It's been said that "words, like eyeglasses, blur everything that they do not make clear." The same holds true for political polling.
So voters shouldn't feel bad if they're confused by poll information that often seems contradictory. For example, news outlets reported yesterday that presumptive Republican nominee Sen. John McCain had eclipsed his Democratic rival Sen. Barack Obama and led him by five percentage points. Today, the headlines blare that McCain is gaining fast on Obama in the polls and the latter's lead among voters had been cut in half.
How is that possible if McCain was already polling ahead?
A few things can greatly affect poll results and give conflicting views of the political landscape. One is timing and over what period people were asked questions about their political leanings. Both of the aforementioned polls––the first a Zogby poll and the second an NBC News and The Wall Street Journal survey––mentioned that events like the Olympics and the conflict between Russia and Georgia were drawing attention to world affairs, where McCain tends to have a stronger perceived position. Also, the Zogby poll was conducted Thursday 8/17 through Saturday 8/19, while Obama was vacationing in Hawaii. The NBC/WSJ poll was conducted Friday through Monday. With the media coverage of both events changing on a weekly and even daily basis, however, the exact timing of when polls are conducted will have an effect on what voters are most concerned with.
Another important factor in interpreting polls is the population being questioned. The Zogby poll questioned voters who intended to vote in November. The NBC/WSJ poll questioned registered voters, regardless of whether they intended to vote. That difference could be responsible for the stronger tilt towards McCain.
An analysis of both polls stress the importance of Hillary Clinton supporters, who are still put-off by their candidate's primary defeat and more likely to sit out the election in protest or indifference. As the election nears and Hillary makes an appearance at the Democratic National Convention, this could swing more of the polled population from simply being registered, to registered and likely to vote. That would skew Zogby's poll back towards Obama.
Polling is an inexact science and as we've seen in the past, sometimes perceived as proof of ballet box tampering. Voters and candidates alike would be well advised to take all polling data with a grain of salt.
Susan Estrich also advises treating poll data skeptically.