Abstract
The mover-stayer model (Blumen, et al. [6]) was proposed for explaining timedependent human behavior. In the originally proposed model, it was assumed that a population is divided into two types of individuals, i.e., “movers” and “stayers”. The mover changes states at any time point according to a time-homogeneous Markov chain, whereas the “stayer” does not change states over time. This model is an attempt to explain time-dependent human bchavior in a heterogencous population. In the present paper, the mover-stayer model is extended to the case where the mover changes states according to a Markov chain with time-dependent transition matrices, and a simple method for obtaining the maximum likeidood estimates of the parameters in the model is constructed. Moreover, the model is applied to real data in the sociometric area to illustrate the estimation procedure.
