Once upon a time (actually about 50 years ago) when I started doing research on social psychological differences between urban and rural people, many authorities dismissed the whole question as out of date. Sure, city-country cultural differences were once vast and important, but in the modern era of interstate highways, telephones, television, nationwide markets, and the like, such differences were gone or pretty soon would be. 
Now, after a half-century more of those distance-shrinking changes plus the internet, cheap air travel, and so on, a Pew Research Center report has gotten media attention by highlighting notable differences in the cultural views of Americans according to whether they lived–or whether they said that they lived–in a city, a suburb, or a rural place. Add this report to the political polarization between city and country highlighted in the Trump victory–he won rural and small towns by about 25 points and lost cities over 50,000 by about the same–and it looks as if city-country differences are alive and well in the 21st century.
There is, however, a technical issue–an interesting technical issue–in the Pew analysis. Whether people say they live in city, suburb, or country partly reflects obvious demographic facts. (Few residents of New York City would tell a pollster that they live in the country–although a vice-presidential candidate did once claim that her Queens neighborhood was just like a small town.) But where survey respondents say they live is also shaped by their stereotypes of city, suburb, and country.[1] So, for example, a socially conservative person who feels that her neighbors share those views might, in an ambiguous setting like a low-density suburb, say she lives in a rural place, because it feels culturally rural. Such subjective answers to “Where do you live?” would make city-suburb-rural differences look greater than they really are.
So, let’s take another look. I’ll ask whether there really are place-based differences in cultural views today, whether they are diminishing as notions about modern technology suggest, and whether those differences really are about the places.
Differences and Trends
I turn again to the General Social Survey to do a rough, back-of-the-envelope look at the data. To simplify matters, I look only at white respondents; we’d have to control for race eventually, anyway. Over the decades the GSS has repeatedly asked questions that tap some of the big political-cultural issues in the U.S. I found a bunch of questions that were asked from the 1980s through 2016; they roughly cluster into two topics, so I constructed two simple scales.
One scale concerns views of race and welfare–are racial differences due to discrimination?; should the government spend more money on welfare?; should the government spend more to improve the conditions of blacks? (It’s no surprise that whites’ views on welfare and race go together).[2] The other scale taps traditional versus modern views of sex and family: is premarital sex ok?; is it better for women to tend the home?; and should abortion for any reason be allowed?[3] The GSS also has an objective measure of where respondents live based on the size and locations of their communities. [4]
I ask, first, What is the connection between where GSS respondents lived and what they said about these topics; and How has that connection changed over roughly 40 years? The two panels below show the answers.
The left-hand graph displays the results for race and welfare, the right-hand graph for sex and family. The residential location scale across the bottom runs from (1) the center cities of the 12 largest metropolitan areas, (2) center cities of the 88 next largest metro areas, (3) suburbs of the largest metros, (4) suburbs of the other metros, (5) small urban places, and (6) rural places. In both graphs, the higher the point, the more conservative the average respondent’s answers.[5] The blue line is for the 1980s and ‘90s, the orange line for the 2000s and ‘10s. What do we see?
We see, first, that there were place differences on both these topics, both before and after the turn of the century. Second, on race-welfare, the differences grew: whites inside center cities and large-metro suburbs became more liberal. (The connection between place and political party affiliation also grew stronger.[6]) Third, on sex-family, the differences shrank; whites outside the center cities, especially the suburban ones, became more liberal–but whites outside the metro areas remained distinctively conservative.
Are These Real Differences?
But are these cultural differences the result of location or just a coincidence? Young and better-educated people tend to live in more urban places; also, rural places are disproportionately in the South. If we take these factors into consideration, do they explain away the place differences? Some, but not much.
The next pair of graphs shows the results of, first, pooling all 37 years, 1980-2016, together–the solid lines–and then adjusting the results to take into account, that is to remove the effects of, period and respondent’s age, education, and region (South vs. rest), as shown in the dashed lines.[7]
What we see in both graphs is that the age, lower levels of education, and region of non-metropolitan whites explain some, but far from all, of their conservatism. After adjusting for those factors, we see that on race-welfare, whites living in the largest center cities remain distinctively liberal and that on sex-family, non-metropolitan whites remain distinctively conservative.
What’s Happening?
So, here we are generations after this city/country distinction was no longer supposed to matter, but it seems to. Why?
An important part of the explanation is self-selection–different types of people choosing to stay in or move to city, suburbs, or rural places. We already saw that one form of apparent self-selection–that by age and education–may explain some of the connection between place and opinion. Most dramatically, young people in smaller towns move away for more education and often don’t come back, leaving their hometowns grayer and with fewer college grads (see, e.g., here). Perhaps, in addition, Americans are picking places to live precisely to suit their lifestyles and cultural views. There is some evidence for that, too (e.g., here). On the other hand, migration rates have been dropping for decades, presumably slowing the process of self-selection, and yet the city-country effect has persisted and in some ways strengthened.
An additional explanation is that there is something about the nature of more urban versus more rural places that encourages liberal versus conservative views, particularly on these sorts of cultural issues (rather than on, say, economic or foreign policy). For over a century, scholars have suggested that the intrinsic size, the density, and the heterogeneity of urban places encourages tolerance for difference and for unconventional views and behavior. How this happens–if it happens–is a subject of yet more discussion.[8] But, if it is so, then the influence of place means that the eternal discussion of city mouse / country mouse will just keep on reemerging.
Update, Nov. 25, 2018:
An analysis of the 2018 midterm elections reveals that the rural-urban divide grew even greater, as outlying districts voted more red and metropolitan districts voted more blue.
Update, August 13, 2021:
A recent analysis by David Hopkins reveals how much the post-Civil Rights sectional gap (the South becoming Republican) has transformed much more into a national urban-rural gap:
NOTES________________________________________________________
[1] I learned this from a 1977 survey I conducted in northern California. Project researcher Carol J. Silverman found that people living in same community differed on whether they said it was a big city, or small city, or suburb, and so on. The labels people gave their communities depended in great measure on factors such as how burdensome their commutes were, their sense of (and taste for) diversity, and especially their fear of crime. If they worried about local crime, they tended to say they lived in a city; if not, more likely a small town or suburb. (See Silverman, “Place Types as Social Constructions,” Working Paper No. 381, Institute of Urban and Regional Development, U.C. Berkeley, 1982; also see David Hummon’s book, Commonplaces.)
[2] GSS items RACDIF1, NATFARE, NATRACE.
[3] GSS items PREMARSX, FEFAM, ABANY.
[4] SRCBELT.
[5] The numbers for the scales are arbitrary, but the y-axes run roughly from – ½ SD to + ½ SD of the grand mean.
[6] The association between PARTYID and SRCBELT went up, by decade, from the 1980s to the 2010s: gamma = -.01, .03, .06, and .11. (For the 1970s, it was .06, which suggests that this is not a linear phenomenon.)
[7] The unadjusted lines do control for period–before and after 2000. In the adjusted results, the scales are regressed on period, age, age-squared, education (4 dummy variables), South (1 dummy), year (1 dummy for 2000-2016), and place (5 dummy variables). The points are the place coefficients versus category 4 (suburbs of smaller metros). The place findings reported in the text are effects significant at p<.01.
[8] The classic works include papers by Georg Simmel (1905), Louis Wirth (1938), and Herbert Gans (1962). My own contributions to the discussion include writings in 1975, 1982, 1984, and 1995.
MADE IN AMERICA 