by Andrew
Responding to hilzoy's post on discrimination, in which I end up making a case quite different to what I expected.
For those who don't recall, the study hilzoy referenced sent virtually identical candidates to interview for jobs, the one variable being ethnicity: each job was applied for by white, hispanic, and black applicants. The results could hardly be called surprising: white applicants received callbacks more often than hispanics, who received more than blacks. More depressing was the discovery that white felons received offers more often than blacks with no criminal record; I can't pretend to be surprised that discrimination still exists in hiring, but that it exists to such a degree that people would rather hire a white criminal than a law-abiding black applicant is rather distressing. It is one thing to discover that, given three equal candidates, whites do better. It's quite another to learn that discrimination remains so strong it can overcome something as significant as a felony. (Although it should be noted that the study points out that the differences between black and white responses are not statistically significant. On the other hand, that's not a big help, either: that means a white felon is only discriminated against the same as a law-abiding black applicant.) In response to this, hilzoy notes that this "makes affirmative action of some sort an obvious remedy."
If that doesn't illustrate the difference in temperament between liberals and conservatives perfectly, I can't imagine what would. Hilzoy and I are in tune when it comes to our emotional reaction to this news: we're appalled and depressed. Yet where my mind immediately leapt to the question 'what can we do to address this,' hilzoy's mind went immediately to affirmative action, calling it an obvious remedy. Well, obvious to her, at least. Given what affirmative action does, I'm a little less sanguine about the odds it will resolve this problem than hilzoy.
Affirmative action is basically a quota system. It's not intended to be one, but it is, because quotas are the only way to protect a business against charges of discrimination. Unlike the study that sparked this discussion, real companies are rarely presented with applicants who are identical save for their ethnicity. In as big a system as the American economy, a normal distribution will mean that we could expect that, in the absence of discrimination, we could expect to see some companies with non-whites overrepresented and underrepresented to a significant degree. But no company can afford to base its defense against charges of discrimination on being at the tail end of the normal distribution. If I'm not mistaken, (and we have more than enough lawyers to correct me where I'm wrong here)[Update: I am incorrect, see here for a helpful precis on the law], discrimination law already allows plantiffs to use disparate results as proof of discrimination. In other words, it doesn't matter if your hiring process is completely color blind; if you don't have a work force with appropriate percentages of non-whites, you're guilty. There is no other choice for businesses but to make sure that they hire a certain minimum of candidates from the right ethnicities to forestall costly legislation litigation.
Now we sail into touchy waters. The problem that a quota system creates is it may be that the number of non-white candidates who are equally qualified for jobs as white candidates may not be proportional to their representation in the general population. In other words, let's say that if the perfect candidate is a 100, and acceptable candidates are scored at 85 or above, the ethnicity of those who score 85 or higher is probably disproportionately white. Doubtless that comment will get me tarred as a racist, but for those who can set aside such reactions, please bear with me. Non-whites graduate from high school and college at lower rates than whites (a Manhattan Institute study showed than 78% of whites received normal high school diplomas in the class of 2002, as opposed to 56% of blacks and 52% of hispanics). That means that businesses would be faced with a difficult choice: hire less-qualified candidates to ensure hiring numbers don't draw lawsuits, or hire the best candidates and accept the risk of lawsuits.
Those options aren't as simple as they sound, either. This study, and general experience, tend to suggest that non-whites face a lot of discrimination that people aren't even necessarily aware of. I'll wager that, if confronted with their disparate reactions to similar candidates based on ethnicity, a lot of employers would be surprised to learn they're discriminating in such a manner. I'll bet we all internalize discrimination a lot more than we realize. Without a quota system that forces businesses to hire a minimum number of non-whites, we end up with a nominally color-blind process that, in fact, is discriminating against non-whites. This is why I think there's so much enthusiasm for affirmative action. I don't know how much actual good many affirmative action programs do; shifting the burden from non-whites to whites may be a net gain given the advantages whites possess in American society, but try explaining that to the people who end up not getting jobs because a business needed to hire a non-white in order to meet its quota. Conversely, how often does this really happen?
If we were to take the 2002 graduation rates and the 2000 census results, let's look at the numbers. In 2000, the population was 64.3% white, 10.6% black, 12.5% hispanic and 12.6% other (these numbers are a SWAG, as the data count is designed so you can be white or black and hispanic. I pulled the hispanic numbers out of the black and white population proportionately, which is a fudge.) So out of every 100 job applicants, we could expect to see 64 whites, 11 blacks, 13 hispanics, and 12 others. (All numbers are rounded to the nearest whole number.) We'll throw out the others since they weren't included in the study. So out of every 100 white/black/hispanic job applicants, we could expect to see 74 white applicants, 12 black applicants, and 14 hispanic applicants. Out of those, 58 white applicants would have high school diplomas, seven blacks would, and seven hispanics would.
Assuming the business was attempting to maintain a workforce that was as close as possible to 74% white, 12% black, and 14% hispanic, how often would a more qualified white applicant be turned away in favor of a less-qualified non-white, using a high school diploma as our definition of qualified? 74% of the time, the business would hire the white over the black or hispanic. The other 26% of the time, the business would hire a black (12%) or hispanic (14%) over a white (assuming a proportional number of applicants). All told, using our methods, we end up with a work force that is 71.7% qualified and 28.3% unqualified, with 5.3% of the unqualified workers being black, 6.7% being hispanic, and 16.3% being white. When a black is hired, there is a 44% chance the applicant is unqualified, against a 22% chance the white applicant is unqualified. So, the 12% of the time this business hires a black, 5.3 of the hires are unqualifed. Had whites been hired instead of blacks, 2.6 of the applicants would have been unqualified, so assuming everyone lines up evenly, we end up hiring an unqualified black over a qualified white 2.7 times. So the odds of a white being passed over for a less-qualified black in this scenario is less than one in 33. For hispanics the numbers are a little higher, 3.6 times the qualified white is passed over for a less-qualified hispanic, which is still less than one in 25. Summing the two, we end up with a 6.3% chance of a white applicant being passed over for a less-qualified non-white.
In the real world, things aren't so simple. If a business knows it needs to make sure that 12% of its hires are black, they are going to take reasonable precautions to make sure that it hires the best available, because to do otherwise is to hurt the business. (Which, ironically, it appears is already happening if a white felon is as likely to be hired as a law-abiding black. Not to tar felons unfairly, but I suspect that businesses staffed by non-felons are, on average, more successful than those staffed by felons.) That doesn't mean that the occasional white won't end up being shown the door in favor of a less-qualified non-white, simply because of the relative dearth of qualified non-whites, but I have enough faith in business to act in its own interests to believe that the number is less than the 6.3% figure I just arrived at, so let's say, for sake of argument, that the 6.3% figure is accurate. When you apply that to the 26% of the time a non-white will be hired, you end up with 1.6%, or less than one time in fifty a white applicant is turned away for a less-qualified non-white.
Given that this study is showing that an equally-qualified black applicant has only 56.5% as much success as a white applicant, the numbers seem to favor affirmative action. Let's say, for the sake of argument, that we create a business that is supposed to only hire qualified applicants, so we throw out the unqualified pool and focus on the qualified pool. Our qualified pool is now 80.6% white, 9.7% black, and 9.7% hispanic. So we should expect that our fictional business will hire a white 81% of the time, a black 9% of the time, and a hispanic 10% of the time. Using current hiring practices, however, out of every 100 applicants, we will have a workforce that is 86.2% white, 5.5% black, and 8.3% hispanic. In other words, a black applicant can expect that almost half the time a less-qualified white will be hired, and a hispanic applicant will lose out about one time in five. All told, a non-white applicant can expect to lose out to a less-qualified candidate 26.8% of the time, or about one time in four. Applying that to the 80.6% of the time we should expect a white to be hired, that's 21.6% of the time non-whites lose out due to ethnicity under the current system, as opposed to 1.6% of the time whites lose out under affirmative action. There is a lot more to the world than math, and my figures are extremely rough in any case, but that's an awfully powerful argument for affirmative action. This is not the first time that I've made a case other than the one I expected to when I started, but that's part of what makes blogging interesting.
I still have reservations about affirmative action. My biggest concern is that it doesn't address the underlying problems. This study demonstrates that there's a lot more racism out there than most Americans realize, for example, and forcing businesses to hire proportionate to the local population isn't going to make racism go away; it may even exacerbate it by giving whites a (generally unjustified) sense of losing out because of their race. Affirmative action will do nothing to fix problems like blacks trailing whites by 22 points in high school graduations. Affirmative action strikes me as a means of reducing total injustice, but perhaps at the cost of setting aside other pressing problems that require far more complex solutions.
I don't have an answer for how to fix those problems. Would that I did. It does seem, however, that hilzoy has a pretty strong argument in favor of reducing total injustice via affirmative action. There are other arguments to be made against such programs, particularly depending on how they are implemented, but I think she's got the better of this particular aspect.
Cross-posted at Andrew Olmsted.
Recent Comments