The U.S. Census Bureau released research from its 2010 Census Race and Hispanic Origin Alternative Questionnaire Experiment, which provides a comparison of different census questionnaire design strategies for collecting census data on race and Hispanic origin.
The Census Bureau follows federal standards for collecting and presenting data on race and Hispanic origin established by the U.S. Office of Management and Budget (OMB) in October 1997. This research tested questionnaire strategies with the goal of increasing the reporting in the race and ethnic categories as defined by OMB, decreasing nonresponse, increasing the accuracy and reliability of the results and eliciting responses for detailed race and ethnic groups.
The results will guide further research on the collection of race and ethnicity throughout the decade, informing OMB and Congress.
“The U.S. Census Bureau is committed to improving the accuracy and the reliability of census results by expanding our understanding of how people identify their race and Hispanic origin,” Census Bureau Director Robert Groves said.
This research is the largest quantitative effort ever on how people identify their race and ethnicity to start off the planning cycle for the once-a-decade census. The study mailed experimental questionnaires to a sample of 488,604 households during the 2010 Census, reinterviewed respondents and conducted 67 focus groups across the United States and in Puerto Rico with nearly 800 people.
The study tested several versions of an experimental combined question on race and Hispanic origin. The current OMB classification treats race and Hispanic origin as two separate and distinct concepts. During the 2010 Census, most households received a census form with separate questions for Hispanic origin and race in accordance with these guidelines. A sample of households received questionnaires with an experimental, combined question.
The results showed that a higher number of individuals were more likely to respond to a combined race and Hispanic origin question than to separate questions. The experimental combined questions had an item nonresponse rate, meaning the percentage of respondents leaving that question blank, of roughly 1 percent, compared with 3.5 percent to 5.7 percent for the race question and 4.1 percent to 5.4 percent for the Hispanic origin question.
The research also aimed to increase reporting in the OMB race and ethnicity categories. The “Some Other Race” category was created to be a small residual category, but as shown in the 2010 Census, “Some Other Race” alone was the third largest race group, after “White” alone and “Black” alone, with respondents of Hispanic origin comprising the vast majority of all people classified as “Some Other Race” alone. The population reporting “Some Other Race” alone was as high as 7.1 percent on the separate race question and roughly 0.2 percent on the combined questions. However, the percent of the population who identified as Hispanic was not significantly different across questionnaires, indicating that the total proportion of Hispanics was not reduced in a combined question approach.
Additionally, another major finding was that removal of the term “Negro” from the “Black, African Am., or Negro” response category did not change the distribution of the black population across the experimental questionnaires. Although 2000 Census results indicated that this term was still relevant to some respondents, this relatively small portion of respondents continues to decrease and, thus, the removal of the term did not have a negative impact on black population estimates.
--The population reporting two or more responses was significantly larger with a combined question, in general, than with a separate question. Reinterview findings suggest that respondents may have been able to more clearly understand the opportunity to report more than one response in the combined format, thus increasing multiple-response reporting. Alternatively, focus group research suggests that the combined question respondents may have been interpreting the question as asking for race and origin.
--The experimental combined questions included write-in lines for all race groups. For the first time, people who answered “white” and “black” could further specify their origin. As much as 50 percent of people who reported as white gave detail when provided with a write-in line compared with only 1 percent to 2 percent in all other questionnaires. Similarly, of black respondents, more than 76 percent reported detail on each of the experimental combined questionnaires, compared with only 3 percent to 6 percent of all other questionnaires.
There was a decrease in Asian and Hispanic respondents providing more detailed responses on the combined questions that did not have checkboxes for specific origin groups. Detailed Asian reporting was 96.6 percent or higher on all other questionnaires, but as low as 92.6 percent when no checkboxes were provided. Detailed Hispanic origin reporting was 86.4 percent to 88.9 percent when specific origin checkboxes were provided within the combined question, but was lower on combined questionnaires that did not contain the specific origin checkboxes (77.7 percent to 80 percent). Detailed Hispanic origin reporting on separate questions was 92 percent or higher.
--There was no change in how people reported within the Asian, Other Pacific Islander, and Hispanic origin examples when different examples were used.
“The findings from this research provide promising strategies to address the challenges and complexities of race and Hispanic origin measurement and reporting issues in our rapidly diversifying society,” Groves said. “This is another step in an ongoing discussion about how we can better understand the changing diversity of our nation. The results will guide upcoming research as the Census Bureau looks toward the 2020 Census.”
These findings will serve as the basis for further research and for a wider discussion among statistical agencies.