May 4, 2012 @ 1:28 pm
Speaking of arms races...
I’ve spent the last two days telling any student who will listen about this piece from Morning Edition, which describes an arms race between colleges, who keep raising their tuition in part to afford . . . scholarships. Not need-based scholarships, but “merit” scholarships, for those very talented students who will eventually make the colleges look good and possibly give money back to the college. So the less talented students are essentially paying more so that more talented students can pay less. They don’t like that (once you explain it to them). Even the ones receiving the scholarships are like “Really?”
But is it purely a case of robbery? This morning, a day later -- in a serious case of mixed messages – this other story described new research suggesting that most of the inventions, most of the economic productivity, most of the art, comes from a “sizable minority” of creative elites. In other words, the distribution is not centered around the mean; it’s skewed to the right. There’s a trickle-down argument brewing there somewhere, suggesting that subsidizing those elites is worth it, and this ecologist doesn't like it. Read his thoughtful post (with graphs, even!). Are people who won ten Grammys really that much more talented than people who won one or two? Or, out of that cohort of very talented people, is it luck who falls into the positive feedback loop?
And, in this clip from The Colbert Report, Jonah Lehrer says that everyone is creative, but our school system squeezes it out of us, echoing the second clip, which compares our educational system to an assembly line, where the idea that there should be an average is artificially enforced. THAT part certainly seems true, and not just on the upper end. It’s also true on the lower end, where professors routinely curve test scores to pass people who have not learned, because making everyone who has not learned repeat the course is not an option, because we’re in competition with other professors, and with other schools. Here's a good website on the issue by Stewart Rojstaczer .
There are so many causes contributing to the one effect -- performance, however you define it -- in complex, interlocking feedback loops that it's ludicrous to try draw conclusions from a single statistical measure. This is the greatest value of mathematical modeling. You have to state exactly what you think all the variables are, what the relationships between them are, and their relative importance. Genetic determinism (Jensen talks about that in his blog post above) basically says genes cause outcomes. That's a bold general statement, and clearly wrong much of the time. But saying that this phenotype is 60% heritable while this other phenotype is only 20% heritable takes a lot longer to say, and is much less memorable.
How to balance the need for simplicity and the need for complexity is a topic for another day.