Current Categories and Anomalies

INFORMATION ON RACE AND ETHNICITY

Of all statistical systems, the decennial census provides the most extensive experience with racial and ethnic data. The 1980 and 1990 censuses each asked several questions about race and ethnicity. Race and Hispanic ethnicity questions were asked on the short form, administered to five-sixths of all households; these and additional questions on country of birth, citizenship, and an open-ended question about ancestry were asked on the long form, administered to one-sixth of U.S. households. Table 3 shows the short-form and long-form race, ethnicity, and ancestry questions from the 1980 and 1990 censuses. All three sets of questions allowed write-in responses.1 Even with the write-in option, the nonresponse rates in the 1990 census were 2 percent for the race question, 10 percent for the Hispanic-origin question, and 10 percent for the ancestry question. For the approximately 5 million Americans about whom the race question was not answered, their race was imputed using the standard census “hot-deck” techniques for nonresponse.2

Of the 9.8 million people reported in the “other race” category, the Census Bureau estimates that some 97 percent were Hispanic. Part of the nonresponse to

1  

As noted above, the Census Bureau obtained a waiver from Directive 15 in order to include the “other” category in the race question.

2  

In this procedure, a missing characteristic for a respondent is found by scanning neighboring households for other persons who are similar on a set of other characteristics and then randomly selecting the missing characteristic from them (see Ford, 1983).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification Current Categories and Anomalies INFORMATION ON RACE AND ETHNICITY Of all statistical systems, the decennial census provides the most extensive experience with racial and ethnic data. The 1980 and 1990 censuses each asked several questions about race and ethnicity. Race and Hispanic ethnicity questions were asked on the short form, administered to five-sixths of all households; these and additional questions on country of birth, citizenship, and an open-ended question about ancestry were asked on the long form, administered to one-sixth of U.S. households. Table 3 shows the short-form and long-form race, ethnicity, and ancestry questions from the 1980 and 1990 censuses. All three sets of questions allowed write-in responses.1 Even with the write-in option, the nonresponse rates in the 1990 census were 2 percent for the race question, 10 percent for the Hispanic-origin question, and 10 percent for the ancestry question. For the approximately 5 million Americans about whom the race question was not answered, their race was imputed using the standard census “hot-deck” techniques for nonresponse.2 Of the 9.8 million people reported in the “other race” category, the Census Bureau estimates that some 97 percent were Hispanic. Part of the nonresponse to 1   As noted above, the Census Bureau obtained a waiver from Directive 15 in order to include the “other” category in the race question. 2   In this procedure, a missing characteristic for a respondent is found by scanning neighboring households for other persons who are similar on a set of other characteristics and then randomly selecting the missing characteristic from them (see Ford, 1983).

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification TABLE 3 Race, Ethnicity, and Related Questions on the Census Questionnaires, 1980 and 1990   Census Year   Form 1980 1990 Short Form Is this person White Black or Negro Japanese Chinese Filipino Korean Vietnamese Indian (Amer): print tribe Asian Indian Hawaiian Guamanian Samoan Eskimo Aleut Other: specify Race White Black or Negro Indian (Amer): print tribe Eskimo Aleut Asian or Pacific Islander: Chinese Filipino Hawaiian Korean Vietnamese Japanese Asian Indian Samoan Guamanian Other Asian or Pacific Islander: print race Other race: print race   IS THIS PERSON OF SPANISH/HISPANIC ORIGIN OR DESCENT? No (not Spanish/Hispanic) Yes, Mexican, Mexican-American, Chicano Yes, Puerto Rican Yes, Cuban Yes, Other Spanish/Hispanic IS THIS PERSON OF SPANISH/HISPANIC ORIGIN OR DESCENT? No (not Spanish/Hispanic) Yes, Mexican, Mexican-American, Chicano Yes, Puerto Rican Yes, Cuban Yes, Other Spanish/Hispanic print one group Long Form In what state or foreign country was this person born? Is this person a naturalized citizen of the United States? When did this person come to the United States to stay? Does this person speak a language other than English at home? What is this language? How well does this person speak English? What is this person’s ancestry? In what U.S. state or foreign country was this person born? Is this person a citizen of the United States? When did this person come to the United States to stay? Does this person speak a language other than English at home? What is this language? How well does this person speak English? What is this person’s ancestry or ethnic origin?

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification the Hispanic-origin question is thought to be linked to the order of the race and ethnicity questions. Individuals responding “other race” often wrote in a response that would indicate Hispanic origin, presumably before seeing the separate Hispanic-origin question; that question may then have seemed redundant and so was Converting the write-in responses to conform with the standard (Directive 15) race and ethnicity categories was a vast project for the Census Bureau. The process was slow, costly, and error-prone in the 1980 census; it was somewhat easier and more accurate in 1990 with a computer-automated coding operation. As noted above, there were about 300 codes for race, 600 for American Indian tribes, more than 70 for Hispanic groups, and more than 600 for ancestry.3 The race codes were aggregated to about 40 groups, and then to the standard five categories (four specified and “other”). Data on the Hispanic-origin groups were reported for about 30 detailed groups in addition to the standard Hispanic or non-Hispanic classification. Of particular difficulty in aggregation were the multiracial responses —such as black/white, white/Asian, Scottish/black—of which there were more than 250,000. The Census Bureau reclassified them according to the first group mentioned, although nothing on the importance of order was indicated on the census form or by respondents. Some other race responses were difficult to reclassify even when only one group was mentioned because they have no clear racial interpretation: these included South African, Guyanese, Moslem, or American for the race item and Filipino, Brazilian, and American for the Hispanic-origin question.4 Successive national censuses have responded to changing preferences for particular terms. Efforts to adapt current usage are important, but since usage is not uniform, no single word or phrase can please everyone. The 1990 census tried to provide acceptable options by listing “black or Negro” in the race question and “Spanish/Hispanic” in the ethnicity question. There has been some research on different types of racial and ethnic classification. Experimental work has been done by the Census Bureau, which field tests 3   Race codes included 75 multiple responses. The rarest responses were black Formosan/Taiwanese, black Hmong, and black Sri Lankan, which each had only 1 response. Other rare responses were for octoroon (2), black Okinawan (2), black Burmese (3), black Bangladeshi (4), black Tongan (5), and white Bangladeshi (5). The most common responses were black and white (47,835), white and black (27,926), mixed (32,505), mulatto (31,708), Eurasian (19,190), Amerasian (18,545), white Japanese (9,329), and white Filipino (7,081) (Rolark et al., 1994). 4   In order to provide comparable data for data sets that do not use an “other race” category, the Bureau of the Census allocates “other race” responses according to Directive 15 race groups to create special files, notably the Modified Age-Race-Sex File. The allocation is done by standard “hot-deck” techniques (see note 13, above). This file is used for demographic analysis, estimates, and projections by the Census Bureau and other federal agencies.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification its race and ethnicity question formats just as it does all questions. Census Bureau research reported at the workshop indicates that question format introduces a substantial amount of variability in responses, particularly in nonresponse rates. As noted, the 1990 race question preceded the Hispanic-origin question in the census, and since the race question lacks a box for Hispanic, many Hispanics checked the “other” category for race and then did not respond to the Hispanic-origin question. The Census Bureau has also field tested a survey with the questions reversed: it resulted in a higher response rate for the Hispanic-origin question and no increase in nonresponse on the race question (Martin et al., 1990). Such experimentation and changes in question format can apparently produce more robust measures of race and ethnicity. Little is known about the labels that people use in their own lives or about the sets of classification that people use for others. Preliminary findings from cognitive research and small focus group sessions conducted by the Census Bureau with Asians and Hispanics suggest that there is variability in responses to the race question by nativity (whether the respondent is native or foreign born), English language ability, and national origin. This research has demonstrated that it is often sensible to focus on nationality by groups, rather than to examine all immigrants aggregated as one group. Analysis of immigrant groups will usually require a more detailed approach (e.g., country of origin and recency of immigration) than a simple scheme of reported race or ethnicity. An example of group self-identity can be seen in studies of Caribbean immigrants, most of whom report that they are black in censuses and surveys. Studies of the black population in selected subareas of New York City reveal that the Caribbean-origin population may account for more than one-third of the black population. Clearly, race data alone would not suffice for research on Caribbean children. For research as well as classification, it may be useful to be able to distinguish second-generation Caribbean black children from native-born black children whose ancestors have been in the United States for several generations. Caribbean blacks have different characteristics than native-born African American families, including, on average, higher median incomes, higher levels of educational attainment, and a higher proportion who are employed in professional and managerial occupations (Billingsley, 1993). Self-reported data depend heavily on an understanding of the social conventions of racial definition in the United States. Research provides evidence that immigrants may have self-perceptions and knowledge of racial definitions that differ from those in the United States. Brazilian immigrants to the United States may wonder, for instance, how to report themselves because racial definitions in Brazil are not the same as those in the United States. Representatives of Middle Eastern immigrant groups have also raised questions about how their members should self-report. Immigrants from the former Soviet Union have primarily an ethnic identity, not a racial one, and are often members of ethnic minorities that are not specifically listed in general social surveys.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification CLASSIFICATION PROBLEMS UNDER DIRECTIVE 15 Several drawbacks of the current directive were pointed out by federal agencies and other workshop participants. Many of these drawbacks can be grouped into two major areas. First, Directive 15 has been in some respects too successful in universalizing the five race and ethnicity categories. There has been little research about the substantive meaning or relevance of the categories, and most agencies do not evaluate the racial and ethnic data they collect. The specification of reporting categories has driven the construction of response categories, with agencies, except for the Census Bureau, deferring to the default categories, instead of constructing finer classifications that might be tailored to more specific needs. Second, the current categories and the reliance on self-identification pose problems for many individuals and groups. Some respondents do not identify with or find applicable any of the available categories, while others encounter technical difficulties resulting from the current wording of the directive. For example, persons of Hispanic ethnicity are explicitly assumed to be either white or black and not Asian and Pacific Islander or American Indian and Alaskan Native. 5 There is no race category that includes persons native to Central and South America. There is no race category for blacks who come from areas in the world other than Africa, and there is uncertainty about many persons from northern parts of Africa. Other examples abound. The classification of Brazilians in the current directive is problematic, since they are South American but are not of “Spanish ” culture or origin. Similarly, persons whose nation of origin is Spain are separated from others of European nationality. Inconsistent Categories and Data Sources Standardization of categories is of great importance to agencies and researchers calculating rates of any kind because rates require compatible data for numerator and denominator. Often, however, numerator and denominator data come from different sources. Workshop participants stated that such data are drawn from both federal and nonfederal agencies, as well as other sources. For example, data for demographic estimates at the national level come from vital statistics collected by individual states, which may use different race and ethnicity categories from Directive 15 and from each other. Often, the census is used as the principal source of numbers for the denominators, which can lead to systematically poor estimation if the collection procedures used for the numerator data 5   Data from recent censuses shows that the latter category of Hispanics with American Indian ancestry is quite large: in 1990, 8.4 percent of people who self-identified as American Indian by race also checked Hispanic origin.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification (e g., birth or death certificate) do not gather data in the same manner as the census. Infant mortality rates by race and ethnicity are the most frequently cited example of inaccuracies that result from inconsistent racial and ethnic reporting. Beginning in 1989, births have been categorized by the race of the mother, as recorded on the birth certificate. 6 If the mother does not state her race on the certificate, then the baby’s race is imputed to be the race of the father. If neither parent ’s race is reported on the birth certificate, the baby’s race is imputed to be the race of the mother on the preceding record with known race (National Center for Health Statistics, 1995a:54). There were several reasons for this change in policy. First, U.S. Standard Certificates of Live Birth were modified in 1989 to include more questions that were directly associated with the mother’s health and health behaviors. Moreover, because of increased incidence of interracial births and nonmarital births, it was determined that “tabulating all Births by the race of the mother provides a more uniform approach rather than a necessarily arbitrary combination of parental races” (National Center for Health Statistics, 1995a:54). Racial classification of deaths, unlike births, is designated by an outside party, usually attending physicians or funeral home directors, on the basis of information provided by relatives or their own observations. Their reliance on visual identification or possibly inadequate information from relatives has resulted in an overassignment of deaths to black and white categories and an underassignment of deaths to American Indian and Asian and Pacific Islander categories (Hahn, 1993). As shown in Table 4, with the pre-1989 algorithm for parents of the same race, almost 10 percent of American Indian infant deaths were incorrectly classified even when both parents were American Indians, and the inconsistent assignment of Asian infant deaths varied from almost 20 percent for Japanese to almost 40 percent for Chinese. Changes introduced with the post-1989 algorithm, whereby infants are assigned the same race as their mothers, have failed to reduce the level of incorrect assignment. As a result, the reported infant mortality rates—even when both parents are Asian or both are American Indian—are substantially lower than the actual levels. Table 4 shows that the level of inconsistent classification of infant deaths by race is even greater for children of mixed heritage. The result of this inconsistent reporting is only a slight overreporting of black and white infant mortality rates, since the numbers of births are so large. 6   Prior to 1989, there were a complicated set of rules for classifying the race of a newborn. Newborns with two parents of the same race were classified as that race; newborns with one nonwhite parent were classified as the race of the nonwhite parent. When both parents were nonwhite but of different races, the newborn was assigned the father’s race; except that if either parent was Hawaiian, the newborn was classified as Hawaiian (National Center for Health Statistics, 1995a).

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification TABLE 4 Proportion of Infant Deaths with Race Reported at Death Different from that Reported at Birth, 1983-1985   Parents’ Races as Indicated on Infant’s Birth Certificate Infant Race at Birth Same Different Pre-1989 Algorithma     All Races Combined 1.8 59.5 White 0.8 — Black 3.0 39.7 American Indian 9.8 76.4 Chinese 38.6 68.2 Japanese 19.4 75.0 Hawaiian 3.5 33.8 Filipino 27.8 80.4 Other Asian 29.5 81.7 Post-1989 Algorithm: Newborns Classified as Race of Mother All Races Combined 1.8 56.7 White 0.8 45.5 Black 3.0 37.9 American Indian 9.7 71.2 Chinese 38.6 78.3 Japanese 19.4 86.7 Hawaiian 3.5 32.4 Filipino 27.8 85.1 Other Asian 29.5 84.3 NOTE: For example, of every 100 deaths of infants who were classified as “Other Asian” at birth and whose parents had different races (from each other), 81.7 percent were classified differently at death. a   See text for discussion. Source: Adapted from Hahn (1993) For the smaller racial and ethnic groups, however, the underreporting of infant mortality is quite large. Race- and ethnicity-specific life-expectancy calculations, relying on these infant mortality rates, will produce inaccurate results and have major ramifications for calculations made by Social Security and private insurance industry actuaries, among others. It is important to understand that the inaccuracies in race- and ethnicity-specific infant death rates stem from a combination of three factors. The first is the difference in the reporting source for births and deaths. This is not a product of the formulation of Directive 15, but rather of the reporting processes used by the state and local authorities who produce the nation’s vital statistics. It might be possible to improve consistency in racial and ethnic reporting, but it would

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification probably involve very detailed rules, and it is not clear that extensive guidance should be part of a standard. The only way to ensure that the reported race on a person’s death or birth certificate is the same as in the census would be to have some kind of requirement for consistent individual identification. Such identification, tantamount to an individual race identification system, would be antithetical to the nation’s history of individual privacy and would also contradict current understanding about the fluidity of racial and ethnic identity. The second factor in the inaccuracies is the great inequality in size between groups. Since infant deaths are relatively rare events —about 1 percent of all births—even minor inconsistencies in the classification of deaths may have a major impact on the rates. The third factor is that intermarriage is more common for American Indians, Asians, and Hispanics than for blacks and whites; hence, a higher proportion of children with an American Indian, Asian, or Hispanic mother have a father of a different ancestry (usually white). For these newborns, given the simple categories of Directive 15, there is necessarily inaccuracy in reported race and ethnicity. One can consider the operation of all three factors, for example, for mothers who are American Indians: small numbers of births, often born to mixed ancestry parents, and shifting racial identity over the lifetime. Calculation of mortality becomes extremely problematic in this situation. Overlap of Race and Ethnicity Questions A major problem in the racial and ethnic classification system is the inconsistency regarding Hispanic origin in the two formats permitted by Directive 15. The line between ethnicity and race is conceptually difficult. The directive attempted to provide a compromise, enabling data users and data collectors to choose the best-suited question and presentation formats for their purposes. However, the two formats present serious problems of incomparability and inconsistency. According to workshop presentations, the incomparability of the two question formats appears in two ways. First, although the one-question format addresses the issue of black and white Hispanics, it does not allow the existence of Hispanics who also have American Indian or Asian and Pacific Islander heritage. Possible responses in the two-question format are not even included in the one-question format. Second, by making Hispanic origin mutually exclusive of all four race categories, the single-question format defines white, black, American Indian, and Asian and Pacific Islanders differently than does the two-question format. The ramifications of such an inconsistency are not only theoretical, but can also have important quantitative impacts. For example, in his review of various combinations of race and Hispanic-origin categories based on the March 1991 Current Population Survey (CPS), del Pinal (1992:3) found that Hispanics ac

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification counted for nearly 10 percent of the white category, but that this figure varied widely, depending on the question format used. In the one-question format, the white population could be reduced by as many as 20.5 million Hispanics; in the two-question format, conversely, the white population would include roughly 20 million more Hispanics. Resolution of this inconsistency is not an easy task. Those favoring a single-format combined race and ethnicity question argue that most users attempt to combine race and ethnicity into the five mutually exclusive categories permitted by Directive 15. In addition, they argue that the Voting Rights Act and the state courts have distinguished Hispanics from blacks and whites. In contrast, proponents of the two-question format, who include some Hispanic advocacy groups, argue that Hispanics, regardless of race, face a special climate of discrimination because of language and culture and so should be separately identified. Finally, from the perspective of individuals, it is clear from census responses and from the investigations of researchers that the conception of race by Hispanics is substantially different than that of non-Hispanic respondents. In part, this is because Hispanics tend to use “race ” as a cultural identification and because Hispanics, as a group, are a mixture of blacks, American Indians, and whites (Nelson and Tienda, 1985; Rodriguez, 1974; Rodriguez and Guzman, 1992). Indigenous Peoples Defining or classifying the indigenous peoples of the Western Hemisphere is a particularly difficult task. American Indian tribes, unlike other racial and ethnic groups, have their own membership criteria, which may carry the sanction of law. In addition, federal laws and government-to-government relations vary across American Indian tribes, Alaskan Natives, and native Hawaiians. The current directive defines an American Indian or Alaskan Native as “a person having origins in any of the original peoples of North America, and who maintains cultural identification through tribal affiliation or community recognition.” Significantly, this definition is the only one of the five race and ethnicity categories that includes cultural identification, raising the issue of how to classify someone with Native American ancestry who does not identify culturally with that group. A notable omission from the current standard is individuals with ancestry in the indigenous peoples of South and Central America. The definition of Hispanic includes persons of “Central or South American or other Spanish culture or origin, regardless of race. ” This does not make clear how individuals from South and Central American indigenous populations should identify themselves. This omission reflects the context in which Directive 15 was developed, which was prior to sizable immigration of people from Latin America whose origins are closer to indigenous peoples or other groups than to those of Spanish origin. In addition, the current directive’s wording may create a potential ambiguity,

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification since “North America” usually includes Mexico, but indigenous people from Mexico do not have the same rights under current law as indigenous people from Canada. Native people from Canada have the right to move to the United States without immigration restrictions; native populations from Mexico are not exempt. American Indians American Indians claim not only a distinct ethnic status, but also a distinct political status. Issues of ethnic categorization are intertwined with issues of the sovereignty and national rights of tribes. Much of the data collected on American Indians is specific to those populations and is thus outside the range of Directive 15. “Indian tribal activities” are listed in the directive as the prime example of government activities that are exempt from the federal standard, on the basis that “the specific program is directed to only one or a limited number of race/ethnic groups.” American Indian identity is rooted in a complex mixture of biology, culture, and politics, which together make the definition of American Indians among the most fluid of racial and ethnic groups. From the perspective of self-identification, American Indians have long been a multiracial population. Among Indians enumerated in the 1910 census, only slightly more than one-half (56 percent) reported full-blood ancestry (Snipp, 1989:167). Culturally, there are individuals who identify themselves as American Indians who speak no native languages and have little or no familiarity with tribal culture. Politically, the latest census data provide strong indication that self-identifying Indians are far more numerous than tribal rolls would indicate. The 1990 census counted 8.8 million people who reported at least part American Indian ancestry and 1.8 million as being of American Indian race. Yet according to the Bureau of Indian Affairs, tribal rolls totaled only a little more than 1 million people at the end of 1993. The 1990 census illustrated dramatically the fluidity of American Indian as a racial category: the increase in numbers of people in the category between 1980 and 1990 is explained more by current trends in self-identification than by natural demographic increase (Eschbach, 1993). Except for the census, which is based on self-identification, American Indians face more stringent standards of proof of racial and ethnic status than individuals of other groups in a variety of settings. For example, General Motors requires written proof of tribal membership for reporting of federal and state equal employment opportunity records. Similarly, applicants seeking hiring preference in the Bureau of Indian Affairs and other federal agencies that primarily serve the American Indian population also are required to show documentation of group membership.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification Alaskan Natives and Aleuts According to federal agencies, there have not been many reported difficulties with the classification of Alaskan Natives and Aleuts. Some agencies ask specific questions about these groups for specific uses and then reaggregate the responses into the American Indian or Alaskan Native category. In an analysis of data from the CPS, however, the Bureau of Labor Statistics found that the category of Alaskan Native is subject to misinterpretation if respondents are allowed to self-identify. Persons born in Alaska or who have resided there for an extended time period sometimes come to think of themselves as “Alaskan Natives,” even though their ancestors are not indigenous people.7 Native Hawaiians Native Hawaiians are currently categorized as Asian and Pacific Islanders in Directive 15, but the state of Hawaii has its own special race and ethnicity reporting formats because of its demographic composition. A suggested change in the directive by some native Hawaiian groups is to reclassify native Hawaiians with American Indians or as a separate subcategory along with American Indians under a “native peoples” category. The argument is that native Hawaiians have more in common with American Indians because of their identity as “original peoples of acquired American lands” than they do with the people who immigrate from the Asian continent. However, the native peoples of other Pacific islands could make a similar claim to be included in a revised broader category of indigenous or native peoples. At the workshop, some groups representing American Indians opposed the addition of native Hawaiians to the category of American Indians. Because of Hawaii’s recent statehood and the special relationship between American Indian tribes and the federal government, the legal status of native Hawaiians differs from that of American Indians. Those groups argued that changing the definition of American Indian to include native Hawaiian would reduce data consistency over time without improving data availability on native Hawaiians. Native Hawaiians would continue to be a subclassification of a major group, with major group membership shifting from Asian and Islander to native American. Inclusion of native Hawaiians, and possibly other native Pacific Island peoples, in a larger category of original peoples of American lands, labeled perhaps “indigenous or native peoples” is an example of a potentially new pan-ethnic grouping. 7   Prior to Alaskan statehood in 1959, these indigenous populations were treated as distinct governmental units. The term “Alaskan Native ” was introduced in 1971 in the Alaskan Native Claims Settlement Act. Prior to Directive 15, the Census Bureau counted each group separately; these earlier counts can easily be reaggregated to accord with current federal policy.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification Multiple Ancestries As noted above, racial and ethnic intermarriage is becoming more and more common in the United States. In a recent review of levels of exogamy (marriage to a person of a different race or ethnicity), Edmonston et al. (1994) estimate current rates of 3 percent for blacks, 8 percent for whites, 20 percent for Asians, and 30 percent for Hispanics. One of the key factors associated with the propensity to marry inside an individual’s ethnicity is foreign birth. Foreign-born individuals, including a high proportion of Asians and Hispanics, have lower levels of acculturation, have fewer opportunities to meet potential marriage partners outside the ethnic community, and are often already married when they come to the United States. As a greater number of people who are Asian or Hispanic are born and raised in the United States, the levels of intermarriage are likely to increase in the future. Overall, 5 percent of the U.S. population in the 1990 census reported an ancestry that differed from their primary race or Hispanic identification. By race, 4 percent of whites and 5 percent of blacks report multiple ancestries. More than 25 percent of those reporting their race as American Indian report a non-American Indian ancestry, and about 10 percent of Asian and Hispanic respondents report a non-Asian or non-Hispanic ancestry, respectively. These results support the contention that a significant minority of the population, arranged by current Directive 15 categories, in fact might report multiple ancestries if it was offered. However, empirical research is needed to know how people with multiple racial ancestries would identify if they were given different options for self-reporting. New Ethnic Groups and Pan-Ethnic Identities The United States has historically been fertile ground for the emergence of pan-ethnic identities.8 In the last century and the early part of this one, immigrants from distinct cultural entities in, for example, northern and southern Italy, gradually found the United States conducive to the formation of a new and unified Italian identity. More recent examples of the emergence of pan-ethnicity come directly from Directive 15. The administrative categories of Hispanic and Asian are also modern examples of the blurring of national origin distinctions. Directive 15’s definition of an “Asian race” category does not exist in Asian countries, nor does it correspond to an Asian identification in any other country. The other categorizations —white, black, and American Indian—are also panethnic identities to the extent that they combine highly heterogeneous subpopulations. Today, new sources of immigration and new pan-ethnic identities make it possible that new ethnic groups will emerge. 8   “Pan-ethnic” refers to broad categories that encompass several ethnic groups, such as “Asian” for the many different ethnic groups from that area.

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification Emerging pan-ethnicities pose challenges to those considering revision of the federal classification of race and ethnicity. The constant tension between the relevance and appropriateness of a classification system to individuals and its usefulness in formulating policy will continue. Workshop participants pointed out that the formation of pan-ethnicities is a complicated process, and official categorization is clearly one influence on changing individual and group self-conceptions. One pan-ethnic identity that may now be emerging is that of Arab American or people from the Middle East. Some Arab American groups have taken the initiative toward developing such a category; they point to the lack of available data that would enable study of discrimination and the special needs of Arab Americans (Samhan, 1994). There were slightly more than 1 million Americans reporting Middle Eastern ancestry in the 1990 census. Historically, however, size has not been the sole criterion for adding a category. For example, according to the 1990 census there were more people of Swedish ancestry than Asian or American Indian ancestry, but no one proposes that Swedish be a new, separate ethnic category. Rather, size has combined with social, economic, and perhaps most important, legal factors in affecting the relevance of a category. New categories that seem most likely to emerge are not those from the simple subdivision of an existing group, but from a combination of subgroups subtracted from one or several categories. For example, an Indochinese category—including Americans of Thai, Vietnamese, Laotian, Burmese, and other origins—may become a more meaningful category than combining third-generation Japanese Americans with new immigrants from Cambodia. Yet such subdivisions and combinations may be difficult to justify in the face of the acknowledged heterogeneity in black, Hispanic, and white populations. Workshop participants pointed out some practical considerations that caution against the casual introduction of new categories. First, from a statistical standpoint, the addition of numerically small groups increases the likelihood of unstable estimates of important social, demographic, and economic variables. Second, the addition of new categories may have the definitional consequence of increasing the number of people with multiple ancestries. Third, from a political viewpoint, the creation of new categories shifts people from existing categories, shrinking their size. Fourth, new categories can change the meaning and interpretation of existing categories. Those who advocate the addition of new categories have a difficult task. To demonstrate the relevance of a new category or additional questions, they must show some statistical evidence, but by definition such evidence is not available in the current statistical regime. Historians of government statistics point out that the only long-lasting categories are those whose inclusion is either specified by law or required by federal agencies to carry out programs mandated by law (Robey, 1989). Other categories, such as “Hindu” and “mulatto,” were introduced in earlier censuses but discontinued as their social relevance became diffi-

OCR for page 21
Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification cult to justify. In the absence of adding finer race and ethnicity categories, open-ended questions like the ancestry ones on the 1980 and 1990 censuses are perhaps most useful in identifying the emergence of new social groups, and they offer a means to evaluate the changing self-identities of Americans.9 By using census data from the Public Use Microdata Samples (PUMS), researchers can identify the socioeconomic characteristics of emerging groups on a national and sometimes even a local basis. Finally, it is important to note that some new categories can be incorporated easily under the current directive as subcategories of the broader racial and ethnic groups. It may be possible to build on the permissiveness of the present directive without changing it. 9   Although open-ended questions are currently expensive to process and require detailed coding, analysis of them promises to become less expensive with new computer technology.