Monday, February 06, 2006

Recent Polling Roundup

I'd been too busy this weekend with the caucuses to post about it, but on Friday, the January State House News Poll was released. I've been following these polls pretty closely since last year, but this was the first one released since Governor Mitt Romney decided not to run for reelection. Not only that, but Bill Galvin finally made up his mind to try for another term as Secretary of the Commonwealth. That means all those fancy charts I had been using these past few months are now worthless. Look for more time-series results next month.

Here are the head-to-head poll results from the State House poll for the general election, with the November results in parentheses.

If the candidates in next year's general election for governor were Kerry Healey running as a Republican and [CHALLENGER] running as Democrat, and the election were being held tomorrow, for whom would you vote?
ChallengerHealeyDon't KnowNeither
Deval Patrick38.9%
(43.5%)
35.7%
(31.8%)
12.7%
(13.6%)
11.0%
(10.2%)
Tom Reilly51.6%
(56.0%)
32.0%
(25.0%)
7.8%
(7.4%)
6.9%
(10.8%)
Both Tom Reilly and Deval Patrick do slightly worse against Kerry Healey than in November, but not enough to ascribe to any factor. One thing to note is that the margin between Reilly's numbers and Patrick's numbers has stayed constant (about 12 points) since they started asking this question. The poll did not ask about all of the various scenarios where Christy Mihos would be a candidate in the general election, although it did find him losing to Kerry Healey 62.2% to 12.4% (+/- 11.4%) in the Republican primary and that he would lose in three-way races with Lieutenant Governor Kerry Healey and either Tom Reilly or Deval Patrick.

Now the State House News did not ask about the Democratic Primary in their November poll, but this time they did. It's not particularly comparable to those previous questions, anyway, since those results almost always included other challengers, Bill Galvin, Mike Capuano and Chris Gabrieli, in particular. Back in May, they measured a head-to-head matchup between Tom Reilly and Deval Patrick at 53.5% Reilly, 11% Patrick. Here are the State House News Poll results from January:

Deval Patrick17.8%
Tom Reilly57.8%
Other3.9%
Don't Know19.4%

Not that much movement over the course of eight months, considering the high margin of error in the subsample (7.2%). That can't be what the Deval Patrick campaign wanted to see. Of course, this poll came out before Tom Reilly was widely criticized for tapping Rep. Marie St. Fleur as his running mate, twenty-four hours before she had to withdraw after her financial difficulties, including nonpayment of taxes, were disclosed. Luckily for us, Suffolk University released a poll of their own today. You can find links to the internals from their website.

Deval Patrick30%
Tom Reilly39%
Other2%
Undecided27%

Why the discrepancy? First of all, as I mentioned above, the State House News poll was taken before the short-lived Lieutenant Governor candidacy of Marie St. Fleur. Tom Reilly got hit pretty bad in the papers last week, and any polls taken before that likely got stale pretty quickly. Not only that, but any polls taken during that time would likely show good news for his challenger. It remains to be seen whether this bump is temporary for Deval Patrick, or the start of something bigger. Another thing that gets overlooked, though, is that the subsample size in the Suffolk poll is better than that of the State House News. Suffolk University screened its respondents so that every one of them was a registered voter, while the other poll did not. That means that the margin of error in the Democratic primary subsample is a full point lower in the Suffolk poll than the State House News poll. This doesn't necessarily mean that the Suffolk poll is more accurate, but generally speaking when you're doing a poll the more responses you get, the better (up to a point of diminishing returns, obviously).

Digging in the marginals of that Suffolk poll, I found something I thought was interesting. Among people who know both Tom Reilly and Deval Patrick, Patrick leads Reilly 37% to 34%. Sure, the margin of error with that size a subsample is 8.2%, but it will be interesting to see how those numbers change as Patrick's name recognition increases. The poll also asks how people's votes would be effected if various well-known sports figures were added to the ticket, including Tom Brady, Curt Schilling, Doug Flutie and Nancy Kerrigan. Those are presumably for measuring the effect of name-recognition, but they're useful for entertainment purposes.