Why are Americans so divided over religious issues? Maybe it's those extreme messages.

The New York Times, "Economic Scene" , November 03, 2004

Have religious issues become more important in politics because too few Americans go to church?

That is the surprising suggestion of a new working paper by the Harvard economist Edward L. Glaeser and two doctoral students, Jesse M. Shapiro and Giacomo A.M. Ponzetto. (The paper, "Strategic Extremism: Why Republicans and Democrats Divide on Religious Values," is online here.)

The paper starts with a puzzle: In a majoritarian system like ours, political economists generally predict that candidates will converge toward the center of the spectrum, so as to attract as many votes as possible. This is the "median voter theory." But it doesn't seem to describe what's happened in American politics. On divisive religious issues like abortion, the two parties aren't hugging the center. They're abandoning it.

While most people know that the Republican Party has taken an increasingly strong anti-abortion position, the authors note that the Democratic Party has simultaneously moved in the opposite direction.

In 1976, the Democratic platform said, "We fully recognize the religious and ethical nature of the concerns which many Americans have on the subject of abortion," while terming a constitutional amendment overturning Roe v. Wade merely "undesirable." In this year's platform, by contrast, Democrats declared that they "stand proudly" for a woman's right to an abortion, "regardless of her ability to pay."

Conventional theory doesn't explain this shift. Even if voters were getting more religiously polarized -- splitting into secular and fundamentalist camps -- candidates would still have an incentive to split the difference.

"If some people are religious and some not, and voters simply choose the politician with views closest to their own, then a right-wing politician would want to take a stand on religious matters that is only just barely to the right of his opponent's position," Mr. Shapiro explained in an e-mail message. "That way, he captures the entire religious vote, but also avoids leaving the middle ground open for his opponent to invade."

So what's going on?

The economists propose a two-part answer.

First, there are actually two important voting decisions -- not just whom to pick but whether to vote at all. Candidates need to get their voters excited enough to come to the polls (or possibly to give money). Extreme positions can do that.

But positions that energize your base may also encourage the opposition to come out against you. That's where the second part of the model comes in. Candidates need a way to target their messages so their supporters are more likely to respond than their opponents.

That's where social groups like churches and unions come in. These groups provide friendly forums for candidates' direct or indirect messages. While outsiders may know something about a candidate's more extreme positions, group members know more -- because the messages are aimed specifically at them.

"When I go out and say, 'I want to tax all the rich and I want to end outsourcing,' I can give that message to an economically left-wing audience without the economically right-wing audience hearing it with exactly the same probability," Professor Glaeser said. "All you need is some ability to target your message, and then you're going to go to extremes."

Those in-group forums work, however, only if the groups are just the right size. They have to be small enough to be homogeneous and big enough to be influential. "The model has this very odd prediction that the power of social groups is most when they're roughly 50 percent of the population," Professor Glaeser said.

If a group is too small, it's not worth courting. But if it's too big, it includes too many of your opponent's supporters, making targeted messages impossible. If everybody goes to church or belongs to a union, membership in either group will not predict voting behavior.

"This is exactly what you see in the data," Professor Glaeser said. "The degree of polarization around religious issues is greatest in the places that are in the middle. It's not the Philippines, which are 100 percent religious, and not Scandinavia, where no one has attended a church in 40 years except for a wedding or a funeral. It's really these places like the U.S. that are in the middle."

Similarly, in states like South Carolina where most people go to church regularly, whether someone attends church does not predict how he or she will vote.

During the period studied, about 62 percent of South Carolina voters attended church at least once a month, but the churchgoers were only 4 percent more likely to vote Republican. In California, by contrast, only 38 percent reported monthly church attendance, and they were about 11 percent more likely to vote Republican.

The same is true for changing church attendance over time. In the 1970's, 57 percent of Minnesotans attended church at least monthly, and churchgoers and nonchurchgoers split their votes about the same way. By the 1990's, only about 45 percent of Minnesotans were regular churchgoers. About 40 percent of them voted Republican, versus 23 percent of nonchurchgoers.

Over the past several decades, the nation's church attendance has shrunk to the political sweet spot, while union membership has become too rare to reward sharp tilts to the economic left. Political platforms have diverged on religious issues and converged on economic issues.

Yet abortion rates show no significant change with the party in office, while tax rates rise significantly under Democrats -- the opposite of what the political rhetoric promises. This result suggests that politicians move away from the social center mostly to get votes ("strategic extremism") and diverge from the economic center because they actually prefer those policies ("nonstrategic extremism").

Since the success of extreme messages depends on keeping your supporters better informed than your opponents, the model suggests that changing news media could be as important as changing social groupings.

"If every time you say something in private to a religious group or a feminist group, it ends up on Drudge within three minutes in screaming headlines," Professor Glaeser said. "It's going to stop people from going to extremes."