Democrats should try appealing to more married voters

Much red and blue ink has been spilled on the narrative that's dominated our politics in recent years. We've all seen "the map" that outlines America's political divide and heard all the stereotypes and caricatures. And we've heard the explanations for the fissions: religion, race, intolerance and bigotry, homosexuality, moral values, and abortion. Others have debunked the whole thesis of a polarized America, saying there are really only purple-hued states and that this theory is a fabrication hyped by political activists, interest groups, and media elites.

Unfortunately, there appears to be little convergence on our understanding of what's going on here. How is one to make sense of it all? Are there red states and blue states? A closer look at the election results by county gives a different picture.

Many counties in blue states are actually red, but dominated by blue urban counties. In other words, the true pattern is blue urban vs. red rural and suburban. The mean population density for counties voting for President Bush was 108 inhabitants per square mile in 2000, and 110 in 2004. This compares to 739 for Al Gore and 836 for John Kerry. This is consistent across the board: As population density steadily decreases from the urban core to the rural periphery, Mr. Bush's share of the vote increases from 24 percent to more than 60 percent. Naturally, the Democratic candidates show the inverse relationship. Oddly, population density is highly correlated with the vote. This raises the question: Why do rural and suburban areas vote Republican and urban areas vote Democrat?

This is usually where the nonsense about religious rednecks and bluenose moral degeneracy begins. But an objective look at the census profiles of counties offers more plausible answers. A simple regression equation matching county characteristics against vote outcomes across 3,142 counties shows two significant variables: population density and the percentage of married households vs. female heads of household. For example, of the 100 counties with the lowest proportions of married households, Mr. Gore won 85 and Mr. Kerry won 90. Of the 100 counties with the highest proportions of married households, Bush won 96 in 2000 and 97 in 2004.

With one exception, age, sex, and race/ethnic characteristics are very weakly correlated with the votes - this includes classifications of Hispanics, Asians, Pacific Islanders, native Americans, and mixed race. But counties with higher proportions of white residents tended to vote for Bush while those counties with higher proportions of blacks tended to vote for Gore or Kerry. However, proportions of black voters are also highly correlated with female heads of household, and this proved to be the more significant factor.

These results offer us more reasonable insights into party polarization. Certainly urban singles, with or without children, have different policy priorities than suburban and rural married couples. These differences are somewhat reflected in the platforms of the two parties. The purpose of politics is to reconcile different preferences and order social priorities. But one suspects there's more to this story.

The issue of religion may give us a clue, but not the one we assume. It's not religious belief that distinguishes red rural from blue urban, but church attendance. Church attendance serves different functions in rural and urban communities. Whereas in rural Oklahoma church is the place to meet neighbors and friends on Sunday morning, in the city, that social function may be better served by Starbucks. The true political importance of religion is found in the nature of belief, between what sociologists call absolutism and contextualism. Religious denominations are divided politically in the US based on whether they are orthodox or not. But this distinction also divides the secular world, between traditional conservatives and progressive moderates. This means we have different ways of understanding the world that we can't merely dismiss as ignorant fundamentalism or misguided modernism.

These three legitimate distinctions among Americans - urbans vs. nonurbans; marrieds vs. nonmarrieds; and absolutists vs. contextualists - help illuminate our political differences over a variety of issues such as gun control, social spending vs. taxes, abortion, stem-cell research, education, foreign policy, immigration, and judicial nominations. I suggest that this framework might be considerably more productive in resolving political differences and setting our social priorities than calling each other funny names.

• Michael Harrington is a political scientist, policy analyst, and writer. He has taught at UCLA and has received national awards for his policy research.