The rising cost of college and demands for accountability have increased interest in providing students with information about the earnings of college graduates by school and major. However, no consensus exists over how to display that information in a way that is most beneficial for students. Some researchers advocate displaying median earnings only, while others advocate showing more detail on the variation in earnings. We argue that an explicit theory of student choice is missing from discussions about the provision of earnings data. We use a survey experiment to assess two models of student choice: one in which students use median earnings, and one in which students use median earnings and earnings variation. We demonstrate that showing respondents the median and variation leads to large and significantly different expectations in earnings and different choices in majors, compared to respondents who see the median only. The results question the use of medians only as a tool to improve student decision making. In contrast, displaying medians and variation provides influential information that allows students to make educational choices that incorporate risk.