The state-by-state margins seem off; Mississippi statistically almost cannot be >60% Romney for the sole reason of the black vote. Alabama being 64% Romney is also highly implausible. In fact, his model seems to be of all states having roughly the same % swing, factoring into my theory that hunches can often be better than statistical models (you know what 14 year-old got only one Senate race wrong when Nate got 3 wrong? This guy!).
Nate, I recall, only got 2 wrong (Colorado & Nevada)...what was the third?
He had Joe Miller beating Murkowski in Alaska.
