– The high-school mathematics curriculum a student completed was unrelated to completing a STEM major.
– They did not find a relationship between any high-school characteristic, including percent minority, and completing a STEM major.
– Completing a STEM major depends on students’ math proficiency and certain student characteristics (e.g. gender).
– ACT mathematics score, years of high-school mathematics, high-school mathematics GPA and gender were all significant predictors of the likelihood that a student graduated with an engineering or mathematics major for a “typical” high school.
– The results provide evidence that, on average, students are equally prepared for the rigorous mathematics coursework regardless of the high-school mathematics curriculum they completed.
2012 - Student and high-school characteristics related to completing a science, technology, engineering or mathematics (STEM) major in college
Data was collected from three sources: the state, an university, and various high schools. The sample is based on data from 3459 students that graduated from a single university. The students that graduated at this university graduated from one of 229 high schools. 1166 students in the sample graduate with a STEM major and 2293 students did not.
The independent variables were student-level variables and high-school level variables. The student-level variables were ACT mathematics score, years of high-school mathematics completed (3,4,5), high school mathematics GPA, high-school percentile rank, gender, African American/Hispanic status. High school variables included percentage of African American/ Hispanic students and Asian students, aggregated ACT mathematics score, percentage of males, high-school location (urban, rural, suburban), and high-school mathematics curriculum (commercially developed, National Science Foundation developed, University of Chicago School Mathematics Project developed).
The dependent variables were completion of a STEM major and completion of an engineering or mathematics major.
The quasi-experimental design does not allow for strong casual inferences; however, many statistical control variables were included to increase the strength of the inferences.