Writing for The New York Times, Nate Cohn recently reported that more Hispanics are identifying as white. The piece -- which even includes a cute graphic in which a (presumably Latino) man steps from one square to another, miraculously becomes white, and then rises up to the sky -- suggests that this may be "new evidence consistent with the theory that Hispanics may assimilate as white Americans, like the Italians or Irish...."
That's interesting (and colorful), to be sure. But as I often tell my data staff, when you discover a surprising fact, you could be on to something -- but you could also just be wrong.
Cohn's analysis is actually a few steps removed, which may explain part of the problem. He bases his discussion on a summary offered by the Pew Research Center, which was in turn reporting on a presentation given at the annual meetings of the Population Association of America.
The underlying research is novel in several ways, one of which is that it links individual answers on the 2000 Census with the answers those same individuals gave when administered the Census in 2010.
It turns out that 2.5 million Americans who marked Hispanic and "some other race" in 2000 indicated that they were Hispanic and white a decade later; while another 1.3 million people flipped the other way, it's still a large net gain -- about 3.5 percent of the Latino population in the year 2000.
So what happened? Perhaps assimilation is indeed alive and well? Maybe the racial threats posed by anti-immigrant rhetoric led some Hispanics to become defensively white? Maybe it's young people who became adults over the course of the decade and finally got a chance to choose their identity rather than have it chosen by the head of their household?
Or maybe the question changed.
That actually seems like the most likely culprit. Demographers have already been aware of the rise in the share of Latinos identifying as white: In the 2000 Census, 48 percent of Latinos identified as white, while in 2010, 53 percent chose that tag. Of course, after a full decade of migration, births, and deaths, those could be very different people -- and that's precisely why the linked record analysis is so interesting.
But you have to ask the same questions. In general, the Census asks two general questions: First, are you Hispanic? And second, what race are you? There's also a specific admonition to the respondent to answer both the questions being asked. And what was different in 2010 was the inclusion of a line -- in bold type -- saying, "For this census, Hispanic origins are not races."
The addition of that phrase could have pushed Latinos into skipping the "other" category and choosing one of the predefined groups -- sort of like whitening through reworking the data rather than the reality. That was, in fact, part of the motivation: The Census wanted what it thought would be a more accurate race count. But could a simple bolded -- albeit nuanced -- statement have really made such a large difference?
One way to examine the issue is to look at American Community Survey (ACS), an annual sample administered by the Census. The yearly ACS has taken the place of what used to be called the long-form census, an instrument that collected in-depth socioeconomic data every decade. There, the question switch -- that is, the addition of the bolded message about racial identification -- occurred in 2008.
So how did the answers in the 2008 ACS compare with those in 2007? While this is not quite as good as individually linked records, one suspects that it's close -- some Latinos died, some were born, and some migrated, but surely the populations were fairly similar in the two concurrent years.
As it turns out, in 2008, the share of Latinos identifying as white in the ACS went up by over eight percentage points -- twice as much as the net shift shown in the linked record analysis for 2000 and 2010. It's hard to think of anything happening that year that would have driven up whiteness quite that quickly -- except for the change in the question.
To be clear, the researchers who made the original presentation were careful to clarify that they were reporting, not interpreting, the numbers, and the commentary by Pew also noted that it could have been a change in the question (something Pew also noted in past postings). That didn't prevent reporters, including at MSNBC and Slate as well as at The New York Times, from speculating on the changing meaning of race for Latinos in America.
I've actually looked at the issue in some detail, most recently in work with my colleague, Laura Pulido. The article, published in American Quarterly, examined how Latinos in Southern California mark their racial identification answers in the ACS.
We're geographers -- so we were particularly interested in whether living in a suburb had a "whitening" effect. (In fact, the paper was title "Where in the World Is Juan -- and What Color is He?") But because housing location is associated with income and education, we used a technique called multivariate logistic regression that tried to control for the impacts of all the variables at the same time.
And before you even ask: We only used data from 2008 to 2010 -- all years after the bolded directions made their way in to the question -- partly because we saw the big jump in the underlying data.
Suburban location was indeed associated with Latinos identifying as white, and this was on top of factors that also pushed in that direction, such as age, income, education, and English ability. But our most novel result may have involved the recency of migration.
Much of the previous work in this area has suggested that new immigrants might be confused by U.S. racial categories and so mark "other"; with more time in country, they will settle into being white. But our regressions indicated that, controlling for other factors, the more time you spend in America, the less likely you are to think you're white.
In our view, there may be something about being "othered" by the society that drives Latino racial identification away from rather than toward whiteness. All this seems relevant in light of the current heated debates about immigration, particularly in places like Arizona and Alabama (and even Congress).
In any case, to those who follow the discussion of race -- and to the raft of colleagues that sent me a note about the piece by Cohn -- a more nuanced understanding of the data and the questions being asked would probably lead one away from a breathless conclusion that a new and fundamental shift in Latino assimilation is occurring.
And for the record, I marked Latino and "other" both years. I value consistency in my answers as well as in the questions.