Across the United States, there is a general pattern – at least among Whites – of urban dwellers tending to be more liberal and rural dwellers tending to be more conservative. Indeed, this pattern is so pronounced that Steve Sailer managed to produce a now well-known (at least in the HBD-sphere) hypothesis of White American voting habits: his affordable family formation theory (see also a longer discussion here). This theory posits that in more sparsely populated areas, cost of land is lower, hence space and goods to raise children are cheaper, leading to people having more of them. This then causes rural Whites – so the hypothesis goes – to become more child-centric in their thinking and voting than their urban counterparts, leading them to favor conservative candidates (hence, the name “the dirt gap” for this hypothesis). Sailer’s theory received some support from an analysis of the effective cost of living across U.S. counties, which found that it is, overall, higher in the “blue” counties than the red ones.
However, even though there is a considerable correlation – particularly between state fertility rates and voting habits – it’s unclear if the causation is as Sailer suggests. Looking at the areas that defy the dirt gap sheds some insight into this pattern. In fact, in so doing, in this post, I will explore the roots of America’s conservative-liberal divide.
The deviations from the dirt gap follow a predictable pattern, which plays into a topic previously discussed on this blog. In particular, there are large areas of the country that have fairly low population density but also tend to lean to the political left, as gauged by voting habits in the presidential election. This map of the vote by population density in the 2012 presidential election (the darker, the more densely populated; the redder, higher the Republican vote, the bluer, the higher the Democrat vote) drawn by Chris Howard (over which I have drawn the borders of the American Nations as described by Colin Woodard) makes three such regions clear.
By contrasting the above map with this one, it becomes clear that, in general, the reddest areas of the country are indeed the emptiest ones and the bluest ones the most crowded. However, three areas in particular jump out, which I’ve highlighted here:
Three areas feature an anomalous level of blueness despite being relatively unpopulated, at least as far as Democratic voting areas go. These areas – northern New England proper, the upper Mississippi valley, and the Minnesota Arrowhead/Lake Superior region and the surrounding area are – unsurprisingly – all in Yankeedom, with some parts in the Midlands and New France. All are overwhelmingly non-Hispanic White. All are relatively sparsely populated, and yet all vote Democrat. All are devoid of and are generally outside the orbit of major metropolitan areas, with the upper Mississippi valley area having only smallish metro areas (Iowa City, Waterloo, and the Quad Cities area) and northern New England having the Burlington, Vermont-Plattsburgh, New York metro as its most populous area.
If we were to believe that population density and the urban-rural divide were the prime drivers of voting habits (second only to race) – as is commonly assumed by many simplistic analysts, then these areas should not exist. Yet they do.
As we have seen with this blog, political orientation, as with all things, is largely heritable. Indeed, a recent meta-analysis of behavioral genetic (class twin) studies by Peter Hatemi et al which observed ~19,000 subjects across five countries found that heredity accounts for upwards of 40% of the variance in political orientation (this is the low end estimate, since measurement error attenuates the results).
Further – and just as significant – the shared environment measure isn’t significantly different from zero, also consistent with other behavioral genetic results. This means that family environment and upbringing (parenting) have no effect on one’s adult political views, which is in line with the notion that parenting doesn’t have much of a lasting effect on how we turn out at all. Because evidence indicates that heredity is of paramount importance in shaping political views, I explore genetic explanations for persistent differences in regional ideological and political differences.
That said, contextual factors can affect the voting habits of individuals and groups even with no change in their heritable dispositions. This is why one cannot focus too much on any single electoral outcome, but must focus on broad trends, especially persistent patterns of behavior. Understanding relevant context may help explain particular electoral results. Hence, I will try to examine these as well.
So what’s the deal with these three regions? Unfortunately, I am only intimately familiar with one of them, as I live deep in the northern New England zone in here in Maine. As well, I have been to the periphery of the upper Mississippi region. I will examine the characteristics of each in turn.
Northern New England (from Maine to northern New York) is perhaps the area of Yankeedom that most consists of the descendents of the original Puritan settlers. However, this area gradually diffuses into New France, with Maine’s Aroostook county hosting a strong Acadian-French speaking population (where I live, French speakers are far from an uncommon occurrence). French Canadians make up a considerable share of the population all along this region. Like the rest of New England, this area received a considerable Catholic Irish influx, but less so than southern New England. Scots are also present, mostly arriving from farther east in Yankeedom (from the Maritimes). This area is not very wealthy, with the major industries (lumber, paper) mostly long since gone. The area is largely dependent on tourism, maple syrup production (particularly in Vermont), and fishing (particularly lobster) along the Maine coast.
The largest urban center in the region (when southern Maine and southern New Hampshire are exempted as being part of the Boston metro area) is in Vermont, the Burlington metro area straddling the beautiful Lake Champlain. This city is home to the University of Vermont (one of the “Public Ivies“). The state of Vermont is the most Left-leaning of the area, notable for being the home of the socialist senator Bernie Sanders. But like Sanders, most of the residents of the state originate from outside Vermont, from both within the region and from without. These newcomers have impressed themselves on to the area’s Puritan culture and have moved it squarely to the political left.
The eastern Iowa area, sitting on the border of Yankeedom and the Midlands, contains elements of both. This area of the Mississippi Valley received a considerable German settlement, as did much of the Midland Midwest. Indeed, this area was one of the strong concentrations of German settlement. A considerable Scandinavian presence was also established here. Catholic Irish are also present. These attest to the area’s Midland tradition of ethnic pluralism.
In keeping with its Yankee characteristics, this area is home to two large universities (indeed, also both “Public Ivies”), the University of Wisconsin–Madison on its periphery and the University of Iowa in Iowa City in its heart. The regions around these schools retain a certain liberal characteristic – appearing as if the entire region was in their sphere.
And finally, the upper Great Lakes area is the most sparsely populated of the three here, being essentially completely devoid of large cities, the biggest being Duluth, Minnesota. All I know about this area is that it interestingly appears to correspond to the area where self-identified Swedish-, and to lesser extent, Finnish- and Norwegian-Americans are most concentrated:
Indeed, the Swedish (keeping the limitations of self-reported ancestry in mind) zone seems to capture the most liberal areas. Considering Scandinavia in general today, and Sweden in particular, perhaps the area’s liberal tendencies aren’t all that surprising.
I invite readers more intimately familiar with these areas to please share what you know in the comments.
So what explains the political inclinations of these areas? What do they have in common? Is religiosity a common theme?
This is a map of religious adherents (from Valpo), which can be taken as a rough gauge of the religiosity of the country. As reported previously, northern New England proper is the least religious area of the country, with the Pacific Northwest (the Left Coast) being next least religious. However the other two anomaly regions along aren’t necessarily as irreligious. In fact…
…the eastern Iowa area is actually quite Catholic (Dubuque, Iowa is a considerable Catholic stronghold). While some of these individuals are from traditional American Catholic groups, such as the Irish, the bulk appear to be German Catholics. Indeed, Catholic Germans appear to explain much the distribution of Catholics in the upper western Midwest seen on this map, corresponding well to the distribution of self-reported German Americans.
(The Catholics in eastern New England are primarily Italians and Irish, in the south; and French, in the north.)
So if population density cannot explain the liberal affinity of these areas, neither can religion or religiosity. Indeed, the eastern Iowa border region appears to liberal and education-minded while remaining at least nominally religious. So what else can we look at? Indeed, for that matter, how, exactly, does the affordable family formation theory hold up, in general, and across these “anomalous” regions in particular?
These are comparison maps from Hawley, 2010. This is a direct examination of the Affordable Family Formation theory. There are some large correspondences, but huge disparities exist.
Indeed, none of the three liberal “anomaly” areas fit here. All are areas where the effective price of housing is low. Affordable family formation cannot directly explain these areas.
In fact, there appear to be large problems with the affordable family formation model. Indeed, much of the Interior (Far) West is incongruent, being racked with high effective land prices but nonetheless solidly Republican. Broadly, we see that it seems to work well for a few large areas: the Great Plains, the Left Coast, and the New York City and Boston metro areas.
Even the South doesn’t necessarily fit into the pattern (even discounting the areas that are majority Black). Indeed, at least one region exists as the reverse of the three liberal anomalies above:
The southern Piedmont is an area of relatively high population density (mostly suburban) and generally conservative voting patterns. This area consists of the suburbs of the metropolises of southern Appalachia and the Deep South/Tidewater. A stark city/suburb split is evident here. While the suburbs of these cities are indeed cheaper to live in than their Northeastern counterparts, even the few relatively expensive areas are solidly conservative.
So what then is the best explanation for the political landscape across America? As Razib Khan once noted when he touched on the matter, the best fit is the American Nations model.
It should be no big surprise that the anomalous areas (in terms of population density) all fit in the respective nations in the current political alliance as they exist today:
Yankeedom, New France, and the Midlands are dominated by groups that today espouse beliefs n line with what are currently modern liberal ethos (such as universalism, faith in government, education, and egalitarianism), as per the original settlers and the people that subsequently came to live among them. By contrast, the Cavaliers and the Scotch-Irish of the current Dixie nations (and to a certain extent the Far West) espouse views generally in line with modern conservatism – restrictive sexual mores, religious faith, an innate propensity towards children and family, and a distrust/de-emphasis of outsiders to the group. This is partly related to the respective levels of historic of inbreeding in each of these groups (see Tentative Ranking of the Clannishness of the “Founding Fathers” and How Inbred are Europeans?)
This split across the American nations is visible across history, as we can see from these maps of historical presidential election results by county (from here):
A look back through these results shows that the alliances between the respective American nations shift with time, and hence voting habits follow the divide of the coalitions of their day. Hence, the recent political regional divide should not be thought of as being absolutely definitive. They do nonetheless speak to deep divisions between individuals and the respective regions/nations.
Colin Woodard noted the importance of ethnonational divisions over the urban-rural divide in his analysis of the recent Virginia gubernatorial race:
Some suggested this was nonsense, the results being better understood in “rural vs. urban” terms, which Democrat Terry McAuliffe dominating the cities, Tea Party Republican Ken Cuccinelli the rural areas. In a postscript I provided some early evidence this was a mistaken idea. Now I have some solid data to prove it.
I asked one of my student research collaborators — Miami University of Ohio’s Nicollette Staton – to run the results both by region and by the National Center for Health Statistics’ six-tiered urban-to-rural spectrum, which categorizes every U.S. county by level of urbanity, from those in major metropolitan regions (1) to the completely rural (6). The results were even starker than I expected.
In Greater Appalachia, Cuccinelli won every category of county, from the very largest cities in the section (where he won 49.1 to 45.7) to counties without so much as a big town (62.8 to 30.8). In every category save the largest (category 2 in Greater Appalachia), he won by more than 20 points.
By contrast, in Tidewater, McAuliffe won by large margins in counties large and small, taking five of the six categories. In the biggest cities he won 56.3 to 37.3. In the most rural counties he won by a convincing 51.0 to 41.1.
Of course, it’s not that the Cavaliers of the Tidewater area are all that Democratic today; rather the Tidewater area contains the state’s Black population, and I suspect this carried McAuliffe to victory. This might be revealed by this part of Woodard’s analysis:
The only category he lost [in the Tidewater] – “category 5” which consists of rural counties with a decent size town within them – was by far the least consequential, accounting for less than 12,000 of the more than 1.6 million votes cast in Tidewater that day and only 6% of the region’s overall rural vote. (Curiously, though, Cuccinelli won this sliver of counties by a staggering 61 to 32.9.)
That all said, the affordable family formation theory does still have much to offer here in terms of an explanation for what we see. Parts of the current “blue” nations, Yankeedom and the Midlands, contain considerable (relatively) conservative districts (at least as seen in the most recent presidential elections). In the Midlands especially, this appears to correspond, to a degree, with the pattern the affordable family formation theory would predict. This is especially visible across the Plains.
Much of the Plains was settled by diffusion from the more populated “seed”/coastal areas. Such areas may have favored individuals with strong family values, as these individuals would need to continue moving in order to continue reproducing. Indeed, this attitude towards marriage, family, and children is one of the modern hallmarks of the psychological differences between liberals and conservatives, as testified by my examinations of their fertility gap (see also Who’s Having the Babies?):
Indeed, the attitude towards sex, marriage, and family are defining traits that distinguish liberals and conservatives today, conservatives being more child- and family- centric. In the past, this difference may not have had the consequences for fertility that it does for the two groups in today’s world (where it is possible to delay fertility). Avi Tuschman wrote a book, Our Political Nature: The Evolutionary Origins of What Divides Us (which I have yet to read) detailing these inherent political dispositions and possible explanations for why they are so (see his Atlantic article, as well as HBD Chick on his work, also here).
The fertility differentials between liberals and conservatives show that the likely primary source of the baby gap Steve Sailer noted across the states is the relative proportions of each of these types of individuals in each state.
So now we see that there are several broad factors that separate liberals and conservatives. These include ethnic origin (the angle explored by David Hackett Fischer and by Colin Woodard) and attitudes towards family and children (the angle of looked at by Steve Sailer and by Avi Tuschman). But what else divides these different type of individuals? Another factor, and one that also lends itself to the pattern Steve Sailer observed, is sensitivity to crowding itself.
Discussion over at the site Neuropolitics once claimed that conservatives have much less tolerance for crowding than do liberals. If so, this would mean that individuals with conservative mindsets are uncomfortable being around large numbers of people, and would tend to be pushed away from urban centers. (John Durant has noted that he believes that conservative attitudes may be a form of pathogen avoidance, which may also explain conservative aversion to high population densities.) Liberals on the other hand appear to be more comfortable in cities, and may remain close to their birth areas if they come from denser areas; indeed, they may preferably move towards them. (I will note for the record that my wife and I, despite being liberals, are not very tolerant of crowds and much prefer our wide-open spaces.)
We have indeed heard these types of ideas here before. These were part of my Pioneer hypothesis. I once posited that natural selection trends towards more conservative minded people on the front of geographically expanding populations and more liberal-minded people in geographically bounded ones. The key limitation for this hypothesis, and why I have been unable to adequately test it, is that I don’t know if conservatives had a fertility advantage in the far past, prior to the 20th century (see Expectations and reality: a window into the liberal-conservative baby gap). Likely, considering, the prevalence of liberals today, if conservatives couldn’t have had much of a fertility advantage in the past, but the traits may have existent in some of semi-balanced equilibrium, being, broadly, equally good strategies. We do see that in geographically expanding populations, certain traits are indeed selected for, as was the case with the French Canadians. However, no group in the U.S. was under exactly the same circumstances, as all regions of the U.S. have received substantial subsequent immigration, particularly Yankeedom and the Midlands. That said, the expansion of the colonial American groups was, for a time, very similar to that of the New French, especially in New England and Greater Appalachia. Today, while Yankeedom and the Midlands, in toto, lean left as part of the Blue state alliance, parts of each are considerably conservative. Many of these areas are regions that were fronts of expansion from their respective colonial bases. Perhaps this is the legacy of individuals who had to spread out into a homestead of their own, away from the filled settlement (if, albeit, often just outside it, given the stronger family ties typical of conservatives). The distributions of some religions that are prominent in the western reaches of the Midlands perhaps attest to this:
If you’re an individual intimately familiar with these regions, I invite you to comment and please share what you know.
Of course, the Far West contains a conservative bunch of Yankees that split off from their cousins in New England: the Mormons. Mormons perhaps exemplify the process of more conservative-minded (at least, what would be such in today’s world) expanding away from more liberal source stock.
Alternatively, these could be the result of simple selective migrations – the founder effect. Perhaps these western areas across the Great Plains and the Far West were just settled by religious pioneers.
Indeed, founder effects – which are broadly responsible for the development of the American nations – works for liberals too. We see that across the Left Coast, which received individuals high on “openness to experience” (see also a critical take on that particular dimension: Openness to Experience – That Liberal Je Ne Sais Quoi | Staffan’s Personality Blog) and introversion. This may be partially responsible for the distribution of personality across the country (see also here for maps of each of the Big Five dimensions):
Self-sorting migration likely reinforces these divides, as noted above. Additionally, these processes alter not just the receiving regions, but the sender regions too (see Boiling Off | West Hunter). The Midwest in particular has lost a considerable fraction of its people – often for the coasts. Perhaps the upper Midwest have lost enough of its “bright lights, big city” folks to render the remaining population more home and family bound in their thinking. Just the same, more family-centric people probably flee expensive coastal metros for the cheaper interior.
In general, the affordable family formation theory is likely correct in that people shuffle themselves around for cheaper/more expensive districts based on their political (and hence, marriage and children) inclination.
We have seen that there are likely several factors involved in shaping the modern American political landscape. Of which, the strongest may be ethnonational origin. This is buttressed by population density pressure, founder effects, internal self-sorting, and perhaps even recent evolution. There remains much that we don’t know, and hence, a lot remains to discover. Indeed, genetic sampling and admixture analysis of White Americans in different parts of the country might go a long way to understand key political divides. However, the heritable roots of these differences mean that the divisions among White Americans are largely intractable, and the divides we see will be with us – in one form or another – for a long time to come.