Researchers disagree on impact of not calling cell phones when polling

By Leah Sorensen and Leah Sorensen

These days, you can’t pass a young American without seeing a cell phones glued to his ear or a Side Kicks glued to her fingers. A lot of these young people have moved into a cell-phone-only lifestyle. They no longer use landline telephones and this could be impacting elections.

This election year we are constantly bombarded by the press with new polls that say which candidate has the most support. Like CNN’s presidential poll the week of Oct. 10. It showed that 53 percent polled supported Democrat Barack Obama, and that 39 percent polled supported Republican John McCain.

Polling influences the perceptions of the voters as well as the campaign tactics of the candidates when it comes to elections.

“[Polling], when done correctly, is the only way you can understand what large groups of people, like registered voters, are thinking about things,” said Rob Daves, a teaching specialist at the Hubert H. Humphrey Institute at the University of Minnesota who previously ran the Minnesota Poll at the Minneapolis Star Tribune.

Polling groups can collect their information in a variety of ways. Random-digit dialing is the typical method used. With the random-digit dialing method, every phone number has an equal probability of being called, so unlisted numbers are not excluded.

Recently many people have speculated that polls about the upcoming presidential election are inaccurate because they are not calling cell phones. According to a survey by the Centers for Disease Control and Prevention, 31 percent of 18- to 24-year-olds live in households that only use a cell phone, as well as 34.5 percent of 25- to 29-year-olds.

Although cell phone numbers are unlisted they can still be called by pollsters.

In fact, the numbers can be obtained through the same means as the landline numbers, which pollsters get from registered voter lists.

“The (cell phone) company that controls how numbers are assigned know which ones are cell phones and which ones are landlines,” Daves said, but “It costs about three times as much to do a cell phone interview as a landline interview.”

Daves said there are two reasons cell phone polling is more expensive. “The first is that there is a federal law that prohibits calling cell phones with an automated calling system, which is a computer program that holds a database of phone numbers that are randomly called. That means calling cell phone numbers by hand, which means paying people to do it, which costs more money,” Daves said. “The second expense comes from a custom that has developed in the polling industry to offer cell-phone users a cash incentive to participate to help offset the cost of using the cell-phone owners’ minutes for a survey.”

Researchers say this does not mean that younger voters are being left out of poll results. “For surveys limited to landline telephone sample, we correct for under-representation of groups more likely to be cell-phone-only through demographic weighting,” said Larry Hugik, of Princeton Survey Research Associates International, the company that does the Minneapolis Star Tribune’s polling now.

“If the U.S. 18- to-24-year-olds make up 24 percent of the population, and you get 12 percent in your survey, then you have to weight your sample to make sure those numbers reflect the population,” Daves said.

Polling companies are keeping a close eye on how the cell-phone-only population might be affecting polls.

“The bottom line is that people who are cell-phone-only are close enough to their demographic counterpoints who are not cell-phone-only that excluding them does not make a significant difference in results,” Hugik said.

The Pew Research Center’s report “Cell Phones and the 2008 Vote” says: “Traditional landline surveys are typically weighted to compensate for age and other demographic differences, but the process depends on the assumption that the people reached over landlines are similar politically to their cell-only counterparts.”

Pew conducted three surveys this summer starting in June and ending in September of that called both landline telephones and cell phones and broke the results down in to three groups: a landline sample, a cell-phone-only sample and a combined sample.

In September, the landline sample was split down the middle with 45 percent of those polled supporting McCain, and 45 percent supporting Obama. In the cell-phone-only sample, 55 percent supported Obama and 36 percent supported McCain. When the samples were blended, it gave Obama a two-percentage-point lead over McCain with 46 percent supporting Obama and 44 percent supporting McCain.

“(The Pew) surveys suggest that this assumption” – that weighting the samples is sufficient to correct for not polling people who only use cell phones – “is increasingly questionable, particularly among younger people,” according to Pew’s report.