Prädiktoren für Gewalt während des Strafvollzugs

Abstract

The factors leading to violent behaviour in prison have been the subject of various research projects. Besides causing financial and structural damage to the penitentiary, violent behaviour during prison sentences is considered to be an important predictor of re-offending. The following article provides an overview of the research literature concerning predictors of violence during imprisonment. The examined criteria include socio-demographic (e.g. age, marital status and level of education) as well as criminological (e.g. previous violent convictions, duration of sentence and type of offence) factors. However, the analysis of the significance of the individual factors is not conclusive. Considering the results of numerous replication studies, young age seems to be connected to an increase in risk for intramural infractions while partnership (marriage) seems to have a protective influence. Regarding vocational training and educational level, mixed results were found. However, being unemployed at the time of the crime seemed to increase the risk for rule violations during imprisonment in most studies analysed. Furthermore, it is generally accepted that duration of sentence and time already served has an influence on the risk of violence as well. Inmates with shorter or finite sentences presented violent behaviour more frequently. The relationship between duration of sentence and violent infractions has repeatedly been described as an inverse U-shaped curve. Accordingly, most violations of prison rules occurred in mid-sentence. The literature examined presents inconsistent results concerning previous violent convictions, type of index offence, ethnic origin and the influence of alcohol consumption and drug use. Evidence seems to be more consistent regarding pre-existing risk-associated socio-demographic variables rather than institutional conditions. Overall comparability of the studies was, however, reduced by the fact that violent behaviour was defined very heterogeneously in the different institutions - sometimes even including verbally aggressive behaviour. Institutions also differed in the likelihood of an infraction being recorded. The heterogeneity of the results can furthermore be explained with the very different cultural and sociological backgrounds of the different institutions examined. As prognostic models are very sample sensitive it is vital to define precisely for which populations the prediction is valid and to examine samples of appropriate size and representativeness as well as control a wide variety of influencing factors with multivariable controlling strategies.