Revisiting the NSW crime decline: New data yield new conclusions from two birth cohorts (1984 and 1994)

This paper offers a rare, but critical reflection on the importance of replication in the quantitative analyses of “best available” data that rely on iterative data matching algorithms – a warning that all crime data-analysts should heed. Using officially recorded police and courts data from the New South Wales (NSW) Bureau of Crime Statistics and Research, this paper traces the longitudinal criminal trajectories of two NSW birth cohorts (1984 and 1994). It updates earlier analysis conducted on the same two cohorts, providing a more comprehensive set of data for an additional 7 years of observation (to age 27). Importantly, the new data presented in this paper are derived from an improved data-matching process that now includes police warnings, cautions, and finalised court outcomes, and the results drastically alter previous conclusions about the between-cohort differences and their likely contribution to the NSW crime decline. Specifically, this new analysis shows that while fewer members of the 1994 cohort had contact with the police by age 27, they accumulated a greater volume of offences and were, in fact, more violent than their 1984-born peers. Much of the apparent differences reported earlier were a consequence of a considerable shift towards informal processing (i.e., warnings and cautioning) and, to a lesser extent, enhanced algorithms that inform data matching principles and strategies.