Polling: A Greater Margin for Error
By Adriel Bettelheim, CQ Staff
To avid followers of election year metrics, a University of New Hampshire Survey Center poll of 519 residents last month could be seen as a sign that the stubbornly independent “Live Free or Die” state, scene of John McCain ’s greatest primary triumphs both in 2000 and this year, was getting swept up in Obamamania. It showed Barack Obama narrowly ahead, 46 percent to 43 percent, with 3 percent supporting other candidates and 8 percent undecided.
But the way pollsters got to these results was perhaps more telling than the top-line numbers. A full 28 percent of those surveyed initially replied that they were still trying to decide whom to support — not so surprising in a state where about 38 percent of registered voters are unaffiliated with either major party. These “don’t knows” are a continual irritation for public-opinion gatherers, particularly in the middle of a presidential race that’s attracting intense worldwide interest. So the interviewers, employing a decades-old tactic of squeezing out clear choices, pressed the holdouts to decide by asking them which way they would vote supposing the election were held that day. Most made a choice.
Did the surveying tactics make the results misleading, or were they part of a well-tested strategy for predicting voter behavior? Is this type of opinion research an effective way of assessing how a campaign is going? Or is it primarily an exercise in generating headlines that detracts from more important questions about public preferences and attitudes?
These questions are the kind that nag at pollsters in most election years. But for a number of reasons having to do with the candidates, voting trends and the voters themselves, pollsters this year face a number of unique challenges in getting to the heart of public opinion.
more...
http://www.cqpolitics.com/wmspage.cfm?docID=weeklyreport-000002933553