Wild poll shifts

There's a new Denver Post/Survey USA poll just out on both the Colorado Senate and gubernatorial races that shows dramatic shifts from previous polls.  I would tend to dismiss these results as outliers, except that we're not likely to see any other polls prior to the primary.  This is pretty much what we've got.  Take it or leave it.

Anyway, the lesson from the governor's race, if the poll is to be believed, is that the plagiarism scandal has substantially wounded Republican Scott McInnis.  McInnis was ahead of his primary challenger, Dan Maes, by 28 points in a June poll, and he's now trailing Maes by four (within the margin of error).  McInnis is also doing worse against his November opponent, John Hickenlooper.  In most of the pre-scandal matchup polls, he was leading Hickenlooper marginally -- he's now trailing by a statistically significant five points.

The polls also shows a pretty profound Tancredo effect.  In the three-way matchups, it's Hickenlooper 44, McInnis 25, Tancredo 26.  (It's basically the same result if you substitute Maes for McInnis.)  That's an enormous vote for a third party candidate, and it looks like Tancredo is almost perfectly splitting the Republican vote.  (I hate to make inferences like that from the topline, but I just can't see Tancredo wooing any Democrats.)  My expectation is that Tancredo's vote share will be considerably lower than that -- third party candidacies tend to fizzle as partisan voters return home -- but still pretty sizable.

But it's the results from the Democratic Senate contest that surprise me the most.  Bennet was beating Romanoff 53-36 in the last Survey USA poll conducted in mid-June.  Romanoff now leads 48-45 (within the margin of error).  That's an enormous amount of movement in six weeks.  And it's not just undecideds going for Romanoff -- Bennet has lost eight points.  This is all the more surprising given that, unlike in the Republican gubernatorial race, nothing that dramatic has happened.  No major scandals, no feisty debate moments, etc.  And unlike Dan Maes, who really came out of nowhere, Democratic primary voters already had some idea about who their candidates were.

So what's doing this?  The most obvious thing would be the Clinton endorsement.  That's the kind of thing that can affect Democratic primary voters, although that's still a surprisingly large effect.  It's possible that Romanoff's continued emphasis on good government issues -- campaign finance reform, not taking PAC money, etc. -- is really starting to pay off among primary voters.  It's also possible that Romanoff's distinctly negative campaign is taking a toll on Bennet, whose Rose Garden strategy isn't yielding much fruit.