Nope. As a matter of fact, I ignored the last poll (due to the fact that Gallup tries to allocate undecideds in their last one). I looked at it from both a month out and the second to last poll for that writeup, and again later (not written) for other timeframes such as two weeks.
What I found is that the undecideds at any of those periods either 1) break for the incumbent in a substantial way, 2) break evenly, or 3) break ever so slightly towards the challenger-- and the times that #3 was the case happened to be in races that were already blowouts.
It wasn't a blowout in 1980 two weeks out was it?