Student factors influencing STEM subject choice in year 12: a structural equation model using PISA/LSAY data

David Jeffries, David Curtis, Lindsey Conner

Research output: Contribution to journalArticlepeer-review

4 Citations (Scopus)

Abstract

This study investigates factors that influenced the science, technology, engineering and mathematics (STEM) subject enrolment decisions of Year 12 students in Australia. Structural equation modelling (SEM) is used to develop a model using Programme for International Student Assessment (PISA) and Longitudinal Surveys of Australian Youth (LSAY) data with participating students (N = 7442) from 356 schools. An adapted version of the theory of planned behaviour (TPB), a behavioural prediction model, is used as the guiding conceptual framework. Students’ demographic background, attitudes towards science and achievement in science and mathematics at age 15 are used as predictors for subsequent enrolment in STEM subjects in Year 12. Gender, socio-economic status (SES) and immigrant status (native vs. non-native) are shown to be contributing factors. The personal value of science, enjoyment of science, self-concept in science and achievement (mathematics and science) are mediating factors in the model. These findings provide schools, policymakers and educational advisors with a greater understanding of the factors that influence Australian students’ decisions of whether to enrol in a STEM subject at Year 12. Evidence provided allows key stakeholders to take a more targeted approach to enhance STEM participation for students from varying demographic backgrounds.

Original languageEnglish
Pages (from-to)441-461
Number of pages21
JournalInternational Journal of Science and Mathematics Education
Volume18
Issue number3
Early online date17 Apr 2019
DOIs
Publication statusPublished - 1 Mar 2020

Keywords

  • STEM subject choice
  • Structural equation modelling
  • Year 12

Fingerprint Dive into the research topics of 'Student factors influencing STEM subject choice in year 12: a structural equation model using PISA/LSAY data'. Together they form a unique fingerprint.

Cite this