The land surface types shown in the map come from the classification system of the National Land Cover Database. That database is generated by a consortium of federal agencies that use satellite data to create a “nationally complete, current, consistent, and public domain” data set describing land cover throughout the United States. As such, it’s the definitive standard for this kind of analysis.
And using this data, you can easily see massive urban areas, like the greater Chicago metropolitan area, with 9 million people and eight predominantly urban districts. Or the Dallas Fort-Worth metro area, with 8 million people and ... just four urban districts? There are also metro areas that cross state borders, like Kansas City, taking in both Missouri’s 5th Congressional District and Kansas’ 3rd. But wait: Is only the Missouri side chiefly urban?
Out West, states often have one major metropolitan area with a district (or districts) unto itself, surrounded by more rural districts, like Seattle, Portland, um, Las Vegas, and… uh… Salt Lake City.
Sometimes things don’t turn out like you expect.
Sometimes you can’t see the city for the trees
With four types of developed land cover, a non-urban type can predominate in a mostly urban district if it is dominant among the non-urban land cover. For example, Pennsylvania’s 12th, a Pittsburgh-based district shown just to the right, appears to be about half urbanized, but the predominant land cover is deciduous forest.
That’s because the vast majority of non-urban area is deciduous forest. It’s like a primary where you have one moderate Democrat winning with just 30% of the vote against four progressives.
In California’s 31st in the eastern San Gabriel Valley, this is even easier to see. The district is more than half urbanized. But it also has a significant chunk of non-urban area, and of that, almost all of it is shrub/scrub. The urban area, meanwhile, has to “split the vote” among four types of land cover.
The map at the top of the post would look a bit different if you consolidated the four types of developed land cover. Overall, of the nation’s districts where the predominant land cover is not one of the four urban types, there are actually six districts where the majority of the land is in fact developed, and another five where the urban area makes up 40-50% of the district.
There are sticky little gerryprints all over this map
But there’s another reason that urban areas aren’t showing up where you’d expect them: gerrymandering. And this reason applies to all the seemingly “missing” urban districts in the introduction. So this map at the top of the post isn’t just for fun after all; it’s actually useful for illustrating the “cracking” of Democratic strongholds—usually done by Republicans to diminish Democratic voting strength, but sometimes done by Democrats to spread their voters out more evenly.
Let’s take a look at Utah. Here are the land surface types for the Salt Lake City area:
The urban area is neatly cut into four parts, diluting the voting power of urban citizens into four large conservative districts with plenty of sagebrush. And the result? Four Trump districts. As you can see on the map below (larger version here), this shows how the 2020 presidential election would have gone using the nation’s new districts, according to our new calculations:
The most dramatic example we can find is in the Lone Star State, thanks to the nefarious slashmaps who gave us the Texas chainsaw gerrymander. Let’s look at the Dallas-Ft. Worth area:
No fewer than 12 districts claim a good chunk of the region in an absolute nightmare of convoluted lines. The majority of them snag a piece of urban environs and then go snaking out into adjacent rural areas to minimize the strength of Democratic voters in the cities and suburbs.
Extracting the state of Texas from the two hexmaps above and showing only the Dallas-Ft. Worth dirty dozen, you can see that the GOP’s gerrymander is quite successful: three very Democratic districts, surrounded by Republican strongholds whose predominant land cover is, in all cases but one, not urban.
Every map involves trade-offs. Traditional maps of congressional districts, as we alluded to at the top of the post, make sprawling rural districts too prominent while making small urban districts almost unreadable—even though they all have roughly the same population.
Our hexmap solves that problem, but it creates another that we readily acknowledge: Districts sometimes cannot be placed within states where you’d expect them to be geographically. As shown just above, for instance, the districts that make up the Dallas area stretch all the way to the Texas panhandle. There just isn’t enough room to show them all within the DFW Metroplex and still have them be visible.
There’s truly no way around this if you want the states to remain recognizable and retain their shapes. The only alternative is to badly distort state outlines, which yields a hard-to-parse blob-like map such as the one on the right at this link. (It’s of Texas counties, rather than congressional districts, but the effect would be similar either way.) Our preferred approach when we use our hexmap is simply to release a traditional map alongside it. You can find one of land cover right here, though you’ll notice right away that the developed regions barely stand out.
Notes on the new hexmap release
This is version 3.0 of the congressional district hexmap, valid for 2022. Files (shapefiles, svg, and png) can be downloaded here; prior versions will remain available. You are free to use the maps as you wish as long as you cite us and link back to the source files (https://dkel.ec/map).
Version 3.0 is in the same style as Version 2.0 and 2.1, updated following redistricting.
New for version 3.0 is a format that should fix the “squished” look some users were getting. Using the Web Mercator shapefile will hopefully eliminate that problem in most cases.
For a peek behind the curtain, this Twitter thread shows a bit of the process of making this map. The reasons for making the map in the first place are discussed here, and the rationale behind the current style can be found here.
Thanks to my colleague Stephen Wolf for looking carefully at each and every district and suggesting numerous improvements.