Minding the store: An internal audit program for demographic programs at the U.S. Census Bureau

Abstract

The U.S. Census Bureau collects and disseminates information products on people, businesses and governments. To ensure the information products are at an acceptable level of quality before release, the Census Bureau has developed and codified a set of statistical quality standards. And to ensure that Census Bureau programs are following the standards, the agency has developed several internal audit programs. This paper examines the audit program which covers demographic data, reviewing five years (2011-2015) of administrative data from the audit program. Analysis revealed four most common areas of noncompliance with the quality standards: Measuring and evaluating measures of nonsampling error; conducting statistical and policy reviews; developing timetables for reviewing files; and following policies for archiving information products. Analysis also reveals that full-scale surveys have both a greater level and share of noncompliance than other types of programs. Reviews of surveys, given their scope, also tend to take more time. While statistical programs vary across countries in organization and scope, the results of this research suggest that data programs may benefit from more organizational support on the back end of data production.