Breaking Down the Expert Testimony of the PA Gerrymandering Case
If a Congressional District looks like it is gerrymandered and behaves like it’s gerrymandered, it’s probably gerrymandered. By that I simply mean if you look at a district and can not locate any communities of interest or make out a sensible geometric shape, and the district consistently goes to the same political party, it was probably drawn for the purposes of intrenching political power. No one can look at the 7th Congressional district of Pennsylvania and say it was drawn to properly represent the residents of the southeastern Philadelphia suburbs. Paired with the knowledge that the 7th has gone to Republican in all three House elections held since the district was redrawn in 2011, the district starts to smell of a gerrymander. When you see that 13 districts have consistently gone to Republicans, and the other 5 districts have consistently gone to Democrats, the overall picture becomes more clear.
Despite the overwhelming evidence provided by this simple “smell test”, the actual challenge to addressing partisan gerrymandering is quantifying its partisan bias and demonstrating its impacts. In a Pennsylvania courtroom, six experts testified to provide a clearer picture of the scope of gerrymandering in the Commonwealth’s congressional districts. Dr. Jowei Chen, who has also testified in the Wisconsin gerrymandering case which has made its way to the Supreme Court of the United States, leant his expertise to modeling what Congressional maps could look like in Pennsylvania if partisan bias was removed, and compared these results to the 2011 plan the Commonwealth currently uses. Dr. John Kennedy testified to the practices of “cracking” and “packing”, and explained how these were both used to draw the 2011 plan. Dr. Wesley Pegden detailed his worked on a computer algorithm which tested the extent to which the districts were politically biased, and Dr. Christopher Warshaw brought up the term “efficiency gap” and how Pennsylvania is an outlier with its 24% efficiency gap. Dr. Wendy K. Tam Cho and Dr. Nolan McCary were brought in a expert testimony for the state to refute the testimony of the other four experts. The testimony of these experts and the effects they had on the Court’s final decision are detailed in the 139 page opinion, but, to save you all some time, I will detail the major points here.
What would a good Congressional District look like? Dr. Chen set out to answer this question by designing a computer model which would draw the 18 congressional districts of Pennsylvania. The program had a few criterion to meet, namely population equality, contiguity, compactness, absence of splits within municipalities, and absence of splits within counties, unless necessary. For good measure, this program was run 500 times, to create 500 distinct randomly drawn Congressional Districts maps which would be compared to the 2011 redistricting plan. Additionally, Dr. Chen ran another 500 simulations with the added parameter of protecting incumbents, to see if this added non-partisan (but still politically motivated) goal could explain for the odd character of the Commonwealth’s Congressional Districts. His findings were as follows:
Of the 67 counties in the Commonwealth of PA, 28 are split up into seperate Congressional Districts in the 2011 Plan. Of Dr. Chen’s 500 simulations, the maximum number of counties in separate districts was 16.
Dr. Chen used the Reock Compactness Score to measure how geographically close a District is, with the higher scores indicating a more compact district. A perfect circle would score a 1 for this measurement. The scores for the 2011 plan range from .278 to .442. For Dr. Chen’s 500 simulations, the scores ranged from .38 to .46.
Dr. Chen also used the Popper-Polsby Compactness Score to have another compactness measurement. The Reock uses a ratio of areas to measure compactness, whereas the Popper-Polsby uses a ratio of perimeters, making the two slightly different measurements of the same concept. Again, the higher the score the better, with 1 being the best possible score. The 2011 plan had an average score of .164, and the 500 simulated plans had an average score of .310.
Using precinct level election data, Dr. Chen was able to predict the partisan vote tally for his simulated Congressional Districts. As mentioned, 13 of the 18 seats have consistently gone to Republicans with the 2011 plan. In the 500 simulations, the maximum number of Republican seats was 10, with the average being 9.
The mean-median gap compares the average percentage of the vote share a party receives per District, and compares this to the median percentage of the vote share. The Republican’s mean vote share was 47.5% per district, whereas the median vote share was 53.4%. Having a median vote share which is higher than the mean vote share insures that over half of the Districts will receive a higher-than-average share of the vote. A 6% buffer as experienced in the 2011 plan, did not occur once in Dr. Chen’s 500 simulations, with the mean-median gap ranging from 0 to 3%.
The other main nuggets of statistical information came from Dr. Pegden and Dr. Warshaw’s testimony, regarding the partisan bias of the Districts and the efficiency gap, respectively. Dr. Pegden ran a computer simulation which tested the change in partisan bias when a small change was made to the district boundary. After 1 trillion (1,000,000,000,000) such changes, Dr. Pegden found 99.999999% of the changes did not increase the Republican bias of the Districts. This means that of the one trillion changes the computer program made to the 2011 plan, only one million resulted in a change which would further help Republicans. Perhaps the 2011 Redistricting Commission did not leave every stone unturned, but they certainly left 99.999999% of them unturned.
The efficiency gap has already been brought up in previous Podcasts here at Three O’Four, and Dr. Warshaw used this measurement to testify how the 2011 Plan is outlandishly favorable to Republicans. He presented the measurement through a historical perspective, noting that in the 1970s the efficiency gap was close to zero, possibly leaning Democrat, and prior to the 2011 Plan, no party enjoyed an efficiency gap north of 10%. However, after the redrawing of Districts in 2011, the Republicans enjoyed a gap of 24%, 15%, and 19% in the 2012, 2014, and 2016 elections respectively. He also noted “it’s very unlikely that some changes in political geography or some other aspect of voting behavior would have driven this change.” Indeed, Dr. Chen’s testimony demonstrated only a 0 to 3% advantage for Republicans with respect to geography.
While the geography of Pennsylvania may lend some advantage to the Republicans when it comes to drawing Congressional Districts, the measurements done by experts indicate that the republicans should not be enjoying it in such a lavish manner. Dr. Chen’s simulations demonstrated how even a non-partisan map could look in the Commonwealth of Pennsylvania, and his comparison along with the testimony of Drs. Kennedy, Pegden, and Warshaw all demonstrate how outlandish the actual maps are in comparison. One cannot be certain how large of a role the expert testimony played in the Supreme Court of Pennsylvania’s final decision, however, it was certainly enough to take up a solid portion of the Majority Opinion as a defense. Through it all, the testimony demonstrates how powerful computer algorithms and programs, while they may be used to draw such gerrymandered maps, can also be used in court cases to strike down such maps as unconstitutional.
Stay Hungry, Stay Foolish
Seamus O'Hara
