Here is another overview of the problems the US Census is having with measuring the Latino population in the United States:

So when they encounter the census, they see one question that asks them whether they identify themselves as having Hispanic ethnic origins and many answer it as their main identifier. But then there is another question, asking them about their race, because, as the census guide notes, “people of Hispanic, Latino or Spanish origin may be of any race,” and more than a third of Latinos check “other.”

This argument over identity has gained momentum with the growth of the Latino population, which in 2010 stood at more than 50 million. Census Bureau officials have acknowledged that the questionnaire has a problem, and say they are wrestling with how to get more Latinos to pick a race. In 2010, they tested different wording in questions and last year they held focus groups, with a report on the research scheduled to be released by this summer.

Some experts say officials are right to go back to the drawing table. “Whenever you have people who can’t find themselves in the question, it’s a bad question,” said Mary C. Waters, a sociology professor at Harvard who specializes in the challenges of measuring race and ethnicity…

Latinos, who make up close to 20 percent of the American population, generally hold a fundamentally different view of race. Many Latinos say they are too racially mixed to settle on one of the government-sanctioned standard races — white, black, American Indian, Alaska native, native Hawaiian, and a collection of Asian and Pacific Island backgrounds.

American conceptions of race usually center on black and white without having much room for middle or other categories. There is a long history of this in the United States as various new groups struggled to become labeled as white.

I like the admission here that the Census needs to find a definition that also fits Latinos’ own understanding. Imposing social science categories on the world can be problematic, particularly if they are not understood in the same ways by all people. Survey questions are not that great if people don’t understand the answers or see where they fit in the possible answers.

This isn’t the first acknowledgment that the Census Bureau has issues here. I would be curious to hear sociologists and others project forward: how will the Census and others measure race, ethnicity, and culture in 2050 when the United States will look very different? Are there ways to measure race and ethnicity in the Census without the pressure of it being tied to federal dollars?