They still did in 2018 to a meaningful extent (though the effect varied by state). I mean just look at how bad the polls were in Ohio, Indiana, etc. that year. So yes.
Though maybe there are some pollsters who have actually figured something out about how to poll these states more accurately that we can pay more attention to than others (Selzer? Atlas Intel?). I still don't personally believe that Trafalgar even conducts actual polls, so I'm not gonna include them...
No? The polls had Democrats winning gov races in MI, PA, and WI. And senate races in OH and WI.
They rated the gov races in IW and OH as tossups and they were really close. Whats the issue
The polling averages showed Cordray winning OH-Gov by like 4-5 points, and he lost by about the same amount. And the OH-Sen polls got the result right but had Sherrod Brown winning by a lot more than he actually did. The polls in IN-Sen were off by like 7 points and showed Donnelly narrowly ahead on average.
Here's a helpful chart:
I agree that the polls weren't so bad in WI/PA/MI in 2018.