Simple demo of gerrymandering for Associated Press

Hosted at https://gerrymandr.github.io/ap-demo/

Contributors:

• Sam Stewart sams@umn.edu @samstewart
• Alex Gutierrez alexg@umn.edu
• Michael Altmann michael.altmann@gmail.com
• Nina Amenta amenta@cs.ucdavis.edu

We use D3 for the front end and QGis, Python and PostGis for data maniulation

Starting with U.S. congressional districts in Lancaster, Delaware, Chester, Montgomery, and Berks counties, PA. Really starting with Chester county.

2010 Census 2016 Election. Election data pulled from: https://github.com/openelections/openelections-data-pa

  Chester    --- 15
  Delaware   --- 23
  Lancaster  --- 36
  Montgomery --- 46
    (source: https://github.com/openelections/openelections-data-pa)

County FP10 Codes

Chester 29
Delaware 45
Lancaster 71
Montgomery 91
Berks  11

Data and Sources

Shapefile of entire area of interest.

County shapefile is in

data/Penn County

Retrieved from

https://github.com/aaron-strauss/precinct-shapefiles/blob/master/pa/2011_pa_precincts.zip

Used QGIS to select just a single county and save it as a separate geojson.

Shapefile and demographics for each MCDS

Complete MCDS shapefile is in

data/Penn MCDS Data

with subfolder for subsets by county.

Retrieved from

https://github.com/aaron-strauss/precinct-shapefiles/blob/master/pa/2011_pa_precincts.zip

Used QGIS to select data for a single county and save it as a separate geojson.

Used the topojson-server node.js package to convert geojson to topojson.

You could also use http://mapshaper.org/

Election returns by MCDS

data/chester_county_votes.txt

Data processing steps

Obtain election results from https://github.com/openelections/openelections-data-pa Note that for each location, there are separate rows for each candidate Import into QGIS as a delimited file with no spatial information

Use GroupStats to create a layer with

name_upper,idIfPossible,democrats,republicans

Obtain shapefile with demographics from https://github.com/aaron-strauss/precinct-shapefiles

Load into QGIS

Create a new String column of width 100 NAME_UPPER = upper(NAME10)

Join spatial layer with voting data layer on NAME_UPPER

As desired, filter by county ID.

Export as geojson

Use

geo2topo -o geo_and_votes.topojson geo_and_votes.geojson 

Adjacency matrix for MCDS (old)

Computed in PostGIS from MCDS shape files

create table "PA_MCDS_graph" as
SELECT a."GEOID10" AS "GEOID101", b."GEOID10" AS "GEOID102",
    ST_GeometryType(ST_Intersection(a.geom, b.geom)),
    ST_MakeLine(ST_PointOnSurface(a.geom), ST_PointOnSurface(b.geom)) AS geom
FROM "PA_MCDS" AS a LEFT OUTER JOIN "PA_MCDS" AS b
ON a."GEOID10" < b."GEOID10"
AND a.geom && b.geom AND ST_Touches(a.geom, b.geom)
AND ST_GeometryType(ST_Intersection(a.geom, b.geom)) not in ('ST_Point', 'ST_MultiPoint')

To start HTTP server:

If you are a developer hosting these files locally, you will need to serve them up from a web server. It doesn't matte rth etechnology, but for Python, here are the instructions.

For python < 3:

python -m SimpleHTTPServer

For python >= 3

python -m http.server