The Next Census Has to Seek Accuracy, Not Political Gain

California lost, big time, in the 1990 census. The Census Bureau believes that a severe undercount missed 834,000 residents, costing the state a House seat and billions of federal dollars.

To prevent another huge undercount in 2000 and to take a more accurate measurement, the Census Bureau wants to use scientific, statistical, computer sampling techniques to augment the traditional head count. The National Academy of Sciences supports this approach. So does the Clinton administration. But House Republicans plan to block the reform when the census spending bill comes up for a vote later this month. At stake is the potential loss of up to 24 Republican seats in the House, some political analysts say. But the fundamental right to equal representation should not rise or fall on such political stakes.

If all California residents are counted in the next census, the state could gain one or two congressional seats and a larger, fairer share of the billions in federal funds that are parceled out on the basis of population.

Undercounts tend to miss immigrants and ethnic and racial minorities, poor people and children. Transiency is a problem. To count more of the hard-to-reach population, the Census Bureau plans to send out thousands of human counters and four mailings, including forms and reminders. Forms will also be available at post offices, churches, convenience stores, homeless shelters and other public places and through community groups. A toll-free telephone line will serve people who prefer to call in. Census officials claim sophisticated computer software should eliminate double counting caused by duplicate forms. This new community-oriented approach would work even better in tandem with computer sampling.

The House Republican leadership opposes the proposed methodology, which is commonly used in public opinion polling, on the grounds of accuracy, constitutionality and potential for political manipulation. They prefer a physical head count only, which tends to favor married homeowners who live in suburbs--the traditional Republican voter base--over single, transient, minority renters who live in cities. The critics insist that the Constitution specifies an actual enumeration, although the Justice Department in the three past administrations has interpreted that language to allow sampling and the National Academy of Sciences offers scholarly approval.

The purely political stakes are high for both critics and supporters of sampling. The heads the Democrats and Republicans want counted are those represented on their side of the aisle. Still, accuracy, not politics, should be the key test for the 2000 census. Sampling is part of a sound strategy for gaining an accurate count.