March 14, 2001

Katherine K. Wallman
Chief Statistician
Office of Management and Budget,
Room 10201 New Executive Office Building
725 17th Street NW
Washington, DC 20503

Dear Ms. Wallman:

The Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (December 15, 2000) is of great interest to the National Committee on Vital and Health Statistics (NCVHS). This Committee has a long history of interest in issues of race and ethnicity in health data and health statistics. It has provided advice to the Secretary of Health and Human Services in this area on a number of occasions. The intent of this letter is to note areas of concern that the Committee intends to pursue further and to alert you that the NCVHS Subcommittee on Populations will explore many of these issues over the next several months. We appreciate the opportunity to provide comments and questions.

The comments and questions of the Committee are in three specific areas: 1) the tabulation plans for the multiple race data collected under the new standard; 2) the rationale and conditions for variances from the guidelines for the collection and tabulation of multiple race and; 3) the continuation and timeline for methodological research to determine procedures for the tabulation, analyses and interpretation of the multiple race data collected under the new standards for use in population estimates and vital statistics. The Committee notes and appreciates all of the work and leadership of OMB on these issues.

Tabulation Plans and Bridging Methods for Multiple Race Data in Population Estimates.
The 1997 OMB Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity provide individual respondents with the opportunity to select more than one race. This change, and the guidelines to handle the resulting health data, pose a set of challenges and questions.

1. As presented in the guidance document, the multiple race and ethnicity data is based on self- definition. There seems to be a number of factors that can result in inconsistency and incompatibility across datasets. Reporting of more than one race will differ geographically and, as some preliminary research has indicated, by individual characteristics such as education, age and income. Variations in methods for creating bridges could also have different effects on the statistics for some of the groups. Preliminary works by some scholars in the field have suggested American Indians and Asian Americans are most affected by variations in tabulation methods. The field would benefit from further leadership by OMB on statistical estimates of group size differences that are attributable to true change versus revised collection and tabulation methods.

Data on mixed race individuals suggests their view of racial identity can be fluid and for some not stable in its reporting. Coupled with this, the guidance indicates that some agencies will receive a variance allowing them to utilize, in addition, a category known as “other races” thereby resulting in a different racial classification scheme. These factors, combined with the permissibility for agencies to vary in the standards used for evaluating the quality and confidentiality of data, may result in difficulties for users in comparing across federal datasets. What guidance is proposed to assist users who must either have uniform data mechanisms in place or are engaging in analyses that will use bridging methods to compare findings or use more than one federal dataset?

2. The variation in self-identification in multiple race also has other implications. According to some current work in this area, individuals who self-identify as multiple race vary relative to education, age, and other sociodemographic variables that are important in health data. One result of this variation is that population estimates of sociodemographics may vary depending on the tabulation process. Is research planned that can examine the results of different racial classification measures on the sociodemographic characteristics of the mixed race vs. single race groups?

3. In the multiple race tabulation process, the provisional guidance indicates that the judgment of the users will be a factor in how the combinations could be collapsed. Insights into the consequences of those choices on population estimates, stability of those estimates, and comparability and uniformity are needed in order to assist users in their choices of the various race classification schemes.

Variances in the Collection and Tabulation of Multiple Race.

1. In reviewing the guidelines, federal agencies can vary in their strategies for racial classification and implementation of tabulation for racial classification. While variances are understandable given that agencies differ in their mission and therefore the use of the data, users are also affected by the variances across agencies. Some data users, particularly those at the state level, mandated to reconcile federal, state and local data reporting systems may in the face of variances, find it difficult to accomplish uniformity or comparability across datasets. Two areas that the Committee will further discuss are how multiple race data will be incorporated into intercensal population estimates, and comparisons of birth and death data with other public health data sets.

Guidance and Future Methodological Research on Multiple Race Tabulation and Data Collection Procedures.

1. The Provisional Guidelines include a section on the interpretation of data collected under the new standards. This section includes extremely important points about the complexity of data collected under the new standards. Helpful to this section of the guidance for users would be greater details of the preliminary and published work of what is currently known about the complexities and implications of the interpretations in the use of multiple race populations. OMB and a number of others from Census, other federal agencies and academic scholars have generated a growing body of work that can serve to help inform users of the complexities of multiple race tabulation and best methods for the collection of this data.

As the number of users and data availability increase, particularly from Census 2000 direction will be needed from research and evaluation that can lead to a better understanding of how individuals report their race and ethnicity, and of multiple race populations in general. Prior to the revision to OMB Directive 15, OMB in 1995 and 1996 commissioned or employed a series of studies (RAETT, CPS Supplement, NCS (National Content Survey), NHIS) that led to the new standards. OMB in preparation for the new standards took an active leadership role in ensuring that high quality methodological work was done to provide the broadest possible understanding of how these changes might affect data collection across the federal statistical system. In light of the issuance of the new standards and the fact that data systems are starting to implement them, we look forward to continued leadership by OMB that will encourage research and evaluation on the complicated process of collecting and analyzing data on multiple race populations. OMB is particularly well positioned to encourage federal agencies to conduct this research. It is hoped that the research mentioned in the report can proceed quickly and swiftly. As this research evolves OMB could assist the field by aggressively pursuing education and training of other federal agencies and data users.

2. While currently provisional guidance is offered on the collection and tabulation process, what does appear to be absent is a discussion of the value of the use of multiple race categories in meaningfully measuring social, economic and health characteristics with the revised data collection measures. This guidance should come from empirical analyses as opposed to opinion.

3. The Department of Health and Human Services as well as a number of other federal agencies create datasets in which information is drawn from administrative data. In regards to the race and ethnic group classification it is not clear if the source of that data is derived from self-report and complies with the self-definition strategy used in other federal data or if the designation is based on observation that the respondent does not confirm. As administrative data is growing in its use and is often used in coordination with other federal datasets, further guidance on the above issues would be beneficial. OMB is encouraged to foster methodologic research on multiple race data collection and reporting for administrative data systems.

Two important questions remain. How meaningful and predictive are the multiple race categories for health risk, health status and public health trends of these populations? Is the quality of the data derived from the multiple race tabulations equal to that for single race groups?

Our Committee looks forward to further pursuing the areas discussed in our letter. We appreciate the opportunity to comment and congratulate OMB on its efforts to provide guidance on these matters. The NCVHS Subcommittee on Populations would welcome the opportunity to meet with you and your staff to further discuss our concerns.



John R. Lumpkin, M.D., M.P.H.