WASHINGTON (MCT) Pollsters taking the pulse of voters this political season are confronting growing obstacles from cell phones and an electorate that is increasingly walling itself off with caller ID and answering machines.
The response rate in phone surveys has plunged from about 40 percent in the 1980s to 20 percent or less now, making it harder and more expensive for pollsters to secure the samples they need.
These changes are causing some to wonder about the accuracy of poll results this year, especially when it comes to young adults, who are 50 percent more likely than the rest of the population to use cell phones but who are voting in ever-greater numbers.
The emerging difficulties pose special challenges for political campaigns because they rely so heavily on polls to detect shifts in public opinion and to target voters.
Pollsters acknowledge that hurdles are becoming higher and insist they are making adjustments enabling them to accurately measure public opinion. But they worry about the days ahead.
"If we'd have said 20 years ago that we'd be getting just a 20 percent response rate, we'd have been horrified," said David Moore, a former senior analyst for the Gallup Organization whose forthcoming book "The Opinion Makers" highlights the challenges.
He added: "In the long run, the industry is very concerned about changes. But in the short run, at least in this election cycle, people feel like they will be able to confront the problems."
Mark Mellman, who polls for many Democratic candidates, bemoaned "the panoply of technologies that allow people to be masters of their own environment" and avoid pollster calls.
Besides making polls harder to take, the diminishing response rate raises the question of whether people who elude pollsters have different attitudes from people who submit to interviews.
"Right now, the obstacles can be surmounted with the right kind of procedures. But it may get to the point in the not too distant future when the obstacles can't be surmounted," Mellman said.
With the deluge of political polls already under way, polling experts say the public needs to understand the fallibility of public opinion surveys. They advise people to be wary of Internet polling, small-sample surveys, polls sponsored by advocacy groups and overstatements about how many people have actually made up their minds.
Susan MacManus, a University of South Florida professor who directs several polls, said that she is troubled when pollsters and their partners draw firm conclusions from samples as tiny as 300 people. (Typical national polls have samples of more than 1,000 people selected randomly.)
"I am astounded at the small sample sizes that people use that statistically aren't very valid," she said. "When polls are really off, and often they are, people start distrusting the entire electoral process."
Darren Davis, a University of Notre Dame professor and polling expert, said people trying to make sense of polls need to beware of broad interpretations about tiny changes in numbers.
"Sometimes in polls, small fluctuations are just random and not always tied to how people view candidates. But these very small changes get taken out of context," he said.
The switch to mobile phones has forced pollsters to adapt.
According to the most recent government figures, nearly 13 percent of American homes were categorized as cell-phone only meaning that they had no land-line. A quarter of young adults are reachable by cell phone only, and that number continues to rise.
Pollsters tend to shun cell phones for several reasons. Among them:
• Cell phone users are able to automatically screen calls and are less likely to answer.
• Area codes of cell phones don't necessarily indicate where the user lives.
• Directories for cell phones are not available, and blocks of numbers available for purchase may not represent the geographical region being polled.
• Mobile phones are typically used by people on the go, perhaps in their cars or in situations where they are distracted and unable to complete long interviews.
Besides narrowing the traditional random sample pool, the switch to cell phones is makes it less likely that people reached by pollsters will be between 18 and 34 years of age a group that is voting more heavily than it used to.
As recently as the 2000 presidential election, pollsters had a nearly 1-in-3 chance of encountering a young adult in a phone call. By the 2006 congressional elections, the number of young people in land-line homes had dropped to 20 percent and earlier this year had dipped into the teens.
For pollsters, this means extrapolating from the responses of young people they do reach meaning they are talking to fewer young people than their surveys might suggest or trying to reach people on cell phones.
Del Ali is president of Research 2000, which conducts polls for the Post-Dispatch and other news organizations. He uses cell-phone numbers in polls, but said he takes extra care to make sure that young people reached on their cell phones are registered voters in the state or locale targeted.
Ali is among pollsters who say a bigger worry is Internet polling.
Pollster John Zogby, a pioneer in Internet surveys, acknowledges that cyberspace polling is still in its infancy. He has been working for years to build a database of 350,000 e-mails that he regards as representative of the nation. When conducting a national poll, he'll e-mail 50,000-75,000 people to take part. From the thousands who respond, he'll randomly select people for the poll.
Skeptics say Internet polling is flawed because an e-mail database may not be representative of the population at large. Zogby, who also uses phone surveys, professes confidence in his Internet results and argues that survey research must change with society.
"I know that it is the next wave," he said. "The telephone is becoming ungainly. It's still a useful tool (for polling), but we're anticipating that it won't continue to be a useful tool."
Meanwhile, Scott Keeter, director of survey research at the nonpartisan Pew Research Center in Washington, says poll-takers must constantly re-examine their methods.
"But we have a final exam every two years in this business, and in the last two cycles, the results were really quite good," he added.