ABSTRACT
Most studies of immigrant assimilation estimate cross-sectional regression equations in which some outcome of interest, such as wage rates or language ability, is regressed on a measure of immigrant experience, usually time since arrival, while controlling for various indicators of human capital, social capital, and household socioeconomic status. By pooling successive censuses, George J. Borjas has carried out a period-cohort analysis to suggest that self-selection does play an important role in immigration. In each community, 200 households were selected using simple random sampling, with the exception of the two smallest communities where fewer households were chosen. Data for the alternative samples come from a detailed series of questions asked of all household heads about their most recent trip to the United States. Sample characteristics and regression results were compared between the census sample and a cross-sectional sample constructed from an alternative data source to measure the degree of potential bias stemming from underenumeration.
