guest post: new census race classifications are about civil rights, not discrimination

The following is a post by Emily Klancher Merchant.

In a recent Washington Post Op-Ed, Mike Gonzalez, a senior fellow at the Heritage Foundation, warns that two changes to the 2020 census, recently proposed by the Office of Management and Budget (OMB), will “aggravate the volatile social frictions that created today’s poisonous political climate in the first place”. One change would combine the currently-separate questions on race and Hispanic origin. The second would add “Middle East or North Africa” as a potential answer to that combined question. Gonzalez disingenuously claims that these changes – along with existing efforts by the Census Bureau and other statistical agencies to classify the U.S. population by race, ethnicity, or national origins – will undermine the unity of the American people. Certainly, in its nearly 230-year history, the U.S. Census has participated in racist projects from slavery to Jim Crow to the 1924 National Origins Act to the internment of Japanese Americans during World War II. Today, however, the census is the government’s primary statistical tool for the enforcement of civil rights legislation and the administration of programs aimed at redressing our country’s long history of race-based discrimination, oppression, and plunder. The proposed changes are an attempt by the demographers of the Census Bureau to make our statistical system better reflect the complex categories of identity in the United States that continue to structure inclusion and exclusion, oppression and privilege. Efforts to block the reporting of race, ethnicity, or national origins in the census aim to undermine scientific analysis of discrimination and policy measures that promote equity. Understanding the changes proposed by OMB and Gonzalez’s opposition requires understanding the history of race in the census both before and after the passage of the 1964 Civil Rights Act.


Census race categories: Before 1964

Through most of the antebellum period, the census placed individuals into three broad categories: “free white persons,” “free colored persons,” and “slaves.” This classification scheme reflected the system of racial slavery and racial inequality among free people. From the end of the Civil War until about 1940, census race categories were in constant flux. The designations adopted and eliminated during this period reflected the demise of one system of racial oppression (slavery), the rise of a new one (Jim Crow), large-scale immigration to the United States, and efforts to restrict immigration. The census experimented with categories for mixed-race individuals (“mulatto” appeared on nearly every census from 1850 to 1920; “quadroon” and “octoroon” appeared in 1890 and were immediately withdrawn) and introduced new identifiers as the sources of immigration changed: “Chinese” in 1870; “Japanese” in 1890; “Native Hawaiian” in 1910; “Filipino,” “Hindu,” and “Korean” in 1920; and “Mexican” in 1930. The appearance of these categories tracked the rights afforded their members in the United States. Any individual belonging to a group classified as other than white could be subject to segregation and denied naturalization.

The Naturalization Act of 1790 limited naturalization to “free white” people, and the Naturalization Act of 1870 extended it to those “of African nativity” or “African descent”, but neither act offered a definition of “white.” Between 1878 and 1944, 52 federal court cases examined the “whiteness” and eligibility for naturalization of individuals from countries around the world (Haney Lopez 2006). The courts found that immigrants from China, Japan, Hawaii, and the Philippines were not white, but that Armenians were. They went back and forth on the whiteness of Syrians and South Asians, ultimately deciding that Syrians were white and that South Asians were not. These decisions referenced race science when it favored judges’ opinions (for example, deciding that Chinese were not eligible to naturalize because race scientists did not classify them as “Caucasian”), and common knowledge when race science did not favor judges’ opinions (for example, deciding that South Asians were not eligible to naturalize even though race scientists did classify them as “Caucasian”). Census race categories mirrored these decisions, providing further weight to both scientific and commonsense arguments. The “Hindu” category appeared shortly before the 1923 case United States v. Bhagat Singh Thind, which declared South Asians ineligible to naturalize (Hochschild and Powell 2008). Groups declared eligible, such as Syrian and Armenian immigrants, were never racialized in the census.

These decisions had material consequences. For example, California barred non-citizens from owning land. Armenian immigrants were able to naturalize, purchase land, and establish lucrative agricultural enterprises, while immigrants from Japan were not. The 1924 Immigration Act sharply limited immigration from Southern and Eastern Europe, but prohibited altogether immigration by people who were considered “racially ineligible” for citizenship (Ngai 1999). The addition of a “Mexican” race category in the 1930 Census accompanied efforts by some members of Congress to limit immigration from Mexico, which was then not subject to quotas (Gratton and Merchant 2016). It was swiftly followed by local efforts in many municipalities to encourage or coerce the repatriation of Mexican immigrants and their American-born children, resulting in the departure of approximately 350,000 Mexican-Americans between the 1930 and the 1940 censuses (Gratton and Merchant 2013). Mexican civil rights groups and the Mexican government protested the designation of Mexicans as other than white and the discriminatory practices that such classification justified and promoted. Enumerators for the 1940 census were instructed that “Mexicans are to be regarded as white unless definitely of Indian or other nonwhite race”. In 1940, the classification of an immigrant group, whether Mexican or Middle Eastern, as white was a major civil rights victory that protected its members from the most egregious forms of official discrimination.

Census race categories: After 1964

Until the mid-1960s, racial categories in the census and other official statistical instruments reflected, tracked, and facilitated discrimination and oppression at both casual and structural levels. This changed dramatically after the passage of the 1964 Civil Rights Act. The federal government began to use data collected by the census and other federal agencies as the statistical basis for the enforcement of civil rights legislation and the operation of other government programs aimed at redressing past and ongoing discrimination (Prewitt 2013). Initially, each federal agency used its own categories and terms to identify protected categories. This multiplicity of classification systems reflected the contingent and constructed nature of race, but created statistical chaos. Accurately measuring discrimination meant that the categories in the denominator (the census) needed to match those in the numerator (data collected by other federal agencies). OMB solved this problem in 1977 with the adoption of Statistical Policy Directive Number 15. OMB 15 standardized the racial and ethnic categories that have classified Americans for federal statistical purposes ever since. Its categories have seeped into state and local governments and private organizations, particularly those seeking government contracts.

OMB 15 established four racial categories that were intended to be exhaustive and mutually exclusive – black, white, American Indian or Alaska Native, and Asian or Pacific Islander – and one ethnic category – Hispanic – whose members could belong to any racial group. The directive defined these categories in a variety of ways: a person was black” if he or she had “origins in any of the black racial groups of Africa,” and was “American Indian or Alaska Native” if he or she “maintains cultural identification through tribal affiliation or community recognition.” The other categories, including “Hispanic,” referenced geography, providing no explanation as to why some origins were racial designations and others ethnic. A 1997 update split the “Asian or Pacific Islander” category into two and directed that respondents be allowed to choose more than one racial identity. The designation of “Hispanic” as an ethnicity rather than a race, in both 1977 and 1997, seemingly reconciled two competing mandates: Public Law 94-311, passed in 1976, which required the government to collect and publish “economic and social statistics for Americans of Spanish origin or descent” (Hattam 2005, p.65), and the federal government’s acquiescence to demands in the 1930s by Mexican American leaders and the Mexican government that those of Mexican descent be classified as white. OMB 15 explicitly stated that its “classifications should not be interpreted as being scientific or anthropological in nature”; they were simply meant to facilitate the collection of data necessary to administer federal programs and monitor compliance with civil rights legislation. Yet the very ubiquity of these categories has resulted in their naturalization, and the distinction drawn by the federal government between “race” and “ethnicity” lends its authority to the stabilization of these concepts individually and relationally.

Currently, OMB 15 requires that data on race and ethnicity be collected separately. This means that each census since 1980 has included two questions, one asking whether a person is “of Hispanic, Latino, or Spanish origin” and the other asking “what is this person’s race?” For decades, this question has confused respondents and driven Census Bureau officials to seek better ways of communicating its request for information. The 2010 census instructed respondents to “answer BOTH Question 5 about Hispanic origin and Question 6 about race,” specifying that “for this census, Hispanic origins are not races”. Perhaps adding to the confusion is the fact that the race question includes many more boxes than the five categories defined in the 1997 update to OMB 15. In 2010, each respondent could select one or more of the following options: “White,” “Black, African Am. or Negro,” “American Indian or Alaska Native,” “Asian Indian,” “Japanese,” “Korean,” “Filipino,” “Vietnamese,” “Other Asian,” “Native Hawaiian,” “Guamanian or Chamorro,” “Samoan,” “Other Pacific Islander,” and “Some other race.” Those who chose “American Indian or Alaska Native,” “Other Asian,” “Other Pacific Islander,” or “Some other race” were instructed to “print race” below their answer. With the exception of “Some other race,” the categories provided in the census can be aggregated to produce numbers for each of the five categories specified in OMB 15.

The number of people who check “Some other race” is a measure of the degree to which the census race categories fail to reflect the identity categories held by respondents and used in their everyday lives. Census data do not transparently illustrate reality, nor do they simply reflect a set of categories imposed by the state on its populace. Rather, they reflect the interactive process by which respondents reconcile the categories provided on the form with those that structure their lived experience (Emigh et al. 2016). According to OMB 15, there is no “other race”; everyone is either white, black, Native American, Alaska Native, Asian, or Pacific Islander. Yet, in the 2010 Census, approximately 21.7 million respondents, representing 7% of the population, checked the “Some other race box,” either alone or in combination. 41% of those identifying as Hispanic checked “Some other race” and 95% of those who checked “Some other race” identified as Hispanic, suggesting that the official distinction between race and ethnicity does not resonate with those it is meant to describe (analysis by author using 10% sample from Ruggles et al. 2015).

Gonzalez describes the changes proposed for Census 2020 as being rushed through by the Obama administration in its last hours, with OMB surreptitiously “slipp[ing] notice of the proposed rule under the door just one day after Congress went on recess in September”. In reality, as sociologist Philip N. Cohen points out in his Family Inequality blog, the recommendations are based on decades of research by demographers and survey methodologists. In its most recent tests, the Census Bureau found that “most Hispanics struggle to answer the separate question on ‘race,'” with many skipping it altogether, “repeat[ing] what they reported for the Hispanic origin question,” or reporting a category that “does not accurately reflect how they self-identify”. In national tests, simply combining the currently-separate race and ethnicity questions reduced the proportion of respondents answering “some other race” from about 10% to nearly zero. Gonzalez incorrectly and misleadingly argues that this change would “eliminate Hispanics’ racial choices.” However, since 2000, the census has allowed respondents to choose more than one race, which means that, with a combined question, respondents could check “Hispanic” and any other category with which they identify. In fact, the change increases choices because it would allow a person to identify only as Hispanic or Latinx, rather than forcing them to choose a separate racial category with which they do not necessarily identify. The proposed “Middle East and North Africa” category originated in requests by people from that region for “a distinctive category to capture forms of discrimination exemplified by the hate crimes and profiling that have occurred as a result of the ‘War on Terror’ and continuing political instability in the Middle East” (Omi and Winant 2015, 123), not a nefarious plot by the Obama administration to reduce the number of people classified as white, as Gonzalez would have us believe. Gonzalez argues that the proposed changes are “silly” because “these ethnic ‘groups’ are not really ethnic at all,” as if “ethnic” had a stable referent. Yet the history of race, ethnicity, and the census in the United States demonstrates that there is nothing natural about these categories, which have shifted continually along with the politics of exclusion and discrimination.

Centuries of discrimination, oppression, and plunder have given material form to categories of race, ethnicity, and national origin, and for many Americans these categories structure nearly every aspect of lived experience. Eliminating these categories from our national statistical systems won’t eliminate difference, nor will it eliminate discrimination or oppression. It will simply eliminate the best tool we currently have for identifying and redressing discrimination and oppression. It is certainly possible that data collected by the census could be used for nefarious purposes in the future, as they have been used in the past. The solution is not to stop collecting data, but to hold our government accountable for how it uses the data it collects.

Emily Klancher Merchant is a postdoctoral fellow in the Neukom Institute for Computational Science at Dartmouth College and Assistant Professor of Science and Technology Studies at the University of California, Davis.

Works Cited:

Emigh, Rebecca Jean, Dylan Riley, and Patricia Ahmed. 2016. Changes in Censuses from Imperialist to Welfare States: How Societies and States Count. Pagrave Macmillan.

Gratton, Brian and Emily Klancher Merchant. 2016. “La Raza: Mexicans in the United States Census.” Journal of Policy History 28 (4): 537-567.

_____. 2013. “Immigration, repatration, and deportation: The Mexican-origin population in the United States, 1920-1950.” International Migration Review 47 (4): 944-975.

Haney Lopez, Ian. 2006. White by Law: The Legal Construction of Race New York: NYU Press.

Hattam, Victoria. 2005. “Ethnicity and the boundaries of race: Rereading Directive 15.” Daedalus 134 (1): 61-69.

Hochschild, Jennifer L. and Brenna Marea Powell. 2008. “Racial reorganization and the United States Census 1850-1930: Mulattoes, half-breeds, mixed parentage, Hindoos, and the Mexican race.” Studies in American Political Development 22 (1): 59-96.

Ngai, Mae. 1999. “The architecture of race in American immigration law: A reexamination of the Immigration Act of 1924.” The Journal of American History 86 (1): 67-92.

Omi, Michael and Howard Winant. 2015. Racial Formation in the United States. New York: Routledge.

Prewitt, Kenneth. 2013. What is Your Race?: The Census and Our Flawed Efforts to Classify Americans. Princeton: Princeton University Press.

Ruggles, Steven, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek. 2015. Integrated Public Use Microdata Series: Version 6.0 [Machine-readable database]. Minneapolis: University of Minnesota.

Author: Dan Hirschman

I am a sociologist interested in the use of numbers in organizations, markets, and policy. For more info, see here.

One thought on “guest post: new census race classifications are about civil rights, not discrimination”

  1. Good post. Good on the history. Recapitulates what most people in the “race” area know, but even PhD sociologists outside the race area don’t know, and is certainly not widely known. Buried in the middle of the history is the decision by the FBI to stop tabulating “Hispanic” arrests separately and to lump them with other races (usually White), a directive not changed until 2013. The result is both the invisibility of discriminatory arrest patterns for Hispanics and an under-estimation of the Black/White disparity in arrests.


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: