Race Counts: Racial and Ethnic Data on the U.S. Census and the Implications for Tracking Inequality

History of the Census: Pre–Civil Rights Movement

The foundation and history of the racial and ethnic categories on the census, which we demarcate as pre–Civil Rights Movement, shaped the political and legal status of people, which in turn shaped the U.S. racial hierarchy (Anderson and Fienberg 2000; Hochschild and Powell 2008; Omi and Winant 1986; Prewitt 2013; Snipp 2003). Hochschild and Powell (2008) suggested that the racial categories for the census were motivated by a debate between Congress and the Census Bureau regarding political control, officials’ commitment to scientific and statistical accuracy, and ideological beliefs about race. Beginning in 1850, census development was guided by a formal advisory board and a superintendent, who consulted with professional organizations such as the American Statistical Association and the American Geographical and Statistical Society (Prewitt 2013; Snipp 2003). The Census Bureau changed category names, wording of questions, and/or instructions regarding race every decade since the first census in 1790. Blacks and Whites have been the only two groups represented (in some form) on every census, and only three categories have remained unchanged in name since 1890: White, Chinese, and Japanese (Farley 2002; Saperstein 2012; U.S. Bureau of the Census 1975). A brief look at the history of some of these changes is instructive for understanding the volatility of racial classification and the centrality of Whiteness in the census.

One of the most contested classification issues was in counting people of the African diaspora. The Census Bureau instituted the first census in 1790 with only the categories of “Slaves” and “All other free persons” for those of African descent. The 1850 census added “Mulatto” (of mixed Black ancestry) because some believed that free Blacks were more likely to suffer from insanity than enslaved Blacks—“Mulatto” reflected a clear political move to find “data” to support plantation slavery. There was also a move to document the inferiority of mixed Black-Whites by showing an association with poor fertility and shorter lives (Hochschild and Powell 2008; Humes and Hogan 2009; Prewitt 2013). In 1890, “Quadroon” and “Octoroon” were added as Rep. Wheeler (D-AL) posited that mixed-race persons were of a lower biological quality. However, these categories lasted only one census because drawing these demarcations was deemed dubious, and therefore the data were unreliable. In 1900, there was just one category of “Black” or “Negro,” but the 1910 and 1920 Census Bureau removed “Negro” and added “Mulatto” again as it was argued that mixed Blacks were distinctly different from Blacks. “Mulatto” was permanently dropped by 1930, and “Negro” was the sole term used from 1930 through 1960 (Humes and Hogan 2009; Pew Research Center 2015; Thompson 2012).

Other categories also fluctuated in relationship to political, social, and economic concerns. For example, Native Americans originally were counted only when they lived among Whites and were subject to taxes. Furthermore, Native Americans were not subject to the one-drop rule as were Blacks and instead, were generally classified by the majority race and culture of the community in which they resided (Hochschild and Powell 2008). “Chinese” first appeared on a California census in 1860 when concerns arose about their large immigration numbers. “Chinese” was then put on the 1870 national census and on every census since then; it was the first category that used a nation-based classification rather than a race-based one (Hochschild and Powell 2008; Prewitt 2013; Snipp 2003). “Hindu” was first mentioned in 1880, appeared on some tables in 1910, was added as a formal category from 1920 to 1940, and then was permanently removed in 1950. The Census Bureau recognized “Hindus” as ethnically Caucasian, but they were not from the European region and therefore were seen as non-White Asiatics; this idea was upheld in the U.S. v. Thind Supreme Court case, which banned South Asians from citizenship (Hochschild and Powell 2008; Pew Research Center 2015). “Mexican” was on the census only once, in 1930, and was determined by one’s birth in Mexico or parents’ birth in Mexico and if one was not “Negro,” “Indian,” “Chinese,” or “Japanese.” The addition of “Mexican” was partly influenced by the anti-Mexican sentiment that took place during the early stages of the Great Depression. However, the Mexican government and officials fought this classification (partly due to the assertion that Mexicans should not be considered subject to Jim Crow), and after 1930, Mexicans returned to being classified as “White” on the census (Anderson and Fienberg 2000; Balderrama and Rodriguez 2006; Hochschild and Powell 2008; Pew Research Center 2015).

These examples show how racial or ethnic categories fluctuate and how such fluctuations have been influenced by multiple forces. During this pre–Civil Rights Movement era of the Census Bureau, race records were largely used to control and track different populations; categories were formed in relationship to Whiteness, and White officials determined the race boxes. It is also important to note that from 1880 through 1960, enumerators, trained census officials, collected information from participants and assigned their race. In 1970, the census began using participant self-identification. In this era, the Census Bureau had a near obsession with maintaining the lines among races, specifically with political, economic, and social concerns about safeguarding Whiteness and maintaining the racial hierarchy (Hirschman, Alba, and Farley 2000; Omi and Winant 1986; Prewitt 2013; Snipp 2003).

Contemporary Census Categories: Post–Civil Rights Movement

Since the Civil Rights Movement, the Census Bureau has moved away from recording racial demographics as a tool of control, and advocacy on the behalf of many racial groups has led to a more equity-oriented data collection (Office of Management and Budget [OMB] 1977; Prewitt 2013). Humes and Hogan (2009:119) noted that the U.S. Commission on Civil Rights “stated that racial and ethnic classification can be justified only if the data produced have a legitimate use in terms of combating discrimination, planning programs, or conducting program evaluation.” Census data continue to be used to apportion seats in the House of Representatives, draw legislative districts of equal population, and plan construction of hospitals, highways, bridges, and schools. The racial categories are also used as a sample frame for other federal surveys, such as the Current Population Survey and the National Crime Victims Survey (Nguyen 2007:228).

In 1974, the Office of Management and Budget (more commonly known as the OMB) determined that there should be a clear and discernable taxonomy for the racial categories. In 1977, the Census Bureau adopted Directive 15, which identified the five basic groups: American Indians and Alaska Natives; Asians and Pacific Islanders; Non-Hispanic Blacks; Non-Hispanic Whites, and Hispanics of any race (Hirschman et al. 2000; Prewitt 2013; Snipp 2003). However, in this same directive it was made clear that “these classifications should not be interpreted as scientific or anthropological in nature” in order to move away from racist biological notions of race (OMB 1977).

Thus, the 1980 census was the first census to have the five groups, often recognized as the “ethnoracial pentagon” (Hollinger 1995) of White, Black, Indigenous, Asian, and Hispanic. In 2000, the Hispanic question was changed to include “Spanish/Hispanic/Latino,” and the order in 2010 was changed to “Hispanic/Latino/Spanish.” Within these five categories, some offer specific racial or ethnic categories to choose from: Hispanic has three options; American Indian or Alaska Native asks participants to write in the tribe; Asian has six options; and Pacific Islander has three options. There is also the option to write in for “another Hispanic, Latino, or Spanish origin,” “Other Asian,” “Other Pacific Islander,” and/or “Some other race.” The Census Bureau added the option to “mark one or more” races in 2000. The option to identify with multiple races was a challenge to the strict “mark one box” rule that had been in place since the inception of the census. Mixed-racial heritage had been previously recognized, but by 1930, the “Mulatto” category had been dropped, and the one-drop rule, whereby one drop of Black blood made one Black, overruled mixed identity. The “mark one or more” option was largely a response to mixed-race individuals and their families advocating for a multiracial identity option on the census. Although the objective of getting “multiracial” was unsuccessful, it still brought forth a significant change in the racial classification scheme: The Census Bureau now reports the number of people who identify with 2, 3, 4, 5, and 6 races. Hence, the census now has 63 possible racial categories and 126 with the Hispanic ethnicity option (Farley 2002; Strmic-Pawl 2016; Williams 2008).

Although under different pressures than in earlier years, the Census Bureau still must satisfy multiple interests: the recommendations of the Office of Management and Budget, participants’ needs for the census to be clear and accessible, and appeals from both Congress and advocacy groups (Farley 2002; Prewitt 2013). In addition, Ngyuen (2007) outlined four specific types of accountabilities that the Census Bureau faces: hierarchical in response to the president, commerce department, and the undersecretary; professional in response to contemporary theories and methods of data collection; political in response to Congress; and legal in response to court decisions (continuous lawsuits are brought against the Census Bureau about who to count and how to count them). At the nexus of these accountabilities, changes continue to be made in the categories used, the order of the questions, and the phrasing and instructions of the questions. To address such issues, the Census Bureau, since the 1970s, has used “content tests to research and improve the design and function of different questions, including questions on race and ethnicity” (U.S. Census Bureau 2017). We now turn to discuss possible race and ethnicity changes in the next census.

Major Changes in the 2020 Census? Hispanics, Middle Easterners, and North Africans

Of particular concern to the Census Bureau is how to account for the problem of “undercounting,” or when a population has a low response rate. Looking at past census results, two groups—the Hispanic population and the Middle Eastern and North African population, the latter collectively referred to as MENA—have been consistently undercounted. Yet the reasons differ; for Hispanics, confusion sometimes arises about what race to choose since “Hispanic” is listed separately as an ethnicity question, while the MENA population is currently grouped with Whites on the census, but often do not identify as such. The Census Bureau has been conducting research to determine possible changes for these groups in the 2020 census.

Issues with Collecting Racial and Ethnic Demographics

Census data on race and ethnicity can be uncritically accepted and then used in the creation of new data, reports, and peer-reviewed publications. Sometimes the justification of the “social construction of race” is used to acknowledge on some level the scientific fallacy that is the racial classification system. Likely, most social scientists now know and work on the premise that race is a social construction, but nevertheless proceed with survey data on race because data are needed in order to address social inequality. “Race is not real but racism is”—so goes the saying that rationalizes the need for race data. The American Sociological Association (2003) stated as such in its statement on race:

Sociological scholarship on “race” provides scientific evidence in the current scientific and civic debate over the social consequences of the existing categorizations and perceptions of race; allows scholars to document how race shapes social ranking, access to resources, and life experiences; and advances understanding of this important dimension of social life, which in turn advances social justice. (P. 4).

Although we agree with this premise, we outline in this section the numerous factors that shape the race data with which we work. The race counts reported by the Census Bureau vary in relationship to (1) whether the word “race” is used, (2) the order and phrasing of the question, (3) self-identification versus outsiders’ assignment, and (4) the categories provided.

European Comparisons: Should the United States Continue to Collect “Race” Data?

If we adopt a comparative frame, it is clear that the United States is in the minority in having racial and ethnic categories on its census. In Europe, for example, the majority of countries do not use race or ethnicity, while some use only ethnicity. Only the anomalous U.K. and Ireland have ethnic and racial categories on their census. However, many Eastern and Central European countries were once (or even twice) part of empires (e.g., Russian, Habsburg, Swedish) and count national minorities mainly by language use, reflecting this multinational composition of their population. However, the largest single cluster of European countries uses country of birth/nationality rather than race or ethnicity. Moreover, the vast majority of all nation states in the world actually use the “equality data method” (i.e., asking the citizens how they identify themselves), with the exception of Northern and Western Europe. The question is therefore whether using categories such as race and ethnicity actually has a progressive effect on patterns of discrimination. Critics of using these categories maintain that they are ultimately unhelpful and divisive.

Indeed, the collection of racial and ethnic data always supposes not only a conversation on the methods and saliency of the categories deployed, but also an appropriate end use for the data. It has already been noted that some groups’ experiences of control and surveillance make them wary of engaging with state-led counting exercises and the use of race categories. So what is offered internationally in terms of rationales? The United Nations provides a clear statement on the issue:

The decision to collect and disseminate information on ethnic or national groups of a population in a census is dependent upon a number of considerations and national circumstances, including, for example, the national needs for such data, and the suitability and sensitivity of asking ethnicity questions in a country’s census . . . . It is important that the responding public be informed of the potential uses and need for data pertaining to ethnicity, as this improves public support for the census exercise. Data on ethnicity provide information on the diversity of a population and can serve to identify subgroups of a population. Some areas of study that rely on such data include demographic trends, employment practices and opportunities, income distributions, educational levels, migration patterns and trends, family composition and structure, social support networks, and health conditions of a population. (2006:2, 160)

Moreover, for the European Union, such data collection is driven by the need to address the following:

discrimination based on racial and ethnic origin in the fields of employment, social protection, including social security and healthcare; social advantages, education and access to and supply of goods available to the public including housing, action against which is framed under three principal agreements; the Racial Equality Directive (RED), 2000; Article 21 of the Charter of Fundamental Rights; and the Council Framework Decision, 2008. (Farkas 2017:4)

The function performed by collecting racial and ethnic data is thus to enable the enforcement of antidiscrimination work undergirded by legislation. Moreover, the U.K.’s Office of National Statistics (ONS 2003) underscores this conclusion stating that the 2000 Race Relations (Amendment) Act gives public authorities a duty to combat racial discrimination. The E.U. report cited here also notes that the ONS makes the guidelines for data collection public, as advocated by the U.N. Moreover, the use of racial and ethnic data is argued to have a significant role in tailoring services and, for example, enabling healthcare and health outcome disparities to be addressed (Fremont and Lurie 2004).

This point, of course, does not overcome either particular groups’ reluctance to provide personal sensitive details for historic reasons or individuals disagreeing with the principle of collecting racial and ethnic data in the first place, as evidenced in ongoing colorblind critiques in North American and European nations. France (Simon 2012) and Sweden (Hübinette and Lundström 2014) are among the prominent European nations to refuse to collect racial data, suggesting that to do so would infringe ideas of belonging to the nation as a set of individuals. This stems from a republican notion of the nation in the case of France and avowed commitment to antiracism in the case of Sweden. However, countries with no such census categories still have a raft of ways to talk about who belongs to the nation by bloodlines and culture. The Netherlands statistics office, for example, in 2016 officially stopped using the terms “autochtonen” and “allochtonen” (Essed and Trienekens 2008), loosely translated as “indigenous” and “nonindigenous,” respectively, only to replace them with “of immigrant background” and “of Dutch background.”

Racial categories, it is argued, reify divisions and therefore the State should help extinguish them. However, in terms of combatting discrimination, there is no neat distinction between nations using racial and ethnic categories in a census and those not using them. Minorities in the nations without these categories still experience racism, and popular ways of conceptualizing the nation are still racialized. The chief argument in favor of collecting race data is to enable observation of discrimination patterns. The debate about whether progress occurs cannot take place without the statistical material. The exercise of collecting the data constitutes a symbolic recognition of social scientists’ arguments that despite their lack of biological basis, “race” and “ethnicity” have social lives and racism has real social consequences. It should be noted, however, that identifying the patterns and remedying them are separate functions. For the data collection to be justified, the monitoring function supposes that state agencies implementing antidiscrimination laws have legitimacy, effective means, and authority to hold departments, companies, and other agencies to account and thus to enforce required changes.

Conclusion: Race Counts

We have examined the ways that the U.S. Census came to be, its roots in racism and control, its evolution toward multiculturalism and group rights, and its struggle with creating racial categories and collecting data. With this examination in mind, we consider five main propositions about race and ethnicity on the census. We are not taking a particular stance on any one specific racial or ethnic category; instead, the following points address the broad principles underlying racial and ethnic data collection on the census: (1) we need to be open to categories consistently changing; (2) the addition of racial groups should be linked to tracking inequality and not necessarily cultural identity; (3) “race” should be the term used; (4) we need more consideration for the addition of a second racial identity question that asks about one’s race as assigned by society; and (5) sociologists need to be more active in the census development process.

U.S. Census racial categories have consistently changed since the inception of the census in the 1700s. As social, political, and economic concerns shifted, the number of racial groups and the terms used to describe groups changed. With the option to mark more than one race, we now have 63 possible racial categories. Rather than criticizing the numerical and qualitative changes in the racial and ethnic categories as problematic, we assert that such change is necessary. Racial representation on the census permits data collection that can help in proving a number of important social inequalities: income gaps, underemployment, residential segregation, low political representation, and more. Therefore, racial representation is much more than identity politics or cultural acknowledgment; it is an important pathway toward tracking discrimination in order to fight it. In this vein, as society changes, so will the racial categories on the census, and these changes should be welcomed, not contested.

However, as we see the census race question primarily as a tool for tracking inequality, we ask whether the addition of categories should be rooted less in the desire for cultural recognition and more on whether the racial or ethnic category is needed to track inequality. In particular, we need to be open to groups advocating for themselves, not necessarily because they have a low public profile but because they are experiencing discrimination. Collecting and categorizing groups does not always serve progressive ends, in that being “culturally” different does not equate 100 percent of the time to discrimination.

It is important to consider at what point, if any, some categories could be removed from the census; simply because a racial group was added does not mean that it will perpetually be necessary. Removal from the census does not mean a group should be disregarded, but we hypothesize a time might arrive when some categories are more necessary than others based on social inequality. The addition of new categories is likely inevitable, so the question must be asked whether at some point categories should be removed to maintain the list at a manageable number. We suggest that a “manageable number” should be evaluated by the ability for categories to be meaningful enough that they allow the tracking of inequality; that is, with too few categories the important differences among groups are lost, but too many categories may cause a dilution effect in that the more detailed the disaggregation, the less clear aggregate forces appear.

For the foreseeable future, “race” should be the term used on the census. We acknowledge the many conceptual difficulties with “race” as well as the reasons that other terms such as “origin,” “ancestry,” or “ethnicity” are advocated (Prewitt 2013). Yet, “race” has a particular history in the United States, and this history has cultivated and shaped people’s understanding of their identities on the micro level and the racial hierarchy on the macro level. Terms like “ancestry,” which could be exchanged for “race,” do not have the same definition or application in people’s lives or in society. Thus, changing the term could possibly lead to distorted numbers and a lack of efficient tracking over time as identities on the census do not match the social reality. Sociologists need to be unwavering in explaining the social construction of race and in debunking contemporary biological reifications of race while recognizing that replacing the term “race” does not get rid of racism and could potentially lessen our ability to address it.

In fact, we could possibly be doing more to track racism with the racial identity question. As social scientists, we know that racial self-identification is just one form of racial identity in society, and research indicates that one’s racial self-identification does not always match the racial identity that society would assign. The social assignment of racial identity is more likely to be the one that correlates with racial discrimination, and therefore this identity might be equally or more helpful in tracking stratification. For example, if Person A self-identifies as White but is viewed in society as a person of color, then with a self-identification question, the census is perhaps not capturing that experience accurately. We are not the first scholars who have brought attention to this issue, and some have suggested more than one question on race (e.g., see Burton et al. 2010; Campbell, Bratter, and Roth 2016; Denton 1997; Saperstein 2012); we call for more conversations on the feasibility of adding a question on the census that reflects racial identity based on others’ perceptions.

Taking the above into account, our final suggestion is a call for sociologists to be more involved in the consultation and decision-making process of the racial categories on the census. The census categories dictate not only the racial data we receive via this outlet, but also the data from a number of other federal and state surveys. The racial categories that appear on the census, therefore, both enable and constrain our analyses, conclusions, and recommendations.

The Census Bureau has a number of advisory committees that “provide a venue and forum for the Census Bureau Director to interact with and solicit input from a variety of external stakeholders to improve overall Census Bureau programs, surveys, data collection, quality and decennial census.” In particular the National Advisory Committee provides feedback on race and ethnicity; based on member biographies, it currently seats four sociologists in academia (Julie Dowling, Kathleen Mullan Harris, Stuart Michaels, and Ann Morning) (https://www.census.gov/about/cac/nac/member-bios.html). All National Advisory Committee meetings are open to the public and take questions. If one cannot be there in person, a link is provided to join the meeting online. Updates on meetings and agendas are posted at https://www.census.gov/about/cac.html. As well, a number of centers work through colleges, universities, and organizations that communicate with the Census Bureau via the Census Information Centers Member Network; these centers and organizations “are recognized as official sources of demographic, economic, and social statistics produced by the U.S. Census Bureau. CICs provide training and technical assistance to local governments, businesses, community groups, and other interested data users in accessing and using Census Bureau data for research, program administration, planning, and decision-making purposes” (https://www.census.gov/about/partners/cic/network.html). On a smaller level, a Respondent Advocate center takes concerns about the survey and its administration; the center can be reached at respondent-advocate@census.gov.

Changes to the U.S. Census not only have helped social scientists measure the racial and ethnic demographics in the country but have been useful in tracking disparities. Although neither simple solutions nor a perfect instrument is available to collecting racial data, as social scientists we must continue to question what the census looks like and in what ways it can better help us capture issues of inequality. As noted, the census and the conversations that surround it have real world consequences. These conversations help to highlight the many ways that race counts.