Out of 50 states, Wisconsin was the only one in which the polls were wrong. Well within the 95% expected accuracy within a margin of error. And Pennsylvania was not wrong. The polls there showed Hillary would win by 1.9 points and she lost by 0.7 points, within the margin of error.
RealClearPolitics - Election 2016 - Pennsylvania: Trump vs. Clinton vs. Johnson vs. Stein
I see you have an article from my bete noir there: RealClearPolitics the shameless. I spent all late summer of 2016 going carefully over all its state polls and doing charts and calculations --- anyone who followed RCP obsessively as I did knows that they had Hillary winning BIGLY in both the polls and the articles. Huge win predicted. They were wrong, wrong, wrong ----------- and an amazing waste of time.
I haven't been on that site from then till now. Because they are demonstratively worthless. People --- when something fails as dramatically and decisively as polling and media predictions failed with our 2016 elections, it IS time to pull back and consider that what you've got here could be a completely nonfunctional process. That happens sometimes, like Theranos' failed bloodtesting, the whole company went under, or Enron's business model, when they suddenly went bankrupt. It happened with polls and predictions in 2016: complete and total failure.
I'd recommend not continuing to believe in what doesn't work.