What are the chances of Trump winning Pennsylvania again in 2020 Election? (Milton: 2014, living in)

[Duderino]

Originally Posted by thecoalman

That's all? I recall one poll having Hillary up by 14 points about week before the election. You still have some ground to make up.

Admittedly there was some polling error of note in 2016, mostly because non-college educated whites were underestimated as a proportion of the participating voter pool, particularly in "blue wall" states like PA.

Nevertheless, an article was posted on FiveThirtyEight yesterday explaining why Biden's poll leads are different from Clinton's at this point in the cycle. Here's the crux of it:

But some people have dismissed Biden’s lead by pointing out that Hillary Clinton also led in most polls of the 2016 election (Clinton, obviously, ended up losing to Trump). While this is true, Clinton’s lead was much smaller. Applying our current polling-average methodology to 2016 polls, Clinton led national polls by an average of about 4.0 points four months before the 2016 election, and 3.8 points on Election Day itself. So while a normal-sized polling error was enough to throw the 2016 election to Trump, it would take a much bigger — and much unlikelier — polling error for Trump to be ahead right now.

https://fivethirtyeight.com/features...ntons-in-2016/

[golgi1]

Originally Posted by Duderino

Admittedly there was some polling error of note in 2016, mostly because non-college educated whites were underestimated as a proportion of the participating voter pool, particularly in "blue wall" states like PA.

Nevertheless, an article was posted on FiveThirtyEight yesterday explaining why Biden's poll leads are different from Clinton's at this point in the cycle. Here's the crux of it:



https://fivethirtyeight.com/features...ntons-in-2016/

"Polling error". Sure.

"We were drastically wrong in 2016 in order to demoralize people enough to not vote, but here is why we aren't lying now".

LOL. I'm surprised that 538 isn't too embarrassed to be in business after their 2016 election fiasco. They should be.

[staywarm2]

Remember that Jim Comey of the FBI came out 10 days before the election and expressed his disdain for her even as he declared that the FBI had found her innocent. That drastically lowered the number of voters who cast their vote for her. He probably lost the race for her and gave us Trump. Gee thanks, Jim!

There was no way 538 could have known Jim Comey would have done that.

[MarketStEl]

Originally Posted by golgi1

"Polling error". Sure.

"We were drastically wrong in 2016 in order to demoralize people enough to not vote, but here is why we aren't lying now".

LOL. I'm surprised that 538 isn't too embarrassed to be in business after their 2016 election fiasco. They should be.

All opinion polls have a margin of error and have since scientific polling began. You will usually find the margin stated in a smaller number next to the poll results, like this: ±3%.

Three percentage points is the typical margin of error for most polls.

And if the polls had Clinton ahead of Trump by 3 percentage points on the eve of Election Day, they wouldn't have been wrong at all, for she did win the national popular vote by 2.1 points. Trump won in the Electoral College, not the popular vote.

So no, Nate Silver has nothing to hang his head in shame about. Nor, really, do most of the other pollsters.

And, by the way, FiveThirtyEight now does a poll of polls with an average figure, like RealClearPolitics does. ISTR RCP's average also had Clinton up narrowly on the eve of the election.

And her vote-total margin was within the margin of error for every poll.

Where FiveThirtyEight blew it was in its probability prediction, where IIRC it had given Clinton something like a 70% chance of winning the election.

[jazzy jeff]

Originally Posted by MarketStEl

All opinion polls have a margin of error and have since scientific polling began. You will usually find the margin stated in a smaller number next to the poll results, like this: ±3%.

Three percentage points is the typical margin of error for most polls.

And if the polls had Clinton ahead of Trump by 3 percentage points on the eve of Election Day, they wouldn't have been wrong at all, for she did win the national popular vote by 2.1 points. Trump won in the Electoral College, not the popular vote.

So no, Nate Silver has nothing to hang his head in shame about. Nor, really, do most of the other pollsters.

And, by the way, FiveThirtyEight now does a poll of polls with an average figure, like RealClearPolitics does. ISTR RCP's average also had Clinton up narrowly on the eve of the election.

And her vote-total margin was within the margin of error for every poll.

Where FiveThirtyEight blew it was in its probability prediction, where IIRC it had given Clinton something like a 70% chance of winning the election.

Actually Nate Silver was never a "pollster". Rather a statistician. He was basing his predictions on what the polls were saying. He analyzes different pollsters over the years and weighs his prediction on their accuracy over the years.

Silver was the guy that brought sabermetrics into Baseball, even though Bill James gets most of the credit. Silver is well respected in the sports world.

[PGH423]

Originally Posted by MarketStEl

All opinion polls have a margin of error and have since scientific polling began. You will usually find the margin stated in a smaller number next to the poll results, like this: ±3%.

Three percentage points is the typical margin of error for most polls.

And if the polls had Clinton ahead of Trump by 3 percentage points on the eve of Election Day, they wouldn't have been wrong at all, for she did win the national popular vote by 2.1 points. Trump won in the Electoral College, not the popular vote.

So no, Nate Silver has nothing to hang his head in shame about. Nor, really, do most of the other pollsters.

And, by the way, FiveThirtyEight now does a poll of polls with an average figure, like RealClearPolitics does. ISTR RCP's average also had Clinton up narrowly on the eve of the election.

And her vote-total margin was within the margin of error for every poll.

Where FiveThirtyEight blew it was in its probability prediction, where IIRC it had given Clinton something like a 70% chance of winning the election.

How does that probability prediction mean they blew it? A 70% chance of Clinton winning means a 30% chance of Trump winning. I see similar responses to weather forecasts. People seem to think anything above 60% means 100% and anything below 40% means 0%.

[MarketStEl]

Originally Posted by jazzy jeff

Actually Nate Silver was never a "pollster". Rather a statistician. He was basing his predictions on what the polls were saying. He analyzes different pollsters over the years and weighs his prediction on their accuracy over the years.

Silver was the guy that brought sabermetrics into Baseball, even though Bill James gets most of the credit. Silver is well respected in the sports world.

Well, that explains why ESPN bought FiveThirtyEight before handing it off to The New York Times.

PGH423: Good point on the probabilities. And that means that everyone who said the polls blew it in 2016 was wrong. And that also means that most of us don't really understand how scientific polling works, including the guy who seemed to think that "polling error" was tantamount to an admission that polls were fake.

[MarketStEl]

I can tell you from the TV ads currently running in the Philadelphia TV market that the Trump campaign is trying to do two things:

Wrap the radical Left around Joe Biden's neck and raise doubts about his mental capacity.

Something tells me that the former won't gain as much traction in the Southeast as it might once have gotten, and he should talk about mental capacity.

But I wonder whether this might not play differently in the Land of the Forgotten around Pittsburgh?

[PGH423]

Originally Posted by MarketStEl

Well, that explains why ESPN bought FiveThirtyEight before handing it off to The New York Times.

PGH423: Good point on the probabilities. And that means that everyone who said the polls blew it in 2016 was wrong. And that also means that most of us don't really understand how scientific polling works, including the guy who seemed to think that "polling error" was tantamount to an admission that polls were fake.

People want to know which models get it "right" with presidential elections but you really can't determine that with an event that happens so rarely. The closest you can do is look at more frequent events - which models get the state and local elections right. You can actually track whether an event predicted to happen 60% of the time happens 60 out of 100 then. You can argue those more frequent elections are fundamentally different from presidential elections but there is no good way to directly test the accuracy of presidential election models IMO.

[blazerj]

Originally Posted by thecoalman

https://lmgtfy.com/?q=hillary+up+14+points

Those were national polls. The post you replied to said Biden was up 10-12% in PA. The problem with national polls is that people in the firmly Democrat states (which are the more populated ones) don't vote if there isn't a real reason to, since their Electoral College goes 100% to the Dems anyway.

Although, Biden is up 5-10% in PA according to RCP