Championship Subdivision Football | FCS Football | Stadiums | Blogs | Forums

It actually happened in May and I didn't catch it:

https://fivethirtyeight.com/features/th ... all-right/

The media narrative that polling accuracy has taken a nosedive is mostly bullshit, in other words. Polls were never as good as the media assumed they were before 2016 — and they aren’t nearly as bad as the media seems to assume they are now. In reality, not that much has changed.

93henfan wrote:

phpBB [video]

It's the present because every time we see a poll come out the response is "well the polls were way off in the 2016 election."

I just saw Michael Caputo do that on CNN. In trying to dismiss the ABC News/Washington Post poll estimating Trump as being way underwater on the job approval front he said the ABC News/Washington Post poll was way off for the 2016 election.

But, actually, the last ABC News/Washington Post poll prior to the 2016 election had a point estimate of Clinton winning the popular vote by 4 percentage points and she actually won it by 2.1 percentage points. So, no, that poll was not "way off." The difference between the point estimate for margin and the actual margin was WELL within what one would expect.

This thing of people dismissing polling data because they think the polls were way off in 2016 is sheer ignorance. You're just kidding yourself if you do it.

I don't get into all the nitnoid minutia that you do, because life is too short, but I do recall that pretty much every poll had Hillary winning the election, and yet she lost.

Let’s not forget how accurate some of the same pollsters were in Great Britain on the Brexit vote...they predicted a rejection...

Col Hogan wrote:Let’s not forget how accurate some of the same pollsters were in Great Britain on the Brexit vote...they predicted a rejection...

Brexit was actual popular vote, though.

Polls that said Hillary was a few points ahead nationally weren't predicting she would win the election. They were predicting she was a few points ahead nationally. And, they were accurate.

Polls










Pols

Skjellyfetti wrote:

Col Hogan wrote:Let’s not forget how accurate some of the same pollsters were in Great Britain on the Brexit vote...they predicted a rejection...

Brexit was actual popular vote, though.

Polls that said Hillary was a few points ahead nationally weren't predicting she would win the election. They were predicting she was a few points ahead nationally. And, they were accurate.

The electoral college polls all (well, at least 98% of them) had Hillary winning the election in the weeks leading up to Election Day. Who cares about the popular vote? Polls are supposed to predict elections and they failed dramatically in 2016. End of story.

Now, if you want to play popular vote in individual states, particularly ones that mattered, how did polls do on the popular votes in Pennsylvania, Michigan, Florida, and Wisconsin? I think they even had Ohio close, and Trump won Ohio by half a million votes. Whiff. They were off.

Let’s not try to complicate something that was obvious for the world to see.

93henfan wrote:

Skjellyfetti wrote:
Brexit was actual popular vote, though.

Polls that said Hillary was a few points ahead nationally weren't predicting she would win the election. They were predicting she was a few points ahead nationally. And, they were accurate.

The electoral college polls all (well, at least 98% of them) had Hillary winning the election in the weeks leading up to Election Day. Who cares about the popular vote? Polls are supposed to predict elections and they failed dramatically in 2016. End of story.

Now, if you want to play popular vote in individual states, particularly ones that mattered, how did polls do on the popular votes in Pennsylvania, Michigan, Florida, and Wisconsin? I think they even had Ohio close, and Trump won Ohio by half a million votes. Whiff. They were off.

Let’s not try to complicate something that was obvious for the world to see.

That's some first class critical thinking right there, 93


Sent from my iPhone using Tapatalk

CID1990 wrote:

93henfan wrote:
The electoral college polls all (well, at least 98% of them) had Hillary winning the election in the weeks leading up to Election Day. Who cares about the popular vote? Polls are supposed to predict elections and they failed dramatically in 2016. End of story.

Now, if you want to play popular vote in individual states, particularly ones that mattered, how did polls do on the popular votes in Pennsylvania, Michigan, Florida, and Wisconsin? I think they even had Ohio close, and Trump won Ohio by half a million votes. Whiff. They were off.

Let’s not try to complicate something that was obvious for the world to see.

That's some first class critical thinking right there, 93


Sent from my iPhone using Tapatalk

It would be more critical if I made the post longer.

JohnStOnge wrote:

93henfan wrote:

phpBB [video]

It's the present because every time we see a poll come out the response is "well the polls were way off in the 2016 election."

I just saw Michael Caputo do that on CNN. In trying to dismiss the ABC News/Washington Post poll estimating Trump as being way underwater on the job approval front he said the ABC News/Washington Post poll was way off for the 2016 election.

But, actually, the last ABC News/Washington Post poll prior to the 2016 election had a point estimate of Clinton winning the popular vote by 4 percentage points and she actually won it by 2.1 percentage points. So, no, that poll was not "way off." The difference between the point estimate for margin and the actual margin was WELL within what one would expect.

This thing of people dismissing polling data because they think the polls were way off in 2016 is sheer ignorance. You're just kidding yourself if you do it.

People who keep focusing on the national popular vote is sheer ignorance. The vote that mattered was the swing state- and the swing state polls were off. Trump won 7 swing states, flipping 5 of them from blue to red (compared to 2012). And the polls in ALL 7 of them under polled Trump.
Trump won Florida by 1.2%, turning it from blue to red. The RCP avg had Trump up .2%, which = 1% off.
Trump won Pennsylvania by .7%, turning it from blue to red.The RCP avg had Clinton up 1.9%, which = 2.6% off.
Trump won North Carolina by 3.7%. The RCP avg had Trump up 1.0%, which = 2.7% off.
Trump won Michigan by .3%, turning it from blue to red. Yet the RCP avg had Clinton up by 3.4%, which = 3.7% off.
Trump won Ohio by 8.1%, turning it from blue to red. The RCP avg had Trump up 3.5%, which = 4.6% off.
Trump won Iowa by 9.5%. The RCP avg had Trump up 3.0%, which = 6.5% off.
Trump won Wisconsin by .7%, turning it from blue to red. Yet the RCP avg had Clinton up 6.5%, which = 7.2% off.

93henfan wrote:I don't get into all the nitnoid minutia that you do, because life is too short, but I do recall that pretty much every poll had Hillary winning the election, and yet she lost.

See, that's the problem You don't correctly recall what the polls were showing. Pretty much every poll showed Hillary winning the popular vote. And she won the popular vote.

BDKJMU wrote:

JohnStOnge wrote: People who keep focusing on the national popular vote is sheer ignorance. The vote that mattered was the swing state- and the swing state polls were off. Trump won 7 swing states, flipping 5 of them from blue to red (compared to 2012). And the polls in ALL 7 of them under polled Trump.
Trump won Florida by 1.2%, turning it from blue to red. The RCP avg had Trump up .2%, which = 1% off.
Trump won Pennsylvania by .7%, turning it from blue to red.The RCP avg had Clinton up 1.9%, which = 2.6% off.
Trump won North Carolina by 3.7%. The RCP avg had Trump up 1.0%, which = 2.7% off.
Trump won Michigan by .3%, turning it from blue to red. Yet the RCP avg had Clinton up by 3.4%, which = 3.7% off.
Trump won Ohio by 8.1%, turning it from blue to red. The RCP avg had Trump up 3.5%, which = 4.6% off.
Trump won Iowa by 9.5%. The RCP avg had Trump up 3.0%, which = 6.5% off.
Trump won Wisconsin by .7%, turning it from blue to red. Yet the RCP avg had Clinton up 6.5%, which = 7.2% off.

Real Clear politics actually had 15 jurisdictions rated as "toss ups." All 7 of the States you mention are among them. What do you think somebody is telling you when they say "toss up?" Do you interpret "toss up" to mean "we say this candidate is going to win?"

Of the 15 jurisdictions rated as "toss up," Trump won 9 and Clinton won 6. The closest spit you could get is 8 and 7.

JohnStOnge wrote:

93henfan wrote:I don't get into all the nitnoid minutia that you do, because life is too short, but I do recall that pretty much every poll had Hillary winning the election, and yet she lost.

See, that's the problem You don't correctly recall what the polls were showing. Pretty much every poll showed Hillary winning the popular vote. And she won the popular vote.

No I recall perfectly. The polls said Hillary would win the electoral college and they were wrong. Kaythanksbyedummy.

JohnStOnge wrote:

93henfan wrote:I don't get into all the nitnoid minutia that you do, because life is too short, but I do recall that pretty much every poll had Hillary winning the election, and yet she lost.

See, that's the problem You don't correctly recall what the polls were showing. Pretty much every poll showed Hillary winning the popular vote. And she won the popular vote.

Immaterial

We are a representative republic with an electoral system. The national popular vote is meaningless.

JohnStOnge wrote:

BDKJMU wrote:

Real Clear politics actually had 15 jurisdictions rated as "toss ups." All 7 of the States you mention are among them. What do you think somebody is telling you when they say "toss up?" Do you interpret "toss up" to mean "we say this candidate is going to win?"

Of the 15 jurisdictions rated as "toss up," Trump won 9 and Clinton won 6. The closest spit you could get is 8 and 7.

I think every prognostication I saw showed that Trump's chance of winning the electoral vote was slim.

I think the problem is that polling errors in each state aren't independent and correlated. You can't assume in some states you'll overestimate the democratic vote and in others underestimate it and it all evens out.

I think that there are indications of a SLIGHT bias towards Clinton but the body of polling was not wildly off. I also think that there was only one State, Wisconsin, such that a confident call might have been made for one candidate then the other candidate won.

I think that if you are going to look at "swing states" the thing to do is something like look at the final RCP electoral maps page. And when you do that you see 15 jurisdictions, not 7, as "toss ups." If you look at the RCP averages for those jurisdictions vs. the Actual results you get this:

1) In 12 of the 15 cases, the candidate with the higher RCP average won the State.

2) There were 10 cases where the RCP average under estimated Trump's performance and 5 cases where it under estimated Clinton's performance.

3) On average, the RCP averages for the 15 jurisdictions over estimated Clinton's performance by 1.6 percentage points.

Another thing I've looked at is what the LATEST polls have to say. Polls are a snapshot. I did it some months ago and looked at it again over the last few days. I did not cherry pick. I established some selection rules prior to doing it. Those rules are as follows:

1) I only considered polls such that the polling period ended in November.
2) I called the poll with the latest end to its polling period the latest unless there was more than one poll with the latest polling date.
3) If more than one poll had the latest polling date I called the one with the latest start date the latest.

There are 23 jurisdictions such that there were polls ending in November. I had to keep two polls for one State, Georgia, because they had latest end date and also had the same start date. So 24 total polls And here is what happened with those "latest polls:"

1) The candidate that got more support from poll respondents won the jurisdiction in 23 of 24 cases. The only exception was...you guessed it...the latest poll of Wisconsin. There was also one case, North Carolina, where the latest poll indicated a tie.

2) Among the 24 referenced latest polls, point estimates of 12 undershot Clinton's performance and point estimates of 12 undershot Trump's performance.

3) On average, the point estimates for the latest polls overshot Clinton's performance by a 1.36 percentage points. And, of course, that is not "significantly" different from 0.

If you choose to continue to believe a myth I guess that's harmless. But understand that if you choose to continue to believe that the polls were wildly off for the 2016 Presidential election you are believing a myth.