How Would Changing Political Sentiment Change Congressional Seats?
​
June 25, 2017
Nate Anderson
​
​
I recently got on Data.Gov to play around a bit. It’s a really interesting place to go in and get ideas for visualizations. I don’t have whole lot of time on my hands these days, and I’d really rather spend them partying or enjoying this gorgeous state I live in, but I started this visualization a while back and figured I would touch it up and load it up on my blog.
I was interested in looking at the concentration and/or distribution of political parties in the US. There has been a lot of talk about gerrymandering, these days. In the United States, the districts within US Congress is typically decided by state governments – with some exceptions. Gerrymandering is the process of the ruling party in a state redrawing political districts in such a way that benefits that party. Any time you have a political environment where the distribution of the partisan levels of districts and not skewed similarly across parties would be evidence of a political environment that benefits one side (the side with a greater disbursement of seats in zones that are safe for one party, but still relatively well mixed), and futher could indicate that this environment would be Gerry mandered.
The first viz I did was the distribution on page two of this worksheet: distributions of congressional seats seats based on the partisan breakdown (Cook PVI) of the congressional seats:
There are clearly other things at play here, and a lot of problems with the viz, which is why I almost didn’t publish it. First, it would be a lot harder to break up congressional districts that are wholly-contained in urban areas, and surrounded by other blue district (i.e. Manhattan), on the other hand, it seems a lot easier to give put a middle-sized urban area or two into a district to reassign some democratic voters into a safe conservative district.
Second, sentiments change… and maps change. Any look at a map would show that rural areas are mostly Republican and urban areas are mostly Democratic. This is pretty much the exact opposite of where we were in the 60’s and 70’s – and before.
If there is a change in the district lines in 2016 or 2010 elections, or if the Supreme Court somehow rules that particular districts are gerrymandered, these maps will change. Most pundits I’ve read would suggest that both the shift in the balance of power within states, and the political climate itself will favor the Democrats in 2018 more than it did in 2016.
Which leads me to the parameters of this map: I brock down every congressional district by its Cook PVI coefficient – a measure of homogeneity within districts. For example a high PVI Coefficient would suggest that a very large percentage of voters in this district vote for the party they skew to.
I assumed three things. First given the polling for Trump right now, that there could be a shift in Governorship power in 2018. Second, there is a possibility of a Supreme Court decision forcing states to realign their districts. Finally that there was higher voter turnout for Republicans in 2016. I assumed that these three things are minor, and if one or two happened, the PVI would shift dramatically. I think it’s fair to assume that the combined possibilities would account for maybe a 1 point shift in PVI, on average. Of course there is a way to control for this that requies some really sophisticated analytics, so I just shifted the PVI by 1 point.
Districts will be different, but I think that we could still use this as a roadmap for looking at the 2018 election. This of course will change when the new lines are drawn, and when that happens, I will upload the new data. Still it’s interesting to play with the slider and see which seats come up as vulnerable.
Needless to say, this analysis was simply made to look at districts are they are now, and there is clearly a more homogeneous population living in districts whose voter partisanship swings one way or the other.