That CNN poll showing Obama way ahead seems suspect. Here is the .pdf of the poll results with all the breakdowns.
The poll’s big news, reported yesterday, was that Obama was coming out ahead 52% to 43% among registered voters, and 53% to 41% among all voters, in response to this question:
Suppose that the presidential election were being held today and you had to choose between Barack Obama as the Democratic Party’s candidate, and Mitt Romney as the Republican Party’s candidate. Who would you be more likely to vote for — Barack Obama, the Democrat, or Mitt Romney, the Republican? (IF UNSURE:) As of today, who do you lean more toward?
But look at the numerical breakdown:
484 to 476, out of 1015 respondents.
484 out of 1015 is about 48%, and 476 out of 1015 is about 47%. (It can’t be registered voters because they say they have only 910, and 484+476 is more than 910).
This means the numerical breakdown, before weighting, is 48/47. Somehow their weighting process takes it to 53/41.
It’s not explained why the weighting is so dramatic. In fact, I can’t find any explanation of their weighting process.
This was pointed out to me by Kevin M, who notes that the poll is an outlier anyway. I see that Hot Air is performing the same analysis with the same results. Ed Morrissey says:
Pollsters weigh results in likely-voter models; in fact, that’s a vital part of the likely-voter modeling process. I’m not aware of the need to weight gen-pop polls unless the demos in the survey are wildly deficient, and since CNN didn’t disclose those, we don’t know — but it hardly imbues these results with confidence.
At the very least, CNN needs to answer some questions about these odd results.