One of the better articles I've read on why US election forecasts are inaccurate. And written by a couple of guys from Brazil. While beyond the scope of this a

One of the better articles I've read on why US election forecasts are inaccurate. And written by a couple of guys from Brazil. While beyond the scope of this article, it would be great see more research done into the relative strengths of each bias, i.e. which biases are skewing polls the most, rather than just using lumped bias terms. Though I did see after Brexit forecasts got it wrong too the British Polling Council is "looking into it". What's most interesting is there wasn't more of an effort by the media to get accurate vote forecasts before the election. Granted, it's more expensive and takes longer but it's not terribly difficult to do. It does involve a research team and statisticians spending time in person in the state and getting deeper data from people. But the time/costs aren't as outrageous as one would think because it only needs to be done in the handful of close swing states. And a sample size of just a few hundred people is enough to get significant results. Seems obvious and worthwhile, maybe in 2020 they figure it out.