Census Method Hurt State, Study Finds

WASHINGTON — California, which would have received $2.2 billion more in federal funds if statistical sampling had been used in the 1990 census, lost more money than any other state, according to a General Accounting Office study.

The study, commissioned by U.S. Rep. Henry A. Waxman (D-Los Angeles), could be used to pressure California’s Republicans to support sampling methods when apportioning federal funds to the states in the coming decade.

A recent Supreme Court decision forbids federal officials to use sampling--a statistical method that projects totals based on a sample--to reapportion congressional districts, requiring instead a literal head count of every American. But the high court’s ruling left unresolved whether the Clinton administration may use sampling in disbursing federal funds to state and local governments.

“California was the state most hurt and it’s inexcusable,” said Waxman, who provided The Times with a copy of the report Tuesday in advance of its formal release today. “Democrats and Republicans from our state ought to be joining together on this issue to make sure we have the most accurate count possible and that our state doesn’t lose money.”

The GAO report is the first government study quantifying winners and losers among the states after the 1990 census. Democrats hope that the study will prod reluctant Republicans in California, Texas and other states that have large undercounts, particularly among Latinos and other minorities, to change their views on sampling.

But GOP leaders in the California delegation said that they are unswayed by the report, partly because they have not seen it, partly because the Democrats are pushing it--and mostly because they do not trust the science of sampling. Republicans so oppose the method that they have threatened to withhold Census Bureau funding if the agency uses sampling.

“Everyone in the delegation is concerned when we’re losing money,” said Dave Bartel, administrative assistant to Rep. Stephen Horn (R-Long Beach), an opponent of sampling methods. “But . . . it’s not a good idea to say sampling is a good idea just because of a report we’ve not seen.”

According to the GAO analysis, California led the nation with 838,000 people not counted in the 1990 census. Focusing its detailed analysis of federal disbursements on 15 state and local programs funded in 1998 based on the 1990 census numbers, the GAO said that $449 million would have been reallocated among the 50 states and the District of Columbia.

Nearly half the total that year would have been shifted from programs and services in smaller states to their counterparts in California. The state “accounted for about 20% of the adjusted population and would receive nearly half of the total reallocated, or $223 million,” the GAO stated.

The $2.2-billion figure for California is for all state programs since 1990.

Waxman, the ranking member of the House Government Reform and Oversight Committee, which has jurisdiction over the Census Bureau, requested the GAO report with an eye toward increasing political pressure on the 24 Republicans in the 52-member California delegation. “They seem to be taking their line from the [GOP] leadership,” Waxman said.

Noting that Republicans fear losing political clout if adjustments are made, Waxman said that the GAO report showed that states with large numbers of minorities were hurt by GOP opposition to statistical sampling.

“The four states that border Mexico [California, Arizona, New Mexico and Texas] accounted for over one-third of the adjusted population and would receive nearly 75% of the total reallocated [federal money], or $336 million,” the study said.

During the decade, Texas was the second-most-harmed state, with 486,000 people not counted. It missed an estimated $93 million.

By contrast, Rhode Island would have received $13 million less in federal aid. Pennsylvania would have been the nation’s biggest loser with an adjusted 1990 census, seeing its share of federal funding shrink by $110 million.

(BEGIN TEXT OF INFOBOX / INFOGRAPHIC)

Winners and Losers

The use of head counts instead of statistical sampling in the 1990 census affected state finances because many federal grant programs use census data to calculate allocations. A General Accounting Office report to be released today documents the difference between what the states received in grants and what they would have been allocated in 1998 if the calculations had been based on sampling.

10 states that would have gained the most money from sampling:

California: $222,833,000

Texas: $93,448,000

Florida: $22,647,000

Georgia: $16,119,000

Louisiana: $11,737,000

Arizona: $10,887,000

Virginia: $9,013,000

New Mexico: $8,554,000

Washington: $8,466,000

Colorado: $7,242,000

****

10 states that would have lost the most money from sampling:

Pennsylvania: $110,396,000

Ohio: $53,230,000

Michigan: $50,602,000

Minnesota: $31,925,000

Missouri: $26,997,000

Wisconsin: $24,119,000

Indiana: $23,667,000

Rhode Island: $13,012,000

Iowa: $11,959,000

Kansas: $8,686,000

Source: General Accounting Office

Researched by Tricia Ford / Los Angeles Times