Most of the contemporary literature regarding classification is predominately American, and most major discussions related to prediction and classification there and elsewhere has ignored women and young people as specific groups with differential need and risk factors. As we have seen in Canada too, rather than develop and implement independent women-centred models, in accordance with the 1990 recommendations of the Task Force on Federally Sentenced Women, the classification and assessment literature, based upon which tools to assess women are developed, tend to be validated on the Canadian male correctional population. As a result, women are still the last two to three per cent of the federal prison population to be examined.

This is the reality despite the fact that attempts have been made during the past few years to apply and validate some of the male instruments for women prisoners (see, for example, Blanchette and Motiuk, 1995, 1997; Blanchette, 1997, 1998). The evidence as to the applicability of male instruments to women appears patchy at best. Some factors, however, must be seriously questioned. For instance, the rating of adult histories of abuse, self-injurious behaviour and suicide attempts as risk factors for violent re-offending have serious implications for women prisoners.

Without contextualizing women's experiences, especially their experiences with violence, it is dangerous and highly problematic to draw conclusions from such correlations. As has been alluded to above, if these were true predictors of violent offending then, given the incidence of violence against women and children, they should be the primary perpetrators of violent offences. The reality belies such insular assessments and research of isolated and unrepresentative groups. Moreover, the lack of attention to gender, race, class, and other areas of diversity results in a variety of layers of systemic discrimination and disadvantage. These range from over classification to a lack of attention to program and service needs and interference with community release and integration strategies/plans.

One of the most compelling, relevant and clear documentation of the shortcomings of the actuarial risk assessment process as predictors of future violent behaviour exists in research generated by the Correctional Service of Canada (CSC). As was discussed earlier, Brian Grant's work regarding the application of existing prediction instruments is especially relevant to these discussions.

Pursuant to the detention provisions of the Corrections and Conditional Release Act (CCRA), CSC identifies prisoners who they believe should be kept in prison for their entire sentence, without any opportunities for conditional release; namely, until the expiration of their respective warrants of committal. These are prisoners who CSC assesses as likely to commit an offence that would likely cause serious violent harm or death prior to the expiration of their warrant of committal to custody. Once a prisoner is assessed as someone who CSC believes should be referred for detention, their cases are referred to the National Parole Board (NPB) for a hearing designed to assess whether in fact they should be detained.