One of the my current work projects is a validation study for a set of cognitive ability (i.e., intelligence) and personality tests used to hire Meter Readers at San Diego Gas & Electric. As part of the study, I’m looking at adverse impact ratios. In other words, I’m determining whether or not people from different races tend to pass the tests at a higher or lower rate.
SDG&E asks applicants to pick their race from the usual list that you see almost everywhere: Black, White, Hispanic, Asian, American Indian, and Other.
While mulling through these data, that list took on a strange quality. It seems deficient to me. Of the 1,800+ people who had taken the test, only seven people claimed to be American Indian. More people chose “Other”, so why have American Indian? And why not have “Arab” or “Indian” (as in from India)? Presumably, Arabs and Indians are supposed to consider themselves White and Asian, but are they? (Hint: No, they aren’t.) Why are we and so many other employers so set on the above list of 5 races and a catch-all “other”?
I did a bit of digging and found out the answer. As usual, it boils down to tradition. One of the documents that Human Resource Professionals treat like the Bible (or Koran; take your pick) is the Uniform Guidelines on Selection Procedures, which dictates a lot of stuff you should and should not do. While thumbing through them, I found this passage:
B. Applicable race, sex, and ethnic groups for record keeping.The records called for by this section are to be maintained by sex, and the following races and ethnic groups: Blacks (Negroes), American Indians (including Alaskan Natives), Asians (including Pacific Islanders), Hispanic (including persons of Mexican, Puerto Rican, Cuban, Central or South American, or other Spanish origin or culture regardless of race), whites (Caucasians) other than Hispanic, and totals. The race, sex, and ethnic classifications called for by this section are consistent with the Equal Employment Opportunity Standard Form 100, Employer Information Report EEO-1 series of reports.
So there you have it, and with an antiquated term like “Negroes” to boot. Personally, I think that’s dumb reason. Not only should we include more races, I think we should change how we collect these data. Currently employers usually ask you to “choose the option that best describes your race.” But what if Tiger Woods is looking to get a part-time job at Wal-Mart to earn a few extra bucks for the holiday? Is he Asian or Black? If he’s half each, no single option “best” describes his race!
The solution, I think, is to give people a list of races and have them fill in percentages so that they total 100%, like this:
Caucasian
Black
Asian
Hispanic
American Indian
Orc
et cetera. For a White guy like me, it’s easy. 100% Caucasian. For someone with a Hispanic mother and a Caucasian father, it’s 50% 50%. Simple.
Of course, what we gain in richness we pay for in complexity. This kind of data would make doing simple adverse impact analyses really difficult, and employment laws aren’t written to deal with it. Still, that could be changed and I think it would be progress.
I think this is a really interesting idea, Jamie. My question is this: what is this data actually used for by employers? I can understand colleges gathering it, for instance, but I’m unclear as to why a company would need that data. Is it just for some interesting stats gathering?
All kinds of things. There are requirements laid down by laws (like the Chapter VII of the Civil Rights Act of 1964) that require to you have a good excuse if your employment tests discriminate on the basis of sex or race (among other things). So if Blacks pass at a lower rate than Whites, you’ve got some ass covering to do.
The most typical (and probably the most legitimate) excuse is that the tests work. That is, they’re valid for predicting future work performance. I’m examining this for two of our tests at the moment. It’s probably still in your interest to reduce the tests’ discrimination, but you’re covered legally and sometimes it’s just the facts of life (like intelligence tests against some minorities and physical abilities tests against women).
As I said in the OP, these legal requirements are what has shaped our data collection methods, even though they don’t quite line up with the rest of reality. I’m interested in ways of expanding that, though.
Not to mention the EEOC requires employers to report annually on the ethnic/racial makeup of their workforce. So it’s kind of this double-edged sword where you (an employer) have to ask for it, but you just can’t make any employment decisions based on it.