Abstract
Understanding the difference between the proportion of sexual offenses that are officially detected versus the total number of sexual crimes that occur has been of interest to decision makers for some time. Previous studies have not considered possible racial differences. More aggressive responses by police and the criminal justice system as well as cultural differences in victim type/reporting rates may result in differences between detection rates of Black versus White men. In the current study (N = 190), we examined the difference between Black (n = 51) and White (n = 139) men committed to a forensic hospital as a Sexually Violent Person on factors that could impact the detected rate. Overall, we found that White men had significantly more total victims, undetected victims, undetected and detected child victims, total male victims, undetected male victims, undetected acquaintance victims, and total (detected and undetected) acquaintance victims than Black men. Black men had significantly more detected adult victims, but this difference did not remain when accounting for undetected victims. We explore reasons for these differences as well as provide recommendations for practice.
Introduction
Race may best be considered a social construct that sorts people into groups based on a perception of shared characteristics such as skin color (American Psychological Association, 2021). One’s race does not necessarily reveal anything about one’s ethnic or cultural identity, specific experiences and accomplishments, and future goals. Yet, race is often associated with attitudes, prejudices, and/or stereotypes against specific persons at either a conscious or unconscious level (e.g., implicit bias; Bergkamp et al., 2023). Professionals in both the mental health and criminal justice systems are not immune from holding such biases. Indeed, those having to make quick judgments in response to potential threats, such as police officers, rely on “gut feelings” to keep themselves safe (Kahneman, 2011). Unfortunately, this may lead to stereotyped judgments and actions, which can result in potential harm for Black urban communities such as over-policing, and a lower threshold for restraint/arrest (Fix et al., 2023). Indeed, Black pedestrians and motorists are stopped more frequently than those who are White, and this remains true when comparing shades of skin tone (Kavanaugh et al., 2023). Prosecutors and courts are apt to be susceptible to similar attitudes and beliefs when determining community risk during sentencing hearings, which may be why young Black males are more likely to receive longer prison sentences (Lehmann, 2023).
Differential responses to defendants based on race by those responsible for law enforcement and criminal judgment, even if not consciously deliberate, could impact the validity of risk assessments. For example, Ahmed et al. (2023) found the Static-99R, a sexual reoffense risk tool based on primarily historical factors, was less predictive for Indigenous individuals in Canada. The authors hypothesized this may be related to historical economic disadvantages due to the harmful effects of colonization, exposure to harmful risk factors, and over-policing, which are associated with higher rates of re-arrest and convictions (Ahmed et al., 2023). Within the United States, Black people also have historical and ongoing racial trauma and stress, generational economic disadvantages, exposure to harmful risk factors, over-policing, and an over-representation in the prison system (Jackson et al., 2023; Kavanaugh et al., 2023; Sreenivasan et al., 2022). Yet, the Static-99R appears to work similarly well for Black individuals as it does for White individuals (Ahmed et al., 2023; Lee et al., 2020). What does appear to differentiate the racial groups is that White individuals with a history of sexual offenses exhibit more paraphilic interests whereas Black individuals exhibit more general criminality (Lee et al., 2020). The exact reason for these racial differences is unknown, although this may be due to stereotypes leading to different detection rates of certain offenses.
A salient limitation of these studies, however, is that they only examine detected sexual offenses (i.e., formal charges and/or convictions for sexual offenses). There may be racial and cultural differences in whether certain sexual offenses are reported, witnesses and those who were victimized are cooperative, reports are found credible by professionals, and cases are processed by the judicial system (Ahmed & Helmus, 2023; Lee et al., 2020). For example, cultures that are more collectivistic and hierarchical such as East Indian societies may be less likely to report sexual offenses due to expectations of “saving face”, patriarchy, blaming those who are victimized, and distrust of law enforcement (Ahmed & Helmus, 2023). On the other hand, there may be real differences in types of sexual offenses and victim types between racial groups. When both detected and undetected offenses are accounted for, would this variation in victim typology and race continue to exist? In the current study we sought to examine whether there were differences between Black and White men regarding types of sexual offenses they committed, the characteristics of the identified victims, and whether there was a difference in the rate of victims resulting in formal sanctions versus those that went undetected.
Sexual Offense Detection Rates
Quantifying the rate of undetected sexual offending has been difficult. An “undetected sexual offense” is defined as an illegal sex act that did not result in an arrest or charge (Kelley et al., 2023). Often, studies rely on self-reported surveys and questionnaires, interviews of individuals admitting to sexual offenses (Abel et al., 1987), and national crime data (e.g., The U.S. Department of Justice; RAINN - Rape, Abuse & Incest National Network). These studies have generally found very low rates of arrests for sex crimes in comparison to reported sex crimes (Rape, Abuse & Incest National Network RAINN, 2023). Based on nationwide surveys of those experiencing sexual assault, for every 20 reports of sex crimes, only one arrest is made. The number of sexual crimes that result in convictions are even fewer (Frazier & Haney, 1996). Unfortunately, relying on detected sexual offenses can create challenges in risk assessment since the estimated risk probabilities associated with actuarial tools may be at best an under-estimation of the individual’s true sexual reoffense risk (Thornton et al., 2021). It may also limit accurate assessments of treatment needs if some types of sexual offenses are less likely to be reported/detected among certain racial and/or cultural groups.
The concept of undetected sexual offending has been researched for some time. There are multiple publications seeking the true rate of sexual offending with estimations in community samples ranging from 15 to 30 times higher than what is known in official reports while estimations in custodial samples range from 1.2 to 7 times higher than official reports (Abel et al., 1987; Bourke & Hernandez, 2009; DeLisi et al., 2016; Drury et al., 2020; Falshaw et al., 2003; Groth et al., 1982; Kelley et al., 2023; Koss et al., 1987; Lisak & Miller, 2002; Weinrott & Saylor, 1991). Remarkably, only one study (Lisak & Miller, 2002), examined differences between racial groups. They found no differences in the proportion of male White and Black college students who self-reported acts that depicted rape or attempted rape.
Kelley et al. (2023) examined the proportion of sexual offenses that led to an arrest, charge, or conviction within a high-risk sample who were ultimately committed under Wisconsin’s Sexually Violent Person (SVP) law. It was believed that the sample’s high non-sexual offending and sexual offending rates contributed to a high detection rate. Specifically, 22.6% of the reported sexual offense victims were detected at the time of the first formal sanction for a sexual offense, and this proportion significantly increased at the time of the second formal sanction (36.2%), although it did not continue to increase with subsequent sanctions. It appears that those who continued to sexually reoffend were caught rapidly and, as such, spent little time in the community before being arrested (or detained) again.
This raises questions when considering potential differences among racial groups. For example, if aggressive policing and longer criminal sentences are more likely to be experienced by Black men, and this is likely to begin during their adolescent years or earlier (Fix et al., 2023), might this increase their detection rates, decrease their time at risk between sanctions, and consequently decrease their undetected sexual offense victims? Or might we see an increase in only types of detected sexual offense victims who are more likely to report as they would not be constrained by cultural expectations (e.g., stranger victims vs. family members)? If there is a material difference among racial groups in the proportion of detected victims, would we be adding to biased practices by misattributing extrapolation formulas derived from a predominately White sample to Black men (Kelley et al., 2023)?
Race and Reporting
Factors affecting a person’s decision to report sexual violence include his or her own misperceptions about the sex offense dynamics, such as self-blame for the assault or not seeing it as a sexual assault, not reporting the crime sooner, the person’s fear of not being taken seriously or the person’s confidence that the case would go to trial, and the belief that it was a private matter that could be addressed without police involvement (Bachman, 1998; Cohn et al., 2013; Frazier & Haney, 1996; Tillman et al., 2010; Wilson & Scholes, 2013). Other factors included not knowing how to report, not wanting others to know (shame), and fear of retaliation by the individual who abused them (Bachman, 1998; Cohn et al., 2013; Tasca et al., 2013). A small qualitative study involving adult female victims of intimate partner violence/sexual violence in an urban community found that Black women reported being reluctant to call police due to fear of retaliation and strong social norms against contacting law enforcement for any reason resulting from historical race-based discrimination and consequences (Decker et al., 2019). Other reasons included fear of police, not wanting their partner incarcerated, and lack of confidence in the justice system or that they would be believed (Decker et al., 2019). Historically (approximately 30–40 years ago), the stereotype that Black women could not be raped or found “credible” pervaded the culture and, consequently, Black women were less likely to report sexual assaults (Wheeler & George, 2005).
According to Gomez (2023), cultural betrayal trauma theory (CBTT), the multi-level sociocultural context, and anti-Black racism oppress Black girls and women from reporting sexual abuse committed by Black boys or men. The broader racist culture creates a stronger bond and intra-cultural trust among Black communities. When a Black man or boy has sexually abused a Black woman or girl, the betrayal of this intra-cultural trust makes the harm within the group (already marginalized) more costly. The within-group trust and bond create pressure for the victims to “keep problems in-house” to protect the Black community from the dominant culture and system from further oppression. Reporting the abuse risks further rejection, alienation, or threats from within the community for the victim. The need of the victim is overshadowed by the needs of the offender and the entire marginalized group.
Those who were victimized appear more likely to report they have been raped when a weapon was used, if the attack occurred outdoors, degrading acts were inflicted, and if the attacker was Black and a stranger to the victim (Bachman, 1998). A study by Greenberg and Ruback (1992) demonstrated similar findings. They found that rape was more likely to be reported to the police when a weapon was used, if the attack occurred outdoors, degrading acts inflicted, the rapist was Black, and there was no prior relationship between the rapist and the victim. The identification of offending Black men being associated with a victim’s increased likelihood to report rape is an interesting finding given that the U.S. Department of Justice’s survey on individuals age 12 or older indicates there is a similar proportion of White victims of sexual assault (1.7%) as Black victims (1.4%), and it is more likely that victims were assaulted by individuals of the same race (Thompson & Tapp, 2022, 2023). Extensive research shows that those committing sexual offenses and those being victimized are far more likely to be of the same race (intra-racial) than of different races (inter-racial) (Gomez, 2023; U.S. Department of Justice, 2000; Wheeler & George, 2005).
Some research examining the variations on child sexual abuse disclosure by race and ethnicity has shown that (1) there are no differences in rates of childhood sexual abuse between White and Black populations (Kenny & McEachern, 2000; Wyatt, 1990), (2) Black children were typically over-represented in child sex abuse cases in 30 states (Luken et al., 2021), and (3) Black mothers are more supportive of their children’s disclosure of sexual abuse than White mothers (Kenny & McEachern, 2000; Pierce & Pierce, 1984). This suggests that Black children who were sexually abused may have higher reporting/detection rates than children of White families. At the same time, research also suggests that Black men are less likely to engage in sexual offenses against minors than White men and Black men are more likely to be prosecuted for criminal behaviors (Fraga Dominguez et al., 2018; Lehmann, 2023). Thus, it appears likely that undetected rates of sexual abuse may be higher for White children.
Policing
Police officers are usually the gatekeepers to the criminal justice system as they determine which cases get taken seriously enough to warrant an arrest and a referral for prosecution. Like others in the community, the police officer can have misperceptions and hold stereotypes of specific sexual offenses, which influence case processing decisions (Venema et al., 2021). For example, when a victim’s behavior may not be consistent with an officer’s stereotype or belief system, the officer is unlikely to believe the victim, minimize the reporting, and dismiss it as a false report (Darwinkel et al., 2013). Often, cultural beliefs about rape, rape myths, and stereotypes of those on the committing or receiving end of sexual abuse can affect the officer’s inclination to find a victim credible. Studies have shown that the officers’ assessment of credibility of the victim are based on (1) The victim’s relationship to the person committing the offense, (2) whether the victim is intoxicated, (3) when the victim delays reporting of the assault, and (4) if there was penetration and if weapons were used (Darwinkel et al., 2013; Frazier & Haney, 1996; Greenberg & Ruback, 1992; Lisak et al., 2010; Venema, 2016). Officers usually estimate that between 50% and 80% of sex crime reporting are false reports, even though the true rate of false reporting is less than 10% (Lisak et al., 2010; Schwartz, 2010). However, rates can differ by context. For example, custody dispute scenarios can yield more false allegations (36%–56%) (Bunting, 2008; Mikkelsen et al., 1992).
Nembhard and Robin (2021) showed that police officer’s behavior is often discretionary, and it can be impacted by implicit biases, which can cause officers to associate people with stereotypes based on identity characteristics rather than the actual behaviors. Research has offered mixed results regarding whether officers are more likely to pursue cases when the person being victimized is White versus Black (Spohn & Holleran, 2001; Spohn & Homey, 1993). However, using a large midwestern sample involving 23,525 sexual abuse cases reported to police between 1999 and 2014 by teenage and adult victims, Venema et al. (2021) found that White victims were found to be significantly more likely to decline to prosecute than Black and Hispanic victims, which led to these cases being closed by police despite having an identified suspect.
There is research on aggressive policing in urban communities that disproportionately affect people of color, which theoretically could lead to increased detection rates, but may contribute to problematic outcomes. Officers working in predominately Black communities have been found to be more likely to label Black adolescents with negative descriptors (Fix et al., 2023). Whether or not individuals are engaged in criminal behavior, Black individuals are more likely to have police contact, to be stopped by the police, and to be seen by police as less innocent than their White counterparts (Jackson et al., 2023). Negative experiences at the individual and community level are associated with an avoidance response to police, which can reduce victim reporting (Decker et al., 2019).
Racial Disparities in the Justice System
According to a publication by the American Civil Liberties Union (ACLU) and The Sentencing Project (July 14, 2022), Blacks and Latino Americans are incarcerated at rates five and three times higher than White residents respectively. One out of every 81 Black adults are in a U.S. prison, compared to one out of every 448 White adults. According to the Bureau of Justice Statistics, Black males were 5.7 times as likely to be imprisoned in 2020 as White males (Carson, 2021). Further, Black males aged 18 to 19 were 12.5% times likely to be imprisoned as White males the same age. The ACLU and The Sentencing Project report (July 14, 2022) also indicated the justice system in the United States disproportionately subjects racial and ethnic minorities to harsher sentences, including imprisonment without the possibility of parole and the death penalty. Lehmann (2023) examined felony sentencing cases in Florida between 1995 and 2006 and found that young Black males were disproportionately more likely to receive a prison sentence and a longer prison sentence than White males, especially when they chose to take their case to trial.
The consequence of Black defendants being more likely to be imprisoned and serve longer prison sentences will artificially suppress further offending by decreasing their time at risk in the community as compared to White defendants. There is little to indicate whether differences in sentencing decisions as a consequence of race occurs in sex offense cases. According to the FBI crime statistics by race (FBI, 2019), among incarcerated men in prison, the ratio of White men in prison for a sex offense (not including rape) was higher than for Black men. This tends to be consistent with data on racial differences for sexual offense type with White men having more sexual offenses involving children (Lee et al., 2020).
Racial Differences in Sexual Offense Trajectories
When an individual sexually reoffends, there are at least two broad factors associated with sexual recidivism—antisociality/criminality and sexual deviance (Brouillette-Alarie et al., 2016). Criminality is usually associated with anger, violence, impulsivity, irresponsibility, and callousness, whereas sexual deviance is associated with sexual preoccupation and paraphilic disorders. A study by Lee and colleagues (2020) showed that Black men were scored higher on criminality/antisociality on the Static-99R items, whereas White men were found to exhibit more indicators of paraphilic interests. Despite these differences, the Static-99R predicted equally well for both groups. However, this does not reflect undetected victims and the sociocultural factors that affect sex offense reporting that ultimately led to detection and possible sanctions. These factors are important to understand because the decision to arrest someone for a sexual crime is heavily influenced by the common stereotypes held by most individuals in the community about Black and White people (Darwinkel et al., 2013; Greenberg & Ruback, 1992; Nembhard & Robin, 2021). There is a paucity of empirical studies on why White men would be categorized as more paraphilic. Based on studies reviewed (Cromer & Freyd, 2007; Nembhard & Robin, 2021) showing that police behaviors are influenced by stereotypes, White men who commit sexual offenses are stereotyped as more paraphilic due to the media coverage of infamous psychopathic rape/murders (e.g., Jeffrey Dahmer, Ted Bundy) or news reports on pedophilic priests. One survey found that people were more likely to rank White men as perpetuating sexual assaults involving children (Livingston et al., 2024). Because of these implicit stereotypes, when children report sexual abuse, officers may be more likely to believe the children and arrest the suspect if the report fits the stereotypes of White males being more pedophilic.
Current Study
Overall, there appears to be sufficient evidence to raise concerns that racial differences in victim reporting rates, police practices, and sentencing decisions may cause differential proportions of sexual offense victim detection. The data that are available regarding racial differences regarding sexual offense motivation (antisociality vs. sexual deviance), diagnosis (pedophilic disorder), victim typology (age, gender, relationship to victim), and validity of the Static-99R has only been based on detected data. Given apparent and likely differences in reporting and detection rates between victim types and racial groups, the data may be a façade of what type of sexual offenses have really occurred. Examining these issues using both detected and undetected sexual offense victim data is one way to elucidate a clearer understanding of what is happening within these two racial groups. Such information is important for making more accurate diagnostic decisions and developing treatment plans that focus on the true areas of need. Lastly, the information will help evaluators better understand whether the detection to true offending ratio is substantially different enough between racial groups to justify different practices when extrapolating for undetected sexual offending in risk assessments and possibly reduce bias and stereotypes.
For the current study we wanted to answer the following research questions: 1. Is there a difference between the number of detected and undetected sexual offense victims reported by White and Black men committed as SVP? We did not have any specific hypothesis about this given that the available research is equivocal; White victims have been found to be less likely to pursue formal charges of sexual offenses while Black victims have been found to be less likely to report sexual offenses involving partner violence but more likely to report violent offenses by strangers (Decker et al., 2019; Venema et al., 2021). Further, the intersection between victim reporting, police response, and differential punishment between the races makes estimating the detection rate difficult. 2. Is there a difference between Black and White men regarding their age at first arrest for a sexual offense and whether they already were known to the criminal justice system due to general offending? We hypothesized that Black men would have an earlier introduction to the criminal justice system and be younger at the time of their first detected sexual offense due to increased risk factors and more aggressive policing styles in predominately urban, Black communities (Fix et al., 2023). 3. Do Black versus White men have less opportunities to sexually reoffend due to differences in punishment/sanctions? Given that young Black men appear to receive longer prison sentences (Lehmann, 2023), we hypothesized that Black men in our sample would have less total time in the community and more time on supervision. 4. Are there differences between White and Black men with regard to the type of detected and undetected victims they have sexually offended against (e.g., age, gender, & relationship to victim)? Given the results of Lee et al. (2020), we made the following hypotheses: a. White men will have more total victims that are associated with sexual deviance. This includes child victims (<12 years old), male victims, and victims with whom they cultivated relationships (i.e., family & acquaintance). Black men will have more total victims associated with general sexual violence. This includes adult female strangers. b. Pedophilic Disorder will be assigned to White men at a greater extent than Black men whereas Antisocial Personality Disorder to be more likely assigned to Black men. c. White men will have higher scores on the Persistence/Paraphilia items of the Static-99R whereas Black men will have higher scores on the General Criminality items of the Static-99R. 5. Do White and Black men differ in terms of their assessed risk for sexual reoffense? To answer this question, we compared their Static-99R scores after their first and last sexual offenses. We hypothesized that Black men would have higher Static-99R scores due to being more likely to incur general and sexual offense charges due to increased community-based risk factors and aggressive policing in urban communities.
Methods
Sample
This study was part of a larger archival study involving detected and undetected sexual offenses among adult males who had been civilly committed under Wisconsin’s Sexually Violent Person (SVP) law and placed at Sand Ridge Secure Treatment Center (SRSTC). The data were focused on men who were residing at SRSTC between 2001 and 2017. Per the SVP statute in Wisconsin, people are eligible for SVP commitment if they have been convicted of at least one “sexually violent offense”, which includes contact and attempted contact sexual offenses, has a predisposing mental disorder that makes it more likely for them to commit future acts of sexual violence, and has a current lifetime risk for sexual violence that is more likely than not (i.e., over a 50% probability for sexual violence at any time in their future lives). Of the original database, only cases identified as having racial designations of White (n = 139) or Black (n = 51) were included in the current study resulting in a total of 190 cases (all men). Of the 51 Black men, the vast majority were originally referred from a large urban city (Milwaukee). There were seven cases including four Black men and three White men who were never released back into the community following their first formal sanction for a sexual offense. These cases were excluded from analysis that was dependent upon individuals having an opportunity to return back to the community.
Of the 190 cases, they were diagnosed with having Pedophilic Disorder (49.5%), Other Specified Paraphilic Disorder with coercive features (13.7%), Other Specified Paraphilic Disorder with hebephilic features (10.0%), Sexual Sadism Disorder (14.2%), Antisocial Personality Disorder (55.3%), any substance use disorder (47.9%), any major mental illness (i.e., Bipolar Disorder Type I and psychotic spectrum disorders) (7.4%), and Intellectual Disability (7.9%). The sample was generally first detected of sexual offenses in their young adulthood (M = 20.15, SD = 6.1) and the majority (72.1%) had two releases back into the community following a formal sanction for a sexual offense. Far less of the sample were released back into the community on a third occasion (38.4%), and this frequency continued to decrease at the time of the fourth release (18.4%), fifth release (8.9%), sixth release (3.7%), and seventh release (2.1%). Only one case was released more than seven times following a sexual offense sanction. Not surprisingly given the select nature of the sample, this was a well above average risk sample as demonstrated by their Static-99R scores at the time they were last taken into custody (M = 6.54, SD = 1.84). The sample continuously spent time in a detention, jail, prison, or forensic psychiatric hospital setting immediately prior to being admitted to SRSTC. No contact with participants was made. Approval of the current study was obtained through the Institutional Review Board at the institution.
Measures
Static-99R
The Static-99R (Helmus et al., 2012) is an actuarial risk assessment measure designed for adult men with a history of sexual offense charges and/or convictions. The Static-99R is widely used in sexual reoffense risk assessment and has been shown to have good reliability and predictive validity (Helmus et al., 2022). There are ten items that make up the Static-99R total score. The total score can range from −3 to +12 and are organized into empirically derived risk classification levels. These include Level I (Very Low Risk), Level II (Below Average Risk), Level III (Average Risk), Level IVa (Above Average Risk), and Level IVb (Well Above Average Risk). In the current study, we identified the Static-99R total score at the time of their first release following a sexual offense sanction as well as their final total score following their last sanction for a sexual offense prior to being civilly committed. Additionally, we calculated two Static-99R subscales based on latent constructs found by Brouillette-Alarie et al. (2016) including “persistence/paraphilia” (i.e., prior sexual offenses, unrelated victims, male victim, & noncontact sexual offense) and “general criminality” (i.e., prior sentencing occasions & prior nonsexual violence).
Procedures
The database from which the current sample was extracted was coded using an archival record review of case files available at SRSTC. A detailed review of the coding procedures including the written detailed coding instructions and interrater reliability results can be found in Kelley et al. (2023). In short, all cases who were admitted to SRSTC, had completed a sexual history disclosure assignment, which was also validated by a truthful finding on polygraph testing, and had sufficient information in their available records regarding their offense histories were considered for inclusion in the study. The sexual history disclosure assignment involved a questionnaire that individuals formally had to routinely complete as part of treatment at SRSTC. The Sexual History Disclosure Questionnaire is an unpublished form developed by Anna Salter and Eric Holden for the Wisconsin Department of Corrections and adapted for use at SRSTC when the facility opened in 2001 (Kelley et al., 2023). In completing the questionnaire, men are asked to identify their detected and undetected contact sexual offenses involving adults and children, relationship to the victim, type of sexual act, and whether force was used. Additional records (i.e., most recent SVP report, original Department of Corrections SVP report, Pre-Sentence Investigation report, and criminal history database) were used to create a timeline for each case including non-sexual offenses, sexual offenses, sentence received, custody time, time in the community, and victim characteristics. The Static-99R was coded at the time of each release from a sexual offense arrest, charge, or conviction. When they were not released due to continued confinement, the Static-99R was coded at the time of their last formal sanction.
Only contact sexual offense victim characteristics were coded in the database, given this was the only type of undetected sexual offense that had this level of information. Although detected non-contact sexual offenses were coded and used for Static-99R scoring purposes, victim characteristics for such offenses were not coded. In order not to double count victims, detected victims were identified on the sexual history questionnaires and not included as undetected victims. In order to be consistent with the Static-99R coding rules, the individual had to be at least 12 years old to be considered as having a sexual offense “victim” that was either detected or undetected. Undetected victims included those who were victims of an illegal contact sexual offense that would have resulted in a formal sanction (arrest, charge, or conviction) had the individual been detected. Any “victims” that were incorrectly identified on the sexual history questionnaire (e.g., peer-aged, consenting intimate relationships within two years of each other) were not included. Sometimes, men had identified ongoing sexual assault of the same victim in the sexual history questionnaire that spanned across multiple release periods. When that occurred, the age and release period when the sexual offending began was coded.
Analytic Plan
Bivariate analyses examined independent effects of relevant variables on committed individuals’ race. Specifically, independent sample t-tests compared means between Black and White men on numerical variables (e.g., victim characteristics and Static99-R score) and Pearson chi-square tests were used to compare group frequencies on nominal variables (e.g., diagnoses). We reported the Welch’s t-test results when the variance between the two samples were found to be significantly different. To assess the independent influence of various victim characteristics while controlling for other victim characteristics, multivariate logistic regression analyses assessed associations between race of the committed individual and the variables of victim sex (male, female), victim age (child, teenager, adult), and victim relationship (stranger, acquaintance, family). Model fit measures are reported as recommended by Field (2009) with both Cox and Snell R2 and Nagelkereke’s R2, unstandardized beta (b), and Odds Ratios (OR) with 95% confidence intervals. To determine model stability and potential multivariate outliers for our sample, we also examined Cook’s distance and standardized residuals as recommended by Field (2009).
Univariate outliers were identified by visual inspection of histograms and boxplots, and by computing standardized scores to identify extreme values (e.g., z-scores > ±3.5). Identified outliers were assigned a score value one unit larger than the next most extreme score within that variable’s distribution (Tabachnick & Fidell, 2019). In this sample, five outliers were identified for undetected frequency count victim variables. Analyses were conducted in SPSS Version 28.
Results
In this sample (N = 190), there were 807 officially detected victims that led to one or more sanctions within the total sample (M = 4.25, SD = 2.10, Med = 4.00, Range = 1 – 13). There were no significant differences between Black (M = 4.37, SD = 2.07) and White (M = 4.20, SD = 2.11) men on the average number of detected victims (t (188) = 0.50, p = .620, d = .081). There were a total of 1786 undetected victims reported across the sanction periods (M = 9.40, SD = 12.41, Med = 4.0, Range = 0–69). White men had significantly more undetected victims (M = 10.58, SD = 13.10) than Black men (M = 6.18, SD = 9.68) according to a Welch’s t-test (t (120.12) = 2.51, p = .013, d = .359). This was a moderate effect size. When examining the total number of victims (detected plus undetected), White men had a significantly larger average number of victims (M = 10.66, SD = 16.64) than Black men (M = 7.06, SD = 5.04) according to a Welch’s t-test (t (183.92) = 2.28, p = .024, d = .249).
When only considering cases who had one or more releases after the first sanction (n = 183), there were a total of 1815 victims including 781 officially detected and 1034 undetected victims. This suggests that 43.0% of the victims were detected in the sample when all the release periods are combined. However, the detection rate differed between races with 59.9% of the victims of Black men being detected compared with 39.1% of the victims of White men.
There were no significant differences between White (M = 20.53, SD = 6.38) and Black men (M = 19.12, SD = 5.10) at the age of arrest for their first sexual offense (t (188) = 1.42, p = .156, d = .233). However, there was a significant association between race and prior nonsexual conviction (χ 2 (1) = 4.04, p = .044, Cramer’s V = .168). For Black men, 64.7% had a prior adult non-sexual conviction. In contrast, only 45% of White men had a prior nonsexual adult conviction. Of those who had at least one release following first sanction (n = 183), there were no significant differences between Black (M = 1936.47, SD = 1783.69) and White (M = 2046.76, SD = 1853.72) men for days spent in the community between sanction periods (t (181) = 0.36, p = .723, d = .060). There was also no significant difference between Black (M = 1265.43, SD = 1238.80) and White (M = 1003.86, SD = 728.04) men on days spent under supervision between sanction periods according to a Welch’s t-test (t (57.36) = 1.37, p = .177, d = .295).
Age, Sex, and Relationship Victim Characteristics
Frequencies of Victim Type by Race of the Offender.
There were no significant differences between Black and White men in terms of female victim count (t (188) = 0.72, p = .471, d = .118). However, there was a significant difference for male victim count, according to a Welch’s t-test (t (176.76) = 2.55, p = .012, d = .308). White men (M = 2.75, SD = 5.58) had more male victims than Black men (M = 1.22, SD = 2.63). There were no significant differences between White and Black men for detected female victims (t (188) = 1.53, p = 1.27, d = .251).), undetected female victims (t (181) = 1.11, p = 2.67, d = 1.88), or detected male victims (t (188) = 1.29, p = .197, d = .212). However, there was a significant difference for undetected male victims per a Welch’s t-test, with White men having significantly more undetected male victims (M = 1.59, SD = 4.53) than Black men (M = 0.45, SD = 1.08) (t (169.61) = 2.73, p = .007, d = .289).
For relationship to the victim status, there were no significant differences between White and Black men for total stranger victims (t (188) = 0.67, p = .503, d = .110), undetected stranger victims (t (180) = 0.07, p = .95, d = .012), or detected stranger victim (Welch’s t-test, t (71.20) = 1.90, p = .061, d = .357). There was a significant difference between White and Black men for total acquaintance victims according to Welch’s t-test, (t (178.47) = 3.41, p < .001, d = .410), with White men having significantly more total acquaintance victims (M = 5.92, SD = 6.57) than Black men (M = 3.53, SD = 3.04). This was also true for undetected acquaintance victims according to Welch’s t-test. White men had significantly more undetected acquaintance victims (M = 3.56, SD = 5.89) than Black men (M = 1.34, SD = 2.22) (t (178.93) = 3.69, p < .001, d = .427). However, there were no significant differences between White and Black men for detected acquaintance victims (t (188) = 0.54, p = .590, d = .088). There was no significant difference between White and Black men for total family victims (t (188) = 1.04, p = .300, d = .170), undetected family victims (t (181) = 0.49, p = .624, d = .083), or detected family victims (t (188) = 0.90, p = .372, d = .147).
Diagnostic Categories
Diagnostic Frequencies by Race of the Offender.
Static-99R Scores
For Static-99R scores, none of the patients had scores in the Level 1: Very Low Risk range and only two individuals had scores in the Level II: Below Average Risk range. At the time of the first release from an arrest for a sexual offense, 47% of patients had scores in the Average Risk range (Level III) and 44% had scores in the Above Average Risk range (Level IVa). Only 8% had scores in the Well Above Average Risk range (Level IVb). At the time of their final incarceration before SVP commitment, the score distribution was 4%, 20%, and 75% respectively for the Average (Level III), Above Average (Level IVa), and Well Above Average (Level IVb) risk ranges respectively.
Black men had significantly higher Static-99R total scores than White men at both first release (Black: M = 4.12, SD = 1.38; White: M = 3.51, SD = 1.21; t (186) = 2.95, p = .002, d = .488) and final incarceration (Black: M = 7.02, SD = 1.67; White: M = 6.34, SD = 1.91; t (188) = 2.40, p = .017, d = .393). When considering the two latent constructs of the Static-99R (Persistence/Paraphilia and General Criminality), White men (M = 1.21, SD = 0.63) had significant higher scores on the persistent paraphilia items at first release according to Welch’s t-test, (t (102.04) = 2.26, p = .026, d = .345) than Black men (M = 1.00, SD = 0.53). Black men (M = 0.58, SD = 0.76) had significantly higher scores on the general criminality items at first release per Welch’s t-test, (t (65.67) = 3.14, p = .003, d = .620) than White men (M = 0.22, SD = 0.51). At last arrest, Black men (M = 1.41, SD = 0.80) still had significantly higher general criminality scores than White men (M = 0.81, SD = 0.81; t (188) = 4.59, p < .001, d = .752); however, there was no longer a significant difference between Black (M = 3.59, SD = 1.24) and White (M = 3.71, SD = 1.16) men on the persistent paraphilia scores (t (188) = 0.64, p = .523, d = .105).
Multivariate Logistic Regression Predicting Race of the Offender.
Notes. R2 = .125 (Cox & Snell), .182 (Nagelkerke), Model χ 2 (8) = 25.22, p = .001.
b = unstandardized beta, CI = Confidence interval.
Discussion
The current paper sought to compare the following between Black and White men who had a history of sexual offenses and were ultimately committed as SVP: (1) the victim detection rate; (2) the presence of non-sexual convictions prior to first detected sexual offense as well as age of first detected sexual offense; (3) time at risk in the community and length of time on community supervision; (4) indicators of sexual deviance including victim type, paraphilia diagnosis, and Static-99R items reflective of persistence/paraphilia; and (5) degree of overall risk for sexual re-offense as measured by the Static-99R total score. Existing research points to racial differences in victim reporting rates, policing, and court decisions that may cause differential proportions of sexual offense victim detection (Bachman, 1998; Decker et al., 2019; Jackson et al., 2023; Kavanaugh et al., 2023; Lehmann, 2023; Venema et al., 2021). Material differences among racial groups in the proportion of detected victims and types of victims detected may lead to misattributing inaccurate diagnoses or applying extrapolation formulas to account for undetected sexual offending derived from predominately White samples (Kelley et al., 2023).
We found that Black men in the sample had a higher proportion of victims who were officially detected by law enforcement than White men. In fact, about 60% of victims of Black men were detected compared to only about 39% of the victims of White men. This is a striking difference in the victim detection rate. Although we could not fully identify why Black men appear to be more frequently detected for sexual offenses, it appears that their detected victims were more likely to be adult women who likely experienced physical force and/or violence given the higher proportion of Sexual Sadism and Antisocial Personality Disorder diagnoses in this group. Based on the victim reporting literature, victims are more likely to report having been raped when the offender was a Black male who used a weapon or degrading acts (Bachman, 1998; Greenberg & Ruback, 1992). Thus, the victim type for Black men in our sample likely increased their detection rate. Additionally, since more Black men than White men were already involved in the justice system due to non-sexual offending, they may have been more easily identified in the commission of subsequent crimes. We do not have any information on the samples’ experience of policing in their community and whether, and in what way, this may have impacted offending and detection rates. The majority of the Black men in this sample come from Milwaukee, which is an urban city with a history of limited residential opportunities for Black citizens and with Black neighborhoods identified by unemployment, poverty, and segregation (Wisconsin Historical Society, 2024). Milwaukee continues to be one of the most segregated cities in the United States (Wisconsin Historical Society, 2024). Such environments are associated with increased crime and more aggressive policing (Jackson et al., 2023; Kavanaugh et al., 2023; Sreenivasan et al., 2022). Both White and Black men had similar time at risk in the community and neither experienced more time on community supervision suggesting that the difference in the detection rate is not specifically due to differences in time at risk.
The results from our study provide validation for the findings offered by Lee et al. (2020). Specifically, White males showed more pedophilic interests and behaviors than Black males. White men had significantly more total (detected and undetected) child victims than Black men, more total male victims and undetected male victims, more total acquaintance victims, a higher proportion of Pedophilic Disorder diagnoses, and higher scores on the Static-99R items indicating Persistence/Paraphilia at their first release. Also consistent with the results in Lee et al. (2020), Black males had significantly higher scores on the General Criminality items on the Static-99R and were more likely to be diagnosed with Antisocial Personality Disorder. However, we also found that a higher proportion of Black men were diagnosed with Sexual Sadism Disorder as compared to White men. Lee et al. (2020) found similar proportions of Sexual Sadism diagnosed in Black and White men, but they found Black men were significantly more likely to use force, physical violence, and threats during the index sexual offense. We were unable to examine the accuracy of the Sexual Sadism Disorder diagnoses in our sample because level of force, violence, and an objective measure of sadistic paraphilic interests was not coded in the dataset. While Doren and Elwood (2009) demonstrated high levels of agreement between SVP evaluators when diagnosing Sexual Sadism Disorder, they did not specifically examine the effect of racial differences within the vignettes. There appears to be a paucity of studies examining whether racial characteristics may affect diagnostic differences regarding Sexual Sadism Disorder, although racial bias has been found to impact diagnoses such as Conduct Disorder, Antisocial Personality Disorder, and others (Garb, 2021). Sreenivasan and colleagues (2022) note racial bias regarding over diagnosing mental disorders and this increases perceptions of risk by laypeople and courts. Overall, this raises an interesting question about whether the force and violence seen in Black offenders best represent general criminality rather than sexual sadism.
The finding that Black men had a higher Static-99R total score at their release from first sanction for a sexual offense should not be surprising given that they also were more likely to have a non-sexual offense conviction prior to their first detected sexual offense. The fact that they continue to have higher total Static-99R scores at their final release/sanction prior to SVP commitment could suggest they present at higher risk due to a higher proportion of detected sexual offenses and non-sexual offense convictions. This makes sense when considering that generational economic disadvantages, ongoing racial trauma and stress, and exposure to harmful risk factors are associated with increased risky behavior (Ahmed et al., 2023; Jackson et al., 2023; Kavanaugh et al., 2023). Indeed, Black men continued to have significantly higher general criminality scores on the Static-99R at their last arrest, but this may either reflect a higher rate of non-sexual offending or may actually reflect a higher rate of detected non-sexual offenses than their White counterparts. When accounting for both total number of detected and undetected victims of sexual offenses, White men have significantly more victims, which appear to be due to the high number of undetected victims. The Static-99R has been shown to have sufficient predictive validity with Black men, which makes sense given that the majority of their sexual offenses are detected (Ahmed et al., 2023). Although Black men might appear “riskier” on actuarial measures like the Static-99R, this may be due to their increased likelihood to be detected for sexual offenses, and to be punished for such, which is in contrast to White men who have a lower detection rate. Indeed, it appears that White men are similarly “risky” but their Static-99R scores likely is an underestimate of this risk due to the high number of undetected victims. Accounting for undetected sexual offending in risk assessment with White men may help to balance this racial inequity.
Although we had hypothesized that Black men would have less time in the community and more time on supervision, thereby suppressing reoffending rates, the current results indicated there were no difference between racial groups. It may be that with an especially high-risk sexual offending sample, any differences in prosecution or sentencing decisions due to race disappears. Further, regardless of sentencing, the sample spent little time in the community before reoffending.
Implications for Practice
Evaluators, attorneys, courts, and other stakeholders who need to make empirically informed decisions about which individuals meet criteria for SVP commitment may best be aided by the recommendations put forth by Helmus et al. (2022). Specifically, evaluations that could result in such high stakes (i.e., deprivation of life and liberty through civil commitment) need to be comprehensive. This includes structured assessments of both static and dynamic measures of risk. However, these tools have limitations since the risk estimates they provide are solely based on officially detected sexual offenses and may under-estimate an individual’s actual risk for reoffense (Thornton et al., 2021).
Kelley et al. (2023) offered a preliminary method for accounting for undetected sexual offending. Specifically, the authors adapted a model put forth by Hanson et al. (2003) in which evaluators can obtain the “true” recidivism rate by dividing the risk probability derived from actuarial measures by the individual’s detection rate. Evaluators can obtain the individual’s detection rate by applying the detection rate for each victim for the median number of victims the individuals in the sample have sexually offended against. Across all the individuals with at least one release back into the community (n = 189), Kelley et al. (2023) found a victim detection rate of 36% and the sample having a median of five victims and offending about once per year. This results in an individual detection rate of 89%. That is, if a sample of 100 sexually offending men had one victim per year for five years, and a 36% chance each victim would be detected, 89% of the sexually offending men would have been caught after five years. 1
The current results indicate that using the method offered by Kelley et al. (2023) continues to be appropriate for White men being evaluated for SVP commitment. However, caution needs to be used when applying the detection rate to Black men. The victim detection rate for Black men in our sample (60%) was notably higher than for White men (39%). If evaluators apply the detection rate offered by Kelley et al. (2023) to evaluations involving Black men, they may grossly over-estimate the sexual re-offense risk rate. For example, imagine we found that the combination of the Static-99R and a dynamic risk tool suggested a future five-year risk probability of 25%. The difference between using a detection rate for a White majority sample (36%) versus a Black sample (60%) would result in risk estimates being 28.1% versus 25.3%. This is because 99% of Black individuals would be detected after five victims when the victim detection rate is so high. We believe that extrapolating for undetected offending will not be necessary when evaluating Black men for SVP assessments, especially for those who predominately have adult female victims and exhibit general criminality. Specifically, the victim detection rate for Black individuals without a pedophilic disorder diagnosis was notably higher (68.0%) than for Black individuals with a pedophilic disorder diagnosis (45.5%). However, these detection rates were both higher than the detection rate for White individuals. Although we had a very small group of Black men in our sample (n = 51), four of whom were never released back into the community following their first sexual offense sanction, this group had a total of 342 sexual offense victims of which 205 resulted in official sanctions. However, we concur that further research with larger sample sizes is needed to better clarify the nature of our results.
Limitations and Future Research Directions
In addition to the small number of Black men in our sample, the dataset did not include specific information that would have been helpful to understanding our results. That is, the level of force or violence used in past sexual offense was not coded, which could have helped to validate the diagnoses of Sexual Sadism Disorder in our sample. Research examining the diagnostic reliability of Sexual Sadism Disorder and other paraphilic disorders among different racial groups may be important in understanding how evaluators differentiate force and violence that may stem from motivations related to general criminality, paraphilia, or both. Further, there is a paucity of research directed at understanding why pedophilic interests and behaviors occur more frequently among White men. Our current dataset did not have variables to directly explore this issue, and we have not been persuaded by current hypotheses on the issue (e.g., Lee et al., 2020). We still do not know if race (and culture) is related to specific types of sexual deviance/paraphilia or whether the cultural influences in victim reporting and rape myths lead to differences in detected sex offenses, thus influencing diagnosis and risk decisions. While some authors have discussed social oppression, discrimination, and pressures that Black men may experience in the culture are related to sexual violence against women (Goldsmith et al., 2005), we still do not know what etiological, psychological, or cultural factors may relate to why White men show higher rates of pedophilic interests.
The dataset was limited to number of contact sexual offenses, which may not provide the full explanation for why some individuals are more easily detected than others. Types of non-contact offending like exhibitionistic offenses may be more likely to result in detection since it is a high frequency offense behavior. Further, the data did not count number of offenses to each victim nor did the data include information on obtaining victim compliance, which could each affect the detection rate. For example, a White man may sexually offend against a teenage acquaintance once. Or he may sexually offend against a teenage acquaintance multiple times. One could imagine that each sexual offense towards the same victim would increase the detection rate. But what if he had cultivated a relationship involving grooming with this victim whereas he did not have such a relationship with the victim he only offended against once? This would likely shift the detection rate.
Our results were based on a highly select sample of those who were ultimately committed as SVPs in Wisconsin. This creates multiple limitations in the analysis of our data. First, the sample is homogenous in that they all continued to sexually reoffend until their final commitment day. There is no comparison with individuals who avoided detection or naturally desisted. Second, we have no information regarding their detection rate after release from SVP commitment. Third, the sample have all been diagnosed with some form of mental disorder that was opined to predispose them to sexual offending and so the sample likely has a higher frequency of paraphilias than what would be found in a routine community setting. Fourth, we had no way to capture undetected general criminal offending between the two races. Lastly, the results would not be generalizable to routine, community samples who likely have different rates of offending and detection. Ideally, researchers with access to similar detected and undetected offense data in community samples would enact similar research projects to see if the differential detection rate between races can be replicated.
Footnotes
Author’s Note
Rachel E. Kahn is now at St. Mary’s Hospital, Madison, Wisconsin, USA. The findings and conclusions expressed are those of the authors and do not necessarily represent the views or opinions of the Wisconsin Department of Health Services – Division of Care and Treatment Services or the California Department of State Hospitals.
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.
