Monday, October 22, 2012

More Polling Nonsense in Ohio

See, this is what I'm talking about. Today a CBS News/Quinnipiac poll was released for Ohio showing Obama up 50-45. Bad news for Romney, right? Hmm, maybe not. Here's the sample they used:

Republican      26%
Democrat        35%
Independent    34%
Other/DK         5%

CBS outsampled Democrats by 9%. In 2008, Democrats had a 7% turnout advantage in Ohio. Is there anyone on the planet that can make an argument that 2012 turnout will be better for Democrats than 2008?

In 2004, Republicans and Democrats were evenly split. In 2010 it was the same. As I argued previously, you could make the case that this year might be something like D+2. So, what happens to the CBS survey if you adjust it for this assumption? Here's what you get*:

Romney       48%
Obama         47%
Other/DK      5%

With Romney's huge lead among independents, I would argue he might get the lion's share of the "Other/DK" also.

This is some serious polling malpractice, something CBS/Quinnipiac has been guilty of this whole cycle. Here's a prediction: look for their samples to get much more reasonable at the very last minute, because the last poll is the one you're judged by.

A second poll came out today by PPP, a Democrat polling firm, that shows Obama leading Ohio 49-48. Seems reasonable, right? Sure, until you get to the fine print. PPP used a D+8 sample. I'm not the only one squawking about this stuff, and it's easy to dismiss as partisan whining until you lift up the hood and do the analysis for yourself. These pollsters have a lot of 'splainin to do.

*Assumptions: all Republicans vote for Romney, Dems for Obama, independents split +8.5 for Romney, consistent with recent national polling.

1 comment:

  1. Follow up to this post: a Suffolk poll just came across that outsamples Dems by 4%. Has Ohio tied. This underscores my hypothesis that Romney wins at D+2.