The role of interaction in generalized linear models arises in the context of analysis of variance, log linear models and logistic models. A unifying formulation is under investigation.

Profile, marginal and conditional likelihoods provide useful methods for determining the evidence from data regarding unknown model parameters. Extension of these methods to hierarchical and other complex models is under investigation.

Biostatistics, Environmental statistics; generalized linear models; linear models; multivariate analysis

Please see my CV for more details

Fellow of the American Statistical Association

Click Here to View Selected Recent Publications