Log-Linear Models for Event Histories

Event history analysisa method for explaining why some people are more likely to experience a particular event, transition, or change than other peoplehas been useful in the social sciences for studying the processes of social change. One of the main difficulties, however, in using this technique is that often information is (partially) missing on some of the relevant variables. Author Jeroen K. Vermunt presents a general approach to these missing data problems in event history analysis that is based on the similarities between log-linear, hazard, and event history models. The book begins with a discussion of log-linear, log-rate, and modified path models and methods for obtaining maximum likelihood estimates of the parameters of these models. Vermunt then shows how to incorporate variables with missing information in log-linear models for non-response. In addition, he covers such topics as the main types of hazard models; censoring; the use of time-varying covariates; models for competing risks; multivariate hazard models; and a general approach for dealing with missing data problems, including unobserved heterogeneity, measurement error in the dependent variable, measurement error in the covariate, partially missing information on the dependent variable, and partially observed covariate values.