U.S. Refuses to Alter Contested Census Figures


Declaring the 1990 count “one of the two best censuses ever taken in this country,” Commerce Secretary Robert A. Mosbacher refused Monday to adjust last year’s population count to reflect the more than 5 million persons estimated to have been missed.

The decision is a blow to California and other states with large minority populations because minorities are more likely to be overlooked in traditional census tallies.

The census figures are used for reapportioning the 435 seats in the U.S. House of Representatives, seats in state legislatures and even city council districts, as well as for distributing federal funds for everything from feeding the poor to running mass transit systems.


If the census had been adjusted, California and Arizona each would have gained one more seat in Congress. Wisconsin and Pennsylvania each would have lost one, according to a study by Election Data Services, a Washington-based political consulting firm.

The decision is also expected to cost California and similarly situated states and cities millions of dollars.

Mosbacher said that the Administration should not abandon its traditional census head count, which the United States has used since 1790, for an estimate based on the latest in statistical science.

“Before we take a step of that magnitude, we must be certain that it would make the census better and the distribution of the population more accurate,” he said.

His announcement is expected to breathe new life into an ongoing court battle by adjustment proponents seeking to force the Administration to use statistics in the compilation of national population counts.

In fact, Los Angeles Mayor Tom Bradley said in a statement issued in Washington that his city would return to court “to protect the interests of the city” and its residents.


Mayors in Los Angeles, New York, Chicago and other big cities with large minority and poor residents initiated a federal suit against the Census Bureau two years ago to compel an adjustment for an expected undercount. A settlement of that suit required Mosbacher to conduct and consider using a statistical adjustment to correct an undercount for minorities in the 1990 census.

Accordingly, the Census Bureau conducted the Post Enumeration Survey--canvassing 171,390 households across the nation. The post enumeration canvass indicated an undercount in the traditional census of about 5.4 million, or about 2.1% of the nation’s total population.

In reaching his decision Monday, Mosbacher rejected the majority opinion of a bureau advisory committee--headed by Bureau Director Barbara Everitt Bryant--that had urged the secretary to adjust the census figures based on the statistical sample.

Mosbacher instead sided with the minority opinion on the bureau’s committee and with four government-appointed representatives on a court-ordered special advisory panel in opposing the adjustment. The advisory panel was split evenly on the issue; four members appointed by plaintiffs in the lawsuit favored the adjustment.

“After a thorough review, I find the evidence in support of an adjustment to be inconclusive and unconvincing,” Mosbacher said during a news conference to announce his decision.

Adjustment proponents reacted to Mosbacher’s decision with predictable disapproval, pledging to continue their fight in the courts to incorporate statistics in the census figures.


“I cannot convey how disappointed I am that . . . Mosbacher decided not to correct the 1990 census from the Post Enumeration Survey,” Bradley said. “The 1990 census--the first census in this century to reflect a larger undercount than the census preceding it--is gravely defective.

“In light of Mosbacher’s decision, we have no other recourse but to return to court immediately in an effort to protect the interests of the city and to ensure (that) our residents receive their fair share of political representation and financial support,” Bradley said.

Leo Estrada, a demographer at UCLA and a member of the court-appointed advisory panel, said that the secretary’s decision went against an adjustment “probably because it was politically difficult” for him to rule in favor of it.

“In my view this process of taking a Post Enumeration Survey was one . . . effort (to obtain an accurate population count),” he said. “It is a disappointment that they took (the second survey) and still did not act on it. It puts the census at a standstill.”

However, Sen. Herbert Kohl (D-Wis.), chairman of the Senate subcommittee with jurisdiction over the census, praised Mosbacher for making “the fairest and most honest judgment that can be made at this time.”

Mosbacher denied that politics influenced his decision. Instead, he said, statistical arguments failed to convince him that the count would be more accurate than the mail-in survey and door-to-door head count.


Before issuing his decision, Mosbacher said that he sifted through a wide assortment of advisory opinions, including about 300 comments generated at public hearings and solicitations, the bureaus’ assessment and separate reports from eight court-appointed experts.

The 1990 Census lists the U.S. population at 248,709,873; the post enumeration count estimated it at about 253,978,000. In California, bureau officials say, the 29,760,021 figure released last year is low by 3.7%, or about 1.1 million, from the PES estimate of 30,888,000.

Moreover, the PES revealed that minorities--notably blacks and Latinos--were most likely to be overlooked in the census. About 5% of blacks and Latinos were undercounted in the traditional census, whereas fewer than 2% of the white population was overlooked.

Mosbacher said that an adjustment would have corrected the undercount in 29 states and produced a less accurate count in 21 other states. But doing so would have meant taking population from one place and giving it to another without fully understanding whether doing so was accurate. The better course was to leave the census intact, he said.

But one of his advisers, Bryant, countered at the press conference that an adjustment was supported by two statistical studies--the Post Enumeration Survey and a demographic analysis of the nation’s population--both of which estimated the population on April 1, 1990, “between 253 (million) and 254 million, not 248.7 million as enumerated” in the traditional census.

Undercounted Millions

The undercount is critical in areas with large minority populations like Los Angeles. The bars below show the estimated size of the undercount by racial group and gender. Table at right shows the actual 1990 census results.


1990 Enumeration

U.S. Total: 248,709,873

Male: 121,239,418

Female: 127,470,455

Hispanic: 22,354,059

Male: 11,388,059

Female: 10,966,000

Black: 29,986,060

Male: 14,170,151

Female: 15,815,909

Asian: 7,273,662

Male: 3,558,038

Female: 3,715,624

American Indian: 1,878,285

Male: 926,056

Female: 952,229

Source: U.S. Census Bureau