Since the release of the 2010 Census Race and Hispanic Origin Alternative Questionnaire
Experiment (AQE) report in August 2012, much has been written about the AQE results
(Compton et al., 2012; Hill and Bentley, 2013; Stokes et al., 2012). Several recommendations
were made based on the AQE findings; one of which was to further test a combined race and
Hispanic origin question. Recently, numerous articles and blogs supporting or arguing against
the use of combined or separate race and ethnicity questions have made national headlines (El
Nasser, 2013); particularly, about the Census Bureau’s recommendation to continue testing a
combined question during the 2020 Census testing cycle (Compton et al., 2012). One concern,
largely stemming from the Latino community, is the potential negative impact on race reporting
among the Hispanic or Latino population (e.g., the undercounting of ‘Afro-Latinos’) if a new
combined question is approved for the 2020 Census.
1
In response to these concerns, the Census
Bureau developed supplemental analysis from the AQE, specifically examining differences in
race distributions by Hispanic origin when alternative questions were tested (Hill and Bentley,
2013). The results from this study are discussed later in this paper.
The Census Bureau is committed to improving the validity and reliability of census data, and
over the last few decades, many census studies have examined race reporting among Hispanics
(Stokes et al., 2012; Ennis et al., 2011; Martin, 2002; U.S. Census Bureau, 1996 and 1997).
However, none examined race reporting among self-reported Hispanics in the decennial census.
In this analysis, self-reported Hispanics are defined as those whose origin was not imputed.
2