Abstract
It is surprisingly hard to fit mixed models to data from multistage surveys, especially if the structure of the model and the structure of the sampling are not the same. I will review why it is hard and some of the approaches. I will describe a very general approach to linear mixed models based on pairwise composite likelihood, which extends work by J.N.K. Rao, Grace Yi, and co-workers, and was originally motivated by modeling questions arising in the Hispanic Community Health Study/Study of Latinos. This approach is implemented in a new R package, svylme, and allows for nested and crossed random effects and for the sort of correlations that arise in genetic models. I will present examples and discuss some computational and inferential issues.
Department students and members are invited to meet with Dr. Lumley. Contact Bryan Shepherd to schedule a meeting.
Thomas Lumley, PhD, is a professor and the chair in biostatistics at University of Auckland's Department of Statistics. His research areas include bioinformatics, medical statistics, design of medical trials, statistical computing, and survey statistics. He is a fellow of the Royal Society of New Zealand and the American Statistical Association, and a member of the International Statistical Institute.