<p>Well, there's two questions.</p>
<p>First, is Espenshade correct? I frankly don't have the math to verify or deny his findings, but 230 SAT points sounds on par with things that I can approximately verify: MCAT and LSAT gaps. 230 SAT points out of 1200 is 19%. The LSAT gap is 11 points out of 60, or 18%. The MCAT gap is approximately nine points out of 42, or 21%. So it sounds about correct.*</p>
<p>Second, assuming he's correct, is that a fair system? This obviously is a matter of value judgment, not statistical analysis.</p>
<p>In any case, whatever your views on affirmative action, it is doubtless only one component of a system that is badly broken in general. If memory serves, the Wall Street Journal once estimated that only about half of Duke's student body was admitted on academic merits -- grades, coursework, SAT, extracurriculars, etc. The other half was admitted for athletics, legacy, "development" (i.e. donations), geography, race, and yield protection.</p>
<p>It is also important to note that according to Espenshade's calculations, nearly 80% of the spots that would be vacated in a race-blind system would go to Asian American candidates, meaning that white candidates would have their admissions rates affected negligibly while Asian American candidates would be significantly more likely to gain admission.</p>
<p>Finally, he also calculates that many of the other components (legacy, athletics, etc.) do not materially affect the proportion of underrepresented minorities, implying that these are mostly white candidates taking spots away from more-qualified but still-white candidates.</p>
<p>*As I've had to discuss previously, just because a test's maximum is 180 does not mean that the test is effectively graded over a 180-point range. A blank SAT will yield a score of 400 (now 600); a blank LSAT will receive a score of 120; a blank MCAT will receive a score of 3. Hence I claim that they are out of 1200, 60, and 42, respectively.</p>