3 ms·
...is it really a fallacy to say that black people are better sprinters? Yes. Data on the Olympics gives very little information on whether the average black p
by yummyfajitas 15y ago
...is it really a fallacy to say that black people are better sprinters?
Yes. Data on the Olympics gives very little information on whether the average black person is a better sprinter than the average white person.
It is not, however, a fallacy to say that it is highly likely that the next top sprinter will be black. (If anyone disagrees with this claim, I'm willing to offer good odds on the race of the next Olympic sprinter medalists.)
Honestly, I'm finding the author's exposition to be rather confused. He seems to conflate two logically distinct propositions ("the best $x come from group $Y" and "a random $y in $Y has a higher probability of outperforming a random $z in $Z") for reasons I can't fathom.
- sambe 15y ago...and in fact I believe that Freakonomics suggests that black athletic ability is more about variance than average.
- nkurz 15y agoI agree with you on "very little information", but I wouldn't want to make the jump to "fallacy". It would be a fallacy to presume that the between-group difference is anything approaching the within-group difference, but I'd guess that the effect is there. Instinctively, I wouldn't be surprised if genetics played at least a 1% role. Would you? What if instead of "black", you pointed to a particular gene? The alternative would be to presume that the variance of sprinting ability with the black population is greater than that of the white. I'd want to see some reason that this is true before presuming it. This also gets us back to the furor regarding Lawrence Summers dismissal from Harvard, which does imply that there some things that are simply "wrong to note".
- yummyfajitas 15y agoIt would not surprise me at all if genetics plays a large role. The only point I'm making is about inferring information about the mean/median of a distribution by observing only it's far tails. In fact, I'd speculate that across the US population, black sprinting ability is probably lower than that of whites. Black women are disproportionately overweight/obese and are likely to lower the average for blacks. Looking solely at olympic level data points would never reveal this. (As far as I'm aware, this latter effect is minimally related to genetics.)
- nkurz 15y agoThe only point I'm making is about inferring information about the mean/median of a distribution by observing only it's far tails. That's the same point I'm making, but I come to the opposite conclusion. I think that in the absence of contradictory information one should prefer simple inferences to complex ones (Occam rephrased). Thus I'd look first to the mean, then to the variance, and only then consider that only one of the groups has a multi-modal distribution. I do think our "overcompensation" for race affects our conclusions here (mine included). If one were to find that the winners shared some other trait other than skin color, I don't think there would be nearly the same desire to find alternative explanations. But this doesn't mean that these alternatives are not useful to consider, and quite possibly correct.