Results by Design: The Artefactual Construction of High Recidivism Rates for Sex Offenders

• NCJ Number: 213535
• Journal: Canadian Journal of Criminology and Criminal Justice Volume: 48 Issue: 1 Dated: January 2006 Pages: 79-93
• Author(s): Cheryl M. Webster; Rosemary Gartner; Anthony N. Doob
• Date Published: January 2006
• Length: 15 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: Canada

Annotation

This article examines the validity of the principal claim of Langevin et al.'s (2004) study of sex offenders' recidivism rate, i.e., that it is exceptionally high.

Abstract

Although Langevin et al. may provide several plausible explanations in justifying their unprecedented results (for example, their use of unusually long followup periods and multiple sources of data), the findings derive from research methods that are so problematic as to make their findings meaningless. Methodological problems range from the researchers' choice of comparison groups to the statistical analyses conducted; however, this article focuses its critique on the sample characteristics and its implications for the validity and integrity of the sample's recidivism rate. The study followed 320 male sex offenders and 31 perpetrators of nonsexual offenses for a minimum of 25 years. The recidivism data were obtained from police records, hospital records, and self-admissions. The sample of offenders was obtained from male sex offenders referred for psychiatric assessment or treatment by the court, police, probation and parole services, defense lawyers, or other mental health professionals between 1966 and 1974. From this sample, the researchers presumed to generalize their findings to all sex offenders. This is done without any showing that the research sample is representative of the wider population of sex offenders. The research also confuses the concepts of recidivism and lifetime criminal history when examining the reoffending patterns of sex offenders referred for psychiatric assessment. By preselecting a sample that includes a substantial proportion of offenders who were already recidivists, the study contained at the outset persons at high risk for reoffending. This inherently biases the study toward an artificial inflation of reoffending rates that are not representative of sex offenders in general; therefore, the recidivism rates reported lack both validity and integrity for any application to sex offenders not in the sample. 6 notes and 25 references