PRB

(December 2008) The U.S. decennial population census attempts to count every person living in every corner of the United States. It provides the numbers that determine how many members each state will have in the U.S. House of Representatives and whether the boundaries of congressional districts must be adjusted because of population change. It establishes “winners” and “losers” in the distribution of federal funds based on population size.

But the census does not find and enumerate every U.S. resident: In the last three censuses, between 0.1 percent and 1.8 percent of the population has been missed. The U.S. Census Bureau can estimate this undercount through comprehensive post-census evaluations and demographic analyses using sophisticated statistical methods. This analysis tells us that some groups—such as black males—are much more likely to be missed than others. Accordingly, the population of geographic areas with a large proportion of these hard-to-count groups (such as Newark, N.J., or Detroit, Mich.) are likely to be undercounted in the census.

Adjustment Issues

The census is not adjusted for this known undercount. Why not, when statisticians are capable of producing such good estimates? This question was addressed by demographer and statistician Kenneth Wachter at a PRB Policy Seminar on Nov. 12, 2008. Wachter, long-time chair of the department of demography at the University California-Berkeley, has worked on census adjustment issues for many years: publishing articles, serving on special panels, and appearing as an expert witness in legal proceedings.

Wachter reviewed the thorny issues surrounding the adjustment of U.S. census results. So far, the Supreme Court has ruled against using adjusted census numbers for apportionment but not against using them for redistricting or the distribution of federal funds. There are currently no plans to adjust the next decennial census, which is less than two years away, but Wachter speculated that the question of adjustment may surface again for the Obama administration.

‘Shares Not Counts’

He pointed out that cities that suffered an undercount because they have a large number of hard-to-enumerate minorities; these cities will not necessarily quality for more federal funds if the census numbers are adjusted. Because the numbers would be adjusted for all cities, an individual city would gain a greater share of the pie only if their undercount is much greater than that of other areas.

Professor Wachter explained the basic approaches for evaluating the census count, and outlined the challenges likely to be encountered in the 2010 Census. He noted the important role of the ongoing American Community Survey in supplementing the information collected on the decennial census form.

About Kenneth Wachter

Wachter served on the Special Advisory Panel to the Secretary of Commerce on (1990) Census Adjustment, as consultant to the Secretary of Commerce on the 2000 Census, and as an expert witness in litigation since 1980. He has published 16 articles on statistical issues in the census. Wachter is a member of the National Academy of Sciences and fellow of the American Academy of Arts and Sciences. He is the outgoing chair of the Committee on Population of the National Research Council. Wachter holds a Ph.D. in statistics from Cambridge University. Among his books are Height, Health, and History and Between Zeus and the Salmon. His research extends across mathematical demography, the biodemography of aging, federal statistical policy, kinship, and microsimulation.

Mary Mederios Kent is senior demographic editor at PRB.

For More Information

Robert Bell and M. Cohen, Coverage Measurement in the 2010 Census (Washington, DC: National Research Council, 2008).

David Freedman and Kenneth Watcher, “On the Likelihood of Improving the Accuracy of the Census Through Statistical Adjustment,” in Science and Statistics: A Festscrift for Terry Speed, ed. D. Goldstein and S. Dutoit (Beechwood, Ohio: Institute of Mathematical Statistics (IMS), 2003).

Kenneth Wachter, “The Future of Census Coverage Surveys,” in Probability and Statistics: Essays in Honor of David A. Freedman, ed. D. Nolan and T. Speed (Beachwood, Ohio: IMS, 2008).