A cluster analysis exploring youth welfare and justice histories linked to later offending
Degree GrantorUniversity of Canterbury
Degree NameMaster of Science
The first aim of this thesis was to investigate the possible existence of subgroups within a Child, Youth and Family dataset based on known risk factors for recidivism using exploratory cluster analysis. A second aim was to examine any association between cluster group membership and later offending (between the years 2003 and 2018) using follow-up conviction data obtained from the Ministry of Justice. The third aim was to ascertain if subgroups identified in this study would bear similarities to the risk groups identified by the McKinlay, James and Grace (2013) study. The dataset contained all New Zealand youths who received a youth justice intake during 2002 (N = 4,307). Exploratory cluster analyses identified 11 variables relating to care and protection and youth justice and K-means clustering identified a four-cluster solution. Cluster one had highest number of females and more likely to have a care and protection intake and little prior offending. Cluster two were prior offenders but not as high risk as cluster four and had little care and protection history. Cluster three had little care and protection and prior youth justice histories and were the largest group. Cluster four was the highest risk, early onset and repeat offenders. Understanding how the care and protection and youth justice histories of children and adolescents cluster together, allows the targeting of programmes to highest needs.