Timing and duration effects in residence histories and other longitudinal data: I. Stochastic and statistical models

A framework for studying the timing of events in migration histories and other micro-level longitudinal data is presented. The framework derives from a general stochastic model of the histories in which moves depend on the past history of the process, time varying individual characteristics, and exogenous constraints and opportunities. The semi-Markov model plays a distinguished role. The framework emphasizes the range of stochastic models available, the different types of time intervals and observational schemes that can be considered, distributions that can be used to characterize intervals, and statistical methodology. The use and crucial importance of the framework in empirical research is illustrated in the sequel.