A Research Note on the pitfalls of sample selection and what (not) to control for in gender wage gap estimations

Estimating discriminatory gaps is one of the core topics of Sociology. A large number of studies use observational data to estimate these gaps, typically by controlling for potentially confounding factors and interpreting the coefficient of the group variable (e.g. female) as the degree of discrimination. However, estimations in this manner require choices both with regards to sample selection as well as the selection of variables to control for the stages where researchers assume discrimination to occur. Against the backdrop of the literature on gender wage gaps, this research note discusses these choices, their impact on the interpretation of the results and provides recommendations for applied researchers on how to deal with these choices in practice.