by Bill Bishop
Politicians and parties have exploited this social evolution, and in doing so, they have exacerbated partisanship and division. Elites have always been more partisan, more extreme, and more ideological than regular voters. But today moderates on all sides are rebuffed, and those who seek consensus or compromise are squeezed out. Paul Maslin, Democratic presidential hopeful Howard Dean's pollster in 2004, explained it this way:
If I had to say one true statement about the entire process you are describing, I think that at the national or state level, it's making life increasingly difficult for people who are trying to thread the needle, to find the swing voter. In a way Karl Rove and Howard Dean and [Dean campaign manager] Joe Trippi were all right here. It's probably one of the things that's driving our politics into a more polarized situation. While the swing vote and the classic vote in the middle still matter, you are much more willing to say now that you ignore at your peril your own base. Because as everything spreads apart, the base becomes more important because they are demographically more together. You don't have a whole bunch of 51–49 communities out there. You have more and more 60–40, 65–35, 70–30 places. Well, you better damn well be sure you maximize your 70–30 votes, whether it's inner-city African Americans or liberal, educated Democrats or whether it's suburban, conservative Republicans or small-town, main-street, or Evangelical Republicans. We have to maximize our base, and they have to maximize their base. Ergo, polarization.
The country may be more diverse than ever coast to coast. But look around: our own streets are filled with people who live alike, think alike, and vote alike. This social transformation didn't happen by accident. We have built a country where everyone can choose the neighborhood (and church and news shows) most compatible with his or her lifestyle and beliefs. And we are living with the consequences of this segregation by way of life, pockets of like-minded citizens that have become so ideologically inbred that we don't know, can't understand, and can barely conceive of "those people" who live just a few miles away.
2. THE POLITICS OF MIGRATION
OPPOSITES DON'T ATTRACT. Psychologists know that people seek out others like themselves for marriage and friendship. That the same phenomenon could be taking place between people and communities isn't all that surprising. "Mobility enables the sociological equivalent of assortative mating,'" explained social psychologist David Myers. Assortative mating—the tendency of similar types to pair up—has been studied as a cause of poverty and autism. But Myers was making a different point. Our wealth, education, and ability to move have allowed us to seek "those places and people that are comfortably akin to ourselves."1
The United States was shaped by migration. Explorers found their way on foot through the Cumberland Gap. Pioneers pushed west in wagon trains. Blacks left the dismal economy and deadly culture of the cotton South in the "great migration" of the first half of the twentieth century. Cubans fled to Florida after the overthrow of Batista in 1959. These mass displacements weren't what Myers was describing. He was identifying a different kind of movement, a migration of self-selection. The Big Sort included an element of personal discretion. People still moved to find good jobs, excellent schools, and safe neighborhoods. But an expanding economy, rising levels of education, and the breakdown of older social groupings had injected more personal choice into the selection of where to move and how to live. Amenities became more important as people sought out a particular kind of church or a special music or art scene. (For instance, Austin is brimming with baby boomers who moved here for the cosmic cowboy sound.) Americans could move to places that reinforced their identities, where they could find comfort among others like themselves. These weren't political choices, but they had political consequences.
Sorting the Evidence
After Bob Cushing and I discovered that Americans were segregating politically, we searched for corroborating evidence that this phenomenon was linked to larger social movements. We hoped not only to confirm the sorting we saw in elections but also to explore the nuances of what appeared to be a massive social and political reconfiguration. So we gathered what evidence was available and devised three tests of the Big Sort's influence. The first measured the voting patterns of communities over a number of presidential elections. If communities were collecting overwhelming numbers from one party or the other, majorities within communities should grow. The power of "assortative migration" would attract more Democrats to Democratic counties and more Republicans to Republican counties. By the same token, as Democrats left heavily Republican areas, those places would become even more Republican and vice versa. To be significant, this couldn't be a regional phenomenon. The sorting should be more than just the South switching from solidly Democratic to staunchly Republican. The whole nation ought to be undergoing the same kind of political separation.
Our second test would calculate the power of place. We wanted to see if geography trumped the measures normally used to designate political leanings. The most talked-about pattern of the past two presidential elections has been the overwhelming support churchgoers gave to the Republican candidate. If geography mattered, we should see a difference in churchgoers depending on the political cast of their home counties. Liberal churchgoers would live in one place and conservative churchgoers in another. If place had a special effect on people's politics, all union members wouldn't be the same either. Union members in Republican counties would have different beliefs from those in Democratic counties.
Finally, if sorting into like-minded communities had been taking place since the 1970s, we figured that we should be able to look back and see some corresponding demographic trends. We ought to be able to take advantage of the fact that hindsight is 20/20 and find the shifts in population that corresponded to the balkanized communities we live in today. Our third test searched for demographic movements that differentiated Republican places from Democratic ones over the past thirty-six years.
Test One: Does Like Attract Like?
For this test, we returned to the county-level presidential votes that had led us to our first story about political sorting and calculated how loyal each county had been to the two major political parties since World War II. Some counties (346, to be exact) had voted for the same party in every presidential election since 1948. In each election thereafter, another group of counties picked a side and stuck with it through the 2004 contest. Fifty-four more tipped in 1952; 536 tipped in 1968.
Before counties tipped, we found they were on average quite competitive. The difference between Republican and Democratic candidates over the years was just 2 or 3 percentage points in untipped counties. But here's the interesting part about the tipping phenomenon: once a county tipped, the spread kept growing. The average vote spread in presidential elections among tipped counties was huge—an overwhelming 20 percentage points in most elections. This was particularly true for Republican counties, which saw the margins for Republican presidential candidates increase over time. In addition, once these counties tipped, they grew more partisan. The trend was stronger in Republican than in Democratic counties. We surmised that this difference was caused by the tendency of Democratic counties to attract a more diverse population—more ethnic minorities, more people born outside the United States, more young people, and more people with college degrees. (I will discuss all of this later in the book.)
We found that Republican counties tended to become more politically segregated than Democratic counties.* This happened in part because Republican migrants were unusually attracted to Republican communities. Between 1995 and 2000, 79 percent of the people who left Republican counties settled in counties that would vote Republican in 2004—and they were most likely to move to counties that would be Republican landslide counties. We don't know the politics of individual movers. We do know that when people left counties that would vote Republican in 2004, they were two and a half times more likely to move to other counties that would vote Republican than to those that would vote Democratic. By contrast, people who left count
ies that would vote Democratic in 2004 migrated to both Republican and Democratic counties without showing much of a preference for either—although they were unlikely to move to counties that would become Republican landslide counties.
As a result of this sorting, most counties were zooming off in partisan directions. Between 1976 and 2004, the gap between the parties increased in 2,085 counties; only 1,026 counties (33 percent) grew more competitive. California is the stereotypical "blue" state. But within California, 17 counties grew more Democratic after 1976, and 30 became more reliably Republican. Only 11 California counties (19 percent) became more closely contested. In 1976, 44 percent of San Francisco County's population voted for Republican Gerald Ford. Over the next seven presidential elections, the percentage of San Franciscans voting for the Republican presidential candidate dropped every four years. By 2004, just 15 percent of San Francisco's voters supported George W. Bush. San Francisco didn't become more Democratic because its population grew; the number of voters in San Francisco County hadn't changed since 1948. San Francisco was transformed because Democrats sorted themselves in and Republicans sorted themselves out. Orange County was always Republican. But despite a population that nearly tripled (and in a state that grew increasingly Democratic), Orange County voted more Republican in 2004 than in 1964, when Barry Goldwater and the John Birch Society were going strong. Literally next door, Los Angeles followed a more mixed path until 1988 and then became increasingly more Democratic (see Figure 2.1).*
This process of self-segregation would be inconsequential if only a few Americans lived in politically homogeneous counties. But the numbers, we learned, aren't small. In 2004, one-third of U.S. voters lived in counties that had remained unchanged in their presidential party preference since 1968. Just under half lived in counties that hadn't changed since 1980, 60 percent lived in counties that hadn't changed since 1988, and nearly 73 percent lived in counties that hadn't changed since 1992, voting consistently Democratic or Republican for four presidential elections in a row. National political choices were being carved into local geographies.
At the same time, we found that the number of counties with landslide majorities continued to increase. In the exceedingly close election of 1976 (Carter versus Ford), 38 percent of the nation's counties had a spread larger than 20 percentage points. In the exceedingly close election of 2004 (Bush versus Kerry), more than 60 percent of all U.S. counties produced landslide elections. As some 10 million Americans moved each year from one county to another, counties clearly were growing less competitive and more politically segregated.
Figure 2.1 Distant Neighbors
The politics of Los Angeles and Orange Counties
are diverging, as seen in presidential voting.
Orange County
Los Angeles County
Source: Dave Leip's Atlas of U.S. Presidential Elections, http://www.uselectionatlas.org.
The tipping phenomenon was fractal—it appeared no matter how large or how small the geography. Political commentators blame much of the nation's ideological polarization on the switch in the South from Democratic to Republican. Indeed, the South has become increasingly partisan since 1976 as it has become solidly Republican. But we found that every region in the country has become more segmented as it has tipped toward one party or the other. The U.S. Census Bureau divides the states into nine regions. All nine grew more segregated politically over the past six elections.2 And the South is not the only region that has switched allegiance. The Pacific Coast, the Middle Atlantic States, and the North Central region were all Republican in 1948. Now these regions are strongly Democratic—and they are more politically lopsided than much of the South. Within the nine regions, there are also fewer competitive states. In the 1976 presidential contest between Carter and Ford, nineteen states had margins of 10 percentage points or more. By 2004, thirty-one states had at least a io-point margin. In 1976, the average presidential election margin in the states was 8.9 points. In 2004, it was a bleak 14.8 points.3
Test Two: Does Geography Matter?
Is it just my imagination, or is Lubbock, Texas, really a very different place politically and culturally from Cambridge, Massachusetts?* Of course. (Lubbock is the Republican town, in case you didn't know.) But generally, are there significant differences in the lifestyles and beliefs of people living in solidly Republican and solidly Democratic counties? To find out, we compiled polls conducted by the Pew Research Center from 1996 through 2004 and analyzed the results by how the counties voted in the 2004 election.4 We compared strong Republican counties (where Bush won by 10 percentage points or more) with strong Democratic counties (where Kerry won by 10 points or more). We found the following:
In strongly partisan Republican counties, 57 percent of the people were married. In strongly Democratic counties, 47 percent of the people were married.
Only 21 percent of the people in Republican counties earned more than $75,000 a year. In Democratic counties, 29 percent earned that.
Republican counties were 86 percent white. Democratic counties were 70 percent white.
In Republican counties, 46 percent of the people said that they went to church at least once a week, and half described themselves as Evangelicals. In Democratic counties, only 34 percent of the people went to church at least once a week, and 32 percent were Evangelicals.
These figures are misleading, however. The standard way to calculate public opinion is to take a group—Evangelicals, the rich, the young—and then describe how this supposedly homogeneous group thinks or votes across the nation. People who go to church once a week or who describe themselves as Evangelicals are thought to be stand-up Republicans and early supporters of the war in Iraq. Nationally, that is absolutely true. But Evangelicals living in counties that voted heavily for Kerry in 2004 are an entirely different breed from those living in Republican landslide counties. According to our analysis of the Pew Research Center's polls, less than half of the weekly churchgoers and self-described Evangelicals in heavily Democratic counties supported the war in Iraq in 2004. In heavily Republican counties, however, this same demographic group supported the war three to one.
Regardless of demographic category—age, gender, religion, occupation—Pew found a difference in support for the war based on geography. Labor union members were against the war in Democratic counties but for it in Republican counties. (Nearly 30 percentage points separated union members in strong Democratic counties versus strong Republican counties.) Women were against the war in Democratic counties but for it in Republican counties (a difference of 23 percentage points). The partisanship of place overpowered the categories that researchers normally use to describe durable voting blocks.
Scott Keeter at the Pew Research Center used findings from a large poll taken in 2004 to conduct his own test of the power of geography. His unpublished report runs 136 pages.5 Each page tells the same story: the differences between partisan counties were wide and deep. For example, 48 percent of the people living in Democratic landslide counties felt "strongly" that homosexuality "is a way of life that should be accepted by society." Only 21 percent of the people in Republican landslide counties agreed. In Republican counties, 49 percent believed "strongly" that homosexuality should be "discouraged by society," compared to 27 percent in Democratic counties. These two Americas, separated by county lines, disagreed significantly on the war in Iraq, the USA Patriot Act, and the use of military force in carrying out foreign policy. In Republican strongholds, half the people had guns in their homes; in Democratic areas, only 19 percent did.
Test Three: The 20/20 Hindsight Experiment
We began our "hindsight" experiment by dividing the nation's more than 3,100 counties into four groups based on the results of the 2004 election.* There were two groups of landslide counties—places where either George W. Bush or John Kerry won by 20 percentage points or more. That left two other groups of counties, one Democratic and the other Republican, that were competitive in the 2004 election—places where t
he vote totals for the two candidates differed by less than 20 percentage points. Each group contained a sizable proportion of the American population. The two groups of Democratic counties contained about 128 million people in 2000. The two groups of Republican counties were home to about 152 million people.
For this test, we examined these groups retrospectively, tracking them through time to see if and how their demographic composition changed. We knew where the groups of counties stood at the time of the highly polarized 2004 election. What we wanted to know was how they got there and whether the four groups had anything more in common than how they voted for president on the first Tuesday in November 2004.
First, we looked to see if the groups had any political coherence. From 1948 to 1960, the four groups jumped about, voting for Democrats in some years, Republicans in others. In 1976, the groups all voted at about the national average (see Figure 2.2). Beginning in 1980 with the first election of Ronald Reagan, the counties began to diverge in their political inclinations, and they continued to separate for the next quarter century.