Not my analysis. It was courtesy of Freeper Dr. Free Market. However, if anything, the subsets would favor MORE conservatives and fewer moderates due to the fact that moderates don't turn out for mid-terms.
Right, LS. I am using exit poll data fromm CNN over time, assuming that Reps and Dems are equally motivated-- benefits Dems-- but allowing for the median historical decrease in midterm turnout in Ohio (~15), then defining likely voters as those that consider the economy as "very important" since that most correlates with actual voters across parties in gubernatorial elections. The 20% that said that the economy wasn't important had unusually high undecideds and young, which are the least likely to vote in a midterm. As I said, the -6.2% is only "backback of the envelope", but it has followed Zogby closely and trended with Rassmussen. So take it for what it's worth.