Suburbia has repeatedly cropped up in the headlines this election season, ever since a more-bizarre-than-usual Donald Trump tweet-pitched to the “Suburban Housewives of America,” pledging to protect them from invasion by evildoers.
Like most things he believes, of course, Trump’s views are divorced from current reality about the suburbs. His image of an idyllic land of refugees from the scary big city—one that’s uniformly white and where the women are all married and don’t participate in the labor force—doesn’t jibe with today’s suburbs. Today, the suburbs are just as diverse as the country as a whole, and a fair number of the residents live there not because they’ve fled the city, but because they’ve been priced out of the city.
Trump, however, is hardly the only public figure to misunderstand who lives in the suburbs. Part of the problem, though, is that there’s little agreement on even the fundamental question of what a “suburb” is in the first place, as well as where suburbs begin and end, both in terms of what’s urban versus suburban, and what’s suburban versus rural.
And that’s a problem, because ever since Trump’s election, there’s been a huge focus on how once-Republican suburbia—especially its most affluent and well-educated parts—has galloped away from the GOP. But where has this phenomenon had the greatest impact?
To help fill this gap in a way that’s of particular use for understanding elections, Daily Kos Elections has applied a new model developed by the Department of Housing and Urban Development to assess how urban, suburban, and/or rural every congressional district in the nation is. Click here for our complete data set, which we’ll discuss in detail below.
Usually the Census Bureau is the ultimate argument-settler in demographic questions like this one, but, surprisingly, the census doesn’t even define “suburb,” let alone try to measure who lives there. Relying on a combination of density and land cover, the census breaks the country down into "urban" and "rural" without any “suburban” category. It considers the nation’s population to be 81% urban without any further gradations beyond that.
However, we’re going to look at several new attempts to remedy that lack of subtlety. One is the recent HUD study mentioned above. The other is a model of our own design focusing on county- and city-level classifications from the Office of Management and Budget. Click here for this second spreadsheet, which we’ll also discuss in more detail shortly.
A number of other analyses have delved into this thorny topic in the past. Pew Research, for instance, has polled a number of times on whether people consider themselves to live in urban, suburban, or rural areas, finding in 2018 that 55% of Americans view themselves as living in the suburbs. However, this national sample isn’t broken down into smaller geographic units like congressional districts or counties.
There are also several federal governmental agencies (like the National Center for Health Statistics and the Department of Agriculture’s Economic Research Service) that have attempted to classify counties at a more detailed level than just “urban” or “rural.” However, their classification schemes don’t actually make reference to “suburbs,” instead using intermediate buckets like “large fringe metro” or “metro, pop. <250K.”
The American Communities Project, a nonprofit based out of George Washington University, has used a statistical technique called “cluster analysis” to sort the nation’s counties into an even more nuanced set of buckets: There are 15 of them, including “urban suburbs,” “middle suburbs,” and “exurbs.” Their classification framework is well worth exploring (their website has a terrific interactive map), but their scheme operates at the county level, not congressional districts.
In 2018, City Lab’s David Montgomery developed a similar cluster analysis technique, but focused purely on the urban/rural spectrum, unlike ACP’s model, which looked at a variety of race, age, and economic characteristics. Montgomery’s model reduced the universe to only six buckets, ranging from “pure urban” to “pure rural” with stops like “dense suburban” and “sparse suburban” along the way. Importantly, he applied his technique to congressional districts, giving analysts a new way to examine and classify elections.
Montgomery’s approach involved looking at the density of each individual census tract (one of the smallest, most granular units of data available), then assembling the constituent census tracts into their respective congressional districts and seeing which levels of density were most prevalent. A number of other pundits who write about politics have referenced Montgomery’s work, and it’s become something of a gold standard for quantifying the relationship between density and politics. Our effort here, in fact, aims to complement and add to his work, rather than supplant or critique it.
The most recent contribution to this type of analysis from June of this year was published by HUD and reflects the work of a trio of social scientists—one at HUD, one at the Census Bureau, and one in the private sector. It combines techniques from several of the above approaches into something of a best-of-all-worlds method.
The HUD approach starts with surveys asking people whether they think of themselves as living in an urban, suburban, or rural environment, but then uses a statistical model layering in other demographic characteristics to estimate how many people within each of the nation’s census tracts would think of themselves as urban, suburban, or rural. Their findings closely mirror Pew’s surveys: They find 52% of Americans are suburban, similar to Pew’s 55%. The advantage, though, is that unlike Pew, HUD has broken the results down to the near-atomic level.
Of course, most people don’t conceptualize the world around them at the census tract level; there are over 73,000 census tracts, and even the most dedicated election nerd doesn’t go around thinking, “Yeah, Cook County, Illinois tract 8457 is trending really dark blue.” However, census tracts are the fundamental building block for all demographic analysis, and if you have the tools and patience, you can assemble those 73,000 jigsaw puzzle pieces into something more useful … like congressional districts, as we’ve done here.
Under this framework, mirroring the fact that majority of the country is suburban, so too are the majority of all congressional districts. There are 81 districts that are predominantly rural, of which only 12 are currently held by Democrats, and generally only either in the Northeast or in districts where whites are in the minority. There are even fewer districts—just 55—that are predominantly urban, of which a grand total of zero are currently represented by Republicans. The remaining 299 seats are all predominantly suburban.
Using HUD’s scheme, the most “urban” district is New York’s 13th (in Harlem and the Bronx), at 97.3% urban, while the most “rural” district is Kentucky’s 5th, in the Appalachian eastern part of the state, at 74.2% rural. The most suburban district is Illinois’ 6th, in the suburbs to the west of Chicago, at 92% suburban. You can visualize this last district as the place where every John Hughes movie in the 1980s was set. And interestingly, in a microcosm of the political shifts taking place in suburbs across the country, Democrats flipped this seat in 2018—and Republicans have almost no hope of taking it back this year.
We’ve also added a separate tab performing the same analysis but but applying it only to the white population of each congressional district. In a rebuke to Trump’s assumptions, white people don’t live in the suburbs at a disproportionate rate: 54% of white Americans live in the suburbs, not much different than the 52% of the total population that lives in the suburbs.
Another way of looking at it, by reversing the numbers, is that 62% of the suburban population is white, which is hardly different than the 61% of the country’s total population that’s white. Again, the suburbs are a very representative cross section of the entire nation. Where the numbers get different are in cities and rural areas: Only 45% of urban dwellers are white while 79% of rural residents are white. That, however, shouldn’t really be a surprise if you’ve ever looked at a map of election results.
Finally, we’ve also included a sheet for the urban, suburban, and rural status of all of the nation’s state Senate districts. While you may not be as initially familiar with these districts as you are with the nation’s congressional districts, this may be helpful for drilling down to more granular parts of any state you find especially interesting. (We haven’t done state Houses, though, because of their small size.)
But wait! That’s not all you’ll get. We’ve also put together our own framework that adds some additional buckets in between “suburban” and “rural” to try and add a little more nuance to the conversation. The dividing line between suburban and rural is often pretty porous: On the one hand, you often see low-density rural-looking areas that are still part of counties in a major metropolitan area; they’re just on the outer periphery of those counties, but still within commuting distance of the big city.
Conversely, you often also see small- to medium-sized towns that aren’t anywhere near major cities, but are fairly dense and, at least within their smallish municipal boundaries, not something you’d immediately think of as being “rural.” These latter two categories are what people tend to think of as being “exurban” and “micropolitan,” and we’ve tried to add those into the mix.
In doing so, we’re relying on the Office of Management and Budget's classification scheme for whether counties are metropolitan, micropolitan, or neither. You might occasionally see reference to, say, the “Gotham Consolidated Statistical Area,” or the “Townsville Metropolitan Statistical Area,” or the “Dogpatch Micropolitan Statistical Area.” These complicated-sounding names come from the OMB, though they rely heavily on census data on density and commuting patterns in developing them.
We’ve divided the country’s population in five ways according to the following method:
- Urban: Population within the city limits of the named cities of each metropolitan statistical area (MSA). (For example, New York City is in the New York-Newark-Jersey City MSA; so anyone in these three cities would be “urban,” even the people who live in the fairly low-density and suburban-looking south shore of Staten Island.)
- Suburban: Population of the central counties within each MSA, minus anyone living in the named cities. (For example, everyone who lives in Hudson County, New Jersey, would be suburban, except for those who live in Jersey City. Unfortunately, this excludes people who live in dense suburbs that are too small to be named cities, with Hoboken being a prime example. In contrast, everyone who lives in Bergen County, New Jersey, would be “suburban” since none of the MSA’s named cities are in that county.)
- Exurban: Population of the noncentral counties within each MSA, also minus any named cities. (There is only one noncentral county in the New York MSA, and believe it or not, it’s in Pennsylvania—Pike County. A better example might be the Indianapolis-Carmel-Anderson MSA, which encompasses 11 counties but four of them—Brown, Madison, Putnam, and Shelby, the ones furthest from the city center—are considered noncentral and thus exurban.)
- Micropolitan: Population of all counties in MSA. (Micropolitan SAs are a separate category, with no overlap with metropolitan SAs. The criterion is they have an urban core of at least 10,000 people but under 50,000, so much of what we think of as small-town but not wholly rural America lies in this category. Micropolitan SAs that are around the national median size for micropolitan SAs include places like Hillsdale, Michigan; Gillette, Wyoming; and Moultrie, Georgia.)
- Rural: Population of counties that are neither metropolitan nor micropolitan. (While this includes many hundreds of counties, many of them have only a few thousand residents, so it’s a small percentage of the nation’s total.)
This model yields very similar urban and suburban numbers to the survey-based HUD model: The HUD model says the nation as a whole is 26% urban and 53% suburban; our approach says 29% urban and 49% suburban. The added nuance mostly comes at the more rural end of the spectrum: Instead of 21% rural like the HUD model, our model says the remainder is 6% exurban, 9% micropolitan, and 6% rural. In other words, the reality is that the large majority of the nation lives either in big cities or in close proximity to them.
One tradeoff for the greater precision on the rural end of the spectrum is a bit less nuance in the vast middle of suburbia; our model, for instance, counts 34 different congressional districts as 100% suburban, while the HUD model doesn’t rate any districts that way. Much of the difference, though, comes around the margins. For example, Georgia’s politically competitive 7th District is 100% suburban in our scheme but 89.6% suburban in the HUD model, so that difference may not have a lot of real-world importance. One potential disadvantage of our model is that unlike the HUD approach, it doesn’t ask what people in the district might actually think; it just asks what the federal government thinks about each place.
By contrast, it may give a more accurate read on a mostly rural district like Kentucky’s 5th. While the HUD model says that this district is 74% rural and 19% suburban (and 6% urban), even that low 19% seems odd when there aren’t any metro areas centered in the 5th, which only takes a peripheral bite out of the Huntington, West Virginia, MSA. Instead, our model says the 5th is 56% rural. That still makes it the nation’s most rural district, but it’s also 39% micropolitan (which reflects that a lot of its population is in towns like Middlesboro and London that aren’t very populous, but still are the centers of gravity for the district’s economy), 4% exurban, and 1% suburban.
But again, that isn’t to say that the Daily Kos Elections model is superior to the HUD model. Really, what you should do is view the two together (and in conjunction with City Lab’s model, too) to get a really holistic, multifaceted view of the composition of each unique congressional district.