WaPo was among the largest errors in 2016.
Their last poll was Clinton +4 and Clinton won the popular vote by 2.1 points. Large error?
Their national poll in 2016 is irrelevent to state polling comparison.
Their last Pennsylvania poll was +6 (46-40) which was off by 7, the average error was 3 for the state.
538 and RCP have no evidence of a Washington Post poll in PA.
It's on Atlas--is it there in error? If not then I have a hard time believing it won't be closer than this.
They were Survey Monkey polls. I'm guessing 538 and RCP did not include them for that reason.
This is a very important thing to distinguish. ABC/WaPo joint polls have been using Langer Research as their pollster for years, and they've been very good. I believe this is the first year Langer has been polling states for them though, previously only did national polls.