Understanding Factors of Engineering Student Persistence Using Predictive Modeling

Student persistence in higher education is a topic of discussion in the academic literature and within our colleges and universities. This is especially relevant as university programs continue to focus on equity, inclusion, and support for student populations that are historically underrepresented in higher education and within specific disciplines. Engineering education has been attempting to address these issues for some time and with the graduation rates for engineering programs averaging up to 50%, understanding why students stay or leave these programs is crucial information. The reasons students persist or leave higher education programs are important data points for any university program. However, traditional statistical analysis methods may not be robust or accessible enough to understand and communicate these factors. To determine these factors, machine learning and predictive analysis software were employed to examine these factors of persistence for engineering education students. Dozens of variables including academic scores, non-cognitive and skill-based assessments, and demographic information for 300 students in an introductory engineering graphics course were used to develop a model capable of predicting whether a student will persist with nearly 94% accuracy. This research indicated that age, gender, three-dimensional modeling self-efficacy, and parental career were the most influential factors of persistence. Using this information, combined with the theoretical underpinnings of these constructs, may provide areas in which to focus and specifically target in order to improve persistence rates in engineering education.