FutureWire - futurism and emerging technology

Friday, October 29, 2004

Accuracy, Future of Phone Polls in Doubt

As the presidential election enters the homestretch, nobody seems able to make sense of incoming polling figures. One minute it's a dead heat, the next minute Bush is breaking out, and the next, Kerry's ahead. What's going on?

Some experts believe that traditional methods of polling -- particularly telephone polls -- are no longer reliable. Brian Vargus, a political science professor at Indiana University, believes that the era of telephone polling is coming to an end. "This may be the last election where you'll see such a proliferation of telephone polls," he says.

Several factors play into the decline in the reliability of telephone polling:

  • More Americans rely on cell phones, and many -- especially young people -- use them as their primary phones. Cell phone numbers aren't listed, and pollsters would hesitate to call them anyway as the recipient has to pay for incoming calls.
  • The prevalence of answering machines and Caller ID allows for the easy screening of calls. In many households, an incoming call from an unfamiliar number is automatically ignored.
  • Many newly-registered voters are poor, and don't have phones at all. If these voters turn out in large numbers, they will invalidate much of the polling, regardless of how they vote.
  • The response rate of telephone polls -- that is, the percentage of people who agree to participate in them -- has dropped dramatically, from as high as 80% in the 1960s to as low as 10% today. Some statisticians believe that the response rate is now too low to provide an accurate sample of opinions.
If not with phone polls, how else could pollsters gauge public opinion? Some have suggested a return to door-to-door polling and "man on the street" interviews. But in this day and age, those would likely be perceived as even more invasive than phone polling. Perhaps, as Internet access increases, pollsters could develop more scientific methods of Internet-based polling. Yet others believe that market-based predictors, in which participants buy and sell "shares" in candidates based on their potential, are the way to go.