Levels of aggression and impulsivity in prisoners using self-report questionnaires

The psychometric assessment of traits theoretically linked to human aggression remains a popular topic of concern to research psychologists. Whilst there is a growing body of research using these measures with forensic populations (e.g. Patton, Stanford and Barratt, 1995; Archer and Haigh, 1997; Ireland and Archer, 2004) the majority of published studies have utilised psychology undergraduates (whose availability and status put them at the mercy of the demands of researchers). Often psychometric measures are used simply to designate high and low aggression groups prior to undertaking some other experimental procedure. Moreover, whilst many of these scales appear to be relatively efficient in distinguishing between male and female undergraduates there is a paucity of empirical data that relates to normative scores in offenders. It is often assumed that high scores reflect an increased likelihood of that individual engaging in antisocial behaviours with little or no corroborating evidence. As part of an ongoing research project examining information processing bias in offenders, the authors recently had the opportunity to collect data from a large sample of male and female prisoners in the North of England using some commonly used questionnaire measures.

During the analysis it struck us that this was an ideal opportunity to examine offender’s self-reported levels of aggression as a within-groups exercise rather than the more traditional (in mainstream psychology) between-groups analysis. We predicted that offenders (differentiated by gender and conviction for a violent or non-violent index offence) would show substantive differences in their self-reports with violent offenders reporting higher levels of aggression and impulsivity. We also predicted that a more detailed analysis of the index offences themselves would produce differences in terms of reported levels of aggression and impulsivity. This paper aims to share these results with an informed readership. The large number of questionnaires administered to each respondent inevitably increases the complexity of any statistical analysis. This article is not intended to provide a comprehensive summary of the findings, this is provided elsewhere (Smith and Waterman, in press). We shall restrict ourselves in this paper to looking at the main group and gender effects for each measure and highlighting those areas likely to be of interest to forensic practitioners.