The proportion of youth offenders who receive an additional conviction at a five year follow up : testing and developing an actuarial risk model for predicting long term recidivism.
Degree GrantorUniversity of Canterbury
Degree NameMaster of Arts
The first aim of this thesis was to determine the percentage of individuals who having received a youth justice intake, went on to be convicted for a further offence five years on. A second aim was to assess the long term predictive validity of the Juvenile Risk Scale (JRS) and to then determine if a statistical model, developed specifically for predicting convictions in the long term, was able to provide more accurate predictions of convictions. An entire cohort of New Zealand Youths, who received a juvenile justice intake in 2002 in New Zealand, were matched to conviction records five years on (N= 4,307) . A nationally representative subsample of this cohort (N = 936), youths aged 13 to 17 years (745 male, 191 female), was utilised to assess the predictive validity of the JRS. Best-subsets logistic regression was used with this sub-sample to produce a predictive model for convictions five years on. Receiver Operating Curve analyses were used to assess and compare the predictive validity of the two models. Of the cohort sample, 54% have received a conviction five years on. The JRS was shown to hold good long term predictive validity for males but not females. The model developed yielded an ‘Area Under the Curve’ of .693, indicating moderate accuracy. Findings suggest that an automatically scored actuarial model for predicting risk of conviction in youths is feasible and may aid in the allocation of intervention resources.