There was a diary earlier today - which seemed to get a strong (and justified in my opinion) "concern troll" backlash - suggesting that internal polling from the Obama campaign was far different from the public polling, and that numerous states where the public polls show large leads are actually very close according to the internals. Additionally, the McCain campaign has specifically indicated that Iowa is tight in its internals in spite of a steady stream of public polling to the contrary.
In light of this, I have tried to do a simple test to show how accurate the public polling was in the 2004 election. Bottom line - it was pretty damned accurate according to my admittedly rudimentary test. All I did was take the 10 closest swing states in 2004 and compared the Real Clear Politics public polling average to the actual result. So here goes:
STATE RCP Average Result Error
Wisconsin Bush 0.9 Kerry 0.4 1.5
New Mexico Bush 1.4 Bush 0.7 0.7
Iowa Bush 0.3 Bush 0.7 0.4
New Hampshire Kerry 1.0 Kerry 1.3 0.3
Ohio Bush 2.1 Bush 2.1 0.0
Pennsylvania Kerry 0.9 Kerry 2.5 1.6
Nevada Bush 6.3 Bush 2.6 3.7
Michigan Kerry 3.5 Kerry 3.4 0.1
Minnesota Kerry 3.2 Kerry 3.5 0.3
Oregon Kerry 4.8 Kerry 4.2 0.6
The average error was 0.92%, so the RCP average predicted the outcome of the 2004 election in these states within one percent. It also got the winner right in 9 of these 10 states. Based on this rudimentary test, it is hard for me to buy into this idea that the public polling is way off from the internals and that Obama is really facing a tight contest in say Iowa, Pennsylvania, Minnesota, Wisconsin, and New Hampshire. I think the truth is that McCain is still fighting for these states because he has nowhere else to go to reach 270, and not because of tight internals. Saying you have a super-secret internal poll showing Iowa close is a way to save face instead of conceding the harsh reality that you are reduced to pursuing a nearly hopeless opportunity to win the election.
SAN DIMAS HIGH SCHOOL FOOTBALL RULES!!!!