Sexual Recidivism

Sexual Recidivism in Europe

Survival Models in the Context of Sexual Recidivism Research

Heinz Leitgoeb, from the University of Linz, made a contribution to the 2012 Annual Conference of the European Society of Criminology, in the category “Crime and Society,” under the title “Survival Models in the Context of Sexual Recidivism Research”. Here is the abstract: The appropriate strategy for modeling sexual recidivism data depends highly on the research question of interest. Since recidivism data generally contain information about event occurrence and the time from prison release to reoffending, OLS and logit regression represent – for a various number of reasons – rather unsuitable modeling approaches for this type of data. In contrast, due to their flexibility, handling of (right-)censored cases, and the possibility of implementing time varying covariates, the application of survival (or event history) models shows great promise for recidivism research. Nonetheless, these models do – at least in contemporary European research – not enjoy great popularity. When applied, the predominant survival model of choice is the semi-parametric Cox model, mostly without further questioning its applicability for the underlying research question. For this reason, I intend the development of a heuristic, allowing the applied researcher to select the appropriate survival model for his/her research question (explanation, prediction, estimation of treatment effects) in the context of sexual recidivism.

Resources

See Also

Further Reading

  • “Survival Models in the Context of Sexual Recidivism Research”, by Heinz Leitgoeb (Proceedings)

Leave a Comment