Using Population Weighted County Centroids to Calculate Migration Distance for Historical Linked Census Data: A Case Study of Male Migrants in the New York and New Jersey Area, 1880-1910

Sula Sarkar, University of Minnesota
Rebecca J. Vick, University of Minnesota

The availability of complete count historical microdata has permitted the linking of individuals between census years for longitudinal analysis. These linked datasets provide very important research opportunities regarding historical population mobility. In previous work, we estimated migration distance with geographical country centroids. This work underlies the MILEMIG variable in the IPUMS Linked Samples. The geographical centroid assumes that the population is evenly distributed throughout the county. This paper extends and refines that measure by calculating migration distance using population weighted county centroids, which we argue relies on more realistic assumptions about how county populations are distributed. We describe our method of calculating population weighted county centroids. We then compare the distances calculated using geographical versus population weighted county centroids. Finally, we describe the effects of population weighted centroids on the migration distances in one of the IPUMS Linked Samples.