Prediction of Violent Behaviour: A Review of the Development of Prediction Research

• NCJ Number: 217063
• Journal: Acta Criminologica Volume: 19 Issue: 2 Dated: 2006 Pages: 27-41
• Author(s): C. Strydom; D. A. Louw
• Date Published: 2006
• Length: 15 pages
• Publication Format: Document (Online)
• Publication Type: Issue Overview
• Language: English
• Country: South Africa
• Note: To access the full text PDF: 1) select the provided link; 2) from the Acta Criminologica Web site, select "Table of Contents"; 3) select the corresponding Volume and Issue (see the NCJRS abstract record for the exact Volume and Issue); 4) scroll the Table of Contents to the exact article; and 5) click on the "full text" icon. Downloaded on January 23, 2007.

Annotation

This paper provides an overview of the existing research on the prediction of violence.

Abstract

The most important conclusions of the recent research on risk evaluation are summarized in the following nine points: (1) rectification of the numerous shortcomings and problems in respect of risk evaluation, as well as progress in this field, must begin with a change in the way of thinking, recognizing the fact that violence and violent offenses do not comprise a unitary concept phenomenon; (2) although the reliability of predictions is still one of the major points of contention, the perfection ideal of the first-generation research was replaced with a more realistic pursuit of a sufficient degree of accuracy that would be feasible, acceptable, and usable; (3) excessive emphasis is placed on false positive and false negative results in the debate surrounding the reliability of risk evaluation; (4) although, as of yet, no consensus exists among experts concerning the factors that should be associated with the risk of violence, there are a number of specific variables which can be linked fairly consistently to the risk of violence; (5) significant progress has been made with the definition and validation of a number of risk-related variables which is evident from the development of relatively new measuring instruments based on these variables; (6) although actuarial predictions tend to enjoy priority among numerous research projects, this does not mean that such predictions are necessarily deemed to be better in all contexts than predictions made in terms of the clinical method; (7) it is highly problematic to test the reliability of predictions of violence; (8) progress in risk prediction is particularly dependent on consistent elimination of the errors that have already been highlighted in research; and (9) the focus of research has shifted from disputes about the reliability of predictions of violence to methods of improving such predictions and using them more effectively. References