There are some substantial issues with the poll released in Wisconsin this week by Marquette Law School's Charles Franklin. For example, the polling sample he used in the May poll is dramatically different than the polling sample he used for the April poll. The most glaring issue is the age demographics which are completely different from what he polled in April. I thought it would be interesting to compare the Public Policy Polling Wisconsin poll that came out the day before with the Marquette poll and also compare the Marquette poll with voting trends in the state. This is cross posted from dane101. More below the fold.
The second poll of the week was from Public Policy Polling on behalf of the DailyKos. PPP has the unique distinction of being the most accurate pollster during the 2011 recall election. PPP has a five point spread between Walker and Barrett at 50-45 with a margin of error of plus or minus 3.4 percent. What makes the PPP poll unique among the three polls released this week is they also polled for independent Gubernatorial candidate Hari Trivedi who managed to take 2 percent in the poll.
According to DailyKos political director David Nir that 2 percent is coming at a cost to Barrett, "their second choice is overwhelmingly Barrett, suggesting that Trivedi is pulling votes away from the Democrat. In a tight race, that could be a difference-maker."
PPP also polled the Lieutenant Governor race which puts incumbent Rebecca Kleefisch statistically neck-and-neck with Professional Fire Fighters of Wisconsin President Mahlon Mitchell 43 to 46 with 14 percent undecided. Kleefisch and Mitchell have the same problem, a large number of Wisconsinites don't know enough about them.
Wisconsin's Lt. Governor is typically a package deal with the Governor, so Wisconsinites don't typically need to pay attention to who's on that line in a general election. Twenty percent say they don't know enough about Kleefisch while 43 percent say they don't know enough about Mitchell.
Marquette released the most recent poll showing a 6 point spread between Walker and Barrett with a margin of error plus or minus 4.1 percent. That's a significant swing from Marquette's April poll when Barrett was leading Walker by one percentage point. Marquette didn't poll for potential Tridevi voters.
In the race for Lt. Governor Marquette puts the distance between Kleefish and Mitchell with likely voters at 47 to 41.
There are some small differences between those polled by Marquette in May as compared to the previous polling by the organization from April. Forty-five percent of those polled were over the age of 60 in comparison to 29 percent in the previous poll. The number of individuals polled between the ages of 18 to 29 was cut in half, dropping from 16 percent in April to 8 percent in this most recent poll. Overall, those 45 and up accounted for 78 percent of May's polling data while 44 and under accounted for 21 percent. In April it was 59 to 39.
The age demographics of the May poll are substantially out of whack with voting trends in Wisconsin. Averaging out the Gubernatorial elections of 2006 and 2010 with the Presidential election of 2008 both the 18-29 age group and the 65 plus age group average a 17 percent turnout. The averaging of those three elections based on the turn out for 45 and up is 54 percent with the 44 and under average a turn out of 43 percent.
If one wants to remove 2008 because it was considered a larger turnout than previous years for young voters and only count the last three Gubernatorial elections (2002, 2006, 2010) the average for the 18 to 29 demographic is 16.3 percent.
PPP's poll sample was more in-line with traditional voting patterns, however, the 18 to 29 demographic only made up 12 percent of their voting pool and the 65 plus group was 21 percent, four percent higher than the average.
Other Marquette polling demographic changes in May over April is the new poll being six percent whiter and five percent more conservative. It also polled five percent fewer union households.
Of those polled by PPP, 98 percent said they were likely voters while Marquette's poll was made up of 91 percent of likely voters.
While both were nearly identical when it came to liberal political leanings Marquette was slightly more conservative and slightly less moderate. Marquette poll was 19 percent liberals (combining liberal and very liberal), 47 percent conservative (combining conservative and very conservative, and 30 percent moderate. PPP was 17 percent liberal, 41 percent conservative, and 41 percent moderate.
Finally, what does all of this polling data matter beyond being wonk wank fodder? Both Marquette and PPP were way off when it came to predicting the results of the Democratic primary with both Barrett and Falk out performing the polls. PPP had Barrett 38, Falk 24, La Follette 9, and Vinehout 6. Marquette had 38, 21, 8, 6. The actual results were Barrett 58, Falk 34, Vinehout 4, and La Follette 3.
Overall 5.7 percent more votes were cast for Democrats and progressive Republican candidate Arthur Kohl-Riggs than votes were cast for Walker and fake candidate Huber.
As has been repeated often, this election is unprecedented in Wisconsin history and it's being waged by a divided state with both sides equally passionate about their candidate. One thing is clear, election night could end any number of ways and the pollster that ends up the closest will be able to claim the pole position for the 2012 November Election.