Preparing for the STEM Pathways? Dual Enrollment and College Major Choice in STEM

Abstract

ABSTRACTGuided by the STEM pathway model, our study hypothesizes that dual enrollment can serve as an effective strategy to improve and equalize college students’ access to STEM programs. We analyzed a nationally representative dataset to disaggregate the influence of dual enrollment course-taking (i.e. participation, dual credits in Math/Science, number of dual credits) on students’ STEM major selection, with a focus on traditionally underrepresented students in STEM. We found that taking dual enrollment courses in general is positively associated with the probability of majoring in STEM, especially at the baccalaureate level. However, taking dual enrollment courses in Math/Science is not associated with the probability of majoring in STEM when compared with students with no dual enrollment courses in Math/Science. The relationship between dual enrollment course-taking and STEM outcomes varies across different student background groups: It is consistently positive for students of higher household income to major in STEM but not statistically significant for low-income students. We discussed practical implications and future research with a focus on the role dual enrollment plays in advancing postsecondary STEM access.KEYWORDS: Dual enrollmentSTEMcollege major choicesociodemographic backgroundHigh School Longitudinal Study Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1. Specifically, the multiple imputation procedure addressed missing values in the control variables: race/ethnicity (3.8%), SES (6.8%), high school GPA (7.4%), educational expectations (7.5%), highest math level (7.2%), the number of AP/IB credits (4.4%), and intent to major in STEM (6.9%), respectively.2. Because the HSLS data does not properly report gender identity (Marine, Citation2011), we use students’ sex assigned at birth as a binary indicator to create the subgroups.3. Due to the relatively small and unbalanced number of treated observations in each racial/ethnic group, we were unable to conduct subgroup analyses using propensity score models based on individual racial/ethnic groups. Thus, we had to aggregate racially minoritized groups based on their student-of-color status. We first defined students of color as students who identify as non-White. In additional analyses, we excluded Asian American students from the sub-sample given their relatively high performance in STEM fields (National Center of Educational Statistics [NCES], Citation2022).4. The readers should be cautioned that, due to the treatment transformation, it is challenging to interpret the coefficients in the context of DE credits earned. Additionally, these commands are not supported by the svy prefix to account for complex survey data, the dose-response model only provides suggestive evidence, without generalization to the broader population. Given these methodological challenges, we present the dose-response findings as supplemental analyses in Appendix D.Additional informationFundingThis research was supported by a grant from the American Educational Research Association which receives funds for its “AERA Grants Program” from the National Science Foundation under NSF award NSF-DRL #1749275. Opinions reflect those of the author and do not necessarily reflect those AERA or NSF.