TCS Daily

A Tale of Two Maps

By Patrick Cox - November 17, 2004 12:00 AM

The now familiar map of the United States, separated into red and blue states, makes the point, graphically, that the coastal population centers tend to vote Democratic while fly-over country leans Republican.

Unfortunately, the map's binary either/or electoral college nature overestimates the philosophical division within the country while failing to show the extraordinary degree to which Americans' voting behavior reflects the degree to which their own neighborhoods are more or less crowded.

A far better illustration, devised by Princeton University mathematician Robert Vanderbei, uses shades of purple to indicate the spectrum of election preferences within counties.

Here is a map, executed by Michael Gastner, Cosma Shalizi, and Mark Newman of the University of Michigan using his procedure:

Here is a map showing U.S. population density in 1990:

Comparisons of these two maps make startlingly obvious the extent to which population density predicts voter behavior. Though not a perfect match, the relationship is undeniable -- and ultimately enigmatic.

What, we are led to ask, could explain this relationship? How does the number of live humans per square mile either influence or reflect political philosophy?

The standard, rather unexamined, assumption is that rural America has more traditional cultural values that are associated with the Republican Party. These include religious, family and pro-military values. Urban population centers and surrounding environs, on the other hand, are associated with more progressive values associated with Democratic Party. These values are assumed to be more secular, progressive and anti-military.

While this may be an accurate description, no one, to my knowledge, has provided a convincing explanation for the differences between lower and higher density regions. Why would, after all, city life cause one to embrace liberal political views? Why would life in the country yield a conservative perspective? What, specifically, are the causative factors?

There has been a surfeit of speculation about psychological factors, but relatively few specifics and even less evidence. Urban areas do have higher crime rates and, while this attribute is widely recognized, there is little real analysis of such as a causative factor in political attitudes. One of the few efforts even to quantify the correlation of population density and crime rates comes from John R. Lott and David Mustard who studied the impact on crime rates of Right-to-Carry Concealed Handgun laws -- which is higher in rural areas.

One of the very few studies to actually examine the effects of population density on behavior and attitudes is "Measuring Helping Behavior Across Cultures" by Robert V. Levine of California State University, Fresno. Levine found, through a series of interesting tests, such as feigning a blind person trying to find and retrieve a lost letter, that

"Far and away the best predictor of helping was population density. Density was more closely tied to the helpfulness of a city than even characteristics like crime rates, the pace of life, economic conditions or environmental stressors like noise and air pollution. Overall, people in more crowded cities were much less likely to take the time to help. New York City was Exhibit A. Crowding brings out our worst nature. Urban critics have demonstrated that squeezing too many people into too small a space leads to alienation, anonymity, de-individuation and social isolation. Ultimately, people feel less responsible for their behaviors toward others -- especially strangers. Previous studies have shown that city dwellers are more likely to do each other harm. Our U.S. results indicate that they are also less likely to do them good, and that this apathy increases with the degree of city-ness."

This, of course, is only one study and may not take into account other less observable forms of "helping" so it may not be safe to read much into the author's conclusions.

Another fascinating and easily verifiable correlation may be tied only indirectly to the characteristics of population density. The red states, that voted for Bush in both of the last elections, it seems, are net receivers of federal tax revenues.

In 2002, Dean Lacy of Ohio State University and the Hoover Institution. published "A Curious Paradox of the Red States and Blue States: Federal Spending and Electoral Votes in the 2000 Election." He found that,

"Thirty of the U.S. states reap more in federal spending than their citizens contribute to the federal government in taxes. The other 20 states provide more in taxes than they receive in spending. In the 2000 U.S. presidential election, George W. Bush won most of the states that are net beneficiaries of federal spending programs, while Al Gore won most of the states that are net contributors to federal spending. A state's ratio of federal spending to tax dollars, particularly non-defense spending, is a statistically and substantively significant predictor of Bush's margin of victory across the states. A state's per capita federal tax burden is also associated with the election result: states with higher tax burdens gave higher vote margins to Gore. Compared to Clinton's state-by-state vote shares in 1996, Gore did worse in states that gained in federal spending per tax dollar from 1998 to 2000."

Even more specifically, Lacy wrote that, "Put another way, Bush's margin goes up by 2 percentage points for every additional dime of federal spending in a state per dollar of taxes paid by that state."

The same results held, in principle, in the Bush/Kerry election, and much has been made of this odd fact by those who want to characterize red states as welfare recipients. It is an odd charge, however, that the states that tend to oppose transfer payments politically benefit most from them financially.

One possible explanation of this seeming contradiction is that it is caused by progressive taxation and higher urban area incomes. Though the cost of living is higher in the cities than it is in rural areas, taxpayers with nominally higher incomes naturally pay more taxes. Most federal programs, however, are paid out based more on the basis of population statistics, regardless of income levels, leading to an outflow of revenues from high-income states to low income states.

The case can also made that the cost of delivering services to states with low population densities is greater than it is in urban population concentrations. It is, for example, easier to deliver mail to rows of consecutive mailboxes on city streets than it is to drive to a series of far-flung farmhouses. It is also true that labor and other costs are also lower in rural areas, so it is probably not safe to assume federal programs are actually higher in rural areas. Additionally, many significant programs, such as agricultural subsidies, are unaffected by the demographic scattering.

I think it is more likely that the net-recipient status of the red states reflects the fact that, in order to secure the support of states whose populations tend to resist transfer payments, supporters of redistributive programs are often forced to buy off red state approval for government spending programs. Still, these characteristics fail to explain the differences in the underlying differences in political attitudes between concentrated and dispersed populations.

All modern differences between rural and urban American life may, in fact, be irrelevant. They could, rather, be holdovers from more traumatic times. As I've lived and worked in both heartland farm country, in Idaho and Florida, as well as and some of America's largest cities, including New York and Boston, I am skeptical that current differences would yield the radical differences in political philosophies that we now observe. The caricatures put forth by both sides, of insular hicks in trailer houses and cowering victims in crime-ridden row houses, do not reflect the experiences of most people.

Less than a hundred years ago, though, the contrasting challenges of city and farm life were far more tangible. One data point is that, before the development of penicillin at the beginning of the first World War, a compound fracture, which was a not uncommon consequence of being thrown from a wagon, resulted in death about half the time. On the other side of the demographic divide, big city politics were often remarkably brutish and intrusive.

Victor Davis Hanson, if I understand him correctly, posits that the legacy of America's farmer/warrior past has had an enduring impact on America's traditional political views, now emanating primarily from the heartland. One might expect, if this is true, that those who live in closest proximity to agricultural activities and communities would maintain more of these traditional attitudes. George Bush's use of the symbols of agriculture, cowboy boots, big belt buckles and pickups, obviously appealed to these people.

If Hanson is correct, however, demography may very well be destiny, as technology has decreased radically the portion of our population that is involved, even peripherally, in agriculture. Moreover, the cultural connection between the modern descendants of agricultural families may soon fade because, as Thomas Sowell suggests, culture lasts only two or three generations, and there are fewer and fewer Americans with agricultural roots as time passes.

Ninety-nine out of a hundred Americans were engaged in agriculture at the time of the American Revolution. In 1953, about 15 percent of the U.S. population, or twenty-three million people, actually lived or worked on farms, and 54 million lived in areas deemed rural. Now, due to the impact of technology on agriculture, only 3 million people, or 1 percent of the U.S. population, live on farms. According to an admittedly revised definition, the rural population has risen to 59 million but has lagged far behind urban growth and is now only 21 percent of the U.S. total. This pattern continues today as growth in rural areas is running at less than a quarter of the urban rate.

On the other hand, agriculture is clearly not the sole political variable. In recent elections, residents of suburbs voted more for Republicans than did residents of more crowded cities -- and they did so in inverse proportion to their population density.

The statistician's perennial caveat is that "correlation is not causation." but there is little doubt that there is connection, largely unexplained, between ideology and demography. Depressingly deterministic as it is, this correlation, if it continues, may mean that future elections will be decided by immigration patterns, reproductive rates and technologies that allow more businesses and workers to locate in suburban and rural locations.

I would be happy to be proven wrong.


TCS Daily Archives