I still don’t understand how virtually everyone (except Trafalgar) botched this race so badly. Even the RGA’s internals apparently never showed a particularly close race.
And it wasn’t just pollsters. Cook always had the race at Safe D.
Yes, we’re beginning to see a pattern in which polls are underestimating Republicans even in states in which they have historically been very accurate (NJ) or have tended to underestimate Democrats (NV, AZ, TX). This is quite a remarkable development, and I don’t think it’s ever been as uniform or pronounced as today. If (college) education and (to a lesser extent) income become better determinants of electoral outcomes than race or even geography and the GOP keeps running up margins among working-class, non-college-educated voters (while winning over more non-white voters of those same groups), I don’t expect this to improve any time soon. There are probably other factors as well which contribute to this — I’m by no means an expert on polling, but I suspect the Democratic base having become more politically engaged than during any decade I can remember has something to do with it. It’s also never been more 'socially accepted' to be a Democrat/liberal than today (see, e.g., the increasing equation of science/reason with liberalism), and it seems like pollsters are having serious trouble reaching less 'traditional' Republican voters (Ann Selzer has addressed this quite extensively). It wasn’t that long ago when the 'landline bias' was working in the Republicans' favor, but I think this has changed dramatically.
A big part of the issue is that public pollsters have started trying to only survey modeled “likely voter universes” (which is what internal pollsters do) and they are absolutely horrible at it. Selzer has maintained her strength because she specifically and vocally avoids doing this.
