VERMONT - Labor Day is the traditional kick-off of the political campaign season. But given the way the November Senate race has been going, most Vermonters with a landline telephone have already received at least one prerecorded message from a candidate. Thus far, cell phone users have largely escaped this minor annoyance. That's because there is no directory of wireless phone numbers for political campaigns to consult, and FCC rules prohibit the use of automated dialers to knowingly call wireless numbers.
But what's a blessing for cell phone customers is a growing conundrum for political pollsters. As increasing numbers of Vermonters rely exclusively on cell phones, it's becoming much more difficult for public research firms to get an accurate snapshot of public opinion. As a result, political pollsters are struggling to find new ways to reach out and touch someone.
"I don't think our industry knows exactly what we're going to do about it," admits Anna Greenberg, a Washington, D.C., pollster who occasionally works for Vermont Democratic candidates. "At the moment, there are problems that are pretty insurmountable."
Consider the numbers. In 2004, just 6 percent of all likely voters had cell phones but not landlines. However, that number was twice as large among likely voters under 30. Today, it's estimated that 34 percent of 18- to 24-year-olds have cell phones only. They tend to be minorities, from lower-income households and in urban areas.
"I have maintained for a number of years that we have a problem regarding response rates and the increasing number of cell phones," agrees John Zogby, president and CEO of the Utica, New York-based polling firm, Zogby International. "I don't think we've reached the crisis point yet, but we're getting there."
Wireless phone companies do not currently sell their customer names and phone numbers. However, public research firms can buy lists of area codes and exchanges - the first three digits after the area code - that are potential cell phone numbers. Pollsters dial them randomly to find working numbers - a labor-intensive and time-consuming process. It's also virtually impossible to get an accurate sample from a specific geographic area since mobile phones are, by definition, portable.
Pollsters are tackling the problem in several ways. Zogby's company is aggressively pursuing its online or "interactive" polling, he says. Zogby International still operates its call center, but since January, it's been conducting more opinion polls online than by telephone. Zogby likens the shift to the one that occurred in the early 1970s, when polling firms switched from mostly face-to-face questionnaires to telephone surveys.
Online polling is "a methodology that is 85 to 90 percent there," Zogby says. "It's sufficient to say that not only is it the wave of the future, it's the wave of the now."
Zogby also says he's "intrigued" by other technologies such as text messaging. In 2004, Zogby International partnered with the cell phone manufacturer Motorola and the youth voter campaign "Rock the Vote" to send out text-message polls to 42,000 cell phone users between the ages of 18 and 34. The poll asked people whether they intended to vote and, if so, would they choose George Bush, John Kerry or Ralph Nader? That poll showed Kerry beating Bush by 14 percent - which, in fact, he did among 18- to 34-year olds in the general election.
"Is that the broken clock that tells the right time twice a day," Zogby asks, "or are we onto something?"
Alas, the day is coming when Vermonters are polled on their cell phones as well, suggests James Dayton, senior vice president of the Burlington-based national public research firm, ORC Macro. Although the company doesn't work for political candidates or conduct election polling per se, it does conduct telephone research on behalf of state and federal agencies, and occasionally asks questions of a political nature.
According to Dayton, in the next few weeks ORC Macro plans to start calling wireless phone numbers at random to test participation rates. The pilot project, which is sponsored by the Massachusetts Department of Health and the University of Medicine and Dentistry of New Jersey, aims to reach at least 400 people, which is considered a statistically reliable sample.
But will people be willing to answer an opinion poll while they're eating out, driving or sitting in a movie theater? And even if they are, how to factor in all the variables around cell-phone use: Is the phone shared among several individuals? It is used for both home and business? Is it reserved for emergencies or outbound calls?
Since federal law prohibits the use of predictive dialers - the automated dialing technology that causes that several-second delay between when you answer your phone and when the caller is connected - all cell phone polls have to be conducted manually. And the wireless customer has to pay for the incoming call. Ultimately, Dayton suggests, the full switch to cell-phone polling won't happen until that changes.
Greenberg predicts that telephone polling will remain the dominant mode of opinion gathering in 2006. As she points out, voters in off-year elections tend to be older, settled, and in the habit of voting regularly - the least likely group to own cell phones.
Pollsters haven't yet called an election wrong due to their inability to reach cell-phone users. But by 2008, when pollsters are stitching together their results from Internet, in-person and text-messaging polling, there could be a different result.