Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification (1996)

Any decision on revising the federal standard for racial and ethnic classification embodied in Directive 15 must take into account many factors. Some of them involve competing objectives or interests. It is almost certainly the case that there is no standard that will satisfy all the people involved: respondents; federal data collectors and users, including statistical agencies, program agencies, and enforcement agencies; state and local data collectors and users; private companies; and researchers. Thus, the task of those who must decide on revision is to determine the most important objectives and the most feasible way of meeting them. In this section, the key issues surrounding the possible revision of Directive 15—discussed both in this report and in the workshop—are summarized.

Eight major objectives for a federal standard for statistical reporting of race and ethnicity were identified at the workshop:

1. fostering the exchange of statistical reports between agencies;

2. ensuring the availability of data for the monitoring of discrimination against minority groups;

3. designing the system for administrative and statistical records so that the data are reliable even when disaggregated by race and ethnicity;

4. ensuring that the categories are mutually exclusive and exhaustive and that the number of categories be of manageable size;

5. incorporating flexibility so that the standard can be adapted to the context of its use;

6. promoting longitudinal consistency for categories over time;

7. producing relevant and meaningful categories for federal policy purposes; and

8. producing categories that are relevant and applicable to individual respondents.

As is clear from the discussions at the workshop and throughout this report, some of these objectives may well conflict with each other. Workshop participants concluded that the primary purpose of a federal standard is to ensure that statistics gathered by various branches of government are compatible and that information on minority participation in government programs is public and available. This purpose is reflected in objectives (1) and (2).

The need for reliable data, particularly the need to ensure that aggregations are large enough to avoid the unreliable or unusable statistics, has obvious, but often neglected, importance. Objective (3) includes concerns about sample size, methods of identification, and harmonization of definitions and coding methods.

Objective (4), concerning mutually exclusive and exhaustive categories, is perhaps the most difficult to achieve in the face of often competing desires for group or cultural recognition and for logical consistency.

Objectives (5) and (6), for flexibility and continuity of a standard, are perhaps not as critical as the first four, but are still important and could be at least partly incorporated in any revision. The last two objectives, (7) and (8), concerning individual and substantive meaningfulness and relevance, may be very difficult to achieve or evaluate.

The categories used in Directive 15 need to reflect the important racial and ethnic dimensions of U.S. society. Ultimately, the classifications are only useful to the extent that they help capture relevant features of society. As social and political circumstances change, continued attention will be needed to ensure that the categories continue to be useful to agencies and to the public.

Because of the constantly changing nature of both the U.S. population and people’s perceptions of their identity, consideration should be given to procedures for regular review and updating of federal standards for racial and ethnic data. Regular review would also ensure that all participants in the process—respondents, data collectors, data users—have access to the process of review, while at the same time providing a forum for agencies to exchange feedback about the directive and ensure that research results are shared.

Even with attention to all these issues, it cannot be overemphasized that the goals for a federal standard can be in conflict. For example, the desires of small groups to have separate statistical breakdowns for all government statistics can conflict with the need for reliable statistics; the need for continuity in categories for longitudinal analysis conflicts with the equally laudable goal of adapting to

the changing demographic composition of the U.S. population; the desire to monitor discrimination can run counter to the wish to deemphasize racial and ethnic labeling of individuals; the desire for logical consistency in category building can run counter to defining categories within the historical context that has indeed defined groups such as American Indians, blacks, and whites by different criteria.

No federal standard will resolve all of these conflicts. Trying to fully achieve every objective and goal without understanding that compromise is necessary will not result in an improved federal standard.

It is also important to recognize that there are some goals that are inappropriate for a federal standard on racial and ethnic classification. First, the federal classification should correspond to the legislative and administrative agenda for minority protection, but it does not constitute that agenda. Classification does not in itself constitute an administrative or legal status. The current directive makes this explicit, stating that the statistical classifications should “not be viewed as determinants of eligibility for participation in any federal program.” This is appropriate as an objective for the standard, even though in practice the distinction is not rigid. Since the process of the determination of protected groups relies on data gathered under the federal reporting standard, if newly defined minority groups emerge from legislative or legal action, then statistical classification will have to include those new groups.

Second, the existence of a federal standard does not preclude consideration of more detailed categorizations for research and monitoring. The existing directive also states this goal explicitly, but it appears in practice that agencies do limit themselves to the Directive 15 categories. Consideration could be given to ways of encouraging a richer set of racial and ethnic information for particular purposes than is embodied in the current or any revised minimum set of categories.

Third, it must be recognized that the directive cannot serve as a substitute for the collection of other relevant demographic information —such as nativity, parental nativity, and language spoken in the home —that is increasingly important for a range of policy issues related to immigration and demographic changes in the United States.

An unintended side effect of the federal standard is that it has not fostered research on alternative categorization schemes. The result is that little is known, except for some work on Census Bureau data, about the effects of different categorizations on response rates, respondent burden, processing costs, public acceptability, and a number of other variables important to the success of any new standard.

Research is needed both in the testing of current categories and in the thor

ough field testing of any proposed revisions or new categories. Several topics seem especially important.

Inventory of alternative ways of collecting racial and ethnic data. In order to evaluate the current standards and comparability of data across the agencies, an inventory of the various ways that racial and ethnic data are collected now is needed. The inventory should include types of instruments, formats of questionnaires (combined or separate), and level of detail of racial and ethnic classification. It should also discuss the sources of errors that may affect different methods of data collection.

Survey field testing. Field testing of question wording and order is needed to assess their effects on response rates and reliability. The Census Bureau has done some research in this area, but there is a need for research on survey instruments other than the census questionnaires that are used for agency data collection (e.g., school reporting, bank reporting, employer reporting). Key issues of concern include: Do respondents understand the new categories? Do new question formats create new classification problems, with new unknowns? What guidelines can be given for comparing data collected under old and new question formats in order to best ensure data continuity?

Self-identification labels and categories. In trying to understand how people identify themselves, Hispanics and new immigrant populations are of particular importance: current research suggests that these groups implicitly use racial and ethnic taxonomies that differ from the Directive 15 categories. The key concerns are the racial and ethnic labels that people use about themselves, the taxonomies used to classify other people, and the comparability of these taxonomies with existing or proposed classification schemes.

Differences between self-identification and observer identification. Research is needed to assess data comparability between racial and ethnic identifications done by self-reports and observer reports. Such research is likely to improve the use of data based on self-reporting in relation to issues of discrimination, for which classification by others is usually of principal importance. Research in this area might be particularly important in the consideration of a multirace category.

Racial and ethnic classification of immigrants. The largest federal statistical source from which racial and ethnic data are not available is the Immigration and Naturalization Service. Racial and ethnic data on immigrant groups will be important for research on emerging minority groups, as well as for agencies’ assessments of their own programs, but this issue is a very sensitive one. Both the U.S. Department of State (which collects information outside the United States on immigration and applications) and the Immigration and Naturalization Service do not now collect racial and ethnic data and have expressed opposition to doing so. Because race or ethnicity is not a legal criterion for admission to the United States, the agencies do not require such data to process applications for immigration. The agencies worry that if racial and ethnic data were collected at

the time of visa application, there would be a perception that such information was being used in admission decisions.

Finally, we note that current research on race and ethnicity categories is quite fractionated, even within the Census Bureau. In addition to the specific topics noted above for testing and evaluation, there is a strong need for coordination among statistical and research agencies so that their work can have a cumulative effect.

Costs are of course an important consideration in a decision about revising the federal standard for racial and ethnic classification. In addition to federal agencies, local and state governments as well as data users will bear the costs of changes. As noted above, specific costs cannot be identified, but potential costs include those for printing and distribution of new forms and explanations, for training clerical and administrative staff, and for harmonizing data gathered under old and new categorizations. In addition, any change in categorization is likely to lead at least to a temporary increase in classification errors, which will also result in costs.

Clearly, there will be a dramatic difference in the costs of changes that can be incorporated within the current standard, such as adding new subcategories within one of the five major categories, and changes to the existing major categories. If the latter kind of changes are made, harmonization of data gathered under old and new systems would be a major undertaking.

In considering revision of the federal standard for reporting race and ethnicity, it should be noted first that there is a distinction in Directive 15 that is important: the question asked and the classification system. Revisions could be made in either the race and ethnicity questions, or in the classifications used in reporting, or in both. This elementary distinction should be maintained because some of the criticisms described in this report can be addressed by questions used in data collection, and others can be addressed in the reporting. For example, respondents might be asked their racial or ethnic identity in an open-ended question, with the data then coded and reported in the existing Directive 15 categories.

In terms of the questions, there are four major options for change:

• Eliminate the two-question format. The most minor proposal for revision that has been discussed is to combine the current two-question format into a single, unified race and ethnicity question with five categories. A drawback of

this procedure is that it eliminates the possibility of identifying different racial groups among Hispanics.

• Add new race and ethnicity categories to the current set of questions. Additional categories might be added for a specific group—such as Creole, European American, or Brazilian—or for a multiracial identity.

• Collect multiple responses. An alternative to adding a multirace category would be to allow multiple responses for a single individual. Procedures would need to be developed for reporting of such responses: either the current five categories could be kept, with multiple responses distributed among them, or new reporting categories could be added.

• Use open-ended questions. Open-ended questions can be asked in a variety of formats. The 1980 and 1990 censuses, for example, included a set of check-off categories with a write-in option for both the race and Hispanic-origin categories. They also had an entirely open-ended response (no boxes to check) for ancestry. It is also possible to have a combination of open-ended and close-ended questions for different data reporting forms.

For reporting, there are also four major options for racial and ethnic data, from making no changes to complete elimination of such data:

• Maintain the current system. Keeping the current statistical directive in force is one option: there is no value in “change for change’s sake.” Advocates of change should be able to shoulder a twofold burden of proof: that there is a significant problem under the current directive that warrants the work needed to develop and implement a new standard and that the revised system would be an improvement.

• Make minor revisions. Among the minor revisions proposed is to report native Hawaiians in a “native peoples” or “indigenous peoples” category, along with American Indians and Alaskan Natives. Such a reporting scheme would also require a change in the questions to include one for native Hawaiians.

• Add major categories. One of the two major changes that has been proposed is to add major new categories to the current classification system. An example of such a category would be “Arab American” or “Middle Eastern.” Information might be reported separately for a multirace category if a question is added to obtain that information.

• Eliminate racial and ethnic classification. The most radical change that has been proposed is the elimination of collection and reporting of racial and ethnic data by the federal government. The arguments for this position involve the unscientific and fluid meanings of race and ethnicity and the tendency for categories to contribute to racial and ethnic distinctions in the nation’s economic, social, and cultural life. But it must be recognized that the United States does have a unique history in which race and immigration from diverse sources have

played major roles. The elimination of racial and ethnic classifications will not abolish the history of discrimination faced by many groups.

A federal standard for racial and ethnic classification cannot change the nation’s history or the real problems and policy choices it faces. It can provide critical data for understanding those problems and informing the policy choices. Those who must decide whether and how to revise Directive 15 will have to carefully consider what such a standard can and cannot do and weigh the important and often conflicting values and objectives inherent in any federal standard for racial and ethnic classification.