From Ralph Stinebrickner and Todd Stinebrickner:
We find that students enter college as open to a major in science as to any other major, but that relatively few students finish school with science as their outcome. This occurs because, relative to other majors, students are both more likely to leave science (if they started in science) and are less likely to change into science (if they started in a major other than science). In terms of major-specific factors that influence the major decision, we find a particularly important role for future grade performance, with future income playing a statistically significant, but smaller role. As to why students leave science, departing students have typically learned that their future grade performance in science would be substantially worse than expected, with this reflecting learning about ability rather than learning about their willingness to study. As to why students do not often change into science, students who do not start in science typically do not learn that they are especially talented in science.
These patterns are consistent with our findings when we use transcript data to examine the process by which students update their beliefs about future major-specific grade performance. We find that taking courses in science is the primary way to learn about one’s ability in science, with science being an outlier in that grades in courses taken outside of the major play a relatively uninformative role in what one learns about his ability in the major. It is possible that requiring additional courses in science during college might lead to more science graduates. However, simulations of our model show that, when we replace each student’s beliefs at entrance about major-specific average grade performance (from survey data) with a distribution representing true major-specific average grade performance (computed from transcript data), students’ beliefs about the probability of choosing science become consistent with the relatively low proportion of students who actually choose science. This suggests that, by and large, students are ultimately choosing science in numbers that are roughly consistent with their abilities at the time of college entrance. As such, if more science graduates are desired, the findings suggest the importance of policies at younger ages that lead students to enter college better prepared to study science.