Election maps, like the elections themselves, have cycles. On election night itself, it’s all about the live election maps. Now that the dust has settled on the election, it’s time for the usual round II: wherein multiple authors try to cut through the terseness of the electoral map and demonstrate that, yes, we are more or less a purple country.
These maps tend to fix three things “wrong” with the usual electoral map: first, it aggregates up all voting patterns to the state level, when we know that there are important distinctions in voting patterns in cities, suburbs, and rural areas. Second, regular electoral maps paint things pure red or blue (thanks, electoral college), but we know that there’s granularity in how strongly someplace voted one way or the other. Thirdly, regular electoral maps don’t demonstrate the massive divide in population density that this country has. The mountain states are really big, but it’s not where most of the people live.
Anyway, let’s get onto the maps that seem to be making the rounds:
First is Mark Newman’s cartograms, which are kindof an institution by this point. You should read his writeup for more maps and more background. There’s lots of ways to scale a map by attribute (cartogram): but this is the Gaster-Newman algorithm. Personally, I have a few misgivings about this map. It’s nice because, if you squint or something, you really do get the impression of the vote being split 50/50. However, if I asked you to point out Minneapolis on this map, or any other more specific spatial pattern, it’d be a challenge just because all the shapes are so darn distorted. It’s still a great map, of course, you just need other representations to help round out the story.
Second is a mapby illustrator Chris Howard, which does the usual ‘purpling’ as then overlays it all onto a map of population density. It’s striking, although I’m pining for a legend: I think the counties are more transparent based on their population, and that’s reinforced by the population density map, but I’m not entirely sure. It’s super-striking, though, and despite the ambiguous symbolization it does a fantastic job showing both big and small scale voting patterns.
Finally, we have John Nelson’s dot density solution. I will admit a personal fondness for this one. There’s a few nitpicks I have that are inherent in any dot density map (the voting in Duluth county, Minnesota is not that geographically spread out, for instance), but otherwise it does a great job of de-highlighting the plains in favor of cities, ‘purpling’ as appropriate, and demonstrating spatial patterns (notice how most cities have a Democratic-leaning center and Republican-leaning suburbs!)
One thing I want to remark on, in closing, though: these post-election maps is how often they’re described as showing the ‘true' America, or providing 'the full story' of the election, and I'm pretty perturbed by that. These maps provide more information, and perhaps more intuitively than others, but I will always caution people from treating any map as undeniable gospel! These maps are awesome, but they have limitations. It’s not the fault of the designers: it’s just that the voting decisions of hundreds of millions of people nationwide is a very complex phenomenon, and you can never condense that into a pithy and perfect image.
It’s election night, and while most people are wondering about trivial things, like who’s going to be the next leader of the free world, us cartographers are interested in one much more humble thing: data visualization! Tonight is not just a race to 270 electoral votes, but also a race for your mouse clicks, as all the major news outlets deploy the most sophisticated, interactive, share-able map displays they possibly can.
So in that spirit, let’s take a closer look at some of the maps that were brought out this year. Don’t forget you can click any of the thumbnails to be taken to actual maps! Unless their links have decayed overnight, which is an all-too-real possibility.
Huffington Post offers our first unique data widget of the night. You might miss it: you have to click on a state to pull it up. But that info window offers some things nobody else does. First, an examination of the propensity for voter suppression in each state. Second, a plot of each county according to its lean in both this and the previous election. A county moving south of the diagonal signals a Republican-wise shift, and moving north of it indicates a Democrat-wise one. Ergo: the flatter the trendline, the more divisive that state has become, or vice-versa.
USA Today's map is stately for sure. In fact, I think this is my favorite of all the election maps in terms of pure design. I love the visual parsimony of it (Free of visual clutter! Responsive web design, so no scrolling needed! Live ticker feed!), but otherwise there’s not too much to remark on. One thing though: it’s one of the few maps to represent ‘leaning’ using a texture (those diagonal hashy lines), instead of color. I honestly have no assumptions about which representation works better, but it’s one of the few ways where the various maps differ in terms of symbology.
The Wall Street Journal is the other map opting to use texture for leaning states. Other than that, there seems to be some interesting info available if you want to drill down state-by-state, but for whatever reason (server load? regular ol’ bugginess?), I can’t get them to work. I do also want to call attention to the bivariate symbology used on the electoral vote counter. The height of each block represents the margin of that candidate’s victory in the state, while its width is its amount of electoral votes. As you might be able to reason out, a sizable chunk of Obama’s electoral victories came from states with a very narrow margin of victory. And the bluest state in the nation continues to be not a state at all: the District of Columbia. One more note: maybe I just failed to notice this functionality in the other maps but the Wall Street Journal’s is the only slippy map in the list. That is, you can freely pan and zoom on this map by using your scroll wheel and then clicking and dragging.
WNYC is a smaller group, but their map is not too shabby at all! The crazy thing here is its emphasis on DIY demography. Are those categories vindicated by any sort of larger theoretical framework or empirical study? Wherever they came from, it’s pretty neat to see a map frame the election results in such a different way from everyone else.
Washington Post is another one that works great, but isn’t particularly unique. I do praise its ‘battleground state dashboard’ beneath the map, though, which does a great job of giving you all the info you need and nothing you don’t. All too often the best way to map something is to not use a map at all.
Speaking of battleground states, CNN.com is the only map to visually highlight the swing states right on the map itself, giving them a bright yellow outline. Clicking on a state leads to the usual county-level view, but also some very cool and thorough exit poll data. I could also do an entire analysis on the use of callouts to solve the problem of representing our geographically tiny east-coast states, but maybe I should stow that for some other time.
I know I’ve said that ‘I don’t anything to remark on’ for a lot of these maps, but Fox News is something I really have nothing to remark on. Don’t get me wrong, it functions just fine, it’s just there’s not a lot of fancy bells and whistles to discuss. Actually, I do have one thing to note: they have by far the most thorough exit polling data out of any of the other maps shown here, but it takes a lot of clickthroughs to get there. Click on a state, then on the thumbnail map that shows up on the right (which opens up a new page), then on the exit poll button (also a new page). Tighten up that UI!
The New York Times gets my award for coolest unique data vis thing: its ‘shift map’ shows how each county in the country has moved relative to how it voted in 2008. There’s a lot of long red arrows near the Mason-Dixon line, indicating that this region has either flipped from Democrat-leaning to Republican, or their Republican leanings has become much more entrenched. (Based on the numbers you get by clicking on the counties, the answer seems to be the latter.) You might also notice the curious case of Wilson County, Kansas, where 46.7% of votes swapped from Republican to Democrat. What’s going on in there?
Finally, I must give kudos to bl.ocks.org, for offering the first post-modernist election map I have… ever seen, actually. Speaking philosophically, this one bears its epistomological argument quite threadbare.
If I missed any major maps, or if you want to remark on which ones you liked, be sure to tweet @maphugger!
I put a question mark after ‘practical’ because I’m not even sure tangents even exist in cartography the same way they do elsewhere. So consider this more of an exploratory thing, rather than straight design advice. But I’ll try to provide some conclusions as appropriate.
But I’m getting ahead of myself: let’s define what a tangent is first. Rather than try to explain it myself, I’ll refer you to this write-up by Chris Schweizer. I won’t try to duplicate it here. Go read it, then report back here. I’ll wait.
So yes, the idea of tangents were derived from illustrators and cartoonists. It occurs when lines intersect in such a way that makes its constituent objects challenging to reify— generally because their edges bump up against each other. Ultimately, our brain can decide on the objects’ relationship without too much effort, but that moment of hesitation is mentally translated into ugliness! Or that’s the theory, anyway.
Left: A square with a circular hole in the middle. Middle: A square with a circle above it. Right: A figure that’s ambiuguous because the square and circle form a tangent.
To test out the role of tangents in cartography, I tried to create the tangent-iest map I could make. The result is below!
My first impression is that it actually isn’t as ugly as I would’ve expected. The corner tangents (where the roads intercept the vertices of the stars) are particularly inoffensive. That legend is pretty ugly, though. Let’s fix the tangents and see if there’s much improvement.
There’s two ways to fix a tangent: first, separate the two objects from each other. Second: (somewhat counter-intuitively?) make the objects explicitly overlap each other. I’ve employed both solutions, as appropriate.
The biggest gains here, I think, come from the legend: giving those elements some breathing room makes it look a whole lot better. One problem that occurred, though: because of the complexity of the basemap (namely, all those roads), it was really hard to move elements around without forming further tangents.
So, the takeaway: tangents probably aren’t as big a deal in cartography as they are in other graphics— maybe it’s because we deal with more visual elements, or maybe because we have a more sophisticated visual hierarchy than, say, a comics panel. However! Items high up on that visual hierarchy are more prone to creating some really gross-looking tangents, so take care when placing your legend and map boundaries and what-not. I suspect most of us are doing this already, though. Maybe now at least you know what the problem is called, and why it looks as ugly as it does.