Online Program

Measuring Undercounts for Hard-to-Reach Groups

*Mary H. Mulry, U.S. Census Bureau 

Keywords: census omissions, post-enumeration survey, reverse record check, record check, demographic analysis

Measuring census undercount is a major way of gaining insight about subpopulations that are hard to reach. Several methods are available for measuring census undercount, so countries are able to choose the method that best suits their situation. The methods for estimating net undercount include a post-enumeration survey, demographic analysis, record check, and reverse record check. These methods are used to form an estimate of the population size that is believed to be more accurate than the census and compared to the census count to form an estimate of the net undercount.

The methods have different data requirements, so not every country has the data needed to apply each method. Also, even when such data exist, some countries have privacy and confidentiality laws that prevent the use of the data in a manner needed to evaluate a census. So, not all methods can be applied in all countries. Demographic analysis uses vital records in aggregate calculations. The record check and reverse record check require high quality administrative records systems and laws that permit their use in the way required for measuring undercounts. A post-enumeration survey is a second enumeration implemented on a sample basis after a census and then matched to the census on a case-by-case basis. Advantages of a post-enumeration survey include not depending on the availability of an administrative or vital records system. Even when such records exist, their quality and the characteristics of individuals that they contain are not issues. Therefore, the method may be applied in developed or developing countries.

This paper discusses the methods for measuring census undercount and their advantages and disadvantages. Also, included will be a discussion of applications of the methods and the subsequent results that identify subpopulations that are hard to reach for censuses and surveys.