The Leadership Conference on Civil and Human Rights

Volume 9 Number 1

CENSUS 2000 APPROACHES

The 2000 decennial census is three years away, and several key decisions affecting the accuracy and utility of the count must be made by the 105th Congress.

Background

The United States Constitution requires the Congress to provide an "actual enumeration" of the "whole number of persons within each state," every ten years and the 2000 census will be the twenty-second national count. While the primary reason for the col lection of census data is the apportionment of representation in Congress, the data also provide the statistical basis for government planners, policy advocates and private industry to shape future domestic policy. Some of the policy decisions affected b y census data include:

• the number of congressional representatives apportioned to each state;
• the number of electoral college votes designated to each state;
• congressional as well as state and local redistricting; and,
• compliance with various civil rights statutes like the Voting Rights Act of 1965, affirmative action, employment, housing, lending, environmental racism, and education anti-discrimination laws.

These data also serve as the basis for the annual distribution of billions of dollars in government funds. Among the issues of concern to the civil rights community are: 1). the habitual census undercount of people of color and low income people; 2). the possible addition of a new "multiracial" category; and 3). whether the funds appropriated to conduct the census will be adequate.

Sampling and the Undercount

One of the most significant issues yet to be resolved regarding the 2000 census is whether sampling techniques will be employed for the first time to achieve a more accurate demographic count of all the nation's people. Counting every individual residin g in the United States is a very difficult endeavor and while the census historically undercounts the population in general, people of color and poor individuals are disproportionately more likely to be undercounted.

The 1990 undercount of racial and ethnic minority groups, referred to as the "differential undercount," was the highest ever recorded. The estimated "differential undercount" was 4.9 percent for Blacks, 5.2 percent for Hispanics, 3.1 percent for Asian Pa cific Islanders, and 5.0 percent for Native Americans, while Caucasians were undercounted by only 1.7 percent. There are several reasons why people of color and the poor are consistently and disproportionately undercounted by the census including:

• mail and door-to-door collection methods have lower response rates in lower income areas;
• lower education levels, illiteracy, or difficulty with the English language affect the ability of many individuals to understand the census;
• a general misunderstanding of the importance of census participation; and,
• distrust or suspicion of government leading to the fear that the census may be used by immigration and/or law enforcement officials to deport or incarcerate or may disqualify one for social welfare programs.

The issue of the undercount was addressed by the Supreme Court in 1996 in Department of Commerce v. New York. A unanimous Court ruled that the Secretary of Commerce was not required to statistically adjust the 1990 census, despite the effect of the under count on minorities and the poor, and the availability of adjustment and sampling techniques to provide a more accurate count. The Court reasoned that the Secretary's decision not to adjust "was well within the constitutional bounds of discretion over th e conduct of the census provided to the Federal Government." The Commerce Secretary's decision not to adjust the 1990 census set the stage for the current debate.

Following the 1990 census, Congress directed the Census Bureau to develop methods to improve the accuracy of the census while also reducing its costs. On February 28, 1996, the U.S. Department of Commerce and the Bureau of the Census unveiled "The Plan F or Census 2000," which calls for sampling procedures to be used in two separate instances: (1) to sample and estimate the 10 percent of the population that fails to respond in the actual enumeration and (2) to use a separate sample of houses to estimate t hose persons missed in the actual enumeration and the sample for nonresponse and revise them accordingly.

Hearings were held on this matter in the 104th Congress. Among those supporting the use of sampling and statistical techniques to supplement an aggressive direct enumeration effort were three panels of experts convened by the National Academy of Sciences who testified that some form of sampling was the only way to reduce the differential undercount while concomitantly containing the costs of the census. In addition, Dr. Barbara Bryant, Director of the Census Bureau under President Bush, asserted in her testimony that the census had reached the limits of what could be done with traditional methods. Representative Thomas F. Petri (R-WI) testified against using sampling to adjust the final count, stating, "To rely on sampling rather than the final census count would be comparable to changing election returns if they are at variance with public opinio n polls." Rep. Petri introduced legislation in the 104th Congress preventing the Census Bureau (or any federal agency collecting data) from employing sampling methods to determine the total population of the states. While no action was taken on the legi slation, a spokesman from his office said the Congressman would introduce similar legislation in the 105th Congress.

In the end, in 1996, the House Committee on Government Reform & Oversight recommended against the use of sampling for the 2000 census in its report on the issue, entitled, Sampling and Statistical Adjustment in the Decennial Census: Fundamental Flaws (Hou se Report 104-821). While the Committee's recommendation is legally non-binding, it represents the first time a House of the Congress has gone on record in opposition to the use of sampling to adjust the census undercount.

The Addition of a Multiracial Category

The most controversial matter regarding the 2000 census is whether a new "multiracial" or "multiethnic" category will be added to the racial and ethnic classification question. Since 1977, the Office of Management and Budget (OMB), the federal agency res ponsible for determining racial and ethnic classifications, has recognized Hispanic as one ethnic category, and Black, White, American Indian or Alaskan Native, and Asian or Pacific Islander as the four racial categories that make up the U.S. population.

The argument for including a multiracial category has been advanced primarily by Project RACE, a national organization advocating on behalf of multiracial children. Susan Graham and Chris Ashe founded the organization in 1991 after realizing "there was n o place for their multiracial children on the 1990 census." At a hearing before the House Subcommittee on Census, Statistics, and Postal Personnel, Ms. Graham noted her son was considered "White" on the U.S. Census, "Black" at school, and "Multiracial" a t home, all at the same time. She claims a multiracial category is needed to reflect changing American demographics, and provide a recognized identity for those individuals who are of mixed parentage. Omission of such a category, she claims, violates th e Equal Protection Clause of the Fourteenth Amendment.

While sympathetic to these sentiments, civil rights advocates contend adding a "multiracial" category would further erode representation and political influence for people of color and low-income individuals. Studies show a switch in data classification from "Black" to "multiracial" would be disastrous to the goal of equal protection before the law because of the important role census data play in ensuring compliance with the Voting Rights Act of 1965, affirmative action, and other anti-discrimination la ws. Moreover, opponents note a majority of all Americans are technically multiracial and/or multiethnic. Adding such categories would only generate confusion by constructing a category a majority of Americans could check.

By early this Summer, OMB will determine whether a new, "multiracial" category will be included. In the meantime, Representative Thomas F. Petri (R-WI) introduced legislation he sponsored last Congress, H.R. 830, which would require a "multiracial" or "multiethnic" category in Federal agency collection of data on race and ethnicity. The bill was referred to the Committee on Government Reform and Oversight.

Funding

The funding appropriated for collection of the census ultimately will determine the success of the 2000 census. As politics continues to play a significant role in congressional decision making, the 2000 census has become a major pawn. The Census Bureau 's plan has a $4 billion price tag for the decade-long census cycle. Further adjustments in sampling could reduce the cost to $3.7 billion. However, the Congress appears to be intent on preventing the Census Bureau from using sampling to adjust the coun t thereby adding millions of dollars to the overall cost. This budgetary situation has led the 2000 census to be designated on the General Accounting Office's (GAO) "High-Risk Series," which is a report produced at the beginning of each Congress designat ing those Federal activities and programs GAO believes have a high risk of failure.

At a recent budget hearing before the Subcommittee on Commerce, Justice, State, the Judiciary and Related Agencies, Frank DeGeorge, Inspector General at the Commerce Department, noted sampling must be part of the design in order to stay within budgetary c onstraints. He also stressed the importance of adequately funding census activities now, calling the Bureau's request for a significant increase in funds an "investment" in a good census.

Additional hearings on all of these matters are scheduled for the second week in March. The MONITOR will keep you posted as updates become available.