hi dig,
there's also the robocall versus interview discrepancy. rasmussen favours landline robocalls and corrects for likely voters. many others also call cell phones and conduct interviews. the latter is actually considered more reliable.
i believe there is also a 1 to 1 1/2% swing to republicans over turnout. but silver corrects for this in his model. his most likely outcome is a 332 count for Obama. same as mine, which i am deriving largely from watching the poll of polls on pollster - while ceding north carolina to the republicans.
for myself, i watch the broad river, not individual polls so much. there's a mean value created by linear regression. and then there are the deviations described by individual pollsters.
but the direction in the swing states is pretty clear. the broad river is rolling in a bluer direction.
i don't think the republican critique is serious. i think it is simple denial. or deliberate obfuscation. they are terrified of confirmation bias. once a campaign is seen to be losing, folks move to support the winning candidate. this looks like a last ditch defence and tells you they know they are in serious doodoo.
| 