The theory of counting heads vs. one, two, three

Under the constitution of 1789, black slaves were counted as three-fifths of a person. In the 1990 Census, blacks and other minorities were counted as 95 percent of a white person.

The problem isn't a lingering constitutional defect, but imperfections in census methodology that result in racial and ethnic minorities being undercounted relative to whites. There are means for reducing this undercount, if the barrier of partisan politics can be swept away.

According to Census Bureau follow-up record-checks and sampling indicators, the 1990 Census missed 8.4 million people, while incorrectly including 4.4 million. The net undercount of about 4 million people disproportionately affected blacks and Latinos, whose undercount was about 5 percent greater than whites.'

A similarly disproportionate undercount has afflicted minorities at least since the Census Bureau began officially measuring the undercount in 1940. The result has been longterm, cumulative harm to the least affluent, least privileged people of our nation. The differential undercount deprives minorities of federal aid distributed by population and of representation in legislative districts that are apportioned according to population counts.

As affirmed by a panel of the National Research Council of the National Academy of Sciences - America's premier scientific organization - scientists have developed sampling techniques that correct the census to count the population more accurately and reduce the differential undercounting of minorities.

Results of a dress rehearsal for the 2000 Census that the Census Bureau conducted last year affirm the value of sampling. The bureau found "that data showed across-the-board that the undercount which has been measured in every census since 1940 persists today, but that scientific methods used at two of the three [dress rehearsal] sites corrected for it."

Unfortunately, the issue of how best to count our population is now entangled in partisan politics. Republicans believe that rectifying the undercount of minorities through sampling methods would harm their political interests by counting additional Democratic-leaning persons.

Opponents of sampling say that the Constitution requires a literal and complete head count of the entire population, not the creation of so-called "fictitious people." Critics also charge that sampling subjects the census to political manipulation. Neither argument withstands scrutiny.

American census-takers have never taken a complete head count of the population.

As the panel of the National Research Council reminds us: "It is fruitless to continue trying to count every last person with traditional census methods of physical enumeration."

Despite technological improvements and increased funding, the 1990 undercount was greater than the undercount in 1980.

Throughout history, the Census Bureau has used methods other than actual enumeration. If census-takers fail to contact people in a dwelling they believe to be occupied, the Bureau may assign people to the dwelling through secondhand reports from neighbors, mail carriers, and building supervisors. The bureau "counted" about 7 million people by such indirect procedures in the 1990 Census.

The use of sampling, moreover, reduces the threat of fraud or manipulation of census counts. Any such interference with the census will likely show up in departures from the expected statistical distribution of sampling data.

Republicans, and the conservative Southeastern Legal Foundation, sued the Clinton administration to block the use of sampling in 2000. Justice Sandra Day O'Connor, writing for a sharply divided Supreme Court, sidestepped constitutional arguments and ruled that the census statute (which Congress could amend) prohibited sampling for the purpose of allocating congressional seats among the states.

But Justice O'Connor also observed that the statute appears to require the use of sampling for all other purposes, including the distribution of federal aid and the drawing of congressional, state, and local districts within states. Her opinion cited the National Research Council's finding that without sampling, the "differential undercount cannot be reduced to acceptable levels at acceptable costs."

Republicans have followed their lawsuit by introducing legislation in several states designed to block the Census Bureau from using sampling even for purposes mandated under current law. They have also announced plans to mount a public relations campaign against sampling that would resurrect their constitutional argument. If these efforts to thwart an accurate census succeed, the result will be the worst kind of gerrymandering: The votes of more affluent whites will count more than the votes of racial and ethnic minorities.

It took a Civil War and the 14th Amendment to the Constitution to invalidate the notorious three-fifths clause of the Constitution.

It would be a sad twist of history if the Constitution became a foil for blocking efforts to reduce the continued undercounting of minorities in America's crucial census tallies.

*Alan J. Lichtman, who has been an expert witness in more than 50 federal voting rights and redistricting lawsuits, is chairman of the history department at American University in Washington, D.C.

J. Gerald Hebert, former acting chief of the US Department of Justice voting section, is an attorney in Alexandria, Va.