ABSTRACT
Despite the reversal of gender differences in educational attainment, women continue to be underrepresented in STEM (Science, Technology, Engineering, Mathematics) occupations. Yet comparative studies indicate that the male advantage in STEM fields varies across countries. To understand how these country variations come about, this study analyzes the gender gap in adolescents’ STEM expectations. While previous research mainly focused on the role of the cultural environment and the education system, this study contributes to the literature by investigating the opportunity structures of the labor market. We study how employment opportunities in science and technology, the post-industrial restructuring of labor markets in both low- and high-status occupations, and women's success in graduating from STEM fields might explain the gender gap in STEM expectations. Empirically, we analyze 15-year-old pupils’ occupational expectations from the OECD's PISA 2015 study linked with macro-level indicators in 35 EU and OECD countries by means of two-step multilevel models. Results indicate that the gender gap in STEM expectations is larger in countries with a more pronounced post-industrial restructuring of the labor market. However, rather than steering girls away from the STEM sector, post-industrial restructuring increases boys’ STEM expectations and thus seems to strengthen their gender-typical tasks preferences.
Acknowledgements
The authors would like to thank the editors and the reviewers for their helpful suggestions, which greatly improved this work. We also thank Juho Härkönen, Laura Menze, Katja Pomianowicz, and Irene Prix for constructive comments on earlier versions of this article.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 It is worth noting that the share of missing values in the dependent variable varies across countries and, to some extent, between the genders, with boys displaying a missing job expectation more often than girls. We follow McDaniel (Citation2016) and exclude students with missing expectations rather than impute their values, as, for example, Blasko, Pokropek, and Sikora (Citation2018) did. Given that the formation of an occupational expectation is a long-lasting process that stabilizes between age 14 and 18 (Gottfredson and Lapan Citation1997), imputing vague and unknown expectations assigns values to students who may not have made up their mind yet.
2 To check for multicollinearity, we estimated pairwise correlation coefficients (see online Appendix C) and variance inflation factors (VIFs), which were below 3 for all variables and the full model and thus cannot be considered problematic.
3 The effects of science and math literacy are estimated by using the mean of the ten plausible values (see McDaniel Citation2016 for a similar approach). Models, where plausible values are estimated as multiply imputed variables, do not change the results (see Model 5 LPM in online Appendix E, Table E).