Tests of Stochastic Models of Timing in Mobility Histories: Comparison of Information Derived from Different Observation Plans

The fundamental assumption about the timing of moves in Markov, semi-Markov, and other renewal-theoretic models of residential mobility—that the intervals between moves are statistically independent—is tested using residence histories of Norwegian men, 1965–1971. The evidence strongly supports the assumption. It is shown that, given basic model parameters, predictions of duration-of-residence distributions are quite accurate; and, conversely, duration-of-residence data can be used in conjunction with the independence assumption to estimate model parameters. The implication of these findings for the design of mobility studies is discussed, and a question raised as to the necessity of gathering detailed histories.