Counting Race and Ethnicity:
Options for the 2000 Census
Judith Lichtenberg, Suzanne Bianchi, Robert Wachbroit, and David Wasserman
Long before Tiger Woods drew public attention to the problem of characterizing Americans of
"mixed race," the federal government had begun to reconsider the classification scheme
employed by the Census Bureau and other agencies in collecting and reporting data on race and
ethnicity. The existing scheme, created in 1977 under Office of Management and Budget
Statistical Directive 15, recognizes five basic categories: American Indian or Alaskan Native,
Asian or Pacific Islander, Hispanic, black, and white. The use of additional categories in data
collection is not precluded by Directive 15, but the reported results must be "organized in such a
way that the additional categories can be aggregated" into those on the basic list. On the Census
form--surely the most visible instrument for federal data collection on race and ethnicity--respondents must select only one of these basic categories, or else choose the nondescript
"Other."
Recently, the Census Bureau has conducted surveys testing alternative schemes, including one
that includes a separate "multiracial" category. As this issue goes to press, OMB is preparing to
release a report by an Interagency Committee for the Review of the Racial and Ethnic Standards,
recommending whether this change or some alternative should be implemented in time for the
2000 Census. After a period of public comment, a final decision will be made this fall.
Much of the impetus for a multiracial category has come from two organizations: PROJECT
RACE (Reclassify All Children Equally) and AMEA (Association of MultiEthnic Americans).
These groups argue that the current Census form forces those with parents of different races to
deny or suppress part of their heritage. For an increasingly large number of Americans, they say,
the government's insistence that individuals fit themselves into one of the existing racial
categories constitutes an unreasonable or even a repellent demand. On the other side, opponents
of a multiracial category argue that the advantages of granting recognition to a small class of
people are far outweighed by its administrative and social costs.
Unfortunately, the debate between these two camps has been governed by the assumption that the
only alternative to the present scheme is a menu of exclusive categories augmented by the
multiracial option. But there is a third alternative, which we believe is preferable to either
maintaining the status quo or adding a multiracial category: namely, a "mark one or more" option
that allows respondents to check more than one box, but does not offer a new category for mixed
race.
To see why this is the best alternative, it will be helpful first to explore the ethical and policy
issues surrounding racial and ethnic classification. We will consider the reasons why data on race
and ethnicity are collected in the first place, as well as the grounds for assessing the adequacy or
appropriateness of a classification scheme. We will then examine the case for creating a
multiracial category, sifting through the arguments on both sides of the public debate. While
some objections to the multiracial category seem to us misplaced, we conclude that the creation
of such a category would have serious drawbacks. We argue finally that the "mark one or more"
option avoids these drawbacks and is thus the superior option.
Purposes and Constraints
Any ethical or policy analysis of Directive 15 must begin with the recognition that there is no
such thing as the correct classification scheme for race and ethnicity (or for anything else, for that
matter). A classification scheme based on skin color is not more or less correct than one based on
ancestry; a scheme containing a composite category--e.g., Asian or Pacific Islander--is not more
or less correct than one that instead breaks the large category into several smaller ones.
Classification schemes are constructed for particular purposes, and so there are at least as many
classification schemes as there are purposes for classifying.
It follows that it is impossible to assess a classification scheme without a clear understanding of
the purposes for which it is designed and (not always exactly the same thing) the uses to which it
is put. Although, in terms of a given purpose, one classification scheme can be better than
another, no single scheme can claim to be the correct grid that articulates our diversity.
What purposes, then, are served by the collection of data on race and ethnicity? One is statistical
or demographic: the government tracks trends and shifts in population growth and distribution, as
well as correlations between different racial and ethnic groups and a variety of socioeconomic,
health, and educational indicators, to meet the needs of statistical agencies such as the National
Center for Health Statistics. A second purpose is legal and political: it is to meet the needs of
federal agencies responsible for civil rights monitoring and enforcement. These two purposes are
to some extent related; federal agencies are interested in correlations between race and various
indicators of wealth or well-being because they want to track the persistence and effects of
racism and discrimination. But since differences between racial or ethnic groups in
characteristics such as disease incidence may arise from sources other than racism and
discrimination, the two purposes are least partially independent.
We shall argue that a change in Directive 15 need not impede agencies in carrying out the
statistical and civil rights purposes of collecting data on race and ethnicity. At the same time, it is
important to realize that the significance of a racial classification scheme extends beyond these
purposes. Two considerations lead us to this view. First, the Census is mandatory: all Americans
are required to respond to it. Second, there is an obvious sense in which the Census conveys the
official view of race and ethnicity--granting recognition to certain categories but not others,
establishing a framework for thinking about diversity and social policy. For these reasons, its
classification scheme becomes a prominent social fact in its own right. As such, it has expressive
or symbolic significance beyond its explicit purposes.
It is in terms of this expressive and symbolic significance that we can best understand the
demand for changes in the present classification scheme by Americans who regard themselves as
of mixed racial or ethnic heritages. While we recognize that self-expression is not a purpose of
the Census, it is essential that the classification scheme in use not force people to violate their
own sense of their racial identity or that of their children. Given the expressive and symbolic role
of the Census, members of PROJECT RACE and AMEA are right to insist that official survey
forms not compel them to misrepresent their racial self-identification. That an increasing number
of people in our society genuinely, and reasonably, feel tied to more than one racial group is a
powerful reason for rejecting any framework requiring exclusive identification.
Some have sought to trivialize the failure of Directive 15 to accommodate people who regard
themselves as belonging to more than one category by arguing that the number who so regard
themselves is insignificant--less than the margin of error in the Census count. But although the
number may be statistically insignificant, it is not socially insignificant. The present
classification scheme reinforces a view of racial identity as exclusive and rigid--a view that has
serious political and cultural consequences beyond the Census itself.
Enforcement Issues
Defenders of the status quo argue have argued that the suggested changes to Directive 15 would
subvert the purposes of the racial classification scheme and have troubling symbolic
implications. One major concern among civil rights advocates and enforcement agencies is that
the inclusion of a multiracial category would hamper the government's antidiscrimination efforts.
They fear that significantly fewer people would classify themselves as "black," thereby creating a
distorted picture of the magnitude of racial discrimination.
The expectation of a large shift may be exaggerated, however. Results of test surveys suggest that
only a minute percentage of people who now classify themselves as black would shift to the
multiracial category. (The effect of including the category is greater on people of other racial
mixes, but that is not a primary concern of the civil rights community.) While the shift in self-identification among African-Americans could be greater in future Censuses as people became
accustomed to the new category, it would probably have only a very slight impact on the racial
data available for civil rights enforcement in the next decade.
Beyond this, we suspect that concerns about lost information rest on a misunderstanding of the
multiracial category--one that has not been addressed by its proponents. The public debate would
lead one to think that the multiracial category was "opaque"--that it provided no information
about constituent races. But in fact, the multiracial category being tested asks respondents to list
their constituent races. Even if many people who now classify themselves as black switched to
multiracial, almost all who did so would list "black" as part of their racial mix. Thus, there would
be no lack of information about their racial composition. (The same would be true, obviously, for
the "mark one or more" approach, which by definition would allow respondents an opportunity to
provide information about all the races with which they identify.)
These facts alone, however, are not enough to assuage civil rights concerns. Even if detailed
information about race continues to be collected, this will hardly matter for enforcement
purposes if this information is not reported or presented in ways that serve those purposes.
Unfortunately, proponents of change have had little to say about this critical issue.
Here is one possibility. Suppose that a classification scheme allowed people to list all the racial
categories to which they felt they belonged--whether under a multiracial category or by marking
one or more. At the data presentation stage, instead of reporting "the number of blacks with
household incomes over $60,000," the Census might report "the number of people with
household incomes over $60,000 who listed `black' as one of their constituent races," as well as
"the number of people with household incomes over $60,000 who listed only `black' as their
race." It has been argued that civil rights enforcement requires exclusive, single-race data
presentation--that we must be able to assign people who check multiple races to one of the
existing categories. But which one? And what about purposes other than civil rights
enforcement?
Clearly, the collected data would have to be transformed in some way. Such a transformation
would involve "assignment rules," so that individuals who checked more than one race would,
for certain purposes, be assigned to a single race for data presentation. Thus, for example, a
person who listed "black" and "white" as constituent races could be considered "black" for civil
rights purposes, but would not need to be reported as black for other purposes, such as health
statistics.
Proponents of change might well object to this strategy. They could argue that the government
was taking back with one hand what it gave with the other--offering the multiracial category or
"mark one or more" as a sop to critics of the present scheme, while continuing to relegate
multiracial Americans to the Directive 15 categories. The civil rights community might object
just as strongly. For if a strategy classifies people as black for civil rights purposes either if they
classify themselves as exclusively black or if they list black as a constituent race, that strategy
might appear to rely upon a version of the notorious "one-drop-of-blood" rule, which says that a
person with even a single black ancestor is to be classified as black.
These objections, however, can be answered. If, in monitoring civil rights enforcement, a federal
agency counted as black those who listed themselves as both "black" and "white" or as
"multiracial: black/white," it would not be deciding that their self-identification was mistaken--that they really belonged to a racial category other than the one they had chosen. It would merely
be recognizing that anyone with black ancestry is at risk of discrimination because of broad
social acceptance of the one-drop-of-blood rule. The pool of at-risk voters or job-seekers should
not be regarded as comprised exclusively of blacks, but of people with black ancestry. There is
no inconsistency in counting a person as a member of this pool while deferring to his self-identification as multiracial, or as white and black.
Of course, assignment rules would not be devised only for those who listed "black" as a
constituent race. In looking at employment or housing discrimination against Asian-Americans,
for example, one would consider all those who listed Asian as one of their races, since
"Eurasians" are likely to be subject to the same prejudices and resentments as "pure" Asians.
We believe that it is a virtue, not a flaw, of a proposal for racial classification that it recognizes
the legitimacy of grouping people differently for different purposes. For example, a person with
one black and one white parent could be counted among the growing number of multiracial
Americans for purposes of tracking intermarriage among traditionally segregated groups, but also
regarded as being at risk of anti-black discrimination in a society where racial bias still follows
the one-drop-of-blood rule. To object to such multiple groupings is to embrace a spurious
objectivity about race, which allows classification in only one racial group for any purpose
whatsoever. If we become accustomed to employing different groupings for different purposes,
we may learn to see racial identity as more fluid, indeterminate, and superficial, with the ultimate
effect of reducing race consciousness. In our judgment, that would be a good thing.
Social Impact and Symbolism
Opponents of the multiracial category have raised other objections, however, that are more
difficult to dismiss. While some of these objections are overstated, they suggest that inclusion of
a multiracial category may have a disturbing symbolism and a divisive social impact. The "mark
one or more" option, we will argue, avoids or significantly mitigates these problems.
First, some civil rights leaders argue that the multiracial category would undermine the solidarity
of a victimized community. They fear that it would encourage African-Americans with mixed
ancestry and/or lighter skin to deny their commonality with other black Americans in the hope of
acquiring a more privileged status. As we have seen, test surveys suggest that in the short term,
inclusion of the multiracial category would not result in widespread defection by people who
formerly classified themselves as black. Over time, however, the shift in self-identification
could grow more significant.
Critics also fear that the mere inclusion of the multiracial category would symbolically denigrate
"unmixed" and darker-skinned blacks. In their view, the multiracial category would be perceived
as an insertion into a racial hierarchy in which "black" is negatively valued, and "white"
positively valued. Instead of challenging these implicit valuations, it would help sustain them by
offering an apparent refinement of the classification scheme--bolstering its spurious claims to
objectivity by making it seem more accurate, like the elaborate, "scientific" schemes of overtly
racist countries.
One might think that the eclectic composition of the multiracial category--encompassing many
different racial and ethnic combinations--would prevent it from contributing to an official or
implied hierarchy of racial types. But the history of the "colored" category in apartheid-era South
Africa suggests otherwise. A category that includes a variety of combinations might still be
viewed as holding an intermediate value between black and white.
Proponents of the multiracial category naturally take a different view of its symbolic import. For
them, a central tenet of American racism has been the one-drop-of-blood rule, which conceives
of blackness as much in terms of taint as in terms of hierarchy. It is this racial outlook that helps
account for the historic American obsession with miscegenation. And it is this outlook, some
proponents say, that the multiracial category would symbolically repudiate, since it would defy
the principle that any trace of black ancestry defines a person as black.
This disagreement about the symbolic effect of the multiracial category is difficult to adjudicate,
since the two sides are emphasizing different aspects of American racism. We believe that an
official scheme of racial classification should strive to repudiate both the one-drop-of-blood rule
and the appearance of a racial hierarchy. In our judgment, the "mark one or more" option
succeeds best at meeting this challenge.
The Case for "Mark One or More"
Like the addition of the multiracial category, the "mark one or more" approach would free
respondents who regarded themselves as racially mixed from having to choose between their
affiliations (or their parents). However, it would not offer up a new category that might be
construed as intermediate between black and white. It would allow lighter-skinned blacks, or
people with one black parent, to opt out of an exclusive identification as black if they wished, but
it would not give official status to a new, competing affiliation.
In the public debate, that status is in fact what proponents of the multiracial category have called
for: they insist that being multiracial is itself a distinctive identity. "People with a combination of
racial and/or ethnic origins are multiracial/multiethnic people," leaders of PROJECT RACE and
AMEA have written. These groups affirm "the individual's right to be called
multiracial/multiethnic." And they specifically reject the "mark one or more" option by insisting
that any listing of constituent races on the Census form appear "under the `multiracial' umbrella."
Otherwise, they say, the "multiracial population" will be "undercounted, miscounted, or rendered
invisible."
We agree that mixed-race people should be free to identify with every part of their racial
heritage. But the insistence on a separate multiracial identity goes too far. A multiracial category
would lump together people with very disparate identities. One cannot assume that a multiracial
black/white has a great deal in common with a multiracial Asian/American Indian. Even their
experiences of dual identity, divided loyalty, and cultural diversity may be significantly different.
Admittedly, this could change over time: multiracial people might develop more commonalities,
more of a distinctive identity, as their ranks swell. And the introduction of a multiracial category
could even help engender such an identity. But at present, a multiracial category would
misleadingly give a single name to an extremely diverse group of people.
Consider those respondents who see themselves as "black and white" or "black and white, but
socially black"--the locutions used by many people with parents of different races. For them,
there would be some distortion in choosing the multiracial option from a list of exclusive
categories. These people regard themselves as having dual racial identity, not as belonging to a
mixed or hybrid group that includes many other racial mixes. Again, the test survey results
suggest that few people currently identify as "multiracial" to the extent of checking that category
on an exclusive list. This suggests that while the "mark one or more" option would be slightly
more procrustean than the multiracial option for a small number of people--those who now
identify themselves as multiracial--it would better fit what is likely to be a larger number of
people--those who identify with more than one race without identifying themselves as
multiracial.
Equally important, by not reifying a multiracial identity, the "mark one or more" option would
avoid some of the symbolic and social effects of that dubious refinement in racial categorization.
Like the multiracial category, the "mark one or more" option appears to repudiate the one-drop
rule. But it does so, we believe, without any suggestion of a more refined racial hierarchy. It can
be seen, instead, as decomposing racial identity, by implying the legitimacy of composite
descriptions.
Some opponents of any change to Directive 15 worry that a "mark one or more" option, and the
plurality of counts of different racial groups that could result, might confuse and anger the public
even more than the current system's failure to recognize multiple racial identities. They argue that
the American people need and want the federal government to provide one number, one set of
racial "facts."
Clearly, public education would have to accompany the change we are proposing. But it is an
untested assumption that the public (or the courts or whomever the relevant audience is deemed
to be) could not comprehend the complexities of multiple counts for multiple purposes. Social
reality is complex. Racial identity is complex. Each generation of Americans is better educated
than the one before it, and it is not unreasonable to think that the average American's ability to
understand nuance and complexity might also be rising. And if it isn't, then perhaps it is time to
improve comprehension of the complexity of racial identity and classification--its meaning, its
purposes, and its implications.
In addition to supporting the "mark one or more" option for the purpose of data collection, we
believe that federal agencies should have the latitude to tabulate racial data in a way that seems
most appropriate to their purposes in gathering that data in the first place. Federal programs
would benefit if their racial classifications were more closely tailored to the purposes they were
intended to serve, and if agencies had to make explicit the assumptions they used in adopting
particular assignment rules. In classifying people differently for different purposes, agencies
would be disabusing the American public of the recalcitrant notion that there is an overarching
"fact of the matter" about racial identity across the many purposes for which racial data are
collected.
This essay has been adapted from a memorandum that the authors submitted to the Office of Management and Budget on
May 2, 1997, concerning suggested changes to Directive 15. Suzanne Bianchi, professor of sociology at the University of
Maryland, joined three Institute members in this research project, which was funded by the National Science Foundation
and the Alfred P. Sloan Foundation. Sources: "Standards for the Classification of Federal Data on Race and Ethnicity,"
Federal Register, vol. 59, no. 123 (59FR 29831-35), June 9, 1994; "Standards for the Classification of Federal Data on
Race and Ethnicity," Federal Register, vol. 60, no. 166 (60FR 44674-93), August 28, 1995; Spotlight on Heterogeneity:
The Federal Standards for Racial and Ethnic Classification, National Research Council, Committee on National
Statistics (National Academy Press, 1996); joint letters to the Office of Management and Budget from PROJECT RACE
and AMEA (September 29, 1995, and January 6, 1995); "Coalition Statement on Proposed Modification of OMB
Directive No. 15," submitted by Lawyers' Committee for Civil Rights Under Law, National Association for the
Advancement of Colored People, National Urban League, and Joint Center for Political and Economic Studies (Nov. 18,
1994); U.S. Bureau of the Census, "Results of the 1996 Race and Ethnic Targeted Test," Population Division Working
Paper No. 18 (May 1997).
