Inferring Transition Probabilities from Repeated Cross Sections

This paper discusses a nonstationary, heterogeneous Markov modeldesigned to estimate entry and exit transition probabilitiesat the micro level from a time series of independent cross-sectionalsamples with a binary outcome variable. The model has its originsin the work of Moffitt and shares features with standard statisticalmethods for ecological inference. We outline the methodologicalframework proposed by Moffitt and present several extensionsof the model to increase its potential application in a widerarray of research contexts. We also discuss the relationshipwith previous lines of related research in political science.The example illustration uses survey data on American presidentialvote intentions from a five-wave panel study conducted by Pattersonin 1976. We treat the panel data as independent cross sectionsand compare the estimates of the Markov model with both dynamicpanel parameter estimates and the actual observations in thepanel. The results suggest that the proposed model providesa useful framework for the analysis of transitions in repeatedcross sections. Open problems requiring further study are discussed.