Census Bureau’s Latino Quandary

The ramifications of the Census Bureau’s recent announcement that Latinos are America’s largest minority will spread to every corner of society. Unfortunately for the bureau, as the Latino share of the population keeps expanding, the ongoing problems of counting Latinos accurately will grow more glaring.

According to the 1990 census, 22 million people called themselves Hispanic. By 2000, the number was 35 million. During this interval, blacks, the largest minority community in these counts, increased from 30 million to 36 million. The bureau’s recent population survey revealed, however, that in the two years since the decennial census, the number of Latinos increased by 10%, to 38.8 million, compared with 38.3 million blacks.

This change probably didn’t surprise a lot of people, especially in Southern California, where Latino growth has occurred not just in traditional Latino neighborhoods but in nearly every city and community. Statewide, Latino newborns constitute the majority. Even in many regions of the country that have not historically had sizable Latino populations, Latinos have increased dramatically. For example, during the 1990s, the Latino population in six Deep South states tripled.

The Census Bureau has not adjusted well to these developments. For 2000, it modified its questionnaire in a manner that failed to count many Latino ethnicities. The bureau instituted these changes without proper testing. Only after numerous Latino community organizations and leaders decried the published figures did the bureau take steps to correct this problem. The new data revealed that, in some areas of the country, the 2000 census undercounted Latino subgroups -- such as Salvadorans, Nicaraguans and Dominicans -- by as much as 20% or 30%.

And a more significant problem looms. Since its inception in 1790, the census has always asked about a respondent’s racial background. As America became more racially complex, the census increased the number of alternative responses. When the Latino population expanded, however, the bureau didn’t add a new category. Instead, in 1980, it added a question about Hispanic origin, rather than a “Hispanic” option to the race question, because it considers Hispanic/Latino ethnicity to be distinct from race. The use of separate questions has advantages. For example, Latinos have a high intermarriage rate, and children of mixed marriages can indicate both backgrounds. That’s important in civil rights enforcement, since separate questions help the bureau avoid undercounting minority populations. Moreover, changing this format after decades of use could reduce the comparability of data among censuses.

Here’s the problem: In the 2000 census, if you identified your race, you were either “white,” “African American,” “American Indian or Alaska Native,” “Asian” or “Native Hawaiian or other Pacific Islander.” Many Latinos, particularly in the Southwest, don’t identify with any of these. So they select “some other race,” in effect, none of the above. In fact, in the 2000 count, more than 40% of Latinos chose “some other race.” Even more telling, of the 15.4 million people who selected “some other race,” 97% were Latinos. No surprise that the 2000 census reported that America’s second-largest racial minority was “some other race.” With the Latino population expected to continue its rapid growth, we can anticipate this category will soon be our largest nonwhite racial group.

There are other effects. The failure to offer Latinos a suitable racial identity discourages them from completing the census process. This is particularly distressing in light of the Census Bureau’s persistent undercounting of minorities. It can also produce a less accurate census, as Latino respondents settle for an identity they consider to be the “lesser of evils.” Furthermore, this mischaracterization can be compounded by the bureau, which imputes -- makes statistical deductions to estimate -- responses that are not provided.

These problems were easier to sidestep when the Latino population wasn’t growing rapidly. But delay is no longer an option. Error in the census is costly. Each nonresponse requires the bureau to follow up with respondents or impute missing data. Follow-up activities constitute one of the largest costs of the decennial census.

The bureau certainly needs to collect race and ethnicity data as accurately and as cheaply as possible. It must also strive to assure that its questionnaires be drafted in a manner that will elicit the information it needs. That will require a thorough review of its questionnaires and who is filling them out. A better understanding of how Latinos identify themselves, especially in the context of a question that may have little meaning to them, should be a key goal of the review. To that end, the bureau, in addition to consulting academics and community organizations, must conduct extensive interviews of and focus groups with the “Latino on the street.” In that way, the bureau will be better able to supply the information the nation needs.