Predicting Undergraduate Passions: An Analysis of Major Migration at Carnegie Mellon
With six undergraduate colleges, Carnegie Mellon University offers a plethora of choices for primary and additional majors, as well as interdisciplinary programs. Such variety has the potential to assist or hinder students’ undergraduate careers, dependent upon how often their interests change, and how quickly they find their fit at CMU. This paper discusses the potential reasons behind major migration at CMU, and whether or not we are able to predict the likelihood that a given student will switch their major before graduation. By looking at the Class of 2015 cohort over twelve semesters, we run linear discriminant analysis and logistic regression on a representative sample of 1136 students. Through these methods, we determine that predicting student migration yields high error rates when working with a small subset of binary demographic variables, yet there is potential for a stronger prediction algorithm with more data and more robust variables such as cumulative QPA and socio-economic status. We also focus on the predictive capability of Dietrich College first-year survey data, and find that additional variables such as the number of interests incoming students have and which interests become their graduating majors are significant in classifying potential migrators. In both cohort and survey data, we find that the time it takes to initially declare is an important variable in determining whether or not a student will switch majors. The likelihood of switching is greatest if one initially declares in the spring of their first year, and drops off thereafter. This suggests that advisors should work closely with students to determine if it is the right time to declare. Declaring early can give students access to classes within the primary department, and declaring too late could leave students stuck in a major they are not truly passionate about.