Poor performance in introductory courses and lack of individualized assistance may contribute to college non-completion. This research aims to identify the effects of increased, personalized instructor feedback on performance in introductory college courses. We conduct an experiment in which poorly-performing students in large lectures are randomized to receive individualized communication through email about their course performance along with a reminder of their instructors’ out-of-class availability. Half of the treated students receive an email from the professor, while the other half receive an email from their teaching assistant. We compare the efficacy of outreach from professors to outreach from teaching assistants. We find that neither treatment measurably increases course performance or perception of instructor quality. Emails from professors decrease attendance at TA office hours, suggesting that students view professor and TA office hours as substitutes. Both types of emails increase the frequency at which students seek help from the email’s sender, but do not increase the frequency at which students seek help from the other, non-sending instructor, even though all emails contain office hour information for both the professor and the TA. Thus, changing the identity of the sender changes the effects of the nudge, which has implications for the scale-up of nudging programs. If messages come from a source that is impersonal or unknown to the recipient, the nudges may be less effective than those in a small-scale study where sender and recipient have a closer relationship.

This paper investigates the connection between a student’s choice of which college to attend and whether or not the student graduates in a STEM field. I use data from the ELS:2002 to estimate a multilevel choice model in which students choose whether to attend college, and whether or not to major in STEM. I find that the effect of college quality on the probability of STEM graduation is decreasing in student academic preparation. Less prepared students experience large gains in the likelihood of completing a STEM major from attending high quality schools; the likelihood that a more prepared student majors in STEM is relatively unaffected by the quality of the college that the student attends.

This study evaluates enrollment in and outcomes from online classes. Most studies that evaluate causal impacts of online college classes are randomized trials. These studies largely find that students taking online sections of a course perform worse than their peers in face-to-face sections, but do not consider the possibility that without the option of an online section some students may choose not to take a course at all instead of taking it face-to-face. I use a regression discontinuity design with waitlist data from the University of Iowa to estimate the effects of online classes in a way that accounts for this possibility. I find that for students on a waitlist for an online section, being offered a seat in that section greatly increases the likelihood that students will take the course at all and also increases the number of related courses that students take. Any negative effect that online sections have relative to face-to-face on these related future courses is small enough to be outweighed by the effect of access to the course. Access to the waitlisted online section has at most a small, positive impact on the number of credits that students take that session, and in most cases does not impact how quickly they graduate. Being admitted to an online section increases opportunities to take the course of interest and has a modest influence on the discipline of future courses.