The following is taken from a post at a web forum soon after creating the cartogram visualization of Election 2004:
...why on earth did I do this ? It's really a combination of interest and opportunity. I am a computational geometer, and the problem of computing a cartogram for a map (where the weight function could be population, income levels, disease prevalence, what have you) is an interesting challenge, mainly because of the problems I described above.
When I saw the electoral maps, I was reminded of work that that colleagues have done, and asked them for their data (hence the opportunity). it seemed like an interesting exercise in practical applications of geometry.
My starting point for this page was Robert Vanderbei's purple map
representing the vote count (at a county level) in Election 2004 for Bush (red) and Kerry (blue).
People had asked him (and I had wondered as well), whether the map could be represented as a cartogram, where each county was distorted to represent its true population, so as to make a more representative picture. Fortunately, I have colleagues who work on cartograms (more on this in the Credits), and was able to use their cartogram to create an equivalent purple map.
Here's the FAQ:
- Can you change the colors ? Yes !! some color changes that
people proposed are rendered below. Send me mail if you want a particular
- Can you change the map ? No ! The map data was given to me by Stephen North (see credits, below)
- Can I have the data ? Contact Stephen North for the cartogram
data. I downloaded the election results from USA Today's website.
Update: Nov 7, 2004
Garstner, Shalizi and Newman have another cartogram rendition of the vote:
They also had a histogram of county-level vote-shares, that had some anomalous values, and was therefore removed. The picture below is a similar histogram that I generated from the vote results.
And now for the maps...
In the maps below, B = fraction of vote Bush got, K = fraction Kerry
got, and T is the third-party vote (almost always negligible). The color
scheme is expressed in these terms. both raw PPM files are GIF files are
linked; be warned that the PPM files are very large.
- Basic purple map: (r,g,b) = (B, T, K). (GIF small/big) (PPM small/big)
- Winner-take-all: if( B > K ), color with red. If K > B, color with
blue. (it never happens that both are equal). (GIF small/big) (PPM small/big)
- A divergent bivariate color scheme generated from Cindy Brewer's beautiful ColorBrewer site (Thanks to Anthony Robinson for suggesting it !). (GIF small/big) (PPM small/big)
- Red-blue (suggested by a poster on the blog). Start with a baseline of white for B = K. as the vote percentage for the winner increases, increase the strength of red or blue appropriately. (GIF small/big) (PPM small/big)
- Grayscale (suggested by Kathryn Myronuk). Start with a baseline of white for B = K. as the vote percentage for the winner increases, make the color grayer. (GIF small/big) (PPM small/big)
- ROYGBIV (suggested by Kathryn Myronuk). Again, white is neutral, but go towards the R side or the V side of the color spectrum depending on who wins and how much (thresholds at .7, 0.6, .53) (GIF small/big) (PPM small/big)