Model Dependency in the Analysis of Women's Electoral Success

In: Politics & gender, Band 17, Heft 2, S. 349-363

Abstract

AbstractMaciej Górecki provides a detailed and interesting discussion of our analysis regarding women's electoral success in the Polish open-list proportional representation (PR) system and the impact of a gender quota implemented in 2011. In essence, he claims that it remains an open question whether the gender quota had a 'paradoxical' effect (i.e., whether the introduction of the gender quota had a negative impact on female candidates) due to 'debatable' methodological choices in our analysis. In this article, we respond to Górecki's critique. First, we demonstrate that we do not necessarily disagree on the paradoxical impact of the gender quota but rather on the strength of the paradoxical effect. Second, we discuss some drawbacks of using the raw number of votes cast for a candidate as the dependent variable. Finally, we respond to the critique of including the ballot position in the regression model. We emphasize that even if one agrees with Górecki's critique that the ballot position is endogenous, excluding this variable leads to a severely misspecified model that does not allow one to reliably identify the effect of gender.