WASHINGTON -- Splitting 5-4, the Supreme Court yesterday rejected the Clinton administration's plan to use statistical sampling to make up for individuals who get overlooked in the 2000 census.
The court ruled that the allocation of House of Representatives seats after the next census must be based on a head count.
The decision was a major setback for cities, for the Democratic Party and for a key part of its following: minorities, who are more often missed in counts by census-takers. The "undercount" in the 1990 census left out more than 4 million people, mainly urban minorities and children, according to the Census Bureau.
The justices left open the possibility that sampling could be used to adjust population figures for two purposes -- dividing up $180 billion in federal aid to the states and redrawing election districts at other levels of government.
It ruled narrowly that the census law does not permit sampling during the count every decade that is used to reapportion the 435 House seats among the states. It did not rule on the broader question of whether sampling is unconstitutional.
However, four of the justices in the majority said "a strong case can be made" that sampling is unconstitutional. It would have taken five votes to establish that proposition.
The challenge to the Clinton administration's plan for the first-time use of sampling in the decennial census was made by the House, by four counties and by individual voters in 13 states. The court did not rule on whether the House could challenge the plan but found that the other challengers had made their case.
The court majority decided an actual count -- in person, or by mail with follow-up, in-person contacts -- is the only method allowed by federal census law for House apportionment.
"From the very first census, the census of 1790," Justice Sandra Day O'Connor wrote for the majority, "Congress has prohibited the use of statistical sampling in calculating the population for purposes of [House] apportionment."
To Baltimore's dismay
Baltimore, one of the cities in which minority population is thought to be undercounted in every census, could lose some access to federal money if sampling is not allowed for federal funding purposes after the 2000 census. The city's relative power in the the General Assembly also could be adversely affected.
News of the decision frustrated officials in Baltimore, where one of every four dollars spent in the city's $1.8 billion annual budget comes from the federal government.
City leaders, including Mayor Kurt L. Schmoke, supported census sampling, noting that the majority of Baltimore's population -- 60 percent -- is black.
In the 1990 census, an estimated 4.4 percent of the nation's African-American population was not counted, the biggest segment being males ages 18 to 34.
Baltimore officials say minorities tend to be undercounted because, if they are poor, they do not own property and are less likely to respond to census surveys.
"It's a shame," said Gloria Griffin, a city planner helping to organize a group to ensure a more accurate Baltimore census. "Those poor souls who really need [the federal aid] are not going to get it."
If sampling is done next year for purposes other than House apportionment, it could result in two versions of the nation's 2000 population: one for allocation of seats in the House, a second for everything else.
But sampling can occur next year only if Congress approves the necessary money. Republicans strongly oppose sampling, because it appears to favor Democrats.
Yesterday's decision, because of its limited scope, reignited that partisan controversy as it bears upon the Census Bureau's legal authority to use sampling techniques to calculate national, state and local populations for these key purposes:
Dividing $180 billion in federal funding for social programs -- an allocation keyed to population;
Calculating where to draw the lines for House election districts, once the seats have been apportioned among the states, and districts for state legislatures and local governing bodies -- keyed to population within states.
House Speaker Dennis Hastert repeated the GOP opposition after the court's ruling, saying: "The [Clinton] administration should abandon its illegal and risky polling scheme and start preparing for a true head count."
President Clinton appears to have no intention of abandoning support for sampling within the limits that the Supreme Court ruling may permit. He reiterated his support for sampling in his State of the Union message last week, and the White House noted pointedly yesterday that the high court had not ruled sampling to be unconstitutional.
House Minority Leader Richard A. Gephardt, a Missouri Democrat, interpreted the ruling to mean that the Census Bureau is required to do sampling for purposes of redistricting and distribution of federal funding.
However, Fred Asbell, director of a group set up by members of Congress who oppose sampling, said: "We do not agree" that sampling is required for any purpose.
'Most accurate census'
Census director Kenneth Prewitt said that sampling would not be used to determine the population for House apportionment, and his boss, Commerce Secretary William M. Daley, said the court's decision was being reviewed to determine the design of the 2000 census.
"Our focus will continue to be conducting the most accurate census possible," Daley said at a news conference.
Justice O'Connor's opinion noted that from 1790 to as recently as 1960, the census was taken every 10 years "through personal visits to every household in the nation." That method was not altered, she said, until Congress changed the Census Act in 1964 to permit use of the mail to contact households and do the count.
But no one suggested then that statistical sampling could be used to fill out the population total for purposes of House apportionment, O'Connor wrote. A 1957 change in census law permitted some sampling, the opinion said, but not for purposes of apportionment.
The majority opinion noted that sampling is authorized to collect other data during the 10-year census, such as figures on unemployment, income and housing.
It was only in 1994, the opinion added, that the Census Bureau began arguing that sampling could also be used in determining House apportionment.
Clinton and the Republican-led Congress fought to a standstill on the sampling issue after the Census Bureau proposed that two forms of sampling be used in the 2000 census to get a more accurate count and to reduce the undercount. Congress then authorized lawsuits to test the legality of those sampling methods.
The two methods are:
After getting census forms mailed back, a process expected to provide a count of 67 percent of the households, the bureau would visit enough of the households that did not respond to bring the overall count to 90 percent of the households. Numbers gathered from that process would be used as samples to project the numbers in the still-uncounted 10 percent.
Then, the bureau would do a survey of 750,000 households within 25,000 small census blocks picked at random from 7 million such blocks, to determine age, race, sex and Hispanic origin, along with other demographic factors.
The sampling from those two efforts would then be compared to head count numbers within the sampled tracts to determine how to revise the population totals to correct the undercount.
Two federal courts ruled that the Census Act did not allow these techniques for House apportionment purposes, leading to the test in the Supreme Court.
Joining O'Connor in the majority were Chief Justice William H. Rehnquist and Justices Anthony M. Kennedy, Antonin Scalia and Clarence Thomas. The dissenters were Justices Ruth Bader Ginsburg, Stephen G. Breyer, David H. Souter and John Paul Stevens.
Sun staff writer Gerard Shields contributed to this article.
Pub Date: 1/26/99