Bouncing Polls (posted September 27, 2000)
Al Gore is ahead! No, Geore W. Bush has come back! With national polls bouncing all over the place in the presidential election, it is important to understand why polls taken at the same time produce different results.
The biggest challenge facing the pollster is voter turnout. The election obviously depends on those who actually turn out (estimated to be around half the electorate in November), but pollsters have great difficulty figuring how who is going to vote. Some polls include the views of registered voters while others are based on likely voters. Even those that emphasize likely voters have different techniques for gauging the likelihood of voting. Candidate preferences based on likely voters are the most accurate.
There also are subtle differences in question wording that can affect the results. Some questions pit Gore versus Bush, Nader, and Buchanan, while others include the VP choices (meaning the question is do you prefer Gore and Lieberman or Bush and Cheney). Some ask "if the election were held today, whom would you prefer" while others say "in the upcoming election in November, for whom would you vote".
Finally, it is important to pay attention to the margin of error. Most national polls have margins of error of anywhere from plus or minus three to five percentage points. If Gore drops two points, it still is within the margin of error of the first survey. With a race that is expected to stay close and competitive throughout the campaign, don't look for polls to be able to resolve the issue of who is going to win the election.
