Politicians who are the parents of daughters provide researchers with a “natural experiment” to test how personal circumstances influence legislators’ roll call votes. Research published in 2008, by author Ebonya Washington, showed that members of the U.S. House of Representatives with daughters cast more liberal votes on women’s issues during the period between 1997 and 2005. But another study published in 2019 by researchers Mia Costa et al., covering the period between 2007 and 2017, did not find such an effect.
In their new paper “Revisiting a Natural Experiment: Do Legislators With Daughters Vote More Liberally on Women’s Issues?” published in the Journal of Political Economy Microeconomics, Donald P. Green, Oliver Hyman-Metzger, Gaurav Sood, and Michelle A. Zee set out to investigate this question further with additional data. Using a larger dataset than those analyzed in the previous studies, they find that having daughters has no effect on how legislators vote on women’s issues.
The authors start their analysis in 1981, prior to the analysis presented in the 2008 paper by Washington, and in a time when, they posit, political polarization was “relatively tame by contemporary standards.” They end their analysis after the presentation of the 2019 paper by Costa et al. Subsequently, the researchers determine the gender proportions of the children of the lawmakers serving in these 20 total congresses, via archives and congressional offices.
From the voting record, Green, Hyman-Metzger, Sood, and Zee determined which roll call votes during each Congress concerned women’s issues, and coded each legislator’s voting record according to whether it was pro-feminist. They also tracked specific legislators’ votes over time, in order to study if the specific roll call votes held in certain Congresses influenced the results of the previous work.
The authors find that, across sessions of Congress, having daughters has little effect on how members voted on women’s issues. This finding holds over time. “If party polarization dampens the daughters effect, going backward in time should increase the apparent average treatment effect,” the authors say. “This prediction is not borne out,” they write – the rise of polarization does not change this pattern.
The researchers note that they successfully replicated the results obtained in the 2008 paper by Washington and the 2019 paper by Costa et al. “The only era during which the daughters effect is positive is the one that Washington happened to study,” they say. “The daughters effect found by Washington is a thought-provoking empirical result that seems to demonstrate that roll call votes are influenced by legislators’ personal circumstances and experiences,” the authors write, while the lack of such an effect reported by Costa et al. indicate that perhaps partisanship has taken precedence over family influences as political polarization has increased. “The findings presented here seem to support a more mundane interpretation,” the authors conclude: “daughters do not seem to have any appreciable effect on legislators’ roll call votes.”