The partisan sample used in the poll is 50%D, 37%R, 10% Indy (D+13) http://images.lancasteronline.com/local_old/662921/franklinandmarshallcollegepoll06062012.pdf (see pg 8 of 19).
The partisan sample also reflects the party registration of the state: http://www.portal.state.pa.us/portal/server.pt?open=514&objID=572645&mode=2
The 2008 actual turnout in PA, a great year for Democrats, was D+7.
The 2008 actual party registration in PA was 51D, 37R, but even in that huge Democratic year, turnout was only D+7. So if registration is now 50D, 37R, AND Republicans are more motivated as many suspect, a D+13 poll probably wouldn't be too accurate.
You really need to stop confusing Party registration and Party ID. They are two totally different questions. This poll asked REGISTRATION not party id which is what the exit polls asked. As a result the numbers should reflect the registration statewide and should not reflect party id. Do you grasp this simple concept yet? Or are you going to be responding with the same clap trap every time??