In the past week alone, the bewildering array of polls have offered encouragement to supporters of both Republican and Democratic candidates.

Several national surveys have suggested that Obama is pulling clear with a commanding double-digit lead; others though have indicated that John McCain is still very much in the race, trailing by only one or two points.

Polling experts, meanwhile, point out that national surveys, while interesting, are not as relevant in terms of predicting the outcome of the election as polling numbers in crucial battleground states.

So while McCain may be neck-and-neck with Obama in a national survey, pollsters are scrutinizing the figures in states such as Ohio, Florida and Pennsylvania for clues to the November 4 election result.

For University of Wisconsin political science professor Charles Franklin, the fluctuating poll numbers are a source of comfort.

"If all the polls gave exactly the same result something would be deeply wrong with the theory of random sampling," Franklin, co-creator of the Pollster.com website, told AFP.

Scientific theory dictates that a sample of 1,000 people should show a margin of error of three percent each way, Franklin said.

"So in that sense the spread in the numbers we are seeing is reassuring, because we know that the random samplings are genuinely random; if it were less varied we'd know the pollsters are cheating."

Pollster.com is one of several of Internet-based polling aggregators that have sprung up in the past decade with the aim of collating and analyzing the endless stream of voter polls churned out by media and research outlets.

Other sites include RealClearPolitics.com, FiveThirtyEight.com and the Princeton Election Consortium.

Henry Brady, a professor of political science at the University of California, Berkeley and director of the institution's Survey Research Center, said aggregator sites helped to pinpoint trends in the election.

"They do an excellent job of reporting the polls," Brady said. "But the key thing to remember is that each poll is a random sample subject to error.

"So you really want to look at the across all the polls to see if there's a pattern. And the pattern at the moment is pretty darn clear -- Obama is ahead."

A survey this week which suggested that only one point separated Obama and McCain highlighted the hazardous nature of polling, Brady said.

"I think perhaps they should have looked at it and said 'Perhaps we got a bad draw' and did the survey again," Brady said.

"When you have so many polls saying Obama is ahead by five, six, seven points and then your survey puts the margin at one, I don't think you've got news, I think you've got error.

"Which is not to say they made a mistake -- it's just that this is what happens with polling. Every once in a while you get a rogue poll. It's the nature of the business."

Franklin, meanwhile, said the possibility of a so-called "Bradley Effect" -- the phenomenon of voters telling pollsters they will vote for a black candidate for fear of being seen as racist, and then voting otherwise -- would give analysts pause for thought.

"It's certainly the kind of thing that would keep pollsters up at night, and you kind of worry about it as a possibility," Franklin said.

"Recent research indicates that it has diminished significantly since the 1990s."

Related Links
Democracy in the 21st century at TerraDaily.com

Memory Foam Mattress Review

SPACE MEDIA NETWORK PROMOTIONS Solar Energy Solutions Tempur-Pedic Mattress Comparison ... Newsletters :: SpaceDaily :: SpaceWar :: TerraDaily :: Energy Daily XML Feeds :: Space News :: Earth News :: War News :: Solar Energy News