I have been trying to get someone at The Bulwark to write about Allan LIchtman, who has accurately predicted nine of the last ten presidential elections. He is a an historian who knows a lot about math and probability. dailykos.com/stories/20…
JFK certainly thought the debates helped him, though the margin was so small it’s hard to tell, and Lichtman wasn’t around to predict that election. Lichtman did predict HW Bush’s defeat before any debates were held. Lichtman predicted Gore’s victory in August of 2000, and Gore’s loss wasn't due to the debates so much as the Supreme Court hijacking the election.
Lichtman used data from many elections to formulate his system, including JFK. Lichtman also insisted post debate that Biden's odds are still better than anyone else's.
That’s true: he did use JFK’s election, but he did not PREDICT JFK’s election. And Lichtman has said repeatedly that debates don't count, particularly ones in June.
I never claimed he predicted JFK's election. He used election data from elections going back to 1840, and the historical environment of those elections. That is how he developed his 13 keys which he began using starting in 1982. I am not sure what you think I am saying.
If we want to talk probabilities, let's do this one:
If you're trying to predict presidents by simply tossing a coin, you have a 1/2 chance on any given election. To get 9 out of 10 right is a 5/512 chance, or a little under 1%. But there are a lot of people in the business of predicting presidential elections. If we have 1,000 of them, (And that particular prognostication market is bigger than that) then we're averaging a little under 10 people who have correctly called the past 9 out of 10 presidential elections, even if their methods are no better than just random guessing.
Now Lichtman might really have the goods. But simply pointing to his record of prediction as proof positive that his methods work doesn't really work out in the probability framework we have; there is a lot of potential for false positives.
But Lichtman is not guessing as if it were a coin toss. No one is saying that his record is proof positive that his methods work. We are saying that his track record means he needs to be seriously considered. Besides which "a little under 10 people" besides Lichtman? Name them. You are completely misunderstanding the probability math at work here.
"But Lichtman is not guessing as if it were a coin toss."
Not what I said. Please respond to the comment, not your strawman.
"No one is saying that his record is proof positive that his methods work. "
I don't know, I think Julie's comment about invoking predicting 9 out of 10 is in fact doing just that. Maybe she can clarify.
"Besides which "a little under 10 people" besides Lichtman? Name them."
I don't think you understood what I wrote. In fact, I'm sure you didn't, because you cannot actually have a fraction of a person making a prediction. I'm not pointing to specific people making guesses. I'm pointing out that with 1,000 people making guesses, you will get about 9.77 who will correctly guess 9 out of 10 presidential elections. There is no actual such thing as .77 of a person who can make such a prediction. These are not actual people I am identifying. I am talking about the probability of false positives even if the method used to guess is worthless.
"You are completely misunderstanding the probability math at work here."
No, I am not. To demonstrate that, I will walk you through all of it.
Odds of calling one particular presidential race completely at random assuming you're only picking between two major party candidates: 1/2.
Binomial formula for the probability of getting K successes (correct predictions) in N trials (total number of predictions) with the success of each trial being p is
P(X=k) =(N) p^k(1-p)^n-k
(k)
For us, N is 10 (total number of predictions), K is 9 for the number of correct predictions, and p is 0.5 The bonimal coefficient of n over k in parenthesis is
n!/k!(n-k)!
At that point it's just arithmetic P(X=9) =10(1/2)^9(1/2)=10(1/2)^10= 10(1/1024)=10/1024=5/512.
The probability of any one person getting 9 out of 10 presidential predictions correctly just by guessing is 5/512, or slightly off 0.977 percent.
Now, if we hypothesize 1,000 people all doing this process, with our 5/512 conclusion from before, we have a simple product of number of people attempting independently by the odds of overall success. 1,000 *0.977 =9.77.
You say I misunderstand the probability math? Show me where I made an error. And please, take your time. I used some statistical language, I wouldn't want to be guilty of gish galloping you.
Now we know you are in in bad faith. A little under 10 people is a couple handfuls of whole people. Name them.
Everything falls apart with your false assumption that Lichtman is "calling one particular presidential race completely at random." He most definitely is not calling presidential races in a random way, like by flipping a coin.
It's funny how you need to literally make a strawman out of my argument to 'prove' that I am acting in bad faith. Meanwhile, you've also accused me of making a gish gallop by using statistical language, and accused me of making a mathematical error. I've challenged you to point out where the flaw in my math was. Do you have one or not?
Again, Lichtman is being brought to the fore because of survivorship bias and because he's pushing against the conventional narrative. People like Lichtman are common but not famous. I don't know who these 9.77 people my model posits are, because they're just people who work at polling houses in obscurity who called the past 9/10 elections with whatever systems they have which might or might not be better than guesswork. Only, they're not getting up and standing saying that you shouldn't worry even though is model is vague and his track record isn't actually as strong as you want it to be.
I want to make this very, very clear, since you are at best fuzzy on what I am saying. I do not think Lichtman is making his model by simple guesswork. I think he has a system, which he applies. I do not know if this system is actually accurate, and bringing up a 9/10 prediction rate is not as strong of an indicator as you might intuitively think, because there is a VERY high probability of multiple people that same point simply by guessing, or more realistically using systems that are no more accurate than guessing. Do you get the difference?
Your math is like solving for 2+2, when the problem is 2+3. Lichtman is not making random guesses no mater how much you wish it were so. There is no "conventional" narrative. There is a narrative being pushed by one faction while ignoring or dismissing out of hand any arguments to the contrary. You THINK he has a system? He has explained his system. By your lights there is no way anyone can analyze their way to a valid or reliable conclusion because it all amounts to an random guess anyway. If ANYBODY else had the same track record as Lichtman, you know they would be guests at all the political podcasts. But they aren't because they do not exist. Anybody can apply his system and get the same result. The only people in these threads getting a different result are people claiming to use his system by using their own definitions of the keys instead of his definitions or otherwise misapplying his system, like the people claiming Trump turns the Charisma key.
"Your math is like solving for 2+2, when the problem is 2+3."
Do you or do you not agree that the odds of someone guessing at random between the two major party candidates will get 9 out of 10 presidential picks correctly 5/512 of the time? Because I don't see how you argue with the math, it's a simple binomial distribution. And if you can't argue with the math, then don't say I made a mathematical error.
"Lichtman is not making random guesses no mater how much you wish it were so. "
I have never said that he was. Stop lying.
"By your lights there is no way anyone can analyze their way to a valid or reliable conclusion because it all amounts to an random guess anyway"
I have never said or implied that. Stop lying.
" If ANYBODY else had the same track record as Lichtman, you know they would be guests at all the political podcasts."
Don't be ridiculous. If someone had gotten the past 9 out of 10 presidential elections and said "well, this one is too early to tell because I don't have the facts in", or "Yeah, by all indications, Biden is going to lose" he would not be on all the political podcasts. That's not the news of the day or what people want to bring in an expert to try to reinforce a counterintuitive opinion. There are almost certainly people working in some polling houses with the same record as Lichtman who are not and will not be called up for precisely that reason.
"The only people in these threads getting a different result are people claiming to use his system by using their own definitions of the keys instead of his definitions or otherwise misapplying his system, like the people claiming Trump turns the Charisma key."
Literally my entire argument has been about applied statistics and the odds of getting to the same 9/10 correct predictions even if your system is not actually any better than guessing. Stop making irrelevant tangents.
By the way, why did you accuse me of making a gish gallop by using statistical language? Was that another lie?
As I said you solved a math problem different from the one before us. Your statistical probability math does not apply to this problem because the assumption of random guessing does not apply to this problem as you admitted in your comment which is also internally inconsistent. You admit Lichtman is not randomly guessing ("I have never said that he was [randomly guessing])" and then argue his system "is not actually any better than guessing." His system is a methodical way of analyzing the election according to thirteen factors. So yes, you most definitely implied exactly what I said you implied.
"As I said you solved a math problem different from the one before us."
It is the one before us if we want to demonstrate that Lichtman's record of predicting 9 out of the past 10 elections is actually meaningful.
"Your statistical probability math does not apply to this problem because the assumption of random guessing does not apply to this problem"
It very much applies. Because Lichtman's expertise is being presented on his track record, and his track record could very easily have been the product of a system that is no better than just guessing, given the numbers involved.
" as you admitted in your comment which is also internally inconsistent."
There is nothing internally inconsistent in my comment. Stop lying.
"You admit Lichtman is not randomly guessing ("I have never said that he was [randomly guessing])" and then argue his system "is not actually any better than guessing.""
Yes. Those are two different things. What I am arguing, is that the 13 keys system he has cannot be proven because he got 9 out of the past 10 elections correct by using it. That is because a completely arbitrary system could also have gotten 9 out of the past 10 elections correct and it is in fact virtually certain that someone using a system that is no better than randomly guessing has also gotten 9 out of the past 10 elections correctly, due to the large number of people making predictive models.
That's why I brought up the 'height and hair' satirical model in the other thread. You can have a system that is predictive, uses real, empirical data, and still be totally worthless. You can even have such a system make correct predictions! But because you are not grasping the probabilities involved, you are taking something that actually isn't that remarkable and elevating it to the level of something fantastical and willing to stake it against countervailing evidence. Then, when you're challenged on it, you lie repeatedly, including making claims that are patently nonsensical. Why, for instance, does bringing up statistical terminology make a gish gallop?
I'm sure that is usually true. But watching Biden was the equivalent of watching someone have a stroke or heart attack on the debate stage. Barack Obama's first debate was Cicero compared to Biden. Also, the whole reason the Biden team asked for the early debate was to put to rest the concerns people had about Biden's age and mental acuity. The only thing he did was to confirm and amplify people's concerns exponentially.
Yeah, he said he was feeling awful. He probably should have postponed it, but then think of how the right would have spun that. He is damned if he does and damned if he doesn't.
I have been trying to get someone at The Bulwark to write about Allan LIchtman, who has accurately predicted nine of the last ten presidential elections. He is a an historian who knows a lot about math and probability. https://www.dailykos.com/stories/2024/7/8/2252438/-Lichtman-s-Keys-to-the-White-House-predict-Biden-win
I saw his reaction on CNN. His formula doesn't include that joint appearance that is laughingly called a debate.
Lichtman watched the debate, and he has said over and over again that no debate has ever affected the 13 keys.
Tell that to JFK, George H W Bush and Al Gore.
JFK certainly thought the debates helped him, though the margin was so small it’s hard to tell, and Lichtman wasn’t around to predict that election. Lichtman did predict HW Bush’s defeat before any debates were held. Lichtman predicted Gore’s victory in August of 2000, and Gore’s loss wasn't due to the debates so much as the Supreme Court hijacking the election.
Lichtman used data from many elections to formulate his system, including JFK. Lichtman also insisted post debate that Biden's odds are still better than anyone else's.
That’s true: he did use JFK’s election, but he did not PREDICT JFK’s election. And Lichtman has said repeatedly that debates don't count, particularly ones in June.
I never claimed he predicted JFK's election. He used election data from elections going back to 1840, and the historical environment of those elections. That is how he developed his 13 keys which he began using starting in 1982. I am not sure what you think I am saying.
I actually think that debate became a large scandal, so that would false another key.
You have to use Lichtman's definitions, not your own. Besides, that is really stretching the word "scandal" almost as if you want keys to fail..
If we want to talk probabilities, let's do this one:
If you're trying to predict presidents by simply tossing a coin, you have a 1/2 chance on any given election. To get 9 out of 10 right is a 5/512 chance, or a little under 1%. But there are a lot of people in the business of predicting presidential elections. If we have 1,000 of them, (And that particular prognostication market is bigger than that) then we're averaging a little under 10 people who have correctly called the past 9 out of 10 presidential elections, even if their methods are no better than just random guessing.
Now Lichtman might really have the goods. But simply pointing to his record of prediction as proof positive that his methods work doesn't really work out in the probability framework we have; there is a lot of potential for false positives.
But Lichtman is not guessing as if it were a coin toss. No one is saying that his record is proof positive that his methods work. We are saying that his track record means he needs to be seriously considered. Besides which "a little under 10 people" besides Lichtman? Name them. You are completely misunderstanding the probability math at work here.
"But Lichtman is not guessing as if it were a coin toss."
Not what I said. Please respond to the comment, not your strawman.
"No one is saying that his record is proof positive that his methods work. "
I don't know, I think Julie's comment about invoking predicting 9 out of 10 is in fact doing just that. Maybe she can clarify.
"Besides which "a little under 10 people" besides Lichtman? Name them."
I don't think you understood what I wrote. In fact, I'm sure you didn't, because you cannot actually have a fraction of a person making a prediction. I'm not pointing to specific people making guesses. I'm pointing out that with 1,000 people making guesses, you will get about 9.77 who will correctly guess 9 out of 10 presidential elections. There is no actual such thing as .77 of a person who can make such a prediction. These are not actual people I am identifying. I am talking about the probability of false positives even if the method used to guess is worthless.
"You are completely misunderstanding the probability math at work here."
No, I am not. To demonstrate that, I will walk you through all of it.
Odds of calling one particular presidential race completely at random assuming you're only picking between two major party candidates: 1/2.
Binomial formula for the probability of getting K successes (correct predictions) in N trials (total number of predictions) with the success of each trial being p is
P(X=k) =(N) p^k(1-p)^n-k
(k)
For us, N is 10 (total number of predictions), K is 9 for the number of correct predictions, and p is 0.5 The bonimal coefficient of n over k in parenthesis is
n!/k!(n-k)!
At that point it's just arithmetic P(X=9) =10(1/2)^9(1/2)=10(1/2)^10= 10(1/1024)=10/1024=5/512.
The probability of any one person getting 9 out of 10 presidential predictions correctly just by guessing is 5/512, or slightly off 0.977 percent.
Now, if we hypothesize 1,000 people all doing this process, with our 5/512 conclusion from before, we have a simple product of number of people attempting independently by the odds of overall success. 1,000 *0.977 =9.77.
You say I misunderstand the probability math? Show me where I made an error. And please, take your time. I used some statistical language, I wouldn't want to be guilty of gish galloping you.
Now we know you are in in bad faith. A little under 10 people is a couple handfuls of whole people. Name them.
Everything falls apart with your false assumption that Lichtman is "calling one particular presidential race completely at random." He most definitely is not calling presidential races in a random way, like by flipping a coin.
It's funny how you need to literally make a strawman out of my argument to 'prove' that I am acting in bad faith. Meanwhile, you've also accused me of making a gish gallop by using statistical language, and accused me of making a mathematical error. I've challenged you to point out where the flaw in my math was. Do you have one or not?
Again, Lichtman is being brought to the fore because of survivorship bias and because he's pushing against the conventional narrative. People like Lichtman are common but not famous. I don't know who these 9.77 people my model posits are, because they're just people who work at polling houses in obscurity who called the past 9/10 elections with whatever systems they have which might or might not be better than guesswork. Only, they're not getting up and standing saying that you shouldn't worry even though is model is vague and his track record isn't actually as strong as you want it to be.
I want to make this very, very clear, since you are at best fuzzy on what I am saying. I do not think Lichtman is making his model by simple guesswork. I think he has a system, which he applies. I do not know if this system is actually accurate, and bringing up a 9/10 prediction rate is not as strong of an indicator as you might intuitively think, because there is a VERY high probability of multiple people that same point simply by guessing, or more realistically using systems that are no more accurate than guessing. Do you get the difference?
Your math is like solving for 2+2, when the problem is 2+3. Lichtman is not making random guesses no mater how much you wish it were so. There is no "conventional" narrative. There is a narrative being pushed by one faction while ignoring or dismissing out of hand any arguments to the contrary. You THINK he has a system? He has explained his system. By your lights there is no way anyone can analyze their way to a valid or reliable conclusion because it all amounts to an random guess anyway. If ANYBODY else had the same track record as Lichtman, you know they would be guests at all the political podcasts. But they aren't because they do not exist. Anybody can apply his system and get the same result. The only people in these threads getting a different result are people claiming to use his system by using their own definitions of the keys instead of his definitions or otherwise misapplying his system, like the people claiming Trump turns the Charisma key.
"Your math is like solving for 2+2, when the problem is 2+3."
Do you or do you not agree that the odds of someone guessing at random between the two major party candidates will get 9 out of 10 presidential picks correctly 5/512 of the time? Because I don't see how you argue with the math, it's a simple binomial distribution. And if you can't argue with the math, then don't say I made a mathematical error.
"Lichtman is not making random guesses no mater how much you wish it were so. "
I have never said that he was. Stop lying.
"By your lights there is no way anyone can analyze their way to a valid or reliable conclusion because it all amounts to an random guess anyway"
I have never said or implied that. Stop lying.
" If ANYBODY else had the same track record as Lichtman, you know they would be guests at all the political podcasts."
Don't be ridiculous. If someone had gotten the past 9 out of 10 presidential elections and said "well, this one is too early to tell because I don't have the facts in", or "Yeah, by all indications, Biden is going to lose" he would not be on all the political podcasts. That's not the news of the day or what people want to bring in an expert to try to reinforce a counterintuitive opinion. There are almost certainly people working in some polling houses with the same record as Lichtman who are not and will not be called up for precisely that reason.
"The only people in these threads getting a different result are people claiming to use his system by using their own definitions of the keys instead of his definitions or otherwise misapplying his system, like the people claiming Trump turns the Charisma key."
Literally my entire argument has been about applied statistics and the odds of getting to the same 9/10 correct predictions even if your system is not actually any better than guessing. Stop making irrelevant tangents.
By the way, why did you accuse me of making a gish gallop by using statistical language? Was that another lie?
As I said you solved a math problem different from the one before us. Your statistical probability math does not apply to this problem because the assumption of random guessing does not apply to this problem as you admitted in your comment which is also internally inconsistent. You admit Lichtman is not randomly guessing ("I have never said that he was [randomly guessing])" and then argue his system "is not actually any better than guessing." His system is a methodical way of analyzing the election according to thirteen factors. So yes, you most definitely implied exactly what I said you implied.
"As I said you solved a math problem different from the one before us."
It is the one before us if we want to demonstrate that Lichtman's record of predicting 9 out of the past 10 elections is actually meaningful.
"Your statistical probability math does not apply to this problem because the assumption of random guessing does not apply to this problem"
It very much applies. Because Lichtman's expertise is being presented on his track record, and his track record could very easily have been the product of a system that is no better than just guessing, given the numbers involved.
" as you admitted in your comment which is also internally inconsistent."
There is nothing internally inconsistent in my comment. Stop lying.
"You admit Lichtman is not randomly guessing ("I have never said that he was [randomly guessing])" and then argue his system "is not actually any better than guessing.""
Yes. Those are two different things. What I am arguing, is that the 13 keys system he has cannot be proven because he got 9 out of the past 10 elections correct by using it. That is because a completely arbitrary system could also have gotten 9 out of the past 10 elections correct and it is in fact virtually certain that someone using a system that is no better than randomly guessing has also gotten 9 out of the past 10 elections correctly, due to the large number of people making predictive models.
That's why I brought up the 'height and hair' satirical model in the other thread. You can have a system that is predictive, uses real, empirical data, and still be totally worthless. You can even have such a system make correct predictions! But because you are not grasping the probabilities involved, you are taking something that actually isn't that remarkable and elevating it to the level of something fantastical and willing to stake it against countervailing evidence. Then, when you're challenged on it, you lie repeatedly, including making claims that are patently nonsensical. Why, for instance, does bringing up statistical terminology make a gish gallop?
I doubt his work includes senile presidents who stare slack jawed at the camera during debates.
Probably not convicted felons and sexual assaulters either.
He would argue, I think, that debates have never determined elections. Barack Obama's first debate was terrible.
I'm sure that is usually true. But watching Biden was the equivalent of watching someone have a stroke or heart attack on the debate stage. Barack Obama's first debate was Cicero compared to Biden. Also, the whole reason the Biden team asked for the early debate was to put to rest the concerns people had about Biden's age and mental acuity. The only thing he did was to confirm and amplify people's concerns exponentially.
Yeah, he said he was feeling awful. He probably should have postponed it, but then think of how the right would have spun that. He is damned if he does and damned if he doesn't.
Good thing Biden isn't senile.