NCPA - National Center for Policy Analysis

Statistician Objects To Sampling

April 19, 1999

Post-census surveys have been used to study undercount since 1940, but have never been used to adjust official population counts. However, the proposed plan for Census 2000 would apply statistical adjustments derived from post-count surveys in sample neighborhoods to every population figure -- down to the level of individual blocks.

However, virtually perfect data would be required, says demographer Kenneth Darga, since small errors would be magnified by statistical adjustments.

But, as in the past, the survey would double-count some people, while missing the same people the census count missed. For example, the 1990 Census appeared to have a net undercount of approximately 1.8 percent, or slightly fewer than 5 million people.

  • Even after a 1991 revision, the apparent undercount rates were still 42 percent too low for black males, 25 percent too low for males of other races, 33 percent too high for black females, and 50 percent too high for females of other races.
  • And although they are nearly identical percentages of the population, the 1990 survey indicated undercount differences of ten percentage points or more between boys and girls.
  • For "blacks in noncentral cities of the Pacific states" the survey indicated the number of boys under ten years of age should be increased 31 percent, while the number of girls should be increased only 6 percent -- a difference of 25 percentage points.

In fact, the Bureau's own unpublished evaluations documented serious problems with matching error, fabrication and ambiguities -- accounting for about 57 percent of the apparent net undercount; outside analysts suggest it was at least 70 percent.

Post-count adjustments based on surveys would introduce new and unpredictable errors into all census data, invalidating comparisons between areas, populations and eras, warns Darga.

Source: Kenneth Darga, "Sampling and the Census: A Case against the Proposed Adjustments for Undercount," Book Summary, March 1999, American Enterprise Institute, 1150 Seventeenth Street, N.W., Washington, D.C. 20036, (202) 862-5800.


Browse more articles on Government Issues