Voting Behavior

What the Patchwork Nation Means for Politics and Policy

Published on by Brynne VanHettinga

In a previous post, we looked at the research of Dante Chinni and James Gimpel, who have identified twelve distinct community types in the United States, and correlated these communities with voting preferences. These communities are scattered throughout the country and—with the possible exception of Mormons and Evangelicals– generally do not conform neatly to state or regional geographic areas. The question then arises as to whether or not the U.S.A. is too politically and culturally diverse to be governed efficiently as a single nation.

patchwork-nation

The researchers did find one unifying theme in all twelve communities. This was the feeling that things were somehow “different” in a post-Great Recession world. That is, one generally could not make predictions of the future based on past patterns, and expectations had to be adjusted downward. However, some of the downward trends had been going on for decades, even if not much attention was paid to them at the time. Between 1972 and 2007, worker productivity grew by 90% while wages declined by 11 percent. As the authors describe it, “The dominoes that began falling during the recession had been lined up for some time.”

The researchers then analyzed correlations between religious attendance, religious denomination, and voting patterns. To a certain extent there is a religious correlation with the red/blue divide, but this is not absolute. High church attendance was found in stereotypically “red” and rural Mormon Outposts, Tractor Country, and Evangelical Epicenters, but also in Emptying Nests and Immigrant communities. The researchers suggest that these differences can be attributed to both the homogeneity of religious membership (high in Evangelical and Mormon communities) as well as the integration of church into local culture and “the views from the pews.” Ironically, when asked about whether churches should stay out of politics, Mormons agreed with this at the highest rate (55%), followed by the more atheistic/agnostic Monied Burbs (52%). The researchers propose that the so-called “separation of church and state” goes both ways—that is, it serves to keep churches from being “infected” by worldly concerns as much as it does to keep religion out of politics.

The next analysis was the relationship between voting patterns and “cultural” factors. There was a “pretty decent correlation between voting patterns in the 2008 presidential election and gun-shop-to-bookstore” ratio. But the stereotypical relationship between guns and Republicans and books and Democrats is also not absolute. Democratic-leaning Industrial Metropolis communities had both fewer gun stores and fewer bookstores, while the Republican-leaning Tractor County—which, not surprisingly had the most gun stores—also had the fourth most bookstores.

Social media users tended to be negatively correlated with age and positively correlated with income. So, in places like Emptying Nests, which have both higher than average age and income, social media usage was low. But in Minority Central, where average age is younger but incomes are lower, social media usage is also low. The oldest communities (Tractor Country, Emptying Nests and Service Workers) use social media the least, while the younger Campus & Careers and Military Bastions use social media the most, followed by Boom Towns and Monied Burbs.

Private companies have much more detailed data about consumer preferences than political or public interest research organizations, which they use to make decisions about store locations. Retail stores such as Walmart and Starbucks will only make an appearance if the demographics are “right” according to corporate market research. While there is some argument that retail chains have homogenized national culture, the Patchwork Nation model makes the argument that the locations of certain stores—or the number of them—in certain communities tends to reinforce cultural differences. “You are what you buy” might be more accurately stated as “you are what you CAN buy.”

The researchers then constructed a “hardship index” which further parsed out community differences. This index was based on county-level data about unemployment rates, gas prices, changes in gas prices, changes in the percentage of household spending for gas and car maintenance, home foreclosures, and changes in home foreclosures, which were then converted to a single “hardship” score. Although Service Worker Centers and Tractor Country both have high majorities of white people, are less densely populated, and share incomes below the national average, Service Worker Centers consistently had the highest hardship index (followed by Minority communities), while hardship in Tractor Country was in second-lowest place—behind the Monied Burbs. The authors attribute this to two things: the low rate of debt in Tractor Country as well as their primary reliance on agriculture. That is, they enjoy a relatively self-contained economy, where the prosperity of Service Centers depends on the prosperity of the broader economy.

The researchers then propose that their “Patchwork Nation” framework provides an explanation for political polarization. This goes beyond the standard Republicans vs. Democrats, but suggests that actual policy proposals may appear to be either helpful or harmful, depending on the idiosyncracies of local economies. For example, extension of the 2008 home-buyers tax credit was welcome in areas like Boom Towns, who were hit hard by the housing market collapse, but viewed as an unnecessary waste by persons in areas where housing prices had stabilized and were even rising. The problem is that policy at the national level is based on aggregated data—that is, “averages” that miss the mark for many of the localized micro-economies.

Is America truly one nation or many? The researchers suggest that American communities may actually have been more different and “localized” before the days of instant communication and mass media. At the same time, society overall was simpler, so in spite of greater regional differences in some things (e.g., spoken accents, choice of entertainment, food availability, etc.), patterns of everyday life were not necessarily that much different. Today’s trends of “niche” or self-selective media offerings that are algorithmically tailored to specific preferences may aggravate cultural and political differences. However, the researchers also argue that the Civil Rights movement would not have been possible without a national media. Moreover, “not every difference in media consumption is about worldview,” as we have seen that use of technology is influenced as much by age and income.

patchwork-flagOne potentially unifying theme is Americans’ distrust of concentrated power. The percentage of persons who agreed with the statement, “The federal government should run ONLY those things that cannot be run at the local level” ranged from 26% (Industrial Metropolis) to 41% (Tractor Country), with an average of 31.6%.  The percentage of persons who agreed with the statement “There is too much power concentrated in the hands of a few big companies” ranged from 32% (Military Bastions) to 40% (Minority Central), with a slightly higher average of 34.4%. Thus it appears that local control and devolution of “power to the people” are important issues for approximately one-third of people in all of the diverse community types. So, perhaps the one thing that unites us all as Americans is the distrust of centralized power and top-down decision-making—regardless of whether this power is political or economic. This might be where the hope of a unified country can be found.

The Patchwork Quilt Becomes More Complex and Colorful

Published on by Brynne VanHettinga

A lot of research has attempted to correlate demographic differences with political preferences. Generalizations (sometimes backed by data) are that younger people and minorities tend to prefer Democrats, and older white people tend to prefer Republicans. Besides the more traditional divisions of age (generational cohort), race, gender and income, other social divisions have been identified based on “cultural” and community factors; i.e., rural vs. urban, higher education (college and above) or not, religious/churchgoing or not, and use of technology or not (the digital divide).

patchwork-america-map

Dante Chinni and James Gimpel have divided the United States into 12 distinct community types based on patterns in census and voting data. They have made these distinctions based on demographics such as population, age, race, income and education as well as voting behaviors over time. Rather than attempting to predict who is more likely to vote Republican or Democrat, these researchers look at how people vote, then analyzed their demographics and community lifestyles—in essence constructing community typologies based on historical voting patterns rather than attempting to predict voting patterns based on the characteristics of individual voters. In descending order of population, these communities are:

Monied Burbs:  286 counties and 69.1 million people. These communities enjoy higher than average incomes and educational levels, and tend to be areas of job (and population) growth. Monied Burbs have the highest percentage of college graduates as well as post-graduate degrees. Other than higher incomes and monied-burbseducational levels, however, Monied Burbs are statistically “average” with respect to the national composition of age, race and religion.  These communities tend to weather recessions better than the rest of the country and also help to “pull the country out of it.”

Boom Towns:  384 counties and 59.3 million people. These are areas which had been growing and relatively wealthy, but were also hardest hit by the Great-Recession housing collapse. They still tend to have higher median incomes, with residents employed in professional and executive positions as well as in the more stable occupations of education and government. They also tend to be younger than the nation as a whole, and slightly more diverse (84% white as opposed to the national average of 87%).

Industrial Metropolises: Only 41 counties, but 53.9 million people. These communities are densely populated, younger, and more racially diverse. They tend to have greater income divergence (very high and very low incomes, with an average that is close to the national average), which is correlated with race.  metropolisIn spite of the “industrial” label, occupations in these areas have bifurcated into professional jobs and service jobs. Industrial Metropolis residents attend college at a higher than national rate (25% versus 16% nationally).

Service Worker Centers: 663 counties and 31 million people. These areas are frequently mid-size towns, neither really “rural” nor really “urban.” Jobs are impermanent, low-paying and lack employee benefits. However, the cost of living is not necessarily low, and many workers here live paycheck-to-paycheck.  Service Worker Centers are slowly losing population—no one really moves here service-workersfor work because of higher-than-average unemployment rates, and the jobs that exist are unappealing. They are concentrated in the Northeast, the Northwest, and upper Midwest, and racially are mostly White. These communities often have a higher-than-average proportion of “tourist” income, so their economic welfare depends to a larger extent on how the rest of the country is doing (and spending). Voters tend to lean Republican, but not by large margins.

Immigration Nation:  204 counties with 20.7 million people.  These areas are concentrated in the Southwest (Arizona, California, New Mexico, and South Texas). They have higher than average Hispanic populations, with lower than average incomes and higher rates of poverty. These areas are hit hardest by “boom and bust” cycles in housing and construction. The authors predict that, “the tensions over labor…will likely intensify if economic times stay tough [and] high unemployment endures in the long term.”

Evangelical Epicenters:  468 counties and 14.1 million people. These communities are concentrated in the mid-South-West (Arkansas, Oklahoma, Alabama, evangelicalsMissouri and East-Central Texas). They are characterized by young families and evangelical Christians. Incomes are lower than national average.  The authors suggest that these are the so-called “values voters,” who tend to believe abortion should be illegal (21%) and seek communities with “strong family values.”

Minority Central:  364 counties and 13.5 million people. These communities contain much larger than average percentages of African Americans or Native Americans. They are located primarily in the Deep South or in connection with Native Nation reservations (the Arizona-New Mexico border, North and South Dakota). They have lower than average incomes and higher than average poverty rates, along with high unemployment and low college enrollment. However, there is a distinct disparity in the incomes of minority and white populations, reflecting vestiges of racial segregation.

Campus and Careers:  71 counties with 13.1 million people. These are so-called “college towns” that are scattered throughout the country. As the name implies, campusthese communities have a lot of college students and young people starting careers.  While incomes are not particularly high, neither are poverty rates. Jobs pay decently and employment tends to be stable. These communities also tend to have higher rates of Internet and technology usage.

Empty Nesters:  250 counties and 12.1 million people.  These communities are located primarily in the upper Midwest (Iowa, Indiana, Illinois, Ohio, Wisconsin, Michigan, and Minnesota) and Florida. Populations are older, many who are living on fixed incomes. Empty Nesters are “solidly white and solidly middle class,” many who had been (or are still) employed in manufacturing.  Although these people are not wealthy, they enjoy slightly higher than average incomes and lower poverty rates. Many of them grew up in an era of strong unions, a healthy middle class, and secure retirements.

Military Bastions:  55 counties and 8.4 million people. These are a small number of communities scattered throughout the country that are located near military bases—which are the major source of jobs in these areas. The soldiers (and their militarypaychecks) are usually welcomed by the regular residents, although the neighboring towns experience disruption when soldiers are deployed overseas for long periods of time. In addition to the soldiers, these communities also attract high-dollar military contractors, which the authors allege could make them “Republican-leaning versions of Campus and Career communities.”

Tractor Country: 311 counties with 2.3 million people. Tractor Country is sparsely populated, and residents earn their livelihoods primarily in farming and agribusiness. They are concentrated in the upper Midwest and middle of the country (North and South Dakota, Montana, Idaho, Colorado, Wyoming, Nebraska and Kansas), with a few counties scattered in Washington, Oregon and Texas. In addition to the Deep South, these are the quintessentially “red” counties. After Mormon communities, it is thtractor-countrye least diverse, being mostly (96%) white. Incomes are lower than national average, but so is the cost of living.  The authors call Tractor Country residents “Progressive Conservatives.” These communities are conservative in their distrust of Washington, their fiscal frugality, and their appreciation for hard work. Yet, they also are strong supporters of their local communities and are willing to help out neighbors in need. Tractor Country was relatively unaffected by the housing collapse and financial crisis—mainly because most of the banks are small and locally owned. However, family-owned farms are increasingly being replaced by industrialized agriculture, which the authors predict will “change the nature of life in these communities.” Tractor Country is also losing about 2% of its population per year, and 40% of its residents are over age 50.

Mormon Outposts: 44 counties with 1.7 million people.  These communities have high proportions of members of the Church of Jesus Christ of Latter-Day Saints. Although Mormons are a religious minority (comprising about 2% of the U.S. population), they have settled in a finite area mainly in the states of Utah, Southern Idaho and Western Wyoming, where their insular communities allow them to maintain their religion and culture. The vast majority are white (96%).  About 11% of Mormon communities work in agriculture, making them the second (after Tractor Country) most rural community group. They are, however, economically better off than their rural counterparts in Tractor Country and their fellow churchgoing Evangelicals, with slightly higher than average incomes and slightly lower than average poverty rates.

The strongest support for Democrats is in Industrial Metropolises, followed by Campus and Careers. Surprisingly, Service Workers tend to support Republicans by a small degree, and both Immigrant and Minority communities were about equally divided in past elections, but the Democrats have been gaining in all three of these groups more recently. The authors allege that Service Worker Centers are “the economy’s canaries in the coal mine”…  When things go bad, they go bad here first.”

The strongest Republican support is in Mormon, Evangelical and Tractor Country communities, which suggests a cultural rather than an economic bias.  Boom Towns, Empty Nesters, and Military communities have favored the Republicans by small amounts in the past, but Democratic support in these communities has also been increasing.

The truest “swing” communities have been the Monied Burbs—favoring Republicans in one election and Democrats in the next, with support for Democrats apparently growing since 1980. Monied Burb constituents are thus the most sought-after voter, both for their higher incomes and for their general lack of diehard party loyalty. Because of their sheer numbers and shifting electoral preferences, Monied Burbs, Boom Towns and Service Worker communities offer the greatest potential for electoral change.

The Patchwork Quilt of Voting Patterns

Published on by Brynne VanHettinga

Having just been through a national election, many of us have seen those maps that show states as being either red (Republican) or blue (Democratic). However, when voting patterns are displayed at the county level, the states do not appear so monochromatic. For example, Texas is a quintessentially “red” state, but there is a string of “blue” counties along its southern border.  Conversely, California is mostly blue along its West coast and red in the interior.  The typical “swing” states of Wisconsin, Michigan, and Florida show a “patchwork” pattern of red and blue. These maps get even more interesting when they have been mathematically distorted based on population density and shadings of purple based on how close the electoral split is at the state and county levels.

2016-election-results-by-county

Some studies have identified divisions that are not necessarily “red vs blue,” but may impact political leanings. These divisions are based on demographics (age, race and income) as well as “cultural” factors such as church attendance and where people shop. Income (and related wealth) is certainly one of the more salient factors, but there is some debate about whether the wealthy favor Republicans or Democrats.  Some of us intuit that wealthy people will favor Republicans based solely on policy preferences. Yet there exists a persistent media-fueled stereotype of opera-season-ticket-holders and chardonnay-sipping Democrats versus NASCAR-attending “Joe six-pack” Republicans.

In Red State, Blue State, Rich State, Poor State, a group of political scientists first present data that supports a correlation between higher income and voting Republican at the individual level, along with increasing stratification of voting patterns. Paradoxically, the richer states tend to vote Democratic. The authors then exhaustively analyze the data in an attempt to answer the following research question:  Why do rich people vote Republican and rich states vote Democratic?  After meticulous parsing (a statistical multi-level modeling) of the data, the authors find that rich people in poor states are more likely to vote Republican (i.e., in accordance with their economic interests), while in rich states, income has almost no correlation with voter preference—suggesting that voter preferences in wealthier states are based more on cultural factors than economics.

Data confirmed a historical pattern of rich counties traditionally supporting Republicans, although this has been steadily declining over the past forty years. Moreover, rich voter support for Republicans varied by region, with the strongest correlation between income and Republican support in the most Republican-leaning of the Southern states. The authors acknowledge that income is also correlated with race, particularly in the South, so preferences might be race-based as much as income-based. Alternatively, there are correlations between higher income and higher education.  Higher education is also correlated with preferences for Democrats, which could result in a higher-income Democratic preference (the so-called “liberal elites”). However, even outside of the South, voting patterns are predictable in the same manner:  higher individual income predicts support for Republicans and living in a wealthier state (regardless of individual income) predicts support for Democrats.

This group of political scientists next analyzed whether “cultural” factors might provide an explanation. They look at trends in “issue polarization” on the basis of economic, civil rights, moral and foreign policy issues.  Party identification has indeed become more associated with economic issues in poor states and with civil rights and moral issues everywhere, with moral issues especially becoming “increasingly correlated with liberal-conservative ideology and with each other.”  One counterintuitive result was that that low-income voters—who also tend to have less education, interest, and participation in politics—are “surprisingly complex,” exhibiting “no clear left-right opinion clusters.”

The next question to be addressed was whether or not there was a religious correlation; e.g., religious observers voting Republican and secular voters voting Democratic. For low income voters, there was little difference in voting patterns when comparing religious vs secular voters. Only among high-income voters does religious observance predict Republican voting. Religious and secular voters are increasingly different in their voting pattern higher up the income scale, which is also manifested in their attitudes on economic and social issues.  Thus it is richer Americans in richer parts of the country—more than the poor and rural—who are creating the cultural divide based on “God, guns and gays.”

The authors then asked whether income inequality may be affecting voting patterns, especially since it has “changed in recent decades,” but conclude (without much analysis) that income inequality does not explain the differences in voting patterns either. However, while income inequality (as far as the authors could establish) does not per se affect voting patterns, it “keeps the economic issues relevant and allows the polarization to continue.” Data at the international level indicates that countries with the biggest differences in voting preferences between rich and poor are those where the left and right parties are furthest apart on economic issues.

Although the authors could not offer an explanation for these results, they were confident in saying that a “typical Republican voter” was a rich person in a poor state and a “typical Democratic voter” was a lower-income person in a rich state. Further discrete findings were:

  • The red-blue divide is sharpest among the richer and more politically influential voters.
  • The voting gap between rich and poor is larger in republican states.
  • The division between rich and poor voters is highest in poor states.
  • Divisions between religious and secular voters are highest among high income voters.
  • The strongest rich-poor divides are those in which the major parties are furthest apart on issues of redistribution.
  • The data does NOT support generalizations that rich people vote based on economic issues while lower-income people are “more likely to be swayed by emotions.”

Here are my own possible explanations for these results:

1. The votes of higher income voters have a more significant impact than the votes of moderate and low income voters. Not only do wealthy people vote at higher rates, they are more likely to be politically active and support political issues and causes (as well as framing them) in public forums other than strictly voting. In summary, the concerns of wealthier voters are more likely to drive overall voter motivation at the polls.

2. When voters are more communitarian (“we’re all in this together”) on both social and economic issues, everyone in the aggregate is better off. This a reverse cause-and-effect explanation. That is, the Democratic-leaning values are what makes the state itself better off as a whole, so economic issues are neither as salient nor as divisive.

Which leads to the suggestion for additional testing for correlations between Gini coefficients, social welfare policy preferences, and voting patterns. Also not addressed by this study (but suggesting subject material for a future study) is the extent to which stricter state voting regulations might be discouraging lower-income individuals from voting.

Toward the end of their findings, the authors pose the question why Democrats have not been more successful than they have been. The economy is a key issue for many voters, not just low income ones, and there is more than ample evidence that the economy has generally performed better during Democratic administrations. The authors allude to some anomaly created by the confluence of electoral and economic cycles as well as deliberately gerrymandered redistricting, but it is not entirely clear, nor does it explain the phenomenon they are looking at. Perhaps it is because neither party has either coalesced or captured the debate about economic angst:  “Conservatives argue that the Democrats have lost relevance and authenticity, while liberals accuse the Democrats of selling out.”

As for predictions, the authors predict that increasing voter polarization could increase the probability of extremely close elections, and that “it will take more to swing voters away from their usual attachments.” Indeed, the recent election—with its extremely close results and sharp partisan rhetoric aimed at increasingly disaffected voters—seems to support this prediction.