The Washington PostDemocracy Dies in Darkness

There’s a big problem with how the census measures race

Activists hold signs during a news conference in front of the Supreme Court in 2015. (Getty Images)

Will the 2020 Census be accurate? A number of observers have been worrying about that question for several reasons. For instance, the Justice Department has been trying to insert a citizenship question on the census form; such a question could discourage many immigrants from completing the form. As a result, cities and regions with large numbers of immigrants could see their populations seriously undercounted, with troubling results for political representation, services and funding.

But there’s another reason to be worried, one that hasn’t gotten much attention. The Census Bureau just announced that its 2020 form will not fundamentally change the questions it uses to ask about ethnic and racial origins. This may seem like a minor technical issue — but it will have major real-world implications. If it does not incorporate already-tested improvements into these questions, the census will deliver a less accurate picture of the United States.

And as a result, census statistics will continue to roil the public discussion of diversity, by exaggerating white decline and the imminence of a majority-minority United States. Political figures and pundits who oppose immigration and diversity could exploit that, peddling an alarmist narrative that doesn’t fit with the long-standing reality of mixing between immigrant and established Americans.

To many Americans, being patriotic means being white

Today’s census questions misunderstand both Hispanic and white identity

Right now, census data are distorting one of the most transformative population developments of the early 21st century. A sizable and growing number of young people come from families with one white and one minority parent, as more adults form families across racial and ethnic lines. By far the largest group among them have Hispanic and white European ancestry.

But you wouldn’t know that from the 2000 or 2010 Census results. While 2000 was the first to allow Americans to report a multiracial heritage, neither it nor the 2010 Census allowed people to check off both part Hispanic and part something else.

Why? The culprit is that the census examines race and ethnicity with two questions: first, race; and then Hispanic origins. This format was created in accordance with a 1997 Office of Management and Budget memorandum that defined the standards for collecting and classifying ethnic and racial data to which all federal agencies must adhere.

When the Census Bureau announced that 2020’s basic categories wouldn’t change, it signaled also that OMB is not prepared to update its 20-year-old standards.

In the existing two-question format, when individuals report having Hispanic ancestry, the Census Bureau assumes, following the OMB standards, that they are only Hispanic regardless of their answers to the race question. It’s certainly true that, as the census often notes, “Hispanics may be of any race.”

But in 2015, the Census Bureau tested a unified question that treated Hispanic origin as like a race, which corresponds to how most Americans think of ethno-racial groups. Respondents could report more than one origin. Almost 30 percent of Hispanics indicated a second ethno-racial origin — which, in most cases, was white. The unified question delivered superior results to the two-question approach in other ways, such as a dramatic reduction in the number of Americans (mostly Hispanic) who claim an unclassifiable “other” option on race.

Sticking with the two-question format means that the great majority of young people with mixed Hispanic and white origins will be categorized only as Hispanic — and therefore as “nonwhite,” in census terminology. This classification will often contradict how they perceive and experience their identity, and how they’re treated by the world around them.

White racial resentment has been gaining political power for decades

And it is sociological nonsense. A growing body of data reveals that individuals from mixed families look more like whites than they do like minorities — except for those who are partly black. The exception demonstrates, it should be emphasized, the persistent and severe racism that confronts Americans with visible African heritage.

Trump’s election is actually a return to normal racial politics. Here’s why.

This blurring of differences from whites is especially true for people from mixed Hispanic and white families. Family mixing weakens attachments to the Hispanic group, research finds. According to a recent Pew study, a growing number of such individuals no longer identify as Hispanic — much as Americans from, say, a mix of Slavic and Italian and Irish backgrounds may now mainly think of themselves as whites.

Using the “one-drop” rule for Hispanic origin leads to inaccuracies

Distorted census data can result in inaccurate statements of “fact” and misleading projections for the future.

For instance, since 2013, the Census Bureau has declared a majority of babies born in the United States are nonwhite — by counting all infants with any mixed origins as nonwhite. But this is only true under a “one-drop” rule. National Center for Health Statistics data shows that more than 50 percent of U.S. babies have a non-Hispanic white mother.

And classifying those from mixed Hispanic and white families as “nonwhites” results in Census Bureau population projections of a majority-minority society by the mid-2040s. But such projections are grossly misleading because of the binary thinking that undergirds them and the misclassification of individuals who are partly white and partly minority.

Of course, the projections assume also that today’s ethno-racial classifications will remain unchanged — and that no matter what Hispanic ancestry gets mixed with, it will remain a primary identity. But a majority-minority society should be seen as a hypothesis, not a foreordained result.

Distorted classifications and projections can increase bias among whites today

These distortions have consequences. Recent social-psychological research — for example, by Maureen Craig and Jennifer Richeson — reveals that many whites react negatively to the idea of a majority-minority society and assert more conservative attitudes because of it.

Demographic data should enlighten us. Instead, the census may have obscured the realities of U.S. racial and ethnic identity — in ways that affect the nation’s politics. In the 2016 presidential election, according to research Michael Tesler has reported here at The Monkey Cage, President Trump appears to have gained many votes from whites because of their anxiety about a rapidly changing society that would soon leave them as part of a minority.

And Trump again sounded this theme of wanting to keep the United States from ethnic and racial change as he discussed immigration proposals with congressional leaders — including his comments calling for less immigration from African countries and more from countries like Norway. The proposals and their justification are reminiscent of restrictive immigration legislation introduced during the 1920s, at a time when pundits and political figures were warning about “race suicide.” The proposed immigration quotas were based in part on faulty data from the Dillingham Commission.

Data matter — for our understanding of who we are now and for how we think about our collective future.

Richard Alba is Distinguished Professor of Sociology at the Graduate Center, CUNY.  With former Census Bureau Director Kenneth Prewitt, he has edited a forthcoming issue of The Annals of the American Academy of Political and Social Science titled “What census data miss about American diversity.”