APPENDIX
K

Quality of Current Data on Race and Ethnicity

REQUIREMENTS FOR RACE AND ETHNICITY DATA SINCE 1970

In recent times, requirements for race and ethnicity data have changed to meet new demands for federal laws and to adjust to a changing population. For racial minorities and persons of Hispanic origin, such data have been required for a range of federal and state laws pertaining to political representation and equal opportunity. For the 1970, 1980, and 1990 censuses, the Census Bureau devised specific research, testing, and evaluation of items related to race and ethnicity. This was consistent not only with new policy needs, but also because the 1970 census was the start of the use of self-identification on the questionnaire. During this period, the Census Bureau engaged in extensive outreach programs to improve the participation of racial and ethnic minorities in the decennial census. One other factor also increased attention to race and ethnicity data: the removal of national origin restrictions on immigration with the passage of amendments to the 1965 Immigration and Nationality Act resulted in the growth of previously smaller ethnic groups, especially among Asians and Hispanics.

These factors taken together resulted in greater visibility of racial and ethnic groups in several ways. First, there has been a growing number of racial minorities and people of Hispanic origin. Minorities accounted for 1 of 4 U.S. Residents

Much of the discussion in this appendix is taken from papers by McKenney et al. (1993) and Cresce et al. (1992).



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 372
Modernizing the U.S. Census APPENDIX K Quality of Current Data on Race and Ethnicity REQUIREMENTS FOR RACE AND ETHNICITY DATA SINCE 1970 In recent times, requirements for race and ethnicity data have changed to meet new demands for federal laws and to adjust to a changing population. For racial minorities and persons of Hispanic origin, such data have been required for a range of federal and state laws pertaining to political representation and equal opportunity. For the 1970, 1980, and 1990 censuses, the Census Bureau devised specific research, testing, and evaluation of items related to race and ethnicity. This was consistent not only with new policy needs, but also because the 1970 census was the start of the use of self-identification on the questionnaire. During this period, the Census Bureau engaged in extensive outreach programs to improve the participation of racial and ethnic minorities in the decennial census. One other factor also increased attention to race and ethnicity data: the removal of national origin restrictions on immigration with the passage of amendments to the 1965 Immigration and Nationality Act resulted in the growth of previously smaller ethnic groups, especially among Asians and Hispanics. These factors taken together resulted in greater visibility of racial and ethnic groups in several ways. First, there has been a growing number of racial minorities and people of Hispanic origin. Minorities accounted for 1 of 4 U.S. Residents Much of the discussion in this appendix is taken from papers by McKenney et al. (1993) and Cresce et al. (1992).

OCR for page 372
Modernizing the U.S. Census in 1990 and are projected to be 1 in 3 for the 2000 census. Second, there is a growing diversity within minority groups by nativity, national origin, and socioeconomic status. Finally, there is a greater recognition that people have multiple racial and ethnic identities. COLLECTION OF RACE AND ETHNICITY DATA IN THE CENSUS Beginning with the 1970 census, the Census Bureau provided data on race and Hispanic origin to meet several statutory requirements, enacted in recent decades. For all practical purposes, the Voting Rights Act of 1965 and as amended requires the director of the Census Bureau to provide decennial census data, including race and Hispanic origin, for bilingual election determinations. Public Law 94-311 (15 U.S.C. 1516a) requires the collection of Hispanic origin data in censuses and surveys since 1976. Race and Hispanic origin data are provided to the states for legislative redistricting (Public Law 94-171) and to the Department of Justice for the review of redistricting plans. The Census Bureau collects and tabulates race and Hispanic-origin data in compliance with the federal Statistical Directive 15.1 In addition, the Census Bureau provides race and ethnicity data to other government agencies for their use in meeting statutory requirements and program needs. In the past 30 years, the Census Bureau has placed increasing efforts on providing detailed and comprehensive classification of race and ethnicity. For the 1970, 1980, and 1990 censuses, additional categories for race and ethnicity were included in the questionnaires. The Census Bureau implemented research and testing programs and extensive consultations on race, Hispanic origin, and ancestry items in response to requests from different groups and laws in response to growing federal data needs primarily related to civil rights compliance. The Census Bureau has utilized an evolutionary process for developing these categories. In terms of understanding Hispanic status, it introduced a Spanish-origin item on the long form in the 1970 census that was later refined in 1980 and 1990 for the short form. With respect to racial status, the 1980 and 1990 censuses became more inclusive of the Asian and Pacific Islander groups, and the American Indian and Alaskan Native populations. The coding of write-ins for the ''other race" category was introduced in 1980 and 1990 to obtain data on smaller groups and to improve the quality of data. In addition, the 1980 census introduced an ancestry item to replace the "place of birth of parents" question that had been on the census since 1870. The question was designed to reflect a more general ethnic identity into the third and later generations. Use of these items for field tests and actual census enumeration has been accompanied by outreach and promotion programs to minority communities. The Census Bureau has invested considerable resources in editing and coding responses into the current categories. Entries in "other race" responses numbered 9.8 million in 1990. Write-ins from the race item numbered over 300

OCR for page 372
Modernizing the U.S. Census separate race responses and approximately 600 different American Indian tribes. The Hispanic-origin item elicited 70 different origin write-in responses, and the ancestry item, an open-ended item, elicited over 600 ancestry groups. Use of a computer-automated coding operation in the 1990 census greatly improved the quality of operation in terms of decreased clerical errors and reduced costs. Even with this automated system, however, the Census Bureau encountered difficulties in aggregating some 250,000 race write-ins that indicated mixed racial or ethnic identities or nationalities and write-ins without a clear racial connotation. The 1990 census produced unprecedented amounts of data on racial groups and people of Hispanic origin, documenting the transformation of the United States to a multiethnic and multiracial society with 1 of 4 being black, Hispanic, Asian and Pacific Islander, or American Indian and Alaskan Native. It also documented the growth of new immigrants with significant impact on the Asian and Hispanic populations. Of the total foreign-born population, nearly 44 percent arrived between 1980 and 1990. One-third of population growth for this period was due to immigrants. If current levels of immigration continue—with the latest projections of 880,000 net immigrants a year—another 10 million or so immigrants will be added to the U.S. population by the year 2000. The Census Bureau's evaluation of data on race and ethnicity from the 1990 census revealed that broad generalizations about the accuracy of the race and ethnicity data are not appropriate. Various statistical measures suggested that the overall quality of the data was good; however, the evaluations also showed persistent problems related to question wording and respondent's understanding of the questions. These problems are of major concern because: (1) they have a substantial impact on the quality of the data for small geographical areas and for such relatively smaller racial and ethnic groups as American Indians; (2) these data are used for drawing political boundaries for very small geographical areas and funding allocations; (3) there is some evidence that a higher proportion of the population had difficulties with answering the race and ethnicity questions in 1990 than in 1980 (Cresce et al., 1992:6); and (4) some studies suggest that the current concepts may not be the most appropriate for a changing population (Hirschman, 1993; Lieberson, 1993). ALLOCATION RATES If a respondent did not provide answers to each question to the census, special procedures were made by the Census Bureau to impute (or allocate) the response. Not all items were allocated, including ancestry. The Hispanic-origin item continued to have the highest allocation rate (10 percent) of any item collected on the short form (see Table K.1). By contrast, the rate was much lower for the race item at 2.0 percent. The allocation rates for the 100 percent items for 1990, particularly for Hispanic origin, were significantly higher than those for

OCR for page 372
Modernizing the U.S. Census the 1980 census. The higher rates were due partially to the decision, based on budget limitations, to cut back on field follow-up for the short-form questionnaires that failed field content edit. The Census Bureau studies, however, show that the national figures hide great variability in the allocation rate for race and Hispanic origin, and that the variability for the national average is likely to be clustered in certain localities. For example, 1990 census allocation rates were substantially higher, above 20 percent, for both race and Hispanic origin in some census tracts of Los Angeles County. Furthermore, the variability has a great effect on some population groups. For instance, over 62 percent of black Mexicans were allocated in the census (del Pinal, 1994). These relatively high allocation rates can have potentially adverse effects on small-area analysis such as those on residential segregation. All major race groups, except American Indians, Eskimos, and Aleuts, experienced increases in short-form allocations. The "other race" category experienced the highest rate of computer allocations in 1990 and the largest increase from the 1980 rate (see Table K.2). The high nonresponse to the ancestry and Hispanic-origin items is also of concern. It may in part reflect respondents' perception that the race, Hispanic origin, and ancestry questions are asking for the same information and may lead some respondents to not answer all three items. Some researchers have suggested that the Census Bureau explore ways of combining these questions to reduce nonresponse rates and perhaps to reduce inconsistent reporting in these items as well (Farley, 1993). Efforts to shorten the census long form might also force consideration of a combined question if the census is still to gather ancestry data. MEASURES OF INCONSISTENCY Measures of inconsistency also provided evidence, however, of some misreporting in certain race and Hispanic-origin categories and of moderate inconsistency in ancestry reporting. Persons identifying in the American Indian, other Asian and Pacific Islander and other race categories of the race item and in the other Hispanic category of the Hispanic-origin item showed high inconsistency rates in their responses to these items. This is of concern because these are the write-in categories for the race and Hispanic-origin items. It may indicate a need for the Census Bureau to further examine the presentation and instructions for the write-in entries on the race and Hispanic items and to better understand how respondents perceive and interpret these questions. The levels of inconsistency are especially important and of concern for the American Indian population because of its relatively small size and the importance of the data to develop and fund programs for American Indian tribal and Alaska native village governments. In addition to this inconsistent reporting, instances of Asian Indians reporting themselves or their children as American

OCR for page 372
Modernizing the U.S. Census Indians and possible overreporting of Cherokees indicate that further research is needed on reporting on response to the American Indian category of the race item. The growth of the American Indian population at a rate greater than can be explained by natural increase and improved methods makes it one of the more important populations for studying how and why persons change their racial identities over time. The level of inconsistent reporting in the other Asian Pacific Islander, other race, and other Hispanic categories is of importance, because it seems to reflect reporting problems among immigrant populations that have grown rapidly during the past two decades. Immigrants are projected to represent ever increasing percentages of the U.S. population. A thorough understanding of how these respondents interpret and respond to other race and Hispanic-origin questions is therefore essential to the Census Bureau's ability to devise and implement questions that provide high-quality data on these groups. With some exceptions, it seems to be foreign-born people and both foreign-and native-born Hispanics that had the highest levels of inconsistent reporting on the race questions. Foreign-born people showed considerably more inconsistency on the race item than native-born ones. White foreign-born people had only moderate consistency in race reporting, perhaps due to the greater ancestral diversity in the white racial group. By contrast, native-born people had the greatest inconsistency reporting in the Hispanic-origin items. Both foreign-born and native-born Hispanics overreported in the other race and other Asian Pacific Islanders categories of the race item, but underreported in the white, black, and Asian and Pacific Islander categories. Shifting, largely by Hispanics, between the other race and white categories led to an underreporting of the white population and overreporting of the other race population in the census. Over 97 percent of people reporting in the other category were of Hispanic origin. Inconsistent reporting by the foreign-born and Hispanic populations may in part reflect unfamiliarity with or different interpretations and meanings of the concepts of race and Hispanic origin. Both race and Hispanic origin are subjective or culturally specific items that lack a universal definition, and foreign-born people may have a different understanding of these concepts. For Hispanics, inconsistent reporting might also reflect ambiguities arising from the listing of several sociocultural or national origin groups in the race item. Some respondents may try to indicate identification with a Hispanic national origin group comparable to other groups specified in the race item. Preliminary findings from cognitive research by the Census Bureau indicates that foreign-born Hispanics, foreign-born Asian and Pacific Islanders, and foreign-born blacks view the concept of race differently from the native U.S. population. For example, foreign-born Hispanics and foreign-born Asians and Pacific Islanders view race in terms of national origin or language.

OCR for page 372
Modernizing the U.S. Census LIMITATIONS OF THE RACE AND HISPANIC ORIGIN QUESTIONS The race and Hispanic-origin questions continue to be confusing to some census respondents. Information provided to census staff in the data collection stages (telephone inquires to the field and processing offices) indicated that a substantial number of people did not understand how to answer the race question. Inquiries about the race question came from persons who: (1) were confused by the listing of sociocultural groups and wanted to provide an ethnic group (e.g., Polish or Jamaican); (2) were Hispanic and did not understand the response categories; and (3) were of mixed parentage or were the parents of interracial children and wanted to report a multiple race. LIMITATIONS OF THE ANCESTRY QUESTION The ancestry question provided data of only moderate quality, in part because of its open-ended format. Unlike the race and Hispanic-origin items, people could identify more than one ancestry. Consistency between 1980 and 1990 in reporting in the general types of ancestry is evidence that the question worked reasonably well in 1990. However, consistency in reporting levels for specific ancestry groups deserves attention. Several ancestries used as examples in the instructions showed implausibly large increases over 1980, whereas ancestries dropped from the instructions decreased substantially. The most noted example was "English," which was used in 1980 but deleted in 1990. In terms of ranking, English was first in 1980 with 49,598,000 responses.2 In 1990, it ranked third with 38,736,000 responses.3 Similarly, "French" responses numbered 12,892,000 in 1980 when it was used as an example but fell to 10,320,000 in 1990 when it was not an example. "Cape Verdean" was a 1990 example that yielded 50,772 responses. This is in contrast to a count of 23,215 in 1980 when it was not an example. "Cajun'' was an example for 1990 but not in 1980; the combined Acadian/Cajun response number in 1990 was 668,271. Unfortunately, there was no comparable figure for 1980 as these responses, deemed too few to warrant their own code, were coded as "French" in the Census Bureau's 1980 code list. Although there may be other reasons for these changes, such as increase or decrease in immigration and changes in coding procedures, such "example effects" must be of considerable concern if write-ins are to be used to capture groups not now listed (for example, Arabs) or to capture an identity that combines racial status and ancestry in one item. This pattern of ancestry inconsistency reporting also reflected high rates of intermarriage and of geographic and social mobility among people of European descent. Waters (1990) notes that European-origin groups have a great range of choices and options in how they choose to identify when there are a wide range of options and the specific ethnic identification may be situational and flexible.

OCR for page 372
Modernizing the U.S. Census The examples presented on the census questionnaire can influence the reporting, as can general social conditions about favorable or unfavorable attitudes toward specific ethnic groups. With the high intermarriage rates for several Asian and Pacific Islander and Hispanic populations, these groups will also grow more important as the pool of people with identity options increases. STRENGTHS OF THE RACE AND ANCESTRY QUESTIONS Racial and ethnicity categories for 2000 and beyond may usher yet new ways of defining group identity as well as disadvantaged populations. The high rates of immigration, from both traditional and nontraditional sources of immigrants, and the increase of interracial unions and births in the last 20 years are the most visible alternate responses to the current race, Hispanic-origin, and ancestry items. The Census Bureau and other researchers initially assumed that these responses may suggest a lack of understanding of current race and ethnicity categories. Subsequent assumptions are that new immigrants and persons of multiple race or ancestry may have a different concept or lack of identification with these categories. Their socioeconomic characteristics also reflect these differences. Research across different groups is instructive in this regard. Snipp's work (1989) on the 1980 American Indian population underscores the multiple and differential meanings of race. Although American Indians are not an immigrant group, they have a high rate of intermarriage. This group has witnessed a tremendous increase (beyond natural growth and an explanation of better coverage due to outreach programs) that is due to shifts in self-identification. Snipp differentiated three groups: (1) persons who identified as American Indian by race and ancestry (numbered 947,500), (2) persons identified as American Indian by race of multiple ancestry (numbered 269,700), and (3) persons who were not American Indian by race but claimed Indian ancestry (numbered 5,537,600). These groups differed in socioeconomic status and patterns of language. Snipp concluded that Americans of Indian descent were more similar to middle-class Americans and did not conform to a common perception of Americans Indians as a culturally distinct population of economically disadvantaged individuals. This common perception was more pertinent to people who consistently identified their race and ethnicity as American Indian. American Indians of multiple ancestry were a diverse group. Utilizing race, Hispanic origin, and ancestry based on the 1990 public-use microdata samples, Farley (1994) concluded that there was a need to define populations by more than race. He classified 63 mutually exclusive categories by Hispanic origin, race, ancestry, and nativity. In a ranking of these groups by percent living in households with incomes below the poverty line, the top 10 groups included 4 that identified themselves as black by race, 4 as Hispanic origin, 1 American Indian, and 1 Vietnamese. In an examination of households

OCR for page 372
Modernizing the U.S. Census of foreign-born people with incomes five times or more the poverty line, 6 identified as white by race and 4 as Asian. Native-born whites and native-born blacks fared less well than their foreign-born counterparts. In the case of Hispanics, preliminary research by del Pinal (1992) based on the Current Population Survey explored the effects of various combinations of race and Hispanic-origin categories on the level of selected socioeconomic indicators. He found that Hispanics are nearly 10 percent of the white category and about 1.5 percent of all blacks. He argues that, on one hand, due to the relatively disadvantaged position of Hispanics, their inclusion in the white category tends to attenuate the difference between whites and other race categories on several important socioeconomic characteristics (e.g., labor force status, educational attainment, poverty level, family income) and that, at a minimum, Hispanics should be removed from the white category. On the other hand, the overlap between black and other numerically smaller race categories with Hispanic origin is small and, for now, does not adversely affect the analytical results of basic socioeconomic indicators. The Census Bureau's analysis indicates that race and ethnicity are complex matters and not likely to be suitably enumerated with one or two questions. The analysis provides some evidence that the question might be more consistent with some respondents' concepts and self-perception, for example, by allowing respondents to identify with more than one race group; to report as multiracial; to treat "Hispanic" as equivalent to a race group; and to combine the race and Hispanic-origin items. At the same time, other respondents might find such changes less consistent with their concepts and self-perceptions. The implications of any such changes for providing data needed for legislative and programmatic purposes clearly require thorough research. In addition, because of the complexity of race and ethnicity, any proposed changes to these items requires consultation with affected groups and data users. Such consultation must occur concurrently with testing. Testing of the race questions before the 1990 census showed that questions with write-in responses for subgroups within a category may produce data as reliable as that from questions with prespecified categories. Questions for write-in responses hold advantages in capturing groups not represented by the detailed categories of the race question. At the same time, benefits of write-in questions must be weighted against their higher processing costs. The Census Bureau's ability to process write-in data in a timely fashion should be considered. Census Bureau evaluations have been confirmed by other data users. In recent years with the growth of new populations, Statistical Directive 15 has been questioned as being outdated and not reflecting the current and foreseeable population. New immigrants often identify themselves by national origin rather than major categories. Other populations of more than one race or ethnicity find it difficult to choose only one category. There is acknowledgment by federal

OCR for page 372
Modernizing the U.S. Census agencies and other data users that some populations do not easily fit into existing categories or want to be reclassified from one category to another. The House Subcommittee on Census, Statistics, and Postal Personnel held a set of hearings in 1993 on the federal standard noting the need for a review of the standards. In preparation for a review of the standard, the Office of Management and Budget requested the Committee on National Statistics to conduct a workshop of key stakeholders to articulate the issues. The workshop, held in February 1994, found that the current categories have broad usage beyond the federal statistical system and have become reified as absolute standards. And yet, many individuals do not readily identify with a single, broad, homogeneous, and mutually exclusive category used in standard classifications. Some prefer multicultural classifications and others prefer subcategories of the broader categories. Such findings are consistent with the Census Bureau's experience with stakeholders. Any revisions to the federal standard will require examination of the Census Bureau's collection, use, and presentation of race and ethnicity data, which are broader than that of most federal agencies. This is of particular importance as the census provides the benchmark for other federal data. It seems clear that to ensure data of high quality and to identify race and ethnicity concepts that are appropriate for the changing racial and ethnic populations of the United States, the Census Bureau must conduct a comprehensive research and testing program for the 2000 census and establish an even more extensive long-range program of research and testing. NOTES 1   OMB Statistical Directive 15 has set forth the race and ethnicity standards for federal administrative and statistical reporting since 1977. Directive 15 acknowledges four racial groups (American Indian or Alaskan Native, Asian or Pacific Islander, Black, and White) and two ethnic categories (Hispanic Origin and Not of Hispanic Origin). OMB is currently in the process of reviewing the race and ethnicity classifications. 2   This number includes people who indicate more than one ancestry. 3   1980 figures are from U.S. Bureau of the Census, 1990 Census of the Population Supplementary Report, "Ancestry of the Population by State." PC80-S1-10. Figures for 1990 are from U.S. Bureau of the Census, Ethnic and Hispanic Branch, Population Division, "1990 Detailed Ancestry Groups for States," CPH-L-97, a supplement to the 1990 CP-S-1-2, "Detailed Ancestry Groups for States."

OCR for page 372
Modernizing the U.S. Census REFERENCES Cresce, A., S. Lapham, and S. Rolark 1992 Preliminary Evaluation of Data from the Race and Ethnic Origin Questions in the 1990 Census. Paper presented at the annual meeting of the American Statistical Association, Boston. del Pinal, J. 1992 Exploring Alternative Race-Ethnic Comparison Groups in Current Population Surveys. Current Populations Reports, Special Studies, Series P23-182. Washington, D.C.: Bureau of the Census. 1994 Social Science Principles: Forming Race-Ethnic Categories for Policy Analysis. Paper presented at the Workshop on Race and Ethnicity Classification: An Assessment of the Federal Standard for Race and Ethnicity Classification, February 17-18. Committee on National Statistics, National Research Council, Washington, D.C. Farley, R. 1993 Questions about Race, Spanish Origin and Ancestry: Findings from the Census of 1990 and Proposals for the Census of 2000. Testimony before the Subcommittee on Census, Statistics and Postal Personnel, Committee on Post Office and Civil Service, U.S. House of Representatives. 1994 The Experience of Federal Agencies Measuring Race in Accord with Directive #15. Comments at the Workshop on Race and Ethnicity Classification: An Assessment of the Federal Standard for Race and Ethnicity Classification, February 17-18. Committee on National Statistics, National Research Council, Washington, D.C. Hirschman, C. 1993 How to measure ethnicity: an immodest proposal. Pp. 547-560 in Challenges of Measuring an Ethnic World: Science, Politics and Reality. Statistics Canada and Bureau of the Census, U.S. Department of Commerce. Washington, D.C.: U.S. Government Printing Office. Lieberson, S. 1993 The enumeration of ethnic and racial groups in the census: some devilish principles. Pp. 23-35 in Challenges of Measuring an Ethnic World: Science, Politics and Reality. Statistics Canada and Bureau of the Census, U.S. Department of Commerce. Washington, D.C.: U.S. Government Printing Office. McKenney, N.C., Bennett, R. Harrison, and J. del Pinal 1993 Evaluating Racial and Ethnic Reporting in the 1990 Census. Paper presented at the 1993 annual meeting of the American Statistical Association , San Francisco. Snipp, C.M. 1989 American Indians: The First of This Land. New York: Russell Sage Foundation. U.S. General Accounting Office 1993 Census Reform: Early Outreach and Decisions Needed on Race and Ethnic Questions. GAO/GGD-93-36. Washington, D.C.: U.S. Government Printing Office. Waters, M.C. 1990 Ethnic Options: Choosing Identities in America. Berkeley: University of California Press.

OCR for page 372
Modernizing the U.S. Census TABLE K.1  Comparative Allocation Rates of Population Questions Asked of Total Population (in percent) Item 1980 1990 Sex 0.8 1.2 Marital status 1.3 2.0 Race 1.5 2.0 Age 2.9 2.4 Relationship 2.1 2.6 Hispanic origin 4.2 10.0   Source:  U.S. General Accounting Office (1993).

OCR for page 372
Modernizing the U.S. Census TABLE K.2  Total Allocation Rates by Race for the United States: 1990 and 1980   100 Percent (short form) Sample (long form) Race 1990 1980 1990 1980 Total 2.0 1.5 1.2 1.5 White 1.7 1.3 1.0 1.5 Black 2.9 1.6 1.6 1.7 American Indian, Eskimo, Aleut 2.5 3.1 1.7 1.5 Asian and Pacific Islander 2.7 2.1 1.6 1.6 Other race 6.4 3.9 3.2 1.8   Source:  Cresce et al. (1992).