Normally, I respect the analytic work of Sean Trende of RealClearPolitics. Today his column argues that the polls did as good a job of predicting the election outcome as they did in 2012. I'm not entirely convinced.
Yes, it is true as he argues that most polls had Clinton ahead by a small margin and the polling margin of error was wider than that margin so a Trump win was possible. He believes the error was in the media's interpretation. I'm still not convinced.
I see a systemic error in the fact that almost all polls showed Clinton ahead by varying amounts. If it were merely random error, you'd expect roughly as many polls to err by overestimating Trump's chances as actually did err by underestimating them. This did not happen. Ergo, systemic bias.
