One reason is that pollsters often survey “likely voters.” But who is a likely voter? Well, the Huffington Post published this article to explain who they are and how pollsters identify them. Some key points from the article are outlined below.
In 2008, 213 million adults were eligible to vote, but only 68% of those adults told the U.S. Census that they were registered to vote and only 62% turned out to vote. We can expect turnout to be lower this year because historically, participation in midterm elections is significantly less than in presidential elections (voter turnout was 40% in both 2006 and 2002). Therefore pollsters target likely voters in an effort to make their numbers more precise, rather than diluting the data by sampling all registered voters--many of whom will not vote.
But how do pollsters know who the likely voters are? It’s tricky because (early voters aside) they can ask people if they plan to vote, but survey respondents often exaggerate their true intentions. In 2006, the Pew Research Center asked adult voters less than a week before the election if they planned to vote. 68% said they were registered and planned to vote in the election, while 54% said they were registered and rated their likelihood to vote as a 10 on a scale of 1 to 10. Ultimately turnout was only 40%.
Thus pollsters must take into account other factors to help them identify likely voters and produce more accurate data. For example, a different Pew study found that eligible voters who report intent to vote, have a history of voting, posses knowledge of voting procedures or who have high interest in or knowledge about politics are more likely to vote than those who do not.
Each polling group utilizes a different method (or a combination of methods) to identify likely voters, which is one of the reasons why their data varies. It doesn’t mean they’re all wrong, but just make sure you're one of the 218 million eligible voters that actually votes this year.