http://www.fivethirtyeight.com/2008/10/bradley-effect-or-elephant-effect.html

Silver echoes many of your points, but really takes Greener to task on his polling analysis.

And, Baysian is correct, Silver has found that even the former Bradley campaign says that they knew (via internal polling) they were running 50/50 in that campaign and were worried that media polls showing them with big leads would lead to complacency among their supporters.

The “Bradley effect is merely the “soccer mom” meme of this election. It allows us to speculate on how racist the swing voters are without having to really come out and say it. There’s always some tortured meme intended to encapsulate the swing voter’s mentality.

Palin has vitality, like her or not, but McCain is simply sad.

I would never speak for Daniel, he who knows an astonishing amount, but:

1) Nate Silver (fivethirtyeight.com), who knows an awful lot about polling, makes a compelling case that the Bradley effect is long gone.

2) Why only subtract six points in the tossups, rather than everywhere? At any rate, a six point subtraction across the board when applied to e.g. the latest RCP averages still gives Obama a fairly clear victory – all the Kerry states plus IA, NM, CO, VA, OH (barely – RCP has it Obama 6.1%).

then add in undecideds

If support strips evenly across the country, it looks like McCain has to get at least to the point where he gets VA back. I’m not sure there are enough “Real Virginians” to do that.