The contribution of behavioural observation to high-risk offender management: results from three pilots

Objectives: When assessing violence risk, evidence on behaviour is often overlooked or given less weight than self-report, and practice has been questioned following a number of high profile absconds and serious further offences. Here I aim to integrate three different studies of behaviour observation in risk management.
Design: Predictions from structured observations were compared to those from a comparison measure (Study 1 and 2), or used to discriminate matched re-offender-desister pairs (Study 3). Study 2 and 3 used ADViSOR, which records offence-related and positive alternative behaviours.
Methods: Study 1 predicted in-prison behaviour using functional analysis of the index offence of 65 life sentenced prisoners, and compared accuracy to that from a randomized set of predicted behaviours. Study 2 used institutional observations to predict re-imprisonment of 25 high-risk sex offenders. Study 3 used observations of 30 delinquent juveniles to predict re-offending or higher education enrolment within three years.
Results: In study 1: a higher concordance in the functions of behaviours, compared with dummy predicted-actual pairs. In study 2: the frequency of negative behaviours strongly predicted re-imprisonment, and prison-community behaviours were highly similar. Meanwhile, MAPPP meeting evidence identified only one-half of the re-offenders. In study 3: negative behaviour frequency predicted reoffending, while the frequency of positive behaviours predicted desistance.
Conclusions: Results of each study provided evidence for the management of offenders that were high-risk but with few/no previous recorded offences. Low level behaviours observed were overlooked by usual practice but added to the probability of risk/protection. ADViSOR can help benchmark changes in risk during a sentence.