The Mills campaign released a poll this morning showing (surprise!) their guy ahead.

The poll is by a new public opinion research firm called the Maine Center for Public Opinion and has a high sample size of 915 respondents.

Here are the results: Mills 22.1%, Otten 17%, Abbott 13.9%, LePage 10.7%, Beardsley 5.9%, Poliquin 4.9%, Jacobson 1.4%, unsure 24%.

In addition to the normal limitations of primary polls, polls paid for by campaigns and polls with large survey windows (this one is more than a week), there are a couple of decisions made by the pollsters here that may have affected the results:

First, they used a very tight voter screen - only registered Republicans with a strong voting history dating back two or three years were called. (Contrast this with Pan Atlantic's approach of apparently allowing respondents to self-identify as "likely" voters.) This screen may have missed some voters who are newly registered or newly energized to vote in this particularly primary.

Second, the question about candidate preference comes at the end of the poll, after a series of questions about Governor Baldacci, the economy, school district spending and the Tea Party. Any of these questions could influence how people think about the governor's race. Most pollsters looking for a clean response on the horse race question ask it first. For instance, people who have just been asked to consider whether or not the Tea Party is hurting the GOP might be more likely to choose a moderate like Mills.

Other parts of the process, from the question wording to the pre-weighting they did by geographic area to how their "virtual phone bank" works may also have affected the results.

The press release from the campaign implies that the firm has done other polls for them, including some earlier ones showing Mills behind. I'd be interested in seeing if their methodology has changed at all from poll to poll.

Here's the full memo: