A NEW METHODOLOGY FOR ASSESSING THE LEVEL OF RISK IN INCARCERATED OFFENDERS

Abstract

A review of the literature relating to the assessment of risk in incarcerated offenders is given. A new methodology for the actuarial assessment of risk based on the prediction of the way in which offending behaviour traits will be manifest in the prison environment is described. A validation of this methodology is described giving a level of accuracy of 65 per cent. A measure of the accuracy, the rated percentage agreement (RPA), is described and the inter-rater reliability of the methodology is investigated. The utility of the instrument in a prison environment is outlined and implications for future refinement and validation are discussed. A long-standing problem for the criminal justice system has been how to assess accurately the risk of re-offending for long-term prisoners, serving indeterminate sentences, convicted of serious offences. The fact that the vast majority of these offenders will eventually be released, and the desire effectively to integrate them into the community, must be balanced against the need to protect the public from the risk of further serious offences. Proposed changes introduced by the 1991 Criminal Justice Act, the decision by the European Courts regarding the right of discretionary lifers to challenge their continued incarceration, and the advent of more open reporting within the English prison system have brought the criteria and methods used to establish risk under close scrutiny. 'It is widely acknowledged that past behaviour is the best predictor of future behaviour,' write Owens and Schoenfeldt (1979). This statement is demonstrably true for criminal behaviour, where virtually all major reviewers (Cornish and Clarke 1975; Farrington and West 1990; Andrews 1983; Hill 1985) have concluded that the best predictor of recidivism is the type and frequency of previous convictions. This finding applies across all ages, offence types, and cultures, and is quite in keeping with current psychological theories. However, when attempting to assess risk of re-offending among a group of serious offenders serving long sentences, several problems are encountered. One problem is that of base rate; the frequency of serious criminal offences (i.e. serious violent or sexual assaults) is fairly low even in such a population. Thus, predicting from past behaviour, one would expect very few of these offenders to re-offend. Unfortun ately, the consequences of a second similar offence are too great for one to accept this general prediction. A second difficulty is that data available at the beginning of sentence do not change over time. So, although one might compute sufficiently accurate actuarial parole risk scales for offenders (Andrews 1983), the level of risk suggested by such tools cannot change over time. Therefore a prisoner assessed with this pre-conviction data is judged to be equally at risk of re-offending the day after sentence and twenty years later. It is possible that this might well be the case, but it