Have seen earlier discussion of the issue but this is the most comprehensive analysis:

Everyone hoping for an accurate 2020 Census breathed a sigh of relief two weeks ago when the Supreme Court ruled 5-4 to block the Trump administration’s cynical attempt to add a citizenship question to the forms—only to experience Twitter-tantrum whiplash when the president ordered his administration to make a last-ditch attempt to include it.

But with so much attention focused on the controversy over the citizenship question, another similarly disastrous Census Bureau decision has gone largely unnoticed: the administration’s choice not to substantively update the decennial survey’s questions on race. As a result, no matter how conscientiously Census Bureau staff administer the survey, a woefully inadequate portrait of the changing face of America will emerge.

The last census, in 2010, became a data disaster when “some other race,” showed up as the third-largest racial group in America. Over 20 million respondents, most with roots in Latin America or the Middle East, selected this none-of-the-above option, making it the most popular choice after white and black. Any time a public-opinion survey asks respondents to self-categorize and “none of the above” comes back as a popular answer, it’s a clear sign that the choices given don’t match up with people’s identities.

Facing this problem squarely, the Obama administration convened the National Advisory Committee on Racial, Ethnic and Other Populations, a panel of academic experts and minority community leaders, to advise the Bureau on improving its race questions for 2020. The committee made myriad recommendations, most crucially suggesting that a “Middle East or North Africa” category sit alongside the “Hispanic origins” box in the upcoming questionnaire. But the Trump administration overruled this advice and, aside from a few minor tweaks, is flying into the 2020 survey without substantive changes. Given continued Latin American and Middle Eastern immigration since 2010, and the more extreme forms of racial “othering” these groups have faced ever since candidate Donald J. Trump came down the escalator in 2015, experts fear that “some other race” will become the second-largest racial group in America according to the 2020 Census.

Every census since the founding of the country has asked about race and ethnicity. Until recent decades, race was not a matter of self-identification; historically, federal census-takers were charged with determining the race of each resident of their assigned census tracts according to their era’s standards. Tracing how race questions have changed over time offers a time-lapse history of American racial concepts in 10-year snapshots. (All of the race questions are conveniently archived on the website racebox.org.)

The most drastic changes to the census race questions took place after the fall of Reconstruction, at the rise of Jim Crow, when America’s mixed-race realities were blotted out and a strict racial binary imposed. Openly mixed-race activists, in particular Charleston’s “Browns” and New Orleans’s “Creoles of color,” had been central to post-Civil War civil rights progress. Their court challenges to segregation, of which Plessy v. Ferguson (1896) was only the last and most famous, assailed the notion that Europeans and Africans remained distinct racial groups in America given centuries of overt and covert race-mixing. At the time, the “one-drop rule” that any African ancestry at all made an American a “Negro” was still new and not widely accepted. This more fluid racial mindset was reflected in the late 19th-century censuses, which all catalogued biracial “mulattos” as distinct from “whites” and “blacks.” The 1890 questionnaire recorded even finer-grained mixed-race categories: “quadroon” (an American with three European grandparents and one African grandparent) and “octoroon” (an American with seven European great-grandparents and one African great-grandparent). But with the firm establishment of the color line post-Plessy, the 1900 census switched to a unitary race. (Not until 2000 would the census again allow respondents to claim mixed-race identities, this time by checking more than one racial box.)

“Only in 1980 did the Census begin to grapple with Latino identity.”

As segregation took root, the stakes of being deemed “white” grew higher. Even as Jim Crow laws proliferated in the early 20th century, the states differed on their official definitions of what exactly a “white person” was and who precisely constituted a “colored person.” Myriad ethnic groups clamored to get into whiteness, often petitioning through the courts. “Semites,” for example, won their way into whiteness using clever, albeit pseudo-scientific, arguments. Their trump card, first argued in 1907 by H. A. Elkourie, a Syrian Christian physician in Birmingham, Alabama, was that if he wasn’t white then Jesus hadn’t been white either. Anglo-Americans’ revulsion at the thought they were worshipping a person of color each Sunday was strong enough that Elkourie and the fellow members of his “Semitic” “race” were deemed “white.”

The next major revamp of the census’s race questions came in the wake of the 1960s civil rights movement. For the first time, the Census Bureau empowered each respondent to choose her own race rather than have a census-taker determine it for her. And embracing the modern understanding that race has no biological reality, only societal meanings, the Census Bureau modified the racial categories to learn more about American society rather than engage in the fool’s errand of sorting humans into some fixed number of distinct races. To this end, the 1970 Census listed eight racial categories, one of which was “Hawaiian”—a useful category for understanding American society but a group so tiny no early-20th-century race scientists ever elevated it into their core “Races of Man.”

Only in 1980 did the census begin to grapple with Latino identity. Rather than add “Hispanic” to the list of races, it introduced a question to stand apart from the various racial choices: “Is this person of Spanish/Hispanic origin or descent?” By noting that Hispanics can be of any race, the Census Bureau hoped to track the growth of this community that comes in all colors. But this well-meaning attempt never fully worked since the Latin American and Anglo-American conceptions of race are fundamentally incompatible.

While the U.S., after Reconstruction, forced Americans to claim a retroactive racial purity, Latin America never denied its mestizo realities. On the most recent Brazilian census, for example, the majority of respondents identified as afrodecendente (Afro-descended). But in Brazil this identity does not in any way suggest that the same person is not also of European, Native American, and/or Asian descent; indeed, over 80 percent of self-identified afrodecendente Brazilians claimed roots on non-African continents as well.

In Mexico, the concept of race (la raza) is even more un-American. The Mexican supposition is that the people of the New World are, in a sense, a new race unto themselves, a mixture of all the world’s peoples. It is these mutually-incompatible conceptions of race between the U.S. and Latin America that has led millions of census respondents to check that they are of Latino origin but are members of “some other race.”

Arab-Americans are similarly migrants from an alternate racial system. Arab identity embraces people of all skin colors and is largely tied to language—people whose mother tongue is Arabic are Arabs even if they don’t live on Asia’s Arabian Peninsula. Though officially white in America since the early-20th-century rulings that “Semites” are white, contemporary American racism has again called Arab whiteness into question.

“The best-case scenario is that none-of-the-above comes out as the third-largest race in America rather than second-largest.”

The most recent federal definition of a “white person,” formulated in 1997 by the Office of Management and Budget and currently used by the Census Bureau—“A person having origins in any of the original peoples of Europe, the Middle East, or North Africa”—clearly includes Arabs. But if “whiteness” has no biological reality and is purely a socially-constructed category in American society for those who enjoy full citizenship, including the presumption of innocence, since 9/11, Middle Easterners have no longer been white. This mismatch between being officially white by the federal definition but not being treated as white in American society has sparked a wildcat campaign among some Middle Easterners not to check the “white” box on the Census (tag-line: “Check it right, you ain’t white”). Indeed, the National Advisory Committee on Racial, Ethnic and Other Populations found many Middle Eastern- and North African-Americans are doing just that, checking “some other race” in defiance of the current federal definition of whiteness.

At this point the 2020 race questions are set, with just a few tweaks from 2010. The 2020 form will include “Lebanese” and “Egyptian” as examples of white ethnicities to remind Arabs to, essentially, “check it right, you are white.” The new wording also adds “Aztec” and “Mayan” as examples of American Indian ethnicities to instruct people with roots in the New World beyond the United States borders that they should still identify themselves as indigenous.

Even with these minor changes, the best-case scenario is that none-of-the-above comes out as the third-largest race in America rather than second-largest. Whiteness in America is in flux today in a way it hasn’t been in a century—even if the Census Bureau’s political appointees, in keeping with the Trump administration’s Know-Nothingism on race, won’t admit it. An administration that has backed border walls and Muslim bans has already shown Latinos and Middle Easterners that, if whiteness means first-class citizenship, they’re no longer white. The painful irony is that the rise of “some other race” at first glance suggests America is becoming post-racial, while its real roots are in rising racism.