The lesson is that the RCP average was more accurate than any individual pollster. And even though the RCP average, on the final day, was giving us bad news (this time around), we need to look at things as they are, not as we wish them to be.
I was on board with the whole “Dems were oversampled” mantra, but, in the end, the skewed polls approach of dissembling samples was way off. Sometimes one has to go where the numbers take one. In two years I expect major Republican gains and I suspect the polls leading into the elections will show that. And in four years I expect a GOP win for the presidency and I suspect the polls will show that as well. So we can sit back, smile, and watch the lefty sites try to explain away the polling numbers.
If I had one guess (sans numbers) about what happened, it was that the Obama turnout machine was remarkably effective given the drop in enthusiasm for his presidency. That allowed for a higher Dem turnout than one would have expected given all the exogenous indicators.
Rasmussen has a pretty good track record, but minor methodological issues can skew a series of polls one way, so hopefully he’ll examine his methods and revise them.
Look, even Peggy Noonan got it last night - as Dan Gainor just tweeted, "Early voting gave Dems weeks to ship low info voters to polls."
These are the people that would formerly slip through the cracks when elections were held on Election Day.
There are only two ways to combat that -- severely restrict early voting, or switch to a national database where it can be instantly ascertained that you only voted once. I prefer the former.