(Welcome, New York Times readers!)
Thanks to commenters on this topic. Your feedback has shaped my thinking on this subject. I recall being skeptical that redistricting could have a major effect. As it turns out, the effects of partisan redistricting helped Republicans far more than I expected.
One reason for my skepticism is that the the effect was clustered tightly in a handful of swing states. My pre-election calculations did not look for state-specific effects (though they were still fairly accurate); it was only after the election that I developed the right statistical tools. All extreme partisan gerrymanders were done in states with GOP-controlled redistricting. Furthermore, they are swing states, putting them on a knife edge and making them places where gerrymandering could help eke out extra wins.
First, some links to previous essays. Then some answers to your questions.
October 4, 2012, “The Very Hungry Gerrymander“: I showed how partisan redistricting tilted the overall playing field for House elections. At the time, I estimated that the effect was equivalent to 13 seats (not too far off) and less than 2% of the popular vote (this was a serious underestimate).
December 30, 2012, “Gerrymanders (Part 1): Busting the ‘both-sides-do-it’ myth“: Here I figured out a general approach to identify specific offending states. The result: seven Republican-controlled gerrymanders, one Democratic-controlled.
January 2, 2012, “Gerrymanders (Part 2): How many voters were disenfranchised?“ Here I estimated the effective disenfranchisement of millions of Democratic voters, hundreds of thousands of GOP voters. There are different ways to do it, but the ratio is always similar.
Some questions with my replies:
It seems that an obvious solution is proportional representation: dividing up each state delegation by the percentage of the vote, and letting the political parties put up slates of candidates.
The multidistrict idea has logical appeal. However, it’s never been done in the US. People like being able to contact their own Congressman. In my view this proposal is a nonstarter. From the point of view of implementing actual change, it is impractical and a pipe dream to suggest it.
However, this does raise a logical issue, which is that clustering of voters is sometimes deemed to be good. The Voting Rights Act as implemented allows the implementation of things like “majority-minority” districts, where a small minority like blacks or Hispanics are packed together. In cases where a minority is far smaller than 50%, this is necessary.
The general principle is proportionality: If you’re 10-30% of the population (minority groups), packing is your only hope. If you’re 40-50% of the population, than packing can leave you out in the cold. I agree that an explicit proportional system addresses this problem. I just think it’s unavailable as a tool given our current legal framework. Of course, that framework could change someday.
Political scientists have claimed that the inherent clustering of Democrats in cities has led to a significant partisan imbalance that is structural (“districting”) rather than redistricting. Is that true?
This claim, made by political scientists, might have been true in past elections at a national level. But state-by-state, and this year nationwide thanks to partisan asymmetry, it is false. I note with emphasis that this essay by Stuart Rothenburg is incorrect.
(1) Here is another demonstration. This graph shows what fraction of the two-party vote would have been needed for Democrats to control the House of Representatives.
The procedure was:
- Calculate the % two-party vote for all 435 districts.
- Calculate the shift in vote needed to make an outcome of exactly 218 Democratic seats.
- Add this shift to the national % Democratic vote.
The colored horizontal line segments indicate which party was in control. Generally, the out-party needs a bit more than 50% of the two-party vote to gain control. This extra barrier is an advantage for the incumbent party. The 2012 value is unusually high – though interestingly, it’s matched by 2004.
I note that dealing with uncontested races is a challenge. For instance, the 2006 data point is distorted by the fact that there were 47 uncontested races won by Democrats (versus only 10 won by Republicans). Forty-seven is an unusually high number. With other definitions, this data point is more comparable to 1996-2004.
(2) Here is another demonstration. This is an example of the resampling analysis I used for the NYT piece.
Green points indicate resampling from all districts (minus at-large and gerrymandered states). Black points indicate resampling from the same districts, but furthermore removing high-density and urbanized states. As you can see, the green and black clouds are basically in line. The red circle is a the actual seat breakdown of seven strong GOP gerrymanders: PA, OH, MI, NC, VA, FL, and IN.
Two conclusions here: (1) urbanized states and all states behave similarly when resampled, and (2) the effect of partisan gerrymandering dwarfs whatever variations are present at a national level.
I’ll post a more questions as they arise.