Endogenous selection bias and gender differences in field of study choices:
Re-assessing the role of preferences and constraints based on prospective panel data
*Alexander Patzina & Carina Toussaint
Institut für Arbeitsmarkt- und Berufsforschung, IAB
Prior research found that pronounced gender differences in field of study choices exist and that a substantial part of gender variations (around 40%) can be explained by differences in vocational interests (i.e., preferences) while constraints (e.g., relative math grade, approval of field choice by peers and parents) only play a minor role (Ochsenfeld 2016). However, as previous work relies on cross-sectional data, endogenous selection bias might heavily distort earlier con- clusions. Such endogenous selection bias might emerge because in previous work preferences are measured simultaneously with the outcome and selection into studying (i.e., sample selection) could not be addressed. To evaluate the impact of endogenous selection on the explanatory power of preferences and constraints for gender differences in field of study choices we employ longitudinal data from the National Educational Panel Study (SC-4). With these data, we are able to replicate findings from earlier work. Furthermore, our preliminary results reveal that the two forms of endogenous selection under study do not distort the explanatory power of preferences. However, when considering sample selection based on propensity score matching, our results reveal that the explanatory power of constraints is particularly important for individuals that are also likely to study while they play a minor role for individuals with a low study propensity. In sum, our study advances the validity of earlier work regarding the role of preferences and adds new insight on the role of constraints.