Essays in the economics of education and behavioral economics

This dissertation analyzes the ways that college applicants and college students navigate through higher education with a focus on ways to improve their experience. In the first chapter, I examine the ways that college applicants change their application and enrollment decisions when using the Common Application. In the second chapter, I highlight the behavioral biases that college applicants reveal in response to navigating the college applications process online. In the third chapter, I examine the ways that instructor feedback to college students affects their performance in introductory courses.

In the first chapter, I find that college applicants are highly sensitive to small changes to the costs incurred in the application process. The Common Application (CA) reduces the non-monetary costs of additional applications by allowing students to send the same application to any member school with the click of a button. Using survey data from the National Center for Education Statistics and administrative ACT microdata, I show that appliers induced to using the CA by membership changes in local public universities send 1.4 additional applications (a 39% increase) and 2.3 additional ACT score reports (a 47% increase). CA usage has no effect on the probability of enrolling in any four-year institution but increases the probability of enrolling in a CA member school by 15 percentage points. The effects of CA usage are strongest for low-income appliers, who send 2.3 additional applications and enroll in private schools and out-of-state schools at higher rates. Further estimates, while imprecise, suggest that induced CA users are more likely to enroll in more selective, higher-quality schools. The estimated changes in enrollment decisions are too large to be explained solely by small decreases in the time costs of additional applications. College-going students' enrollment decisions are sensitive to changes in application methods for behavioral reasons. Applicants are oversensitive to small, short-run costs relative to the long-run investment of college attendance and lack information about costs of attendance. Centralized application systems that lower application costs may be beneficial because applicants respond by expanding their college search, and universities are able to provide information directly through acceptance letters.

In the second chapter, I show that the proliferation of online college applications paired with centralized application services such as the CA dramatically increase the scope for choice architects to affect the decisions of college applicants. Choice architects provide the information that allows students to make choices about where to apply. I provide evidence for two settings in which students are affected by the way choices are listed for them in online settings. First, when local schools join the CA, whether or not ACT-takers expand their college search depends on the options presented by the joining school's website. If applicants can only apply to the local school through the CA or the website lists the CA option first, test-takers exposed to the CA send more than two additional ACT score reports (a 46% increase). In comparison, applicants increase score report volume by only 0.6 when exposed to the CA by a joining school that lists the CA option second (a 13% increase). Second, I show that CA-joining schools lead local ACT-takers to send an outsized number of score reports to schools at the top of the website's alphabetized search results. These findings imply that college applicants exhibit clear biases that may limit the effectiveness of their college search. The results also indicate that applicant decision-making is sensitive to the amount of information provided and how information is provided.

In the third chapter, we identify the effects of increased, personalized instructor feedback on performance in introductory college courses. We use an experimental method where poorly-performing students in large lectures are randomized to receive additional feedback through email about their course performance along with encouragements and methods for improvement. A third of students receive no such email, a third receive the email from the professor, and a third receive the email from their teaching assistant (TA). We then compare the efficacy of feedback from professors to feedback from teaching assistants. We find that neither type of feedback measurably increases course performance or perceptions of instructor quality. Both types of emails increase the frequency of help-seeking, though estimated effects are imprecise. Professor emails measurably decrease TA office hour attendance, suggesting that students view professor and TA office hours as substitutes. Our findings suggest that more intensive interventions are necessary to generate improvements in course performance and that students respond differently to feedback from professors and teaching assistants.