Using Logistic Regression Analysis to Predict Student STEM Majors

• In Event: 47.078 - Longitudinal Studies of STEM Interests, Choices, and Outcomes

Abstract

Logistic regression was used to develop a model using high school factors to predict a student's decision to major in a STEM field at the undergraduate level. Data consisted of 8,366 student responses from the High School Longitudinal Study of 2009. The final regression model included seven statistically significant predictors of STEM group membership: student's sex, student's math proficiency, student's self-rating of his/her math identity, science identity, math self-efficacy, science self-efficacy and level of school belonging. The developed regression model was able to correctly predict future student STEM majors in 79% of the cases examined. The model had a specificity of 93% and a sensitivity of 39%. The strongest predictors were student's sex, math proficiency and science identity.

Authors

• Christina Witta Gillespie, Valencia College

• Haiyan Bai, University of Central Florida