The Supreme Court may have gotten something right

Today, the Supreme Court handed down a decision in a discrimination case where some city in Conn. had administered a written test for promotion in its fire department and then, when the results failed to qualify enough minority fire fighters, decided to toss out the test results. The Supremes said 5-4 that you can't do that.

I caution that I have not read the decision.

But it strikes me from what I have read that it makes a certain amount of sense to say that you can't discriminate solely on the basis of race, which seems to be what the city did. They got back test results and discovered that only one Hispanic and no African Americans had passed the test. If, on that basis ALONE, they threw out the test, then the Supreme Court is probably correct. The reason I think the Sc might have gotten that one right is because the city failed to probe further into why the test results were so unequal; they just assumed and threw out the test. This means they threw out the test results on the basis of RACE ALONE; not because of differences in test administration or some provable bias in the test itself. Moreover their argument seem to be that the unequal results prove the unfair bias, almost arguing that using a written test biases the results. That argument assumes the desired result; a species of post hoc ergo proctor hoc logical fallacy. I have, and I suspect the majority of Justices would have, no problem tossing out the test if the city had bothered to establish HOW the test was discriminatory beyond the fact that it achieved a politically undesirable result.

At the end of the day, the law must be applied equally as well; and had the result been that not enough Caucasians qualified and the city then tried to throw out the results on that basis, I think quite clearly this lawsuit would have gone against the city. Moreover, I think the question begged here is why are we doing this kind of analysis anyway? Are we to decide the validity of any (and every) test or hiring result based on whether or not the results from that activity exactly match the racial distributions of the population? Oddly enough, random chance rarely produces results exactly on the numbers; it almost always produces some variation and must be allowed to generate results at the ends of the bell curve some time. In this case, you could achieve randomization by randomly selecting names form a hat for promotion or by going strictly on seniority, but those systems would not tend to necessarily put the most competent candidate in the position. Still, if the goal is "fairness" over competence, that would work.