Lawsuits Over Method : Census Mired in Dispute Over Counting the Hidden

In the beginning, the U.S. Census did not even try to count everyone.

The Constitution exempted hostile Indians and said that each slave should be counted as three-fifths of a free person. Historians believe that the first census, conducted 200 years ago, missed about 2.5% of the nation’s inhabitants.

The modern census is not supposed to leave anyone out--but it does.

An estimated 3 million people were missed in 1980. This year, moreover, the Census Bureau is being sued by those who say it ought to count more people and by those who say it is counting too many.

The Census Bureau is used to being sued by groups who want a larger share of the political power and federal money that ebb and flow with population changes. But this time around, census officials had hoped to avoid a fight on at least one front.

Since 1980, the bureau has been trying to perfect a new technique of identifying the nation’s hidden population, including illegal immigrants, homeless people, families living in garages and vans and a host of others with no fixed addresses. The majority of them live in cities and, traditionally, many of them do not get counted.

The Census Bureau’s new technique, crudely stated, involves treating these people like fish--not such a strange idea if you think of a big city as a dark ocean full of wary, elusive creatures. Taking a fish census is simply a way of counting what you cannot always see.

With a favorable review from the National Academy of Sciences, the new estimating technique was supposed to be ready for use in the 1990 census.

Instead, it has become mired in litigation. Rejected by the Reagan Administration, the new technique is now the subject of a lawsuit by several states and cities, including California and Los Angeles, which still want it adopted in time for the next census.

Playing Politics

Those bringing suit argue that a Republican White House is playing politics with the census. Critics accuse the Administration of trying to stem the flow of money and increased political representation to the places that would show the greatest gains in population if everyone was counted. Many of those places would be urban Democratic strongholds.

At the same time, the Census Bureau is being sued by the states of Pennsylvania and Kansas, along with 40 members of Congress, who want illegal aliens removed from the population base used for reapportioning Congress. The two states bringing suit are likely to gain congressional seats if illegals are excluded.

Besides affecting the apportionment of seats in Congress and in state legislatures, the census bears directly on the distribution of more than $30 billion in federal aid. A city’s share of the money is often determined by its number of poor, homeless or unemployed people. In 1980, New York City claimed that it lost more than $50 million a year in government money because its black and Latino population was under-counted.

People Omitted

In 1980, census takers failed to count less than 2% of the entire population of the country. But they admit to missing a much larger proportion of city dwellers, particularly transients and minorities.

In the last census, the average under-count for large cities was 4.5%, compared to 0.5% for the rest of the country.

The 1980 census missed 1.3% of the nation’s white residents but failed to count 6.2% of the black population. The rate of omission for black and Latino inner-city residents has been estimated at 12%. Nationwide, estimated under-counts for black males run as high as 18%.

At least one top Census Bureau official said he did not expect to see a big improvement in 1990 results if the Census Bureau continues to rely solely on traditional methods of counting people.

“No matter how well we conduct the 1990 census we do not expect to eliminate the differential under-count (the difference between the percentage of missed whites and blacks),” said John Keane, the Census Bureau director until he left early in the year as part of the change in administrations.

Traditionally, the Census Bureau has relied heavily on commercial mailing lists to find out where people live.

Critics maintain that those lists are no help in finding people who lack driver’s licenses, telephones, credit cards or checking accounts.

The head of a firm that provided mailing lists for New York City in 1980 told a Congressional committee that his company “had probably not mailed anything to the South Bronx in 10 years.”

The controversial new methodology was designed to count people who are not known to the list makers.

Method Used to Count Fish

The technique grows out of a 100-year-old way of counting fish in a lake. A certain number of the fish are caught, tagged and thrown back. An estimate of the total number of fish is then obtained by comparing the number of tagged fish that show up in subsequent catches with the number of untagged fish. Over time, this method of random sampling also has proved to be a dependable way of estimating the number of human births and deaths in Third World countries where accurate records were not kept.

Applied to the census, sampling would be used to estimate the number of people missed in the initial head counts.

As advocated, the sampling would take the form of a post-census survey, a nationwide canvass of 300,000 households immediately after the census.

Among the targets of the survey would be city neighborhoods where the census habitually under-counts people, especially minorities. The targeted neighborhoods would be subjected to an intense recount. Afterward, the list of recounted residents would be compared with those found in the same area during the census. The percentage difference between the two counts would be projected to other similar areas and, ultimately, provide the basis for estimating the nation’s hidden population.

Advocates of this method say it would eliminate the national under-count and, with it, the inequities that have dogged the census, such as the comparatively high number of blacks who are not counted.

“We would eliminate the racial bias on a national level and also the urban-rural bias (the relatively higher number of city dwellers who are not counted),” said Eugene P. Ericksen, a sociologist at Temple University who has testified on behalf of the lawsuit calling for the post-census survey.

But the survey technique may have its own potential for error.

For David Freedman, a professor of statistics at UC Berkeley, the trouble starts with trying to treat people like fish.

“You can’t tag a person the way you tag a fish,” Freedman said.

In other words, Freedman contended, you cannot be sure that a person interviewed during the post-enumeration survey was or was not first identified during the census.

“Someone uses a different name. An address is written down wrong. Someone moves. Someone doesn’t want to be counted,” Freedman said.

Confusion and error are inevitable, he said. Worse, what seems like a minor mistake confined to one block of one neighborhood would be compounded, he said, if it were factored into a formula for adjusting the national census.

Pending Litigation

There have been lawsuits over the census in the past. But the pending litigation over the post-census survey marks the first time, officials said, that such a fight has caused a rift within the Census Bureau. Two veteran officials who helped refine the new technique resigned after it was rejected. One of them, former Associate Director Barbara Bailar, who spent 29 years with the Census Bureau, has testified in favor of the lawsuit to force adoption of the new technique.

Bailar acknowledged that there were professional differences within the Bureau over the usefulness of the new approach.

“Altering methodology is bound to arouse doubts and debate,” Bailar said, recalling that she opposed using the new technique in 1980, when it was first proposed, because she believed that more work needed to be done on it.

But this time, Bailar said she thinks that the Administration had its own reasons for resisting a methodology she said had been proven effective through years of testing, including a trial run in East Los Angeles in 1986. “The arguments against it were so vague, they left us feeling that they just didn’t want to do it. I guess I ended up thinking that their motivation was political.”

Undersecretary of Commerce Robert Ortner, who oversees the Census Bureau, decided against using the post-census survey to adjust the census in 1990. He has vigorously denied that his motivation was political, and said he made his decision after a majority of ranking Census Bureau officials questioned the reliability of the survey.

“Adjustment would be controversial, even among statisticians. Techniques are available to adjust, but there are questions about the validity of their results,” Ortner told a Congressional committee last year.

Traditional Approach

Without the post-census survey, the bureau will have to rely on more traditional approaches to solving the under-count problem.

The Census Bureau has made strides. It estimates that the number of black people omitted has dropped from 10.3% in 1940 to 6.2% in 1980. During the same period, however, the five-percentage-point difference between the number of blacks and whites missed has remained about the same.

Plans for the 1990 census call for special efforts to reach people who have been left out in the past. Bureau employees will augment their lists of addresses by checking for additional households in person. Census forms will be delivered by hand to the occupants of some inner-city apartment buildings where mailed forms that are left in a common entryway can be easily lost.

The Census Bureau said it is undertaking an ambitious program to make people aware of the census. Going to work earlier than ever before, the bureau has begun enlisting church and civic groups and mayor’s offices across the country in an effort to inform people of the importance of the census.

It already appears, however, that the publicity strategy needs more work.

After rehearsing its advertising campaign in St. Louis and East-Central Missouri, the bureau found this year that one in five black residents still had not heard of the census. A report on the rehearsal also found that nearly one-half of both rural and urban residents did not believe that census information is confidential, and a majority of residents thought that the census is used to locate illegal aliens.

Officials concede that the 1990 census will not be fail-safe.

“We have not eliminated all of the problems,” Ortner told members of Congress.

For example, there is the problem of reaching non-English-speaking residents with a census form written in English. (The form will include instructions in Spanish on how to receive Spanish-language assistance.)

John Reeder, the bureau’s regional director in Los Angeles, talked about the bureau’s self-imposed limits and acknowledged that they may cause some people to be missed.

“We’re not going to open dumpsters to see if there are people sleeping in them,” Reeder said. “We’re not going to open cars. We are not going into buildings where our people would be put in danger.”

Money at Stake

Precisely because the bureau can not go everywhere, say its critics, it ought to employ a technique, such as the post-census survey, which can reliably estimate the number of people who cannot be found.

Plaintiffs in the suit calling for use of the post-census survey plan to argue that cities such as Los Angeles stand to lose up to $12 million a year in aid if the under-count of minorities and other residents is not eliminated.

Leo Estrada, a professor of urban planning at UCLA who is chairing the Hispanic Advisory Committee to the 1990 census, said the consequences of an incomplete census are not only financial.

“The under-count makes it hard for a nation to measure its own progress,” he said. “You don’t know how many people remain in poverty from one decade to the next. You can’t tell how well immigrants and ethnic groups are being assimilated. You don’t know who is getting ahead and who is falling behind.”