How Spatial Polygons Shape Our World
Amelia McNamara, OpenVisConf 2017
This repo contains a PDF version of the slides for my 2017 OpenVisConf talk, How Spatial Polygons Shape Our World. The video of me giving the talk is available on youtube, and the full transcript is on the conference website.
References and links:
All maps are wrong:
- Choropleth maps color polygons by the value of some variable. This one shows Prop 1 election results from Seattle.
- All maps of parameter estimates are misleading, paper by Andrew Gelman and Phillip Price.
- An illustration of the issue of mapping parameter estimates: Avoiding Data Pitfalls, Part 2: Fooled by Small Samples (Kidney cancer rates by county).
- A solution to the issue of mapping parameter estimates: Surprise! Bayesian Weighting for De-Biasing Thematic Maps, by Michael Correll and Jeff Heer.
- The standard presidential election map: 2008 Presidential election map.
- One way to avoid being fooled by large areas with few people is to make a cartogram, like the 2004 Presidential election cartogram.
- Square and hexagon cartograms are another way to avoid being fooled by area.
- If you're making cartograms, you might care about Whose map is better? Quality metrics for grid map layouts.
- Evaluating cartogram effectiveness is another resource.
Combining incompatible data
- With rectangular data, you could use the RStudio data wrangling cheatsheet to join two types of data. With "incompatible spatial units" this doesn't work.
- The Modifiable Areal Unit Problem is the problem that aggregating point data to different polygons can have huge effects on the visual distribution.
- It shouldn't surprise us that this happens. Check out the histogram essay I've been working on with a collaborator to see this in a one-dimensional case.
- In the context of elections, we call the MAUP Gerrymandering.
- John Oliver on Last Week Tonight explains gerrymandering in a much funnier way than I could.
- A commonly-cited Washington Post article, "The best explanation of gerrymandering you will ever see."
- For a more realistic experience, try the Redistricting Game.
Downscaling, upscaling, sidescaling
- Disser could help you move from aggregated data to point data.
- Jon Kimerling has thoughts on Dotting the dot map, revisited.
- Another project with my collaborator: Spatial aggregation explorer.
- In the electoral context, you might want to Redraw the States by moving counties from one state to another.
- How zip codes almost masked the lead problem in Flint by aggregating by inappropriate polygons.
- The pycno package in R performs pycnophylactic interpolation.
- Aran Lunzer, my collaborator, who has been thinking about how aggregation impacts histograms and maps with me for many years.
- Pierre Goovaerts, whose talk on geostatistics in practice at UCLA in 2014 started me thinking about the change of support problem.
- Friedrich Hartmann, who introduced me to dasymetric mapping and the MAUP at IEEEVis in 2014 and pointed me toward cogran.js right before my talk.
- Moon Duchin, one of the mathematicians solving gerrymandering.
- Richard Casey Sadler, who works on public health problems in Michigan and discovered many problems with the Flint water data.
- Matt Brehmer, for sending me references to cartogram effectiveness measures.