Fix the Census’ Archaic Racial Categories

By Kenneth Prewitt, NY Times

STARTING in 1790, and every 10 years since, the census has sorted the American population into distinct racial groups. Remarkably, a discredited relic of 18th-century science, the “five races of mankind,” lives on in the 21st century. Today, the census calls these five races white; black; American Indian or Alaska Native; Asian; and Native Hawaiian or other Pacific Islander.

The nation’s founders put a hierarchical racial classification to political use: its premise of white supremacy justified, among other things, enslaving Africans, violent removal of Native Americans from their land, the colonization of Caribbean and Pacific islands, Jim Crow subjugation and the importation of cheap labor from China and Mexico.

Of course, officially sanctioned discrimination was finally outlawed by civil rights legislation in 1964. The underlying demographic categories, however, were kept. Securing civil rights required statistics. Thus resulted an uneasy marriage of preposterous 18th-century racial classifications to legitimate 20th-century policy goals like fair electoral representation, anti-discrimination programs, school desegregation, bilingual education and affirmative action.

But the demographic revolution since the immigration overhaul of 1965 has pushed the outdated (and politically constructed) notion of race to the breaking point. In June the Supreme Court struck down a core provision of the Voting Rights Act, taking note of changing demographics. I disagree with the court’s ruling, but agree that society is changing. And our statistics must reflect those changes.

Fast-growing population groups — mixed-race Americans, those with “hyphenated” identities, immigrants and their children, anyone under 30 — increasingly complain that the choices offered by the census are too limited, even ludicrous. Particularly tortured is the Census Bureau’s designation, since 1970, of “Hispanic” as an ethnicity or origin, thereby compelling Hispanics to also choose a “race.” In 2010, Hispanics were offered the option to select more than one race, but 37 percent opted for “some other race” — a telling indicator that the term itself is the problem.

Indeed, anyone who filled in “some other race” that year was allocated to one or more of the five main groupings. Many absurdities have resulted.

America has about 1.5 million immigrants from sub-Saharan Africa — some 3 percent of the nation’s black population. Like President Obama’s father, who was Kenyan, their experience differs vastly from that of African-Americans whose ancestors were enslaved, yet they are subsumed into the same category — one that, until this very year, continued to include the outdated term “Negro.”

The census considers Arabs white, along with non-Arabs like Turks and Kurds because they have origins in the Middle East or North Africa. Migrants from the former Soviet nations in Central Asia are lumped in as white along with descendants of New England pilgrims.

An indigenous person from Peru, Bolivia or Guatemala is Hispanic, but if she “maintains tribal affiliation or community attachment,” she might also be counted as part of a racial group that includes the Inupiat and Yupik peoples of Alaska.

Are Australian immigrants whites or Pacific Islanders? (The Census Bureau’s own documents are unclear on this.)

The census has no second-generation question, leaving Congress to debate immigration reform with inadequate statistics about which new Americans are learning English, finishing school, living in segregated neighborhoods or staying out of jail. Social scientists closely track intermarriage as an indicator of assimilation, but the census reports intermarriage only among whites, blacks, Hispanics and others — overlooking unions between, say, Japanese and Chinese, Cubans and Mexicans, Nigerians and native-born blacks. These marriages may have as much to tell us about where the nation is headed as the rate at which whites intermarry.

Much attention has been paid to the news that non-Hispanic whites now account for less than half of births in the United States and that deaths now exceed births among non-Hispanic whites. These projections are oversimplified and misleading because they rely on the outdated “five races” concept. The far more significant turning point is the shift from a nation of a few large racial blocs into a hybrid America of numerous nationalities, ethnicities and cultures, unprecedented in human history. It is this hybrid, multivalent, dynamic America that is not reflected in the census. We cannot, however, fix this at the expense of abandoning racial categories, which are still needed for legitimate policy purposes.

The Census Bureau has begun to consider what changes it will recommend for the 2020 census. It will focus, appropriately, on operational improvements, like increasing response rates. But there are also political decisions to be made.

I URGE three actions. First, drop the current race questions, which misleadingly conflate race and nationality, and ask two new questions: one based on a streamlined version of today’s ethnic and racial categories, and a separate, comprehensive nationality question. (The 2010 census asked Hispanics, Asians and Pacific Islanders to specify a national origin and allowed American Indians and Alaska Natives to put down their tribe.)

These two questions would allow for much-needed flexibility. Broad racial groupings are significant for protecting voting rights, but information on national origin is more useful for understanding health disparities in a metropolis, or for diversifying a university’s student body. Indeed, the failure to appreciate rising inequality within the country’s white majority and to distinguish, say, inner-city blacks from African asylum-seekers, or Southeast Asian refugees from well-educated East Asians, have contributed to the criticisms of affirmative action as too blunt a tool of social policy.

Second, add parental place of birth to the census. One-fourth of Americans under the age of 18 are children of immigrants — a proportion that will increase sharply over the next quarter-century.

Third, slowly phase in the use of the data to make policy. There is a precedent: in 2000, there was strong opposition to the new option of selecting more than one race. It was feared that this would reduce the size of various racial minorities. The government responded by counting those who are white and of one minority race as minorities for the purposes of civil-rights monitoring and enforcement. The new comprehensive statistics on national origin would be put to use judiciously. The five races would not disappear from the statistical system, but neither would they be the only policy tool available.

Americans may hope for a colorblind future, but we know that the legacy of discrimination continues to haunt us; that some new immigrants are assimilated even as others are left behind; that new versions of racism crop up, within as well as among the five “races.”

Faced with these empirical realities, statistical ignorance is a moral failure. It is also a political failure to ignore the arrival of a hybrid America. Even the questions on race we use in 2020 will be wrong for 2100. It will take decades of gradual re-engineering to match census statistics to demographic realities. The Census Bureau is prepared; what’s missing is public awareness and political leadership.