I have gathered the 13 Senate races identified by either RealClearPolitics or Rasmussen Reports tonight as tossups. I then recorded the current Rasmussen poll numbers, along with the remaining Undecided % (which is, of course, partially 3rd-party %). Finally, I hypothesized what the races would look like if there’s 1%, 2%, 3%, and 4% liberal bias in Rasmussen’s current poll numbers. Take a look:
As you can see, it would only take 1.5 – 2% liberal bias in the current polls for the actual election results on Nov 6th to turn out favorably for Republican majority control of the U.S. Senate. The seats that would hypothetically swing to Republican hands if there’s a 2% bias would be Montana, Ohio, Pennsylvania, and Virginia. And all of these states theoretically have enough Undecideds to enable such a 1.5 – 2% shift between the current polls and the election results on Nov 6th.
Of course, Rasmussen Reports is not known as a liberal-biased outfit. But even if there’s no “liberal bias” per se in their polling numbers, there’s still an interesting reason why they could be inaccurate. Take a look at this 4-minute interview of Michael Barone from November 1st, entitled “Could ‘Hidden Vote’ Lead to Landslide Romney Win?”