From a recent washingtonpost.com headline:

Poll: Many Obama 2008 supporters defecting to GOP

It's true that the accompanying AP story by Liz Sidoti made the following claim [emphasis added]:

Nearly two years after putting Obama in the White House, one-quarter of those who voted for the Democrat are defecting to the GOP or considering voting against the party in power this fall.

But as I noted yesterday, that's not quite what the AP polling data found. Instead, it found that exactly seven percent of voters who supported Obama in 2008 plan on voting Republican next month. In order to get to the much more eye-catching “one-quarter” figure, you have to take into account all the 2008 Obama voters who said they didn't know how they would vote in local House races next month: 13 percent.

But here's the kicker. Based on the same AP poll, guess what percentage of John McCain supporters from 2008 are going to “defect” and vote Democratic in November? That's right, seven percent. And guess what the number basically balloons to when you take into account how many McCain voters haven't decided how they're going to vote next month? That's right, one-quarter.

So instead of there being some sort of significant migration away from Democrats by Obama voters on the eve of the midterm elections, what the AP poll found is there's a tiny defection rate among both parties, along with a moderate number of undecideds.

Do editors at the Post want to rethink the “many” headline?