Hey, posting an LMGTFY is basically always extremely condescending and unnecessary. Here's a link that explains the quality of polling in 2016 and talks about some of the misconceptions both before and after the election: https://fivethirtyeight.com/features/the-polls-are-all-right...

Trump lost the popular vote. National polls captured that perfectly.

The polling failure came in the rust belt where the polling models were not tuned to turnout correctly. Turnoit model is the secret sauce of political polling and the most likely source of failure.

I was betting on the election and this proved an expensive failure for me. However, your notion of deliberately skewed polls by 'oversampling' Democratic areas is utter hogwash.

> I see you're trying to gaslight

Would you please stop breaking the site guidelines so we don't have to ban you?

https://news.ycombinator.com/newsguidelines.html

You don't understand what oversampling means in the context of that email.

They were oversampling to get larger than necessary sub groups so that they could get statistically meaningful results of those subsamples.

As for your first comment swing state polling was more mixed. The RCP polling average gave Trump a lead in Florida. Which he won. It gave him a lead in Ohio which he won, it gave him a lead in Nevada, which he lost.

The polling failure was in the rust belt. Where similar demographics met similar flawed turnout model and result in identical polling failure across the States.

But the irony was the polling failure was over estimating Clinton support not under estimating Trump. Trump got less votes than Romney in Wisconsin yet still won because the Democrats vote collapsed spectacularly.