Malcolm Gladwell's most recent New Yorker article uses the example of pit bulls to argue that racial profiling doesn't work, and that there are better methods of screening out troublemakers.
Steve Sailer then points out the irony in Gladwell's article...that its message of trying to look beyond first impressions seems to directly contradict Gladwell's own bestseller, Blink, which argues for the value of thin slicing and rapid decisionmaking.
He then argues that opposing racial profiling is a matter of political correctness, not science, given the clear evidence of different crime rates for different races:
"In total, blacks had the highest incarceration rate at 7.2 times the
[nonHispanic] white rate, followed by Hispanics, at 2.9 times the white rate.
[American] Indians and Pacific Islanders were imprisoned at about twice the
white rate, and Asians at only 22 percent of the white rate."
In other words, Asian-Americans are incarcerated only 1/33rd as much as blacks. So, if you are cruising in a squad car, which group of teenagers on the corner are you going to pay more attention to: blacks or Asians?
Far be it from me to be politically correct, but it seems to me that both stances are exaggerations, and that we can learn more from a moderate position.
Racial profiling does work. Even if we assume that police racism is epidemic, and that half of all black men are unjustly imprisoned (I'm not arguing this, just assuming it to prove a point, so please don't start flaming me yet), it is still the case that blacks are correctly incarcerated 16 times as often as Asian-Americans.
And if incarceration is in fact a good proxy for lawbreaking, that means that the probability that a given group of black teenagers is 16 times as likely to be trouble as the corresponding group of Asian-American teenagers.
However, this analysis ignores two things: the absolute rate of incarceration, and the costs of improper profiling.
According to the Department of Justice, in 2002, 0.6% of white men were behind bars, versus 5% of black men (presumably this means that about 0.11% of Asian-American men were behind bars).
Even with a worst case scenario assumption that 5% of black men are criminals (again, I don't believe this, I'm just assuming this for the sake of argument), racial profiling would misidentify law-abiding citizens as criminals 95% of the time.
In other words, for every time that racial profiling correctly fingers the black guy, there are 19 cases where an innocent man gets hassled.
As a society, we should decide whether or not the benefits of racial profiling are worth that price. I don't know what the answer should be, but I do know that simply arguing about the validity of racial profiling is a waste of time.
A tip of the hat to Ben Casnocha for alerting me to Steve's article.