Lancaster-Warwick-Stirling node

Can we identify offence patterning - that is, offences which tend to be committed together at the same point of an offender's criminal history? How do such patterns change over time? Can we identify generational effects? How dengerous are serious offences such as blackmail and kidnapping?  Can we identify patterns in the changing spatial distribution of recorded crime over time? What is the reconviction experience of white-collar offenders? Statistical methods involve the development of dynamic latent class and latent transition analysis models, spatio-temporal models and hazard rate analysis.

How are value systems changing over time and across countries? Is there evidence of an increase in post-materialism? How are attitudes to gender roles changing over time? What is the best way to model attitudinal data based on ordinal likert scales, ranked data and paired comparison data? 

Statistical methods under development with colleagues at the Economic University of Vienna include models for longitudinal partial ranks, modelling multivariate likert scales, and random effects models for ranked data.

How do cognitive skills develop in young children? Can we distinguish between competing psychological models? For example, when preschoolers develop skills in executive function, what causes such change? Likelihood-based statistical methods are being developed using latent variable models for multivariate longitudinal data.

This research strand addresses statistical problems which arise in analysis of conflict data (sociology), longitudinal voting patterns (political science) and other non-linear models where a random effects structure is needed. The research will investigate the use of pairwise likelihood to reduce an intractable high dimensional problem to a less complex tractable problem.