Espenshade’s paper on admissions preferences, which I just skimmed, is 21 years old and the latest cohort data used is 1997 although he includes earlier years as well. The aggregate admit rates in that paper are about 30% again confirming this from another time and likely not applicable today. I’ll readily grant that legacy preference was a more significant thing in the past and may be a slight benefit now when choosing amongst kids who share the same demographic but compared to the class as a whole, they have stronger applications accounting for the vast majority of their higher admit rate. As mathmom states above, Harvard, which seems to have stronger legacy preference than other Ivy’s, shows higher objective standardized test scores in legacy vs. the average applicant. There are some schools like MIT that state they give no legacy preference but I think it is safe to say most schools want their alumni to think there is at least some legacy preference for fundraising and I’m sure those admits yield at higher rates which helps with selectivity statistics and both those outcomes are desirable at an institutional level.