<p>
</p>
<p>Well, the US News online edition tells us what percentage of students at each school submit SAT scores. At Harvard it’s 98%, at Princeton it’s 98%, and at Yale it’s 92%. All these figures are very high, by the way. But I don’t think the numbers of students not reporting SAT scores is much of a factor at these schools.</p>
<p>But in a way you just make my point. By adding 75th% CR to 75th% M, US News concludes the 75th% CR+M is 1580 at all three schools, Harvard, Yale, and Princeton. I say preposterous because only 2,146 students in the entire country scored 1580 or higher in 2008, and if the 75th percentile at all three schools is 1580 that means half the kids in the country with 1580+ CR+M would need to be at Harvard, Yale, or Princeton, and that’s just wildly implausible. You say, not unreasonably, that many thousands scored 780+ on CR and even more thousands scored 780+ on M, so it’s plausible that the reported 75th percentile CR and M scores for these schools are right. And that’s true; I’m not for a minute doubting their reported 75th percentile CR and M scores. But look what we’re saying here: we’re looking at roughly an order of magnitude difference between the number of students who scored 790+ on EITHER CR or M, and those who scored 790+ on BOTH. In other words, roughly 90% of those who scored 790+ on one part of the test did NOT score that high on the other. Consequently, while it’s entirely plausible that Harvard, Yale and Princeton could all have 75th percentile CR scores of 790 and 75th percentile M scores of 790, it’s wildly improbable that they have 75th percentile CR+M scores (properly measured) of 1580. As everywhere, most of their 790+ CR scorers probably scored lower than 790 on M, and most of their 790+ M scorers probably scores lower than 790 on M—though probably not by a 9- or 10-to-1 margin as in the entire national sample, because they’ll get a lot of double-high scoring applicants and generally these will be favored in admissions, other things equal. Nonetheless, my best guess is that a 1580 CR+M more likely represents something like the 90th percentile at HYP—not the 75th percentile. And that’s a big difference. </p>
<p>And the distortion could be just as great at the low end: the figure US News reports as the 25th percentile SAT CR+M at HYP could actually be something like the 10th percentile. I don’t believe for a moment that a full 25% of Harvard’s 2008 freshman class had CR+M scores under 1390. That’s because the number of applicants who get in despite being in the bottom quartile in BOTH CR and M is likely to be quite small. Recruited athlete, legacy, URM, I don’t care who you are, you’ve got to have something in your record to redeem you if a school as selective as Harvard is going to admit you. They may overlook a weak CR score if you’re strong in M, or vice versa. But very few people are going to be weak in both and get admitted—a few coveted athletes and big-time development cases maybe, but certainly not 25% of the class.</p>
<p>In short, US News is wildly misleading on this whole subject. Don’t be misled. Don’t trust US News.</p>