I wouldn't be surprised that they got NY, CA, TX, etc. right as to outcome, but if (just for example) they were consistently over-estimating Dem turnout by, say, 3%, that would be a more interesting data point. I know I never answer phone polls, and nearly every person who votes like me has said the same. I sometimes wonder how Trump does as well as he does in the polls, given that I don't know a single person who answers polls in his favor.