Recently my colleague Liz Scheltens argued that the standard electoral map we see every four years is pretty useless. The problem, she says, is that the maps you typically see favor geographical accuracy over electoral importance. Scheltens proposes alternative solutions, like a list of states with bar charts or a cartogram, which is a kind of map that distorts geographical areas based on some variable.
I think about maps a fair bit, and I'm lucky that it's part of my day job: I built Vox's maps tool (examples of the choropleth maps it produces are here and here) and have created some cartograms too. My thoughts on data visualizations are not static; they evolve because the times evolve.
Scheltens's video raises a great point. The typical electoral map you see every four years has its weaknesses. But so do her suggested solutions — and that matters to journalists who design information graphics. We have to weigh the trade-offs in what we make to ensure clarity, accuracy, and fairness. How to map data is a contentious debate in data visualization, and not just with electoral maps: A geographically accurate map can underrepresent minorities or put more visual weight on large western states. These kinds of maps have their problems, but so do cartograms.
The case against cartograms
A cartogram is a kind of map that distorts geographical areas based on some variable. The electoral map below shows states portrayed as squares sized by their number of electoral votes. The squares are then arranged in a way that loosely captures neighboring relationships. This is the cartogram that Liz suggests to improve the typical election map:
Here's the thing: Cartograms are an alternative, but they're not necessarily better than geographically accurate maps. At least with the latter, you get, well, geographical accuracy. The supposed benefit of cartogram-based maps over standard choropleth maps is that you're able to understand scale or significance better. But can you really?
Take a look at that proportional representation electoral cartogram one more time. Can you tell me:
- What color is Iowa, a key battleground state?
- How many electoral votes does Iowa get versus California or New York?
- Who's winning, red or blue?
I'll start with question one: It took me a couple of seconds to find Iowa, which on a typical state map is located a little north of the center of the map but in this case is located a third to the left. It's blue, but I would have found it faster on a typical map.
On to question two. Just eyeballing it, I'd guess Iowa has about a sixth of California's electoral votes and a fifth of New York's. In reality, Iowa has six electoral votes, California 55, and New York 29. So I'm in the right margin for New York but pretty far off from California. That's still not too bad if the intent is to show me scale: California has more sway than New York, which has more sway than Iowa. This is an interesting fact, but it doesn't tell me the thing I want to know the most, which leads into the last question.
This is where the map really fails for me. The proportioned squares aren't arranged in a way that tells me which side is winning. It looks pretty even to me, but Obama won by 126 electoral votes. Granted, that's not clear in a geographically accurate map, either. But that doesn't mean a cartogram is the solution.
When a map is a dictionary
Since Scheltens has already made the case for cartograms (and it's a good one!), I'll make the case for geographically accurate electoral maps.
When I think about geographically accurate state maps, I find that their usefulness is really more of a visual dictionary. Rhetorically, these maps are lookup tools that enable you to easily match up a state and the value you're trying to show.
I get that this is nearly a tautological argument: A map shows you where a place is, so it's easy for you to find that place. But look at Politico's map and bar chart list below and tell me: Which one makes it easy to find out how Wisconsin voted? (The list is so long that I had to crop it so you can view it in your browser. Click on the images to view them larger — or, better yet, check out how Politico laid out its 2012 results on its site.)
The fact that I had to crop the super-long list so you can't see Wisconsin only proves my point. Politico's bar charts represent the popular vote and not the number of electoral votes, but it doesn't matter if each state's electoral votes were represented by an appropriately colored and sized bar — it's far easier to find the state you're looking for in a map, not a list.
Cartograms are okay, too. I just don't think they're the end-all, be-all solution. They can be an attractive alternative to the monotony of choropleth maps, especially with the proliferation of tile-grid maps (an equal-area cartogram useful for comparing states) — but they're just an alternative.
Alternatives to maps when we're talking about electoral significance
So what are some great ways to show electoral significance?
I loved the Guardian and RealClearPolitics collaboration that depicted states as balloons:
The individual states don't matter so much as the separation of those state balloons into clusters, making the collective number of electoral votes easy to parse visually.
My favorite visualization for electoral significance — definitely in my top three favorite viz of all time — comes from the New York Times:
Mapping out every scenario to take the swing states is as fresh an approach today as it was when the visualization first debuted in the 2012 election.
Making visuals is a series of compromises on how to best relay information to a reader. And for elections, I don't think cartograms are better or worse than geographically accurate maps at all. Cartograms are very useful tools to explain the Electoral College, but not necessarily the better visual as results trickle in on November 8.
On election night, I know what we'll see. Big numbers that tell me how many electoral votes the candidates are racking up, plus a geographically accurate map that shows how each state is leaning. And it's going to be just fine.