Among candidates for U.S. House seats in the 2018 general elections:
Emily-endorsed Democratic women meaningfully overperformed other members of their party, and
Susan-endorsed Republican women well underperformed other members of their party.
Democrats overperformed (held higher voteshare margins) and Republicans underperformed (held lower voteshare margins) relative to the partisan leans of the districts they ran in.
xare women candidates endorsed by the named organizations
oare all other Democrats (blue) and Republicans (red).
Neither dataset I used has candidates’ genders, so I compared Democratic candidates endorsed by Emily’s List (
Emily Endorsed?) and Republican candidates endorsed by the Susan B. Anthony List (
Susan B. Anthony Endorsed?). Both organizations endorse only women, but not all women from their preferred parties, so this method is only a very approximate proxy for candidate gender. Plenty of women candidates were endorsed by neither organization.
|All Republican candidates||-9.9%||181|
Susan-endorsed Republicans underperformed by -5% (p = 0.018) - compared to Republicans not endorsed by the organization
|All Democratic candidates||+6.3%||176|
Emily-endorsed Democrats overperformed by +3.5% (p = 0.004) - compared to Democrats not endorsed by the organization
I used FiveThirtyEight’s dataset on 2018 primary candidates and only included candidates who won their primaries and advanced to their respective general elections. This dataset includes only candidates who had primary challengers. Moreover, I only looked at candidates for U.S. Representative/House seats (no U.S. Senators or state-level candidates).
As the above findings could be caused by several factors, and correlation does not equal causation, one should not draw the conclusion that a
Susan endorsement tanked a candidate or that an
Emily endorsement helped a candidate to victory. It could also be, for example, that
Susan made “riskier” endorsements, perhaps endorsing candidates that were very likely to lose, while
Emily made safer endorsements. But still, this is not a conclusion that can be drawn directly from this analysis.
Last Updated: Feb 16, 2019