ABSTRACT

This chapter presents a Markov chain model for the estimation of individual-level binary transitions from a time series of independent repeated cross-sectional (RCS) samples. Although RCS samples lack direct information on individual turnover, it is demonstrated here that it is possible with these data to draw meaningful conclusions on individual state-to-state transitions. We discuss estimation and inference using maximum likelihood, parametric bootstrap, and Markov chain Monte Carlo approaches. The model is illustrated by an application to the rise in ownership of computers in Dutch households since 1986, using a 13-wave annual panel data set. These data encompass more information than we need to estimate the model, and this additional information allows us to assess the validity of the parameter estimates. We examine the determinants of the transitions from have-not to have (and back again) using well-known socioeconomic and demographic covariates of the digital divide. Parametric bootstrap and Bayesian simulation are used to evaluate the accuracy and the precision of the ML estimates, and the results are also compared with those of a first-order dynamic panel model. To mimic genuine repeated cross-sectional data, we additionally analyze samples of independent observations randomly drawn from the panel. Software implementing the model is available.