Please enable JavaScript to view the comments powered by Disqus.

Washington Matters

Multiple Choice Polls

If you want another sign that this election is simply unlike any we've seen, look at the trouble pollsters are having in figuring out the samples they need to be as accurate as possible. Cell phone only users, a huge bump in voter registration, the possibility of young and minority voters turning out in record numbers and of course the unpredictability of race -- these are all factors bedeviling pollsters and making results tougher for readers to interpret.

And one of the oldest and most respected polling outfits firms, Gallup, has really bollixed things up now by basically announcing that it couldn't figure out the single best way to determine who is likely to vote, so it is taking the New Coke and Old Coke route and letting the consumers decide which they prefer.

Old Coke (or "traditional" in Gallup's phrasing) is the formula in use for years, "which determine respondents' likelihood to vote based on how they answer questions about their current voting intention and past voting behavior." Or, in other words, this formula simply doesn't take some voters at their word that they'll actually cast a ballot and excludes them from the bloc of likely voters.

New Coke ("expanded") believes new voters who say they are committed to showing up Election Day. "This model would take into account increased voter registration this year and possibly higher turnout among groups that are traditionally less likely to vote, such as young adults and racial minorities" is how Gallup explained its new methodology.

This all means that anyone following Gallup's carefully watched daily tracking poll will have three sets of results to choose from: registered voters, likely voters as measured for eons and likely voters as adapted to the new fangled electorate that Barack Obama and his enthusiastic backers may have put into play.

Not only do these three different samples produce different outcomes, the differences are exactly the kind that pollsters have expected. Today's poll, for example,shows Obama leading John McCain 50%-43% among registered voters. But Obama's lead shrinks by more than half -- and within the poll's margin of error -- when the traditional screen for likely voters is used: 49%-46%. And the expanded sample of likely voters that includes more of the groups whom pollsters believe lean toward Obama indeed gives the Democrat his largest lead, 53%-44% -- a pattern consistent since Gallup announced the change Monday.

The decision to publish both likely voter results has caused a stir -- and vigorous arguments -- among pollsters and interested onlookers. It's causing some problems, too. Political websites that use various formulas to determine trends from all major polls have to decide which method they believe is most accurate. Or not.  RealClearPolitics decided to split the difference by using both Gallup numbers, weighing each at 50% when factoring them in with other polls. is going with New Coke. Why? Largely because Gallup's traditional approach includes significantly fewer 18-to-29-year-olds "than both the likely voter models of other pollsters and available estimates of the 2004 electorate. While no one can be certain about who will vote, the least likely outcome is a 2008 electorate that is older than those who voted in 2004," explains the director of the site, pollster Mark Blumenthal.