Early prediction of violent and non-violent youthful offending

Conclusions Violent and non-violent offending can be predicted more accurately than is generally believed. It is not difficult to identify a high risk category of people at age 8–10 who have an elevated probability (three or four times that of the remainder) of becoming offenders. This identification can be based on either antisocial behaviour or a deprived background or both. Depending on the prevalence of offending, the false positive rates may be high or low, but the odds ratio is a better measure of predictive efficiency. Overall, the early prediction of offending in the Cambridge study was quite impressive.