Comments on: An Undistorted Election Results Map

By: Matt Healy

Matt Healy — Sat, 13 Dec 2008 22:38:10 +0000

Over in the EagerEyes blog is an interesting discussion of ways to visualize electoral trends over time; my two cents can be found here:
http://eagereyes.org/blog/2008/swing-states.html#comment-1094

My picture can be found here:
http://mdhealy.home.sprynet.com/Elections_1904_2008.png

Making it took me just a few minutes in Excel. First I downloaded the zipped data through 2004 that the maintainer of the EagerEyes blog made available. Second I manually added a 2008 line and sorted in reverse order by Year. Then I color-coded with Conditional Formatting.

The trickiest part was how I sorted the States to emphasize both recency and consistency. Below the Year column I put an exponentially-decaying series of numbers beginning at 1.0, below that 0.8, below that 0.64, etc. Then in the cells to the right of these numbers and below the main data table I put a bunch of formulas along the lines of =IF(K5=”D”,$B57,0) so that each cell in this region had either a zero if a corresponding cell in the primary table wasn’t “D” or a value that decayed exponentially by rows for a “D” in the main data table. Below all these numbers I put a bunch of SUM formulas, thereby making a series of weighted “D” counts with the most recent elections getting the most weight.

Finally I selected the weighted sums, then used Paste Special/Values to put their values above the main data table, then I selected the table contents, picked Sort, clicked “Options” and specified Sort Left to Right.

I manually tweaked the year 1964 a little to emphasize that key year.

I think the result pretty clearly brings out various aspects of the data, making both major national swings and cases when a single State differed from its overall trend quite visible.

By: derek

derek — Sat, 15 Nov 2008 09:57:55 +0000

I was very impressed with the interactive New York Times President map. Two of the options are “County leaders”, which is a pure heat map tinting the counties according to the percentage lead, but not correcting for population, and “County bubbles”, which corrects for population but only shows the bubbles as red or blue, not tinted for percentage lead. In theory I suppose you could combine both “retinal variables” (to use a Bertinism), but I’m not sure how much extra value that would add.

You can adjust the bubble size. I prefer the smaller bubbles. so there’s less overlap. You can also step through the last four elections, and it’s striking how many more people are recorded as voting in recent elections than in 1992.

By: Jon Peltier

Jon Peltier — Fri, 14 Nov 2008 14:08:38 +0000

Andy -

I think there’s room for both of these visualizations, and more that I haven’t thought about. There are certainly benefits to keeping the map undistorted, but there’s something about the striking cartograms and the not-so-subtle way that they remind us how much diversity our country exhibits.

By: Andy Woodruff

Andy Woodruff — Fri, 14 Nov 2008 13:51:18 +0000

Hi Jon, glad to see you’re still thinking about this! I’ll certainly try to make a couple of maps with your suggestions. Hopefully I can post a handful of examples on the Axis Maps site within a few days. A white background was actually the way I did it in original experiments with the idea. We opted for the black background (Mark Harrower’s idea) for aesthetics, though I believe it also has the effect of changing the visual variable involved, from saturation to brightness.