Quinnipiac (7/14-19, registered voters, 5/20-26 in parens):
Arlen Specter (D-inc): 45 (46)
Pat Toomey (R): 44 (37)
Undecided: 14 (10)Joe Sestak (D): 35 (37)
Pat Toomey (R): 39 (35)
Undecided: 30
(MoE: ±2.8%)
Primary:
Arlen Specter (D-inc): 55 (50)
Joe Sestak (D): 23 (21)
Undecided: 19 (27)
(MoE: ±4.3%)
Snarlin’ Arlen finds himself on slightly more comfortable footing in the primary, but the general election is a different story. I can’t tell if the sample’s been skewed a bit in the GOP’s favor or not — Quinnipiac did poll an almost equal amount of Dems as they did Republicans this time, which is not something they did in May (and not something that reflects the Democratic registration advantage in PA), but it’s also possible that these are oversamples made for the purposes of getting a clearer primary picture on both sides. It’s hard for me to tell without being able to pop open the hood. It’d be surprising to me if Quinnipiac fell victim to a mistake that any Polling 101 student would spot, but stranger things have happened.
RaceTracker Wiki: PA-Sen
There polls always seem a bit off to me. If Pennsylvania got rid of Santorum, there is no way theyd ever elect Toomey.
They oversampled in order to get better numbers for the primaries. Nate Silver ran the numbers with the 2008 exit poll numbers and found a 46-43 lead for Specter, so the party ID that Quinn used had to be somewhere around those splits.
so that Toomey can act crazy. Without a primary Toomey can act moderate for the whole campaign.