Empirical challenges in assessing the “leaky STEM pipeline”: how the research design affects the measurement of women’s underrepresentation in STEM

Abstract Background Despite a well-documented underrepresentation of women in STEM majors and occupations, empirical evidence on whether there really is a “leaky STEM pipeline” is mixed due to a lack of methodological consistency. Studies vary by (1) the definition of STEM, (2) the decision to measure choices alone or to also include aspirations, and (3) the use of longitudinal or cross-sectional data. Results In order to analyze how variations in the research design affect the measurement of women’s underrepresentation in the field of STEM, we critically reviewed relevant literature on the “leaky pipeline” and identified three central features in the designs of existing empirical studies. We illustrate how the variation of these affects the results by applying them to the German context. Our results support the “leakage” perspective for Germany only during the transition to the labor market. Changes in STEM aspirations between grades 9 and 12 do not follow a clear pattern. Indeed, a comparison of grade 12 aspirations to actual college major choices even shows an increasing share of women in STEM. Conclusion Germany does not exhibit a significant “leaky STEM pipeline”. Due to more men choosing STEM, gender gaps widen in higher education, while the pipeline remains stable. Therefore, we challenge the “leaky pipeline” metaphor, advocating a life-course perspective to better understand STEM trajectories. We call for refined measurement standards, emphasizing official STEM definitions, long-term observations from aspirations to career entry, and the use of longitudinal data.