Let's Build Better Election Visualizations
In the wake of the U.S. election, a ton of different outcome visualizations have made the rounds online. The traditional winner-take-all electoral map gets the job done on election night…
…but doesn't tell a very nuanced or useful story about the American electorate.
For starters, this map doesn't account for population distribution. It also makes the country look like a game of Risk between Blue and Red. Maybe that's accurate on some level. But the way we visualize ourselves informs the way we think about ourselves. The Risk board is built for competition -- and when we look at that election night map, we're reading it in fundamentally competitive terms, terms that are more concerned with points on the scoreboard than facts on the ground.
For example: If you simply look at the results on a state-by-state level, Alabama, Mississippi, Georgia, and the Carolinas are all fire-engine red. But this overlooks an important phenomena revealed by county-level analysis: a band of reliably Democratic African-American voters stretching through each of those generally conservative states.
This matters to election strategists, of course, and it matters to any citizen who wants to understand their state and country more meaningfully. Finer details reveal larger narratives: the geographical distribution of these southern Democratic voters was shaped in part by prehistoric plankton over 100 million years ago.
The more granular you get with your visualization, the more geopolitical subtleties you can discern. That's obvious. So the question becomes: how can you most accurately and informatively depict a highly granular electoral dataset? How do you represent a nation to itself in a way that's both factual and meaningful?
In order to address the population distribution problem, some visualizers have turned to cartograms -- maps with dimensions distorted in order to weight for population. Here's Chris Houston's state-by-state model from Wikipedia's Electoral Collage page :
Mark Newman of the University of Michigan's Department of Physics and Center for the Study of Complex Systems created a number of different cartograms -- the one below distorts each county in the U.S. according to population (while attempting to retain the basic shape of the country as a whole) and colors the counties in shades of blue, red, and purple according to vote percentages:
Then there's this map, which Gizmodo's Jesus Diaz declared "the real political map of America." It indicates population density according to color saturation:
It's a good step toward nuance that preserves the accessibility of the familiar and demarcated map of the United States. This approach from John Nelson of IDV Solutions is even better:
You can see a much higher-resolution version of this map here. In conversation with io9, Nelson describes his map as "a pointillist look at the 2012 election results, which does a fairer job of illustrating where, how many, and how people voted in the election than the more typical full-color generalization." He goes on to say that he was inspired by "seeing so many horrible cartograms and pseudo-extruded county maps, and even vanilla solid-color maps that miss the boat on variations in population density."
Still, we're not representing third-party votes or, maybe more crucially, non-voters. The American Presidency Project at the University of California at Santa Barbara estimates that in 2008, 57.5% of Americans old enough to vote actually did. Whether you consider this a travesty or no big deal, it's significant information -- abstention defines a democracy as much as participation.
How do we build a better map? For starters, I'd like to see a pointillist approach expanded to include non-voters -- let a yellow dot represent 100 abstentions. I'd also like to see votes for the Green and Libertarian parties represented in a way that doesn't assign them to a red-blue continuum.
Are there other data points an election map should represent? Which visualization method balances accessibility and data-richness most successfully? Maybe we can build a better map and tell a better story.