By Steve Schifferes
The 2004 election has revealed a deeply divided electorate which is polarised more than ever over cultural issues as well as the war in Iraq.
Religion - rather than class, ethnic origin or education - has become the key determinant of voting in the 2004 presidential race, according to an exit poll conducted by the Associated Press news agency.
And moral issues were more important for voters than Iraq, the war on terrorism, or the economy.
According to the exit poll, 22% of the electorate said "moral values" was the issue that mattered most in how they voted - compared to 20% who cited the economy, 19% who cited terrorism, and just 15% who said Iraq was the key issue.
Not surprisingly, four out of five voters who cited moral values as their key issue voted for President Bush - as did the same proportion of those who cited terrorism.
In contrast, those most concerned about the economy voted four to one for Senator Kerry, as did three in four of those who cited Iraq as their main concern.
During the campaign itself, polls showed a steady rise in the number citing moral issues like abortion as the top issue facing the country.
In the 2000 election, many analysts saw a nation split down the middle - with Democrats increasingly concentrated in urban, coastal regions, with liberal values, while Republicans were more rural, southern or Middle Western, and held conservative values on issues like gun control and abortion.
That division has persisted in this election - for example, Senator Kerry carried urban areas by 56% to 43%, while President Bush won 56% of rural votes - but that difference weakened compared to the 2000 election.
Gay marriage may have been a decisive issue for many voters
In big cities, where the Democrats are ahead by two to one, Mr Bush polled 11% higher than in the last election.
What has divided voters in this election, however, are views on the Iraq war, and on new moral issues like stem cell research and same-sex marriage.
Those against gay marriage, for example, voted strongly for Mr Bush, as did those opposed to abortion.
And the electorate divided sharply over Iraq, with the 47% disapproving of the decision to go to war strongly backing Senator Kerry.
The social and cultural divisions in the American electorate are best expressed by the sharp divide in voting patterns by church attendance.
Two-thirds of voters who attend religious services regularly (once a week or more) backed President Bush rather than Senator Kerry - and they make up 40% of the electorate.
Those who never attend services, in contrast, backed the Democrats by the same margin - but they make up only 15% of the electorate.
Democrats also ran strongly among unmarried and young people, families with incomes under $30,000 a year, and among the highly educated.
Republicans voters tend to have higher incomes, be located in the South, be married, and are more likely to be white Protestant and male.
But none of these divisions is as sharp as religion in explaining people's votes.
In a close race, with turnout the key issue, both sides have sought to mobilise their core supporters as much as possible - and turnout has increased sharply in this election.
But it appears that the Republicans may have been marginally more effective in energising their base.
For example, turnout rose sharply in many of the 11 states (including Ohio and Michigan) which also rejected gay marriage where it appeared on the same ballot on Tuesday.
The ability of the Republican party to mobilise its religious base could prove to be the decisive factor in what will still be one of the closest elections in US history.
However, this approach risks further dividing the nation.
Recent polls suggest that supporters of the two rival candidates have little respect for their opponent.
An NBC News-Wall Street Journal poll asked people whether they agreed or disagreed with the statement: "No matter who wins the election, America will have a good president."
Just 30% said they agreed while two-thirds said they disagreed.
The sharp division revealed by the election, then, could persist well beyond voting day.
The exit poll was conducted for AP and the US networks by Edison Media Research/Mitofsky International during the voting on 2 November. The sample size was 13,531 and the margin of error is +/-1%.