First, if that number could be read directly as a refutation of my hypothesis, it would be tied, rather than showing an advantage to the legacy students.
Now, a few guesses: first, to the extent that affirmative action is even partially successful in targeting disadvantaged groups, affirmative action students should be less likely to be legacy, meaning that applications of legacy students should at least reflect a numerical superiority consistent with the strength of the affirmative action bias. Second, I will say that from personal experience I valued, more highly than was deserved, the schools that my parents attended; in this way those schools were the beneficiaries of highly effective advertising, through my parents, that made me more focused on those particular schools than they deserved (so much so that two of the three were two of the four schools I intended to apply to.) The aggregate outcome of that effect across all legacy applicants should be a higher-than-usual applicant quality in the pool of legacy students.
Now, a few guesses: first, to the extent that affirmative action is even partially successful in targeting disadvantaged groups, affirmative action students should be less likely to be legacy, meaning that applications of legacy students should at least reflect a numerical superiority consistent with the strength of the affirmative action bias. Second, I will say that from personal experience I valued, more highly than was deserved, the schools that my parents attended; in this way those schools were the beneficiaries of highly effective advertising, through my parents, that made me more focused on those particular schools than they deserved (so much so that two of the three were two of the four schools I intended to apply to.) The aggregate outcome of that effect across all legacy applicants should be a higher-than-usual applicant quality in the pool of legacy students.