<p>My point is that if Columbia and Penn draw a higher percentage of students thru ED, then that means they’re taking the students most llikely to favor their schools out of the equation. So the empirical finding that the majority of cross-admits prefer Columbia/Penn over Duke is underdetermined because the students who are most likely to prefer Columbia and Penn are not accounted for in the cross-admit battles. So the fact the Columbia and Penn use ED more than Duke actually counts against them. In any event, the point is moot because the RP study does not seem to include ED students.</p>