I've had several people ask me why I think we and most pollsters significantly underestimated Barack Obama in Nevada.
It's actually a question with a pretty easy answer. We had whites at 70% of the electorate and McCain winning them by 10. The exit poll had them at 69% of the electorate and McCain winning them by 8. Not much of a disparity there. We had blacks at 8% of the electorate and Obama winning them by 90. The exit poll had blacks at 10% of the electorate and Obama winning them by 89. So Obama picked up a couple points there relative to what we had projected.
The real difference was with Hispanic voters though. We pegged their proportion of the electorate basically correct- 16%, as opposed to 15% in the exit poll. But we had Obama winning them only 61-36 when he actually won them 76-22. That caused us to underestimate him by about five points right there.
I'd say Hispanic voters are harder to poll, especially if you only poll in English, but we did a better job than most of projecting how dominant Obama's victories would be in the similar states of Colorado and New Mexico. So I'm really not sure what caused the problem. If anyone has any useful tips for the future in Nevada I'm open to hearing them.