Thursday, August 11, 2005
Haha! You lose at statistics!
Warning!
The researchers found that "the number of AP and honors courses taken in high school bears little or no relationship to a student's later performance in college."
However, students with AP classes on their transcripts have a higher rate of admittance, said Richard Black, assistant vice chancellor of admissions and enrollment.
Oops. See if you can spot the problem. Anyway, unlike a newspaper, I can actually point you to the study. Well, a newspaper can, too, but that would be waaaaay too much effort for people whose job it is to bring news to the public. (I Googled the research associate's name, and it was the third link)
As I've warned before, trying to do statistics on current admissions practices in this way is challenging, because you're dealing with a selected sample. Despite the researchers' attempts to control for most relevant variables, it's simply not possible to control for admission. Would the people who have been rejected by the current policy but would have been accepted under some new policy (and vice versa) show these same results? You can't really tell, simply because only those who were admitted have data for "how they did in college."
I'm not saying the criticism doesn't have merit. I'm only saying that this study (and, in my opinion, most studies) should not be treated as some kind of great truth-delivery system upon which to overrule the "well, that makes sense" method of policy-making.
. . .
|
. . .
|