Do static risk factors predict differently for aboriginal sex offenders? A multi-site comparison using the original and revised Static-99 and Static-2002 Scales.

• Author: Babchishin, Kelly M.
• Position: Canada

The over-representation of Aboriginal offenders within the Canadian criminal justice system continues to cause serious concerns, despite attempts by the federal government to lessen this disparity (Criminal Code, s. 718.2(e); R v Gladue 1999). Although Aboriginals represent 4% of the Canadian adult male population (Statistics Canada 2008) they disproportionately represent 19% of incarcerated men in Canada (Public Safety Canada 2009) and 18% of men incarcerated for sex offences (Correctional Service Canada 2008). Despite the continued over-representation of Aboriginal offenders within the criminal justice system, the effective management of Aboriginal offenders is made difficult due to a general lack of knowledge regarding the similarities and differences between Aboriginal and non-Aboriginal offenders, in general, and within specific offender populations, such as sex offenders.

Research has mainly focused on comparing and contrasting characteristics of Aboriginal and non-Aboriginal offender populations. It has been consistently reported that Aboriginal offenders exhibit more risk factors compared to non-Aboriginal offenders. For example, Aboriginal offenders are younger and less educated (Statistics Canada 2006), have lengthier criminal histories (Dell and Boe 2000; Holsinger, Lowenkamp, and Latessa 2003), commit more violent crimes (Trevethan, Moore, and Rastin 2002), and are more likely to be classified as higher-risk and higher-need (Rugge 2006; Statistics Canada 2006) compared to non-Aboriginal offenders. In addition, Aboriginal offenders are more likely to experience poverty and family dysfunction in childhood compared to non-Aboriginal offenders (Johnston 1997; Trevethan, Auger, Moore, MacDonald, and Sinclair 2002).

In terms of predicting the risk of recidivism, the majority of Andrews and Bonta's (2010) central 8 risk factors (i.e., antisocial attitudes, antisocial peers, history of antisocial behaviour, antisocial personality, and substance abuse) predict recidivism equally well for both groups (B.C. Public Safety and Solicitor General 2004; Bonta 1989; Bonta, LaPrairie, and Wallace-Capretta 1997). Similarly positive findings have been reported for the predictive validity of risk scales designed to assess risk of future criminal behaviour. The Level of Service Inventory (LSI) and its subsequent adaptations (e.g., LSI-R, LSI-OR, and YLS/CMI), for example, have been shown to predict recidivism equally well for Aboriginal and non-Aboriginal offenders (Bonta 1989; Brews, Wormith, and Guzzo 2009; Gossner and Wormith 2007; Gross and Sroga 2008; Tanasichuk, Wormith, and Guzzo 2009).

From these results, it appears that, within general offender populations, established risk factors and risk scales are equally predictive of recidivism for Aboriginal and non-Aboriginal offenders. It is important to note, however, that this review is unique to Canadian Aboriginal offender samples. It is possible that these findings may not replicate in Aboriginal samples outside of Canada. In fact, the LSI-R did not predict recidivism for Aboriginal offenders in the United States (AUC = 0.61, 95% CI = 0.49 to 0.71, n = 100), albeit the predictive accuracy of the LSI-R was similarly low for the non-Aboriginal offenders in this sample (AUC = 0.63, 95% CI = 0.53 to 0.71, n = 162; Holsinger, Lowenkamp, and Latessa 2006).

Although there is some research on general Aboriginal offenders, much less has been conducted on Aboriginal sex offenders. The available studies have identified similarities between Aboriginal general offenders and Aboriginal sex offenders. First, like Aboriginal non-sexual offenders, Aboriginal sex offenders score higher on a large number of risk factors compared to their non-Aboriginal counterparts. For example, Aboriginal sex offenders are significantly younger and have lengthier criminal histories compared to non-Aboriginal sex offenders (Ellerby and MacPherson 2002; Olver and Wong 2006; Rastin and Johnson 2002). Also, consistent with the literature on general Aboriginal offenders (e.g., Statistics Canada 2006), Aboriginal sex offenders are more likely to have lower educational achievement, higher rates of unemployment, and a history of substance abuse (Ellerby and MacPherson 2002; Olver and Wong 2006).

When looking at risk factors specific to sexual offending, however, Aboriginal sex offenders actually exhibit fewer risk factors than non-Aboriginal sex offenders. For example, Aboriginal sex offenders have fewer sexually deviant interests (e.g., violent sexual fantasies) and paraphilias (e.g., exhibitionism) compared to non-Aboriginal sex offenders (Ellerby and MacPherson 2002) and are less to likely to have male victims (Ellerby and MacPherson 2002; Nahanee 1996; Rastin and Johnson 2002; Rojas and Gretton 2007). Another notable difference is that Aboriginal sex offenders are less likely to target younger victims than non-Aboriginal sex offenders (Ellerby and MacPherson 2002).

In sum, the available studies have found key differences between Aboriginal and non-Aboriginal sex offenders as to the presence of criminogenic factors associated with both general and sexual recidivism.

On the one hand, Aboriginal sex offenders have more risk factors related to general antisociality and criminality, such as having lengthier criminal records, being unemployed, and having substance abuse problems. On the other hand, they also have fewer risk factors related to sexual recidivism, such as deviant sexual interests. In addition, Aboriginal sex offenders show higher rates of sexual recidivism (Rastin and Johnson 2002; Rojas and Gretton 2007; Williams, Vallee, and Staubi 1997), violent recidivism (Rojas and Gretton 2007), and general recidivism (Rastin and Johnson 2002; Rojas and Gretton 2007) compared to non-Aboriginal sex offenders. It is unclear, however, whether these observed differences in recidivism rates are due simply to Aboriginals being higher risk and, consequently, would disappear after accounting for differences in risk level.

For evaluators involved in the risk assessment of Aboriginal sex offenders, an important issue to consider is the extent to which the available tools (and their items) are equally predictive for both groups. Differences between Aboriginal and non-Aboriginal sex offenders on key characteristics related to criminal behaviour (e.g., deviant sexual interest, substance abuse) do not mean that available risk scales are invalid for Aboriginal sex offenders. These findings certainly should, however, motivate researchers to examine possible differences in the predictive accuracy of scales assessing these factors.

Preliminary research has suggested that Static-99 is equally predictive of sexual recidivism for Aboriginal (n = 109) and non-Aboriginal sex offenders (n = 254; AUC = 0.67 for both groups; Nicholaichuk 2001). Studies examining differences on scores and predictive accuracy of risk scales with larger and more varied samples continue to be sorely needed to help understand the applicability of risk scales, such as Static-99, to Aboriginal sex offenders.

In a recent study, we examined the predictive validity of the STABLE-2007 (a dynamic risk scale designed to predict sexual recidivism), in an attempt to further extend the research on the predictive accuracy of risk scales for Aboriginal sex offenders (Helmus, Babchishin, and Blais 2011). The STABLE-2007 is a 13-item scale divided into 5 subsections: social influences (e.g., positive influences in the offender's life), intimacy deficits (e.g., capacity for relationship stability), general self-regulation (e.g., impulsive acts), sexual self-regulation (e.g., deviant sexual interests), and cooperation with supervision. In this study, Aboriginal (N = 89) and non-Aboriginal (N = 520) sex offenders were first compared on their STABLE-2007 item scores and total score. Results indicated that Aboriginal sex offenders scored significantly higher on the STABLE-2007 total score as well as on items indicative of general antisociality (i.e., cooperation with supervision, impulsivity, lack of concern for others, and poor problem-solving). There were no differences between Aboriginal and non-Aboriginal sex offenders on social-influence items or sexual self-regulation items. In the second part of this study, we examined differences in the predictive validity of both STABLE-2007 item scores and total score. Despite the fact that Aboriginal sex offenders scored, on average, higher on the risk factors measuring general criminality compared to non-Aboriginal sex offenders, these items were significantly less predictive for Aboriginal than for non-Aboriginal sex offenders. In contrast, items assessing sexual self-regulation, social influences, and relationship stability predicted similarly for both groups.

These findings are surprising considering that the research findings on general offenders, albeit few, have found risk factors indicative of antisociality and general criminality to be equally predictive for both Aboriginal and non-Aboriginal offenders. One plausible explanation for these results is that psychologically meaningful risk factors (e.g., antisociality) are not adequately represented by the items in the STABLE-2007. Specifically, Mann, Hanson, and Thornton (2010) have proposed that risk factors generally measured in research studies and risk instruments are markers of underlying propensities. For example, antisociality may be theorized to be an underlying propensity that plays a causal part in offending. Items such as number of past criminal offences or poor problem-solving are included in risk instruments in an attempt to capture or tap into the propensity for antisociality. It is possible that the causal factors (propensities) for recidivism are similar for Aboriginal and non-Aboriginal offenders but that the risk factors (markers) as assessed by certain items are not adequately measuring these propensities for Aboriginal offenders.

Our...