Corruption of Census Endangers Apportionment

THE 19TH-CENTURY BRITISH PRIME MINISTER Benjamin Disraeli warned, “There are lies, there are damn lies, and then there are statistics.” Conducting the 2000 census by deliberately not counting at least 10 percent of the U.S. population is a questionable proposition that could challenge the very fabric of our constitutional order.

The Clinton Administration is proceeding with an unprecedented plan to use the methodology known as “statistical sampling” for the census in the year 2000. The plan calls for an actual count of 90 percent of the U.S. population and a statistical count of the remaining 10 percent. In short, this means that approximately 26 million people will be statistically projected.

Why would this set off a constitutional crisis? The U.S. Constitution mandates an “actual Enumeration” by “counting the whole number of persons in each state” to be taken “in such manner as they [Congress] shall determine” for purposes of apportioning congressional representation among the states. Statistical sampling makes a hard, distributive count of the U.S. population an impossibility. More important, the opportunity for partisan manipulation of the census count will become a political reality.

Congress, as the constitutionally appointed guardian of the census, has made clear its will. Title 13, Section 195 of the U.S. Code specifically prohibits the use of statistical sampling for purposes of determining congressional apportionment. And although Congress has properly delegated the census-taking responsibility to the Commerce Department, it has not given away—and indeed cannot give away—the requirement that the census be an “actual Enumeration.” The President, however, along with his Commerce Secretary Bill Daley, who oversees the U.S. Census Bureau, is pursuing the unconstitutional and yet unproven sampling scheme for use in the decennial census that is slated to occur in the year 2000.

Under the Clinton sampling scheme, actual head count figures will be replaced by computer-generated population numbers, which are the “best guess” of bureaucrats in the Commerce Department. Cast adrift on the endless sea of theoreticals, the potential impact of a sampled census on congressional districts, particularly those in states containing large urban centers, is staggering.

In February, 16 plaintiffs from 14 states, along with three counties, brought suit in federal district court in Virginia to challenge the unconstitutional sampling scheme. Southeastern Legal Foundation attorneys, along with attorneys with Cooper, Carvin & Rosenthal in Washington, D.C., launched the suit for the purpose of enforcing both the U.S. Constitution and existing federal law. Several days later, the U.S. House of Representatives, as a body, filed a similar challenge in D.C. federal district court. Both suits will be heard by special three-judge panels in June, and any appeal will be fast-tracked to the U.S. Supreme Court in early fall.

At issue is the 1990 census undercount, which by most reports was 1.6 percent. Urban minorities, Hispanics in border states, migrant workers, and the homeless comprise a significant portion of those hard-to-reach undercounted. At the same time the Census Bureau is preparing to update its computers and mailing lists, hire hundreds of thousands of ethnic enumerators to reach the communities where the undercount is high, and streamline its data collection systems in an effort to count better, its leaders are pushing ahead with the sampling scheme, which a recent GAO report describes as “doomed to failure.”

Arguing that national census accuracy is the issue, Bureau leaders and their liberal congressional counterparts have suggested that sampling is the only method which can adequately compensate for the failure to count everybody. After all, the National Academy of Sciences has endorsed statistical sampling as a legitimate methodology for counting the U.S. population. How could they be wrong?

In Federalist No. 54, James Madison describes the actual head count methodology as the “permanent and precise standard” for census-taking. The Framers of the Constitution understood the temptation for states to inflate their population numbers in order to increase political power in Washington, and implemented the check and balance of direct taxation as a disincentive for fudging population numbers. The Sixteenth Amendment, which instituted income taxation, eliminated the constitutional disincentive. It is therefore more critical than ever that we enforce the constitutional and statutory mandate to protect the objective, actual head count census. This issue, and the pending lawsuits, are about the law, not about the accuracy of statistical sampling.

The integrity of the U.S. census count is every bit as critical to representative democracy as the integrity of the ballot box is to honest and free elections. We do not guess at the number of votes in an election, and we should not guess at the number of people in America. The Clinton Administration’s “best guess” plan offers a poor solution to a legitimate problem, and flies in the face of a clear constitutional and statutory mandate.

Should the 2000 Census be comprehensive and accurate? Of course. Will it reflect the true population of our nation? By law, it must. “Actual” versus “estimated” enumeration is a distinction with significant legal consequences. We must avoid the politicization of the national census. We must encourage national participation in the census, support the Census Bureau in its efforts to improve the count, and win the court challenges against sampling. Indeed, we must count everybody, because in America, everybody counts.


Mr. Glavin is President of Southeastern Legal Foundation.