The rural-suburban-urban divide in U.S. politics is becoming increasingly salient for more and more Americans. All you have to do is turn on cable news or peruse a popular press article to hear or read about the deepening geographic division among American voters. Social media is awash in conversations, accusations, and prognostications among elites and non-elites alike about the depth of the divide (is it political or social or both?) and how it is changing our politics.
Academia is also not immune to pondering this seemingly new phenomenon. In recent years, social science scholars—including Kathy Cramer, Robert Wuthnow, Kal Munis, and Nicholas Jacobs among others—have begun to tease out the dynamics of place-based consciousness/identity and how it affects Americans’ political behavior. Their studies demonstrate that this divide is genuine and not a figment of the pundit class’ collective imagination.
The question is, then, how exactly did aggregate rural/urban-ness affect the most recent presidential election? I will not bore you with statistical minutia (you can scroll down to the regression table at the bottom of the page if you choose to drown in minutia), but the take-home message of my analysis is that the impact of the rural-suburban-urban divide was real.
On second thought, brace yourselves for some social science gobbledygook as I explain: Using a USDA categorization of county-level rural/urban-ness (with 1 being completely urban and 9 completely rural), the model calculated a 0.5-point difference in Donald Trump’s county-level support moving up and down the scale level-by-level (see the figure at the top of this post). For example, the analysis predicts completely rural counties supported Donald Trump 3.9 points greater than completely urban counties, holding any racial/ethnic, economic, household income, educational attainment, or pandemic-induced differences constant.
And, yes, this is admittedly a small number; however, given the tight election margins we are accustomed to seeing now, any single element that has this much of an independent impact is important to understand better. Also, do not let the specific predicted percentages distract you; they are simply a function of the prediction specification. Let me demonstrate with two more examples.
You could, for instance, ask what is the impact of rural/urban-ness on election results in counties that are economically dependent on the mining industry (e.g. oil and gas, coal extraction, etc.)? Or put differently, does a county’s location matter if its economy is dominated by oil and gas, or does the economic question trump all other factors influencing the election (no pun intended)? If you look at the Figure 2 below, rural/urban-ness still clearly matters.
The analysis predicts that mining intensive counties (red line) are across the board 11.4 points more supportive of Donald Trump than counties whose economies are dependent on other industries (blue line), regardless of their locations. This makes sense given the rhetoric heard during the campaign around fracking and coal mining, with Donald Trump signaling greater support of these industries than Joe Biden.
But rural mining intensive counties are even more supportive of Donald Trump than urban mining intensive counties. This difference in Trump’s support between completely urban (think: Harris County, Texas) and completely rural counties (think: Fallon County, Montana) is again 3.9 points.
Another way to demonstrate the impact of the rural-suburban-urban divide on the 2020 election is to examine the electoral dynamics in majority-Hispanic counties. Hispanics as a voting bloc—albeit one with vast internal diversity—featured heavily in media stories about their potential emergence as a new swing voter group. Whether this characterization is accurate or not is beyond the scope of this article, but it does raise an interesting question: What did the Trump-Biden race looked like in counties populated by large numbers of Hispanics, and did rural/urban-ness play any part in these election results?
As you can see in Figure 3 above, the impact of the rural-suburban-urban divide is not unknown in majority-Hispanic and near-majority Hispanic counties. For both rural and urban counties, the analysis predicts that as they become increasingly Hispanic, from 40% to 70% of the county’s population, the aggregate electoral support of Donald Trump declines. This difference in support of Donald Trump’s candidacy between 40% Hispanic counties and 70% Hispanic counties is 5.67 points (comparing the space between the blue and orange lines). Generally speaking, this squares with the conventional story told in the media about Hispanic voting behavior, with Cuban Americans in South Florida being the exception.
But again, the analysis predicts rural majority and near-majority Hispanic counties to be more supportive of Donald Trump than demographically similar urban majority and near-majority Hispanic counties. In other words, the 3.9-point difference between completely rural and completely urban counties exits in these electoral results as well.
So, in the end it fair to say that rural matters. The rural-suburban-urban divide exists and will most likely continue to affect future elections, especially if the two major political parties stick with a similar message and approach to rural voters.
I end with this caveat: This analysis was based on the results of one election using county-level data. I cannot say how location influenced an individual’s decision to support Donald Trump or Joe Biden. But there is clearly a trend in both the aggregate data and individual-level data (see the scholars’ studies that I mentioned above) to suggest that the rural-suburban-urban divide is here to stay.
Daniel “Ben” Bailey, PhD, is assistant professor of political science at Presbyterian College in Clinton, South Carolina.
Regression Table
| Table 1: | (Model 1^) |
| Dependent Variable: % Republican 2020 | |
| Rural/Urban Continuum Code | 0.00478*** |
| (0.000839) | |
| COVID-19 Deaths, Cumulative | -0.00000598 |
| (0.00000795) | |
| % BA, 2018 | -0.876*** |
| (0.0323) | |
| Unemployment, Sep. 2020 | -0.000000278 |
| (0.000000185) | |
| Economic Dependence: Farming# | 0.0476*** |
| (0.00618) | |
| Economic Dependence: Mining# | 0.0584*** |
| (0.00729) | |
| Economic Dependence: Manufacturing# | -0.00453 |
| (0.00449) | |
| Economic Dependence: Federal/State Gov# | -0.0158* |
| (0.00639) | |
| Economic Dependence: Recreation# | -0.0533*** |
| (0.00716) | |
| % Hispanic, 2019@ | -0.189*** |
| (0.0175) | |
| % Asian, 2019@ | -0.830*** |
| (0.0868) | |
| % Black, 2019@ | -0.478*** |
| (0.0142) | |
| Median House Hold Income, 2018 | 0.00000154*** |
| (0.000000216) | |
| _cons | 0.809*** |
| (0.0119) | |
| N | 3110 |
Standard errors in parentheses
* p < 0.05, ** p < 0.01, *** p < 0.001
^ Unit of analysis: US county election returns, 2020 presidential election
# Comparison category: nonspecialized counties
@ Comparison category: % White, 2019
Republish our articles for free, online or in print, under a Creative Commons license.
