Creating a Longitudinal Data Infrastructure at the Census Bureau

Catherine Massey, U.S. Census Bureau
Trent Alexander, U.S. Census Bureau
Amy B. O'Hara, U.S. Census Bureau
Todd K. Gardner, U.S. Census Bureau

In-depth analyses of population dynamics require longitudinal data spanning multiple decades. The Census Bureau is collaborating with the Minnesota Population Center to create a core set of linkable census, survey, and administrative records that would provide data on the American population across seven decades. The core linkable data files consist of the 1940 Census, the 2000 Census, the 2010 Census, the American Community Survey, and the Current Population survey. Using the core as a foundation, researchers can then integrate administrative records or other external data sources, depending on their research objectives. In this paper, we discuss the development of this project and provide an overview of the record linkage techniques that enable the creation of longitudinal data of this magnitude. We also report our progress on building a linkable version of the 1940 Census, as well as the scope of the linked data across the core data sources.