Numerous public opinion polls taken recently in the United States show the race for president between Republican Party incumbent George Bush and Democratic Party nominee John Kerry is virtually tied - much too close to predict which candidate will win on Election Day, November 2. While the campaigns and the news media follow the results of these polls closely, what role do they play in the election?

Politicians in the United States are fond of saying the only poll that really counts is the one on Election Day.

While that may be true, the campaigns, news organizations, and private companies spend millions of dollars each election cycle in an effort to determine what voters are thinking.

The media focus on the so-called horserace, the head-to-head contest between the candidates for president, as well as the top issues, which surveys say in this campaign are the war on terrorism, the continuing conflict in Iraq and the U.S. economy.

Political Science Professor Dennis Johnson of George Washington University says the campaigns of both presidential candidates use public opinion polling for another reason - to plan their strategies. "Polling is extremely important because it will tell you whether an ad (TV advertisement) is working. It will tell you whether a speech is the right thing to say. It will tell you whether you should go into a different state because your (poll) numbers are going down in popularity. It helps you in determining strategy, in terms of the message you have and if things are working. So polling, all together, is sort of like an instantaneous weather vain. It tells you whether things are going right or wrong and what you have to do to improve," he says.

While the public opinion polls have shown President Bush and Senator Kerry in an extremely close race, many of those surveyed give Mr. Bush higher marks on which candidate would do a better job leading the war on terror, while Mr. Kerry is usually preferred as the candidate who can best handle domestic issues.

Frequently, different polls show different results in what are called battleground states.

These are states where surveys say the race is very close and they are likely to play an important role in how the election is decided.

Political science professor Stephen Wayne of Georgetown University says polling is not an exact science and that is why some surveys disagree about which candidate is ahead. "Now we have a problem with polls in the United States today and the problem stems from how the polls are conducted. Polling operations use a technique where a computer spits out randomly a phone number, not a person who has a cell phone, but a regular line phone and those people are called by groups that do this between six and nine at night. And the hang-up rate of people who don't want to be disturbed by these calls is well over 50-percent. So we don't know if people who hang up the phone are any different from the people who answer all the questions," he says.

Professor Dennis Johnson of George Washington University agrees, saying poll results are based on trust. "We never really know for sure if people are telling the truth. We suspect they are, or we suspect enough people are telling the truth when they answer a poll. There can be all kinds of problems with a poll. You can poll the wrong people. You can poll too many women, for example, or you can poll too many people from a certain part of the country. You can ask misleading or leading questions. So there are a lot of things that can go wrong with polling," he says.

Larry Sabato, a political analyst and Director of the University of Virginia's Center for Politics, says in a very close election, some polls can actually have an impact on how people cast their vote. "You know there has been a lot of research in that area. Sometimes there is a bandwagon [follow the leader] effect. It is possible for the candidate who is ahead to gain a little among undecideds simply because they want to be with a winner. So I think we need to be very cautious about polls. The problem is almost no one is, and certainly we don't find that news organizations generally are very cautious," he says. "Look at the polls. They are all over the lot. On the same day you have polls released which show a wide Bush margin, a narrow Kerry margin, some dead ties. It makes no sense and it makes no sense because polling is not a science. It is an art."

For the first time the major television networks in the United States have agreed to be more careful in using polls to project winners in various states on Election Day.

Four years ago the organizations were embarrassed after declaring Democratic Party candidate Al Gore the winner in Florida.

Hours later the networks were forced to reverse themselves, awarding the state, and the presidency, to George Bush.

The race was so close it triggered a lengthy legal battle and a major debate over how polls influence voters in U.S. elections.