That was some election!
This tweet caught my eye:
Texas has surpassed 2016 even if you account for population growth. (Many sunbelt states have seen 4% to 6% population growth since 2016.) In September, my Trump win prediction was based on three factors, each worth close to 3%; Electoral College bias, polls narrowing, and poll bias due to shy Trump voters. (Many Trumpistas are beta males, who are too timid to proudly state their beliefs in the way that you and I do.)
Obviously things are not looking good for my hypothesis. The polls actually got worse for Trump in October, before narrowing recently to about 8%. While the EC bias is still about 3%, Trump will need a 5% polling error to win. Even worse, many of the votes were cast when the polling margin was a bit higher.
Another problem for Trump is that we’re headed for a big turnout. That doesn’t necessarily favor the Dems, but it tends to. Hillary was clearly hurt by a low turnout of minority voters. For instance, Texas is typically a low turnout state. It’s hard for me to believe that Texas would still be strongly red if turnout were very high, given that only 41% of Texans are white non-Hispanic. (I emphasize “strongly”, as Texas Hispanics are less Democratic than in California.)
Another problem for Trump is that polls suggest that Dems are very enthusiastic about voting this year. And other problem is that polls show the Dems likely to do very well in House races, where the shy Trump voter bias is not much of an issue. So he has his work cut out for him in the next two days.
[And he’s being outspent, which seems impossible for a man who claims to be worth $10 billion. But . . . ]
Some people say “Don’t look at the national polls; look at the swing state polls. In my view, that’s only useful for ascertaining the EC bias. Once you’ve done that, the national polls are what you want to look at. Here’s why:
Let’s say the national polls are roughly correct, or at least within 3%. In that case Trump loses even if he wins Pennsylvania, and even though PA is the mostly like swing state. That’s because if the national polls are within 3%, then any extremely surprising result in one state (like PA) must be offset by a weaker Trump performance somewhere else (like Arizona or North Carolina.) If Biden wins the popular vote by 5%, he will win the election. If he wins by 4% he’ll probably win the election. If he wins by 3% then Trump will probably win the election.
Here’s another way to think about polling errors. Last time around, the national error was 2% and the polling error in midwestern swing states was 4% or more. But that means the polling error in places like Arizona was less than 2%. So unless the national polling error is really large, another big miss in the Midwest would imply that Biden is likely to win states like Arizona and North Carolina.
If we assume that Wisconsin and Michigan are out of reach, then Trump doesn’t need to just hold Pennsylvania; he needs to win all the states on the list above except Nevada.
[Technically speaking, Biden picking up WI, MI and AZ would yield a tie, but Biden’s also leading in Omaha. Believe it or not, Omaha might well be the “swing state” in an election where Trump wins PA and loses AZ. I only know one Omaha resident (Buffett) and I assume he’s voting for Biden.]
It’s probably pointless for me to predict the presidential election, given that I’ve lost touch with American politics. (I got them all correct from 1968-2012, before missing badly in 2016.) Instead I’ll just predict a very strong pro-Trump EC bias, even stronger than in 2016 (when it was about 2.9%.) Just to make things interesting, let’s say an EC bias of 3.5%. I base that prediction on the assumption that the Dems continue to gain ground in states that they would likely lose in a close election, such as Texas and Georgia. (They might win Georgia, but they’d lose it in a close election.)
Totally off topic, a week ago I forgot to link to this amusing NY magazine story about Jay Powell:
While Powell is a Republican, all three of these shifts are victories for the mostly left-leaning crew of Fed watchers who spent the past decade pushing the central bank to focus less on inflation and more on employment. This movement has included organized groups, like Fed Up, that exist to lobby the bank and make it more accountable to the public. It also includes academics and journalists who seemed a decade ago to be mostly a bunch of borderline cranks talking about nominal GDP targeting in op-eds and on Twitter. But you never know who you’ll reach on Twitter — and Powell, like Romney before him, has a secret Twitter account, and he pays attention to what is being said there.
The borderline crank comment probably doesn’t refer to me, as I’m definitely not “borderline”. Probably they were thinking of the more respectable member of the Mercatus team, David Beckworth. I’m an out and out crank, approaching MMT levels of weirdness.
Tags:
1. November 2020 at 19:28
Although trump is closing the gap, I still think biden has an advantage. A lot of the D R difference in the 2016 national tally is from California. Over the past few years the CA libs have registered over 1M democrats. I am not sure how predictive are national polls (in regard to electoral college outcome) when one state, CA, contributes so much to the D column.
1. November 2020 at 20:28
If Trump wins again, with even more evidence against him in the supposedly adjusted-from-2016 polls, we have to conclude that these polling techniques simply aren’t effective in the way they used to be.
Maybe that’s due to cell phones, or the internet, or something. But clearly they don’t capture what they purport to.
1. November 2020 at 20:58
I just looked up some numbers from 2016. NY and CA alone gave Hillary a net 6M votes. Of course , at a certain point the delta in national vote will transfer to electoral college victory, but, perhaps, the chance of EC victory compared to deficit in national tally is greater than some assume.
1. November 2020 at 21:25
Look at the ground tho:
https://twitter.com/KyleMartinsen_/status/1322603604616876033
https://twitter.com/RachelP07413367/status/1322539172067004417
https://twitter.com/Jamie43050057/status/1322972859069812744
https://twitter.com/DedrianaPerry/status/1322604060365783045
https://twitter.com/NYCdeb8tr/status/1322288067873484801
https://twitter.com/Shannon20421184/status/1322689029607952385
I still give Trump better odds, but honestly I have no idea.
1. November 2020 at 21:40
The more knowable question for me is, what’s your burner Twitter account, Scott? Lately you’ve been pointing to a lot of tweets and @MoneyIllusion doesn’t follow anyone.
1. November 2020 at 21:42
msgkings. Absolutely—the polls would be discredited.
Jg, Good point, but also keep in mind that NY and CA give Dems a lot of electoral votes. Biden doesn’t need to win 50% of the other 48 states, more like 35% or 40%. And if Biden wins by 5% nationally, he wins by about 8 million votes.
Cartesian, Yes, the turnout could help Trump, but I tend to discount these anecdotal stories right before an election. I’d say the odds are 60-40 that a big turnout helps the Dems. But I suspect that turnout will be big for both parties.
1. November 2020 at 21:43
Rajat, What is a burner twitter account?
1. November 2020 at 22:08
Scott, I think a burner Twitter account is meant in analogy to a ‘burner phone’.
Btw, interesting article in the NY magazine. But the headline is pretty insulting and/or parochial. Powell might be the best bureaucrat in the US, but he’s probably far from the best in the world.
1. November 2020 at 22:23
Why do the political parties run their own polling for internal use?
If official polls are unbiased (or merely believed to be unbiased), there would be no reason to reproduce their work.
It is possilble that internal polls are fabricated for the purpose of being leaked, but if that were true I would expect them to be flaunted more.
2. November 2020 at 00:13
Scott Sumner’s economic views are probably as sound as any, although the profession is a house built without keystones and with aging mortar, atop rotting balsa-wood stilts.
Sumner’s political observations might be cranky-crackpotty.
But again, who can tell anymore?
Biden calls for an “end to shareholder capitalism,” and no one bats an eye. Trump has an excruciatingly bad personality for a president.
Sane people suggest armed mobs are possible in the streets after the elections, while others note that four battleground states could go into protracted vote counts, re-counts and tenacious legal challenges.
The lawyers might outnumber the mobs, though with less street courage (beta males as lawyers?).
Good luck everybody.
2. November 2020 at 01:24
You are the definition of a beta male. Have you seen yourself on tape? You are a complete wreck, and total joke, so I don’t think you should be talking about beta males supporting Trump – since YOU ARE ONE.
What you failed to show in your suppression article – blinded by your bias – is that at this time last year Donald Trump was down 2 to 1 in early voting. In 2020, he is up in Michigan in early voting, and leading in Arizona and Nevada, which hasn’t happened to a conservative since 1988. He’s tied in Florida.
2. November 2020 at 01:43
Trump will win. He is not only leading in the early voting, but the enthusiasm is on his side. Trump supporters created a caravan in Arizona 96 miles long. In Pennsylvania he had 45000 people show up. In addition to that, one can look at the recent shift in news ratings. Tucker Carlson is averaging 5M viewers. Hannity 5M viewers. All of the top 5 are Fox News. This is a movement. It’s movement towards our principled values, away from the “woke” and “corrupt” who want to destroy capitalism in favor of some sort of totalitarian socialism. What’s remarkable, and historic, is that the force of personality of ONE MAN saved America and the free world from the brink of Klaus Schwab’s Great Reset. Unbelievable. He’s an American Hero. I suspect Biden will be charged eventually, within the next year or two, and at that point the democratic party will begin to collapse.
2. November 2020 at 02:00
Did you really call Trump supporters Beta? You do realize that his supporters are farmers and blue collar workers right? I mean, these guys did just force the Biden Bus off the road in Texas. And in terms of showing support, his crowd sizes are not small! Even Nuns are at his rally. I mean, no offense, but you don’t look like Arnold or Sylvester. You don’t have amazing looks and an extrovert personality 🙂 I wouldn’t exactly put you in the Alpha category of men 🙂 I think that is an extremely poor word choice and observation on your part.
2. November 2020 at 02:47
I think you’re wrong on the national vs state poll question.
> Let’s say the national polls are roughly correct, or at least within 3%. In that case Trump loses even if he wins Pennsylvania, and even though PA is the mostly like swing state.
Not necessarily. If the polls are wrong in favor of
* Biden in Carolina and New York
* Trump in Pennsylvania, Florida, and North Carolina
Then the national polls could be perfect, yet Trump could still win.
You might expect that this is unlikely because polling errors are likely going to be correlated across all states. I agree. I think looking at national polls is likely an okay way to assess probability for that reason. But that doesn’t make it better than state polls.
I think a good way to look at it is that national polling is worse than state polling insofar as the vote distribution has the property that moving votes around between states can change the result. In your example
> If Biden wins the popular vote by 5%, he will win the election. If he wins by 4% he’ll probably win the election. If he wins by 3% then Trump will probably win the election.
this is not the case (moving votes away from one state to another means both flip and we’re back to where we started), and that’s why looking at national polls works.
So insofar as you expect that the distribution does not have this property, national polls are a fine indicator. However, that’s not a reason to privilege them over state polls; at best they would work equally well. You haven’t given any argument why state polls are worse.
In fact, state polls just tell us directly what we want to know. We trivially can’t do better than directly estimating the result that will decide the election.
I guess one reason to privilege national polls is that they’re more accurate because sample sizes are larger. That’s true, but sample sizes of polls in Pennsylvania are so large that I’m pretty sure this factor is negligible. I haven’t done the math or anything, but I have some decent intuition about binomial distributions. Even if we just take all the good polls, their combined sample size should be so large that, assuming the polls are perfectly unbiased, the chance for the right value to be off by a relevant factor should be far below 0.1%. They’re not perfectly unbiased, of course, but errors from bias don’t depend on the sample size, so they’re going to be as strong in national polls as in state polls.
2. November 2020 at 02:52
Totally unrelated, I’m very worried about Trump declaring victory prematurely. I expect this to happen if the election is at all close (which it may well not be, but it’s possible). Pennsylvania has a law that mail-in ballots can’t be counted until election day, so early counts may consist primarily of in-person ballots. Even if Biden wins the state by 7 points, early counts could have Trump ahead by 10. He could come out, say he won the election, and frame counting mail-in ballots as Democrats trying to steal the election.
I don’t expect the state to stop counting votes if he does that, but it’s still a bad place to be in.
2. November 2020 at 03:50
If Trump voters are so shy, then why is Trump over performing so many incumbent Republican Senators?
It’s worth noting that district level polling indicate a Biden lead in the 10-12 pt range. Dismiss those polls as you will but a lot of party spending decisions are made off those polls.
Overall, the polling has been pretty stable this cycle with Biden having a consistent and sizable lead.
Anyway, I guess we will soon find out and this is a good reminder that today of all days is a good one to stay off social media.
2. November 2020 at 08:28
Some commenters have stated “Trump is ahead in early voting”—-I don’t think so—-unless they mean ahead of where he was in 2016—-but I don’t think that is true either.
GOP rooters are desperately clinging to ideas like a terminal patient seeking magical treatments in third world countries.
Maybe rust belt are those magical vote “on election day” republicans where Trump supposedly has a 20-30 point lead—-never heard of that before a few weeks ago.
I have read that in previous elections early voters were 25% of total vote. Had never heard that before either
All these things we have (or I have) never heard before are coming out of the woodwork.
I have to admit, those PA crowds did seem kind of large—particularly at the Pitts airport. Secret service said it was 57K (I think they always overstate). Local news said 10k—now that is clearly wrong. But 25k seems really plausible—and it was VERY cold.
It is true no one wants to see Biden. But where are the crowds against Trump? Sitting at home watching TV–they voted a week ago.
The other magic potion is “Trump ahead in Iowa”! by 7!!. but Iowa does not have “correlation juice” like PA if Nate is right—he likely is. Nebraska 2 which Trump won by 2% in ’16—has enormous correlation juice—if we KNEW that Trump would win N2 (it is the right way to Phrase 538’s model)—nis model would give hi a 30% chance versus 10.
More magic potions
2. November 2020 at 08:39
Why has PA only have 38% early vote? Doesn’t that bode well for Trump? That does sound right to me—if you believe “GOP votes more than Dems on election day”
Since they no longer have to prove mail in were post marked by election day—-against state law—-but who cares—and they do not need signature verification wouldn’t a crook wait until you need to know how many fake votes you need before mailing it in? If its that easy–won’t both parties cheat? of course the Dems control to much of state apparatus
If PA is close and it is the deciding state—–what a mess that will be
2. November 2020 at 11:06
I’m standing by my prediction:
Biden wins the popular vote with > 2% margin
Biden wins the EV unless mail votes are thrown out or the count stops via court order
2. November 2020 at 11:32
Well, social justice warriors and advocates for socialism and communism are more worthy of the title “beta” since most are degenerates that need the government to assist in every facet of their lives. They don’t have the intelligence to innovate, and they are jealous of those that do.
I’d normally vote Libertarian. But this year is too important. Don’t waste your vote.
Trump 2020.
2. November 2020 at 12:17
Since people are making predictions on here I’ll give mine for fun:
Biden 413 Electoral Votes, Trump 125 Electoral Votes.
Obviously I’m predicting Biden takes Texas. Another prediction that I have much more confidence in:
If Biden wins Texas, all of the sudden many Republicans will be open to repealing the Electoral College, and many Democrats will be more reluctant to repeal it.
2. November 2020 at 12:35
It is indeed incredible hard to see how Trump is going to win. The vast majority of facts seem to speak against him.
There are only very few straws left. Since I am a contrarian, I will try to list some.
The enthusiasm at Trump’s campaign events was again greater than that of his competitor. Like in 2016, Trump’s voters are electrified and loyal, the motivation of Biden voters often “just” seems to be that they don’t want Trump, similar to 2016.
From the perspective of his supporters, Trump has delivered. Scott and others may not see it that way, but that’s not really relevant, since they never even came close of being potential Trump voters.
My lucky guess from September that Pennsylvania is the key state seems to come true. The prediction tool from 538 is really great, if one sets that Trump will win Pennsylvania by a narrow margin, his chances of winning suddenly jump to 60% and more.
The dirties battles over absentee ballots might take place in this very state. The Supreme Courts are staffed and dinner is ready. Bon appetite.
2. November 2020 at 15:04
Xu, Jayne and tyrese, Take a deep breath everyone. It’s a JOKE. It’s Trumpistas who insist that Trump voters are too shy to admit their preferences. I’m just reporting their claims.
Of course I’m a beta male; almost all the best men are.
sty.silver, Don’t confuse what’s mathematically possible with what’s plausible. It’s not plausible that Trump wins in the EC if he loses by 5% in the popular vote.
Christian, You said:
“From the perspective of his supporters, Trump has delivered.”
I completely agree. They could care less that he didn’t reduce illegal immigration, bring back manufacturing jobs, reduce the trade deficit, build the wall, repeal Obamacare, etc. The elected him to be a troll, and he was.
2. November 2020 at 15:38
The people who looted and destroyed cities look to be near winning a Presidential election. The side that corrupted intelligence agencies and federal enforcement agencies while spreading disinformation through a corrupt press may soon control all three branches of government. Then they promise to stack the Supreme Court so that it will become their rubber stamp as they remake the country.
Group rights and the color of your skin will matter more than the content of your character. Once the ideal was a color-blind society, now it is statistical justice and group rights enforced by mobs.
Equality of opportunity is to be replaced with a redistribution based on what the squad and the mob thinks is fair. Your children will be re-educated to remove racists thinking, like showing up on time and working hard. “Subconscious” racism will be removed from their thinking and they will apologize for the color they were born. Incorrect thinking will no longer be tolerated.
If you think you had rights protected by the Constitution think again. The new regime thinks the Constitution is a racist document that needs to be destroyed. They will tell the Supreme Court what the “living” Constitution means under the new order. If you disagree the mob will remind you who is in charge.
Look at our major cities being boarded up. That has become the new normal. Democrats and the media are telling you that the only way Biden loses is if the system is corrupt, so prepare to riot. And then they tell you that Trump is a threat to Democracy. Each of those boarded-up windows tells you who the real threat to Democracy is.
Ronald Reagan once said:
Freedom is never more than one generation away from extinction. We didn’t pass it to our children in the bloodstream. It must be fought for, protected, and handed on for them to do the same, or one day we will spend our sunset years telling our children and our children’s children what it was once like in the United States where men were free.
We have lost an honest free press, we have tech billionaires who censor the truth, we had a President Obama who corrupted the levers of federal power against political opposition. This slide has not been sudden, it has been steady.
Biden is the frail, old frontman elected from his basement by billionaires who expect to cash in under the new order. Their children, like the Biden children, will prosper under the new order. Then they will hide behind walls that they will build.
Our cities will see crime escalate and violence becomes ever more common. Corrupt prosecutors already release the looters and arsonists without consequences. America may drift into armed camps.
Pelosi has already researched the removal of a President by Congress. Biden’s days are numbered. Soon the frail frontman will have served his purpose and he will be sent back to his basement with some ice cream. Then the new order begins.
https://spectator.us/american-media-failing-russiagate-obamagate-hunter-biden/?fbclid=IwAR1oNOw1yvb9jTq-m2I_JVKMLi3EnoDYy_orIJYar7YEERWNV1DXhx4G784
2. November 2020 at 22:58
LB,
I mostly share your fears. But I honestly don’t know if the outcome you describe is more likely under a Trump victory or a Biden one. None of these problems go away if Trump wins, in fact he may be a useful catalyst for the solidification of the cathedral. Trump supporters are explicitly regarded as idiot peasants by today’s elites. They need a much better leader who’s simply way more impressive than Trump is. In four years, it’s almost impossible to imagine that happening if Trump wins. It’ll be a socialist revolution.
(Off topic) To me the best hope moving forward is something like Unity2024. Yes, this is a bit out there. But imagine some kind of dual ticket, for example Yang+Amash running together. We need some way to work through our differences towards solutions.
3. November 2020 at 04:29
Today the 538 closed with their prediction of Trump winning as 10%. I thought it was way low so I decided to go to search for some betting sites to put some money on my hunch that Trump’s chances are way higher.
As it is the implied probability of Trump winning is around 40% on around 4 betting sites I checked and actually slowly creeping higher. Which is about where I was seeing it so in the end I decided against the bet.
This may be much closer election that most see it. Just one of the things that people underestimate: Democrats have almost 50% mail ballot registrations compared to Republicans. However mail votes have notoriously higher rate of being invalid – we are talking 2-5 percentage points with the higher estimate for people/voting districts where people have little prior experience with it. This by itself can swing some states.
3. November 2020 at 10:06
LB, You said:
“The people who looted and destroyed cities look to be near winning a Presidential election.”
When comments start this way, I stop reading after the first sentence.
Cartesian, You said:
“To me the best hope moving forward is something like Unity2024. Yes, this is a bit out there. But imagine some kind of dual ticket, for example Yang+Amash running together. We need some way to work through our differences towards solutions.”
That sort of utopian solution is blocked by our two party system. We need proportional representation.
JV. Maybe. I’ve given up predicting US politics.
3. November 2020 at 13:18
Scott’s recent book review was interesting and thoughtful if not fully convincing. It is hard to imagine he wrote it.
Scott seems to have no understanding or insight into how people respond to incentives. People and their motivations seem to be a mystery to him. Just get monetary polict right and individual liberty becomes trivial. You must laugh at Milton Friedman for arguing that economic freedom and individual liberty go together – even if you have a good monetary policy.
You clearly consider some 40-50 million Americans deplorable and beneath you.
Senator Harris this weekend said that our system must guarantee equality of outcome. Vice President Biden argues that he can just make an executive order on masks and police forces around the country must enforce his will, Our future is bright with these two in office. (That is sarcasm Scott.)
Democrats will now take steps to destroy the Republican Party. They will do their best to disenfranchise perhaps 50 million people. Scott will cheer them on as they remake the “living” Constitution” into something new and stange. Stacking the court to assure that nothing stands in their way.
What did the recent Hoover study say? $12,000 increase for the average car. Loss of income of $6,500 on average. Less efficient more bureaucratic economy. Every job Biden creates with his new energy policies will cost 1.1 jobs in other sectors. Now that is a plan Scott can support.
Scott supports the death cult called abortion. My daughter was born at 8 months. She is a wonderful gifted human being. The abortion crowd say she was just a collection of cells that could have had her spinal cord cut to kill her – she wasn’t really human. Perhaps we should begin to do that with all those the left considers defective or in the way.
Europe is having a major outbreak of Covid. Interesting how that will become Trump’s fault. Or every country that has an outbreak is just run by people who are stupider than Scott. Of course that is true, just ask him. Why didn’t those leaders just ask Biden for his secret miracle cure? I guess those leaders want to kill people. America had bent the curve until the riots started and new cases erupted again. That was Trump’s fault too I guess. 290,000 people died of the flu when Biden was vice-president, why didn’t to stop that?
I suggest you read Charles Lipson’s article on the failure of media the last four years. Nevermind, you don’t read what you don’t agree with. Explains your gaping lack of understanding on many issues.
I’m done here. Trump was flawed. His trade policy seemed to come from the Democrats of the 50’s-70’s. His foreign policy was isolationists which the revolving door between government and the defense industry couldn’t stand. The racist attacks against him were laughable but sad. He talked and tweeted often without thinking. Some prefer Obama who rarely said anything that wasn’t carefully plotted. Harris who is the most amoral politician I have ever seen. Or Joe Biden who has a hard time remembering the talking points his staff gives him.
Pray that moderate Democrats, the handful that still exist, can slow down the Social Democrats. Miracles might still happen.
But as you look at all the plywood on businesses around the country, ask who sent the mobs into the street. The mob has won. Perhaps we all move into armed camps now.
3. November 2020 at 16:18
“Bush dropped more bombs than Clinton.
Obama dropped more bombs than Bush.
Trump dropped more bombs than Obama.
This trend has much less to do with who’s president than with the nature of global metastatic imperialism.
And it ain’t changing with Biden.”
https://twitter.com/caitoz/status/1323687673521827841?s=20