Reading the study (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2763720) I'm amazed it did not consider parental income. Not only does parental income correlate strongly to child income as adults, but also to college acceptance. Most importantly in this case, parental income is usually used to pay fraternity dues.

Fraternities might give a strong sample set of students with poor parents being separated from students who have parents with moderate to high incomes. Not to mention the obvious cultural divisions that might keep a high percentage of low income students from being included in many fraternities.