In the 2008 election, at least at pollster, the threshold for the polling average to have a chance to be wrong about the election was 1.1 to 1.2%.
Even below 1.1%, they held up pretty well. Although pollster categorised a number of races as swing states, within the swing state category, only Missouri and Indiana moved differently than anticipated by the polling averages. Nate Silver got one state wrong. Pollster got two (one to blue, the other to red - so no obvious bias in the numbers).
This time, I see Ohio, Nevada, Iowa, Wisconsin, Pennsylvania and New Hampshire all above 2% pro-o at pollster. And North Carolina pro-r. At that kind of number, the polls are very accurate in predicting outcomes.
So barring last minute changes, the states in play are Colorado, Virginia and Florida.
| 