Search for: All records

The underrepresentation of women in science, technology, engineering and mathematics (STEM) fields has been a subject of extensive research and policy debate. However, there is limited clarity regarding the specific mechanisms that generate these disparities, and which interventions are most effective in reducing the gap. In this study, we use causal decomposition analysis to estimate how the gender gap in STEM participation would change if we were to intervene on women’s self-efficacy beliefs in mathematics. Women tend to underestimate their abilities in math-related fields, which can affect their educational and career choices. The question we ask is to what extent the gender gap in individuals’ enrollment in STEM majors and identification with mathematics would be reduced if self-efficacy in mathematics were set to be equal across gender categories. The results suggest that equalizing this target factor will reduce the observed disparities in math identity by 53%, and in the enrollment of STEM majors by 2.5%. The modest influence of self-efficacy on enrollment disparities suggests that it is not the predominant factor. We discuss the implications of our empirical findings, as well as how causal decomposition analysis can benefit social and behavioral disparities research. 

Causal decomposition analysis is among the rapidly growing number of tools for identifying factors (“mediators”) that contribute to disparities in outcomes between social groups. An example of such mediators is college completion, which explains later health disparities between Black women and White men. The goal is to quantify how much a disparity would be reduced (or remain) if we hypothetically intervened to set the mediator distribution equal across social groups. Despite increasing interest in estimating disparity reduction and the disparity that remains, various estimation procedures are not straightforward, and researchers have scant guidance for choosing an optimal method. In this article, the authors evaluate the performance in terms of bias, variance, and coverage of three approaches that use different modeling strategies: (1) regression-based methods that impose restrictive modeling assumptions (e.g., linearity) and (2) weighting-based and (3) imputation-based methods that rely on the observed distribution of variables. The authors find a trade-off between the modeling assumptions required in the method and its performance. In terms of performance, regression-based methods operate best as long as the restrictive assumption of linearity is met. Methods relying on mediator models without imposing any modeling assumptions are sensitive to the ratio of the group-mediator association to the mediator-outcome association. These results highlight the importance of selecting an appropriate estimation procedure considering the data at hand.