The more I think about this, the less it makes sense. I mean, I've never worked in a campaign, but my surmise is that one of the things campaigns do with polls is decide where the candidate needs to go so that they can jack up turnout as is necessary to win. If the Romney campaign was consistently overestimating turnout to begin with, then how could they make such decisions reliably? Sure there are reasonable precedents, but the electorate changes all the time too for all kinds of reasons, so you still have to rely on polling, which means you have to find out where the best estimates are at and trust them. I just find the story Crawford was fed here to be very suspicious.
Paul Ryan was sent to Minnesota for no good reason. They ignored Nevada and New Mexico despite similar polls in MN because it was more obvious there that there wasn't enough whites to win. It's obvious they skewed the white samples in the states to 2004 levels and as a result PA and the Midwest looked like they were going to vote for Romney.
It's clear their strategy thought it was 2004 electorate. Joke campaign.