The Analysis of Residence Histories and Other Longitudinal Panel Data: A Semi-Markov Model Incorporating Time Varying Exogenous Variables

Northwestern University, Evanston, IL 60201, USA
Regional Science and Urban Economics (Impact Factor: 1.01). 05/1983; 13(2):271-285. DOI: 10.1016/0166-0462(83)90017-0


The analysis of residence histories and other longitudinal panel data is fraught with methodological problems. Much recent progress has been made in methods of analysis within discrete time. This paper extends the development of empirically tractable mixed continuous time stochastic models. Analysis of a sample of intra-urban residential histories identifies the effect of tenure type, age of household head, size of household and duration of stay on movement probabilities. Surprisingly, no further variation, as represented by a gamma mixing distribution over a hazard rate parameter, may be identified.

1 Read
  • Source
    • "For a first-order Markov model, it would be necessary to condition upon the first outcome and only outcomes for t 2 2 would be modeled. For a renewal process, it would be necessary to condition either upon the duration of stay prior to t = 1 or upon the outcome sequences up to and including the first recorded event, although commencing analysis from the beginning of the process enables us to avoid conditioning (Pickles 1983). In any application, the actual conditioning adopted will thus depend upon the type of feedback included in the model and the data that are available or feasible to collect. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Unless a longitudinal analysis is adopted including both adaptive behavior (feedback) and control for omitted variables, the behavioral modeler risks obtaining seriously misleading results. Marginal likelihood methods provide a suitable inferential framework but present largely unfamiliar problems. These problems are explained and assessed.-Authors
    Preview · Article · Sep 2010 · Geographical Analysis
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper reviews the problems of omitted variables in panel data analysis. Results obtained from previous analyses have been shown to be sensitive to the particular parametric forms assumed for the distribution of omitted variables. An alternative non‐parametric approach is described in which the effects of omitted variables enter the model through the non‐central moments of an error term distribution. A reparametrisation ensures that the constraints on the moments of a probability distribution are not violated. The model is applied to data on the residential mobility history of a sample of households in Cardiff, Wales.
    No preview · Article · Nov 1983 · Journal of Mathematical Sociology
  • [Show abstract] [Hide abstract]
    ABSTRACT: Theories of choice, which geographers frequently entertain, suggest that an individual’s choice is not only a function of the measurable characteristics of the individual and the choice set, but may also be a function of some unmeasurable component such as tastes, motivations and natural propensities (e.g., ‘chronic movers’, Morrison, 1970). In addition, choice is suggested to be a function of past behavior. Examples of such a relationship are the changing of choice probabilities as the result of learning from previous choices, growing loyalty or increasing inertia to change with repeated use of the same alternative and circumstances in which a change of choice is associated with a set of varying costs. In studies of migration, where choices are areas of residence, all of these forms of relationships may be expected to exist.
    No preview · Chapter · Jan 1984
Show more