There was quite a big argument between Nate Silver, from 538, and G. Elliott Morris, the author of this model, which was quite an entertaining read if you’re into statistical drama.
> "The first thing to say is that 72% and 89% can correspond to vote forecasts and associated uncertainties that are a lot closer than you might think."
I'm with Nate Silver on this one, 72% and 89% are as close as 75% and 50% in terms of being double the likelihood for a Republican win.
We will likely never know which model is more accurate but given Nate Silvers history on these predictions I trust him a bit more than others.
His tone comes off as very defensive about his own work rather than constructive criticism of the Economist model.
>In that sense, the best tip-off that the forecasts are different is they have Biden at 97% to win the popular vote, and have sometimes been as high as 98% (and 99% before they revised their model) whereas we are at 82%. You're getting up to 97-98%, you're getting VERY confident.
> His tone comes off as very defensive about his own work rather than constructive criticism of the Economist model.
Most of the linked thread isn't about his own work or the Economist model, but about the false description of their forecasts as being pretty close, made as part of the (true) description that fairly small changes in intermediate results in the Economist model would lead to the same bottom line forecast as Silver's model.
Reading through both Nate’s twitter thread and the article, it reads like Nate never read the article. It seems to address every point he brought up in a reasonable manner. Odd.
Some of the other models for the last election were really bad. Giving something like a 95-98% chance to Hillary was arguably a fundamental failure. I found it very odd how they arrived at those numbers by e.g. treating every state as a separate chance while mostly ignoring that those results are not uncorrelated.
I think he does a better job at emphasizing the uncertainty while still showing that polls can be pretty reliable.
I don't agree, I think he speaks and writes with nuance and intelligence. What is your issue with his Bayesian statistical approach? He went to UofC and LSE so he clearly had top notch training.
His 2016 model and he himself were much more predictive of the trump EC win, he repeatedly stated it was a possible outcome, something the vast majority of other forecasters completely missed.
I believe they got every single state right in three consecutive elections before 2016? And even in 2016, their predictions align reasonably well with the outcome, easily beating all competitors.
For anyone with a history of probabilistic predictions, the goal is to be well-calibrated. For predictions that have 70% likelihood, you want to be right 70% of the time. 90% likelihood, you want to be right 90% of the time. There's a mathematical way to analyze an entire set of predictions and determine how well-calibrated that set of predictions is. Silver claims 538's predictions are well-calibrated.
There are also tools and websites out there that people can use to make predictions and track their own calibration over time. They're pretty fun in terms of encouraging one's own sense of rationality.
I was very unimpressed with his democratic primary model, which he basically fine tuned every week when new data came in in order to get a result that was closer to his own preconceptions of the race.
Doesn't this suffer from low n? There's only been so many presidential elections, and there's only been so many in a similar environment, whatever you think the relevant factors are.
Then there's the one - off problem of evaluating how good the prediction was. If you think it's 95+ for one side then yes, you're embarrassed when it goes the other way. If you think it's 75 then you should be wrong one out of four anyway, but people will still think you messed up.
Impressive work regardless, I always enjoy these stats things.
Not at all. These models all have a multilevel structure, accounting for election level data, district data and poll data. The main source of uncertainty is probably how hard representative polling is, not to mention the polls are carried out by many different organizations.
The logistical challenges are considerable factor this election. Many people may want to vote for say Biden, but their ballots could be disregarded.
Even without the political fight over USPS funding, it is going to be extremely hard for them to turn around in < 1 week as required in many states, there are also the issue with votes that which are postmarked earlier but delivered later than election date( this is with good reason as you don't want results delayed if you wait for votes to come in)
I am not sure how anyone, without detailed data access to postal operations can accurately predict how much of this going to be a problem.
The podcast is for a general audience so it doesn't go into details of statistical modelling used in this model, but it does touch on polling, failures of polling to accurately predict results.
Models are great but doesn't this all hinge on quality of the data inputs? In 2016 polls fielded the day before the election had Clinton up by 10 points in Wisconsin. Why did the polls and the election differ so greatly?
Note that these polls are of Likely Voters. A big source of systematic error in 2016 state polls was Likely Voter models, used to weight raw poll results by voting propensity. In the Midwest in 2016, lots of people without college degrees said they preferred Trump - but based on previous elections, pollsters expected few would vote. This kind of shift is hard to predict - pollsters do not intend to make the same mistake again, but they might make another. (National polls are much less noisy because these kinds of regional errors often cancel out when averaged over the country.)
So do 538 and The Economist try to take these polls, back out their biases, and reapply different ones? Or, to borrow a phrase from Trump, "unskew" them?
Even though there were more people supporting Trump in Wisconsin (or rather any non-left candidate), they couldn’t tell that in polls openly because is goes against vocal minority.
I say this with all seriousness, the main flaw I see is the models not accounting for the levels of voter suppression being applied this year.
One party is on record as saying that their own successful efforts in voter suppression last election won them the rust belt and as a result the presidency.
After seeing it being so successful at tipping the scales last time it appears it will be even more heavily relied on this year.
There are the most recent familiar hints in the news regarding mail in ballots, but so many more examples since the overturning of the voting rights act. How this isn't more of a scandal our democracy is frankly shocking.
It's extremely difficult to talk about this in a purely non-partisan way because there really a case of "both sides are doing it". One party is effectively utilizing this to obtain, maintain and increase their hold on govt positions.
> None of this is a coincidence. Republicans responded to the election of Barack Obama in 2008 not by trying to broaden their own base or appeal to a changing nation, but by modernizing voter suppression tactics out of the old Jim Crow playbook.
Chief Justice John Roberts’s shameful 5-4 decision in 2013’s Shelby County v Holder ripped the heart out of the prime enforcement mechanism of the Voting Rights Act. Roberts argued that the South had changed, and such protections were no longer necessary. 1965, Roberts suggested, was a long time ago, in a different nation, across a changed South. The tens of thousands of Georgians whose voting precincts were closed, forcing them to endure six-hour lines to vote during a pandemic would like a word with him Roberts’s calamitous ignorance has done grievous damage.
"Crosscheck" app/system which is frequently run against democratic counties was found to have a 200 to 1 false positive to true negative rate. It's still being heavily pushed.
> Crosscheck is a national database designed to check for voters who are registered in more than one state by comparing names and dates of birth. Researchers at Stanford University, the University of Pennsylvania, Harvard University, and Microsoft found that for every legitimate instance of double registration it finds, Crosscheck's algorithm returns approximately 200 false positives.[108] Kobach has been repeatedly sued by the American Civil Liberties Union (ACLU) and other civil rights organizations for trying to restrict voting rights in Kansas.
> Often, voter fraud is cited as a justification for such measures, even when the incidence of voter fraud is low. In Iowa, lawmakers passed a strict voter ID law with the potential to disenfranchise 260,000 voters. Out of 1.6 million votes cast in Iowa in 2016, there were only 10 allegations of voter fraud; none were cases of impersonation that a voter ID law could have prevented. Only one person, a Republican voter, was convicted. Iowa Secretary of State Paul Pate, the architect of the bill, admitted, "We've not experienced widespread voter fraud in Iowa."
We're reaching a very concerning threshold of election integrity that is being under appreciated. Our image of ourselves versus our reality is starting to diverge drastically.
I believe there was a much clearer more charitable reading of the above than assuming the person posting doesn't understand the difference between a political party and a Supreme Court.
There was a action. That action was a effective repeal of the voting rights act. The Supreme Court performed that action.
A political party responded to that act by rolling back voting rights.
Will pull up the quote I remember, but a few additional examples showing it is by design beginning early last decade. And well known what the effects are, and it's intentional to win elections.
From 2012:
> House Majority Leader Mike Turzai (R), who said even more clearly in a 2012 speech that voter ID would help Romney carry his state.
> "Voter ID, which is going to allow Governor Romney to win the state of Pennsylvania: done," Turzai said while listing his legislature's accomplishments.
From 2016:
> "We battled to get voter ID on the ballot for the November '16 election," Schimel told conservative host Vicki McKenna on WISN (1130 AM) on Thursday.
>"How many of your listeners really honestly are sure that Senator (Ron) Johnson was going to win re-election or President Trump was going to win Wisconsin if we didn’t have voter ID to keep Wisconsin’s elections clean and honest and have integrity?"
It is well established that voter ID is not preventing any form of widespread in person voter fraud because there is essentially none. Even the president's own commission couldn't find anything of note. But it is extremely effective at suppressing voting of certain groups. For example Texas where valid forms of Id were military Id and concealed carry permits (thinks more likely held by republicans), but invalid forms were state employee photo IDs and university photo IDs (things more likely to be held by those voting democratic).
From prior studies voter roll purges like Crosscheck, eliminate 200 false positives from voter rolls for every 1 true positive. So instead of stopping it, it was rolled out nation wide.[3]
Trump is really good at saying the quiet part out-loud - even if indirectly. Just today leaked quote from 2016 about lower black turnout.
"The president acknowledged in a 2017 meeting with civil rights leaders that he benefited from Black voters staying home."
Combined with Trump campaign's deliberate (they bragged about it in the press) paid media to convinced AfAm voters to sit the election out it's pretty damning.
Might not be as direct a quote as you're looking for but it's literally the first off the top of my head, from today only!
Another good recent one is a lot of vote-by-mail quotes from Trump.
A couple go out on a picnic with a young man. The moment they set out, the young man begins predicting rain. The couple begin to hate him for his gloomy disposition. They meet an old man on the road. He scrounges up his face, looks up at the sky, and says there wont be any rain. He has seen many days like this in his long life and it usually clears up. The couple cheer up at once and praise him for his wisdom.
No sooner had they set up their picnic, then it begins raining heavily. On the drive back completely drenched, they look at the young man with anger. Like somehow he was responsible for the rain. They think of the old man fondly, "well atleast he tried".
This actually happened in the 1988 Mexican election. The PRI was losing the vote and their opposition was predicted to win using live vote tabulation. They hid the results by saying the system crashed and then declared themselves as winning. This kept many people from bothering to even vote.
However, they went much further and burned legitimate ballots and even made fake ones.
The difference here is the PRI actively used (fake) predictions to discourage voters.
Before 2004, I was a "victim" of voter fraud after signing a petition at my university in a swing state. (I put victim in quotes because I did not feel particularly wronged, just annoyed.)
They registered me as Republican, and I received an unexpected voter id with my name misspelled on it in the mail.
I view this as similar to polls or predictions. They'd use the same data, e.g. "In Florida, 29% of voters are registered as Republicans," in an attempt to make other people feel 'safe' joining them.
With voting, some people want to feel a moral victory, that they voted for the winning team, and knowing that there are 29%, instead of 26 or 24 or however much they'd have without cheating (I'm making these numbers up for the example), would have some marginal psychological effect on people to believe that Florida is a partly Republican state, and you're not a complete nutball if you vote for a Republican in Florida.
As an aside, I was actually able to vote with that id, and I'm not sure if that should have given me more or less confidence in the system. More in the sense that, despite someone else messing with the system, I was still able to cast my vote. Less in the sense that, I voted with an incorrect name, and I wonder how easy it might be to conjure up nonexistent people or use dead people to vote.
Just wanted to give another example of how this is true, and the scummy lengths people will go to, to affect an election outcome.
I often notice this trend in my country. No obvious contender one year before the presidential election. Then some random politician is favorite in some poll. Then a feedback loop happens, media focus on this politician, who becomes more popular and so on...
I suppose there are PR firms acting in the shadow to sell their candidates to the media, but nonetheless it seems there's a lot of randomness at play.
Readit News
A piece of it:
https://twitter.com/NateSilver538/status/1294263127668924416...
I'm with Nate Silver on this one, 72% and 89% are as close as 75% and 50% in terms of being double the likelihood for a Republican win.
We will likely never know which model is more accurate but given Nate Silvers history on these predictions I trust him a bit more than others.
Deleted Comment
>In that sense, the best tip-off that the forecasts are different is they have Biden at 97% to win the popular vote, and have sometimes been as high as 98% (and 99% before they revised their model) whereas we are at 82%. You're getting up to 97-98%, you're getting VERY confident.
Most of the linked thread isn't about his own work or the Economist model, but about the false description of their forecasts as being pretty close, made as part of the (true) description that fairly small changes in intermediate results in the Economist model would lead to the same bottom line forecast as Silver's model.
Deleted Comment
Silver made the same point about probability before the 2016 election, and got into a similar Twitter argument (https://www.politico.com/story/2016/11/nate-silver-huffingto...). One guess on who proved to be correct.
I think he does a better job at emphasizing the uncertainty while still showing that polls can be pretty reliable.
His 2016 model and he himself were much more predictive of the trump EC win, he repeatedly stated it was a possible outcome, something the vast majority of other forecasters completely missed.
..and that's only presidential elections. See https://projects.fivethirtyeight.com/checking-our-work/ for their quantitative reflections on accuracy. Looks pretty good.
There are also tools and websites out there that people can use to make predictions and track their own calibration over time. They're pretty fun in terms of encouraging one's own sense of rationality.
Dead Comment
Then there's the one - off problem of evaluating how good the prediction was. If you think it's 95+ for one side then yes, you're embarrassed when it goes the other way. If you think it's 75 then you should be wrong one out of four anyway, but people will still think you messed up.
Impressive work regardless, I always enjoy these stats things.
Even without the political fight over USPS funding, it is going to be extremely hard for them to turn around in < 1 week as required in many states, there are also the issue with votes that which are postmarked earlier but delivered later than election date( this is with good reason as you don't want results delayed if you wait for votes to come in)
I am not sure how anyone, without detailed data access to postal operations can accurately predict how much of this going to be a problem.
https://www.economist.com/podcasts/2020/06/12/modelled-citiz...
The podcast is for a general audience so it doesn't go into details of statistical modelling used in this model, but it does touch on polling, failures of polling to accurately predict results.
Here's the last poll Google collected in Wisconsin in 2016: https://datastudio.google.com/s/kcPRGFaMaO0
Even though there were more people supporting Trump in Wisconsin (or rather any non-left candidate), they couldn’t tell that in polls openly because is goes against vocal minority.
The only question is which way does the majority lean, while the edges pull further apart.
One party is on record as saying that their own successful efforts in voter suppression last election won them the rust belt and as a result the presidency.
After seeing it being so successful at tipping the scales last time it appears it will be even more heavily relied on this year.
There are the most recent familiar hints in the news regarding mail in ballots, but so many more examples since the overturning of the voting rights act. How this isn't more of a scandal our democracy is frankly shocking.
It's extremely difficult to talk about this in a purely non-partisan way because there really a case of "both sides are doing it". One party is effectively utilizing this to obtain, maintain and increase their hold on govt positions.
> None of this is a coincidence. Republicans responded to the election of Barack Obama in 2008 not by trying to broaden their own base or appeal to a changing nation, but by modernizing voter suppression tactics out of the old Jim Crow playbook. Chief Justice John Roberts’s shameful 5-4 decision in 2013’s Shelby County v Holder ripped the heart out of the prime enforcement mechanism of the Voting Rights Act. Roberts argued that the South had changed, and such protections were no longer necessary. 1965, Roberts suggested, was a long time ago, in a different nation, across a changed South. The tens of thousands of Georgians whose voting precincts were closed, forcing them to endure six-hour lines to vote during a pandemic would like a word with him Roberts’s calamitous ignorance has done grievous damage.
"Crosscheck" app/system which is frequently run against democratic counties was found to have a 200 to 1 false positive to true negative rate. It's still being heavily pushed.
> Crosscheck is a national database designed to check for voters who are registered in more than one state by comparing names and dates of birth. Researchers at Stanford University, the University of Pennsylvania, Harvard University, and Microsoft found that for every legitimate instance of double registration it finds, Crosscheck's algorithm returns approximately 200 false positives.[108] Kobach has been repeatedly sued by the American Civil Liberties Union (ACLU) and other civil rights organizations for trying to restrict voting rights in Kansas.
> Often, voter fraud is cited as a justification for such measures, even when the incidence of voter fraud is low. In Iowa, lawmakers passed a strict voter ID law with the potential to disenfranchise 260,000 voters. Out of 1.6 million votes cast in Iowa in 2016, there were only 10 allegations of voter fraud; none were cases of impersonation that a voter ID law could have prevented. Only one person, a Republican voter, was convicted. Iowa Secretary of State Paul Pate, the architect of the bill, admitted, "We've not experienced widespread voter fraud in Iowa."
We're reaching a very concerning threshold of election integrity that is being under appreciated. Our image of ourselves versus our reality is starting to diverge drastically.
https://time.com/5852837/voter-suppression-obstacles-just-am...
https://www.npr.org/2018/10/23/659784277/republican-voter-su...
https://www.nytimes.com/2016/09/17/us/some-republicans-ackno...
https://en.wikipedia.org/wiki/Voter_suppression_in_the_Unite...
https://www.vanityfair.com/news/2020/08/new-postal-service-p...
There was a action. That action was a effective repeal of the voting rights act. The Supreme Court performed that action.
A political party responded to that act by rolling back voting rights.
Do you have a link to the record where that party says this?
From 2012:
> House Majority Leader Mike Turzai (R), who said even more clearly in a 2012 speech that voter ID would help Romney carry his state.
> "Voter ID, which is going to allow Governor Romney to win the state of Pennsylvania: done," Turzai said while listing his legislature's accomplishments.
From 2016:
> "We battled to get voter ID on the ballot for the November '16 election," Schimel told conservative host Vicki McKenna on WISN (1130 AM) on Thursday.
>"How many of your listeners really honestly are sure that Senator (Ron) Johnson was going to win re-election or President Trump was going to win Wisconsin if we didn’t have voter ID to keep Wisconsin’s elections clean and honest and have integrity?"
It is well established that voter ID is not preventing any form of widespread in person voter fraud because there is essentially none. Even the president's own commission couldn't find anything of note. But it is extremely effective at suppressing voting of certain groups. For example Texas where valid forms of Id were military Id and concealed carry permits (thinks more likely held by republicans), but invalid forms were state employee photo IDs and university photo IDs (things more likely to be held by those voting democratic).
From prior studies voter roll purges like Crosscheck, eliminate 200 false positives from voter rolls for every 1 true positive. So instead of stopping it, it was rolled out nation wide.[3]
Or just check out wikipedia.[4]
[1] https://www.jsonline.com/story/news/politics/2018/04/13/atto...
[2] https://www.washingtonpost.com/news/the-fix/wp/2016/04/07/re...
[3] https://www.washingtonpost.com/news/wonk/wp/2017/07/20/this-...
[4] https://en.wikipedia.org/wiki/Voter_suppression_in_the_Unite...
"The president acknowledged in a 2017 meeting with civil rights leaders that he benefited from Black voters staying home."
Combined with Trump campaign's deliberate (they bragged about it in the press) paid media to convinced AfAm voters to sit the election out it's pretty damning.
Might not be as direct a quote as you're looking for but it's literally the first off the top of my head, from today only!
Another good recent one is a lot of vote-by-mail quotes from Trump.
https://www.politico.com/news/2020/08/21/trump-black-voters-...
https://www.theverge.com/2016/10/27/13434246/donald-trump-ta...
https://www.washingtonpost.com/politics/2020/03/30/trump-vot...
Deleted Comment
A couple go out on a picnic with a young man. The moment they set out, the young man begins predicting rain. The couple begin to hate him for his gloomy disposition. They meet an old man on the road. He scrounges up his face, looks up at the sky, and says there wont be any rain. He has seen many days like this in his long life and it usually clears up. The couple cheer up at once and praise him for his wisdom.
No sooner had they set up their picnic, then it begins raining heavily. On the drive back completely drenched, they look at the young man with anger. Like somehow he was responsible for the rain. They think of the old man fondly, "well atleast he tried".
Thats human psychology.
However, they went much further and burned legitimate ballots and even made fake ones.
The difference here is the PRI actively used (fake) predictions to discourage voters.
https://en.m.wikipedia.org/wiki/1988_Mexican_general_electio...
They registered me as Republican, and I received an unexpected voter id with my name misspelled on it in the mail.
I view this as similar to polls or predictions. They'd use the same data, e.g. "In Florida, 29% of voters are registered as Republicans," in an attempt to make other people feel 'safe' joining them.
With voting, some people want to feel a moral victory, that they voted for the winning team, and knowing that there are 29%, instead of 26 or 24 or however much they'd have without cheating (I'm making these numbers up for the example), would have some marginal psychological effect on people to believe that Florida is a partly Republican state, and you're not a complete nutball if you vote for a Republican in Florida.
As an aside, I was actually able to vote with that id, and I'm not sure if that should have given me more or less confidence in the system. More in the sense that, despite someone else messing with the system, I was still able to cast my vote. Less in the sense that, I voted with an incorrect name, and I wonder how easy it might be to conjure up nonexistent people or use dead people to vote.
Just wanted to give another example of how this is true, and the scummy lengths people will go to, to affect an election outcome.
[1]: https://www.bbc.com/news/uk-politics-35350419
[2]: https://www.hkupop.hku.hk/english/report/freedom/FTP_2012.pd...
I suppose there are PR firms acting in the shadow to sell their candidates to the media, but nonetheless it seems there's a lot of randomness at play.