Integrating Canadian and U.S. historical census microdata: Canada (1871 and 1901) and the United States (1870 and 1900).

A t the beginning of the twenty-first century, the field of historical demography has acquired new energy .through the proliferation of historical census microdata projects. Wedding traditional skills and new technology, census microdata projects in Canada, the United States, the United Kingdom, Norway, Denmark, and elsewhere are increasingly making historical and contemporary microdata available to researchers. One promising development in this work is the harmonization of census microdata across both time and space to facilitate comparative historical and demographic analysis. A group of census researchers drawn primarily from North and South America and Europe have formed the International Microdata Access Group (IMAG) to encourage the harmonization of historical census microdata. MAG facilitates the international consultation and cooperation of scholars who wish to unite multiple data files with a common set of comparably coded variables.' In a parallel initiative, the Social Sciences and Humanities Research Council of Canada (SSHRCC) and the Canadian Families Project (CFP), which created a machine-readable sample of the 1901 Canadian census, have sponsored the integration of these microdata with similar microdata from Canada in 1871 and the United States in 1870 and 1900.* The resulting integrated series is known as the Integrated Canadian-U.S. Historical Census Public Use Microdata Series (ICAPUMS), 1870-1901. I undertook this project on the heels of a similar initiative conducted between 1995 and 1997 in which I integrated the 1871 Canadian census microdata with U.S. microdata from 1850 and 1880. Integrating census microdata files from different nations and years enables scholars to compare individual, family, and social behavior across both space and time. Fully harmonized census microdata allow researchers to consider time and nation as independent variables together with others, such as sex, marital status, occupation, and age. Scholars can then analyze the strength with which individual behavior is associated with time and nation versus personal and subnational contextual variables. Such research allows historians to explore the extent to which social, economic, and cultural behaviors transcended national boundaries (Dillon forthcoming). This article describes the challenges of integrating the 1871 and 1901 Canadian and the 1870 and 1900 U.S. census microdata, considering some of the theoretical issues inherent in combining data from two countries that expressed somewhat different intentions for their censuses. The article is also informed by my experiences working with both the Minnesota and Canadian census microdata projects, which have treated their data in both similar and different ways.