The use of estimating functions as tools for statistical inference in the presence of nuisance parameters. Particular areas of application are models for measurement error in regression covariates and methods for missing data in models for highly stratified or longitudinal data.
Applications wherein it is useful to model association (e.g., correlations, odds ratios) in a regression framework.
Areas of substantive interest include psychiatric epidemiology (chidren and adults), aging research, genetic epidemiology, environmental impacts on health.