<p>A few months ago, someone (oldolddad? I’m not sure) wrote a very cogent post about estimating the effect of superscoring. I don’t remember the precise reasoning, but the bottom line of which was that it was highly unlikely that the number of superscored scores of 2250 and above was greater than the number of single-test scores of 2200 and above. Anyway, I was convinced.</p>
<p>Here’s an interesting way to look at these numbers, though. On the pre-2006 test, the number of 1600 single-test scores usually hovered around 1,000. So about 25% of those people were also getting 800 on the Writing section. On the one hand, that’s pretty high; on the other, I would have expected it to be quite a bit higher because (a) the tests don’t test anyone’s maximum capacity, so a lot of those 1600 kids might have scored higher on a tougher test, and (b) the Writing and Critical Reasoning sections aren’t completely testing different things. The number of people scoring 2390 is about twice the number scoring 1590 before.</p>