Dueling Minnesota Polls

I'm hoping to get a full senate polling update out later today, but in the meantime, I've been getting a lot of e-mails about the conflicting results in Minnesota, where SurveyUSA shows Norm Coleman 10 points ahead of Al Franken, but the Minnesota Star Tribune has Franken 9 points ahead of Coleman. (The one thing the polls agree on? Third-party candidate Dean Barkley is a factor).

While SurveyUSA provides plenty of crosstabular detail, there's very little from the Star-Tribune in the way of demographics, making comparison of the polls difficult. The Star-Tribune does, however, provide their party ID split (as does SurveyUSA):

Poll           D     R     I    D +/-
Star-Tribune   42    26    27   +16
SurveyUSA      37    30    21   +7

The Democrats hold a 16-point edge in partisan ID in the Star-Tribune poll, as compared with a 7-point edge in the SurveyUSA poll. In 2004, Democrats had a 3-point party ID edge in Minnesota. Since then, however, Democrats have gained about 8 points in party ID nationwide. If the trend can be extrapolated linearly to Minnesota, that would imply that their edge in the state should be about a +11 -- roughly in between the two surveys.

There's something else weird going on, too. Look at that table above and add the three numbers together. In the Star-Tribune's poll, they add up to 95 percent, which means that 5 percent of voters are "missing". And in the SurveyUSA poll, they add up to 88 percent, which means that 12 percent of voters are "missing".

Those "missing" voters aren't actually missing, of course; they probably identify themselves as part of some sort of third party, or refused to provide their party ID. There may also be some ambiguities related to Minnesotans identifying themselves as part of the Democratic-Farmer-Labor party as opposed to straight Democrats.

So who's right? I'm not sure that I'd trust either of these polls. The Minnesota senate race, most likely, is a toss-up.