ADVERTISEMENT

Election Prognosticator Season officially opens

And again, the amount they were off overall in 2016 was historically less than average. The polls were pretty good. The large errors were limited to a handful of states, and some of those states had very little good late polling, so it's quite possible the polls we had were very accurate at the time they were actually taken.

One more thing, a poll today is interesting in its own right, but it's informative value for November is low. You have discussed the model's take on uncertainty. It is a fair assumption COVID will play a role in how people vote (maybe even literally). No one knows what will happen with it come November.
 
One more thing, a poll today is interesting in its own right, but it's informative value for November is low. You have discussed the model's take on uncertainty. It is a fair assumption COVID will play a role in how people vote (maybe even literally). No one knows what will happen with it come November.
Correct. Silver points out that the model currently is 77% polls and 23% "fundamentals," and that will gradually shift to 100% polls on election day (under the theory the polls will capture the fundamentals as we approach the election), but Covid could throw everything off. What if Trump announces a vaccine on Nov. 1? There's no way for the polls to accurately reflect that in all 50 states. We could be looking at a situation where 538 is telling us on the eve of the election, "Well, Biden does have a 12 point lead, but the model still thinks Trump has a 30% chance of winning, because, quite frankly, no one knows what the hell is going on."
 


Choo chooo.. It's the KAG2020 train!!
lol.gif
lol.gif
rolleyes.gif
rolleyes.gif


They should have had the 'Destination' as Trump Casino.
 
Last edited:
  • Like
Reactions: TommyCracker
Wisconsin's being VERY heavily invested in, same with Michigan/Pennsylvania. Because of Biden's roll in the auto bailout in Michigan and being from Scranton, PA, he should do better in those two states then Clinton did.

Biden's also booked ad time in Minnesota, Nevada, New Hampshire, and Colorado which are pretty much all the states Trump could have a chance of flipping potentially from 2016.
 
Wisconsin's being VERY heavily invested in, same with Michigan/Pennsylvania. Because of Biden's roll in the auto bailout in Michigan and being from Scranton, PA, he should do better in those two states then Clinton did.

Biden's also booked ad time in Minnesota, Nevada, New Hampshire, and Colorado which are pretty much all the states Trump could have a chance of flipping potentially from 2016.

Trump's not going to win Nevada or Colorado. FL will decide the election IMO. Trump will keep Ohio and NC.
 
  • Like
Reactions: Tacoll
I’m going to admit straight up that I haven’t read the article. sorry/not sorry.

I’m far more concerned with who will actually leave the house to vote, not, what is their vote. not to dismiss such an endeavour. polls are important, and I’m hanging on them nearly daily. I feel though that the decision for most voters is pretty well baked in at this point. The question for me now is, who’s actually going to vote? the more salient metric is turnout, not, who is voting for whom. ive heard the term “likely voters” many a time, and somehow I’m not impressed by that alone.

so again, sorry/not sorry for failing to read the article. when this series of polling manages to include the activation level for voters, only then will i really start paying attention. until then, the random opinions of voters leaving the local walmart or post office who may or may not vote come Election Day dont really move the needle.
 
I’m going to admit straight up that I haven’t read the article. sorry/not sorry.

I’m far more concerned with who will actually leave the house to vote, not, what is their vote. not to dismiss such an endeavour. polls are important, and I’m hanging on them nearly daily. I feel though that the decision for most voters is pretty well baked in at this point. The question for me now is, who’s actually going to vote? the more salient metric is turnout, not, who is voting for whom. ive heard the term “likely voters” many a time, and somehow I’m not impressed by that alone.

so again, sorry/not sorry for failing to read the article. when this series of polling manages to include the activation level for voters, only then will i really start paying attention. until then, the random opinions of voters leaving the local walmart or post office who may or may not vote come Election Day dont really move the needle.
I want to say "Read the article before commenting!" but you bring up a valid concern, which I hope I can assuage. Polling isn't what you think it is. It's not just finding random people and asking who they like. "Likely voter" screens are not alchemy. Pollsters have different methods - some better than others - for screening out who is actually likely to vote. Because of the uncertainty of Covid, this year will really put those methods to the test.

But polling is a lot more scientific than you imagine it to be. It's never the random opinions of voters leaving Wal-Mart.

And 538 is a model that is more than just polls. It adjust all the polls for various factors, ranging from demonstrated bias from that pollster to economic conditions. It does so on multiple levels - state, regional, national. There's a lot that goes into it.
 
  • Like
Reactions: PhyloeBedoe
I want to say "Read the article before commenting!" but you bring up a valid concern, which I hope I can assuage. Polling isn't what you think it is. It's not just finding random people and asking who they like. "Likely voter" screens are not alchemy. Pollsters have different methods - some better than others - for screening out who is actually likely to vote. Because of the uncertainty of Covid, this year will really put those methods to the test.

But polling is a lot more scientific than you imagine it to be. It's never the random opinions of voters leaving Wal-Mart.

And 538 is a model that is more than just polls. It adjust all the polls for various factors, ranging from demonstrated bias from that pollster to economic conditions. It does so on multiple levels - state, regional, national. There's a lot that goes into it.


I appreciate the response, and I’m happy to read that. Ive heard the axiom in the past that basically says we get out and vote FOR issues, and FOR candidates, and not against candidates. i accept that in principle, and i do think (hope is a better word) in this election we will have the impulse to get out and vote out this bad man. we will find lout the limit of how much shit one can stir and still remain in office.
 
Back to the topic at hand, Silver has this post about not counting Trump out too early. In it, he points out that going into Election Day, the 538 model also gave Trump a 29% chance of winning in 2016, but then goes on to describe how Trump's 29% in November 2016 is very different than Trump's 29% chance in August 2020.

Basically, Trump had a chance going into Election Day because the race was close, and a modest polling error in key states could easily swing him the election (which is, in fact, what happened). Now, however, Biden has a much bigger lead than Clinton did, and Trump's 29% chance of winning is rooted almost entirely in uncertainty. To put it simply, the model suspects there is a decent chance the race will become closer as the election approaches. But if it doesn't, if Biden maintains his current lead, then Biden's 71% odds of winning go up to something like 93%.

In other words, it's a good lesson that models are, to put it mildly, volatile. As the next few months go by, Biden could turn this into a rout, or Trump could close the gap. We still have the conventions and the debates and who knows what will happen with the economy and Covid. It's a model, not a prediction. That's a distinction that too many people didn't clearly make in 2016.
It will be Biden in a rout.
 
It won’t affect my life either way, but you’re probably right. We need sanity back.
Thanks for that. Really. We need more people of your leanings to come to that realization.

And here's some fodder for another thread while I've got your attention...

What do people like yourself think of the situation with the NRA leadership? I don't believe I saw you comment on that when it went down.
 
Thanks for that. Really. We need more people of your leanings to come to that realization.

And here's some fodder for another thread while I've got your attention...

What do people like yourself think of the situation with the NRA leadership? I don't believe I saw you comment on that when it went down.
TBH, I haven’t really paid much attention to it. But upon further reading, it looks like it started with a pretty acrimonious split between the NRA and it’s former ad firm. There’s also a sizable rift between Oliver North and LaPierre, the organizations 2 most powerful people.

Numerous lawsuits have been filed by all parties and now the state of New York has jumped in, which I have no doubt is politically motivated.

What will come of all this, I have no idea.
 
USC panel poll, claim to fame is that they are a poll that showed Trump leading going into the 2016 election.

They have Biden up 11 points nationally and have an interesting breakdown of voter preferences compared to 2016 votes. From George Conway's twitter feed, retweet of an email.

9% of 2016 Trump voters now going Biden
4% of 2016 Clinton voters now going Trump
41% of 2016 3rd party voters now going Biden.
30% of 2016 3rd party voters now going Trump.
52% of 2016 nonvoters now going Biden.
32% of 2016 nonvoters now going Trump.

take-home: People on the sidelines in 2016 (3rd party & non-voters) seem to be strongly favoring Biden.



EfyfSbgWAAAOiqM
 
  • Like
Reactions: brianiu
I have a number of Libertarian friends who seem to follow this trend, leaning very much anti-Trump, citing the secret police force urban actions, postal shenanegans, and fiscal irresponsibility. They are for he most part deciding to lean Biden rather than voting Libertarian.
 
  • Like
Reactions: UncleMark
Minnesota is now tied. A state Obama won by I believe 10 points on average.
Minnesota is not "tied." It's probably closer than Democrats would like, but you need to be careful about taking a single outlier poll as indicative of the actual status on the ground, especially when it's from a partisan polling firm.

There has been only one other poll taken in Minnesota in August. Truth is, we don't really know who is leading there right now.
 
ADVERTISEMENT