The media have been jumping for joy since an NBC News/Wall Street Journal poll discovered a supposedly big bounce in President Obama's approval rating.
But as Hot Air's Ed Morrissey discovered, the pollsters involved used some interesting methodology to reach this conclusion:
NBC/WSJ’s pollster, Hart/McInturff, polled 1000 adults without restricting respondents to registered or likely voters. In fact, 19% of those responding claimed not to have voted in the 2008 elections, which is almost 1 in 5 and tends to make the predictive value of the poll rather weak. The D/R/I sample split is 31/21/39, a ten-point Democratic advantage when Gallup and Rasmussen both show the electorate evenly split or with Republicans taking a slight edge in affiliation. A 21% sample composition for Republicans is a gross underrepresentation. The previous poll has a 33/23/36 split.
On top of that, 42% of respondents say they voted for Obama, and 29% for John McCain, while 10% split evenly say they voted for someone else or can’t recall for which candidate they voted. Since Obama beat McCain by seven points in the popular vote, this appears to oversample Obama voters badly. Compare this sample to their previous poll, in which the split was 41/32 and 17% said they didn’t vote for anyone at all, and one can see where at least some of the bounce originates.
Yeah, I'd say including only 21 percent Republicans is a tad absurd given what just happened in November.
But the end does justify this means, doesn't it?