The so called Bradley Effect is baloney. LS and I followed the polls religiously on a daily basis in 2008 and if I took one thing away from that election cycle it was that. Total baloney for head-in-the-sand types and since I’m sure you aren’t that, you’d do well to drop that angle. The polls were right. If there is any such thing it’s already built into the margin of error making it effectively unmeasurable.
The other difficulty this time around is that so much of the micro data seems to point to a landslide, or a very big victory for Romney. For ex, the fact that by almost every poll shows CT close and "in play", the poll from northern VA showing Obama barely up in a region he won by 23%, the agreement of every poll that the indies have swung heavily toward Romney. On the other hand, while the trend is good, Romney still trails in NV and hasn't opened up any room in OH.
So my optimistic side goes with the former data.