified as having a sexual interest in children and self-referred for confidential assessment. There were more similarities than differences between these groups on several risk factors for sexual recidivism, but those who were undetected were more likely to report sexual preoccupation than those who were detected. Furthermore, undetected men were more likely to report other paraphilic sexual interests than detected men, among those who committed both child sexual exploitation material offenses and contact sexual offenses against children.

Differences in Sexual Interest in Children

Prior studies have not examined the difference between undetected and detected men on sexual interest in prepubescent and/or pubescent children (see Seto, 2018). This is noteworthy because sexual interest in children is emphasized in virtually all models of sexual offending against children and is one of the best predictors of sexual recidivism, leading to its classification as a psychologically meaningful risk factor (Mann et al., 2010). Furthermore, there appears to be an assumption that many individuals who commit undetected offenses have a sexual interest in children, given that many programs targeting undetected men in the community were specifically developed for men with sexual interest in children (Prevention Project Dunkelfeld; Beier et al., 2009).

Because all men in Neutze and colleagues’ (2012) study had a sexual interest in children, researchers were unable to examine group differences on this key variable, though they found no differences in the reported degree of exclusivity of sexual interest in children (Neutze et al., 2012). Furthermore, sexual interest in children in Neutze et al.’s (2012) study was assessed through self-report, but other measures are often used, especially in clinical or forensic settings where men who have sexually offended might be less forthcoming (see Seto, 2018). These other measures include phallometric assessment of penile responses to sexual stimuli depicting children, relative to sexual stimuli depicting adults, and inferences about sexual interest in children on the basis of victim characteristics such as having boy victims, multiple child victims, younger child victims, and unrelated child victims (Seto, Stephens, et al., 2017).

Differences in Willingness to Admit Sexual Offense Details

An important methodological consideration in research comparing undetected and detected men is that detected men vary in how forthcoming they are about their offending (e.g., Abel et al., 1987), ranging from denial of all known offenses to admission of undetected offenses beyond their criminal records. Denial of offense history may distinguish undetected and detected men, as by definition, undetected men are admitting to sexual interest in children or sexual offenses that were otherwise not officially known. The consideration of denial is important as Neutze and colleagues (2012) specifically postulated that group differences may have been driven by undetected men being more forthcoming than detected men, especially because it may have increased the likelihood of direct benefits in their study (i.e., inclusion in group treatment).

Past studies have not been able to control for the likely possibility that undetected men might be more forthcoming when disclosing their offending history compared with detected men (Yates, 2009). In support of this assertion, men who deny their offending history were more likely to engage in impression management and thereby present a distorted picture of themselves (Baldwin & Roys, 1998). We may find that undetected men are more similar to detected men who also admit to undetected offenses than they are to those who deny their known criminal history. The inability to control for denial in past studies is because much of the research that has compared undetected and detected men has utilized data from the Dunkelfeld sample (e.g., Kuhle et al., 2017; Neutze et al., 2012). The Dunkelfeld is a unique research and clinical setting because individuals can present for assessment and treatment without concerns that their confidentiality will be breached because of mandatory reporting laws. As such individuals in the Dunkelfeld studies would be expected to be more forthcoming in disclosing potentially incriminating information compared with a North American sample where mandatory sexual abuse reporting laws apply (see McPhail et al., 2018).

The Present Study

This study advances past research by providing an in-depth examination of sexual interest in children in undetected men, which has not been previously investigated. It further adds to the literature by comparing undetected men to different groups of detected men who varied in denial. Specifically, we compared undetected men to four detected groups who differed in the extent to which they admitted to their official victim history: men who denied the complete extent of their known victim history; men who admitted to offending against some victims but denied the full extent of their known offending; men who admitted to all of their known offending; and men who admitted to more victims than their known offending history.

We hypothesized that undetected men would be more likely to report sexual interest in children, show relatively greater phallometric responses to children than to adults, and to have more sexual victim characteristics suggestive of a sexual interest in children. For all comparisons, we expected larger differences between undetected men and those who denied their entire offense history compared with those who admitted to more than their known victims. If there were fewer differences between undetected men and detected men who admitted to undetected victims, this would suggest that the degree of denial in detected men might be more important in understanding differences between undetected and detected men.

Method

Sample

The sample was comprised of 2,238 men assessed between 1995 and 2011 at a sexual behavior clinic located in a university-affiliated hospital in a large metropolitan Canadian city. All men in the sample had committed at least one sexual offense. The operationalization of a sexual offense is consistent with the definition in the Canadian Criminal Code and involved any sexual contact with a non-consenting adult or with a child (defined as under age 14 up to May 2008, when the age of consent was changed to 16). In the definition of a sexual offense, both contact and non-contact offenses (e.g., exhibitionism) were included. Child pornography offenses were excluded from the official victim count data and are captured elsewhere in the dataset. Sexual offending history was based on both official file information that accompanied the referral and/or self-disclosures.

Although most of the samples were referred from the criminal justice system and were known to law enforcement (e.g., referred by the courts or by probation or parole officers), a subset of men had sexually offended but were unknown to law enforcement. These men could be referred by other professionals (e.g., physicians, professional colleges) because their sexual behavior was problematic (e.g., excessive pornography use) and/or they were distressed or impaired by their sexual interests or behavior (e.g., unhappy about having sexual fantasies about children). They may also have been involved in the legal system for non-sexual offenses. Regardless of their referral status, all men underwent the same assessment procedure. Descriptive information for the total sample is contained in the first column of Table 1.

Table 1.

Descriptive Statistics for the Total Sample and Groups.

Demographic & referral variables	Total sample[n = 2,236]	Undetected[n = 96]	Detected complete deniers[n = 629]	Detected partial deniers[n = 281]	Detected admitters[n = 951]	Detected disclosers[n = 279]	F or χ²	ES
Age at assessment	39.1 (13.2) [n = 2,236]	34.9 (10.7) [n = 96]	40.8 (13.0) [n = 629]	40.0 (13.9) [n = 281]	38.1 (13.1) [n = 951]	39.3 (13.5) [n = 279]	6.90***	.01
Age at first known sex offense	29.3 (12.9) [n = 1,905]	18.2 (8.0) [n = 76]	33.3 (13.0) [n = 555]	26.7 (11.6) [n = 240]	30.7 (12.8) [n = 805]	21.1 (9.4) (n = 229)	61.54***	.11
Age at most recent known sex offense	35.7 (12.9) [n = 1,909]	29.7 (11.0) [n = 76]	37.1 (12.9) [n = 558]	36.2 (13.3) [n = 239]	34.9 (12.8) [n = 804]	36.3 (12.8) (n = 232)	6.92***	.01
Referral status							603.37***	.63
Probation or parole	48%[n = 731]	5%[n = 3]	57%[n = 262]	43%[n = 87]	48%[n = 303]	45%[n = 76]
Lawyer	17%[n = 259]	2%[n = 1]	22%[n = 101]	14%[n = 29]	17%[n =105]	14%[n = 23]
Institution	28%[n = 420]	8%[n = 5]	19%[n = 90]	39%[n = 78]	32%[n = 202]	29%[n = 45]
Self-referred	8%[n = 120]	86%[n = 53]	2%[n = 10]	4%[n = 7]	4%[n = 26]	14%[n = 24]
Ethnicity							69.96***	.18
Caucasian	79%[n = 1,748]	87%[n = 83]	72%[n = 446]	81%[n =226]	79%[n = 751]	87%[n = 242]
Asian or Pacific Islander	3%[n = 70]	4%[n = 4]	4%[n = 22]	1%[n = 4]	3%[n = 32]	3%[n = 8]
Black	7%[n = 145]	1%[n = 1]	9%[n = 57]	6%[n = 16]	7%[n = 66]	2%[n = 5]
East Indian	4%[n = 92]	1%[n = 1]	7%[n = 46]	4%[n = 10]	3%[n = 26]	3%[n = 9]
Aboriginal	3%[n = 62]	1%[n = 1]	2%[n = 14]	5%[n = 15]	3%[n = 27]	2%[n = 5]
Other	5%[n = 104]	6%[n = 6]	6%[n = 36]	3%[n = 9]	5%[n = 45]	3%[n = 8]

Note. For all age variables, we report mean values and standard deviations in brackets. For categorical variables, we report the percentage of men with that specific characteristic, with the numbers in square brackets representing the number of men with that specific characteristic. There is variation in the sample size for different analyses due to missing data. The reported effect size was partial eta-squared for ANOVAs and Cramer’s V for chi-square analyses. ES = effect size; ANOVA = analysis of variance.

***

p < .001.

There were five groups in the study. Our undetected group (n = 96 or 4% of the total sample) was comprised of men who had no known victims according to official records but admitted to sexual offending during a clinical interview.¹ We had four detected groups who differed according to the extent to which they admitted to the known victim information on file from police, court, or criminal record documents: detected men who denied the complete extent of their known victim history (detected complete deniers; n = 629, 28%); men who admitted to offending against some victims in the interview but denied the full extent of their known offending history (detected partial deniers; n = 281, 13%); men who admitted to all of their known offending history (detected admitters; n = 951 or 43%); and men who admitted to offending against more victims than their known offending history (detected disclosers; n = 279, 13%). We excluded two men with missing victim information, resulting in a final sample of 2,236 men; however, sample sizes varied between analyses because there is some missing data for dependent variables.

Measures

Self-report

Prior to undergoing phallometric testing, men ranked their sexual interests for different sex and age groups during an interview. Men ranked their sexual interest in male and female prepubescent and pubescent children from no sexual interest (score of 1) to strongest sexual interest (score of 5). Higher scores indicated stronger sexual interest in prepubescent or pubescent children; for the purposes of these analyses, gender was ignored.

We took the highest rank score for prepubescent or pubescent children. For example, if an individual ranked their sexual interest in prepubescent children as a 3 (moderate sexual interest) and their sexual interest in pubescent children as a 5 (strongest sexual interest), the score of 5 was used. Our research team has previously shown that these rankings were positively correlated with other measures of sexual interest in children and had predictive validity for sexual non-contact recidivism (Stephens, Cantor, Goodwill, & Seto, 2017; Stephens, Seto, Goodwill, & Cantor, 2017).

Phallometry

Phallometric testing involved the measurement of penile blood volume change to photos that were shown at the same time as audiotaped stories. The audiotaped stories described sexual activities with a person resembling the person in the photos. Stimuli differed by sex (male or female) and sexual maturity (prepubescent, pubescent, and fully mature adults). Men viewed four trials per stimulus categories: prepubescent boys, prepubescent girls, pubescent boys, pubescent girls, men, and women. A neutral category (landscapes) was also shown. In total, there were 28 trials (four trials for each of the seven stimulus categories) and each trial lasted 1 to 2 min. The interval between trials was dependent on the individual’s return to their baseline level of response. The four trial scores per stimulus category were averaged and the data were transformed to z-scores for the purposes of analyses. Men were excluded from analyses if they exhibited only low responses (less than 1 cc to any sexual stimulus) or if they had their greatest response to the neutral category.

A phallometric pedohebephilia index was created by subtracting the maximum response to adults from the maximum response to prepubescent or pubescent children. A positive score on the index indicated greater sexual interest in prepubescent or pubescent children than adults. A negative score indicated greater sexual interest in adults than prepubescent or pubescent children. A score of 0 indicated equal responding to prepubescent or pubescent children and adults.

Several research studies that have examined the psychometric properties of phallometric testing have found support for its use as a measure in the assessment of sexual interest in children (McPhail et al., 2019). Phallometry has shown adequate-to-moderate internal consistency or test–retest reliability (Harris & Rice, 1996; Seto, 2018). There is good evidence for face (Merdian & Jones, 2011), convergent (e.g., Laws, 2009; McPhail et al., 2019), divergent (e.g., Seto, 2008, 2018), and predictive validity (Hanson & Morton-Bourgon, 2005; McPhail et al., 2019). Given variation in phallometric procedures across labs, it is crucial that clinics establish the psychometric properties of their phallometric procedures. The phallometric procedure used in this study has demonstrated concurrent and divergent validity in past studies and has good sensitivity and specificity (e.g., Blanchard et al., 2001; Cantor & McPhail, 2015); its reliability has not been assessed, but reliability constrains validity so previous research suggests at least acceptable reliability.

The Revised Screening Scale for Pedophilic Interests

Information about sexual victim age, gender, and relatedness as well as child pornography offending was contained in the archival database (Seto, Stephens, et al., 2017). Victim characteristic information was solely based on self-report for the undetected group, whereas victim characteristics were recorded from self-report and file information for detected individuals. When there was a discrepancy between self-report and file information, the information that led to the higher victim count was entered into the database; for example, if the individual denied official information that stated he had two victims, the information for the two victims was entered.

Sexual victim and child pornography information comprised the five items of The Revised Screening Scale for Pedophilic Interests (SSPI-2): boy victim below the age of 15, multiple child victims below the age of 15, victim below the age of 11, extrafamilial victim below the age of 15, and child pornography offending. Each item was scored as present (1) or absent (0). A total score was calculated for each individual that could range from 0 to 5, with higher scores more indicative of sexual interest in children. Although originally developed as a measure of pedophilia, the SSPI-2 might be better conceptualized as a measure of pedohebephilia (Stephens et al., 2019). Inter-rater reliability of the SSPI-2 is in the acceptable range, and there is evidence for convergent, divergent, and predictive validity (e.g., Seto, Sandler, & Freeman, 2017).

Procedure

The data used in this study came from men who consented to the use of their clinical assessment data for research purposes; however, we do not know the number of individuals who refused this consent. Briefly, men participated in a brief semi-structured interview, review of file information, and phallometric assessment during their initial assessment. The data from the assessment were collected by an evaluator and entered into a database by a research technician. The semi-structured interview involved gathering a range of information such as, demographic information, self-report information about paraphilias, sexological information (e.g., number of sexual partners), and self-reported offending history. The semi-structured interview and review of file information were completed prior to phallometric testing. Phallometric testing was completed in a private room. The individual was seated in a comfortable chair with a volumetric apparatus that recorded penile blood volume change to stimuli presented on screens in front of him and audio-recorded stories presented through headphones. All information from the assessment was recorded on a sheet for input into the archival database.

The use of clinical assessment data for research purposes was approved by the institutional research ethics board. This archival database has been used in previous studies examining different research questions about the assessment of sexual interest in children and the validity of various assessment measures used at the clinic (Barbaree, Blanchard, & Langton, 2003; Blanchard 2011; Blanchard & Barbaree, 2005; Blanchard, Kuban, et al., 2009; Blanchard, Lykins, et al., 2009; Blanchard et al., 2001; 2006, 2007, 2012; Cantor & McPhail, 2015; Lykins et al., 2010; Seto, Cantor, & Blanchard, 2006; Seto, Stephens, et al., 2017; Stephens, Leroux, Skilling, Cantor, & Seto, 2017; Stephens, Seto, Goodwill, & Cantor, 2017, 2018; Stephens et al., 2019). The database is not publicly available because it contains sensitive clinical data and there is a potential risk that participants could be identified by triangulation of data. All pertinent data from the data set are included in this study, and we report all the analyses conducted for this study.

Data Analysis

Only men who offended against children (defined as victims age 15 or younger, because the legal age of consent in Canada is 16) were included in these analyses. This criterion was used so that group differences could be attributed to detection status, rather than a difference in the likelihood of having child victims and therefore being suspected of having a sexual interest in children. Groups were compared using univariate analyses of variance (ANOVAs) and significant differences were followed with post hoc Bonferroni tests. To examine the association between detection status and individual SSPI-2 items, we utilized chi-square analyses and examined standardized residuals to see whether men who were undetected were more likely to have specific victim characteristics than their detected counterparts. Given the number of statistical comparisons, we adjusted our alpha to p < .001 and focus more on effect sizes in our interpretation and discussion of the results.

Results

Characteristics of Undetected and Detected Men

We first compared all undetected and detected men in the sample on referral status and sociodemographic variables (see Table 1). As expected, undetected men were more likely to be self-referred and less likely to be referred through the criminal justice system than expected. Undetected men were significantly younger than all detected groups, except for detected disclosers, at the time of their first sexual offense. Similarly, undetected men were younger at the age of their most recent sexual offense and at the time of the assessment compared with detected complete deniers. Finally, undetected men were less likely to be Black than expected.

Sexual Interest in Children

Undetected men endorsed a greater degree of sexual interest in children than all detected groups (see Table 2): detected complete deniers (p < .001, d = 2.92, 95% CI = [2.86, 2.98]), detected partial deniers (p < .001, d = 1.37, 95% CI = [1.23, 1.57]), and detected admitters (p < .001, d = 1.43, 95% CI = [1.34, 1.51]), except for detected disclosers (p = .004, d = 0.35, 95% CI = [0.16, 0.55]). The self-report findings were in contrast to the phallometric data, where there were no significant difference between undetected men and all detected groups: detected complete deniers (p =.038, d = 0.37, 95% CI = [0.30, 0.44]), detected partial deniers (p =1.00, d = 0.21, 95% CI = [0.12, 0.31]), detected admitters (p =.057, d = 0.35, 95% CI = [0.29, 0.41]), and detected disclosers (p =1.00, d = 0.00, 95% CI = [–0.12, 0.12]).

Table 2.

Sexual Interest in Children by Detection Status.

Sexual interest in children indicator	Total sample[n = 1,883]	Undetected[n = 69]	Detected full deniers[n = 516]	Detected partial deniers[n = 252]	Detected admitters[n = 789]	Detected disclosers[n = 207]	F-value	ES
Self-report	1.69 (1.31)[n = 1,823]	3.43 (1.62)[n = 68]	1.13 (0.60)[n = 511]	1.60 (1.25)[n = 251]	1.64 (1.22)[n = 786]	2.85 (1.66)[n = 207]	79.80***	.16
Phallometric responding	0.07 (0.86)[n = 1,833]	0.31 (1.11)[n = 69]	−0.01 (0.82)[n = 516]	0.12 (0.83)[n = 252]	0.01 (0.84)[n = 789]	0.31 (0.94)[n = 207]	7.73***	.02
SSPI-2 score	2.36 (1.21)[n = 1,833]	3.19 (1.03)[n = 69]	2.18 (1.12)[n = 516]	3.06 (0.99)[n = 252]	2.02 (1.19)[n = 789]	2.99 (1.12)[n = 207]	71.34***	.14

Note. The numbers represent the average values and the numbers in brackets are the standard deviations. The numbers in square brackets are the sample sizes and as can be seen, there were missing data for self-reported sexual interest, but not for phallometric responding or the SSPI-2. The effect size for the ANOVAs was partial eta-squared. Self-report was gathered during a clinical interview with a score of 1 suggesting no sexual interest in prepubescent or pubescent children and a score of 5 suggesting that their sexual interest in children was strongest. For phallometric responding, we calculated the pedohebephilia index as the maximum score to fully mature adults subtracted from the maximum score to prepubescent or pubescent children. The SSPI-2 score ranges from 0 to 5, with higher scores more indicative of sexual interest in children. ES = effect size; SSPI-2 = Revised Screening Scale for Pedophilic Interests; ANOVA = analysis of variance.

***

p < .001.

There were some differences on the SSPI-2. The undetected men had the highest SSPI-2 score; however, they only differed from the detected complete deniers (p < .001, d = 0.91, 95% CI = [0.82, 1.00]) and the detected admitters (p < .001, d = 0.99, 95% CI = [0.92, 1.07]). There was no significant difference between undetected men and two detected groups: detected partial deniers (p = 1.00, d = 0.13, 95% CI = [0.02, 0.24]) and detected disclosers (p = 1.00, d = 0.18, 95% CI = [0.05, 0.31]).

The results for individual SSPI-2 items are presented in Table 3. There were significant group differences for all SSPI-2 items. Men in the undetected group were more likely than expected to have boy victims and child pornography offenses. Interestingly, it was the detected disclosers who were more likely than expected to have positive scores for all SSPI-2 items, except for the presence of a child victim below the age of 11. Furthermore, the detected partial deniers were more likely than expected to have a boy victim, multiple child victims, and a victim below the age of 11.

Table 3.

Detection Status and SSPI-2 Items.

SSPI-2 item	Total sample[n = 1,883]	Undetected[n = 69]	Detected complete deniers[n = 516]	Detected partial deniers[n = 252]	Detected admitters[n = 789]	Detected disclosers[n = 207]	χ²	Cramer’s V
SSPI-1: boy victim	29%[n = 516]	44%(2.2)[n = 30]	27%(–1.1)[n = 137]	45%(4.8)[n = 114]	22%(–3.9)[n = 170]	40%(2.8)[n = 82]	73.29***	.20
SSPI-2: multiple child victim	51%[n = 927]	58%(0.9)[n = 40]	41%(–3.1)[n = 211]	84%(7.5)[n = 212]	38%(–5.2)[n = 296]	81%(6.2)[n = 168]	265.65***	.38
SSPI-3: victim below 11	67%[n = 1,228]	86%(1.9)[n = 59]	70%(0.8)[n = 360]	77%(2.0)[n = 195]	60%(–2.4)[n = 474]	68%(0.1)[n = 140]	41.90***	.15
SSPI-4: extrafamilial victim	79%[n = 1,453]	96%(1.5)[n = 66]	75%(–1.0)[n = 388]	89%(1.8)[n = 225]	74%(–1.7)[n = 582]	93%(2.2)[n = 192]	69.32***	.19
SSPI-5: child pornography	10%[n = 182]	36%(6.9)[n = 25]	5%(–3.4)[n = 27]	10%(0.2)[n = 26]	9%(–1.2)[n = 68]	17%(3.4)[n = 36]	80.55***	.21

Note. We report the percentage of men with that specific victim characteristic, with the numbers in square brackets representing the number of men with that specific characteristic and the number in round brackets representing the standardized residuals. The victim characteristics were represented by a dichotomous variable based on official file information and self-report for all but men who were undetected for which we only had self-report data. SSPI-2 = The Revised Screening Scale for Pedophilic Interests.

***

p < .001.

Supplementary Analyses

Given that self-referral status could confound the comparison of undetected and detected disclosers, we examined whether restricting the analyses to those who were self-referred influenced effect sizes, putting aside statistical significance given the smaller group numbers. When examining the results for sexual interest in children, the magnitude of the difference for self-report decreased in magnitude (d = 0.35 to d = 0.11) but increased in magnitude for the pedohebephilia index (d = 0.00 to d = 0.19). The magnitude of the effect was still small for the SSPI-2 (d = 0.18 to d = 0.22). The direction of the results remained the same for self-report; however, while there was previously no difference before between undetected and detected disclosures on the pedohebephilia index, the detected disclosers now exhibited a higher score than undetected men when the analyses were restricted to those who were self-referred. Similarly, the direction of the effect changed for SSPI-2 where the detected disclosers had a higher SSPI-2 score than the undetected men when the analyses were restricted to those who were self-referred.

When examining the SSPI-2 items, most of the results were in the same direction, except for the SSPI-2 boy victim and extrafamilial victim items. The effect sizes were similar in magnitude when the sample was restricted to men who were self-referred except for the presence of child pornography, which had an increase in effect size when analyses were restricted to those who were self-referred. These supplementary analyses suggest that even when restricting analyses to those who were self-referred, the comparisons for the undetected and detected discloser groups were relatively similar to when the results were not restricted (Table 4).

Table 4.

Detection Status and Victim Characteristics for Self-Referral Only.

SSPI-2 item	Undetected[n = 42]	Detected disclosers[n = 17]	Cramer’s V	Cramer’s V when not restricted to self-referral
SSPI-1: boy victim	41%(–0.2)[n = 17]	47%(0.3)[n = 8]	.06	.03
SSPI-2: multiple child victims	54%(–0.8)[n = 22]	88%(1.3)[n = 15]	.34	.23
SSPI-3: victim below 11	88%(0.4)[n = 37]	71%(–0.6)[n = 12]	.21	.17
SSPI-4: extrafamilial victim	98%(0.0)[n = 41]	100%(0.1)[n = 17]	.08	.05
SSPI-5: child pornography	45%(0.1)[n = 18]	41%(–0.2)[n = 7]	.04	.20

Note. We report the percentage of men with that specific victim characteristic for those who were self-referred. The numbers in square brackets represent the number of men with that specific characteristic, and the numbers in round brackets represent the standardized residuals. SSPI-2 = The Revised Screening Scale for Pedophilic Interests.

Discussion

We found that undetected men reported greater sexual interest in children than all detected groups. The effect size was largest for the comparison between undetected and detected complete deniers and decreased in magnitude, as detected men were more forthcoming about their offending history. Nonetheless, there were few statistically significant differences when examining objective measures of sexual interest in children. Furthermore, there were no differences when comparing undetected men with the detected disclosers on the different indicators of sexual interest in children.

Despite a greater willingness to admit sexual interest in children among men with undetected victims, there were no differences in phallometric responding based on detection status. These results are consistent with an older finding that suggested men who denied or accepted personal responsibility for offending against children differed in self-reported sexual interest, but not in phallometric responses (Baldwin & Roys, 1998). Furthermore, there were relatively few differences on the SSPI-2, although the undetected men had a higher SSPI-2 score than the detected complete deniers and detected admitters. Most importantly, across both objective measures, there was no significant difference between the undetected men and their closest comparison group, detected disclosers.

Based on the many similarities between undetected and detected disclosers, the degree of denial of offending explained group differences more than detection status in this study. These findings add to the literature on denial in sexual offending, as previous studies have primarily focused on categorizing different types of denial, examining the function of denial, and exploring the relationship between denial, recidivism, and treatability (e.g., Schneider & Wright, 2004; Yates, 2009). Given that most group differences observed in this study seemed more dependent on the degree of denial, our findings challenge those that suggest undetected men are particularly prolific in their offending (Asahina, 2010; Lisak & Miller, 2002).

An important caveat to our interpretation of the results for sexual interest in children is that several past studies examining differences between undetected and detected men were conducted in Germany, where there are strict laws that protect client–clinician confidentiality (e.g., Neutze et al., 2012). As a result, it is likely that detected men in those studies were also relatively forthcoming, given that they had presented for a comprehensive assessment of their own volition. In contrast, our study was conducted in a jurisdiction where mandatory reporting laws might inhibit some participants from admitting to officially undetected victims. Based on these important socio-legal differences, readers should be cautious when comparing our results to previous studies on detection status that have been conducted with men who were self-referred to the Dunkelfeld project (Kuhle et al., 2017; Neutze et al., 2012).

Reaching Undetected Populations

The findings suggest that undetected men do not likely differ in sexual interest in children compared with detected men in a North American sample, which has clinical implications. Most notably, this overarching conclusion is important to consider when delivering services to undetected men, as the development and delivery of comprehensive assessment and treatment for undetected offenders has been the focus of more research attention. Germany has been a leader in this regard, as it has a comprehensive program for individuals who identify as having a sexual interest in children who are not currently involved in the legal system (Beier et al., 2009). This approach could be useful, as a pilot evaluation found reductions in risk factors in the treatment group (Beier et al., 2015), although a subsequent re-analysis of the crucial interaction between treatment group and time produced non-significant results (Mokros & Banse, 2019). In addition, outcomes based on official offense records were not examined, so further evaluation is needed. This type of programming is more difficult to offer in North America because of differences in mandatory reporting legislation and the socio-political climate; however, a recent review of legislation, ethical guidelines, and case law concluded that it is possible to provide similar services in countries with mandatory reporting laws (McPhail et al., 2018). In addition, the development of anonymous online programs (e.g., https://troubled-desire.com/home.html; Stop it Now! https://www.stopitnow.org/) offers another avenue for those who are undetected to seek help anonymously.

As part of the development of services for undetected men, some additional discussion on the ways in which undetected and detected men may differ is warranted. Neutze and colleagues (2012) argued that there are two types of factors associated with detection status: preceding factors that allow undetected men to evade detection and resulting factors that may be responses to detection or reflect biases in self-report. Our results are likely most relevant to resulting factors that potentially reflect biases inherent in self-report measures, given the smaller effects on objective measures of sexual interest in children. It is important that researchers and clinicians distinguish between preceding and resulting factors. Research on preceding factors may aid in an improved understanding of how some men evade detection and highlight important targets in treatment for undetected men.

Limitations

One limitation in this study pertains to the variability between the groups in the reasons for presenting for assessment, which may have affected the veracity of participants’ self-report and the information available for analysis. Although we attempted to control for this by examining different groups of detected men, the limitation is still present. It is likely that at least some men who denied or minimized their officially recorded sexual offense history had additional unknown victims at the time of the assessment. This same limitation likely applies to the undetected or detected disclosers, as they may have elected to not disclose their entire offense histories either. In addition, we did not have inter-rater reliability data available to us given the way the data were entered into the archival database.

Another limitation pertains to our sample, which was mostly comprised of men referred by the legal system who had official victim information. In the dataset, we had a small proportion of men who were self-referred (4%), yet most of our undetected group came from this referral source. This means that for the undetected group, victim characteristic information was largely obtained through self-report, as there was no file information; however, many of the effects were similar or at least in the same direction when we restricted our sample to those who had only been self-referred. Similarly, there were relatively few men in the undetected group, compared with the detected groups, which was further reduced when we examined differences among individuals with child victims. It is also possible that the undetected group differ from those who are undetected and do not seek help. For example, it is possible that undetected men who self-refer have a higher degree of distress or sexual preoccupation compared with other undetected men, prompting their referral to a specialized clinical program. Finally, there are also important differences in studies on detection status due to differences in laws surrounding clinician–client confidentiality, which makes it difficult for us to generalize our findings to research conducted in a jurisdiction, such as Germany.

In addition, we do not know if the men who were undetected at the time of their assessment remained undetected after their assessment was completed. We expect that some men who were undetected were eventually charged, perhaps when past victims came forward or when they committed additional offenses that were reported and prosecuted. This is a significant consideration, as the passage of time could be a potential confound in the study of undetected men because some men may have felt more comfortable disclosing offenses that occurred long ago. For example, it is possible that sexual offenses that were perpetrated when the person was an adolescent were more likely to be reported than more recent offenses, and this could drive differences in SSPI-2 victim characteristics. This may in part explain why men in the undetected and detected discloser groups were significantly younger at the onset of their offending. We had no way to control for this in our analyses as we did not have details about when the offenses were committed, just that they had been committed prior to the assessment.

Conclusion and Future Research

Much of what we know about detected men may generalize to those who are undetected in a North American sample, given that the degree of denial appears to be driving group differences in this study. It would be useful to determine if a similar result would be obtained on measures of antisociality, the other major risk dimension for sexual recidivism (Hanson & Bussiere, 1998; Hanson & Morton-Bourgon, 2005). We would hypothesize that undetected men would be less antisocial than those who are detected, as prosocial characteristics might enable them to evade detection and/or encourage them to disclose their offending during the assessment. Nonetheless, a previous study suggests that undetected men may have a more significant criminal history for non-sexual offenses (Asahina, 2010), which may suggest that they would have a greater degree of antisociality. It would also be worthwhile to compare men who are undetected and detected on other dynamic risk factors, such as sexual preoccupation and intimacy deficits (e.g., Mann et al., 2010), while accounting for the degree of denial in the detected group.

As part of this research, it would be interesting to distinguish the impact of degree of detection (completely undetected, investigated but not charged, charged but not convicted, and convicted) on the results and to carefully control for potential confounds such as the passage of time between when the sexual offense was committed and when the individual was assessed. It is important to determine the risk of further sexual offending among undetected men and to determine if assessment and treatment approaches designed for detected men are applicable to undetected men. This line of research is important given the recent development of prevention programs that attempt to treat individuals who have committed undetected sexual offenses or are at risk of committing such offenses.

Footnotes

Acknowledgements

The authors would like to thank Dr. James M. Cantor for facilitating access to the data for this study.

Authors’ Note

The authors take responsibility for the integrity of the data, the accuracy of the data analyses, and have made every effort to avoid inflating statistically significant results. They were compliant with APA ethical standards in the treatment of human subjects.

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.

ORCID iD

Skye Stephens

Notes

References

Abel

Becker

J. V.

Mittleman

Cunningham-Rathner

Rouleau

J. L.

Murphy

W. D.

(1987). Self-reported sex crimes of non-incarcerated paraphiliacs. Journal of Interpersonal Violence, 2, 3-25.

Asahina

(2010). Japanese sexual offenders: A descriptive study. International Journal of Comparative and Applied Criminal Justice, 34, 351-365.

Baldwin

Roys

D. T.

(1998). Factors associated with denial in a sample of alleged adult sexual offenders. Sexual Abuse, 10, 211-226.

Barbaree

H. E.

Blanchard

Langton

C. M.

(2003). The development of sexual aggression through the life span: The effect of age on sexual arousal and recidivism among sex offenders. Annals of the New York Academy of Sciences, 989, 59-71.

Beier

K. M.

Ahlers

C. J.

Goecker

Neutze

Mundt

I. A.

Hupp

Schaefer

G. A.

(2009). Can pedophiles be reached for primary prevention of child sexual abuse? First results of the Berlin Prevention Project Dunkelfeld (PPD). The Journal of Forensic Psychiatry & Psychology, 20, 851-867.

Beier

K. M.

Grundmann

Kuhle

L. F.

Scherner

Konrad

Amelung

(2015). The German Dunkelfeld Project: A pilot study to prevent child sexual abuse and the use of child abusive images. The Journal of Sexual Medicine, 12, 529-542.

Blanchard

(2011). Misdiagnoses of pedohebephilia using victim count: A reply to Wollert and Cramer (2011). Archives of Sexual Behavior, 40, 1081-1088. https://doi.org/10.1007/s10508-0119835-0

Blanchard

Barbaree

H. E.

(2005). The strength of sexual arousal as a function of the age of the sex offender: Comparisons among pedophiles, hebephiles, and teleiophiles. Sexual Abuse, 17, 441-456. https://doi.org/10.1177/107906320501700407

Blanchard

Klassen

Dickey

Kuban

M. E.

Blak

(2001). Sensitivity and specificity of the phallometric test for pedophilia in nonadmitting sex offenders. Psychological Assessment, 13, 118-216. https://doi.org/10.1037/1040-3590.13.1.118

10.

Blanchard

Kolla

N. J.

Cantor

J. M.

Klassen

P. E.

Dickey

Kuban

M. E.

Blak

(2007). IQ, handedness, and pedophilia in adult male patients stratified by referral source. Sexual Abuse, 19, 285-309. https://doi.org/10.1177/107906320701900307

11.

Blanchard

Kuban

M. E.

Blak

Cantor

J. M.

Klassen

Dickey

(2006). Phallometric comparison of pedophilic interest in nonadmitting sexual offenders against stepdaughters, biological daughters, other biologically related girls, and unrelated girls. Sexual Abuse, 18, 1-14. https://doi.org/10.1007/s11194-0069000-9

12.

Blanchard

Kuban

M. E.

Blak

Cantor

J. M.

Klassen

P. E.

Dickey

(2009). Absolute versus relative ascertainment of pedophilia in men. Sexual Abuse, 21, 431-441. https://doi.org/10.1177/1079063209347906

13.

Blanchard

Kuban

M. E.

Blak

Klassen

P. E.

Dickey

Cantor

J. M.

(2012). Sexual attraction to others: A comparison of two models of alloerotic responding in men. Archives of Sexual Behavior, 41, 13-29. https://doi.org/10.1007/s10508-010-9675-3

14.

Blanchard

Lykins

A. D.

Wherrett

Kuban

M. E.

Cantor

J. M.

Blak

Dickey

Klassen

P. E.

(2009). Pedophilia, hebephilia, and the DSM-V. Archives of Sexual Behavior, 38, 335-350. https://doi.org/10.1007/s10508-008-9399-9

15.

Cantor

J. C.

McPhail

I. V. M.

(2015). Sensitivity and specificity of the phallometric test for hebephilia. The Journal of Sexual Medicine, 12, 1940-1950. https://doi.org/10.1111/jsm.12970

16.

Department of Justice. (2017). Justice Canada knowledge exchange on the criminal justice system’s response to sexual assault against adults—Summary of proceedings. http://www.justice.gc.ca/eng/cj-jp/victims-victimes/cal/summ-resu.html

17.

Hanson

R. K.

Bussiere

M. T.

(1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362.

18.

Hanson

R. K.

Morton-Bourgon

K. E.

(2005). The characteristic of persistent sexual offenders: A meta-analysis of recidivism studies. Journal of Consulting and Clinical Psychology, 73, 1154-1163.

19.

Harris

G. T.

Rice

M. E.

(1996). The science in phallometric measurement of male sexual interest. Current Directions in Psychological Science, 5, 156-160.

20.

Kuhle

L. F.

Schlinzig

Kaiser

Amelung

Konrad

Beier

K. M.

(2017). The association of sexual preference and dynamic risk factors with undetected child pornography offending. Journal of Sexual Aggression, 23, 3-18. https://doi.org/10.1080/13552600.2016.1201157

21.

Laws

R. D.

(2009). Penile phelthysmography: Strengths, limitation, innovations. In Thornton

Laws

R. D.

(Eds.), Cognitive approaches to the assessment of sexual interest in sexual offenders (pp. 7-29). John Wiley.

22.

Lisak

Miller

P. M.

(2002). Repeat rape and multiple offending among undetected rapists. Violence and Victims, 17, 73-84.

23.

Lykins

A. D.

Cantor

J. M.

Kuban

M. E.

Blak

Dickey

Klassen

P. E.

Blanchard

(2010). The relation between peak response magnitudes and agreement in diagnoses obtained from two different phallometric tests for pedophilia. Sexual Abuse, 22, 41-57. doi:10.1177/10790632-0352094

24.

Mann

Hanson

Thornton

(2010). Assessing risk for sexual recidivism: Some proposals on the nature of psychologically meaningful risk factors. Sexual Abuse: A Journal of Research and Treatment, 22, 191-217.

25.

McPhail

I. V.

Hermann

C. A.

Fernane

Fernandez

Y. M.

Nunes

K. L.

Cantor

J. M.

(2019). Validity of phallometric tests for sexual interests in children: A meta-analytic review. Assessment, 26, 535-551. https://doi.org/10.1177/10731911117706139

26.

McPhail

I. V. M.

Stephens

Heasman

(2018). Legal and ethical issues in treating clients with pedohebephilic interests. Canadian Psychology, 59, 369-381. https://doi.org/10.1037/cap0000157

27.

Merdian

H. L.

Jones

D. T.

(2011). Phallometric assessment of sexual arousal. In Boer

D. P.

Eher

Craig

L. A.

Miner

M. H.

Pfafflin

(Eds.), International perceptives on the assessment and treatment of sexual offenders (pp. 141-169). West Sussex, UK: Wiley & Sons.

28.

Mokros

Banse

(2019). The “Dunkelfeld” project for self-identified pedophiles: A reappraisal of its effectiveness. The Journal of Sexual Medicine, 16, 609-613. https://doi.org/10.1016/j.jsxm.2019.02.009

29.

Neutze

Grundmann

Scherner

Beier

K. M.

(2012). Undetected and detected child sexual abuse and child pornography offenders. International Journal of Law and Psychiatry, 35, 168-175.

30.

Neutze

Seto

M. C.

Schaefer

G. A.

Mundt

I. A.

Beier

K. M.

(2011). Predictors of child pornography offenses and child sexual abuse in a community sample of pedophiles and hebephiles. Sexual Abuse: A Journal of Research and Treatment, 23, 212-242.

31.

Perreault

Brennan

(2009). Criminal victimization in Canada, 2009 (Catalogue No. XIE85002, Vol. 30, No. 2). Canadian Centre for Justice Statistics.

32.

Schneider

S. L.

Wright

R. C.

(2004). Understanding denial in sexual offenders: A review of cognitive and motivational processes to avoid responsibility. Trauma, Violence, & Abuse, 5, 3-20. https://doi.org/10.1177/1524838003259320

33.

Seto

M. C.

(2018). Pedophilia and sexual offending against children: Theory, assessment, and intervention (2nd ed.). American Psychological Association.

34.

Seto

M. C.

Cantor

J. M.

Blanchard

(2006). Child pornography offenses are a valid diagnostic indicator of pedophilia. Journal of Abnormal Psychology, 115, 610-615. doi:10.1037/0021-843X.115.3.610

35.

Seto

M. C.

Sandler

J. C.

Freeman

N. J.

(2017). The Revised Screening Scale for Pedophilic Interests: Predictive and concurrent validity. Sexual Abuse: A Journal of Research and Treatment, 29, 636-657.

36.

Seto

M. C.

Stephens

Lalumière

M. L.

Cantor

J. M.

(2017). The Revised Screening Scale for Pedophilic Interests (SSPI-2): Development and criterion-related validation. Sexual Abuse, 29, 619-635. https://doi.org/10.1177/1079063215612444

37.

Stephens

Cantor

J. M.

Goodwill

A. M.

Seto

M. C.

(2017). Multiple indicators of sexual interest in prepubescent or pubescent children as predictors of sexual recidivism. Journal of Consulting and Clinical Psychology, 85, 585-595. doi:10.1037/ccp0000194

38.

Stephens

Leroux

Skilling

Cantor

J. M.

Seto

M. C.

(2017). Taxometric analyses of pedophilia utilizing self-report, behavioral, and sexual arousal indicators. Journal of Abnormal Psychology, 126, 1114-1119. https://doi.org/10.1037/abn0000291

39.

Stephens

Seto

M. C.

Cantor

J. M.

Lalumière

M. L.

(2019). The Revised Screening Scale for Pedophilic Interest (SSPI-2) may be a measure of pedohebephilia. The Journal of Sexual Medicine, 10, 1655-1663. https://doi.org/10.1016/j.jsxm.2019.07.015

40.

Stephens

Seto

M. C.

Goodwill

A. M.

Cantor

J. M.

(2017). Evidence of construct validity in the assessment of hebephilia. Archives of Sexual Behavior, 47, 301-309. https://doi.org/10.1007/s10508-016-0907-z

41.

Stephens

Seto

M. C.

Goodwill

A. M.

Cantor

J. M.

(2018). Age diversity among victims of hebephilic sexual offenders. Sexual Abuse, 30, 322-339. https://doi.org/10.1177/1079063216666837

42.

Yates

P. M.

(2009). Is sexual offender denial related to sex offense risk and recidivism? A review and treatment implications. Psychology, Crime & Laws, 15, 183-199. https://doi.org/10.1080/10683160802190905