Our goal is to get one race variable to use for analysis.  Below are options I thought of, but I’m sure there are lots of others.

1. We could just use the first racial identity the people reported (the variable race3m1). This has the benefit of simplicity but means that people of mixed race get assigned only the first race they happened to mention. One could argue, however, that the first race a person mentions might be the one they identify with most.

2.  We could use the recoded variable racecomb but as discussed above, you don’t know how Pew created it AND Hispanic identification is not included.

3.  We could use the recoded variable racethn but it excludes mixed race people and groups other than whites, blacks, and Hispanics.

4. We could create your own variable! One way to do this would be to assign single-race people to the category they report and give everyone else a “multiracial” code. This is a small step toward recognizing racial complexity.

5.  We could create a variable with every single possible combination of races.