Tuesday, May 5, 2009

Bizarro Affirmative Action

Ricci v DiStefano is being heard this week by the Supreme Court. This has extreme implications.

A good explanation of what happened:

How the Supreme Court Judges seem to view it:

One opinion (the comments are kind of interesting, too):

The two points that made the most sense to me.
From the Times article:

"Near the end of the argument, Justice Breyer asked a series of hypothetical questions to test the contours and limits of the white firefighters’ position.

What if, Justice Breyer asked, an employer sets an application deadline but receives an insufficiently diverse mix of applicants? May the deadline be extended?

What if, he asked, a university is dissatisfied with the number of female professors gaining tenure under its usual requirements? May it suspend the requirements?

What if Texas, which admits high school students graduating in the top 10 percent to its public universities, becomes dissatisfied with the resulting racial mix? May it switch to 15 percent?

Justice Breyer, characterizing the white firefighters’ argument, more or less answered his own questions. If employers know the identities of those who would benefit under the old rules and nonetheless adopt new rules based on racial or similar grounds, Justice Breyer said, the new rules would have to be struck down."


"Individuals, not groups, have rights in American law."

It is my opinion that a "fair" test is one that gives everybody a chance to pass that is equally commensurate with the combination of their innate ability, experience and level of effort for preparation, and is directly related to the privilege for which you are testing to be able to do.

Firefighting does entail a lot of facts that you need to be able to pull up under pressure. You need to know about structural engineering, you need to know about fire and the rate of its spread, for which you need to know about chemistry and electricity. The best way to do that is either oral exams or written tests, and written tests entail much less bias in grading. In other words, you're graded on important skills most equally through a written test.

Thus, giving a written test to firefighters probably does fall under the "directly related" portion the same way a written test is appropriate for a license - maybe answering on paper is not directly what you're doing when you're driving, but if you can't answer it on paper, it's a safe bet you're not going to be able to do it when you're driving.

The question then is, does this test give everyone a chance to pass that is equally commensurate with the combination of their innate ability level, experience and level of effort in preparation - in other words, their intellectual capital, stripped of other components (like race).

The evidence here says yes, it does. The test was designed by a company whose specific intention is to scrub tests of racial bias. Additionally, a man with severe learning disabilities managed to overcome them to score very highly on the test through hard work. This was a test where every indication is that no matter what your race, you could do well.

As a result, reverse engineering the results (throwing out the test) isn't justified by a notion that the test was somehow unequal solely because you don't like who scored well. "Unequal" results occur all the time among potentially equal actors - it's quite likely, statistically, that a test can be fair to blacks and whites and have white firefighters score highest by far. If you see that a coin flips heads five times in a row, you cannot use the results to say that the coin is unfairly weighted. Perhaps the results mean that you reexamine the coin, but the results are not conclusive evidence in and of themselves - there's a very decent chance (1/32, or about 3%, probability) that you just happen to have picked 5 flips that are heads. Larry Bird's 3 point % is only a little above average for his career, but he's regarded by any fair basketball observer as one of the most lethal 3 point shooters to ever live, because of the situations in which he had to play and the way other teams played him. Results are significant only in light of the particular circumstances of the measurement, not because of the results themselves. (Side note, this is why I believe every college student should be forced to take statistics). The results of test themselves are not evidence that the test is biased; it could be a product of the particular firefighters taking the tests and how hard they're working. All other circumstantial evidence (The fact that Ricci managed to score 6th on the exam, the fact that it was created by a company that specializes in race-scrubbed test design) points away from the notion that the test is unfair, so a conclusion of an unfair test does not seem to be warranted.

Additionally, saying it's "fair" to throw out the test results because you're treating everybody equally is misleading. Yes, you are throwing out the test for everybody, but you are throwing out different results for everybody. It is no different than throwing out everybody's resumes when applying for jobs, or throwing out everybody's behavior records when applying for parole - technically, you're throwing out the exact same document, but the existence of a record doesn't imply the same inputs by every actor into what's on that record. Throwing out every resume means you're throwing out the extra effort, experience or ability of people who are more qualified or worked harder. That's completely unfair by any definition of fairness. If you're doing so because the people who happen to have scored higher are white, then that's a racial reason for doing so, and does constitute reverse racism.

As a depressing side note, it looks like the frontrunner for the new Supreme Court spot, Sonia Sotomayor, voted against the white firefighters and for the city of New Haven's decision.


No comments:

Post a Comment