Many American Dilemmas: The statistical politics of counting by race and ethnicity

In recent years Americans have been understandably anxious that their increasingly diverse society is fragmenting along group lines and identities. Now a massive new federal government survey sheds considerable light on such concerns. And while this study offers some comforting evidence that these group identities are hardly as fixed or rigid as we typically assume, it also suggests that their very fluidity will fuel controversies over how and where to draw lines between groups.

The occasion for this survey is the federal government's reevaluation of how it collects and publishes racial and ethnic data. The current statistical regime dates from 1978, when the Office of Management and Budget issued Statistical Policy Directive No. 15 ("Race and Ethnic Standards for Federal Statistics and Administrative Reporting") in an effort to standardize across departments and agencies the volumes of new data required by affirmative action, the Voting Rights Act, and other race-conscious policies.

Explicitly denying any scientific or anthropological authority for its determinations, Directive 15 nevertheless established five basic categories to be used by federal agencies: white, black, Hispanic, Asian or Pacific Islander, and American Indian or Alaskan Native. OMB went on to stipulate that these are racial categories--except Hispanic, which was defined as an ethnic category. In other words, as is often stated in the fine print at the bottom of government documents, "Hispanics can be of any race."

In recent years, this framework has been challenged from a variety of directions. First, there have been objections to the group names--for example, that "Hispanics" should be referred to as "Latinos," or "blacks" as "African Americans." Then, there have been efforts by groups not specifically listed in Directive 15 to be explicitly designated. Thus, Arab Americans have lobbied that a separate category be established for them, while Hawaiians have argued that they are not adequately served under the Asian and Pacific Islander category and are therefore entitled to separate designation. Finally, there has been pressure from those of mixed backgrounds for a new multiracial category.

The focal point for such concerns will undoubtedly be the federal government's largest, most expensive, and most visible data-gathering effort-- the decennial census. The Census Bureau, still smarting from the controversy over its undercount of Hispanics and blacks in 1990, is moving cautiously ahead with plans for the millennial census in 2000. For as will be evident shortly, the revision of ethnic and racial categories will have significant consequences not simply for how groups define themselves, but for how numerous they are.

To address such concerns, OMB established the Interagency Committee for the Review of Racial and Ethnic Standards, which then sponsored a one-time supplement (in May 1995) to the Bureau of Labor Statistics' monthly Current Population Survey. With a sample of about 60,000 households (approximately 100,000 individuals) selected randomly without regard to racial or ethnic origin, the CPS offers a unique opportunity to see how revised racial and ethnic categories would affect the ways all Americans identify themselves. Despite its soporific title, the "CPS Supplement for Testing Methods of Collecting Racial and Ethnic Information" provides an abundance of data that will inevitably fuel the fire under America's melting pot as we approach the millennium.

Racial Self-Identification

Discussions of racial and ethnic identity typically proceed warily, often because it is unclear how individuals from the groups being discussed wish to be identified. Discussion based on the CPS supplement will have sounder-- though not entirely unproblematic-- footing, since all 100,000 respondents were asked how they preferred their racial or ethnic group be identified.

It turns out that Americans who trace their antecedents back to Africa do not necessarily prefer to be called "African American." Only 28 percent in the CPS survey opted for that term, while 12 percent preferred "Afro-American." A plurality of 44 percent preferred to be identified simply as "black," the term that will be used here.

In the same vein, the CPS supplement reveals that 58 percent of Americans tracing their lineage back to Mexico or other Latin American nations prefer to be identified as "Hispanic." Barely 12 percent prefer the currently voguish "Latino."

At its core, the CPS supplement was concerned with the effects of two possible changes to the present list of OMB-approved racial and ethnic categories: adding a multiracial category and treating Hispanic as a racial category, thus transforming Hispanic from its present status as a nonracial ethnic category. The supplement tested the four possible combinations of these two options by dividing the random sample into four equivalent groups, or panels, of 15,000 households.

Respondents in each panel were then asked questions about their racial and ethnic identity based on one of the four possible combinations. In panel 1, questions were based on the present OMB-approved racial and ethnic categories (with Hispanic-origin as an ethnic category and no multiracial category). Questions for panel 2 were the same as for panel 1, but included a multiracial category. Questions for panel 3 specified Hispanic-origin as a racial rather than as an ethnic category and included no multiracial category. Questions for panel 4 were the same as for panel 3, but added a multiracial category.

Table 1 presents the results. It is immediately striking that, at most, 1.7 percent of respondents chose the multiracial identification. It is also striking that blacks and Asians or Pacific Islanders were basically unaffected by this option. The responses of these two racial groups hardly varied in any of the four panels.

At the other extreme were American Indians, Eskimos, or Aleuts; whites; and Hispanics. As other researchers have noted, American Indian identities are extremely volatile and difficult to pin down. Because they constitute a relatively unique and complicated case, however, I cannot explore it here.

On the other hand, the self-identification of Hispanics was a central concern of this survey. And not surprisingly, changing the question format did affect Hispanic self-identification. As table 1 indicates, panels 1 and 2 did not present Hispanic as a racial category. But panels 3 and 4 did, with the result that from 7.5 to 8.2 percent of individuals, depending on the specific format, identified themselves as racially Hispanic.

Table 2 puts this effect into perspective by comparing how many individuals identify themselves as Hispanic either ethnically or racially. These data reveal that the shift from the ethnic (panels 1 and 2) to the racial category (panels 3 and 4) actually reduced the number of individuals identifying themselves as Hispanic. Indeed, overall Hispanic self-identification fell 30 percent from panel 1 to panel 3, 18 percent from panel 2 to panel 4.

But what about the shift in those identifying as "white" in table 1? As a careful reading of the table suggests, white identifiers (as well as those identifying racially as "something else") decreased when Hispanic racial identification was an option. What these data, in conjunction with other findings from the CPS supplement, confirm is that given the option of identifying as a racial group, fewer Hispanics identified themselves as white. Indeed, while substantial numbers of Hispanics identified themselves as racially white under OMB's currently approved categories (with Hispanic as a separate ethnic category), when forced to choose between a white racial identity and a Hispanic racial identity (as in panels 3 and 4), fewer Hispanics opted for white.

How many fewer, and which Hispanics? Table 3 reveals the size of this shift among specific Hispanic subgroups by comparing how Mexicans, Puerto Ricans, and Cubans identified themselves racially in the supplement's four panels. The results are impressive. When Hispanic was not presented as a racial-category option (panels 1 and 2), clear majorities of all three subgroups identified themselves as racially white. But when Hispanic was presented as a racial-category option, the overwhelming majority of Mexicans and Puerto Ricans (85.2 percent and 71.5 percent, respectively) identified themselves as racially Hispanic. Substantial numbers of Cubans also shifted to the Hispanic racial identity, though a majority (58 percent) continued to identify as racially white.

How Will Minority Group Leaders Respond?

What are we to make of these results? Black leaders have already signaled their opposition to a multiracial category, arguing that any such change would undermine black solidarity and siphon off numbers from black group totals. Yet the data presented in table 1 suggest that such concerns are exaggerated and, indeed, that blacks (along with Asians) would be the least affected by any of these proposed changes. Of course, the multiracial category might over time become more popular and, consequently, prove to operate just as black leaders fear. Moreover, faced with pressures against affirmative action and other preferential racial policies, black leaders are likely to view any diminution in their numbers with alarm.

Such black anxieties will only be heightened by the continued growth and visibility of Hispanics, whose emergence as the nation's "other minority" is an important cause of the suggested changes in OMB racial and ethnic categories. Yet the implications of these changes for Hispanics (or for Americans generally) are hardly self-evident.

On the one hand, Hispanic numbers were maximized when, as is currently the case, there was a separate Hispanic-origin ethnic category. Particularly in light of the substantial fall-off (18 or 30 percent) in Hispanic identifiers when "Hispanic" was treated as a racial category (as shown in table 2), Hispanic leaders would presumably reject any such change. Anyone doubting this need only think back to the negative response from Hispanic leaders to the much smaller loss when the 1990 census failed to count approximately 5 percent of their group.

Yet there is another side to this coin. Well over 60 percent of all those identifying themselves as Hispanic in the CPS supplement actually preferred that the designation be treated as a racial category. This preference undoubtedly reflects what has come to be common usage. For despite the fine print at the bottom of government documents, we have gotten into the habit of regarding Hispanics as a distinctive racial category--as when we refer to "whites, blacks, and Hispanics."

This habit has definitely been nurtured by Hispanic leaders, who have fought to ensure that their group is afforded the same racial minority status and resultant racial preferences accorded blacks. With such considerations in mind, Hispanic leaders would presumably favor treating Hispanic as a racial category. Indeed, as we have seen, such treatment dramatically reduced the total number of Hispanics identifying as racially white. But it also reduced the number identifying themselves as Hispanic.

Is Ethnicity a State of Mind?

What of the implications of these survey findings for American society as a whole? I have argued in Mexican Americans: The Ambivalent Minority that Mexican Americans, who constitute almost two-thirds of all Hispanics, are not accurately or prudently regarded as a racial minority group similar to black Americans. For much the same reason, I would argue against proposals to recategorize Hispanic as a racial designation.

Nevertheless, it is not obvious how our existing racial and ethnic categories apply to Hispanics. As table 3 shows, many Mexicans and Puerto Ricans in particular do not see themselves fitting into any of the established racial categories. Thus they opt for a vaguely defined "other" racial category. This tendency is still another factor in the push for an explicitly multiracial category. But Hispanic leaders, like their black counterparts, would not likely approve of such a change.

In response to such convoluted findings, many Americans may well recoil in dismay and conclude that the whole effort to count by groups--whether ethnic or racial--is a divisive farce, a power grab abetted by social engineers. I empathize with such reactions but consider them excessive. Despite our self-conscious individualism, Americans have often identified with various subnational categories and groups--whether ethnic, racial, or regional. Though sometimes carried to unfortunate and even dangerous extremes, such group ties and identities have usually served as bulwarks against the rootlessness and anomie of modern life.

In any event, the continuing drive for equality will not permit us to dispense with group categories. The fact is that for today's disadvantaged (as it was for yesterday's) some form of ethnic or racial identity will be the basis of their aspirations for a better life and inclusion in the American dream. Acknowledging this, however, does not mean we have to accept the extremes of multiculturalism and group entitlement that now threaten to overwhelm us.

The dilemmas posed here will not be soon--or easily--resolved. But a few useful insights can be derived from the findings of this intriguing survey. We should bear in mind that there are limits to our efforts to respond to demands for individual choice in this, as in other, spheres of social life. The distinguished Harvard historian Stephan Thernstrom has criticized the increasingly subjective basis of our ethnic and racial data. As he puts it, we've made "ethnicity a state of mind." He continues: "It doesn't matter if you don't think I look Chinese. I feel Chinese; ergo I am Chinese." It is worth recalling that our emphasis on subjective self-identification is in part a reaction to our past, when we did not afford individuals any such choices about the ethnic and racial groups to which they were deemed to belong. Nevertheless, I agree with Thernstrom that we have now gone too far in the other direction. But in any event, the reactions of black and Hispanic leaders to the proposed multiracial category indicate there are limits to individual choice and subjectivity in this realm.

Yet there is much more encouraging news to be discerned in these survey data. As I indicated at the outset, the very fluidity of these categories--at least for some groups--highlights that the group boundaries and identities with which we are preoccupied are not as rigid as we usually think. Certainly not for Hispanics, soon to be the nation's largest minority. On the other hand, as my analysis strongly suggests, the very fluidity of these lines--along with the insecurities generated by such changes--will be the grounds on which many political battles will be fought. Therein lies the hope, the reality, and the challenge of racial and ethnic relations in the United States at the end of the 20th century.