In my first Assembly post, I had identified a Wisconsin (AD-31) seat at Democratic when it really was Republican. Part of the reason I try to get as much public input as possible on this project is because when it comes to state legislatures, it is not always easy to get the most reliable data on the spot. I've crosschecked those lists again, but I am always going to rely on readers to let me know if I've got something wrong - the help is invaluable.
A few other folks started raising concerns about data that didn't line up with their on-the-ground knowledge of Wisconsin. Immediately, I found there were two real causes behind this, both very different in nature.
The first was a simple Excel problem - the dynamic spreadsheet I use to put all my data in was supposed to auto-update in areas, but failed to do so - that quick fix made up for some of the misleading data we were seeing.
The second problem, and the larger one, was one structural with my model. The level of data I use largely depends on how many people the legislators represent - and in most cases, county level data is accurate, and much more efficient than having to go ward by ward for every single seat. My prior exception was only for major American cities, just because they have such a high concentration of districts that if I did not, you would wind up with a chunk of districts all with identical scores - not a very accurate reflection of reality.
During my testing, I look at a wide ideological range of districts to make sure my process was accurate. One thing I failed to account for, however, is the serious urban-minority/suburban-white divide that exists. I use state legislative elections as a local check in the system to account for community differences in party preference - but that simply isn't a strong enough check when the communities have so little in common - economically, ethnically, and politically - with each other. As a result, I've decided that I'm now going go ward-by-ward when possible for all counties over 200,000 residents.
The result is a picture that I think my Wisconsin readers will find as a much more accurate assessment of the two chambers, and will prove a better model for future states. I want to thank everyone who raised concerns and sent questions my way for making it this much better.
Let's talk about the Senate first. In my first projections, I urged caution over getting too optimistic about the recount this summer - other than Sen. Dan Kapanke (SD-32), the rest of the recalled Senators (Ballotpedia has an easy to read list you can use to match names to districts) represent areas that appeared fairly reliably conservative - a major ground change would be required for the Democrats to get the three seats they need in order to flip control of the chamber.
Sen. Kapanke's district (D+6) has moved even further into Democratic territory, and is now the 9th most liberal seat in the Wisconsin Senate. As I'm sure the Senator would attest to, the frustrating thing about recalls is that they really don't offer a graceful political exit - you can't make up an excuse to not run in a race to fill out a term that you previously signed up to serve. And now, a year removed from being a serious challenger for Congress against Rep. Ron Kind, Sen. Kapanke is going to lose his state senate seat - not the desired trajectory for a political career.
However, the effort it will take to claim the two other necessary seats is remains high. As discussed last time, if you apply a universal seven percent swing on the best Democratic result from the fall, Dems would be on track to win a R+4 seat - but my fixes to the index haven't brought the other seats any closer. Commenters have noted that union membership is very uneven across these contested districts - and so the bar to beating Republican support may be lower.
Questions have also been raised about turnout, but my guess is that with all that is at stake, there is going to be some pretty serious mobilization - on both sides - and assuming Democrats are going to have a remarkable turnout advantage seems to be an inappropriate strategy. But between the personal troubles of Randy Hopper (SD-18... before his wife kicked him out and sent him to live in Madison after learning of an affair with an aide) and the varying union membership, flipping control of the Senate appears to be a very attainable objective.
On the house side (name for districts here, you can see the real changes. To start, the most heavily Democratic districts are twice as Democratic as before - they are the heavy minority districts that the previous model was missing out on. It also shows the amount of Democratic votes that are sunk into these seats. Only 42 of the 99 seats are more Democratic than the state on average. The current districts are not an overt gerrymander - they date back to a 1992 compromise plan between then-Governor Tommy Thompson and the Democratic Legislature. In 2002, the districts were merely altered by the courts to balance the population shifts. The result is a fairly high amount of Democratic vote sinks that seem destined to only grow Democrats can grab control of some part of the redistricting process this summer.
Also, I have got some news going forward - over the next few days I will be releasing data for Nebraska and Ohio, where another one of Governor Scott Walker's friends has been running into popularity problems lately, jeopardizing his legislative majorities. I'm also working on embedding spreadsheets into the blog via Google Docs so that people can quickly input the rankings for their own calculations. For updates on the project's progress, and any random questions you want to throw at me, follow my Twitter account, TheLDI.
No comments:
Post a Comment