Hexes, Tiles, and Districts
A conversation with Daniel Donner on designing a cartogram of United States congressional districts
Recently, my team at Graphicacy collaborated with Grassroots Analytics to build a tool called PrimaryCast, to display Grassroots’ predictions for who might win upcoming Democratic primary races. As the designer on the project, I led the discovery process, where we explored various ways to display this data to users.
We knew we wanted the primary visual to be a map. While the statewide races (such as Senate and Governor races) were more straightforward and could be displayed with a standard U.S. state choropleth map, displaying House races proved much more difficult. For the purposes of representation in the House of Representatives, states are divided into multiple congressional districts (CDs). In order to ensure each CD represents roughly equal numbers of people, they vary widely in their geographic extent. Rural CDs can be large, as in Montana, where the entire state is one CD. On the other hand, urban CDs can be small, as in the New York City metro area, which comprises 17 CDs.
We wanted a map that was interactive and explorable, where the small urban CDs were just as easy to find as the large rural CDs. That quickly led us to tiled cartograms, a cartography method where a map is constructed out of equal-sized tessellating shapes. Rather than land area, another metric is used to determine size (such as population, or electoral college votes). After reviewing several cartogram methods for mapping CDs, we found a particularly nice example put together by the elections data team at Daily Kos. I became fascinated with the various ways to tackle the same mapping and design challenge and reached out to Daniel Donner, who created the map, to ask him about his process.
Below is our conversation, edited for length and clarity.
Origins and Inspiration
Carni Klirs: Can you introduce yourself and how you ended up doing what you do?
Daniel Donner: My name is Daniel Donner, and I’m a contributing editor at Daily Kos Elections. I work mainly on collecting, interpreting, and presenting elections-related data. I’ve been making elections maps since 2005. I started with a map of President Bush’s approval ratings in November 2005 and it blew up — to my complete surprise. I’d found there was a demand for this sort of thing. And since then I’ve just worked on it more and more and taught myself the basics, especially once QGIS came online, which was awesome.
CK: Why use a cartogram to map congressional districts?
DD: So the problem is, how do you display anything to do with congressional districts in any sort of map form that people can understand? Because if you use a map with standard land-based geography, you’re not going to be able to see any urban districts, and urban districts represent a huge number of people, of course. It was the 2015 UK election, when we saw The Guardian’s hexagon-based map, that we were inspired, and we thought, “Wow, we know we can do that for the United States. Why aren’t we doing that for the United States? Why is nobody else doing that for the United States?” And, actually, it turns out, now I know that there is a long tradition of cartograms for elections in the UK. Just as there is in the United States for the electoral college. But not so much for congressional districts. So our map is part of a long tradition of showing election results by constituency.
Iteration and Compromise
CK: What kind of reception did you get for your first version of a CD map?
DD: So people were kind of intrigued at first when we put it out, back in 2015. At the same time, there’s lots of joking around and lots of criticism. Constructive criticism, thankfully. And then other media outlets started adopting cartograms for congressional districts, too. Now everybody has their own version, which is great. It’s awesome to see all these different things out there and especially I see the different ways that people attack the problem. It comes down to what compromises you want to make. You can’t make a perfect map. You’re not going to be able to get it so that everybody can understand the entire concept at first glance, because it’s a bit of a tricky concept if you’re not used to it already.
CK: How has your map evolved?
DD: While our current map is noncontiguous in order to preserve state shapes, our first version was contiguous. The logic behind that was, when you have one contiguous mass, it’s a little easier to identify the proportion of colors within that mass. And we also had fun with the concept that the Yellowstone super-volcano blew a hole in the middle. But then, with feedback, we realized we really needed state labels, and there wasn’t any place to put labels. In the original version the state shapes were not always readily identifiable so you needed labels unless you sat there and studied it for a while.
So, for Version 2.0 of the Congressional District Map, we decided to switch over to noncontiguous, so we could fit in labels to help people identify their states right away. We also thought it was more pleasing to the eye, to be honest; it’s a more familiar form, less abstract. When people look at a map, one of the first things they want to do is find themselves, right? And when you use the familiar state shapes that makes that really easy. Of course, the drawback is they see the state shape, then they think, well, the districts must be in their correct relative positions within a state. And so every time you post a map like this on Twitter you get one or two comments that say, “You put that district in the wrong place, there’s only one district in the Upper Peninsula of Michigan.” I think what many people do is they don’t pay too much attention to the words that accompany a map when it’s such a familiar form — which makes sense, really. So no matter how many caveats or explanations or little diagrams we add in, somebody’s going to miss them.
CK: Well, do you think that the weirder maps are, the more it forces someone to have to stop and read the text to understand it?
DD: Yeah, probably. You have to process it, think through it a little bit more.
Hexes vs. Squares
CK: So your map is tiled together from hexagons. I’ve seen other cartograms using square tiles. Can you discuss your decision for which basic shape to use for a cartogram?
DD: So hexagons are more flexible for conforming to irregular shapes, as they can make pseudo-curve-ish shapes. So that’s why we went with hexagons, as we were trying to preserve identifiable state outlines. Squares, obviously, are more simple and a little more intuitive, if you’re talking about counting units. If you have abstract state shapes then squares are a really good way to go, because then you’re just counting up units within the state. You don’t have to put them in any sort of geographical order. And it makes sense to people, since they’re already seeing that it’s an abstract map, they’re not expecting the districts to be in their proper position, either.
CK: Why did you decide to make the maps open source and how do you keep them up to date?
DD: Well, we try to make as much of our data available to as many people as possible. Because we think it’s useful data and we want everyone to use it if they want to. So this is just part of that ethos. As far as the evolution, we have to tweak it all the time, with redistricting lawsuits every year. So we throw up the files on the page and try to make sure that people understand what’s new and what’s old and leave it at that.
CK: What’s the next major round of redistricting that you’re expecting to happen?
DD: The new 2020 census data coming out, hopefully in a few months, will result in reapportionment of congressional districts among the states, and congressional redistricting within nearly every state, and most states will have to be changed on our map. Because within each state we also try to arrange the districts as best we can, even if it is rough at times. So, when states change their districts, we will also change our map, even though they may have the same number of districts.
CK: Thanks so much for this deep dive into mapping Congressional Districts! Before we wrap up, any recent projects you worked on that you’d like to plug?
DD: Last year I made the Congressional District Atlas, which I’ll be updating as soon as we finish calculating presidential election results by congressional district. It shows standard maps for every congressional district, as well as demographic and physical geography maps. And then I also visualized the data that we tend to use most frequently, and made a static snapshot, in order to make this data available to everybody in a fairly easy-to-understand format.
Carni Klirs is a Washington, DC based designer and musician, currently the Senior Information Designer at Graphicacy, where he creates interactive, data-driven applications for mission-driven organizations.