PGP and DRA "competitiveness" scores are worthless

The "competitiveness" scores many people have pulled from the Princeton Gerrymandering Project and Dave's Redistricting App to accuse the preliminary House and Senate maps of being uncompetitive and therefore gerrymandered are very poor measures of the ability of challengers to change the partisan outcome in each district, and do not present a valid critique of these maps. PGP's description of its Redistricting Report Card metrics ( describes its competitiveness rankings as follows: "We use statewide elections to calculate a partisan index of each voting precinct which we can then use to calculate district-by-district estimates for Democratic vote share percentage (it is common practice to study estimates from the Democratic point of view, but one could easily flip the analysis and consider it from the Republican point of view). To calculate the partisan index, we consider an average of the most recent statewide elections for the U.S. President, U.S. Senate, and Governor. [ ...] Number of competitive seats: Calculated by counting the number of districts whose Democratic and Republican vote share percentages estimate is between 46.5-53.5%" The most common "competitiveness" metric people find in DRA is similarly derived by counting the number of districts in the 45D/55R-55D/45R 2-party vote share range in their default election dataset ("Composite 2016-2020"). Both of these two datasets have strongly Democratic-favoring outomes relative to the set of all recent statewide election datasets in PA. The district-by-district electoral outcomes in these dataset are not necessarily reliable representations of how districts will perform over the course of a decade, even ignoring issues like incumbent advantage and random variations in candidate quality. Moreover, these single averaged measurements tell you nothing about how the districts actually vary from election to election, even ignoring those factors. A 45D/55R district may have 45% of its voters who will always vote for Democrats no matter what and 55% who will always vote for a Republican no matter what, or it may have 35% of its voters who alwasy vote D, 45% who always vote R, and 20% who frequently flip from one party to the other or who only vote sporadically. The first district is in no way competitive, no matter how "close" the electoral outcomes appear, while the latter is _highly_ competitive, even if sometimes the results come out 35D/65R and sometimes they come out 55D/45R. The only way to actually reliably determine how voter behavior may change with changing electoral circumstances, and, thus, whether a district may genuinely be competitive, is to look at many different electoral outcomes _separately_ and see which districts _actually change_. Dave's Redistricting offers many different electoral datasets and thus makes this check feasible. Looking at those datasets, it's easy to see that the "competitiveness" measure varies significantly depending on which electoral dataset you use. "Composite 2016-2020" is a good example of a D-favoring dataset, while "Senator 2016" is a reasonable example of an R-favoring dataset. With these two datasets, the "competitive" district counts for the current and proposed maps are: dataset Comp Comp.Dem Comp.Rep house-current-composite 45 20 25 house-current-senator2016 35 14 21 house-proposed-composite 29 13 16 house-proposed-senator2016 43 22 21 senate-current-composite 13 9 4 senate-current-senator2016 16 6 10 senate-proposed-composite 8 4 4 senate-proposed-senator2016 15 10 5 (The "Comp.Dem" and "Comp.Rep" columns break down the total count of "competitive" districts into those that lean D and those that lean R.) Using the better measure of looking at districts that _actually flip partisan outcomes" from the composite to 2016 senate datasets_, we see that the current and proposed maps aren't much different, and many are missed by the other ranking.: map D2R R2D D2D R2R house-current 21 0 77 105 house-proposed 22 0 84 97 senate-current 9 0 17 24 senate-proposed 5 0 21 24 Don't be fooled by partisan talking points.