This dissertation focuses on social interactions in classrooms to assess the extent that race and gender of classmates and graduate teaching assistants (TAs) affect student outcomes in behavior and performance. Two chapters rely on quasi-experimental variations for social interactions while the third relies on a randomized experiment.
The first chapter utilizes administrative data from a large California community college. I examine whether minority students are affected by the race and ethnicity of classmates. To identify social interactions, I leverage classroom level enrollment variations in racial and ethnic compositions on more than 186,000 course enrollments by 83,000 first-time students with limited class choices. The setting and richness of the data provide a robust examination of interactions with the inclusion of individual and class fixed effects. I find that Hispanic and African-American students are more likely to persist in, pass classes, and continue with the same course subject when there are more classmates of their own race and ethnicity. This is especially true for Hispanic students in courses that are transferable to the University of California and California State University systems.
The second chapter is the first-ever large-scale experiment of interactions between female and male students in an important gateway course for the Sciences. Universities around the world struggle to attract more women into STEM fields. A major concern is that female students face gender bias, discrimination, and related barriers in male-dominated STEM fields. To investigate this concern, we randomly paired every student enrolled in an introductory Chemistry lab (3,902 students and total N=5,537) over the past four years at a large public research university. Although students work very closely the entire term in the labs, we find no evidence that female students react negatively to male students. When assigned a male partner, female students do not receive lower scores or grades, and they are no more likely to drop the course, or lose interest in continuing in a STEM field. We also find no evidence that academically weaker female students are negatively affected by male students and no evidence that female students are negatively affected when paired with academically stronger male students.
The third study explores how a lack of minority role models affects the post-secondary STEM achievement gap between minority and non-minority students. I leverage a quasi-experimental setting in introductory Chemistry labs to estimate the extent the race of TAs affect student performance and decisions. Although I do not detect effects on grade performance, I detect that a racial match between students and instructors decreases the probability of dropping a course by 1.2 percentage points (on a base rate of 5.0 percent). This suggests that the race of TAs is an important component in student decisions to persist in science.
