Sunday, May 21, 2017

Numbers showing that WisGOP gerrymander getting worse over time

Bruce Thompson's Data Wonk column in Urban Milwaukee is back with another excellent analysis, this time about the Wisconsin GOP's gerrymandering of the State Assembly. To review, this gerrymander was ruled to be illegally one-sided by an Appeals Court in November, and that court came back in January and ordered the maps to be redrawn.

GOP hack/Attorney General Brad Schimel has appealed the case to the US Supreme Court to try to maintain the GOP's electoral advantage, but unless SCOTUS rules in the GOP's favor and overturns the Appeals Court, the State Legislature will need to draw new maps and have them in place in November 2017. One of the arguments that WisGOPs and Bradley Foundation front groups have argued is that Wisconsin has a unique electoral population where Democrats are heavily concentrated into urban areas, and so there will be a "natural gerrymander" that favors Republicans no matter how you draw the maps.

What Thompson points out in the article is that any "Big Sort" that may cluster Democrats together doesn't explain the distortion of Wisconsin's results, where a near-even split of votes for the Legislature has resulted in Republicans having overwhelming majorities in the State Assembly throughout the 2010s. A central argument behind the case that got WisGOP's gerrymander overrules was the usage of statistic called the "efficiency gap", where it was calculated how far away the results of the State Assembly's elections differed from what a fair map would have given.

Thompson includes the following graphic to show just badly this was skewed in 2016, where Donald Trump only got 47% of the state's presidential votes, but Republicans won 64 or the 99 Assembly races. In fact, Thompson notes that an already-bad gerrymander in 2012 has gotten increasingly worse as the decade has gone on.

Now compare that figure to 2008's outcome, which were under maps drawn by a federal court. In this instance, a 50-50 split would give the Republicans an advantage in about 52 Assembly seats, not 64.

Thompson also brings up the work of University of Michigan political scientist Jowei Chen, who did 200 computer simulations drawing up Wisconsin's 99 Assembly districts, and overlayed it with Wisconsin's 2012 vote. What he found was that the GOP's gerrymander created an efficiency gap well beyond anything that could be explained by "The Big Sort." In his analysis, a - number indicates districting that favors Republicans, and a + indicates districting that favors Democrats.
Figure 3 reveals that the simulated districting plans are reasonably neutral with respect to electoral bias. About 72% of the simulated plans exhibit an efficiency gap within 3% of zero, indicating de minimis electoral bias in favor of either party. In fact, 23% of the simulations produce an efficiency gap between -1.0% and +1.0%. These patterns illustrate that a non-partisan districting process following traditional criteria very commonly produces a neutral Assembly plan in Wisconsin with minimal electoral bias.

It is important to note that the simulations produce plans with both positive and negative efficiency gaps. Although the efficiency gap of every simulated plan is relatively small in magnitude, 90% of plans exhibit a negative efficiency gap, indicating slightly more wasted Democratic votes than wasted Republican votes. But 10% of the plans exhibit a positive efficiency gap, reflecting more wasted Republican votes. Hence, it is not extraordinary for Wisconsin’s political geography, combined with traditional redistricting criteria, to naturally produce a districting plan that somewhat favors Republicans.

The blue star in the lower left corner of Figure 3 represents the Assembly plan enacted by Act 43. (I can't figure out how to link the picture fromm the PDF, sorry.) This blue star depicts the enacted plan’s efficiency gap of -15.1%, reflecting significantly more wasted Democratic votes than wasted Republican votes. Thus, the level of electoral bias in the Act 43 Assembly plan is not only entirely outside of the range produced by the simulated plans, the enacted plan’s efficiency gap is well over twice as biased as the most biased of the two-hundred simulated plans. The improbable nature of the Act 43 efficiency gap allows us to conclude with high statistical certainty that neutral, non-partisan districting criteria, combined with Wisconsin’s natural political geography, would not have produced a districting plan as electorally skewed as the Act 43 Assembly plan.
One of the ways Chen notes that GOPs pulled as part of their extreme gerrymander was to split up counties that didn't have to be split up. Only 14 counties are kept together in one district under the current WisGOP maps, while Chen's computer simulations kept between 18 and 25 counties intact, which could make for more uniform interests in a district and likely less skewed outcomes.

Given the unpopularity of Republicans in DC along with a lagging economy in Wisconsin, it seems possible a Dem wave will hit in 2018. And if so, no wonder why Republicans in Wisconsin want to keep their rigged maps to give themselves as much of a cushion against such a wave, and get a free pass on backing legislation that a majority of Wisconsinites disapprove of. What Chen's and Thompson's work shows is that the lengths that WisGOP have gone to protect themselves from the will of the people is obscene, and should be ruled illegal based on the numbers.

Just because WisGOPs have rigged the maps, it doesn't mean that Wisconsin Dems don't have a lot of problems of their own to fix, as evidenced by the large amount of rural counties that flipped from Barack Obama to Donald Trump in 2016. But banning the gerrymander might be an important first step that would alert the average Wisconsin voter to how badly WisGOPs have acted, and lead them to understand a big reason why their elected officials seem to be doing things in Madison that the voters back home didn't ask for.


  1. Inclusion of micro-bordering of districts should be included in an Amicus.
    If you look at the micro-jagged nature of assembly borders, it's clear these borders reflect drawing borders to capture specific homes with regard to voting values, Republican or Democratic-leaning.

  2. It's troublesome that the proposed solution to the problem is to hand the process back to the people who created the problem. This is not an error on their part; the intent was clear. New maps should be drawn by an outside party.

    At this point, I don't think any political goal is more important than taking one house in both the Congress and the Wisconsin legislature in 2020.

    The Republican wave of 2010 handed them out of scope legislative majorities for what will likely be a full decade. That cannot continue.

    1. Win the Governorship 2018, and you have a veto in Wisconsin. And that's true in a lot of GOP-run states that voted for Obama.

      And I agree that giving the drawing tools back to the Legislature is stupid and will cause another set of problems. This was proven today when SCOTUS threw out a "corrected" North Carolina map because it was still a gerrymandered piece of garbage.