I take half of that 6+ over sampling for Democrats to bring it down to 3+ and then cut that in half for margin of error.
What to do with that other 1/2+ ?
So actually Romney is 50 1/4% ? to Obama's 42 3/4%.
That's a 6 1/2 point margin for Romney in PA.
I'm afraid that's what's wrong with this study. In terms of registration in PA, the numbers as of right now are: Republican 3,126,257; Democrat 4,235,483; Non-affiliated 608,963; Other Party 489,764; So, in terms of the typical breakdown that's D/R/I = 50/37/13. +6 Dem is simply
not realistic, and cutting it in half goes the wrong way, sad to say. Considered "active" as of 10/15/2012, our numbers are slightly better: D/R/I=50/38/12.
Based on 2008, which is the model most pollsters have been using, I don't think any model is realistic that has anything less than D+8 or even D+10.