Has any poll oversampled Republicans by 6%?
Nope! If the D+ oversampled polls are all wrong, we are looking at a GOP landslide!
Of course it won’t happen without a big turnout operation. Let’s hope the GOP has its ground game up and running today.
"Has any poll oversampled Republicans by 6%?"
No, you can't oversample Republicans (and conservatives and libertarians) very easily! They're always more difficult to poll than Democrats (and liberals, socialists, communists, etc.) Pollsters "correct" their raw numbers to obtain the numbers they publish. Exactly how each pollster tries to do this is not something they're upfront about, but they all do it. Otherwise, they'd always predict that the Democrat would win. My guess is that they weigh the gross numbers with a formula that utilizes the last presidential election exit polling party ID results.
A "tie" usually means that the Republican is ahead by a couple of points. Gallup had Kerry-Bush tied at 49% / 49% in 2004. The election result was 51% Bush / 48% Kerry. They also predicted 53% Obama / 42% McCain. The actual election day vote was 53% / 46%. They understated the Republican vote by 4%, and that was with low Republican voter turnout!
Gallup has been doing polling since 1936 so they're as good as anyone. The partisan polling outfits and college polls don't warrant more than a skeptical eye roll. I'd look at a poll's party ID and mentally adjust the numbers under the assumption the roughly 1/3 of voters are Democrats, roughly 1/3 are Republicans, and roughly 1/3 are independents right now. As has been stated many times, if the independents are breaking significantly for Romney, it shouldn't be tied.
Yeah 1 ,when they wanted to keep Atkins in the race I think