Low enrollment and high attrition of women in science, technology, engineering, and mathematics (STEM) continues to be an issue for postsecondary institutions. Improvements in representation of women has been seen in some of the agricultural and biological sciences; however, in many of the more math intensive areas such as geosciences, engineering, mathematics/computer science, and physical sciences (GEMP), women continue to be underrepresented leading to underrepresentation in the workforce and further exacerbating gender gaps. Studies suggest the lack of representation is not due to a gap in math ability between men and women, yet underrepresentation remains predominantly within math intensive STEM areas, suggesting something like math self-efficacy (MSE) and math self-concept (MSC) may be impacting enrollment and retention. The research presented here investigates the link between enrollment in GEMP STEM and retention in STEM with the factors of MSE, MSC, and gender. Structural equation modeling (SEM) with Bayesian estimation is used incorporating additional factors from previous research. Study results indicated that MSE and male were both positive and significant indicators of enrollment in GEMP STEM and retention in STEM. MSC was not a significant indicator of retention in STEM but was shown to be significant for GEMP STEM enrollment; however, it was negatively associated with GEMP STEM when combined with MSE. Several program related factors were also shown to be significant indicators of GEMP STEM enrollment and STEM retention. This study highlights the importance of MSE and gender for enrollment and retention and should encourage future efforts towards improving MSE as a possible method of increasing representation of women in underrepresented areas of STEM.

This study explored the factors that influence public high school students' advanced math course-taking. The factors investigated were parental involvement, peers' educational aspirations, students' own educational aspirations, and math self-concept. These factors were further examined for students in different settings as defined by school demographic variables of urbanicity, minority concentration, and poverty concentration. The study analyzed longitudinal data from the National Education Longitudinal Study of 1988 (NELS: 88), using structural equation modeling. Results indicated that parental involvement was much more important than peer influence for students' educational aspirations, and in turn, for their advanced level mathematics course-taking. Parental involvement had a larger effect for students in high-minority, high-poverty urban schools, who, on the average, had taken the smallest number of advanced mathematics courses, compared to students in other settings. Results from the study indicated that African-American students' math self-concepts were not affected by their previous math achievement, suggesting the lack of feedback about their mathematics performance. Recommendations based on the findings included improving parental involvement for all students, especially for students in high-minority, high-poverty urban schools, and providing more feedback to African-American students about their level of performance in mathematics and its consequences in terms of advanced math course-taking. / Ph. D.