The underrepresentation of racial/ethnic minorities and students from low socioeconomic backgrounds in college majors that promote social mobility is problematic. Relative risk aversion theory predicts that disadvantaged students will choose college majors that promote social mobility since they are more secure educational investments. However, the theory of effectively maintained inequality predicts that privileged students, not disadvantaged students, will obtain more secure degrees. To test these theories, I utilized the NC Roots of STEM dataset to model choice of college major. The NC Roots of STEM dataset is a multivariate, longitudinal dataset that followed NC high school seniors from 2004 through 2010. This thesis utilizes a series of multilevel logit models to examine the relationship between race, SES, educational opportunities and students’ interest, odds of declaration and odds of graduation with a STEM degree. The results give evidence for both theories at work within STEM majors. Disadvantaged students, particularly Black students, are more likely to have interest in STEM majors but are the least likely to graduate in these majors, once controlling for declaring a STEM major. While SES did not appear to have much difference on STEM interest and major declaration, low-SES students were significantly less likely to graduate in STEM majors. These findings give support for effectively maintained inequality within higher education.