Backward Step for Census

The Supreme Court’s decision Monday to block the use of sampling techniques to improve the accuracy of the decennial census is akin to insisting that congressional reports be written with quill pens because that’s the way they were done when the Constitution was drafted. California lost out big in terms of House seats and federal funding because 1990 census takers using outdated methods undercounted the state’s population. Now, without the use of sampling methods in obtaining counts that will be used for reapportionment, California could be an even bigger loser in the 2000 census.

While the Constitution requires an “actual enumeration” every 10 years, the Census Bureau has not exclusively used head counts since at least 1940. Instead it has used statistical estimates as a check on the accuracy of the head count itself, adjusting it as required. Used in tandem with the Census Bureau’s traditional counting methods, these statistical techniques are not hocus-pocus; they allow for the most accurate, fairest and most cost-effective count.

Yet a 5-4 court majority invalidated the use of statistical sampling. The 1990 undercount severely hurt Southern California and other major urban areas; it’s estimated that 4 million people went uncounted across the nation. The result in California was that the state lost an estimated $500 million in federal funds and one House seat. It could be shorted as much as $3 billion in federal funds and two additional House seats as a result of next year’s count.

The legal dispute before the high court centered on an ambiguous line in 1976 amendments to the Census Act on the scope of the Census Bureau’s ability to use sampling. At its core, the case involved a debate over the meaning of the 18th century constitutional requirement for head counts, and it thrust the court into the middle of a contemporary partisan wrangle. Republicans take refuge in the wording of the 1976 amendments, arguing that Congress did not intend that an “actual enumeration” for purposes of congressional apportionment would include sampling. The high court Monday agreed with this interpretation. Democrats accuse Republicans of wanting a poorly performed census and of resisting sampling because minorities and the poor--groups most likely to be missed by census takers--tend to vote Democratic.

The court would appear to have decided the question of whether next year’s census can rely on modern sampling to determine congressional apportionment. Absent specific authorization from Congress or a constitutional amendment, it cannot. Congress will now have to spend much more money on what will surely be less accurate methods, such as door-to-door canvassing to find those who don’t mail back the census questionnaire.

Yet the court decision appears to permit the use of sampling to verify traditional population counts for other census purposes, such as for the allocation of federal funding. President Clinton has asked Congress to fund this sampling technique. Congress ought to consent.

The upshot of Monday’s decision could be more confusion. The old head count methods, with their documented flaws, will determine the size and composition of Congress into the 21st century. But sampling techniques might determine federal funding figures.

California is almost certain to lose out on congressional representation in the next decade. It need not lose out again on its share of federal funds. The state’s delegation in Washington can help see to that.