March 3, 2010

Census History: Counting Hispanics

Despite the long history of Hispanic residents in the United States, there was no systematic effort to count this group separately in the Census until the late 20th century. An analysis of changes in Census question wording over recent decades reveals the challenges in trying to count and describe this fast-growing population.

An estimated 48 million Hispanics are now living in the U.S., or almost 16% of the population. Hispanics are the nation’s largest minority group, having surpassed African Americans in number in 2001. The growth of the Hispanic population this century is due mainly to births in the United States, not immigration from abroad, a reversal of the pattern over the previous four decades.

There was a one-time inclusion of a “Mexican” race category in the 1930 Census, when forms were filled out by census-takers who went door to door. The first major attempt to estimate the size of the Hispanic population for the entire nation was in the 1970 Census, in which forms were completed by residents themselves. The question appeared on one of the two long-form questionnaires sent to a sample of the population, not the short form that everybody answered. The question asked: “Is this person’s origin or descent—“ and the response categories were: “Mexican,  Puerto Rican, Cuban, Central or South American, Other Spanish,” and “No, none of these.”

This question did not work very well.  The total count of 9.1 million reported in that census was about 500,000 less than other estimates for the Hispanic population. (See this page on the Census Bureau website for data and more history)  Further, even this 9.1 million count was about 1 million higher than responses to the question by people of Hispanic origin. According to later research, a major problem was that hundreds of thousands of people living in the south or central regions of the U.S. mistakenly were included in the  “Central or South American” category.  As is its usual policy, Census reports on the Hispanic population in 1970 use the originally reported figures.

Hispanic Question on the Short Form

In 1980, the question was moved to the short form that went to all households, and specified that it pertained to Hispanics: “Is this person of Spanish/Hispanic origin or descent?”  The possible responses were: “No (not Spanish/Hispanic); Yes, Mexican, Mexican-Amer., Chicano; Yes, Puerto Rican; Yes, Cuban; Yes, other Spanish/Hispanic.”  The Hispanic origin question followed the race, age, and marital status questions.  (The previously problematic “Central or South American” category did not appear.)

This question counted 14.6 million Hispanics and worked reasonably well.  A few hundred thousand non-Hispanics apparently misinterpreted the question and attempted to identify as “American” by marking the “Yes, Mexican, Mexican-Amer., Chicano” category.  Many actually circled the abbreviation “Amer.” to so indicate on the form.  There also were several hundred thousand people whose place of birth, language, origin or descent suggested that they were Hispanic but who failed to indicate that they were of Hispanic origin. (This and other apparent errors were documented by researchers after the census was completed; as has generally been the case with decennial censuses, the uncorrected numbers appear in official census reports.)

In 1990, the Hispanic-origin question was virtually identical to the 1980 question and counted more than 22 million Hispanics. This census form included a new write-in line to specify a group for “other Spanish/Hispanic” origins.  The form also shortened the category name “Mexican-Amer.” to “Mexican-Am.,” which helped to eliminate the problem of a decade earlier in which respondents tried to identify as “American” by circling “Mexican-Am.”

The 2000 Census, which counted more than 35 million Hispanics, saw some significant changes in the Hispanic origin item.  The term “Latino” was added, so the question read, “Is this person  Spanish/Hispanic/Latino?”  There were major changes in instructions to respondents and question placement.  The Hispanic-origin question preceded the race question, rather than following it, and respondents were instructed to answer both questions.  Immediately after the question was the instruction to “Mark the “No” box if not Spanish/Hispanic/Latino.”  The Mexican category remained the same but both the “No” and “other” categories added the term “Latino.” The examples given in 1990 for the “other Spanish/Hispanic/Latino” category were eliminated.

A major purpose of the new placement and instructions was to persuade Hispanic respondents to specify a category in response to the race question and not mark “some other race” with a Hispanic write-in (e.g. “Mexican” race).  This attempt was largely unsuccessful, as about 43% of Hispanics did not specify a race. Moreover, a new issue arose: The proportion of Hispanics who specified that they were some “other” Hispanic origin, without specifying a country, was much higher than in other surveys and sources.

For the 2005-2007 American Community Survey (ACS), the Hispanic origin question was identical to the one in the 2000 Census.

Changes for the 2010 Census

The question that is being used in the 2010 Census (and in the American Community Survey, beginning in 2008) had two changes in wording. The order of the terms is different (“Spanish” is the third option, not the first) and the word “origin” has been added.  The question asks whether the person is “of Hispanic, Latino, or Spanish origin.”  The wording of response categories has been changed to reflect the question wording. The last response category also has been reworded to say: “Yes, another Hispanic, Latino, or Spanish origin,” and a list of examples is provided (“Argentinean, Colombian, Dominican, Nicaraguan, Salvadoran, Spaniard and so on”) in an attempt to elicit a specific response.

In the lead-in, respondents are instructed to answer both the Hispanic origin and race items (with the items named, not just numbered as they had been in 2000). The instruction specifically says that “Hispanic origins are not races.”  Additionally, the instruction to “Mark No if not Hispanic” was eliminated.

These changes apparently had a significant impact on responses in the 2008 ACS, so it seems likely that 2010 Census responses also will be affected.  The share of Hispanics who gave a specific race, rather than marking “some other race,” dropped by about 10 percentage points.  The pattern of “other” Hispanic responses changed markedly.  The new question wording also may have induced more U.S. natives to identify as Hispanic than had been the case in earlier years of the American Community Survey.