(November 2012) The U.S. Census Bureau has been studying a very different method for collecting data on race and ethnicity. This potential method may provide a more accurate picture of the nation's ethnic makeup, particularly improving the classification of Hispanics.

Tracking the evolution of how the Census Bureau collects data on race and ethnicity is a useful reminder of how statistical data are actually defined. While some characteristics, such as age and marital status, are straightforward and objective, others are less clear and certainly, race and ethnicity is one of them. In a real sense, the Census Bureau does not itself "count" people—people count and identify themselves.

The Hispanic Category

Prior to the 1970 decennial census, Hispanics were not enumerated as a single group. However, there were reports on "persons of Spanish surname" in five southwestern states and one on Puerto Ricans living in the United States. The question on Hispanic origin was added to the survey in spring 1969, late in the census planning process. It appeared only on the long form, sent to a 5 percent sample of all households, and the question was not in a very conspicuous spot on the form. The concept of "Spanish origin" was largely unknown to most Americans and the question did not actually use the term and instead asked about "origin or descent": Mexican, Puerto Rican, Cuban, Central or South American, Other Spanish, None of These.

Subsequent census reports grouped the five ethnic groups under the category "persons of Spanish origin." The question in the 1970 Census was added so late that the group does not appear in the four main census volumes. Some people who lived in the central or southern United States filled in the "Central and South American" circle although they were not of Spanish origin. Remarkably, this new definition combined various Hispanic groups, such as Puerto Ricans in New York and multigenerational Mexican-heritage residents of the Southwest, into a single group.

The 1970 count of "persons of Spanish origin" was 9,072,602, 20 percent of whom were foreign-born. But how good was the count? The Census Bureau compared the results to responses in other categories, such as language use, although they had not originally been intended to count only Hispanics. Table 1 shows just how well the Hispanic question fared among other potential ways to identify the group. Compared to other categories, "Spanish origin" performed poorly. Because the question wasn't clear, some Hispanics did not self-identify themselves as such, while other non-Hispanics did.

Table 1
Hispanic Population Counted in the 1970 Census

Identifier United States Southwestern States1
Spanish origin 9,072,602 5,008,556
Spanish surname not applicable 4,667,975
Spanish language2 9,589,216 5,662,700
Spanish heritage 9,294,509 6,188,362
Spanish language or surname 10,114,878 6,188,362
Spanish birth or parentage 5,241,892 2,321,642

1 Arizona, California, Colorado, New Mexico, and Texas.
2 This group consists of all persons of Spanish mother tongue and all other persons in families in which the head or wife reported Spanish as his or her mother tongue.
Source: Jacob S. Siegel and Jeffrey S. Passel, "Coverage of the Hispanic Population of the United States in the 1970 Census: A Methodological Analysis," Current Population Reports, Special Studies, Series P-23-82 (1970): table 1.

In the 1980 Census, several significant changes were made in the way Hispanics were counted. The question was moved to a prominent position on the short form that was sent to all households, the term Hispanic was included, and the "No" response was placed first so that people would be less likely to choose a wrong category. Chicano was added to accommodate regional differences in terminology.

However, the Hispanic question immediately followed the question on race, possibly influencing the rate of response, especially for Hispanics who had already checked "Other Race" and written in a Hispanic nationality.

In 1990, a write-in box was included to encourage specific responses.

A major change was made in 2000: The Hispanic question was placed before the question on race and a reminder was added to answer both the Hispanic question and the race question. In addition, Latino was added to several choices. On the race question, respondents were not limited to one race and could check all that applied.

In 2010, the Census Bureau made only a few small changes in categories. The official census counts of Hispanics are given in Table 2. Implied growth rates give some idea of how the question performed from census to census. The high rate from 1970 to 1980 is likely due in part to considerable outreach to "stand up and be counted" in Hispanic neighborhoods.

Table 2
Counts of Hispanics in U.S. Censuses

Census Year Hispanic Population (millions) Implied Annual Growth Rate (%)
1970 9,072 -
1980 14,609 4.8
1990 22,354 4.3
2000 35,306 4.6
2010 50,478 3.6

Source: U.S. Census Bureau.

The New Proposal

The Hispanic and the race questions have been two separate questions since 1970. Thus, there has always been a problem with these percentages totaling to 100 percent because Hispanics can be of any race. The Census Bureau is now testing a combined "race or origin" question in order to simplify self-identification and improve coverage. This alternative questionnaire experiment revealed improvements in response rates and no reduction in the size of the Hispanic population. Should the new question be adopted, the race/origin categories would then total 100 percent, including the multiracial group as the first option.

The categories being tested are listed below, and respondents are asked to write in their specific race or origin.

  • White.
  • Black, African American, or Negro.
  • Hispanic, Latino, or Spanish Origin.
  • American Indian or Alaskan Native.
  • Asian.
  • Native Hawaiian or Other Pacific Islander.
  • Some Other Race or Origin.

Carl Haub is a senior demographer at the Population Reference Bureau.