I am interested in the modelling of complex psychosocial systems, attempting to describe and understand the circumstances, attitudes and behaviours of individuals in their social environment. I'm particularly interested in the application of these methods to issues of serious mental health and social enfranchisement.
Latent Variable modelling
I specialize in statistical analysis using latent variables. This provides a way of studying many of the important aspects of psychosocial systems that aren't directly observable or measurable. Common terms such as 'mental health', 'social capital' and 'ethnicity' actually describe latent variables, hypothetical constructs that we use to make sense of the things we do see in the world around us.
Specifically, I use general latent variable modelling software (e.g. gllamm, Mplus) to fit:
- Structural Equation Models, to assess causal, mediation hypotheses
- Latent Growth Curve Models, to analyse change over time
- Mixed Multinomial Logit Models, to analyse preferences and choice, e.g. voting
- Factor- and Item-Response Theory Mixture Models, to analyse patterns of individual behaviours and questionnaire responses.
Follow the link below to watch me give a five minute introduction to Structural Equation Modelling, using a tea towel: