Today is Martin Luther King, Jr. Day, as I’m sure you know. MLK Day was the only federal holiday we got off at Duke, or at least the only one that fell during the semester. It was always marked with a big celebration and my dance group often performed. I always liked that celebration.
But I’ve gotten totally off-topic. An article this week in the NYT highlighted the issue of choosing a race, particularly on census forms, for Latinos in the US. Latinos, who are incredibly diverse in physiognomy and heritage, are, according to the article, choosing to mark ‘other’ instead of one or more of the categories provided.
The issue is of particular importance to economists because in most microeconomic work, we control for race. The implication of this, of course, is that by including someone’s race in a regression, we are separating out some aspect that is predictive of whatever behavior or outcome we’d like measure. And not only are we separating it out, we’re separating it out in a measured, specific way such that we think it applies to all respondents.
For example, we might see a regression that says, all other things equal, the average black person receives one more year of education than a white person. (I saw a statistic like this the other day, saying that black people of similar wealth and socio-economic status get more education than their white peers, I wish I could remember where it came from.) Though the statement is necessarily couched with “on average”, if a number of people are choosing other instead of white or black or some combination of these, we’re not actually seeing the true average. This is called measurement error, and can have pretty significant effects on esimation.
In my own work, for instance, black mothers and white mothers in the Fragile Families Data display different characteristics and decisions regarding investments in children when controlling for whether they’ve received a promise of financial support. But if I were able to capture more of the group that self-identifies their race as other, this effect may be reduced or even disappear.
The question of whether to even ask about race, or ethnicity, is a sticky one. It may give us information that gives different groups more “clout” as the NYT article argues, or it may reinforce stereotypes and feed the flames. Regardless, if research continues the way it currently goes, having a large group of people opt out because they don’t find something that fits them is problematic.
There’s still lots of thinking to be done about it, and perhaps today is a good day to mull it over a bit. I hope you enjoy your MLK Day!
–“The arc of the moral universe is long, but it bends towards justice.” -MLK, Jr.