The only thing worse than newspapers failing to report on other polls is the way they (mis)interpret their own findings.
The Los Angeles Times is currently running a story titled “Obama slightly widens lead after debate, poll finds” (via Yglesias). The basis for this claim is a 1% increase in Obama’s support and a 2% increase in Obama’s lead among voters in their poll who watched the debate (presumably a non-random group):
The much-anticipated first presidential debate between John McCain and Barack Obama on Friday appears to have helped Obama slightly widen a lead over his Republican opponent, a post-debate Los Angeles Times/Bloomberg survey shows.
Registered voters who watched the debate preferred Obama, 49% to 44%, according to the poll taken over three days after the showdown in Oxford, Miss.
That is a small gain from a week ago, when a survey of the same voters showed the Democratic candidate with a 48% to 45% edge.
An increase of 1% in Obama’s poll numbers (or a 2% increase in his lead among the group that watched the debate) is meaningless. There are two problems. First, we can’t be sure that the increase among the group who watched the debate is meaningful. Any time you survey people twice on their vote preferences, there may be minor fluctuations. A difference of 1-2% is almost surely not statistically significant. Second, we can’t extrapolate anything meaningful from that group’s change to the larger population of registered voters as a whole. The reason is that the voters who didn’t watch the debate may also change their preferences as a result of media coverage, external events, etc.
Who allowed this nonsense to be published in a major newspaper? It’s the statistical equivalent of 2+2=5.
Update 9/29 8:41 AM: The description of the problems with the article above has been improved and corrected with help from Joel Wiles.
Update 9/29 9:10 PM: Per Jonny Drake’s comment below, the wording of the post has been clarified to distinguish between a 1% increase in Obama’s support and a 2% increase in his lead among voters in the Times poll who watched the debate.