Voters are bombarded with presidential polling numbers during election season. But many of the polls show different results, and voters have trouble deciding which polls — if any — are accurate.Different polling results come from the different methodology of polling organizations. Most are weighted based on voter turnout predictions, but this method could prove problematic, said Kirby Goidel, Public Policy Research Lab director.”The tricky part is filtering out who is mostly likely to turn out on election day,” Goidel said. “You would expect variation. There would be something really odd if they came up with the same numbers.”He said the numbers are attributed to different models used to predict voter turnout. The traditional model for turnout, which is based on likely voters, is still in place. But with all the hype of this election, there is another model which assumes a higher voter turnout, he said.There are many factors in determining a likely voter, said Richard Nelson, public relations professor. Pollsters look at voting history, age, ethnicity and income levels to make an educated assumption as to who is more likely to vote.”The likely voter issue is a problem that plagues this year’s polling data,” Goidel said. “Younger voters usually support Sen. Barack Obama, so it is an issue as to whether or not they will actually vote.”Older voters, who tend to lean slightly conservative, historically have a better turnout, Nelson said. They get into the habit of voting and are more concerned with health care, benefits and taxes, he said.Goidel said there are a few factors in this election that makes polling accuracy problematic, such as the Bradley effect.The Bradley effect occurs when white voters, worried that they will be seen as racist, publicly support a black candidate while secretly voting for another candidate.Another problem is many polls do not account for cell-phone only users, said Robert Hogan, associate professor of electoral politics. Because so many college students — who typically support Democrats — do not have a land line, they are left out of most polls, Hogan said. That may be giving Sen. John McCain an artificially high number. Accurate registration numbers often cause problems, Goidel said. Pollsters rely on the respondent to tell them if they are registered, so any dishonesty may affect data.”One problem we have with accuracy is turnout,” Nelson said. “People say they will turn out to vote and then make up excuses for not voting.”As Election Day draws near, candidates use the polls more for the issues, Goidel said. The candidates change the importance and message of an issue based on the strength of voter concern to persuade voters, he said.Goidel said supporters of the leading candidate are usually energized by high poll numbers. The other side tends to be discouraged.”When there is an increase in momentum, you have a higher voter turnout because one side gets excited about going out and voting for their candidate,” he said.Ryan Dwyer, mechanical engineering junior and McCain supporter, said he does not pay extremely close attention to the polls but does look at them from time to time. “It’s all going to change come election time,” Dwyer said. Josh Wilson, finance senior and Obama supporter, said he keeps up with polling numbers, but he does not put much faith in them.  The best way to look at polling numbers is to look at an aggregate of polls rather than putting too much weight on one poll. Doing so gets rid of the bias of one poll, Hogan said. But even looking at an aggregate may be difficult, given the high number of polls, he said.”If you look at multiple polls you will have a better sense of where the election is than if you just looked at one poll,” Goidel said. “I would never look at a single poll.”Goidel said the narrative is often interpreted in terms of declining numbers. When a candidate starts to decline, the angle is always “why are they falling,” just as the story for those with higher numbers is “what is he doing to cause an upward trend,” he said.”Most polling organizations see their credibility on the line in terms of the election results,” Goidel said. “Any of the long standing organizations will usually be pretty accurate.”Hogan said all polls strive to be non-partisan, but each poll has a different methodology, causing numbers to differ. Most polls are done by random digit dialing between 600 and 1200 people, he said. “Because you are taking samples, you’re going to get a sample that is slightly skewed,” Nelson said. “That’s why you get samples that are taken frequently over time.”  Goidel said it is hard to tell the accuracy of polls because it could all change on Election Day. The polls are just a snapshot of what is happening at a given time, he said. “If there was ever a period in which polls could be wrong, this is the election cycle, because of factors such as having an African American nominee, cell-phone only voters and predicting voter turnout.” Goidel said.—-Contact Steven Powell at [email protected]