In a study of a large state university, students were randomly assigned, roommates. Researchers found that, on average, males assigned to roommates who reported drinking alcohol in the year before entering college had GPAs one-quarter point lower than those assigned to non-drinking roommates. For males who drank frequently before college, being assigned to a roommate who also drank frequently before college reduced their GPAs by two-thirds of a point.
The marginal social benefit curve will be:
