location: publications / books / Justice Behind the Walls / Sector 4 / Chapter 3 The Arbour Report: The Indictment of a System / Postscript: "The Most Comprehensive Empirical Review . . ."

It is difficult to see how practice effects would influence tests that do not rely on speed or correct answers. For example, a test designed to measure depression might ask if the subject has trouble getting out of bed in the morning, or a test measuring anxiety might ask if the subject has sweaty palms. There is no right answer to these questions so it is difficult to see how practice would make perfect. In his thesis Dr. Zinger suggests that the practice effects phenomena in his research operates so that "Participants lose interest in answering repeatedly to identical questions and tend to report less problems overtime." In response to Kate OíBrienís probing of this explanation, he augmented this explanation by saying that subjects get bored with taking the same test over and over and they began to think less about their answers and hence became more truthful over time. Ms. OíBrien, like myself, found this explanation more confusing that helpful. Why would prisoners become more truthful or why would their answers necessarily show an improvement in mental health?

Two other social science researchers have raised another significant problem with the reliability of Zingerís findings relating to the limited confidentiality which Zinger and his researchers offered research participants in obtaining their informed consent. The research assistants in obtaining that consent gave the prisoners the following warning:

Before we begin, I need to tell you that although the information you provide today will be confidential, there are limits. I have an obligation to disclose any information you may provide if itís in regards to your safety or that of the institution. These areas include suicide plans, plans of escape, injury to others and the general security of the institutions.

Dr. Ted Palys and Dr. John Lowman of Simon Fraser School of Criminology have argued that this limited confidentiality is directly related to the research validity of Zingerís results. Given that Zingerís research focussed on two symptoms (1) "internalizing" symptoms, e.g. suicidal tendencies; and (2) "externalizing" symptoms, e.g. anger directed towards others, he hypothesized that if segregation has debilitating effects, prisoners should express greater hostility and aggression to themselves and others than prisoners who are not segregated. Palys and Lowmanís critique is contained in the following paragraph:

A credible test of this hypothesis requires honest and unfettered responses in interviews and any written tests that require behavioural or emotional self-report. These conditions are most likely to be met when anonymity can be guaranteed or, if anonymity is not possible, when the participant completely trusts the researcher. Anonymity would be difficult to achieve in a prison solitary confinement unit. Therefore, prisoners had to divulge information on the basis of trust . . . Instead of making an unconditional guarantee of confidentiality, the interviewers made it clear that information relating to institutional security would not be confidential but would be reported to prison authorities. Researchers effectively said to prisoners, "We want to know about your aggressive tendencies towards yourself and others in order to reveal the Ďtruthí about solitary confinement, but if you tell us about any, we must report you." This kind of qualification clearly could -- and we believe would -- influence what participants would be prepared to reveal to researchers about their behaviour and emotional state in both interviews and psychological tests.

Why would a researcher choose to proceed with eyes wide shut, guaranteeing confidentiality except for information that is crucial to the study? If there is no difference between the "isolated" and "general prison population" groups in their self-reported propensity for suicide and aggression toward others, it may well have something to do with information that was excluded from the study as a result of Zingerís limitation on confidentiality. (Ted Palys and John Lowman, "Social Social Science With Eyes Wide Shut: The Limited Confidentiality Dilemma" (2001) 43 Canadian Journal of Criminology 255 at 257)

Palys and Lowman point out that at the time when Zinger was conducting his research for his thesis he was an employee of the Correctional Service of Canada and they therefore "presumed his felt obligation to limit confidentiality originates in his role as an employee. The problem is that his role as a CSC employee and his role as a researcher conflict." (at 264)

In that response to Palys and Lowman, Dr. Zinger and his research associates argue that on the basis of research participantsí score on the Balanced Inventory of Desirable Responding [BIDR] "we can report with confidence that segregated prisoners employed fewer tactics to deceive researchers (intentionally or not) and likely responded accurately throughout our data collection." (Zinger, Wichmann and Gendreau (2001) 43 Canadian Journal of Criminology 269 at 273). In reply to Zingerís response, Palys and Lowman write:

The problem with his claim is that BIDR barely speaks to the problem we have identified. We suggested that, because of limited confidentiality, prisoners simply would not disclose certain kinds of information to researchers. There is a vast difference between "self-deception" and "impression management," the constructs underlying the BIDR, and the wholesale omission of information about what prisoners actually do or intend to do. Information about prisonersí actions and intentions is vital to assess "effects" of segregation. Why would they divulge this information to researchers knowing that it could have tangible negative consequences, such as further segregation or loss of privileges? (Lowman and Palys, "Limited Confidentiality, Academic Freedom, and Matters of Conscience: Where Does CPA Stand?" (2001) 43 Canadian Journal of Criminology, 497)

The Palys and Lowman critique raises a much larger issue that goes to the heart of the debate that was before Madam Justice Arbour during her inquiry. A quantitative analysis, even like Dr. Zingerís, that takes into account the methodological shortcomings of other such research, but virtually excludes qualitative data gathered from interviews where prisoners feel free to voice their real feelings and experiences, is subject to significant limitations in drawing conclusions about the effect of administrative segregation. As Roberts and Gebotys point out, from the very beginning Zingerís research had built in administrative constraints. The measures that were selected were ones which had a short administrative time. They point out "there are many psychological measures that take longer to administer, and which might have detected effects of segregation, but were not employed because they were inconvenient for the researchers." (at 92) They fault the research also because no in-depth interviews were conducted even though previous research using this methodology had produced disturbing findings. Roberts and Gebotys in their commentary make this appeal to Zingerís "sense of empirical modesty."

For all their sky-high alpha levels, the psychological tests used in this research have their limits. Isolation, particularly repetitive isolation, may have subtle effects that cannot easily be captured by the current measures of psychological functioning . . . To conclude, we would draw the following rather more modest conclusion from the research conducted by Zinger et al. With the restrictions noted in their article, as well as those identified here, the findings suggest that administrative segregation under two months in duration did not generate detectable psychological deterioration, using measures which consume " a short administration period " using a small sample of male prisoners with considerable previous isolation experience. (at 95)

Page 3 of 4