Showdown in Census Feud

House Republican leaders left Washington last week after inserting wording in an appropriations bill that essentially prohibits the Census Bureau from using “statistical sampling” to improve the accuracy of its traditional head count in the 2000 census.

A showdown seems inevitable when Congress returns next month, for President Clinton has threatened to veto the bill unless the language is removed. The bill funds not only the census but the departments of Commerce, State and Justice as well.

The census has been a popular target since 1792, when George Washington issued the first presidential veto in disagreement with Congress’ interpretation of the 1790 census figures. The numbers are critical, for they determine how congressional districts are drawn and federal dollars are distributed.

Demographers and statisticians agree that sampling is the most cost-effective way of preventing a recurrence of the embarrassing errors that afflicted the 1990 census, which missed more than 4 million Americans and counted another 4 million twice. Those errors have hurt California disproportionately, for its undercount of 2.7% was far higher than the average national undercount of 1.6%. Los Angeles had the second-highest undercount of all major U.S. cities, at 3.83% or 138,808 people.

Politics enters the picture because the kind of people whom sampling should catch--those owning no homes--lean toward Democratic candidates, while the folks overrepresented in a traditional head count, those owning multiple homes, tend to favor Republicans.

House Speaker Newt Gingrich is leading the Republican opposition against sampling, calling the census “an issue of great importance to our party.” But while the use of sampling in the year 2000 could cost the GOP some votes, Gingrich’s opposition could cost it more by alienating the two constituencies traditionally slighted when sampling is not used, Latinos and African Americans.

In 1991, Gingrich defended one of those very constituencies in a census undercount, writing a letter urging the Bush administration to use sampling to correct for a low enumeration of African Americans in Georgia in 1990. “If the undercount is not corrected, it would have a serious negative impact on Georgia,” Gingrich wrote, for “minority voting strength would be greatly diluted.”

Gingrich should reread that letter and support the Census Bureau’s current attempt to do what he wished it had done in 1990.

(BEGIN TEXT OF INFOBOX / INFOGRAPHIC)

Census Shortfall

Average undercounts in the 1990 census:

National: 1.6%

California: 2.7%

L.A.: 3.8%