Virtually the same: Using Bayesian methods to investigate the relationships between online course delivery and postsecondary student enrollment, course outcomes, and degree attainment
Skinner, Benjamin Thomas
Over the past number of years, postsecondary students have increasingly enrolled in online courses. In this dissertation, I use human capital theory and Bayesian statistical analyses to investigate how students choose these courses, whether students perform as well in them as they do in face-to-face courses, and what the long-term degree outcomes among online course-takers may be. I first examine a potential mechanism of a student’s choice to take online courses: access to high speed broadband. In preferred specifications, I find that every tier increase in download speed is associated with a 41% to 56% average increase in the number of students who enroll in at least one online course. I next estimate the effect of online delivery format on course outcomes among students enrolled in the University System of Georgia, finding that those in online courses were 2.8 percentage points more likely to withdraw and, conditional on completing the course, 5.4 percentage points less likely to pass with a C- or better. Finally, I produce state-level estimates of differences in aggregate degree outcomes between sometimes-online and never-online students using survey data that are only nationally representative with a technique from political science, Bayesian multilevel regression with poststratification (BMRP). On average, I find that never-online students were slightly more likely to earn a bachelor’s degree in six years, while students who took any online courses were slightly more likely to earn an associate degree in three years. Results vary considerably across the states, however, justifying the use of the method. My dissertation contributes to the education policy literature both through its empirical findings and its use of Bayesian methodologies. With each question, I offer new insights into postsecondary online courses that support a better understanding of how these courses may fit into students’ human capital enrollment decisions.