Postsecondary Stem Paths of High-achieving Students in Math and Science: A Longitudinal Multilevel Investigation of Their Selection and Persistence

Soohyun Yi, Purdue University

Abstract

This study used a quantitative approach to investigate high-school students’ talent-development pathways in STEM from 10th through 12th grades and for 8 years thereafter. The purpose of this study was to longitudinally investigate three important choices and accomplishments on the STEM talent development trajectory: a) selecting a STEM major in college; b) persisting with the STEM major until graduation; and c) selecting a career in STEM after college graduation. Given that students with gifts and talents are more likely to persist and succeed in STEM fields than average achievers, and understanding their unique needs may be the first important task to promote their talent and career development, this study concentrated on college bound high school students who achieve at high levels in math and science. I operationally defined students identified as high-achieving in math and science as those who scored in the 95th percentile or above in math or science in college entrance exams. Through an investigation, I used the longitudinal data of the Education Longitudinal Study of 2002 (ELS:2002) of a nationally-representative cohort of U.S. students. Two inferential analytic methods were used to estimate the probabilities associated with each binary outcome variable: multilevel logistic regression model and discrete-time hazard model. Students identified as high-achieving by the criteria of this study were more likely than students who did not meet the criteria to enter postsecondary STEM education and to persist in STEM after college graduation. However, there were severe disproportions in the numbers of students identified as college bound high-achievers. Female, Black, Hispanic, Native American, and other-race students, students from families of lower-quartile SES, and students who attended schools with higher levels of academic pressure were less likely to be identified as high-achievers than students in the corresponding reference groups. Mathematics self-efficacy and advanced courses in math and science, as moderators, increased the probabilities of STEM entrance, regardless of the identification as high-achieving. In terms of STEM persistence and graduation, fewer Black, Hispanic, Native American, and other race students graduated from college with a STEM major compared to White and Asian students. The disparities in the probabilities of further persistence also existed by student- and school-level covariates. Unlike prior studies in STEM education, I controlled for the effects of high achievement in college entrance exams, thus, the results revealed the effects of some covariates were unique for students identified as high-achieving. Based on the baseline estimates of probabilities provided by this study, more research needs to be conducted to investigate reasons for the significant effects promoting or preventing desirable outcomes on STEM pathways.

Degree

Ph.D.

Advisors

Gentry, Purdue University.

Subject Area

Gifted Education|Educational psychology

COinS