Research Activities

Our specific research strands, each of which involves the development of new statistical models, cover four topical issues. These are concerned with social and developmental change both through the substantive topics and through the statistical and computational issues related to them.

Patterns in Criminal Careers.

Can we identify offence patterning - that is, offences which are committed at the same point of an offender's criminal history? How do such patterns change over time? Can we identify generational effects? Is there evidence of patterns in serious offences? Statistical methods involve the development of dynamic latent class analysis models to represent changes over age, period and cohort.

Changing values and value systems.

How are value systems changing over time and across countries? Is there evidence of an increase in post-materialism? Can we find better ways to measure postmaterialism? 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.

Development of executive function in young children.

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.

Efficient methods for fitting complex random effects models.

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.

Funded By