A Beginner’s Guide To Creating Small-Area Cross-Tabulations

Haydar Kurban , Howard University
Ryan Gallagher , Northeastern Illinois University
Gulriz Aytekin Kurban , University of Chicago
Joseph Persky , University of Illinois at Chicago

Data Shop

Data Shop, a department of Cityscape, presents short articles or notes on the uses of data in housing and urban research. Through this department, PD&R introduces readers to new and overlooked data sources and to improved techniques in using wellknown data. The emphasis is on sources and methods that analysts can use in their own work. Researchers often run into knotty data problems involving data interpretation or manipulation that must be solved before a project can proceed, but they seldom get to focus in detail on the solutions to such problems. If you have an idea for an applied, data-centric note of no more than 3,000 words, please send a one-paragraph abstract to david.a.vandenbroucke@hud.gov for consideration.

Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent the views of the U.S. Census Bureau or the Bureau of Labor Statistics. The research in this article does not use any confidential Census Bureau information.

This short article introduces two techniques of generating cross-tabulations in small areas (for example, block groups or tracts) for which only univariate distributions are available. These techniques require either a microsample or a cross-tabulation from a larger geographic area (for example, a Public Use Microdata Area). One technique uses hill-climbing algorithms, and the other is based on iterated proportional fitting. In this article, we identify the general characteristics of both techniques. We present and evaluate an example (generating cross-tabulations of households by housing value and number of children enrolled in public school), briefly discuss extensions of both techniques to synthetic population construction, and test the synthetic populations by comparing the estimated microsamples with the actual population.