RACE II: THAT EMPLOYMENT STUDY. Earlier this month, the New York Times reported on a careful study of racial discrimination in hiring practices in Boston and Chicago. Here’s the study in PDF. The researchers found significant discrimination: White “applicants” (there were no real applicants–I’ll explain this momentarily) had a 10.08% chance of getting called in for an interview after sending in a resume, whereas black “applicants” had a 6.70% chance. That may not sound like much, but it cashes out to a 50% difference. To put it more plainly, most applicants don’t get called back for most of the jobs they apply for. In order to get one callback, the average white applicant would have to send out 10 resumes, whereas the average black applicant would have to send out fifteen. Nine percent of employers favored whites, whereas 3.7% favored blacks. Having a higher skill range also helped white applicants much more than it helped black applicants–thus there’s less pressure from the market (in these two cities) for black job-seekers to improve their skills, since better skills are proportionately less likely to result in better jobs. Again, black job-seekers had to put in more effort for less payoff. Even living in a “nice” neighborhood was more helpful for white applicants than for black applicants.

The left side of the blogosphere had quite a bit to say about this study–here’s J. Bradford Delong; here’s Armed Liberal; here’s CalPundit; Ted Barlow and (I think) Ampersand also linked it.

Here are some links to discussions of the study from the conservative or libertarian wings:

[chirp chirp]

Oh wait, now that I’ve started poking around, I found two, but both are, in my opinion, almost totally unhelpful. (For example, why assume that employers are, reasonably, sketched about black applicants because the black applicants might sue under EEOC rules, but not note that employers might just as reasonably be worried that white applicants would engage in racial harassment? We know that most people do neither; why is the first worry judged significant, neutral, a non-racist concern one might have, whereas the second is not even considered? Similarly, why is “I bet his A’s in college don’t mean much because he’s black and his professors probably gave him an easy time” any more “reasonable” or uninflected with racism than “I bet his A’s in college mean a lot, since he probably faced racial discrimination”? [What about “I bet his A’s in college mean a lot, since he probably faced teasing from peers who thought he was ‘acting white'”??] Neither judgment seems to me to be a particularly helpful way of sorting potential employees. But to assume that the former rather than the latter is the natural assumption is just odd.) I do think there are other explanations than racism for some of this data; for some of it, too, there’s a complex interplay between stereotype and reality that I’ll get to in a minute; but it’s hard to believe that the whole discrepancy can be explained away without any reference to racism. And “explaining away,” rather than explaining, is what those two links seem to me to be doing.

Now, it may be that I just haven’t run across better discussions. It may well be that once I’ve put this post up, I’ll get tons of emails from bloggers of the right, saying, “Hey, man, I was on this study like ugly on an ape!” And I certainly don’t hold people accountable for not blogging about something that they might not have seen… might have seen but didn’t have time for… etc. (I mean, I’m joining this party pretty late myself.) Nonetheless, there’s a lot of interesting stuff going on with this study, and it would be cool to see other people take a look at it from a right-leaning perspective. If it is being ignored because it doesn’t fit with the fairly Pollyannaish view of race relations that many conservatives hold, that’s a shame, because both what the study reveals and what it doesn’t and can’t reveal are intriguing and don’t play to any one side’s expectations or stereotypes.

Oh, and before we begin, I should note that I’m pretty firmly convinced that there are a lot of white racists in this country. This is largely because I’ve met a bunch of them. I don’t, however, think that white racists are the biggest problem facing black people trying to get ahead, or even in the top five, although I do think white racists exacerbate some of the top five problems. More on this tomorrow.

Anyway, here’s the study’s basic setup: Researchers created a slew of fake resumes, sorted into two skill levels (determined by things like how much education the “applicant” had received). They then sent resumes in response to 1300 classified ads. For each ad, they sent in four resumes: two high-skill, two low-skill. The resumes were randomly matched with (fake) local addresses, and also randomly matched with names. The names are the key thing. The researchers had two lists of first names and two lists of last names, which, when matched, would produce a total name that most people would read as “white” or “black.” For example, “Jamal Jones” was a possible black name; “Brendan Baker” was a possible white name. The names were first selected based on both the popularity of the names, and the unlikelihood that the names crossed racial boundaries–in other words, Jamal was picked because it’s really popular with black families and not at all popular with white families, and vice versa for Brendan. Then the names were winnowed by running them by a focus group that was asked to guess whether the names were “white” or “black”; names that were overwhelmingly given to black children rather than white, but not perceived as “black,” were discarded. (The study gives Maurice and Jerome as examples.) So by the time the names are placed on the resumes, the researchers could be pretty confident that the black names would be perceived as black and the white names as white. Each job received, as I said, four resumes: a low-skill white resume, a low-skill black resume, a high-skill white resume, and a high-skill black resume. Then the researchers waited to find out who would get called for an interview.

The researchers describe the ways their study improves on previous studies–for example, previous hiring-discrimination studies used an “audition” model, where a white actor and a black actor would actually go in for interviews for the same job. (The actors would usually be matched to have similar heights and so forth.) But this is sub-optimal not just because people, unlike resumes, can’t be mix-and-matched or standardized–no matter how hard you try to match the pairs, sometimes the black guy will just be more attractive, more personable, or more in tune with the employer’s mindset–but also because the actors knew they were in a study, so there was a strong danger that they would (consciously or subconsciously) skew their performances to produce the results they believed were most likely.

The fact that the “black” names were chosen because they were distinctively black (i.e. they had to be not just popular among black families but unpopular among white families) may have introduced some other kinds of skew into the data, however. The black women’s names are disproportionately “non-traditional” or “unusual” (among non-blacks; obviously they’re “usual” among black city folk or they wouldn’t have been picked), and people generally do discriminate in favor of traditional, solid-seeming names. (Which will no doubt be a great disappointment to the parents here.) Several of the black men’s names sound Arab/Muslim–Hakim, Kareem, Rasheed, and maybe Jamal, thus four out of nine–and the study was conducted after 9/11, so we may be seeing several different kinds of assumptions at play.

Non-black employers may also assume that black applicants with “unusual” names are more likely to come from the ghetto. (That’s an even bigger guess on my part than the previous two things.) Within races, the researchers did check whether some names were more likely to be given by mothers who hadn’t completed high school, but there weren’t any correlations between the diploma rate and the callback rate–in other words, Ebony got more callbacks than Lakisha, but the mother of the average actual Lakisha was not less educated than the mother of the average Ebony. If you follow me. It’s neat that the study takes that into account, but what we really want to know is whether the average Lakisha’s mom is less educated than the mother of the average black girl named Lisa’s mom (i.e. are poorer families more likely to give distinctively black names), and whether the average employer thinks “Lakisha” is not just a black name, but a ghetto name. The researchers tried to show that the discrimination was not about class; they noted that black applicants from middle-class neighborhoods still had a harder time finding jobs. But I think they may be assuming an overly-rational employer, one who knows which neighborhood the fake addresses come from, is more swayed by address than by the unusual name, can accurately assess the frequency of a name, and so on.

The white names have some weird quirks as well. Six of the nine white last names are… Irish. (All the last names: Baker, Kelly, McCarthy, Murphy, O’Brien, Ryan, Sullivan, and Walsh.) Given that Boston showed more discrimination than Chicago, does this mean simply that Boston is more pro-Irish than Chicago, rather than that Boston is more anti-black?

Two interesting things that we can’t know from this study: 1) Which races discriminate most against blacks, and by how much? What proportion of black employers discriminates in favor of blacks, and what proportion (if any) against? (It’s in no way unthinkable that black employers would discriminate against black applicants–black cab drivers often bypass black men trying to hail cabs, etc.)

2) Who else would get discriminated against? If Kristen Walsh and Chava Rosenstein try for the same jobs, how often will Kristen get a callback as vs. Chava? Kristen vs. Chava vs. Lakisha? How much discrimination is considered a huge problem–would your view of the study change if you found that Ebony (one of the less-discriminated-against black names) was discriminated against only as much as Chava, or Yahyin? (I’d be pretty shocked if any other race was as discriminated against as the most-discriminated-against black names, with the possible exceptions of Arabs and, in areas near reservations, Native Americans. Aisha, Keisha, Tamika, Lakisha, and Tanisha all did worse than the lowest-callback-rate white women’s names. But Ebony, Latoya, Latonya, and Kenya received callbacks only slightly less often than the average white women’s names.)

The other big interesting thing we can’t learn from this study, of course, is why the callback gap appears. I’ve outlined some partial explanations, but I don’t think they explain the whole gap. (I’m not a social scientist, so take this all with a grain of salt, of course.) A chunk more of the gap is explained, I suspect, by the very existence of ghettos–black names are associated with poverty, poor schooling, and violence. This is one of the problems of being a minority: Members of the majority may only know the minority from the evening news, and what gets on the evening news is rarely good. This association likely persists even when the actual resume in front of the employer shows a college degree or a fairly high skill level.

There’s a good quick explanation of the process in this article (link requires registration), which InstaPundit linked and which I’ll talk more about in a moment: “Parsing the statistics, it can be hard to untangle cause from effect, hard to know whether to blame the segregation on poverty, or the poverty on segregation. Demographer Roderick Harrison, who studies these issues at Howard University in Washington, D.C., suspects that blame is due all around.

“‘When you have these disparities’ in income, education and housing, ‘it feeds the stereotypes: Black people equal poverty, crime, welfare — all the things that whites moved out to the suburbs to escape,’ Harrison said. ‘That increases white resistance and fear to having even middle-class blacks move into the suburbs.’

“And that increases segregation — which in turn widens the gulf between black and white, by keeping African Americans from better schools and jobs in the suburbs.”

That sounds right to me. Note that the equation of “black people” and “poverty, crime, welfare” is still racist–it judges an entire race based on the actions of a few. It assumes, even in the face of specific evidence to the contrary, that individual black people will be gangbangers or illiterates or whatever. It’s pretty messed up to judge a resume based more on the name at the top than on the credentials in the middle. But part of the tangled problem here is that you have to address both things at once–both the discrimination and the half-rational assumptions that underly it. You have to say, “Judge applicants as individuals, question your snap racial judgments, pay attention and check to see whether the resume actually fits the stereotype you have”–and you also have to reduce the poverty, and the crime, and the welfare rate. Just repeating, “Don’t assume,” isn’t going to work. We can argue about to what extent reducing racism will help clear up the other problems in poor black communities. I think it’s pretty obvious that it would help; it certainly couldn’t hurt. I also think it’s pretty clear that reducing other problems in poor black communities will make racism seem less “reasonable,” less natural, and thus it will become less prevalent.

I’m going to blog more tomorrow about solutions, now that I’ve sketched out some parts of the problems. I strongly agree with CalPundit’s statement that improving education is key. I’m going to offer a bunch of things that I think will help a) reduce racism and b) improve conditions for the black people most likely to be discriminated against.

I’m also going to blog about my reasons for opposing affirmative action. Here’s something to chew on while you wonder what I’m going to say about that–it’s a result of the callback study that really surprised me: Employers that claim to be “equal opportunity employers,” and “federal contractors, who are more severely constrained by affirmative action laws,” do not discriminate less than other employers. Repeat, (federally-contracting) affirmative-action employers do not discriminate less. Bizarre.