Measurement Issues of Victim-Offender Overlap

• In Event: New Insights into Youth Crime in a Digital World: ISRD4 Results from 21 countries

Abstract

Based on first results of the ISRD4 data (21 countries) the presentation will address the problem of the measurement of victim-offender overlap. Already in 1912, presenting a coefficient of association known as Yule’s Y, Udny Yule did show that his previously developed measure of association (Yule’s Q, also known as Goodman & Kruskal’s Gamma) for two dichotomous variables is not well suited to compare the association (or overlap) when comparing groups. More recent work did demonstrate that Loevinger’s H, which has also been used for the development of scales with dichotomous items, is a promising alternative, especially when the focus is on a positive association (such as the victim-offender overlap). Demonstrating the advantage of using H instead of simple percentages of overlap the presentation will compare the victim-offender overlap in different countries. Additionally, we look at the association of victimization by parental violence and violent vs. non-violent offending. Results show that in the case of juvenile delinquency, the idea that “violence begets violence”, which seems obvious at first glance, is far too simple.

Author

• Dirk Enzmann, University of Hamburg, Germany