The resignation last month of a top Census Bureau official has brought to a boil the long-simmering dispute over whether a statistical adjustment should be made in the 1990 census to correct for the undercounting of minorities.
At stake in the dispute are not only the number of U.S. Representatives each state can claim, but also the allocation of more than $30 billion in federal grants, including education programs, whose funding formulas depend on census data.
So political are the ramifications of an undercount adjustment that many here are charging that the Commerce Department’s October decision to forgo one may be based more on the fear of creating Democrat-heavy Congressional districts than on the technical challenges such a maneuver would present.
In an interview with Education Week last week, Barbara Bailar, who resigned her post as the bureau’s associate director for statistical methodology and research last month, said the agency had made a decision in June to go ahead with plans to use an adjustment method. Perfected after the 1980 census, it was intended to ensure that the 1990 census would give a more accurate picture of the number of minorities in the population.
She said her resignation had been prompted by the action of the Commerce Department--the Census Bureau’s parent agency--to overrule that position.
“It was a political decision, and if they had said it was a political decision, I might have lived with it,” Ms. Bailar said. “But they tried to make it look like a technical decision, and I couldn’t live with that.”
Since the 1950 decennial census, the bureau has acknowledged that blacks, Hispanics, and other minorities--particularly those living in inner cities or isolated rural areas--have been undercounted.
Reasons given for such undercounts range from the reluctance or inability of census takers to canvass the areas in which many in these populations live, to economic and social problems, such as fear of eviction, that may make some minorities leery of divulging census information.
After the 1980 census, the bureau estimated that, overall, the U.S. population had been undercounted by between 1 percent and 2.2 percent. For blacks, it estimated, the undercount was from 5.6 percent to 6.5 percent.
The estimates were arrived at through a method called demographic analysis, which adds to the baseline count estimates on births, deaths, immigration, and emigration. But the official census count has never been adjusted to reflect such estimates.
A second method, one involving a sizable post-census survey whose results could then be factored into the original count, was tried after the 1980 census, but yielded widely varying results.
But under the guidance of Ms.Bailar, who is president of the American Statistical Association, bureau officials have worked to perfect this method, which they say could now be used to statistically adjust the 1990 decennial count.
Moreover, the method could allow for undercount corrections down to the local level, a development that could aid some school districts in the determination of their share of federal funds from such programs as the Chapter 1 compensatory-education program.
Ms. Bailar said the decision not to use the adjustment method was made by a top Reagan Administration official, Clarence J. Brown, deputy secretary for the Commerce Department. She said she believed the implications of the decision were obvious.
“If you look at the people who are usually undercounted--blacks, Hispanics, Indians, the homeless, and people who live in urban centers--well, they are not the kind of people who are registered Republicans,” Ms. Bailar said.
Funding Relies on Data
In addition to members of Congress, some of whom have offered bills on the subject, civil-rights groups, organizations representing state and city governments, and some education associations have joined the fight to have the census adjusted.
The reliance on census data for the distribution of federal education grants has attracted to the fray such groups as the Council of the Great City Schools, the Council of Chief State School Officers, and the National Education Association.
“We figure education funds to our cities are under by an average of 5 percent because of the undercount,” said Mike Casserly, legislative director for the Council of the Great City Schools. “In many places, that figure is greater.”
Chapter 1, which provides financial assistance to state and local education agencies serving concentrations of children from poor families, is the program that can be most crucially affected by a census undercount of minorities. As a 1986 report from the Congressionally mandated national assessment of Chapter 1 points out, a greater proportion of minority children are apt to be poor.
Education Department statisticians rely on census data to determine how much Chapter 1 money will go to each county nationwide. “The census poverty count,” said Diana M. Whitelaw, Chapter 1 coordinator for Connecticut, “is the largest and most important factor in the Chapter 1 funding formula. We are dependent on the accuracy of the data.”
Ms. Whitelaw said that in some states, when welfare benefits move a poor family above the federal poverty line, a count of welfare families is added to the formula. But that count does not offset any inaccuracies in the original census data, she noted.
“There are already some problems with Chapter 1 not reaching everyone it should,” said Ramsay Selden, director of the State Educational Assessment Center of the Council of Chief State School Officers. “If the children are not counted properly, then there are some real problems.”
Children Most Undercounted
Mr. Selden pointed out that, generally, the highest undercounts are for minority children. The Census Bureau’s estimates for the the 1980 census, for example, show that black males between the ages of 5 and 9 were undercounted by 6.1 percent.
By comparison, the undercount for white males in the same age group was under 1 percent.
A large number of black children were among those not counted in Detroit, where a minority undercount as high as 12 percent was estimated, according to TerriAnn Lowenthal, staff director for the House census subcommittee.
A Detroit school official who asked not to be named confirmed last week that parent groups and educators there have raised ques4tions about the census data, saying the figures do not reflect the degree of poverty in the city.
“Maybe census takers are hesitant to knock on every door in the inner city, " he said. “I think I speak for the community when I say we would welcome some adjustment.”
Rural Districts Complain
The problem is not confined to big cities, however. William Dallam, Chapter 1 coordinator for Pennsylvania, said the districts in his state that have complained most about inaccurate census counts have been rural ones.
“We struggled hard to get the Census Bureau to adjust their counts for those districts, but they couldn’t do it,” Mr. Dallam said. “So we suggested to the school districts that they encourage teachers to become census takers.”
In addition to Chapter 1, funding formulas used to allocate vocational-education grants for the disadvantaged, Chapter 2 block grants, and mathematics and drug-education grants rely to varying degrees on census data.
Until now, the issue of whether or not the undercount could be corrected rested on the development of a reliable technique for calculating more accurate numbers.
The bureau had developed a statistical method that was considered accurate for large areas. But for small localized areas, an error of just a few people in an adjustment could produce less reliable figures than the original census count.
“We started in 1981 improving that method,” Ms. Bailar said. “And we have.”
Last May, a National Academy of Sciences panel headed by Benjamin King of the Educational Testing Service concurred. It concluded that a census adjustment was technically feasible and urged the bureau to seek funds in the fiscal 1989 budget to proceed with plans for the adjustment.
Ms. Bailar told the annual convention of the American Statistical Association last August that “the consensus of the statisticians from government, industries, and academia is that an adjustment will provide more accurate data on the size, location, and demography of the minority population in this country.”
“It’s time to get on with the job,” she said.
On Oct. 29, Representative Mervyn M. Dymally, Democrat of California and chairman of the House Subcommittee on the Census and Population, introduced legislation that would require the bureau to make such a statistical correction in the 1990 census.
The bill currently has 57 co-sponsors, and companion legislation has been introduced in the Senate by Daniel Patrick Moynihan, Democrat of New York.
But the day after the Dymally legislation was introduced, Commerce Department officials announced their decision that no adjustment would be made. They argued that, although techniques were available to adjust, “there are questions about the validity of their results.”
The department also said that the adjustment’s reliance on a large-scale post-census survey would drain resources and “raise suspicions in the public mind about the reliability and integrity of the census.”
High Political Stakes
But other legislation in the Congress has added layers of complication to the issue--and demonstrated the high political stakes involved.
Bills seeking to disallow the practice of counting illegal aliens in the decennial census have been introduced by Representatives Thomas E. Petri, Republican of Wisconsin; Barbara B. Kennelly, Democrat of Connecticut; and Thomas J. Ridge, Republican of Pennsylvania. They would bar the inclusion of such persons in counts used for the reapportionment of House seats.
In the view of many civil-rights advocates, the issue ultimately is one of equal representation. Because the number of seats a state has in the House of Representatives depends on the size of its population, any shifts occurring between the decennial censuses can result in political loss and gain.
The census bureau’s population estimates for mid-1987, for example, showed New York, Massachusetts, Pennsylvania, and Ohio in danger of losing Congressional seats, while population increases in Texas, Florida, and California put those states in line for additional seats.
But an undercount adjustment--or the exclusion of illegal aliens--could change those estimates. Hence, the flurry of census bills in the Congress--and the threat of suits against the Commerce Department, should the legislation fail, from officials in some affected areas, such as New York State.
Other ‘Seats’ at Issue
But, in addition to House seats, the 1990 census will affect the boundaries of state legislative districts, school districts, and water districts.
At a recent meeting sponsored by the U.S. Conference of Mayors, a coalition of government groups, religious and civil-rights organizations, education associations, and legal scholars was formed to lobby for passage of the Dymally legislation.
To achieve that, the coalition participants said they would try to prevent the issue from becoming a battle pitched along party lines.
But that may not be a simple task. Many Republicans, coalition leaders said at the mayors’ meeting, believe that an adjustment of census data would benefit Democrats at every level of government, because blacks and Hispanics tend to live in areas that vote Democratic.
According to many observers, the courts--not the Congress or the Census Bureau--may be where the ultimate decision on the census adjustment will be made.
A version of this article appeared in the January 20, 1988 edition of Education Week as Tempers Flare Over Plan To Adjust Census Undercounting