The analysis of recurrent events for multiple subjects

Methods of analysis for recurrent events for subjects in longitudinal studies are reviewed. Examples concerning the occurrence of tumours in laboratory animals undergoing a carcinogenicity test and the occurrence of automobile warranty claims are discussed. Different ways of introducing covariates and the distinction between marginal (e.g. rates and mean functions) and conditional (e.g. intensity functions) event process characteristics are emphasized. The usefulness of robust methods for estimating rates and mean functions is demonstrated.