In Germany, secondary school students have to choose at least one STEM subject (mathematics, biology, chemistry, and physics) for their Secondary School Leaving Examinations. In a representative sample of students in grade 13 in one federal state in Germany, we explore male and female students’ subject choices in an expectancy-value as well as dimensional comparison framework by considering prior performance, ability self-concept, and values in the chosen subject. We extend previous research by including dimensional comparisons that students make between the varying subjects they have to choose from. We discriminate between two opposing groups. One group shows a science-avoidance choice pattern by selecting only one science subject: biology (n = 439). The other group shows a science-oriented choice pattern by selecting either physics or chemistry or two STEM subjects of which one is physics or chemistry (n = 248). We measured achievement test scores, relative and absolute midterm grades, ability self-concepts, as well as attainment and utility values in chosen and non-chosen subjects and calculated logistic regressions as well as multigroup models. Our results showed that science-oriented final exam choices depended on two mechanisms. Within the expectancy-value framework, a science-oriented choice pattern was predicted by ability self-concept in mathematics for male and female students. However, attainment and utility values appeared to be irrelevant for this specific choice. Within the dimensional comparison framework, the relative mathematics-English midterm grade was relevant, but only for male students. Our findings raise the question whether male and female students should be encouraged differently in order to stay in the STEM pipeline and how structural conditions may shape pathways into or out of this pipeline.
