About the Quinnipiac Poll showing Biden ahead of Trump by 13 points

Quinnipiac Poll ^ | Matthew Maschinot

[MMaschin]

This poll has been in the news everywhere, but no one seems to be pointing out how flawed (and obviously so) the sample is.

For polls to be accurate they modify their sample to match previous elections exit polling data. This poll attempts to match the 2016 national election poll data and at first glance appears to do so. Here is what 2016 national exit poll party identification looked like.



The above data is how people respond to the question "To what party do you most closely identify? Republican, Democrat, or Independent?". Notice those are the only three options, and respondents must choose one of those 3 responses.

For Quinnipiac to have a valid sample they need to have their sample match the exit poll data as close as possible. To accomplish this they ask a Party Identification question, and modify the sample to match exit poll data. Below is the Quinnipiac Poll's Party ID question -



Notice they added the option of "or what?"

In the 2016 exit polls 'progressives' and 'socialists' were not given an 'or what' option, they would have answered 'Democrat', they do not consider themselves "independent". Yet in the Quinnipiac poll they are given an option to proudly declare they true identification of "progressive" or "socialist". This means that in 2016 exit polling they would have been included in Party ID as 'Democrat' but in this poll they are included in 'Independent'.

The effect this has is dramatic, they are moving there most reliable Democrat votes from 2016 out of the Democrat sample and moving them into the Independent sample. If 8% of the sample answered 'progressive' or 'socialist' then moving them out of the Democrat sample allows that 8% to be replaced by 'Democrats'. It also means that 1/4 of the Independent sample that Trump won by 4% is now composed of 100% guaranteed Democrat votes. This can be seen in the Independent sample, in 2016 Trump won them by 4% yet in this poll Biden wins them by 30%!

Overall this poll is VERY good news for Trump, he's gained in both African American and Hispanic voters. If you look at the demographic responses, understanding that the sample is skewed heavily not only toward Democrats but towards their most alt-left voters, this poll is very good news.

[july4thfreedomfoundation]

SeeBS reported last night that Biden was leading Trump 53 to 40.

SeeBS also said Biden was beating Trump in Florida, North Carolina and Arizona.

Like I said, SeeBS.

Next they’ll be telling us the Easter Bunny is real.

[EternalHope]

Many years ago I taught statistics, including survey sampling, at the University of Washington.

The author’s analysis is a bit simplified, but correct.

Thanks for posting.

[Vaquero]

I call bull$h!t. Quinnipiac is a bunch of leftist scum.

[Alberta's Child]

I would throw any poll results out the window. Not only were the 2016 polls highly inaccurate, but the lamentations of pollster Frank Luntz during the election night news coverage of the results were very telling.

What he said, in effect, was that the 2016 polling data was useless because a huge number of people responding to the polls deliberately lied in their answers.

He was particularly upset because the EXIT POLLS were incorrect. He pointed out that this could only mean one thing: Not only were many Trump supporters unwilling to tell pollsters who they INTENDED to vote for; they also simply lied about who they DID vote for.

[bryan999]

Sshhh! Don’t let the RATs know....

Let them keep thinking they’re winning until the final, bitter end.

[ReaganGeneration2]

1) poll taken too early (e.g., before any Dem indictments occur, before the Wall is more significant, etc)
2) Biden and the others have not had any exposure to voters for what they really are
3) not clear if registered or likely voters or what

[samtheman]

This is why President Hillary became President, because she was so far ahead in the polls.

[DoughtyOne]

Thanks so much for your support to this point... I personally apprecaite it...
FReepers, it's far beyond time to wrap up this FReep-a-thon.  Lets do it today.  Please chip in.

President Donald J. Trump and the Free Republic of the United States of America
President Donald J. Trump's address to the United Nations on 09/19/2017.

Ramirez political cartoon:  Google LARGE VERSION 06/09/2019: LINK  LINK to regular sized versions of his political cartoons (archive).
Garrison political cartoon:  Socialism the Melinial Flytrap LARGE VERSION 06/2019: LINK  LINK (scroll down) to regular sized versions of his political cartoons (archive).

FReepers, 95.945% of the Second Quarter FReep-a-thon goal has been met.  Click above and pencil in your donation now.  Please folks, lets end this FReepathon.  Thank you!
...this is a general all-purpose message, and should not be seen as targeting any individual I am responding to...

Just $48.00 dollars to 96.00%

[USS Alaska]

Many years ago I taught statistics, including survey sampling, at the University of Washington.

Most national polls are designed to skew toward the demonRATs, and they do it out in the open.

If you poll the country based on population, kalifornikator will get 11% of the calls.

The poll is going to lean demonRAT before the first call is made. The same with NY or IL.

Because there are so many heavily populated states that are blue, a national poll can not be fair or predictive.

[SamAdams76]

Remember that over 95% of the media hate and despise Trump. So when you see their "polling", bear in mind that they will do every possible thing that they can to skew their polling to Trump's disfavor.

So long as Trump is in office, we'll never see an honest poll in mainstream media. They will never allow Trump to be over 50% approval either.

The internal polls (which we will never see) will have it right so as the election season gets into gear, look to see where the candidates are campaigning.

If Trump is still campaigning in PA, FL and WI in the final days of the campaign, it's going to be a nail-biter. If however Trump is in OR, VA, CO and NH in the final day, the question becomes how large will Trump's victory be.

I think Hillary's campaign knew she was in trouble in those final days of 2016 (when MSM was busy putting together glossy covers of "Madame President") because they were seeing her internal polls - and probably afraid to show them to Hillary herself.

[MMaschin]

Sshhh! Don’t let the RATs know....

Let them keep thinking they’re winning until the final, bitter end.

The problem with that approach is that it affords the 'RATs' a MUCH better opportunity to win.

First, bogus polls do not suppress voter turnout. Democrats were assured a Hillary win and they had a massive turnout. The only reason Trump won is that Hillary has always been hugely unpopular going back to Hillarycare, and he had an even larger turnout. Any other Democrat would have most likely beat Trump. Also, against Hillary any Republican was almost certain to win. I do thank God that we have Trump instead of an establishment Republican :)

Second, the enthusiasm these fake polls produce among the "useful idiots" results in massive amounts of donations to the Democrats.

[Innovative]

We know there are Democrats in the Republican party as moles and for precisely instances like this one: “I am a registered Republican”, but I still prefer Hillary and hate Trump”...

[Ann Archy]

WHO answers POLLS these days??

[Innovative]

Another phony poll. Nothing new.

As I and others keep saying, ignore all polls, just be sure to get out and VOTE in the actual election.

[Buckeye McFrog]

This poll is agitprop intended to get Millenial Democrat Radicals to shaddyap and get behind Biden.

[Vigilanteman]

If you drill down a little further, you can see that they interviewed REGISTERED voters.

I took a senior level statistics class from one of the professors who worked for Richard Wirthlin, Ronald Reagan's pollster.

One of the more memorable lectures was on games pollsters play. There are a lot of them, including those in play here. A good pollster always asks their client if they want an accurate poll or one which sounds accurate and will confirm what they are selling. If the later, then they use a lot of weasel words in the poll and the design of questions

[edwinland]

I think you’re onto something and your point about Aftrican American and Hispanics is a good one.

According to the odious NBC article linked below, Trump got 8% of the African American vote and 29% of the Hispanic vote. In the poll, he’s at 12% and 33% in a head-to-head with Joe Biden, plus there’s an additional 4% of “Don’t Know”. 4% of African Americans “don’t know” if they would vote for Trump. I think we can safely conclude that it’s really 16% (ie double) and 37%.

Seems pretty unlikely that he could do that much better with African Americans and Hispanics and be so far behind Biden overall.

https://www.nbcnews.com/storyline/2016-election-day/trump-did-better-blacks-hispanics-romney-12-exit-polls-n681386

[TexasGurl24]

Quinnipiac polls should be disregarded out of hand after their huge misses in 2018. They were terrible. I’ve mentioned in several threads their final Florida polls which had Gillum and Nelson each winning by 7. That’s so far outside the margin of error and such a bad miss, that none of their polls should be taken seriously after that.

Their polls have been consistent outliers in regards to the President.

[yoe]

"Remember that over 95% of the media hate and despise Trump."

I think that number is in its self Fake News...Hate and despise for what reasons? Parroting is more like it...repeating and repeating until the "story" is almost fact when there is NO basis for hating and despising other than the Deep State lost and the DS pretty much owns the MSM....