The harmonization of international census microdata for demographic research: the IPUMS project

In recent years, international comparative population research has expanded steadily as more and better demographic data have become available. Scholars are increasingly able to compare not just aggregated summary data, but individual-level microdata, across countries. But this promising area of research is hindered by uneven access to data, differing survey instruments, uneven documentation, and incompatible technical specifications. Even when individual researchers can overcome these difficulties, the result is substantial duplication of effort and limited reproducibility of results. The Integrated Public Use Microdata Series (IPUMS) is designed to address these issues for world population census data. The data series contains 111 national census samples from 35 countries and has agreements with 40 additional national statistical offices. The data are consistently coded and documented across countries, and a web-based extraction system allows users to pool selected variables and censuses for downloading and analysis. Since the project began ten years ago, the IPUMS research team has been forced to confront numerous technical and intellectual challenges related to international harmonization. This paper describes some of the key innovations, the reasoning behind them, and the implications for research. Some persistent difficulties true for the census samples -- and to some degree inherent to all harmonization efforts -- are discussed. Finally, we present some examples of research results that suggest the possibilities and perils of cross-national research using harmonized data.