Comment by gwern | Hacker News Reader

Comment by gwern | Hacker News Reader

Parent Full thread

I didn't think these were journalists writing; Hambrick and Meinz were involved in some of these studies, if you google for them.

> Once you're in the top 10% in terms of IQ, it's really just about hard work!

Did we read the same article? IQ still matters past the top 10%!

> The remarkable finding of their study is that, compared with the participants who were “only” in the 99.1 percentile for intellectual ability at age 12, those who were in the 99.9 percentile — the profoundly gifted — were between three and five times more likely to go on to earn a doctorate, secure a patent, publish an article in a scientific journal or publish a literary work.

That is, a 0.8% difference in percentile ranking increased the specified output by 300-500%. Now what do you think a 10% percentile ranking difference would imply?

> That is, a 0.8% difference in percentile ranking increased the specified output by 300-500%. Now what do you think a 10% percentile ranking difference would imply?

http://en.wikipedia.org/wiki/Standard_deviation

The difference in intellectual capacity between the top 10%, 1% and 0.1% (or 90, 99, 99.9 percentile) isn't necessarily proportional to the "percentile ranking difference".

I realize that, but it doesn't matter to the mistaken point the parent was making. If anything, considering the diminishing returns on IQ points or percentiles strengthens the point: by the usual argument that each percentile measures less real difference in intelligence (which I agree with), that means that the change in output is even more astounding! Going up 0.8% of the population yields you 3-500% increase in output, and this is in the part of the population where that IQ change ought to be least important. We ought to see much larger gains going from, say, 91th to 92nd percentiles, with less gains for each step upwards - but we don't.

No matter how you cut it, shifting less than a percentile and getting the measured increase is dramatic. (So dramatic that I can't help but wonder if there are non-intelligence explanations - could it all be due to Matthew or network effects?)

Reply on news.ycombinator.com