PROCEEDINGS OF THE 18TH ANNUAL CONFERENCE ON RESEARCH IN UNDERGRADUATE MATHEMATICS EDUCATION
2015
Pittsburgh, Pennsylvania
Cluster analysis of STEM gender differences
Page: 793
In this project, we form, describe, and study groups (or clusters) of students based on their academic history prior to their first semester in college. These clusters allow us to examine the effect of gender on a student’s academic career decisions, such as course and major selection, while controlling for the level of preparation. We begin with an overview of standard hierarchical clustering and discuss the pitfalls of a straightforward application with our data. We then describe how to adjust the technique in order to form stable clusters. Using these clusters, we find that course selection and STEM retention are related to a student’s gender, with female students more likely to leave STEM early than male students with the same level of preparation and college grades.