Quinnipiac: Biden +3 in FL, Biden +7 in PA, Biden +5 in OH, Trump +1 in IA

[BudgieForce]

Quinnipiac exposes for the frauds they are.  Now herding/massive undecideds to try and make it look like they weren’t issuing JUNK all cycle. 

This does feel like Quinnipiac's trademark. Release an absurd outlier then come back down to earth.

[kwabbit]

-8 in Biden margin in Florida, -6 in Iowa, -1 in Pennsylvania, +4 in Ohio.

A scary trend just 5 days before Election Day.

-7 for Greenfield in Iowa as well.

Basically no change in PA, moved farther above 50 in fact, and +4 in Ohio is a “scary trend?” Give me a f—king break. The previous Iowa and Florida polls were outliers anyway, they are just coming back down to Earth.

Why do you seem so fundamentally incapable of grasping the concept of a margin of error and of outliers? Not every “movement” in every poll is evidence of a real “trend.” That’s why we have averages in fact; individual polls are useless without context given MOEs and outliers.

I mean MOE implies that the poll is the best guess of the state of the race and there is a confidence interval around it. Qpac's MOE is usually around 3 points. So the first Florida poll is Biden (48, 54) to Trump (37, 43) and the second Florida poll is Biden (43, 49) to Trump (40, 46). The margin then is (5, 17) for the first poll and the second is (-3, 9). The trend is then (-4, 20) shift in the margin towards Trump/against Biden with a mean of 8. This would imply about a 90% percent chance that it shifted towards Trump, a 50% chance that it did shift 8 and a 10% chance that it shifted 16 towards Trump.

These are all 95% confidence intervals.

Of course it could be nothing and the likely real cause of the shift is Qpac herding instead of putting out another wacky result, but if you take these polls at their word then they would likely signal a large actual trend towards Trump. And saying otherwise is bad statistics and polling analysis.

[Skill and Chance]

Quinnipiac exposes for the frauds they are.  Now herding/massive undecideds to try and make it look like they weren’t issuing JUNK all cycle. 

This does feel like Quinnipiac's trademark. Release an absurd outlier then come back down to earth.

Really ironic that Quinnipiac could be Biden's worst polls of the day.

[TiltsAreUnderrated]

Quinnipiac exposes for the frauds they are.  Now herding/massive undecideds to try and make it look like they weren’t issuing JUNK all cycle. 

This does feel like Quinnipiac's trademark. Release an absurd outlier then come back down to earth.

Really ironic that Quinnipiac could be Biden's worst polls of the day.

It's almost election day so they had to release a message like to prime the base for the public reveal of JFK Jr.'s endorsement of Trump

WHERE WE POLL ONE WE POLL ALL

[mijan]

44% Iowans will vote on E day according to Q poll, which is unlikely as 36k people are voting per day in IA. It will only increase as election day comes closer.

851k people have voted so far.Which is 55% of 2016.
Another 200k people will most probably vote early in 5 days .

That means 1051 k people will vote early . Which is 67% of 2016 and 51.1% of all Iowa active  voters.
In 2016 there were 1996 k active Iowa voters, 1566 k voted in 2016. Turnout is 78.45%. Among Republicans turnout was 87%, which will be difficult to replicate this year by the way.

According to Q poll early voting will only 55%, election day voter will be 44%. Then Q poll is estimating 860 k people  will vote on election day . With 1051 k early voters , total turnout  will be 1913 k .
Today in IA there are 2056 k active voters. So according to Q turnout will be 93% in IA.




Turnout in IA will most likely about 1600-1650k  . Q poll massively underestimating early voting.

[GeorgiaModerate]

Q will be downgraded to C- after the election

538 only considers polls with a median field date of 21 days or less before the election when they compute their ratings.  So a pollster that's way out there early in the campaign, but comes back to a reasonable range at the end, isn't going to suffer in the ratings.

[KoopaDaQuick 🇵🇸]

Trump and Ernst winning IA? IA voting to the right of OH? Biden winning FL in a D+8-9 environment? Impossible.

I'm suing your avatar for copyright infringement.

[Hammy]

Comparison with 2016:

Florida was D 46-45, final result was R 48-47, so not too far off from the margin. Still a lot of undecideds but if we account for how they broke in 2016, we'd get about D 50-49.

Iowa was 44-44 and Clinton actually underperformed here, while Trump overperformed by 7. R 52-42 could be drawn from that though I doubt it'll be that wide. Either way probably not good for Biden.

Ohio was R 46-41, final result was R 51-43. Both over-performed and it came down to undecideds, and a similar error would result in Biden 50-48.

Pennsylvania was D 48-43, final result was R 48-47 (though technically only off by a half point with Hillary)--not horrible compared to some of the other Pennsylvania polls. 51-49 can easily be taken from this, too close for comfort but doesn't have room for calling the wrong winner as 2016 as Biden is over 50 with less undecideds.

[Eraserhead]

Those Ohio numbers are quite nice.

[Hope For A New Era]

OH 6 points to the left of IA? Doubt.

I agree with the whole "undecided between

• and not voting" thing. I think those are the new swing voters in the polarized era - not those who swing between candidates but those who swing between voting and not voting.

[Progressive Pessimist]

Florida actually looks reasonable here this time...but then we get that doubtful Ohio result. We just can't win with Quinnipiac.

[Badger]

Those Ohio numbers are quite nice.