|Anti-establishment Democrats outperformed establishment Democrats by an average of 1.9 points relative to their districts’ partisan leans in the 2018 U.S. House general elections|
The observed effect is relatively small at just 1.9 percentage points. 8.2 point overperformance for anti-establishment candidates (n = 60) compared to 6.3 point overperformance for establishment candidates (n = 149). Both groups significantly overperformed their districts’ partisan leans.
The effect is barely statistically significant (p = 0.049).
A lot depends on how one defines “anti-establishment”. Several fields in the dataset were used to categorize a candidate as “anti-establishment” or not:
Yeson any of the following fields indicated the candidate was anti-establishment:
Our Revolution Endorsed?,
Justice Dems Endorsed?,
Noon any of the following fields indicated the candidate was anti-establishment:
No Labels Support?
Party Support?- Political parties often throw their support behind viable candidates in competitive races (in general elections), even if the candidates don’t line up completely with the party establishment.
Obama Alum?- Originally, I had thought that a candidate who previously worked in the former president’s administration couldn’t possibly be considered anti-establishment, but after some research, I concluded that having served in an administration does not mean one is ideologically aligned with said administration.
Biden Endorsed? were ignored, and/or
Party Support? and/or
Obama Alum? were considered indicative of establishment status, the observed effect becomes even smaller (closer to 1 point) and the effect is no longer statistically significant. This illustrates that the effect is very small and highly dependent on how one categorizes a candidate as “establishment” or “anti-establishment”.
I used FiveThirtyEight’s dataset on 2018 Democratic primary candidates. This dataset includes only candidates who had primary challengers, so incumbent U.S. Representatives who did not face a primary challenge were excluded from this analysis.
Last Updated: Feb 8, 2019