I got quite a vociferous e-mail today saying the only reason our Colorado polling was showing Michael Bennet and Bill Ritter in such bad shape was that we had our party weighting all wrong.
We don't weight for party- we weight for fixed demographics like gender, race, and age and let the party ID fall where it may- because unlike those other things people can change their party.
It was actually pretty ironic to get an e-mail from a Democrat about this because last year every time we put out a Colorado poll it was the Republicans complaining about the very same thing.
Given that both the 2008 exit poll and the state's registration statistics show a 1 point advantage for Republicans, the 3 point advantage for them we found in this poll seems perfectly reasonable in this political climate.
Anyway this person asked me to recalculate our results using the state's registration breakdown of 35% Republicans, 34% Democrats, and 31% other.
Here's what you get using those numbers. Bennet's approval rating is 33/35 rather than 31/38 and he trails Bob Beauprez 41-39 instead of 42-39. Ritter's approval rating is 41/45 rather than 40/45 and he trails Scott McInnis 45-38 instead of 46-38.
In other words changing the party id breakdown helps the Democrats, but only by a very small amount. Bennet and Ritter aren't doing poorly in our poll because of the party breakdown, they're doing poorly because a) Republicans disapprove of them more strongly than Democrats approve of them and b) Republicans are more committed at this very early stage to voting for Beauprez and McInnis than Democrats are to supporting Bennet and Ritter.