This has me more worried about PPP's reliability than enthusiastic about the 2012 race. What is the explanation for Feingold's and Johnson's approvals being this way so soon after an election where Johnson won by 5 points? Can a changing electorate make that much of a difference?
Likely Voters vs. Registered Voters makes the difference.
Yup, keep in mind those gaps were pretty large in the RV & LV polls in Wisconsin before Election Day. The question basically is will the Democrats that turned out in 2008 but didn't turn out in 2010, turn out in 2012?
PPP uses some kind of weird soft-likely voter model until they switch to a "real" likely voter model around September or so. It's not a pure registered voter model.
I don't see why turnout won't be more like 2008 than 2010; turnout was about the same in 2008 and in 2004, as far as raw numbers go.