"Taking down the straw man" or building a house of straw? Validity, equity, and the Custody Rating Scale.

• Author: Webster, Cheryl Marie
• Position: Response to article by Kelley Blanchette and Laurence L. Motiuk in this issue, p. 621

Given the powerful impact of security level assignment on both the female offender and the correctional facility, as well as the serious implications of (mis)classification for broader legal, economic, and humanitarian issues, we welcomed Blanchette and Motiuk's reply to our article (Webster and Doob 2004). Indeed, we would have been pleased to be persuaded that our concerns regarding the lack of validity and equity of the Custody Rating Scale (CRS) used to assign initial security levels to federally sentenced women offenders in Canada were incorrect. Unfortunately, we were not convinced. On the contrary, this rejoinder will demonstrate that our principal concerns were either not addressed or not convincingly refuted. Reluctantly, we are forced to conclude that rather than "taking down the straw man," Blanchette and Motiuk (2004) have simply reinforced the classification of their own house as fragile and unprotected.

In this rejoinder, we have opted not to enter into linguistic distinctions of whether the CRS "constitutes the foundation of the security classification system in Canada"--as we note (2004: 396), along with others (Canadian Human Rights Commission 2003)--or represents one of the "offender classification instruments [that] are critical to the effective and efficient risk management of women offenders while under federal sentence"--as claimed by the Correctional Service of Canada (CSC) (Research Branch, Correctional Service of Canada 2003: 11). Similarly, we have chosen to ignore the misguided criticism levelled at us for not having provided any new data, since the entire point of our prior paper was to demonstrate that CSC's own data do not support its repeated claims of the validity and equity of the CRS. Further, while we concede - despite similar affirmation by others (Canadian Human Rights Commission 2003)--that the CRS may not originally have been designed for a homogeneous (white) male population, we simply note that the veracity of this statement is irrelevant to the persuasiveness of our critique.

Rather, we prefer to focus our rejoinder exclusively on our principal conclusions. For the sake of simplicity, we have opted to summarize our key findings, followed by a discussion of Blanchette and Motiuk's response to each of them.

I--Overall scale validity

While a significant linear relationship exists between overall CRS security level designations and institutional incidents (CSC's outcome measure for assessing the predictive validity of its classification scale) for women as a group, we note that this relationship collapses for Aboriginal female offenders. More specifically, the institutional incident rates for Aboriginal women classified in minimum and medium security are essentially the same. Clearly, the CRS (as a whole) lacks predictive validity for this sub-sample of female offenders, as it is unable to discriminate accurately between these two security classification designations according to differing probabilities of inmate risk.

Blanchette and Motiuk begin by reiterating--as we note--the existence of an overall linear relationship between CRS designations and security incident involvement over a six-month follow-up period for women offenders in general (623). However, they are unexpectedly silent on the absence of a linear relationship for Aboriginal women, leaving us to continue to assume that the CRS lacks predictive validity for this sub-population. Subsequently, they provide new data demonstrating the same general linear relationship between CRS security level designations and their newly presented measures of "risk to public safety" for women who have been released (625). Unfortunately, they continue to present no separate validity data for Aboriginal and non-Aboriginal sub-populations. Particularly given the absence of a linear relationship between the overall scale designation and institutional incidents for Aboriginal inmates, it would seem imperative to examine this possibility for these new outcome measures. In the absence of additional evidence to the contrary, we are forced to continue to conclude that the CRS lacks predictive validity for this group.

II--Sub-scale...