Saturday, November 17, 2012

Combining the Race and "Hispanic ethnicity" questions?

After each decennial Census, the Census Bureau produces reports about the data -- but it also reflects on the process and technical aspects of the Census survey and methods.  To test out possible changes to future Censuses, the Census Bureau conducts "Alternative Questionnaire Experiments (AQEs)."  The AQEs are small-scale experiments and research projects that ask, "What might happen if we changed X about the Census?"  

Some of the AQEs explore potential changes to survey questions about race and ethnicity.  The questions about race and ethnicity change pretty much every Census.  Take a look at the questions from the 1790 Census, the 1890 Census, the 1990 Census, and the 2010 Census and you'll see that the questions about race are different each time.  Why?  That's a big question, but one short answer is the questions about race on the surveys change because the questions about race in larger society change.  So, the Census Bureau conducts some AQEs to explore the possible effects of various potential changes to the questions.  The Census Bureau has recently released a lengthy report detailing what they learned from a set of AQEs about race and ethnicity questions.  This press release provides a brief summary.  

Today, I want to draw attention to one AQE and possible option, in particular: a combined race and Hispanic ethnicity question.  Historically, the Census has long asked questions about race on its decennial surveys.  But, the separate "Hispanic ethnicity" question is relatively new; it was first introduced in 1970.  Since then, the Census has asked about "race" and "ethnicity" as two separate questions.  But, the Census Bureau is now testing what might happen if it combined the two questions. One reason to do this is that the Census has found that having two separate questions confuses many people and that leads to less accurate data.  For example, the Census Bureau has found that most people who check the "Some Other Race" box for the race question, if you ask them, will say they're Latin@ or Chican@ -- but because those aren't options in the race question, they check "Other."  When the Census introduced the "Some Other Race" category, they figured their race options were broad enough that very few people would check "Some Other Race."  Clearly, they were mistaken.  The population of people who check "Some Other Race" is significant and growing.  So, a change is needed.  But what change?  Well, one possibility is to include a "Latino" category in the race question and phase out the separate question "Are you Hispanic: yes or no?"  

What might happen if the Census combined the race and ethnicity questions?  The AQE suggests a few possible consequences.  First, it would likely increase the accuracy of counts of Latinos.  Second, it might reveal that the White population is actually smaller than we thought, by maybe 6-8 percentage points (because some of them are actually Latinos being miscounted, given the way the questions are asked).  And third, it might increase the number of people who check Latino AND another race -- that could mean that the number of people who identify as Two Or More Races could increase and it could also mean that, among the TOMR population, there could be an increase in the number of people who identify as Latino-and-another-race.  

-What do you think of a possible combined race-and-ethnicity question?  What questions do you have?
-Who do you think might _oppose_ a combined question?  Why might they oppose it?  And, knowing that people may not always be forthright about their real motives, what arguments might people pose for opposing a combined question?

I'd like to give a shout-out to Thomas Lopez putting this issue on my radar, well before I joined the NAC.  Thomas is a leader with Multiethnic Americans of Southern California (MASC) and founder of MASC's Latinas and Latinos of Mixed Ancestry (LOMA) project.