Hearing on Multiracial Identification

Statement of Nancy M. Gordon
Associate Director for Demographic Programs
U.S. Bureau of the Census

Before the Subcommittee on Government Management, Information and Technology
of the House Committee on Government Reform and Oversight

Hearing on Multiracial Identification
25 July 1997


Mr. Chairman and Members of the Subcommittee:

Thank you for the opportunity to testify today on the recommendations made to the Office of Management and Budget (OMB) by their Interagency Committee for the Review of Racial and Ethnic Standards, which included the Census Bureau as one of more than 30 federal agencies. The Census Bureau complies with the guidance to all federal agencies for collecting and reporting data on race and ethnicity that the OMB issues in Statistical Policy Directive No. 15. If, as a result of the Interagency Committee's recommendations and public comment, the OMB makes any changes to the existing directive, the Census Bureau intends to collect and produce data consistent with those changes. We believe that it is essential that there be such standards for use by all federal agencies to ensure that data are consistent and comparable.

The Census Bureau's major role in this process has been to conduct research on concepts and questions about race and ethnicity, consult with a wide variety of data users, and undertake two major tests that helped the Interagency Committee to develop its recommendations. These tests also provide information on the wording and placement of questions for use in Census 2000. I described the results of the first of these tests — the 1996 National Contest Test — when I testified before this Subcommittee on April 23. I have been asked to describe briefly the results of the second test — the Race and Ethnic Targeted Test — in the context of our overall research program. Then I will turn to implications of the Interagency Committee's recommendations for Census Bureau programs.


We are confident that we can collect data on race and ethnicity in conformance with all of the recommended revisions to Statistical Policy Directive No. 15, because we have been undertaking a multi-year testing program of various formats and questions on race and ethnicity. These tests are an integral part of our program to structure the questions for Census 2000 so that they meet the government's need for data on race and ethnicity.

We began our research program for Census 2000 by evaluating the adequacy of data on race and ethnicity from the 1990 census. Although the evaluations indicated the data were of overall high quality, we identified several problems with the questions, mostly related to misreporting, nonresponse, and inconsistent reporting.

We consulted with a broad array of advisors and stakeholders, including members of our four census advisory committees on race and ethnicity; a census advisory committee of professional associations; and the Secretary of Commerce's 2000 Census Advisory Committee. We also sought advice from a diverse panel of experts on data about race and ethnicity; representatives of federal, state, tribal, and local governments; representatives of various groups with differing views of classifying data about race and ethnicity; attendees at a conference on the measurement of ethnicity organized jointly between the Census Bureau and Statistics Canada; and hundreds of data users to whom we have sent our plans for comment.

Over the last several years, we have also conducted congnitive research, focus groups, and classroom experiments to test the questions. Several of the proposals for testing came from the research agenda developed by the Research Working Group of the Interagency Committee. In both the National Content Surve and the Race and Ethnic Targeted Test, we asked the race and Hispanic origin questions in the traditional way (as in the 1990 census) and also alked them with numerous changes in question format, sequencing, and terminology, including several of those being recommended to OMB by the Interagency Committee.

The National Content Survey, which we conducted from March through June 1996, is the principal vehicle for testing and evaluating the full subject content for Census 2000. Its sample consisted of thirteen panels spread over 94,500 households, but our analyses of ways to collect data on race and Hispanic origin were based on four panels with about 4,500 households each. The National Content Survey was designed to provide nationally representative data, but not to collect data specifically to examine small population groups, such as American Indians, Alaska Natives, detailed Asian and Pacific Islander groups (such as Chinese or Hawaiians), or detailed Hispanic orgin groups (such as Puerto Ricans or Cubans).

In order to do that, we conducted the Race and Ethnic Targeted Test in the summer of 1996. This test consisted of a sample of about 112,000 housing units drawn from census tracts with high concentrations of racial and ethnic populations including American Indians, Alaska Natives, Asians and Pacific Islanders, Blacks, Hispanics, and White ethnics. Because of the "targeted" design, the samples are not representative of the total population; however, they are designed to detect differences in responding to various questionnaire designs among those specific populations in areas where they are highly concentrated. It should also be noted that since targeted samples are drawn from the entire population in selected census tracts, they contained a diversity of population groups. That is, a census tract selected because it had a high concentration of Asians and Pacific Islanders might also contain a number of Whites, African Americans, and so on.


Except for changes in terminology, there are no changes recommended by the Interagency Committee to the broad categories of racial and ethnic classification. (There is one proposed realignment that would place Central and South American Indians in the American Indian category.) The four major race categories would continue to be American Indian or Alaska Native, Asian or Pacific Islander, Black or African American, and White and the categories of ethnicity would remain Hispanic origin,and Not of Hispanic origin. In addition, the Interagency Committee has recomended that when race data are tabulated, at a minimum, an additional category — "more than one race" — be added and that more detail be provided whenever possible without compromising confidentiality or data quality.

As in the current Directive No. 15, the recommendations are designed to provide minimum standards for federal data on race and ethnicity. The recommendations continue to permit the collection of more detailed information on population groups to meet the needs of specific data users, provided the additional detail can be aggregated to comply with minimum standards.

Several specific recommendations by the Interagency Committee were based in part on results from the Census Bureau's research program. They include:

  • When self-identification is used, a method for reporting more than one race should be adopted;
  • When the two-question format is used, the Hispanic origin question should precede the race question, when self-identification is used;
  • The name of the Black category should be changed to "Black or African American," and additional terms, such as Haitian or Negro, can be used if desired; and
  • The term "Hawaiian" should be changed to "Native Hawaiian."


I will now briefly report on results from our research program that relate to the Interagency Committee recommendations listed above.

Wording of the Instruction on the Race Question

Perhaps the most interesting finding from the Race and Ethnc Targeted Test was that some respondents provided unrequested multiple responses to the race question even when they were asked to "mark one" box. These percentages were highest in the Alaska Native targeted sample (about 5.2%) and lowest in the Black targeted sample (less than one percent). In the 1990 census, we estimate that perhaps half of one percent of the general population gave multiple responses to the race question when asked to "mark one."

We also tested separate Hispanic origin and race questions and asked respondents to "mark one or more" categories in the race question. In this case, multiple responses ranged from a low of 1.4 percent in the White targeted sample to a high of 10.0 percent in the Asian and Pacific Islander targeted sample.

However, we found that the instruction "mark one or more" races had no statistically significant effect on the percentage who reported a single race of White, Black, American Indian, or Asian and Pacific Islander in their respective targeted samples. The instructions for multiple reporting was not included in the Alaska Native targeted sample because the sample of Alaska Natives in the targeted test was too small.

The Sequencing of the Questions on Race and the Hispanic Origin

Research results show that placing the Hispanic origin question before the race question seems to ameliorate, but not eliminate, two problems that existed in the 1990 census. Asking the Hispanic origin question before the race question reduced the percentage not answering the Hispanic origin question from about 9 percent to about 5 percent in the National Content Survey. This ordering also decreased the percentage reporting as "other race" in the race question from 25 percent to 16 percent in the Hispanic targeted sample of the Race and Ethnic Targeted Test.

Changes in Terminology

Results from the National Content Survey and from other tests indicated that a plurality of Black respondents still prefer the term "Black," but a sizeable number prefer the term "African American." In the Race and Ethnic Targeted Test, we tested the substitution of "Native Hawaiian" in place of "Hawaiian", combined with listing this category immediately after the "American Indian and Alaska Native" category as the first of the Asian and Pacific Islander groups. The changes combined increased reporting of Hawaiians in the Asian and Pacific Islander targeted sample.


We have reformatted the forms we currently plan to use in the Census 2000 Dress Rehearsal (and in Census 2000) to determine the feasibility of accomodating changes recommended by the Interagency Committee, should they be adopted by the OMB. We have, therefore, placed the Hispanic origin question before the race question, have used the instruction "Mark one or more races," and have made the proposed changes in terminology. We were able to do so without any technical difficulties or lengthening of the form. We discussed the proposed form for the Census 2000 Dress Rehearsal, which is planned for 1998, with the Secretary's 2000 Census Advisory Committee, the Census Bureau advisory committees of the professional associations, and the four committee on race and ethnicity earlier this month and published a Federal Register notice about the questions on race and ethnicity on July 17. Public comments will be accepted during the following 60-day period.

We plan to capture multiple responses to the race question with the census data capture hardware and imaging technology, regardless of whether Statistical Directive No. 15 is modified or not. We also expect to be able to capture unrequested multiple responses to the Hispanic origin question. This technology can "read" written characters as well as marked circles. While some technical issues remain about the exact coding of the write-in responses and the exact format of the permanent electronic census file, we intend to maximize the amount of information retained.

As in the past, the Census 2000 will collect more detailed data on race than the minimum required by OMB and those data will be processed in such a way as to maintain maximum flexibility for data users. Census data, including those on race, will be available to users through the Census 2000 Tabulation and Publication Series (the predetermined, standard publication for the census), all of which will follow whatever standards and guidelines the OMB ultimately issues. The Data Access and Dissemination System for Census 2000 (and other data products) will allow even more options and broader access for users to generate customized tabulations. This emphasis on meeting the needs of users in an integral part of the vision for all the Census 2000 topics and products. Selected microdata files also will be available, but the confidentiality of individual respondents will always be maintained.

Mr. Chairman, that concludes my testimony. I will be happy to answer any questions.

Leave a Reply

Your email address will not be published.