Betsy McCall
Gerrymandering has been going on longer than the Republic, and long before there was a name for it. District maps are drawn by those in power to maintain that power. Preserving incumbency has always been an important motivator, and with the rise of political parties, maintaining party control is also important to the legislatures drawing the maps. Various methods have been developed, including cracking and packing, to produce the legislature’s desired results. But while those in power are happy with this situation, voters would prefer fair maps that were capable of reflecting the preferences of the voters, including as they change from election to election. But the methods to thwart gerrymandering have had almost no research to determine what actually works. Some methods like compactness, have been proposed, but even researchers in the field note in the literature this gap in their knowledge even as they design technical solutions for creating maps. The goal here, therefore, will be to examine the fairness of current district maps and to determine if compactness could be a viable solution to the fairness question.
This project will rely primarily on data from two sources:
United States Congressional District Shapefiles and MIT Election Data.
The Congressional district shape files go back to 1789. MIT’s election data seems to only go back to 1976, but this will be enough for an initial start. From the shapefiles we can calculate area and perimeter of the districts which will go into our calculation of compactness. We can probably also use the area of the districts as a proxy for relative density since Congressional districts have to have about the same number of seats during the time period to be considered. The election data will have to be joined to the shapefiles to determine the winners of the elections.
Maps can be produced compare the statewide winning party (based on vote totals), and compared to the outcomes by seats in the Congressional delegation to determine how closely (or not) that they align. A larger deviance, would suggest worse gerrymandering.
And finally, classification and regression methods will be employed to determine if some of the proposed definitions for compactness actually correlate to fairer district outcomes.
Code: Working on it!
Data: Available in the links above.
Here’s my first code chunk.
R code will come later. All hiding now.
Add any additional processing steps here.
Coming soon!
coming soon
All sources are cited in a consistent manner Project Proposal