Wednesday, November 11, 2009

Do Even "Fairly" Drawn Congressional Districts Favor Republicans?

I had a very interesting paper make its way into my inbox today from the Political Methodology section of the American Political Science Association. Jowei Chen and Jonathan Rodden examine the inherent bias against urban -- in this case Democratic -- parties in the redistricting process. As they describe it [pdf]:
"Our central claim is that a substantial, systematic bias against the urban party does not require any intentional manipulation of maps by its opponents. On the contrary, our contention is that under political geography conditions that are quite common in industrialized societies, virtually any districting scheme that privileges compactness and contiguity will produce a bias against the urban party."
In other words, if you were to take an evenly divided state with some number of urban centers and randomly divide the state into congressional or state legislative districts -- while adhering to the court mandated principles of compactness and contiguity -- the party most representative of the urban areas would garner substantially fewer than 50% of the seats in the congressional delegation or in either state legislative chamber. If, for example, you were to take, say, Florida and its basically tied election in 2000 and just randomly draw some districts (Well, not randomly. You'd have to keep the population in each district proportionate to the other districts.), the urban party wouldn't receive 50% of the seats (to approximate 50% of the statewide vote). That party would be more likely to get between 39-42% of the seats.

And in fact, that is what Chen and Rodden have done. They took the Florida 2000 election data and simulated thousands of redistricting plans. The result? Democrats, not through any nefarious plot to pack their partisans into as small a number of districts as possible, were disadvantaged. The bias results from the fact that those higher density population centers so homogeneously Democratic, it takes more less heavily Republican districts spread out in suburban/exurban and rural areas to round out the representation. As such, a state can end up with a tie in terms of the statewide, two-party vote, but end up with the non-urban party taking a significantly higher percentage of the congressional and state legislative seats.

And yes, this assumes there wasn't an overtly partisan redistricting plan put into place in the first place. Stated differently, there wouldn't be any of Elbridge Gerry's salamanders on the map.

This one is well worth your time with a new redistricting cycle on the horizon. Read away. I'll be revisiting some of the issues discussed in this piece in future posts. It really is rich with very pertinent information.

Recent Posts:
If it's a vote on the internet, Ron Paul wins.

Gallup Poll (Nov. '09): Huckabee Continues to Garner the Most Support

40 Passes, 39 Used: What's Wrong with This Again?


Jack said...

Sigh ... I've made this observation before, and now I can't publish it!

Josh Putnam said...

Yeah, I don't know that it is a new idea -- the literature review in the paper speaks to that. I think the data and the way it was tested was what was most important here. This is a better way to test this particular hypothesis.

Robert said...

This observation seems to re-enforce the previous indication that Republicans benefited from redistricting in the South in response to the Voting Rights Act that required minority-majority districts. Equally relevant questions become how the changing demographics will affect future elections and whether a party that appeals to a predominantly white male population can obtain support among white females and multicultural populations who happen to live in suburban settings.