Making Meaningful Comparisons:
Transforming Data on Race and Ethnicity in Postsecondary Education, Pre- to Post-Directive 15

Kristen A. Renn and Christina J. Lunceford
Michigan State University
AERA ~ San Diego, California ~ April 14, 2004



In October 1997, the United States Office of Management and Budget (OMB) issued revisions to its Directive 15, changing the federal racial identification process to expand the number of racial categories and to include the option for respondents to indicate more than one race (OMB, 1997). The 2000 Census marked the first time in U.S. history that individuals had the option to self-identify in more than one of the five racial categories (American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, White) in addition to indicating Hispanic or Latino ethnicity[1]. Of all respondents, 2.4% indicated more than one racial category; while these respondents represent a small minority of the total population, it is important for higher education policy that 4.0% of those under age 18 and 7.7% of those under age 18 reporting Hispanic or Latino ethnicity indicated more than one category (U.S. Census Bureau, 2001). The U.S. Census Bureau (2001) projects that by 2020, nearly 30% of people with Hispanic or Latino ethnicity, 20% of Asians, 15% of Blacks, and 12% of Whites will be mixed race; by 2100, the projections increase to 70%, 45%, 35%, and 35%, respectively. Without a change in data collection procedures, institutional record keeping based on the monoracial assumption that instructions to “check one box only” accurately capture racial demographics would soon be confounded by individuals for whom the one-box option is inappropriate. In fact, OMB mandated an immediate change in federal data collection and the full implementation of the change in data reporting by January 2003.

On the surface, the inclusion of an option to indicate more than one racial category seems a minor change in data collection, to be accomplished with a word processor and new forms; beneath this simple change lays a political minefield from institutional to national levels. It involves the National Center for Education Statistics (NCES) and the Office of Civil Rights (OCR), state legislatures and higher education coordinating boards, institutional policy and legal compliance, a societal trend away from self-reporting racial and ethnic categories, and a tendency among mixed-race youth to report their race differently in different contexts (Harris & Sim, 2002). Anticipating the challenge of implementing the changes and the political consequences of such changes (e.g., implications for congressional redistricting, incomparability of data pre- and post-revisions, etc.), representatives from a number of federal agencies have been working to develop guidelines for collecting racial and ethnic data under the revised standards, to create “crosswalks” from old data to new, and to keep stakeholders informed of the changes. Representatives of postsecondary institutions and organizations have been involved in the process, and the stakes for this constituency are high.  

In a time and culture of legal challenges to affirmative action and growing campaigns (particularly in California and Michigan) to ban classifying people by race and ethnicity in public institutions, racial statistics matter more than ever. The implementation of revisions to OMB Directive 15 could have a significant impact not only on how data are collected, but also on how data are used. There are many challenges involved in collecting and reporting data accurately. It is not clear that data collected before, during, and after the transition can be compared in any meaningful way, yet these data will become very important in assessing the impact on access, enrollments and graduation rates of various racial and ethnic group members. Finally, there are questions about how the shift might run afoul of the Federal Educational Records Privacy Act (FERPA), in cases where individuals might be identified by a unique combination of race and ethnic categories. Clearly there is more at stake than revising and reprinting institutional forms to meet the new federal standards. Issues of how data collection will be transformed intersect with what the new data will say about the status of racial and ethnic groups in postsecondary education.

In this paper, we discuss the policy environment in which data on race and ethnicity are collected through a history of federal collection of racial data, the current status of racial data collection, and the processes that led to current and proposed data collection requirements. We then describe the impact of the changes on institutional data collection as illustrated by data from a preliminary study of current and planned data collection practices at a national sample of institutions. Finally, we propose strategies for comparing data from before and after the change to multiple-race reporting and topics for future research in the area.

Policy environment in which postsecondary data on race and ethnicity are collected

Background of federal collection of data on race and ethnicity

The U.S. Bureau of the Census has counted individuals by race since the eighteenth century, though definitions and methods of identifying members of racial categories have varied substantially over time (Lee, 1993; Spickard, 1989). Figures based on a complex system of labels for individuals with varying levels of black and white “blood” (e.g., “full-blooded Negroe,” “mulatto,” “quadroon,” “octoroon”), all based on observation of skin tone, were admitted by the 1890 census report to be “of little value” (Spickard, 1989, p. 433, note 27). By the 1950 census this system had given way to three racial categories: White, Black, and Other (U.S. Census Bureau, 1957).

During the 1960s, the federal government began collecting racial data for reasons other than the decennial population count. Title VII of the Civil Rights Act of 1964 directed the Census Bureau to collect registration and voting information by race, color, and national origin, believed to be “the first Civil Rights era instance of Congress mandating the collecting of racial data” (Farley, 2001, p. 3). The Department of Health, Education and Welfare (HEW) Office of Civil Rights used Title IV of the Civil Rights Act of 1964 to collect information about the race of public school students. Title VII of the Civil Rights Act of 1965 established the Equal Employment Opportunities Commission, which required data on race and sex in order to insure employers’ compliance. The test of compliance in employment, and often in public education settings, was the presence of black employees, students, or both.

The 1970s brought additional changes to federal record keeping related to race and national origin. Increasing numbers of immigrants from nations other than those in Western Europe swelled the Asian, Caribbean, African, and Latino populations; the 1970 census included a question to identify those whose descent was Mexican, Puerto Rican, Cuban, Central or South American, Other Spanish, or none of the above (Farley, 2001). The 1975 renewal of the Voting Rights Act of 1965 provided an opportunity for Latino activists to push for expansion of protected categories, and “language minorities” were added to include Asian Americans, Alaskan Natives, American Indians, and persons of Spanish heritage.

Prompted by Secretary Casper Wineberger of HEW, the Federal Interagency Committee on Education (FICE)[2] in 1973 began the process of creating government-wide standards for racial classification. In 1977 the Office of Management and Budget (OMB) issued Directive 15, which mandated that federal agencies gathering demographic data must use four major racial categories (White, Black, Asian or Pacific Islander, American Indian, or Alaska Native), with only one per person. Spanish heritage (yes or no) was also to be determined for each individual, either by a separate question or as a fifth “racial/ethnic” category (OMB, 1977). A rider amended to a military appropriations bill by Congressman Robert Matsui delineated the Asian or Pacific Islander category on the 1980 census as Japanese, Chinese, Filipino, Vietnamese, Asian Indian, Hawaiian, Guamanian and Samoan (Farley, 2001).
Recent history of the changes in Federal Standards for Racial and Ethnic Data Collection and Reporting (known as OMB Directive 15)

A small but influential movement of multiracial individuals and parents of multiracial children emerged in the late 1980s and quickly drew attention in the media and from some key politicians. After working at the state level to add “multiracial” as a data collection category in Ohio, Illinois, and Georgia, activists began to address national data collection. A counter-movement immediately emerged from monoracial activist groups (e.g., National Urban League), and there began a serious debate about the potentially harmful effects of a separate multiracial category. The debate lasted through much of the 1990s[3] and involved such unlikely advocates as House Speaker Newt Gingrich, in whose congressional district lived an influential white activist for adding a separate multiracial category.

In 1993, OMB began a formal review of the standards for collecting, analyzing and reporting government data on race and ethnicity. In 1994, OMB created the Interagency Committee for the Review of Racial and Ethnic Standards (ICRRES), representing more than 30 federal agencies. ICRRES was charged with making recommendations about how information on racial and ethnic backgrounds might more fully be collected and reported. The resulting OMB standards apply to “all federally collected data and reporting, including all levels of education, the national census, medical research, disease statistics, drawing boundaries for Congressional districts, the Voting Rights Act, and compliance with federal law and statutory regulations” (Davis-Van Atta, 1998, Summary, ¶ 3)

The new standards, published in final form in October, 1997, included the change from five “acceptable racial and ethnic categories” (American Indian or Alaskan Native, Asian or Pacific Islander, Black, Hispanic, White) to “five minimum categories for data on race: American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, White” and “two categories for data on ethnicity: Hispanic or Latino, Not Hispanic or Latino” (OMB, 1997, Summary, ¶ 2). The new standards also stipulate that there be no “multiracial” or “multiethnic” option, but that respondents are offered the option of indicating more than one category. The preferred method of racial/ethnic data collection is through a two-question format that asks first about ethnicity, and then about race, though a one-question format is permissible. Self-identification is the preferred mode for collecting racial and ethnic data, though “observer-collected” data on race and ethnicity are permitted, in which case the observer should use six categories, combining the racial and ethnic categories as in the one-question format (OMB, 1997).

Reporting requirements were also changed, to include the minimum of reporting in the five racial groups and two ethnicities when data were collected by self-identification or the six racial/ethnic groups when data were collected by observation. In addition, “data producers are strongly encouraged to report detailed distributions, including all possible combinations, of multiple responses to the race question. If data on multiple responses are collapsed, at a minimum, the total number of respondents indicating more than one racial category must be provided” (NCES, 2000). OMB directed all changes to be effective “as soon as possible, but not later than January 1, 2003” (OMB, 1997, Effective Date, ¶ 1).

In February 1999, the Tabulation Working Group of the Interagency Committee for the Review of Standards for Data on Race and Ethnicity (formerly ICRRES, the group formed in 1994 whose recommended changes to the racial/ethnic data collection and reporting resulted in OMB’s revision of Directive 15) issued Draft Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (Tabulation Working Group, 1999). The final version of these guidelines was presented on December 15, 2000 as Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (Tabulation Working Group, 2000). (See Appendix for a timeline of events related to the development of policy related to racial/ethnic data collection in postsecondary education.)
  Relationship of new requirements to collection and reporting of racial and ethnic data in postsecondary education

Although the new standards apply legally only to data collection and reporting by federal agencies, multiple interfaces among postsecondary institutions, related organizations, and the federal government all but mandate changes in institutional data collection and reporting. Not adopting the new methodologies at the institutional level would lead to fundamental incompatibility not only between one’s institutional data and all federal data (e.g. census data, IPEDS [Integrated Postsecondary Education Data System] data), but to lack of comparability between institutions that have adopted the new standards and those that have not. Further, the result of not adopting the new standards would mean an inability to report institutional racial and ethnic data that more accurately reflect the U.S. population and would also forfeit an opportunity to provide prospective and current students, faculty, and alumni with options to capture for their [sic] racial and ethnic backgrounds more fully and accurately. (Davis-Van Atta, 1998, Implications for Institutions, ¶ 2)

Furthermore, it was expected that “adopting the new standards will result in not only new data, but also in a different structure for national and institutional databases” (Davis-Van Atta, 1998, Implications for Institutions, ¶ 3, emphasis in original). For example, rather than use a single data field with a limited set of single values to represent one racial/ethnic background per person, databases might have to include multiple permutations of possible racial/ethnic combinations, each with its own data field and dichotomous values. Overhauling institutional, state, and national databases to reflect the new OMB standards was predicted to be “likely to require considerable investment of both time and money” and “very different analytic treatment of racial and ethnic data, as well as of all other data associated with racial and ethnic backgrounds” (Davis-Van Atta, 1998, Implications for Institutions, ¶ 3).

Given the complexity of the tasks of complying with Directive 15 as revised by OMB (1997) and then bridging from old data to new, the National Postsecondary Education Cooperative (NPEC)[4], NCES, and the National Science Foundation (NSF) co-sponsored a Policy Panel on Racial/Ethnic Data Collection, which met in March, 1998 and February, 1999, and issued recommendations in April 1999 (see Appendix for a timeline of events involving changes to racial/ethnic data collection and reporting). The Policy Panel identified the two major changes as: “students and staff will be able to indicate if they are members of one or more races; and Asian or Pacific Islander racial category was split into two separate categories – ‘Asian’ and ‘Native Hawaiian or Other Pacific Islander’” (Westat, 1999, p. 1). The purposes of the 1998 meeting were

• To bring together appropriate organization representatives to decide how and when to implement the transition to the new Federal standards, and
• To ensure that institutional researchers, admissions/registrars, and computer specialists at colleges and universities come to a consensus about the Federal collection of data in IPEDS. (p. 1)
  The purposes of the 1999 meeting were
• To review the draft OMB Tabulation Guidelines Report (and administrative data collections) and to provide comments to OMB;
• To make final recommendations regarding the changes to the IPEDS survey forms, and the timing and process for implementing these changes; and
• To develop a plan to disseminate information to all stakeholders on the implementation of the new OMB racial/ethnic standards. (pp. 1-2)

Based on the assumption that it would take two years to create and adopt appropriate software and two years to become consistent in collecting and maintaining racial/ethnic data across institutions, the panel predicted, “It would probably take institutions 4 years to have complete, consistent data, beginning from the time of their commitment to make the changes” (Westat, 1999, p. 5). Furthermore, in terms of readiness as of 1999 to comply with the revised OMB Directive 15, according to the Policy Panel report,

Institutions of higher education are generally aware of the classification changes, but they are all waiting on official, definitive guidance from OMB and NCES before they begin implementing the changes. However, not all information technology staff are aware of these changes, especially since many institutions are focusing their technology resources on Y2K efforts. (Westat, 1999, p. 4)

So by February 1999, it was already fairly clear that it would not be possible for institutions to meet the Directive’s January 1, 2003 deadline for implementation, even if guidelines for collecting and reporting racial/ethnic were immediately forthcoming from OMB.

Because of the central role of IPEDS in postsecondary data collection and reporting, a consensus emerged among higher education stakeholders (e.g., NCES, NSF, institutions, national testing organizations, etc.) that a planned redesign of the IPEDS survey could serve as both model and method for facilitating institutional conversion to the new standards. The Policy Panel recommended a timeline for transformation of the process, proposed to begin March 1999 with NCES issuing written recommendations from the Policy Panel and to end in fall of the 2002-2003 academic year with the IPEDS Fall Enrollment survey. The Policy Panel further recommended that “implementation of the changes should occur simultaneously in all institutions” (Westat, 1999, p. 9), with the exception of the Graduation Rate Survey, which should be phased in because of difficulty in ascertaining under the new standards the race/ethnicity of students who had left institutions prior to graduation.

Data collection. In response to Directive 15’s allowance for a one- or two-question format for collecting data, the Policy Panel recommended the one-question format, unless states required the two-question format. Use of this format would result in six racial/ethnic categories: American Indian/Alaska Native, Native Hawaiian & Other Pacific Islander, Asian American, Black/African American, White, and Hispanic/Latino. The Policy Panel (1999) further recommended that institutions use eight data fields – these six categories plus a “bridge field” from the old coding scheme or the new data and a “flag field” to indicate whether the bridge field was from old or new data (p. 6).

The NCES Taskforce for IPEDS Redesign (1999) ultimately recommended that institutions could choose either the one- or two-question format for obtaining race/ethnicity data and a modified version of the Policy Panel’s two-table reporting format. The two-question format better meets the OMB requirement that ethnicity be reported independently from race and is posed as:

1. Which best describes you? (Choose one)
            a. Hispanic/Latino
            b. non-Hispanic/Latino
2. Which of these best describes your background? (Choose one or more)
            a. American Indian/Alaska Native
            b. Asian
            c. Black/African American
            d. Native Hawaiian/other Pacific Islander
            e. White
(NCES Taskforce for IPEDS Redesign, 1999, p. 25)
The one-question format is:
            Which of these best describes your background? (Choose one or more)
                        a. Hispanic/Latino
b. American Indian/Alaska Native
            c. Asian
            d. Black/African American
            e. Native Hawaiian/other Pacific Islander
            f. White
(NCES Taskforce for IPEDS Redesign, 1999, p. 26)

In either case, institutions would have sufficient information to report respondents of single racial groups, multiple combinations of races, and either Hispanic/Latino or non-Hispanic Latino (either by choosing the appropriate response in the first question of the two-question format or by choosing or not choosing Hispanic/Latino in the one-question format). Both formats clearly meet the requirements of OMB Directive 15 (1997) for collecting data on race and ethnicity.

Data reporting. In response to the OMB Draft Provisional Guidelines (Tabulation Working Group, 1999) for reporting data, the Policy Panel recommended:

A new reporting format that includes a total of 17 columns in two tables. The first table consists of six categories of individuals reporting each race/ethnicity alone, Hispanic/Latino plus one or more races, and Non-Hispanic plus more than one race. This first table provides ‘minimum/lower bound’ count for each group. The categories for the first table are:

            • (1) All Non-Resident Aliens
            • United States Citizens and Permanent Residents Heading
                        (2) Unknown Race/Ethnicity
                        (3) American Indian and Alaska Native only
                        (4) Asian American only
                        (5) Black/African American only
                        (6) Native Hawaiian & Other Pacific Islander only
                        (7) White only
                        (8) Hispanic/Latino only
                        (9) Hispanic/Latino plus one ore more races
                        (10) Non-Hispanic plus more than one race
            • (11) TOTAL (Computed field: sum of the above 10 columns)
            • (11a) Total number reporting more than one race regardless of ethnicity. (NCES
  should only add this category if required by OMB.)
The second table includes six categories for individuals reporting each race/ethnicity alone and in combination with other categories. The second table provides a “maximum/upper bound” count. The categories for the second table are:
            • Any selection of race/ethnicity alone or in combination (U.S. citizens and
  permanent residents only)
            (12) American Indian and Alaska Native
            (13) Asian American
            (14) Black/African American
            (15) Native Hawaiian & Other Pacific Islander
            (16) White
            (17) Hispanic/Latino (Computed field; sum of 8 and 9 above).
(Westat, 1999, pp. 10-11)

In essence, the first table provides an unduplicated count and the second allows for multiple counting of individuals who indicate more than one race. Through the second table, the framework allows mixed race individuals to have all of their races counted, but the first table fails to distinguish among mixed race people in the two ethnicity categories (indicated in fields 9 and 10). Furthermore, both tables conflate the six OMB categories for race and ethnicity (i.e. there is no separate count of Hispanic/Latino and Not Hispanic/Latino by race).

Having recommended either the one- or two-question format, and noting that the answers to either would result in a “very large” number of possible race categories, the NCES Taskforce for IPEDS Redesign recommended:
that NCES adopt the 16 race/ethnicity categories recommended by the NPEC/NCES/NSF Policy Panel on Racial/Ethnic Reporting to report aggregated data for the Fall Enrollment Survey, the Completion Survey, the Fall Staff Survey, and the Graduation Rate Survey. These 16 categories are listed below.

1. Non-resident aliens
            U.S. citizens and resident aliens
2. Unknown race/ethnicity
3. American Indian/Alaska Native only
4. Asian only
5. Black/African American only
6. Native Hawaiian/other Pacific Islander only
7. White only
8. Hispanic/Latino only
9. Hispanic/Latino and one or more races
10. Non-Hispanic and more than one race
            Computed unduplicated total count (sum of 1 through 10 above)
11. American Indian/Alaska Native alone or in combination
12. Asian alone or in combination
13. Black/African American alone or in combination
14. Native Hawaiian/other Pacific Islander alone or in combination
15. White alone or in combination
16. Hispanic/Latino alone or in combination
Categories 11 through 16 are for reporting maximum counts of individuals with a particular racial/ethnic background. For example, “white alone or in combination includes all individuals who report “white only” and white and any other race/ethnicity.
(NCES Taskforce, 1999, p. 26)

Known as the “16-category framework,” this recommendation, one of a number of recommendations made in August 1999 for revisions to IPEDS, was to be implemented as optional for institutions in 2000-2001 and 2001-2002, but required for all in 2002-2003.
A hitch in the giddyup: The NCES 16-category framework is put on hold while an OCR format moves forward

Only three months after the NCES Taskforce for IPEDS Redesign made its recommendations, which were in turn based on the recommendations of the NPEC/NCES/NSF Policy Panel, NCES alerted the postsecondary education community that “the status of the changes in race and ethnicity reporting that were to be implemented with the 2002 IPEDS Fall Enrollment Survey are currently being revisited” (NCES, 1999). According to the notice, the 16-category framework was being countered with “several alternative strategies for reporting aggregated race/ethnicity data…which might be more acceptable to the various federal civil rights agencies (e.g., the Office of Civil Rights and the Equal Employment Opportunity Commission) and/or the Office of Management and Budget” (NCES, 1999)[5]. As a result, NCES was “strongly recommending that institutions do nothing at this time to change their current race and ethnicity reporting systems and formats, but should wait until more definitive decisions on the direction of aggregate reporting of race and ethnicity have been made” (NCES, 1999, italics added). The notice went on to indicate that NCES had delayed plans to have draft IPEDS reporting forms available, would be meeting with OCR, and was urging OMB “to provide more guidance on this issue to the federal statistical community with the hope of getting some resolution to this situation as soon as possible” (NCES, 1999).

The abrupt halt to IPEDS revisions revealed a flaw in the apparently seamless development of new mechanisms for collecting and reporting data on race and ethnicity in postsecondary education. From the formation of NPEC in 1994 to the NPEC/NCES/NSF Policy Panel reports in 1998 and 1999, to the adoption of the 16-category format for reporting race/ethnic data by the NCES Taskforce on IPEDS Redesign, the progression of ideas and recommendations had proceeded smoothly, if not speedily. And because the Department of Education was a member of the OMB’s Interagency Committee for the Review of Racial and Ethnic Standards (ICRRES), of which the proposal to revise Directive 15 was a product and whose work continued in the form of the Tabulation Working Group of the Interagency Committee for the Review of Standards for Data on Race and Ethnicity, it seems somewhat surprising that the Policy Panel and NCES Taskforce got as far as they did before learning of objections that would fully derail the process of adopting the 1997 revisions to Directive 15 in the postsecondary education sector.

In fact, the matter remains unresolved. In July 2002 the Association for Institutional Research (AIR) issued a briefing alert that included the following:

The new federal guidelines will allow students and employees to indicate more than one racial category, using the following five categories as a minimum (more detail will be allowed within each category): American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian of Other Pacific Islander, and White. The American Indian or Alaska Native category has been changed to include individuals from South American and Central American countries. In addition, students and employees will choose one ethnicity using the following two categories as a minimum: “Hispanic or Latino” and “Not Hispanic or Latino.” “Multiracial” or “Multiethnic” will not be acceptable response categories. The aggregate reporting requirements for individuals who choose more than one race have not been determined. (Sapp & Fuller, 2002a, Current Status)

The alert indicated that although various proposals have been offered for standards for aggregate reporting, “No decision has been made. Furthermore, a target date for such a decision has not been announced. Until it is, institutions should not take any further action (Sapp & Fuller, 2002a, Summary, italics added). Specifically,

Institutions should not make any further changes at this time in forms, databases, or computer programs until a final decision has been made. After the collectionand reporting processes have been finalized by the federal government, collection instruments (e.g., for admissions, registration, and employees) will need to be updated, and students and employees may need to be re-surveyed. Changes will be needed in databases (allowing for all possible combinations of race and ethnicity categories), and computer programs used to generate reports will need to be updated. In addition, a mechanism for retaining “old” codes should be created for internal crosswalk purposes and for Graduation Rate Survey (GRS) cohorts that have matriculated but not yet reached 150% of normal time to award. Coordination among a number of offices (e.g., Admissions, Registrar, Human Resources, Information Technology, and Institutional Research will be necessary. (Sapp & Fuller, 2002a, Implications for Institutions)

Because of the extended timeline from adoption of new IPEDS requirements to institutional transformation (a total of four years, according to the NPEC/NCES/NSF Policy Panel), there was no possibility of meeting the January 2003 deadline for adoption of the changes to Directive 15. The latest word (Sapp & Fuller, 2002b) indicated that the “deadline is no longer in effect” (Summary), aggregate reporting decisions have not been made, and “reporting for postsecondary institutions is currently ‘on hold’” (Current Status). The most recent AIR alert ended: “Unknown – Date of a final decision on guidelines for IPEDS reporting and actual deadline for compliance with the new OMB standards on race/ethnicity” (Sapp & Fuller, 2002b, Timeline, ¶ 2).

While the IPEDS proposal for a 16-category framework was put on hold, OMB issued a bulletin Guidance on Aggregation and Allocation of Data on Race for Use in Civil Rights Monitoring and Enforcement (OMB, 2000) to assist various federal agencies in their efforts to work with data on race used in relation to civil rights. This bulletin was also incorporated in the Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (Tabulation Working Group, 2000). It stipulates a framework for reporting race only (i.e. not including information on Hispanic/Latino ethnicity), including the five single races, the four most common double race combinations from the 2000 Census, any additional combinations that represent 1% or more of the population in a jurisdiction, and the balance of individuals reporting more than one race[6]. The bulletin provides this example:[7]

1. American Indian or Alaska Native
2. Asian
3. Black or African American
4. Native Hawaiian or Other Pacific Islander
5. White
6. American Indian or Alaska Native and White
7. Asian and White
8. Black or African American and White
9. American Indian or Alaska Native and Black or African American
10. > 1 percent: Fill in if applicable                            
11. > 1 percent: Fill in if applicable                            
12. Balance of individuals reporting more than one race
13. Total
(OMB, 2000, Aggregation Guidance, ¶ 2)

The specificity of the four most common combinations – as well as any others that might apply in particular circumstances – could not be met in aggregated data by the 16-category framework proposed by NCES, though it could be met at the individual respondent level. For example, a respondent who marks Asian and White would be counted in items 10, 12, and 15 of the NCES framework, and in item 7 of the OMB framework. But once the data from multiple respondents were aggregated at the institutional level, it would not be clear from the NCES table how to allocate those individuals in the OMB table. Asian-White mixes would be indistinguishable from other Asian mixes, and the presence of individuals indicating three or more races would further confound the data. This example illustrates the importance of developing guidelines for aggregating data on race and ethnicity so that data can be compared within and across institutions and agencies. Though the OMB guidelines were imposed from outside the postsecondary education data community, they provide a more clear system for obtaining an unduplicated count that captures more of the specific racial backgrounds of individuals than does the NCES 16-category framework and may ultimately prove more useful, especially if they are adopted as the standard by other agencies for reporting data on race.

Challenges for data analysis: Bridging old (1977) and new (1997) data formats

If the changes were only a matter of data collection and aggregate reporting, the delay in adopting them would not seem to be more than a bureaucratic inconvenience. But racial/ethnic data are needed to inform important public policy questions in a number of arenas, including higher education. The need to “provide the most accurate and informative body of data” to “those Federal Government officials charged with carrying out constitutional and legislative mandates, such as redistricting legislatures, enforcing civil rights laws, and monitoring progress in anti-discrimination programs” (Tabulation Working Group, 1999, pp. 8-9) is identified repeatedly in OMB documents related to the development of guidelines for implementation of revisions to Directive 15. And for the higher education community, the ability to compare data within and across institutions, state systems, and nationally is essential to understanding issues of access, equity, and success in the postsecondary education sector. As with proposals to collect and report data, proposals for bridging data have been put forward by a number of stakeholders, culminating in recommendations from the Interagency Committee for the Review of Standards for Data on Race and Ethnicity’s Tabulation Working Group in Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (2000). In this section, we briefly summarize these recommendations as they apply to postsecondary education data[8].

The Tabulation Working Group’s guidelines provide for a “bridge period,” during which agencies are permitted to use both the new data collected under the 1997 standards and a “bridging estimate,” which is “a prediction of how the responses would have been collected and coded under the 1977 standards” (Tabulation Working Group, 2000, p. 85). This process allows for new data to be compared more meaningfully to the older format, though it does not attempt to estimate how old data might be seen under the new standards (which would not be possible, since respondents could indicate only one racial category and there is no way to determine how many individuals might have done so). In essence, the bridging estimates attempt to collapse the new data, with multiple race responses, into the old categories for race and ethnicity.

Several acceptable bridging methods were proposed, all of which involve the use of individual-level responses. The methods fall into two major types: “whole assignment,” which places each individual into a single racial category, and “fractional assignment,” which places individuals indicating more than one race into multiple categories (Tabulation Working Group, 2000, p. 87). Within whole assignment methods, individuals can be assigned on a deterministic set of rules (e.g., all White and Black/African American responses are assigned to the Black/African American category) or a probabilistic rule that randomly assigns a certain percent of multiple race responders to specific single race categories (e.g., 75% of White and Black/African American responders are assigned to Black/African American and 25% are assigned to White). A fractional assignment method could either assign a straight fraction of each response to the appropriate categories (e.g., a White and Asian respondent is assigned .5 to White and .5 to Asian) or in some ratio representing the amount of time that individual might identify with one group compared to another (e.g., a White and Asian respondent could be assigned .25 to White and .75 to Asian). Attempting to estimate how often an individual might identify in different groups is a messy and political business and therefore not likely to be incorporated often in strategies for reporting data on race and ethnicity.

Although whole assignment methods may be easier to manage statistically and fit better with a sense of one-body-one-response (avoiding the “half a person” effect of fractional assignment), they fail to meet a primary goal of the OMB’s revision to Directive 15, the ability of respondents to self-identify. Fractional assignment promises to better meet this goal, although the identities of mixed race college students are not necessarily well represented by simple division (see Renn, 2000, 2003; Harris & Sim, 2002; Rockquemore & Brunsma, 2002; Wallace, 2001). Allen and Turner (2001) offer strategies for fractional assignment based on 1990 census responses regarding ancestry. Allen and Turner’s strategies hold promise for those institutions that want to engage in probabilistic fractional assignment when reporting aggregate data while representing, to the best extent possible, the identities of mixed race respondents.

It is crucial that institutions determine a common method to bridge data across time and sources. In an analysis of six different bridging methods using race data from Census 2000, Grieco (2002) found that the various methods did not produce consistent estimates for all race groups. The NPEC/NCES/NSF Policy Panel recommended that NCES should build a 2-3 year bridge from new to old data, in part because a national bridge would be useful to institutions needing to bridge their own data. The need to bridge institutional data would continue unless and until institutions resurveyed continuing students for IPEDS compliance and staff for EEO reporting requirements; the Policy Panel therefore recommended, “institutions should be strongly encouraged but not required to resurvey continuing students” (Westat, 1999, p. 14). The Policy Panel further recommended that NCES should report data in only one format, that institutions report all data to NCES using only one format in a given year, that institutions should convert old data to the new format before reporting it, and that new data should be reported as new data rather than attempting to collapse it back into the old formats (p. 14). These recommendations were made with the suggestion that the 16-category framework be adopted and work less well with the 2000 version of Provisional Guidance, but they convey in clear terms the recommendation that NCES provide leadership to institutions in bridging between new and old data.

With the derailment of the plan for the 16-category framework and the subsequent silence from NCES regarding revisions to race and ethnicity data collected and reported for IPEDS, institutions are left with no additional guidance on collecting, reporting, or analyzing data. While this silence hardly represents a crisis of national policy, there are a number of important cases around the country that would benefit from resolution of this issue. Even where data on race and ethnicity are not evidence for public policy debates and civil rights monitoring, institutions attempting to track their own progress in terms of access, participation, and success across racial and ethnic categories wait in statistical limbo for guidance from NCES. In the next section, we use results of studies of data collection and reporting practices to illustrate challenges of interstate and inter-institutional comparisons of racial/ethnic data in education.

Racial and ethnic data collection and reporting in postsecondary education

In February 1997, the State Survey on Racial and Ethnic Classifications was conducted for NCES and the Office for Civil Rights (OCR) to review OMB’s Directive 15 and assess the quality and efficacy of the use of the data collected by the state departments of education using the five standard federal categories for race and ethnicity. The results included information in every state except Hawaii. Is this study, eight states reported using categories other than the five standard categories mandated by the federal government; some states included a “multiracial” category while others added separate categories to identify specific populations within one of the five categories requested by the government (e.g. “Filipino” was a separate category in California). Seventeen states reported that changing to the Directive 15 standards would affect their ability to project enrollment and may even “prevent comparisons across time.” In addition, 31 states reported requests from schools and parents to add a “multiracial” category; 20 states agreed with these requests to add a “multiracial” category to “reflect the nation’s increased diversity” or to “reflect the growing population of mixed-race individuals.” Since 1997, primary and secondary school systems have been working with NCES and OCR to represent their student populations and comply with new federal standards; evidence demonstrates that similar cooperation in the postsecondary sector is lacking.

In a fall 2002 study of 127 U.S. postsecondary institutions (Renn & Lunceford, 2002), the majority of institutions had not yet implemented the 1997 revisions to OMB’s Directive 15 or expanded the 1977 racial/ethnic categories. Sixty-two percent of institutions in the study used the “old” reporting categories. Although there were differences in how institutions collected and reported racial and ethnic data by Carnegie classification, region in which the institution was located, institutional type (public or private), and by the racial composition of the student body, the differences were not statistically significant.
The overall finding that 62% of institutions had not yet implemented OMB’s 1997 revisions to its Directive 15 is striking. The Policy Panel estimated that it would take institutions four years for institutions to implement changes to their methods of data collection (Westat, 1999). Five years after OMB issued the 1997 revisions to the standards for data collection and reporting, less than 40% of institutions had made changes. In addition, only 17% of the institutions in the study offered students the option of marking more than one category for race and/or ethnicity – a critical element of the new standards. Four out of five institutions will have to make this substantial change in the near future in order to comply with Directive 15.

Currently NCES recommends that institutions not change their current race and ethnicity reporting systems, but rather wait for a definitive decision from the federal government (Sapp & Fuller, 2002b). The majority of the institutions contacted in the 2002 study reported that they were waiting for instructions from IPEDS administrators at NCES to make any changes to the categories used to collect student racial/ethnic data, as advised by OMB. While we applaud the willingness of institutions to hold off individual changes in anticipation of guidance from NCES, we are concerned that the spirit of the OMB directive, which could have been honored at the institutional level, has been set aside in favor of bureaucratic processes.

Changing social contexts for the collection of data on race and ethnicity

Often omitted from discussions on racial classification is why individuals self-identify as they do and how self-identification changes over time and across contexts. It is outside the scope of this paper to describe all of the influences on racial/ethnic self-identification, but it is important to understand some particular challenges to the reliability of data on race and ethnicity in postsecondary education. Many mixed race adolescents and college students do not identify unilaterally across social contexts (e.g., Lopez, 2003; Renn, 2000, 2003; Rockquemore & Brunsma, 2002; Wallace, 2001). Instead, they may identify situationally according to a range of sociocultural influences. In a school-based survey with a nationally representative sample of adolescents, Harris and Sim (2002) found that 12% of middle and high school children provided inconsistent answers to almost identical questions asking them to indicate their race and/or ethnicity. School-based surveys were conducted with 83,135 high- and middle-school children and 18, 924 interviews were conducted at home with the same students. While 8.6% of the adolescents reported being multiracial at home or in school, only 1.6% reported multiracial in both contexts, and only 1.1% selected the same combination of two or more racial groups in both contexts. As a result, there are discrepancies in multiracial reporting in school and at home – 54% of those reporting as multiracial at home are not multiracial in school data, and 75 % of the school multiracial population is not multiracial in home data.

Fluidity in racial identification may be influenced by an individual’s age and place or region of residence. For example, individuals who grew up when law or common practice dictated that anyone with any African American heritage, even a very small fraction, was classified as Black may still continue that classification, or pass their views to family members; multiracial individuals may thus be influenced to identify in only a single monoracial category. Differences between socially distinct monoracial groups may also cause fluidity of self-identification (e.g., Renn, 2000, 2003). Harris and Sim (2002) found different patterns of those who provided inconsistent self-reported data on race by different racial/ethnic groups. While a larger percentage of individuals who identified as both “white” and “black” at school identified as only “black” at home, a larger percentage of individuals who identified as “white” and “Asian” at school identified as “white” at home. There is no common understanding or practice of self-identification of race or ethnicity, especially among multiracial individuals or in families that have multiracial children, who may identify differently from one another.

Over a lifetime, individuals may change how they identify in terms of race and ethnicity. When collecting data from one census to the next, if the person who completes the information for an entire household has a different view of race from that of the person who completes the next census, racial classifications, even in this single household, may be vastly different. In attempts to determine the best method to collect and bridge data across time and sources, it is important to understand that concepts of race and ethnicity change over time and context, and methods of treating and interpreting data must account for – or at least acknowledge – these changes.

Implications for policy and practice

It is clear that higher education is in the early process of a major change in policies and procedures surrounding the collection, analysis, and reporting of data related to race and ethnicity. The postsecondary sector failed to meet the January 1, 2003 deadline of the OMB’s 1997 Directive 15, and it is not yet clear when changes will occur. The consequences of missing that deadline do not appear dire; there is nothing in the directive and no indication elsewhere that institutions or agencies will be penalized for failing to meet this deadline. Given the advance time required to implement such a shift in educational institutions (estimated at four years by the NPEC/NCES/NSF Policy Panel), it would have been a fairly close call to implement the changes in any case, but the apparent disconnect between NCES and other stakeholders rendered the task unfeasible by January 2003. And although the deadline came and went without systemic changes in postsecondary data collection and reporting, the changes are inevitable and will impact data at the institutional, state, and national level.

Decisions about data collection seem reasonably solid; there has been no public discussion of changing from the option of a one-question or two-question format for collecting data from students and employees. These formats are compatible with the 2000 Census and should yield data that can be made comparable across data from a number of federal sources, although the fluid nature of multiracial students’ identities casts doubt on the stability of any static data set. It is not clear whether or not institutions will be required to re-survey all students and employees. We anticipate that the number of individuals who refuse to answer a question on race would be significant and could have an adverse effect on the value of data collected. Still, a requirement to re-survey could provide valuable information for institutional data bridging purposes and we concur with the NCES Taskforce on IPEDS Redesign that institutions should engage in this process. Students could be required to re-identify themselves (or to indicate that they decline to identify themselves) prior to registration and employees could be required to do so as a condition of employment, perhaps in conjunction with annual benefits review and subscription. At many institutions, an online process would not be difficult to implement or manage.

The current situation for data collection, as evident through a preliminary study of 127 postsecondary institutions (Renn & Lunceford, 2002), is nearly consistent in timing of data collection (all but two collect from students on application and do not re-survey), but quite divergent in how questions are asked, how many responses students may supply, and what categories are available. While there will certainly remain some diversity among categories available, standardization of how the information is requested will be an asset to the data community, institutions, and students themselves. Institutions have by and large heeded the NCES call to stay the course and have not changed questions in anticipation of the new standards; indeed the vast majority still requires students to indicate only one racial category. While we are somewhat frustrated with the resistance to adopting a multiple response option, at this time we believe that the interest in consistency in data collection outweighs the interests of individual respondents and, perhaps, political activists seeking information on multiple responders.

Data aggregation and reporting however, represent much more complex tasks that must be coordinated within and across institutions. Comparability of data is critical to conducting peer analyses; to analyzing institutional, systemic, and national trends; and to Civil Rights monitoring and enforcement (see Lopez, 2003). Two options are currently in discussion in the postsecondary education community – the 16-category framework and the OMB Civil Rights format. Although there are certainly other formats for reporting data, we discuss the strengths and weaknesses of these two.

The OMB format has been adopted and is in use nationally; there are strong arguments for having postsecondary data reported in this format. It facilitates comparisons across agencies and purposes; it is a fairly straightforward approach that includes an unduplicated count; it can be adapted to specific populations whose racial combinations may vary from the national norm; it provides specificity in racial combinations in ways that the 16-category framework does not. Furthermore, bridging from pre-1997 data is not terribly complicated using this format. What it lacks, however, is any accounting for ethnicity. It is a race-only framework that requires additional tabulation to indicate Latino or Hispanic heritage. It also fails to recognize the specific combinations of races falling below one percent of the total, which while not appearing to be a significant disservice to individuals does violate the “self determination” philosophy of OMB Directive 15 (1997). While OMB very clearly stated that this framework was for the purposes of Civil Rights monitoring and enforcement only, the adoption of this policy across agencies indicates that it will be a presence and a precedent for federal agencies.

The NCES 16-category framework could still be adopted for IPEDS and other purposes. It is especially useful to higher education because it includes non-resident aliens, who constitute an important segment of the postsecondary population; it includes an “unknown” category, which can be important in understanding the relative proportion of other individuals counted; and it includes ethnicity. Although somewhat bulky, the combination of unduplicated total (lines 1-10) and duplicated figures (lines 11-16) provides a more accurate picture of campus populations and, more so than the OMB unduplicated format, meets the spirit of self-determination. What it lacks, though, is specificity. It is not clear from this framework how many of the “alone or in combination” respondents are “in combination” with other specific races. There is no way without examining individual records to determine the number of, for example, Asian-White respondents. On some campuses, thisinformation could be important, as could the ability to compare this information across institutions, especially within state systems.

Although neither is perfect, and there are undoubtedly other models, we recommend the adoption of the OMB model. It includes more specificity than the NCES framework, it is less confusing than the unduplicated-duplicated dual system, and from it the duplicated information could easily be constructed if it were of interest. We recommend, however, that the OMB model be modified to include Hispanic/Latino as an equal category to the five “races” and to include Hispanic/Latino as appropriate in the four most common combinations reported in the table. Because 7.7% of those under age 18 indicating more than one race also reported Latino or Hispanic ethnicity (U.S. Census Bureau, 2001), it is critical to provide a means to express this identity in educational demographics. Our recommendation challenges OMB’s differentiation between race and ethnicity and, although we again acknowledge the contested nature of the discussion, asserts that the lived experience of students identified as Latino or Hispanic constitutes that on par with students identified with so-called racial groups.

Current practices vary widely, with no consistent method for institutional presentation of data outside the IPEDS process. Even within systems that collect data through common admissions applications (e.g., the California State Universities), data presentation is inconsistent. There will always be reasons for institutional variation in data presentation, and some institutions will want to present far more detail than the 16-category or OMB formats. We recommend, however, that institutions make available data in whatever format is adopted by IPEDS, as well as institutional reporting formats, with information on how individuals who indicated more than one race were assigned to categories (if this is not apparent from the final model adopted by IPEDS). At present, it is difficult to compare data across institutions from information available publicly. And it is even more difficult to determine how multiracial individuals are represented in the data presented.

Recommendations for bridging and analyzing data hinge in part on the framework adopted for aggregation and reporting. Some important philosophical elements, however, apply to any plan for bridging old and new data. These bridge calculations involve highly charged political stakes and personal identities, as the process essentially involves re-assigning individual responses from the categories indicated by individuals into the “old” single response categories.
As Allen and Turner (2001) indicate, there are a number of ways to assign the responses of mixed race individuals to single categories that are acceptable under the OMB’s Provisional Guidance (Tabulation Working Group, 2000). We are least comfortable with “whole assignment” models that call for unilateral assignment of all individuals with White and other racial heritage into the category of that other race. This is, in fact, the OMB’s method in Civil Rights monitoring and enforcement calculations, and we are troubled by the ways this model inflates the number of people in racial categories other than White, no matter how those individuals might look, feel, or act.

We support Allen and Turner’s (2001) recommendations for “fractional assignment,” although in responding to the individual’s right to self-determination they interfere somewhat with the institution’s need to have data that are readily analyzed by whole numbers. If institutions adopted an overlapping data collection process (e.g., collecting data with “old” and “new” questions for a few years), an institution-specific scheme for fractional assignment could be determined that would account for regional racial and ethnic history and trends in self-determination. In the absence of the resources or will to conduct such a bridging study, many institutions could rely on Allen and Turner’s work for guidance on fractional assignment. Once the final IPEDS categories are established, this issue should be revisited for further clarification.

Recommendations for future research

Our article represents two strands of work – the policy context and current practices – on the topic of postsecondary data collection and reporting related to race and ethnicity. Yet we are left with a number of questions and a number of suggestions for future research in each of the study’s areas. And, of course, we are eager to see what will happen when NCES issues final recommendations for IPEDS revisions in the area of race and ethnicity. The research represented in this paper could be expanded in a number of interesting ways, some of which we suggest here.

First, an in depth qualitative study of the change process would be useful and interesting. More information on decisions within the NPEC/NCES/NSF Policy Panel and the NCES Taskforce on IPEDS Revision could reveal insights into political processes and stakeholder influences. How, for example, was the 16-category framework developed, whom would it benefit, and why was it abandoned? Can a resolution be reached that meets the same needs yet also meets OMB and other agency concerns and mandates? How can similar situations (such as developing and then abandoning a solution) be avoided in the future? After NCES issues new IPEDS guidelines, a follow up study to see how these guidelines were ultimately developed and are being received by a number of stakeholders would be in order.

A broader study, placing the postsecondary data community in the context of a larger federal data community could also be useful in understanding how postsecondary education is seen and how it might exert its influence at the federal level. Certainly every agency within the government, and the many external constituencies that must collect and report data on race and ethnicity, are affected by OMB Directive 15 (1997). Whose concerns are being heard most clearly? How do their concerns match – or conflict with – those of higher education? Should members of the postsecondary sector form political alliances differently to achieve mutually beneficial outcomes? How might they do so? A thorough document review of other agencies’ processes related to Directive 15, and then interviews with key informants, could provide useful information for future decision making around interagency initiatives and cross-agency mandates. And once the changes are in place, a thorough review of the place of NCES and NPEC in the final decision would be instructive as well.
With regard to current practices for data collection and reporting, there are a few questions that would benefit from additional research even before changes are finalized. For example, why have some institutions changed from “old” to “new” and/or “expanded” data collection? What forces operate on institutions to cause them to make these changes? How do institutions respond to local constituencies in this regard? How are they aggregating and reporting data that do not fit the prescribed IPEDS categories?

Once changes are mandated, a follow up study of how institutions ask for data on race and ethnicity (the one- or two-question format or other variation), as well as how they aggregate and report data would be important to understanding the national picture of data collection and reporting. It may be the case that after the changes are made, data remain incomparable across institutions, systems, and states. While many data analysis needs can be met without exactly comparable data, the needs of the postsecondary community and the public may not be best served if data cannot be aligned for easy comparison.
The issue of bridging will require further research to ascertain the effectiveness of different proposed bridging methods. The ability to compare old and new data sets for purposes of tracking access, retention, and completion is critical. The ability to compare trends across broad historical timeframes is already impeded by inconsistent data collection methodologies and terminologies, but it will be impeded further by inconsistent bridging practices. It will be important to examine bridging proposals and then actual bridging practices to determine the viability and effectiveness of various methods.


Allen, J. P., & Turner, E. (2001). Bridging 1990 and 2000 census race data: Fractional assignment of multiracial populations. Population Research and Policy Review, 20, 513-533.

Army Information Paper. (n.d.). Federal Interagency Committee on Education Information Paper. Retrieved October 20, 2002 from

Bers, T. (1997, November). National postsecondary education cooperative. AIR Alert: A briefing on emerging issues in higher education from the Association for Institutional Research, #3. Retrieved October 19, 2002 from

Davis-Van Atta, D. with Arnold, C., & Lyddon, J. W. (1998, June). New federal standards for racial and ethnic data collection and reporting. AIR Alert: A briefing on emerging issues in higher education from the Association for Institutional Research, #6. Retrieved April 23, 2002 from

Farley, R. (2001). Identifying with multiple races: A social movement that succeeded but failed? PSC Research Report no. 01-491. Ann Arbor: Population Studies Center at the Institute for Social Research, University of Michigan. Retrieved October 17, 2002 from

Grieco, E. M. (2002). An evaluation of bridging methods using race data from Census 2000. Population Research and Policy Review, 21, 91-107

Harris, D. R., & Sim, J. J. (2002). Who is multiracial? Assessing the complexity of lived race. American Sociological Review, 67(4), 614-627.

Lee, S. (1993). Racial classifications in the U.S. census: 1890-1990. Ethnic and Racial Studies, 16 (1), 75-94.

Lopez, A. M. (2003). Mixed-race school-age children: A summary of Census 2000 data. Educational Researcher, 32, 25-37.

National Center for Education Statistics (NCES). (1999, November 5). What’s new in racial/ethnic reporting for IPEDS. Retrieved April 17, 2002 from

National Center for Education Statistics (NCES). (2000). Appendix L: Revisions to the standards for the classification of federal data on race and ethnicity. Retrieved online April 23, 2002 from
NCES Taskforce for IPEDS Redesign. (1999, August 31). Integrated postsecondary education data systems (IPEDS): An improved system. Final report. Washington, DC: National Center for Education Statistics. Retrieved April 17, 2002 from

Office of Management and Budget. (1977). Standards for the Classification of Federal Data on Race and Ethnicity (Statistical Policy Directive No. 15). Washington, DC: Office of Management and Budget.
Office of Management and Budget. (1994). Standards for the Classification of Federal Data on Race and Ethnicity. Retrieved October 28, 2002 from

Office of Management and Budget. (1997). Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity. Retrieved April 23, 2002 from

Office of Management and Budget. (2000, March 9). OMB Bulletin No. 00-02: Guidance on aggregation and allocation of data on race for use in civil rights monitoring and enforcement. Retrieved April 22, 2002 from

Renn, K. A. (2000). Patterns of situational identity among biracial and multiracial college students. The Review of Higher Education, 23 (4), 399-420.

Renn, K. A. (2003). Understanding the identities of mixed race college students through a developmental ecology lens. Journal of College Student Development, 44, 383-403.

Renn, K.A., & Lunceford, C. J. (2002, November). Because the numbers matter: Transforming racial/ethnic reporting data to account for mixed race students in postsecondary education. Presented at the 27th meeting of the Association for the Study of Higher Education, Sacramento, CA.

Rockquemore, K. A., & Brunsma, D. L. (2002). Beyond black: Biracial identity in America. Thousand Oaks, CA: Sage.

Sapp, M., & Fuller, C. (2002a, July). Update on federal race and ethnicity reporting changes. AIR Alert: A briefing on emerging issues in higher education from the Association for Institutional Research, #14. Retrieved September 7, 2002 from

Sapp, M., & Fuller, C. (2002b, August). Follow up to Alert #14, Update on federal race and ethnicity reporting changes. AIR Alert: A briefing on emerging issues in higher education from the Association for Institutional Research, #14. Retrieved September 7, 2002 from

Spickard, P. R. (1989). Mixed blood: Intermarriage and ethnic identity in Twentieth-Century America. Madison: University of Wisconsin Press.

Tabulation Working Group, Interagency Committee for the Review of Standards for Data on Race and Ethnicity. (1999, February 17). Draft provisional guidance on the implementation of the 1997 standards for federal data on race and ethnicity. Washington, DC: Office of Management and Budget. Retrieved April 20, 2002 from

Tabulation Working Group, Interagency Committee for the Review of Standards for Data on Race and Ethnicity. (2000, December 15). Provisional guidance on the implementation of the 1997 standards for federal data on race and ethnicity. Washington, DC: Office of Management and Budget. Retrieved April 20, 2002 from

U.S. Census Bureau (1957). Statistical abstract of the United States. Washington, DC: Government Printing Office.

U.S. Census Bureau. (2001). Mapping Census 2000: The geography of diversity. Retrieved March 15, 2002 from

Wallace, K. R. (2001). Relative/Outsider: The art and politics of identity among mixed heritage students. Westport, CT: Ablex.

Westat. (1999, April 2). Recommendations of the Policy Panel on Racial/Ethnic Data Collection, February 11-12, 1999. Rockville, MD: Author.

Williams, K. (2000). Changing race as we know it? The political location of the multiracial movement. Unpublished manuscript. Cambridge: Harvard University, John F. Kennedy School of Government, Taubman Center for State and Local Government.
Timeline of Events Related to Development and Implementation of OMB Directive 15 and Subsequent Revisions
1973: HEW Secretary Casper Wineberger asks the Federal Interagency Committee on Education (FICE) to begin process of creating government-wide standards for racial classification
1977: Office of Management and Budget (OMB) issues Directive 15, mandating that federal agencies gathering demographic data use four major racial categories (White, Black, Asian or Pacific Islander, American Indian or Alaska Native). Spanish heritage (yes or no) was also to be determined for each individual.
1980: U.S. Census includes a delineation of the Asian or Pacific Islander category as Japanese, Chinese, Filipino, Vietnamese, Asian Indian, Hawaiian, Guamanian, and Samoan
1993: OMB begins formal review of standards for collecting, analyzing, and reporting federal data
1994: OMB creates Interagency Committee for the Review of Race and Ethnic Standards (ICRRES), representing over 30 federal agencies
October 1997: OMB issues revision to Directive 15; changes from “five acceptable racial and ethnic categories” to “five minimum categories for data on race: American Indian or Alaskan Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, White” and “two categories for data on ethnicity: Hispanic or Latino, Not Hispanic or Latino).” There is no Multiracial or Multiethnic option, but respondents must be offered option of indicating more than one racial category; mandated to take effect for Census 2000 and for all federal data by January 1, 2003
March 1998 and February 1999: National Postsecondary Education Cooperative (NPEC), National Center for Education Statistics (NCES), and National Science Foundation (NSF) sponsor meetings of a joint Policy Panel on Racial/Ethnic Data Collection (Policy Panel)
February 1999: Tabulation Working Group of Interagency Committee for the Review of Standards for Data on Race and Ethnicity (formerly ICRRES) issues Draft Provisional Guidelines on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity for comment and review by interested parties and agencies
April 1999: Policy Panel issues final report, including the 16-category framework for reporting data on race and ethnicity; recommendations incorporate recommendations of Tabulation Working Group’s Draft Provisional Guidance report
August 1999: NCES Taskforce for IPEDS Redesign accepts 16-category framework for use in IPEDS data collection and reporting
November 1999: NCES puts hold on recommendation for 16-category framework and tells institutions not to make any changes until additional guidance is forthcoming from federal government
March 2000: OMB issues bulletin on the collection and reporting of racial data for Civil Rights monitoring and enforcement; contains a format for reporting race that is incompatible with NCES 16-category framework
December 2000: Tabulation Working Group issues Provisional Guidelines on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity; includes the March 2000 OMB format for reporting race
July and August 2002: Postsecondary Statistics Division of NCES informs postsecondary education community that “the status of the changes in race and ethnicity reporting that were to be implemented with the 2002 IPEDS Fall Enrollment Survey is currently being revisited,” and “is strongly recommending that institutions do nothing at this time to change their current race and ethnicity reporting systems and formats.” January 1, 2003 deadline no longer applies.

[1] Definitions of “race” and “ethnicity” vary by source and interpretation. For the purposes of this paper we are using the federal government’s terminology and categories (five races and two ethnicities), which we acknowledge to be highly contested constructs.

[2] FICE was established in 1964 by executive order (OMB, 1994, Background, ¶1), and members have included the Departments of Education, Agriculture, Commerce, Defense, Energy, Health and Human Services, Housing and Urban Development, Interior, Transportation, and Treasury, as well as a number of non-cabinet agencies (Army Information Paper, n.d.).

[3] For a more complete discussion of the evolution and dissolution of the multiracial movement and the debate about adding a “multiracial” category to the census, see Farley (2001) or Williams (2000).

[4] In 1994, Congress authorized NCES to form NPEC “to identify and communicate on-going and emerging issues germane to postsecondary education, and to promote the quality, comparability and utility of postsecondary data and information that support policy development, implementation and evaluation” (Bers, 1997, Summary, ¶ 1). NPEC is “a compact of co-equal partners of postsecondary education data users and providers,” including “a broad spectrum of organizations and agencies representing all levels of postsecondary education that provide and/or use data: colleges and universities and proprietary institutions; state and federal government agencies; national associations; the media; accrediting associations; coordinating boards; and more” (Bers, 1997, Summary, ¶ 2).

[5] The 2000 update of Provisional Guidance (Tabulation Working Group, 2000) contains substantial requirements for the nature of tabulations (e.g., “data should be presented in as much detail as possible … subject to agency criteria for statistical reliability and confidentiality,” and “to the extent possible, Federal agencies should report data using the same set of standardized categories to facilitate comparisons across subject-matter areas and data systems” [p. 48]) on pages 48-59.

[6] OMB (2000, Allocation Guidance, ¶ 1) also provided guidelines for the allocation of multiple race responses for use in civil rights monitoring and enforcement as follows:
• Responses in the five single race categories are not allocated.
• Responses that combine one minority race and white are allocated to the minority race.
• Responses that include two or more minority races are allocated as follows:
• If the enforcement action is in response to a complaint, allocate to the race that the complainant alleges the discrimination was based on.
• If the enforcement action requires assessing disparate impact or discriminatory patterns, analyze the patterns based on alternative allocations to each of the minority groups.

[7] Although the bulletin was issued before the four most common racial combinations could actually be determined from the 2000 Census, the combinations provided in the OMB example were, in fact, the four most common reported by census respondents. Therefore, the bulletin’s example represents OMB’s actual final guidance on the matter.

[8] Lopez (2003) provides an overview of data related to mixed-race school children and a comparison for the tabulation of race and latino origin data from the state of California. She compares several options for data tabulation, including two “catch-all mixed heritage” categories, two combination categories, fractional assignment and multiple group assignment (p. 32).