Thought I would post this in advance of what's sure to be another crap poll from Gallup today.
--------
Difference in method may explain polls' divergencesnip
Every pollster strives to weed out people who have opinions but who may not bother to vote. Some pollsters do so differently and more rigorously than others, one possible reason for conflicting findings.
Peter Hart is the Democratic pollster in a bipartisan team that found a 48-48 tie among likely voters in the presidential race in a poll published last week by the Wall Street Journal and NBC News. Hart says that he's confident in his method of simply asking people how interested they are in the election, as compared with Gallup's more complicated method of asking seven questions.
"The difference is that it (the Gallup Organization) relies on ideas that George Gallup had back in the 1950s. I think the Gallup Poll is having a harder time this year because they are eliminating a lot of people," he said.
Frank Newport, Gallup's editor in chief, disputed what Hart had to say. He said his organization uses a time-tested series of seven questions to gauge voters' interest. Then, when computing the survey's results, Gallup eliminates those whose responses did not suggest a strong likelihood of voting according to a formula that factors in the projected turnout.
http://www.stltoday.com/stltoday/news/stories.nsf/election2004/story/E448B3850CF0AF9986256F3700456B48?OpenDocument&Headline=Difference+in+method+may+explain+polls'+divergence