| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
The authors are with the Office of Analysis, Epidemiology, and Health Promotion, National Center for Health Statistics, Hyattsville, Md.
Correspondence: Requests for reprints should be sent to Jennifer D. Parker, PhD, Infant and Child Health Studies Branch, Office of Analysis, Epidemiology, and Health Promotion, National Center for Health Statistics, 6525 Belcrest Rd, Room 790, Hyattsville, MD 20782 (e-mail: jdparker{at}cdc.gov).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. Race-specific health statistics are routinely reported in scientific publications; most describe health disparities across groups. Census 2000 showed that 2.4% of the US population identifies with more than 1 race group. We examined the hypothesis that multiple-race reporting is associated with interracial births by comparing parental race reported on birth certificates with reported race in a national health survey.
Methods. US natality data from 1968 through 1998 and National Health Interview Survey data from 1990 through 1998 were compared, by year of birth.
Results. Overall multiple-race survey responses correspond to expectations from interracial births. However, there are discrepancies for specific multiple-race combinations.
Conclusions. Projected estimates of the multiple-race population can be only partially informed by vital records.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
In 1997, the Office of Management and Budget (OMB) issued a revision to the long-standing directive for the collection of race and ethnicity data within the federal statistical system, known as OMB-15.4,5 Among other modifications designed to reflect the changing racial and ethnic profile in the United States, the 1997 standard requires that new data collections allow individuals to report 1 or more race groups when responding to a query on their racial identity. Analysts examining previously available data hypothesized that up to 2% of respondents to surveys or administrative collections would report 2 or more groups under the new standard.6–11 About 2.4% of the US population, nearly 7 million people, reported 2 or more race groups in the 2000 decennial census.11
The impact of multiple-race reporting on statistics used for health policy and research is not yet known. It is likely that multiple-race respondents differ from each other and from their single-race counterparts on many measures of health and access to care.7,10,12,13 The extent of these differences will depend on many factors. All considered, multiple-race reporting will influence public health policy for both the newly tabulated multiple-race groups and the remaining single-race groups, which will be changed as a result of a wider choice of racial identification.
Interracial births have increased over the past 3 decades.14,15 In the early 1970s, 1.4% of infants were born to parents who reported different race groups; by 1998, this percentage had increased to 4.3%. It would be reasonable to assume that individuals with parents of different races would identify with and report more than 1 group when responding to surveys and other data collections. However, how interracial births affect multiple-race reporting is unclear.
This report compares year- and racespecific national estimates of interracial births with year-specific survey estimates of multiple-race reporting. We compared the distribution of parental race for births from 1968 through 1998 with the reporting of more than 1 race for survey respondents in the 1990–1998 National Health Interview Survey (NHIS) who were born from 1968 through 1998. If all individuals with interracial parents reported both race groups on the survey, we would expect the distribution of multiple-race responses on the NHIS to coincide with the distribution of interracial births from birth records for the appropriate age–year combination. For example, the race distribution for births in 1970 would correspond to the race reported among the respondents who were aged 20 years in the 1990 NHIS, who were 21 in the 1991 NHIS, and so on. We would also expect that the inclusion of individuals with 1 or both parents who themselves identify with more than 1 race group may increase the percentages of multiple-race responses in the NHIS even more. Although neither the NHIS nor the birth certificate were developed to provide national race distributions, both data sources are routinely used to provide national estimates of race-specific health outcomes.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Although the categories used to collect race data changed within the study period, race could be recoded into the groups defined under the 1977 OMB-15 standard: White, Black, American Indian or Alaska Native (AIAN), and Asian or Pacific Islander (API). A residual category, "other race," was used before 1989. Births were considered interracial if the races of the parents differed.
From 1978 to 1988, "other race" was divided into 2 groups: other non-White and other API; it has been estimated that before 1978, 85% of other-race responses were for parents identifying with the "other API" subgroup. Although we retained the original other-race response for the primary analysis, estimates were recalculated for the earlier years after reassignment of the other-race responses to API. To more generally assess the influence of the other-race category on these estimates, interracial birth estimates were recalculated without counting combinations that included "other race"; these supplemental findings are discussed in the text but are not included in the tables.
The National Health Interview Survey
The NHIS for 1990 through 199817–19 was used to provide a large enough sample size for stable estimates of birth-year-specific multiple-race responses. The NHIS is a continuing household survey of approximately 40 000 households (100 000 respondents) each year; only 60% of the sample was collected for 1996. To compare the race distribution with the corresponding distribution from the natality files, our study sample was limited to the 339 342 US-born respondents born in 1968 through 1998 with valid data for race; of the 962 056 survey respondents included in the NHIS for 1990 through 1998, 368 207 were born between 1968 and 1998 and were eligible for inclusion. Of those, 23 625 were foreign born and 5236 were missing race information and were therefore excluded. Before the 1993 survey, nativity data were collected only for surveyed adults. Although this study criterion was differentially applied across survey years, race distributions for the entire survey sample were compared with those for the 1996–1998 native-born sample to assess potential bias from the combination; these results are described in the text but are not shown in the tables.
Since 1976, the NHIS has allowed respondents to choose more than 1 race group. Although survey responses are ideally selfreported, many are proxy responses; parents typically provide survey information for their children. To be congruent with the natality files, the choices for race on the NHIS were recoded to the four 1977 race groups. The comparison was limited to the 4 largest multiple-race groups (Black–White, AIAN–White, API–White, Black–AIAN); even with several years of the survey combined, very few respondents reported API–AIAN or API–Black. Before 1997, only 2 reported race groups were retained on the National Center for Health Statistics computer files; however, given that the 1997 and 1998 data show that over 90% of multiple-race respondents report only 2 groups, the bias associated with this loss of detail is small.
The NHIS codes a residual category, "other race." As with the interracial birth estimates, the multiple-race estimates from the NHIS were calculated both with and without multiple-race combinations that included "other race" to evaluate the effect of this residual category on the multiple-race estimates; these findings are reported in the text.
Comparison of Data Sources
We compared the data sources in 2 ways. First, we compared the racial distribution from the natality files with the birth-yearspecific race distributions from the NHIS, presenting the comparison by year of birth. Second, we calculated the estimated race distribution for the NHIS using the age distribution of the survey respondents and their corresponding birth-year-specific race proportions from natality files, presenting the comparison by survey year.
For the first comparison, we divided both the natality and the NHIS data into 6 birth cohorts (1968–1972, 1973–1978, 1979–1982, 1982–1988, 1989–1992, and 1993–1998). For the second comparison, we divided the NHIS into three 3-year survey groups. We calculated expected race group proportions using single-year race estimates from natality files and single-year age distributions from the NHIS.
We calculated percentages from the NHIS using the survey weights for all analyses. Applying the sample design across several years of survey data for standard error estimation for small subsets of the sample was deemed untenable. To ensure stable estimates, however, year–race groups were limited to those groups with at least 35 observations; preliminary calculations found that groups of this size had relative standard errors of about 25% for multiple-race estimates. Of the 339 342 eligible NHIS respondents, there were 7816 multiple-race respondents, ranging from 744 in 1990 to 1053 in 1994; in 1996, a year with fewer NHIS respondents overall, there were 617 who reported 2 or more race groups.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
). Multiple-race reporting among NHIS respondents born in the corresponding years also increased. Among those born before the early 1980s, however, the percentage in the NHIS was greater than the corresponding percentage of births; the opposite relationship was true for those born in later years. Previously published estimates show that overall multiple-race reporting in the NHIS has remained relatively even since 1982, ranging from 1.2% to 1.8% (Figure 2).6 Although the percentage of respondents who reported more than 1 race group among those aged younger than 30 years (1.8%–2.6%) was slightly higher than that for the whole survey population, the pattern was similar throughout the 1990s.
|
|
). For example, whereas 0.77% of NHIS respondents born between 1968 and 1972 chose AIAN and White race groups, only 0.22% of birth certificates for 1968 through 1972 recorded AIAN and White for parental race groups. The concordance for the AIAN and White was greater for those born more recently.
|
Black–White race, by year of birth. Similarly, the percentage of infants with 1 Black and 1 White parent increased greatly, from 0.33% to 1.77%. The corresponding percentage of respondents reporting 1 Black and 1 White parent on the NHIS was slightly more than half the corresponding interracial birth percentages during the study period.
AIAN–Black race, by year of birth. During the study periods, the percentage of infants who were AIAN and Black increased from 0.01% to 0.04%. The corresponding percentages of AIAN–Black respondents in the NHIS did not increase but were much higher, fluctuating from 0.9% to 0.14%.
Other multiple-race combinations, by year of birth. The other multiple-race groups show an actual decline in interracial births, which can be attributed, in part, to differential coding of the "other" category over time. When interracial births with 1 parent reported as "other race" were not included in the other multiple-race groups, the percentage of interracial births dropped considerably more for earlier than for later years, reversing the trend (not shown). Similarly, when NHIS respondents who reported other race and a more specific race were taken out of the other multiple-race groups, the percentages decreased by about one half for all year groups (not shown).
Effect of Nativity, by Year of Birth
The estimates of multiple-race responses in the 1996–1998 NHIS were similar with and without the inclusion of foreign-born respondents (not shown), suggesting that the differential exclusion of foreign-born respondents between surveys did not bias the results.
Multiple-Race Reporting, by Survey Year
Multiple-race reporting in the NHIS varied across survey years (Table 2), particularly for the AIAN–White group. For the Black–White and API–White groups, the observed percentages of multiple-race respondents were less than we would expect given the reported interracial births; the difference between the estimates was larger for the Black–White than for the API–White group. The differences varied across survey years as well, although no statistical tests were done to assess statistical significance.
|
). Indeed, in the Black–AIAN group, the NHIS estimate is about 6 times as high as for the corresponding interracial births. In comparison, the observed and expected percentages for single-race White, Black, and API groups were similar; the single-race AIAN responses reported in the NHIS were about 50% higher than natality data would predict (not shown).
The higher NHIS estimate for these groups and the lower estimate for the others combined into overall multiple-race percentages that were similar between data sources (Table 2, last column). For 1993 through 1995, the percentage is slightly higher than that reported for the same years in the OMB’s provisional guidance document.7 As shown in Figure 2, this discrepancy is due to the younger age cohort used for this study.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
The correspondence between the NHIS reporting and the birth distribution differs between race groups. The relatively large proportion of multiple- and single-race AIAN responses in the NHIS, coupled with the relatively low proportion of AIAN interracial and single-race births, is consistent with the documented, but not wholly explained, increase in the reporting of AIAN since 1960.20 Strategies that have been suggested for improving data for single-race AIAN births will need to be implemented and adapted for their multiple-race counterparts.21,22
In addition, this finding is more marked among older NHIS respondents, who are more likely to be responding for themselves. Although we use the term "respondent" in this article, in the NHIS, children younger than 18 years do not respond to the survey themselves; a proxy report, usually from a parent, is obtained. In the comparisons for the earliest birth cohort, none of the survey respondents were younger than 18 years; for the 1973–1977 birth cohort, about one third of the corresponding survey respondents were children; in the 1978–1982 cohort, nearly 90% of the corresponding survey respondents were children; and after 1982, all of the respondents were children. Data for adults in some households may also be reported by proxy. How proxy reporting vs self-reporting and age contribute to the overlapping lines in Figure 1 is unclear. However, the differences between interracial births and multiple-race reporting across birth cohorts were often similar within specific race groups (Table 1), suggesting that differing distributions of race by age may contribute more to the overall results than proxy reporting.
Aside from the fact that the survey was limited to those born in the United States, the effects of Hispanic origin and immigration/emigration were not fully considered. Hispanic origin, although correlated with racial identity in the United States, is collected by a separate question under both the 1977 directive and the 1997 standard. The number of infants identified on birth certificates as being of Hispanic origin has increased over the past 30 years, but data on Hispanic origin were not uniformly collected during the study period. For these reasons, Hispanic identity was not considered for this comparison. Both natality files and the NHIS routinely assign Hispanics of unknown race to White race; consequently, the impact of Hispanic origin on these findings may be small. On the other hand, more Hispanics than non-Hispanics report more than 1 race.11 A closer look at the relationship between race and ethnicity reporting on administrative records was beyond the scope of this study.
Given the large number of birth records with missing information for father’s race, the true percentages of interracial births are probably slightly higher than those known from vital records.14,15 These higher percentages would lead to larger differences between the percentages reported from the vital records and the survey than the differences reported here. Because the percentage of birth records missing the father’s race has been increasing over time, the bias is probably greatest for the comparisons between the most recent natality data and the younger survey respondents. Unfortunately, the small number of multiple-race respondents in the NHIS does not permit a more detailed study of these limitations.
With full implementation of the 1997 OMB standard, more data for multiple-race groups will be reported. For most health statistics, the number of multiple-race respondents will be too small for stable estimates, at least in the next few years; however, as the number who identify with more than 1 group increases, and years of data are combined, the stability of such estimates will improve. The lack of concordance between systems for trend lines, survey controls, and vital statistics dependent on population denominators will be an issue for several years.23 Our findings suggest that inferences regarding multiple-race groups cannot necessarily be extrapolated to all individuals with interracial parents; studies of interracial births cannot be used to generalize to multiple-race groups. As with all public health studies that document racial disparities in outcomes or services, other factors underlying racial differences will need to be considered.
|
Acknowledgments |
|---|
Human Participant Protection
No institutional review board approval was needed for this study.
|
Footnotes |
|---|
Accepted for publication March 12, 2002.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
2. Changing America. Indicators of Social and Economic Well-Being by Race and Hispanic Origin. Washington, DC: Council of Economic Advisers for the President’s Initiative on Race; September 1998.
3. Kington RS, Nickens HW. Racial and ethnic differences in health: recent trends, current patterns, future directions. In: Smelser NJ, Wilson WJ, Mitchell F, eds. America Becoming: Racial Trends and Their Consequences. Vol 2. Washington, DC: National Academy Press; 2001:253–310.
4. Office of Management and Budget. Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity, Federal Register 62FR58781–58790, October 30, 1997. Available at: http://www.whitehouse.gov/omb/inforeg/statpolicy.html. Accessed July 10, 2002.
5. Race and Ethnic Standards for Federal Statistics and Administrative Reporting. Washington, DC: Office of Management and Budget; 1977. Statistical Policy Directive 15.
6. Office of Management and Budget. Recommendation from the Interagency Committee for the Review of the Racial and Ethnic Standards to the Office of Management and Budget Concerning Changes to the Standards for the Classification of Federal Data on Race and Ethnicity, Federal Register 62FR36873–36946, July 9, 1997. Available at: http://www.whitehouse.gov/omb/inforeg/statpolicy.html. Accessed July 10, 2002.
7. Office of Management and Budget. Provisional Guidance on the Implementation of the 1997 Standards for the Collection of Federal Data on Race and Ethnicity. December 15, 2000. Available at: http://www.whitehouse.gov/omb/inforeg/statpolicy.html. Accessed July 10, 2002.
8. Results of the 1996 Race and Ethnic Targeted Test. Washington, DC: Bureau of the Census; 1977. Population Division Working Paper No. 18.
9. Tucker C, McKay R, Kojetin B, et al. Testing Methods of Collecting Racial and Ethnic Information: Results for the Current Population Survey Supplement on Race and Ethnicity. Washington, DC: Bureau of Labor Statistics; 1966. Statistical Notes No. 40.
10. Sondik EJ, Lucas JW, Madans JH, Smith SS. Race/ethnicity and the 2000 census: implications for public health. Am J Public Health. 2000;90:1709–1713.
11. Overview of Race and Hispanic Origin. Washington, DC: Bureau of the Census; March 2001.
12. Parker JD, Makuc D. Methodologic implications of allocating multiple race data to single race categories. Heath Serv Res. 2002;37:203–215.
13. Parker JD, Lucas JW. Multiple race reporting among children in a national health survey. Ethn Dis. 2000;10:262–267.[Medline]
14. Atkinson J, MacDorman M, Parker JD. Trends in births to parents of two different races in the United States: 1971–1995. Ethn Dis. 2001;11:273–285.[Medline]
15. Parker JD. Birthweight trends among interracial black and white infants. Epidemiology. 2000;11:242–248.[Medline]
16. Vital Statistics of the United States: Natality 1998. Technical Appendix. Hyattsville, Md: National Center for Health Statistics; March 2000.
17. Benson V, Marano MA. Current estimates from the National Health Interview Survey, 1995. Vital Health Stat 10. 1998;No. 199:1–428.
18. Massey JT, Moore TF, Parsons VL, Tadros W. National Health Interview Survey, 1985–94. Design and estimation. Vital Health Stat 2. 1989;No. 110:1–42.
19. Botman SL, Moore TF, Moriarty CL, Parsons VL. Design and estimation for the National Health Interview Survey, 1995–2004. Vital Health Stat 2. 2000;No. 130:1–41.
20. Thornton R. Trends among American Indians in the United States. In: Smelser NJ, Wilson WJ, Mitchell F, eds. America Becoming: Racial Trends and Their Consequences. Vol 1. Washington, DC: National Academy Press; 2001:135–169.
21. Rosenberg HM, Maurer JD, Sorlie PD, et al. Quality of death rates by race and Hispanic origin: a summary of current research, 1999. Vital Health Stat 2. 1999;No. 128:1–13.
22. Burhansstipanov L, Satter DE. Office of Management and Budget racial categories and implications for American Indians and Alaska Natives. Am J Public Health. 2000;90:1720–1723.
23. Durch JS, Madans JH. Methodological issues for vital rates and population estimates: 1997 OMB standards for data on race and ethnicity. Vital Health Stat 4. 2001;No. 31:1–30.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
