Methodological Considerations in the Analysis of Residential Mobility: A Test of Duration,State Dependence,and Associated Events

This paper provides evidence that a longitudinal research methodology is particularly well suited to disentangle life-course explanations of residential mobility while controlling for the duration-of-residence effect. Much of the literature on residential mobility has yielded conflicting evidence regarding the effects of cumulative inertia, cumulative stress, and duration dependence due in large part to different methodologies and different operational conceptions of the life cycle. I argue that the lack of analytical attention directed toward the concept of the risk period, as well as the persistent use of the household, rather than the individual, as the unit of analysis, further serve to confound our understanding of residential mobility. The empirical analysis uses the Panel Study of Income Dynamics and discrete-time logistic regression models to include duration dependence, state dependence, associated events and nonstationarity in models of residential mobility for both renters and owners. Using an event-oriented data structure that links household careers and housing careers, the study is conducted in three stages: (1) important tenure differences are established using comparative survival rates, (2) discrete-time logit models of the hazard of moving, that include duration dependence and state dependence, are fitted by tenure group, and (3) the models are extended to include a change in household type as an associated event that increases (or decreases) the likelihood of moving. The findings indicate the potential of event-history analysis to advance the field of residential mobility by providing a means to assess issues that once seemed intractable.