2020 Trump polling thread

[Northwoods]

2016 polling was rather ... off.  They showed HRC so far ahead that 538.com was giving her something like 90% chance of winning right up until election day. 

Now polls are supposedly showing Biden with a significant and growing lead.  The political class is all excited/worried depending on which side you're reading.

https://amp.cnn.com/cnn/2020/06/04/politics/electoral-map-2020-election-donald-trump-joe-biden/index.html

https://hotair.com/archives/allahpundit/2020/06/04/trump-summons-campaign-staff-meeting-polls-show-bidens-lead-growing/

So, here's a thread to discuss the polling and if we think it's plausible, and why.

Fwiw, my opinion is that polling now is almost meaningless.  Not only are they surveying registered voters rather than likely voters, but they are probably missing a lot of subgroups due to how they conduct the polling, but also because the situation is so fluid now that the effects of the pandemic, George Floyd effects, etc won't really be known for a while.

[Ron]

All news media, all lies all the time.

Even when they say a true thing it is framed in a narrative of lies.

[DittoHead]

538.com was giving her something like 90% chance of winning right up until election day. 

Polling sucks, but at least be accurate about how much it sucks.

Clinton is a 71 percent favorite to win the election according to our polls-only model and a 72 percent favorite according to our polls-plus model.

Nationwide polling for the presidential election is pretty much meaningless, yet it seems to get the most attention. Statewide polling doesn't seem to be as accurate, it's usually done by smaller outfits and with a wider margin of error. Combining all of those margins of error to get a decent electoral college prediction doesn't seem easy - I certainly don't have the statistical background to do that kind of thing. Polling this far out is also pretty meaningless, so much can change and when it's a close election (like they all seem to be) a good October surprise can make a difference that polling won't have time to reflect.

A lot of people now seem to put more trust in betting markets, I don't know if their track record is any better though.

[Ben]

I don't know - polls are polls. They have most always to me, been barely better than flipping a coin on many things that are not of a documented high probability ("Will the sun rise tomorrow?").

I will say though, that I certainly give close to zero credence to any polls run by the MSM, as they have, more and more, seemed to me to be almost intentionally wrong and just another tool of "journalistic activism".

I mean, pretty much every MSM poll indicates that like 80% of the country wants ARs banned and general gun licensing. What just happened to gun sales over the last few months? Record numbers, including a good percentage of liberals who ostensibly want everybody's guns taken away.

Also, to the OP, as mentioned above, it's still early presidential poll-wise, and there are plenty of dummies blaming Trump (or "X" incumbent president) for the virus, among other things that actually have nothing to do with him. Polling will change based on whatever "outrage" is happening or not happening around October.

[makattak]

Polling sucks, but at least be accurate about how much it sucks.
Nationwide polling for the presidential election is pretty much meaningless, yet it seems to get the most attention. Statewide polling doesn't seem to be as accurate, it's usually done by smaller outfits and with a wider margin of error. Combining all of those margins of error to get a decent electoral college prediction doesn't seem easy - I certainly don't have the statistical background to do that kind of thing. Polling this far out is also pretty meaningless, so much can change and when it's a close election (like they all seem to be) a good October surprise can make a difference that polling won't have time to reflect.

A lot of people now seem to put more trust in betting markets, I don't know if their track record is any better though.

This final numbers are not indicative of the state of polling prior to the election. IF, as many people believe, most early polling is about SHAPING public opinion, rather than reflecting it, the polls by election day have to more closely represent reality and claim that there has been massively variable shift in public opinion.

On August 15, 538 predicted an 89.2% chance that Hilary won the election.

Interestingly, on July 30, they predicted basically 50%, which is the only time the model accurately predicted the outcome. On June 8, it was 66% Hilary.

That is to say, polling is basically worthless at this time. Likely because the pollsters have an agenda other than presenting data.

[HankB]

. . . That is to say, polling is basically worthless at this time. Likely because the pollsters have an agenda other than presenting data.

This.

IMHO there's a perception among pollsters that by favoring a particular candidate, polls can influence more of the voting electorate to favor that same candidate.

Pollsters also don't factor in non-responses to their polling or control for those who will, quite simply, LIE to a voice on a phone asking their opinions.

[DittoHead]

Interestingly, on July 30, they predicted basically 50%, which is the only time the model accurately predicted the outcome.

Saying there's a 50/50 chance is not accurately predicting anything 

IMHO there's a perception among pollsters that by favoring a particular candidate, polls can influence more of the voting electorate to favor that same candidate.

While pollsters might believe that, I don't. If anything, I think the most effective way to energize your voters would be to show a close race with your candidate just slightly behind. Even that wouldn't do much to influence voters.

[Perd Hapley]

Saying there's a 50/50 chance is not accurately predicting anything 

It predicted a close race, which it was.

[makattak]

Saying there's a 50/50 chance is not accurately predicting anything 

Very small changes in Hilary's direction would have tipped the election to her.

I'd say that putting the odds around 50/50 is a MUCH better prediction than any of the other ones.

[DittoHead]

It predicted a close race, which it was.

They often predicted a close race, that's not the same as a 50/50 probability of a specific outcome.  Even when the predictions looked lopsided there was still plenty of uncertainty.

To say that Clinton has (roughly) an 85 percent chance of winning is not to say that she’s going to glide to victory 85 percent of the time. The 85 percent figure includes its share of close calls, along with a (larger) share of blowouts. It includes a few oddball scenarios, like Clinton winning the Electoral College while losing the popular vote.

[cordex]

They often predicted a close race, that's not the same as a 50/50 probability of a specific outcome.  Even when the predictions looked lopsided there was still plenty of uncertainty.

I don't think the sentence you quoted says what you think it says.

[grampster]

Ummm The Stupid Republican Party, if they had any sense whatsoever, every ad in every sort of media would be based on the fact that America is witnessing, twice, in 4 months what an abject failure that LibProg "governance" has proven to be.  Then tick of point by point every failed program and policy, what party was in charge of what states and cities for how long where those failures happened.  Included in that is a point by point revelation of how the Ds have failed with every program and policy they've pushed for has failed.

[HankB]

Ummm The Stupid Republican Party, if they had any sense whatsoever, every ad in every sort of media would be based on the fact that America is witnessing, twice, in 4 months what an abject failure that LibProg "governance" has proven to be.  Then tick of point by point every failed program and policy, what party was in charge of what states and cities for how long where those failures happened.  Included in that is a point by point revelation of how the Ds have failed with every program and policy they've pushed for has failed.

Could be very effective if done right. Take the state that provided and excuse for the current kerfuffle and rioting:

Minneapolis
Mayor: Democrat
City Council: 12 Democrats, one Green
5th Congressional District representing Minneapolis and some surrounding areas: Democrat Ihlan Omar, who married her own brother in an immigration scam
Hennepin County Attorney: Democrat

Minnesota
Governor: Democrat
Senators: Both Democrats
Representatives: 5 of 8 Democrats
Attorney General: Democrat

I used to live in PROM (People's Republic Of Minnesota) just east of St. Paul. The Democrats of Minneapolis have been anti-cop for decades; at one time, they had a resolution that all cops should be disarmed, like British bobbies.