Make the Census Count

Every decade since 1790, the U.S. government has taken a population count. The process of compiling the census has evolved with a changing nation, from gathering information by horseback to door-to-door questioners to mail-in household surveys. The plan has been to include statistical sampling in the 2000 census in order to better count minorities, immigrants, renters and others who for a variety of reasons are often not picked up by traditional census methods.

Each nose matters, and California benefits from each resident’s presence on the rolls. That’s reason for displeasure with Monday’s ruling by a three-judge federal panel that rejected the use of sampling, agreeing with House Republicans who sued to block use of the valuable statistical tool. Census data is used to draw or realign congressional districts and shape allocations of billions of federal dollars to states. If the decennial head count fails to find elusive citizens, they are inadequately represented politically and in terms of government programs. The solution lies in sampling, say Democrats, who traditionally attract the votes of many of those who tend to be uncounted.

This issue of sampling and representation is front and center in the GOP lawsuit and demands a prompt Supreme Court review. The GOP, led by House Speaker Newt Gingrich, specifically challenged the Clinton administration’s contention that the laws governing the census allow sampling for political apportionment. Meanwhile, the Census Bureau is appropriately proceeding with a two-track preparation, using sampling in one version and excluding it from the other.

Getting the numbers right by using sampling is critical. In the 1970 census, a small undercount was identified through sampling and the census was adjusted accordingly without dispute. There was no sampling in the 1980 census, but sampling conducted after the 1990 census identified nearly 9 million people as missed in the traditional head count--including 1 million in California--and 4.4 million who were counted twice.

The 1990 undercount cost California an estimated $500 million in federal funds and shorted it one House seat. An undercount in the 2000 census could cost California $1 billion in federal funds and two additional House seats.

From day one the dispute over sampling has been about power, not accuracy. A neutral judgment from the Supreme Court is warranted, promptly.