Granted, you can't put too much stock in any one poll. But the dems internal polling not withstanding, Walker has held a consistent 5 to 7 pt. lead in most recent polls higher in a few others. But the Appleton Post Crescent doesn't want you to be fooled by that:

Polling puzzles

The latest Marquette University poll, released Wednesday, showed Walker with a 52 percent to 45 percent advantage over Barrett.

That means a slam dunk for the governor in Tuesday’s vote, right? Not so fast.

The poll of 600 likely voters appears to show a tremendous lead for Walker, but with a margin of error of plus or minus 4.1 percentage points, the lead isn’t statistically significant. Put another way, that same poll could reflect a 49 percent to 48 percent Barrett lead based on the margin of error.

Barrett’s campaign blasted the poll, saying it is an outlier that undersamples younger voters and oversamples heavily Republican counties.

That might not be just campaign posturing. Polls in advance of the 2008 Democratic presidential primary in New Hampshire, for instance, vastly underestimated support for Hillary Clinton, who won the state. The American Association for Public Opinion Research issued a report following the primary that pointed out how quickly voters’ opinions can change and the failure of polls to accurately sample all voters. (See report: bit.ly/NHPollFail)

Unmentioned by Barrett, but relevant to the polling association’s point about quickly changing voter attitudes, are the dates the poll was conducted: May 23-26. The candidates debated for the first time Friday, three days into the polling period, so any effect of the candidates’ performances is not effectively measured.

This might work, if they throw in clicking their ruby slippers three times as they recite this.