The USA has an established urban–rural division in numerous aspects of political behavior and public opinion, including vote choice (Scala and Johnson, Reference Scala and Johnson2017; Rodden, Reference Rodden2019; Johnson and Scala, Reference Johnson and Scala2022), partisanship (Gimpel et al., Reference Gimpel, Lovin, Moy and Reeves2020), anti-establishment candidate support (Cramer, Reference Cramer2016), racial resentment (Nelsen and Petsko, Reference Nelsen and Petsko2021), and willingness to protest or put up a political sign (Lin and Trujillo, Reference Lin and Lunz Trujillo2022). Furthermore, the urban–rural political division can be found in other countries, such as Canada, the UK, Denmark, Germany, and more (Ford and Jennings, Reference Ford and Jennings2020; Armstrong et al., Reference Armstrong, Lucas and Taylor2022; Huijsmans, Reference Huijsmans2023).
Given the intertwining of political affiliation and urban–rural designation, it could be that rural political attitudes are simply a reflection of the geographic distribution of partisanship across the urban–rural spectrum. In the USA especially, partisanship is a predominant force in public opinion, and Americans increasingly live in places that mostly share their political views (Brown and Enos, Reference Brown and Enos2021). In other words, some existing literature implies that the average policy stances of partisans should be the same regardless of respondent location (Bishop, Reference Bishop2008; Hopkins, Reference Hopkins2018), with context itself playing a minimal role in determining individual issue stances. Conversely, other work suggests that rurality is more strongly conservative on at least some issue stances (Fennelly and Federico, Reference Fennelly and Federico2008; Fudge, Reference Fudge2020), controlling for partisanship.
How distinct, then, are rural attitudes from Republican attitudes on policy issues? Are rural attitudes simply Republican attitudes? Using survey data on American adults from the 2020 American National Elections Study (ANES) (N = 8280), we examine this question by evaluating average respondent support for 22 policies along a number of issues, by geographic location (urban, rural) and partisanship (Democrat, Republican). First, we find that rural Democrats comprise a significant proportion of rural Americans; the proportion of rural residents who identify as Republican versus Democrat is around approximately three to two. Further, the number of rural Republicans and urban Republicans are approximately the same. These descriptive numbers imply that equating rural with Republicanism is potentially problematic, and that researchers and the broader public should not underestimate the prevalence of rural Democrats and urban Republicans.
Second, we find that partisanship (and, to a lesser extent, other demographic factors), accounts for a substantial share of most urban–rural division in political issue stances. In general, rural Democrats are just as liberal on most issues as their urban counterparts, and urban Republicans are just as conservative on most issues as their rural counterparts, once other demographic factors are accounted for. Rural issue attitudes are therefore not necessarily Republican attitudes; rather, rural issue attitudes tend to be partisan attitudes. With that said, four policy stances on immigration, transgender individuals, and income inequality are an exception; for admitting more refugees, providing pathway to citizenship for undocumented immigrants, allowing transgender people to serve in the military, and supporting policies to lessen income inequality, rural Democrats are less supportive compared to urban Democrats on average, but are much more supportive than urban and rural Republicans. This aligns with previous work in the USA linking urban–rural residence with immigration attitudes (Fennelly and Federico, Reference Fennelly and Federico2008) and LGBTQ policy issues (Thompson, Reference Thompsonn.d.).
These results imply that across several major national issues, urban–rural division is essentially reflective of partisan division. However, we find some issue areas that are not; we speculate these areas still hold urban–rural differences due to the demographic composition of urban areas—thus creating contextual pressures on opinions—and on the more immediate experience of economic inequality in urban areas. Further, there may be other issues not examined here that lend themselves to a more specific urban–rural split, echoing scholarship emphasizing the importance of place in politics (Munis, Reference Munis2022; Borwein and Lucas, Reference Borwein and Lucas2023). Future work should further examine the complexity that certain elements of urban–rural division are the result of the partisan nationalization of politics, while others reflect the relevance of local and place-based considerations. In addition, it is inaccurate to assume rural interests are necessarily Republican, as many rural residents who identify as Democrat hold similar issue stances to their non-rural counterparts.
1. Literature review
Existing literature on the US public suggests that rural issue attitudes are overwhelmingly Republican and conservative. Much of this assumption stems from election results, where rural areas have increasingly been more supportive of Republican candidates over time (Scala and Johnson, Reference Scala and Johnson2017; Rodden, Reference Rodden2019; Gimpel et al., Reference Gimpel, Lovin, Moy and Reeves2020; Johnson and Scala, Reference Johnson and Scala2022). This urban–rural split may have stemmed from geographic sorting along partisan lines, where urban centers increasingly attracted either left-leaning individuals or demographic groups that tend to support Democrats (Bishop, Reference Bishop2008). Others find no partisan sorting effects along geographic lines (Mummolo and Nall, Reference Mummolo and Nall2017), alongside other critiques of the geographic sorting argument (Darmofal and Strickler, Reference Darmofal and Strickler2016). Additionally, some scholars find evidence of contextual effects where specific locales themselves foster tendencies toward the right (or left) (Martin and Webster, Reference Martin and Webster2020).
Other work finds that urban–rural residency is significantly associated with political outcomes, including certain issue stances, controlling for political affiliation. There are consistent urban–rural differences in immigration attitudes across time and throughout Western democracies. In the USA, rural residency predicts support for restrictive immigration policies than suburban and urban residents (Fennelly and Federico, Reference Fennelly and Federico2008). The relationship between immigration stances and rurality may be due to populist-related impulses (Lunz Trujillo, Reference Lunz Trujillo2021; Huijsmans, Reference Huijsmans2023) and values-based differences like multiculturalism (Fennelly and Federico, Reference Fennelly and Federico2008) or cosmopolitanism (Maxwell, Reference Maxwell2019). In addition, rural areas in the USA are less supportive of government spending than their urban counterparts, at least up until 2008, controlling for ideology (Fudge, Reference Fudge2020).
That said, political attitudes have become increasingly nationalized and subsumed by partisan identity. This implies that partisans hold similar issue stances regardless of sub-national context. Politics and political behavior in the USA have become more nationally oriented and homogeneous (Hopkins, Reference Hopkins2018). This has occurred for various reasons, including the decline in local news (Moskowitz, Reference Moskowitz2021). Further, partisanship is becoming increasingly homogeneous with respect to issue stances and ideology (Mason, Reference Mason2018). However, Americans’ issue stances—including their operational ideology—oftentimes do not align with self-reported (or symbolic) ideology (Popp and Rudolph, Reference Popp and Rudolph2011; Ellis and Stimson, Reference Ellis and Stimson2012). Partisanship also dwarfs other demographic and identity-based divisions in values, vote choice, and issue stances (Maxwell, Reference Maxwell2019; Doherty et al., Reference Doherty, Kiley and Asheer2020). Regarding issue stances, people even tend to establish their partisan affiliation first and then subsequently adopt attitudes associated with that affiliation (Achen and Bartels, Reference Achen and Bartels2017). Finally, these other politically relevant demographic and identity-based divisions—such as race, education, religion, rurality, etc.—are increasingly predictive of partisanship, which can be seen as a “mega-identity” that organizes other identities into one of two camps (Mason, Reference Mason2018). Taken together, this body of literature implies that partisanship should determine attitudes in similar ways, regardless of context or locale.
Given the above discussion, we propose a pair of competing hypotheses. The first is a null hypothesis based on assumptions about rural attitudes being more like Republicans’ in nature. Rural residents have been found to be more conservative and right-leaning (Gimpel et al., Reference Gimpel, Lovin, Moy and Reeves2020), and people's issue preferences often do not align with their partisanship (Popp and Rudolph, Reference Popp and Rudolph2011; Ellis and Stimson, Reference Ellis and Stimson2012) despite the increasing ideological homogenization of the two major political parties in the USA. This points to the urban–rural context having an association with certain issue positions above and beyond partisanship. In other words, rural Democrats may hold more conservative stances while urban Republicans would hold more liberal ones:
H0: Rural Democrats will be significantly more conservative on political issue stances than urban Democrats, on average and controlling for other factors. Conversely, urban Republicans will be significantly more liberal on political issue stances than rural Republicans, on average and controlling for other factors.
Alternatively, given the substantial effect of partisanship on political attitudes (Achen and Bartels, Reference Achen and Bartels2017; Mason, Reference Mason2018), alongside the nationalization and homogenization of political behavior (Hopkins, Reference Hopkins2018), we might expect average issue stances of partisans to be the same across the urban–rural spectrum. For this reason, we propose an alternative hypothesis:
HA: Rural Democrats will not be significantly more conservative on political issue stances than urban Democrats, on average and controlling for other factors. Further, urban Republicans will not be significantly more liberal on political issue stances than rural Republicans, on average and controlling for other factors.
2. Methods
2.1 Data
The data for this study come from the 2020 ANES (N = 8280). This is a nationally representative study on political attitudes and behaviors conducted before and after each presidential election in the USA.
For this study, we are interested in items that measure support for one's political attitudes toward pressing issues in this country today. These are (full question wording can be found in the Supplementary materials): transparency for journalists, separating children from parents at the border, requiring COVID-19 vaccinations, requiring identification to vote, regulating greenhouse gas emissions, reducing income inequality, providing paid family leave, providing citizens $12,000 yearly, providing a pathway to citizenship for undocumented migrants, increasing spending on health care, ending birthright citizenship, deporting undocumented migrants, building a wall on the Southern border, banning assault-style rifles, requiring background checks for gun purchases, assault rifle buyback, approving the Affordable Care Act, allowing free trade agreements, allowing felons to vote, allowing transgender individuals in the military, allowing refugees to come to the USA, and increasing efforts to combat the opioid epidemic.
Respondents rate each item on a 7-point Likert scale where 1 is strongly favor and 7 is strongly oppose. For the analysis, we reverse code the responses so that higher scores indicate greater favorability.
We rely on self-reported place of residence, e.g., whether they currently live in a rural area, small town, suburb, or city. Since one's perceived place of residence is often more indicative of politically relevant attitudes and group-based affect than objective designations (Nemerever and Rogers, Reference Nemerever and Rogers2021; Lunz Trujillo, Reference Lunz Trujillo2022b), such measure is suitable for classifying respondents based on where they live (see Supplementary materials for question wording).
To assess partisanship, we use standard self-reported party identification measures, defining Democrats and Republicans as those who are partisan and partisan leaners. Pure independents are excluded from the analyses.
2.2 Analysis plan
To address our hypotheses, we first create measures that represent the interaction of partisanship and residence. We dichotomize the place of residence variable so those indicating that they live in a city or suburb are considered “urban” and those living in a small town or rural area are considered “rural.” We justify combining urban and suburban together because we find that within-party urban–suburban differences in issue stances are largely indistinguishable from one another.Footnote 1 Then, we add partisanship to create four categories representing Urban Democrats, Rural Democrats, Urban Republicans, and Rural Republicans. The weighted distribution of respondents in these categories are shown in Figure 1. Note that approximately one-third of non-independent rural respondents are Democrats, and that there are roughly equal numbers of Republicans in rural and urban areas.
Figure 1. Distribution of respondents from each party-place of residence category.
After generating the place-party measure, we calculate descriptive statistics for each political attitude variable. For the items where there is a statistically significant difference between urban and rural residence for either party, we then compute ordinary least squares (OLS) regression models with the particular attitude item as the outcome variable and predict this using the place-party categorization. In these models, we control for political ideology, support for Donald Trump, gender, income, education, racial resentment, and church attendance.Footnote 2
3. Results
First, we examine descriptive statistics to see how attitudinal positions relate to partisanship and place of residence. From here, we select items to conduct regressions based on those with a statistically significant difference in attitudes between places of residence. Since we know that the parties are likely to differ on these issues, given that they are widely debated in the media and general public, the more interesting difference on this step is the extent to which urban and rural residents within each party differ in their attitudes on each of the issues.
As such, Figure 2 shows the mean and 95 percent confidence intervals for each of the issue positions by place of residence and party identification. Of the 22 issues examined, 14 have a statistically significant intraparty urban–rural splits based on a lack of 95 percent confidence interval overlapFootnote 3 : transparency for journalists (Republicans), regulate greenhouse gas emissions (both), provide paid family leave (Democrats), provide citizens 12K a year (Democrats), provide a path to citizenship (Democrats), deport immigrants to native country (both), build wall on Southern border (both), ban assault rifles (Republicans), assault rifle buyback (both), approve Affordable Care Act (both), allow free trade agreements (both), allow felons to vote (both), allow transgender people to serve in the military (both), and allow refugees to come to the USA (Democrats). Of these 14 issues, ten had intraparty urban–rural splits among Republicans, while 12 had intraparty urban–rural splits among Democrats. Among the significant urban–rural splits, rural attitudes are nearly always more conservative than urban attitudes. Further, rural Democrats were always more conservative then urban Democrats, and urban Republicans were always less conservative than rural Republicans.
Figure 2. Positions of key issues by party identification and place of residence.
Next, we use OLS regression models to predict all 22 issues. The main independent variable is a categorical designation of respondent residency and partisan identity. Figure 3 displays these results, using “Democrat-Rural” as the base category (control variable results are not shown in the figure; see Supplementary materials for full results). Notably, none of the Republican intraparty urban–rural splits are statistically significant. However, among Democrats, four issues have statistically significant (p < 0.05) urban–rural splits, controlling for other factors. Two of these are immigration issues—allowing refugees and building a wall—where rural Democrats are more conservative than urban Democrats. The other two issue areas with statistically significant urban–rural splits within Democrats are income inequality and allowing transgender people to serve in the military, with the models predicting that rural Democrats are less supportive of these issues than urban Democrats. In addition, these four issues also display statistically significant within-Democrat splits in means not controlling for other factors (see Figure 2). However, these differences are all substantively small: moving from rural Democrat to urban Democrat is predicted to be a 0.5 or less shift in issue position on a one to seven scale.
Figure 3. Regression results for key issues by party identification and place of residence.
4. Discussion
These results suggest that, for national policy stances, rural attitudes are not necessarily Republican attitudes. Rather, urban–rural differences in most issue stances are largely partisan splits (e.g., 18 out of 22), supporting the alternative hypothesis (HA). Rural Democrats are not significantly more conservative on political issue stances than urban Democrats, and urban Republicans are not significantly more liberal on issues than rural Republicans, on average and controlling for other factors. Notably, urban attitudes are largely comparable to suburban attitudes on these issues, suggesting that the lack of urban–rural differences found here is not simply because suburban attitudes pull the “urban” category in a more conservative direction.
The exceptions to our main finding are attitudes toward immigration, income equality, and transgender rights, where rural Democrats are more conservative than their urban counterparts. It is perhaps not surprising to find place-specific tendencies in immigration attitudes given that previous literature finds urban–rural differences, controlling for other demographic and political factors (Fennelly and Federico, Reference Fennelly and Federico2008). There are several possible mechanisms here. Two of these include greater perceived competition for rural jobs, or perceived placed-based threat to values/a distrust of outsiders (Maxwell, Reference Maxwell2019; Gimpel et al., Reference Gimpel, Lovin, Moy and Reeves2020; Lunz Trujillo, Reference Lunz Trujillo2021; Huijsmans, Reference Huijsmans2023). It could also be that urban areas in particular are more tolerant of immigrants and transgender individuals due to the relative diversity of urban areas; previous work finds rural identity to significantly correlate with rural LGBTQ attitudes as well (Thompson, Reference Thompsonn.d.). In addition, urban areas are more economically unequal than rural areas (hence, heightened contextual concern for reducing inequality). These explanations, however, are speculative; additional studies are needed to explain the mechanism that might be driving within-party urban–rural splits on issue attitudes.
That said, these are exceptions to the overall finding that little urban–rural differences occur for major national issue stances once accounting for partisanship and other demographic factors. This points to the idea that policy issue stances are nationalized, particularly in our current area of heightened partisanship and unified parties (Hopkins, Reference Hopkins2018). However, we emphasize that this does not mean local context is irrelevant to political behavior, as noted in the exceptions to our findings above. Beyond this study, we expect that issue areas that are less prominent in or specific to the national agenda would have more contextual effects.
Our results also highlight the distribution of residency and partisanship. There are substantial numbers of urban Republicans and rural Democrats; Figure 1 shows a near-equal number of rural Republicans and urban Republicans. Given the institutional setup of the USA, urban Republicans may have less impact on elections; however, in terms of what they might want on issue stances, they are similar to their rural counterparts. Further, a significant minority of rural Americans identify as Democrat; researchers and scholars should therefore be cautious in viewing the rural voter as a monolith (see also Scala and Johnson, Reference Scala and Johnson2017).
This study is limited in a few ways. The ANES presents a quality survey sample of American adults, but it is cross-sectional and thus we cannot ascertain causality. Though causality is not the focus of the paper, over-time or experimental evidence could supplement the analysis here. In addition, this study does not look at variation within the groups examined. As just one example of many, it is possible that partisans in certain types of rural areas hold different stances on issues compared to their co-partisans. Previous work finds that rural voting behavior varies depending on the predominant economic sector of the area (Scala and Johnson, Reference Scala and Johnson2017), and something similar could occur with issue stances. Or, respondents classified as rural by the ANES tend to be drawn from metropolitan adjacent areas, leading the results here to be a conservative estimate of differences (or lack thereof) between urban and rural, compared to a rural sample that includes more remote individuals. Future studies should endeavor to examine such within-rural (or within-urban) differences.
Furthermore, the data used here comes from 2020, which was a unique year for several reasons. These include the COVID-19 pandemic, Black Lives Matter protests, and the 2020 presidential election between Donald Trump and Joe Biden, which had an unprecedented number of votes that were mailed in and/or submitted before election day. For these reasons, the uniqueness of 2020 may have altered the relationships found here compared to other years, particularly since issue stances tend to be unstable.
Finally, as noted above, despite our main findings urban–rural residence may still predict other relevant political attitudes and behaviors. For instance, rurality significantly predicts certain types of political participation (Lin and Trujillo, Reference Lin and Lunz Trujillo2022). The present study is limited to the items available in the data set. In addition, given the predominant role of identity in politics (Mason, Reference Mason2018), place-based identity along the urban–rural spectrum might be a stronger predictor of certain political outcomes (including issue stances or group-based attitudes) compared to residency alone (Lunz Trujillo, Reference Lunz Trujillo2022b; Thompson, Reference Thompsonn.d.). That said, place-based identity may be to some extent divorced from physical location (Lunz Trujillo, Reference Lunz Trujillo2022a). Future work should further delve into the relative role of place identity versus partisanship in predicting relevant phenomena in political behavior.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2023.48.
To obtain replication material for this article, visit https://doi.org/10.7910/DVN/2MZE9D
