Tampa, Florida -- Pick any day of the week and you can probably find multiple polls for the same race showing different results.
Why the discrepancies?
Experts say several factors can lead to different results.
First, who is being asked? Is the poll of likely voters, registered voters or someone who believes it's every citizen's duty to vote? Asking someone about whom they support and if they will vote are two entirely different things.
Second, is the poll based on a previous election turn-out model? This is the flaw the Connie Mack campaign is citing as why some polls show Bill Nelson ahead in their U.S. Senate race by double digits while the Republican's polling suggest a much tighter race. The Mack campaign claims the polls showing Nelson up by 14 points are based on 2008 turn-out levels.
"It looks like there is some credibility to that argument, because it assumes there will be the same percentage of younger voters who actually did vote as in 2008 and a lot of people are questioning that based on other polls that ask younger people how likely they are to vote," said University of South Florida Professor of Political Science Dr. Susan MacManus.
Third, how many cell phone users were polled? Experts say neglecting this segment of the population is a mistake given more than a quarter of all Americans now have just a cell phone and no landline.
"Individuals that have the landlineat home do not... they don't reflect the overall electorate," said 10 News Political Analyst Frank Alcock.
Experts add cell phone users are typically younger and more demographically diverse than landlineusers which can therefore skew results. Remember, robocalls to cell phones are also illegal so how does the poll adjust for that or are live interviewers used to gather data?
Fourth, when was the poll taken?
"Certain organizations might do their polling in a particular time slot each night, others might mix it up in terms of days of the week, I mean you're going to capture different demographics with those samples," Alcock said.