Paper
Green, Donald P., Oliver Hyman-Metzger, Gaurav Sood, and Michelle A. Zee (2023). Journal of Political Economy Microeconomics, 1(3): 506–
Read the original paper → Opens at the publisher; use the DU library / JSTOR login if it asks for access.
In one paragraph
Washington (2008) found that legislators with daughters cast more liberal roll-call votes on women’s issues. Costa and co-authors (2019) found the pattern fading in later congresses and suggested polarisation might be the reason. Green and co-authors widen the window: eight congresses before the four Washington studied and eight after, including three not covered by Costa et al. They find no daughter effect before Washington’s period and none after it — which does not fit the polarisation story and leaves the original finding looking like a feature of one window.
Setting and question
The same natural experiment — the sex of legislators’ children — applied to a much longer run of US congressional voting. The question is whether a published effect generalises across time.
Data and identification
The identification strategy is Washington’s; what changes is the sample. The paper is useful for thinking about pre-registration, multiple comparisons, and what counts as a successful replication when the original design is sound but the estimate does not travel.
Main results
No statistically detectable daughter effect in the pre-period or the post-period; the original window remains the exception.
Why it is on the reading list
The companion to Washington (2008): same design, opposite conclusion about generality. Together they are a compact lesson in external validity and in how the literature corrects itself.
No interactive explainer for this paper yet. If you build one, see Contribute to get it added.