Some of you may have noticed that Nate Silver has become an insufferable bad-hot-take machine, obsessed with driving clicks by talking about the hour-by-hour minutiae of his team’s forecast models. We now have published scholarly evidence that this may be bad for democracy.
In a very robust-sounding paper*, which I admittedly have not read in its entirety, researchers suggest that:
an increase of 20% over even odds in this study lowered voting by 3.4%, and an advantage of 40% [around what Clinton had in the 538 2016 forecast] lowered the voting by 6.9%. If as the evidence provided above suggests, Democrats were more affected [than Republicans]…, probabilistic forecasts may have a strong enough effect on turnout to constitute an important factor influencing the election.
Similar to the observer effect in physics, where you change the result by measuring it, the widespread consumption of these forecasts may well undermine the very things they seek to report. Importantly, this effect was not seen in research subjects who were only shown poll results.
Here’s a good thread on the paper!
Unlike polls that show candidates’ expected vote share, prob. election forecasts convey the estimated probability that a candidate will win. Problem: folks donât understand probabilities. This paper demonstrates severity of this confusion, and its political consequences. (2/n) pic.twitter.com/MEt9GoUhZs
â Jonathan Mummolo (@jonmummolo) February 28, 2020
When a draft of the paper was published in 2018, Nate Silver dedicated an entire episode of his podcast to “rebutting” it. You can see the rebuttal of the rebuttal here. I’m inclined to believe the authors, for Nate Silver has become a terrible hack. Apparently he was more recently urging scholarly journals not to publish this piece, which is super classy.
Open thread!
*This link is to last year’s draft of the paper; the paper itself was published this month, and is behind a paywall.
James E Powell
As Scott Lemieux at LG&M has argued since that dark November of America’s soul, the belief that Hillary Clinton was going to coast to victory was a major factor in her loss.
Betty Cracker
Fascinating. I’ll admit I look at the polls, but I try not to let the probability forecasts influence my thinking, and this is all the more reason to ignore them completely. Thank you!
Baud
So proof that people stayed home because they thought Hillary was going to win anyway?
New Deal democrat
This is what is known as a âsecond order chaotic system,â where the subject under observation has the ability to observe right back, and alter its behavior as a result. All forecasts of human behavior, e.g., stock market and economic forecasts, are plagued by this paradox.
When Nate Silver was a small fish at Daily Kos, he was more accurate than he has been since he became widely known.
By contrast, my economic forecasting record is impeccable. Nobody pays attention to me. Bastards! ( /s)
Major Major Major Major
@Baud: social science experiments do not provide âproofâ, but this paper seems to have very robust experimental evidence, both natural and laboratory.
feebog
If you did a “man on the street” show and asked who Nate Silver was, about one in a thousand would be able to tell you.
Baud
@Major Major Major Major: Understood.
This paper is also
proofsupport from something I and others have been saying all along — liberals, progressives, the left, etc. need to push the idea that voting is a civic duty and not means of self-expression or self-actualization. If people are committed to voting regardless of whether they are personally inspired by particular candidates, savvy poll predictions will have less of an impact on the result.MattF
Itâs too bad. Statistics is trickyâ and Silver knows that.
Major Major Major Major
This was maybe my favorite part.
Butch
@New Deal democrat: Starting with the caveat that I’m not a fan of Silver’s but a small point; I thought he actually started at the NYT mostly working with baseball statistics, and not at Kos?
New Deal democrat
They donât need to know Nate Silver. They just need to know that âall the polls say Hillary is going to win.â
Jim, Foolish Literalist
@Baud: anecdotally, a lot of people said they voted for Stein or Johnson or wrote in Dumbledore to “send a message”. I’ve never seen any data on that, but the margins in PA, WI and MI were so small, who knows…
As I recall, in the last couple weeks of the ’16 campaign, it was Silver’s pessimism vs Sam Wang’s confidence about Clinton. I remember mocking Silver myself for factoring in mood or momentum or something else that struck me as equally un-scientific. I trusted the quants.
New Deal democrat
@Butch: Correct, but of course as Publius at Kos, he was anonymous and didnât have measurable impact.
Rommie
“I’m gonna be edgy and vote for Brexit, because I know it’s going to lose anyway!” /surprised_Pikachu
Felanius Kootea
Completely off topic; I received the following today about Nigeria’s first case of the coronavirus:
I hope that the resources that Nigeria used to defeat Ebola are still in place – that was one of the most impressive feats of contact tracing, and helped prevent disease spread beyond the hospital staff that treated the victim.
randy khan
@Butch:
He actually started doing baseball before he did politics. He designed the original PECOTA, a tool for projecting player performance, which is used at Baseball Prospectus. (It turned out to be not as predictive as they hoped, but they still use a variation of it today.) He started doing politics at dKos, and then created his own site.
Felanius Kootea
Completely off topic: Nigeria just identified its first case of coronavirus.
I hope the resources that were in place for the Ebola crisis are still intact; Nigeria had one of the most impressive contact tracing programs, which confined disease spread mainly to hospital workers who treated the victim.
LongHairedWeirdo
Um. Point of note: in physics, the reason “the act of observing affects the observed” is because, essentially, that rule holds when measuring things that are just barely “there”.
For example, electrons appear as a wave or a particle – they can (or “usually do?”) move at relativistic speeds, and anything – *ANYTHING* – you do to measure them requires some form of physical affect. An electron is *so* tiny, and has *such* tiny mass, that it’s *impossible* to measure an effect without that effect having changed the electron.
For example, if you watch for a magnetic or electric field change, the field has affected the movement of the electron. If you bounce a photon off it, it’s like trying to measure the speed of a car by throwing a bowling ball (or, possibly something bigger – I honestly *do not* know the scale, though I’ll bet Randall Munroe could spell it out) at it, and measuring how the bowling ball moves – once you try to account for what happens to the car, because someone just fired a high speed bowling ball at it. Obviously, if you let the electron impact on something, you know damn well you’ve altered it.
I’m not criticizing – the idea of “observation affects the observed” *is* a handy meme. It’s like the boiling frog – a frog in a pan of water, being slowly heated up, *will* try to escape before the water is dangerously hot for the frog. It doesn’t *exist*, but it shows a fundamental truth: it’s easy to ignore slow changes, because by the time the entire GOP is *proud* of having covered up lawlessness, it’s *really* late in the game. The time to have started fighting back against the hate was probably ’94, or ’96.
Now, you literally – *literally*, and I don’t mean “emphatically” – have people thinking that Dems are baby killers who need to be opposed to save the BAAAYBEEES!!!!!1!!!!!!!!11!!!!eleventyone!!!!!
You have people *literally* thinking 20 judges will ignore a woman’s basic civil rights, and “murder” her by following her clearly stated wishes about end of life.
You have people *literally* thinking it’s not a lie to say “The UK thinks something,” without adding “but we already know it’s BS”, if we’ve already proven it’s BS, and the administration has corrected 3 earlier speeches, to remove that BS claim. (I was going to say “they had to correct” but, face it, the Bushies didn’t “have” to do anything, they could have shot someone on fifth avenue and just said IT WAS A TERRIST!! and gotten full wingnut cover.
You have people *literally* thinking Trump is law abiding and Obama was lawless.
And you have people *literally* thinking that’s just fine and dandy, if demanding an occasional furrowed brow, and softball statement, because OMG they might lose their legislative job, and have to go on and become a highly paid lobbyist, YOU CAN’T EXPECT PEOPLE TO DO THE RIGHT AND DECENT THING IF THEY MIGHT FAIL TO WIN REELECTION, THAT’S CRAZY TALK!!!!1!11!!!!!!!(etc.)
This early morning rant was brought to you by someone who actually thinks facts matter.
J R in WV
I have never liked Nate, nor his hot-dogging statistics work. I think he got his start analyzing baseball stats, and then decided numbers are numbers, why not get into political stats where the real money is. Baseball is not politics, and while numbers are numbers, people are not numbers…
Politics is not baseball, those numbers are fundamentally derived in different ways and have different meanings! Beliefs change political numbers, but believing you are gonna hit that ball won’t change your batting stats.
MattF
@Butch: Wikipedia has the story. Silver started in sports statistics, then added politics. NYT published his politics blog for a while.
JMG
Silver is bitching loudly about this paper on Twitter today.
randy khan
I have spent a surprising percentage of my work life worrying about how to express probabilities to people so that they understand them. It’s really hard. There’s research that says that people are strongly inclined to hear only three things – very unlikely, 50-50 chance, and very likely – no matter what you say. I try to use explanations like “this probably will happen, but there’s a decent chance that that will happen instead,” but I’m not convinced that works any better. So this research makes a lot of sense to me.
Silver actually probably was one of the more conservative predictors in 2016. His final projection gave Trump about a 1/3 chance of winning, which definitely was closer to right than anybody else, and probably more significantly, he more or less identified the most likely path for Trump to win. But I agree he’s gotten more click-baity since he left the NY Times.
Goku (aka Amerikan Baka)
Because nothing says integrity like trying to suppress contradictory evidence! You’d think integrity would be a good thing to have in Silver’s field. Will no longer be looking at 538
randy khan
@J R in WV:
There actually is way more money in baseball than politics, particularly as teams have adopted analytics in their player evaluations and have begun to hire analytics people.
LongHairedWeirdo
@Jim, Foolish Literalist: Respectfully, on the day of the election, Silver called the race as 2-in-3, Clinton wins, 1 in 3, Trump wins.
Any gamer can see “that’s a 5 or 6 on a standard d6 (or six sided die; gamers may use “polyhedral” dice, with 4, 6, 8, 12, or 20 sides, or irregular shapes with 10 and 30 sides (and IIRC, someone invented a “golf ball” with 100 sides, all numbered, which shows some people have a lot of time on their hands). And every gamer knows that 2-in-3 is *pretty* good, but *way* uncertain.
Most other predictions showed a much higher probability of a Clinton win, and thus, Silver’s was *probably* the most accurate.
The complaint above isn’t about accuracy; it’s about how those top line numbers affected the electorate. I haven’t followed Silver’s coverage of the primary, so I can’t say he has, or hasn’t, become a click-whore. I’ll still generally trust his accuracy, so long as he’s reminding readers, right now, that “of course, anything can happen, only a few states have voted, and these predictions *will* change as time moves on.”
ETA: I do not consider “a lot of time on their hands” to be anything other than playful – whoever invented the d100 was obviously clever! But gamers had long used two 10 sided dice (or two 20 sided, each of which was numbered with 0 to 9) to generate random numbers from 1 to 100 (or 0 to 99, if you insist).
Eunicecycle
When the Cleveland Indians were up 3 games to 1 in the World Series in 2016, I think it was Nate that said that Donald Trump had the same chance of winning as the Chicago Cubs did. Gave me a chill at the time. Still does.
MattF
@randy khan: Thereâs all sorts of tricky thingsâ For example, astronomers have to be trained to distinguish clumpiness in random processes from clumpiness thatâs not random, or lack of clumpiness thatâs not random. You donât make the distinction by looking at the dataâ you do a calculation.
Hoodie
@Baud: Civic duty or just plain interests as opposed to “principles.” One thing the right has done with thing like Fox News is conflate politics with ideology, causing a lot of folks to internalize a notion that it’s wrong to have interests. I think this has accelerated the nationalization of elections and diminished the effectiveness of Congress.  This all took off with the bullshit about earmarks and other traditional forms of legislative horse trading.
Jim, Foolish Literalist
Open thread? I see StockMarketCrash2020 trending on twitter. I wonder what effect social media will have on the Panic. Twitter was still “that weird micro-blogging thing” in 07-08, wasn’t it?
Major Major Major Major
@New Deal democrat:
The experiments found that only probabilistic forecasts decreased voting behavior, not polls.
dnfree
@James E Powell: I think this was especially true at the state level. All the talk of âsafeâ Democratic states not only led to people staying home; it led to them feeling confident they could cast a protest vote for a third-party candidate without affecting the electoral college outcome. Imagine their surprise in Wisconsin and Michigan at least. I know people who did that in Illinois.
Major Major Major Major
@LongHairedWeirdo: itâs an analogy.
Baud
@Major Major Major Major: I missed that nuance. Interesting.
pamelabrown53
@New Deal democrat:
I thought Nate Silver was Poblano at Kos, not Publius?!
Baud
I hate our savviness.
Suzanne
@Baud:
I agree… and yet I fear that this message will be a huge depress-er of turnout. Like it or not, many people are self-interested. Jury duty is a civic duty, too, and that hasnât gotten more popular.
Nate Silver is insufferable. I like Clare Malone and Micah Cohen at FiveThirtyEight, though. I do think they make a good contribution to the discourse, and I do think that more journalists would do better if they were more data-driven. Like, we had Cletus Safaris coming out of our asses, and I have serious doubts that any of them told us anything useful.
Jim, Foolish Literalist
@Baud: Republicans, as a whole, rarely lose sight of the prize. I remember current never-trumper Mona Charen addressing some conservative group in 92 saying her slogan for the Bush campaign was “Only Four More Years”
though I guess that was the one time in my political adulthood that purity politics (from David Souter to Pat Buchanan) did wound them
VeniceRiley
Funny. I just saw a California poll putting Bernie far ahead and Warren in 2nd. And I made up my mind then and there to vote Warren because defeating Bernie is my #1 issue in the primary.
smintheus
@New Deal democrat: Silver has always been prone to grandstanding and making grandiose pronouncements about politics that don’t stand up to much scrutiny. His expertise is in manipulating statistics, but he has for years worked under the delusion that he has expertise in politics. Ironically, that delusion is so strong it has repeatedly caused him to massage what the polls seem to forecast, because he imagines that his political spidey-sense tells him we need to zag when the numbers are saying to zig. A clear example has been his terrible record with regard to British elections (where he understands even less about politics).
Major Major Major Major
@Suzanne: the problem with âjournalists being data-drivenâ is that most journalists are innumerate and just parrot whatever data is handed to them. Same with most non-specialty-outlet science and tech âjournalismâ.
hitchhiker
My recollection is that Nate Silver correctly identified the unusually high number of “undecideds” before the 2016 election as a critical factor.
The fact that most of them chose “against Hillary” at the last minute was the surprise. I’m looking at undecideds in SC and it’s 15% — election tomorrow and 15% of voters haven’t decided what they’re going to do.
That’s a huge number. What the hell does it mean that they haven’t decided yet? Could be anything. Hoping for clarity from the debate. Not paying attention. Waiting for a sign from their morning tea leaves. Torn between two options. Leaning toward X but open to Y or Z. Need to check in with their aunt who’s really smart.
My own prediction, (based on nothing), is that the virus is going to scare people who might have been feeling like taking a chance not to take that chance. I think in this week’s chaos there will be a sense that we need someone familiar and steady. I think Biden’s numbers will be high.
Martin
In his defense, but not really in his defense, because the paper in question also misses the point.
The problem with the polls is that people interpret the wrong polls. Hillary had a 3-4 point lead in the national polls going into the election.
135 million votes were cast. Hillary ‘won’ by 2.1% in the popular vote. She won California by 4.2 million votes, which is 3.1% of the total votes cast. So, if we subtract out the 4.2 million unnecessary votes for Clinton to win California, she’s now down 1% in the popular vote for the other 49 states, and her 3-4 point lead in the polls is erased. She won New York by 1.7 million votes. That’s another 1.3% of the total votes cast. Subtracting out those unnecessary votes, she’s now down by 2.3% in the popular vote in the remaining 48 and down in the national poll.
What about Trumps excess votes? His largest surplus was Texas – 800,000 votes. 0.6%. Tennessee is next, 600K. 0.4%.
This is how gerrymandering functions. Small margins in a large number of districts, and losing in blowouts in a small number of districts.
The horserace at the very least tells us that the popular vote runups in a handful of states won’t lead to an electoral victory. So while I agree that the probabilistic model is bad for us, they mainly exist to correct for the even worse national polls. If you want to burn down Nate, burn down the national polls first, which Nate is trying to compensate for.
And my warning for continuing to use the national polls are in the head-to-heads. Bernie may trot out a 5-6 point advantage over Trump, but that’s only if youth voting turnout is quite strong. But the states that put Trump over the top aren’t  young voter states. California and New York are  the  young voter states, which is why Trump lost so badly there. Bernie just gives us a 6 million vote margin in CA and a 2.5 million vote margin in New York. So what? We already have those states. What do the head to heads look like in MI, PA, WI? I bet they don’t look so good for Bernie. I bet they look way better for Biden.
Suzanne
@Major Major Major Major: That’s true. But then the answer has to be to make them smarter, not advise them to be dumber.
LongHairedWeirdo
@Major Major Major Major: Agreed! As I said, I just wanted to explain a bit about the physics, because
1) iIve heard too many people who were 100% certain that looking at a car was the same as looking at an electron, and
2) it took me a long time to understand the concept, and I was tickled pink when I realized I’d correctly sussed it out.
Ain’t no one as ready to talk, as someone who could have been tripped up earlier, and learned differently – it’s like the old rule “never light up around an ex-smoker; no one else can quote how horrible smoking is than they.” (Not *all* ex-smokers, of course.)
Martin
@pamelabrown53: He was poblano.
Orange Is The New White
@New Deal democrat: I read your site close to every day. And your posts at Seeking Alpha.
Back when you and Bonddad were at DKos you called the coming crash in 08. And got drummed out of the community, with staggering viciousness, for doing so. It’s the reason I left and ended up here, I don’t like it when people who know what they’re talking about get silenced because what they’re saying is inconvenient for some folks’ agenda.
You guys saved my 401k, because I put it all into cash based on what you were reporting, the market tanked, and I lost nothing. So, thank you.
Major Major Major Major
@Martin: how did the paper miss the point? It had a question and a hypothesis and studied them.
Yutsano
@JMG: Methinks the Silver doth protest too much…
New Deal democrat
@Major Major Major Major: that doesnât refute my point. âPoll x says Hillary is going to winâ probably wonât affect behavior. âAll the polls say Hillary is going to winâ conveys much more certitude. And we donât have to overlook the evidence in front of our own eyes that both Comey and Obama took (or in the second case, failed to take) action because âeveryone knows Hillary is going to win.â
Thatâs a second order chaotic system.
smintheus
@Major Major Major Major: Even simple polling has had this effect in the past – when that polling and what the news media was saying about the polling seemed to overwhelmingly and consistently favor one candidate. It was said in 1948 that Truman may have benefited from the over confidence of Republican voters, and the willingness of some people in the middle to cast a vote for Truman because they were sure he was going to lose anyway.
Seems entirely reasonable to me that the public has invested way too much of themselves in poll aggregation. I know peope who understand little about politics who talk confidently about where things stand in the horserace by citing their favorite aggregator.
The other factor which the aggregators mostly refuse to deal with and that almost nobody else is talking about, is that poll aggregators are easily and now fairly consistently being manipulated by partisans. There’s an increasing number of polls with a strong partisan bias that seem to exist exclusive to push the aggregators in the preferred direction. Silver and others will tell you they can control for that by applying algorhithms to different polls results, but the reality is that doesn’t really control manipulation – it just tries to minimize the bad effect it has upon models. The correct approach would be to use only the best and most reliably unbiased pollsters, and exclude the evidence you know to be sketchy. The reason the aggregators don’t do that, however, is that to admit that a lot of polls are garbage would be to sow seeds of doubt in the minds of their audience. It’s the last thing they want because it’s bad for business. The aggregators want to persuade people to keep coming back obsessively to check on their latests horserace numbers. So they pretend that nearly all polls have something valuable to contribute to our knowledge of where races stand.
The wider problem is that elections should not be treated as horseraces, because inevitably that distracts from issues and personalities.
piratedan
@Major Major Major Major: data without context are just numbers sans meaning. That is apparently my biggest beef with these guys is that they are constantly in Veccini mode… these numbers do not mean what you think they mean.
MP
@New Deal democrat: Hey, I do! Whenever I get nervous about the markets and the economy, I look to you and Bill McBride for a cogent, systematic analysis of the data. And then I turn on CNBC and watch Cramer and it all goes out the window.
joel hanes
@Suzanne:
There’s some evidence that an effective framing is “you’re a good person if you vote, and should be proud of being a voter”.
Framing it as an obligation seems to me less likely to inspire infrequent voters.
New Deal democrat
@Orange Is The New White: Thanks. I do have a few influential followers in the financial community now, so my record will probably start going to hell….
Major Major Major Major
@New Deal democrat: havenât read the paper but based on what Iâve read *about* the paper I wouldnât be surprised if it addresses this. But my short response would be that âeverybody knows Hillary will winâ fits nicely into the âpredictionâ category, not the âpollâ category.
janesays
Silver is the person who is most responsible for making election forecasting a thing (namely because of his near perfect predictions in 2008 and 2012), but in fairness to him, he was actually the most (small “c”) conservative of all the most well-known forecasters in terms of the odds he was giving Clinton on the eve of the election in 2016. He had he with a 71% chance of victory, and I remember a lot of people being irritated that he had her odds that low. The New York Times had her at 85% chance of victory, Slate had her at 90%, Daily Kos had her at 92%, Sam Wang had her at 93%, and Huffington Post had her at 98%. They were all obviously terribly wrong, but FiveThirtyEight was the least wrong of the major forecasters. Where Silver holds thew bulk of the blame is in the fact that he is the one who created this whole movement in the first place. Election polls have existed for many decades, but the business of analyzing them in microscopic detail to predict the odds of particular outcomes is a very new thing.
joel hanes
@MP:
I turn on CNBC and watch Cramer
You might as well hit yourself in the face with a hammer.
CNBC has been pushing the line that the Democrats are ginning up SARS-CoV-2 hysteria as a weapon against Trump, and that the threat of pandemic is manufactured.
MP
@joel hanes:
You might as well hit yourself in the face with a hammer.
It’s a little unnerving that you’re familiar with my Cramer-watching ritual. I’m gonna have to get rid of that Ring camera.
Goku (aka Amerikan Baka)
@joel hanes:
How can they possibly say that? It’s in 40+ countries right now. Are all them in on the plot too? The WHO? What makes them think it’s not already here silently spreading? A person in NorCal tested positive and they don’t know how they got it
This kind of talk is what’s going to kill people and CNBC should STFU
cokane
Is “apparently” the way you can get away with writing things about people that you cannot back up with evidence? I mean if you’re going to write a blog post slamming Silver, get the links for your claims! Or leave out the claims that you can’t substantiate. I swear, this blog did not use to be a haven for weakly sourced theories like it’s becoming these past couple of years.
I’m not sure your post makes a lot of sense really. I don’t doubt there’s some, marginal effect of these forecasts. But they literally did not exist in the past. But more importantly these analyses have been helpful for targeting states and certain Congress races in a more efficient way. So I’m not sure you can say that the net result of Silver and his copycats has been negative.
Also, turnout is higher than it’s ever been in recent elections! Obama’s elections. 2018, one of the highest turnout midterms.
janesays
@Baud: Absolutely correct.
I am convinced to the core of my being that if America had been given 48 hours notice that there would be a do over election on November 10th, Clinton wins every state she won the first time plus Michigan, Pennsylvania, Wisconsin, and Florida (she was within 1.2% in each of these states). I think there were enough people who may not have been her biggest fans but nonetheless preferred her over Trump who stayed home in those states because they were convinced she had it in the bag to change the outcome of the election.
cokane
@janesays: For sure, there was some perfect storm kind of timing with how the last two weeks or so of the election played out. And we can see in all the three key states, that turnout was significantly lower in the big cities than in 2012.
Another Scott
Sean posted the paper at Dartmouth (37 page .pdf, dated February 27, 2020).
Cheers,
Scott.
Brachiator
Yawn. Non story. I would only care if Silver and other pollsters were lying or fudging results. Otherwise, what are you going to do, ban information and discussion?
People stupidly want to make polls into oracles. Not much to be done about this kind of nonsense.
Next, people will be grumbling about uninformed speculation on blog sites influencing results.
trollhattan
@JMG:
Pretty sure his continued income requires it. Zut alors!
If y’all had, like me, forgotten about Bad Lip Reading, resolve the error of your ways by viewing this BLR State of the Union, also featuring Nancy SMASH and Mister Mother.
Do it now, do it for Friday, do it for America.
trollhattan
@Goku (aka Amerikan Baka):
China blew their chance for containment by mis-assuming it was not transmissible prior to symptoms presenting. We’d do well to learn from that.
I crack myself up.
Martin
@Major Major Major Major: Because the paper acknowledges the dynamic I’m referencing, but doesn’t address it.
But national polling has the same effect, at least it did in 2016. Nate was one of the people pointing out that Trump had more paths to victory than one might assume from the national polling data. Yes, Nate was calling for a Clinton victory, and yes, I acknowledge the paper’s conclusions on turnout, but the national polls were also giving an inflated confidence in a Clinton victory – especially their treatment of undecided voters.
Now, the probabilistic forecasts certainly magnified the scale of the result which voters confuse for vote margin, and therefore affect turnout, but the turnout would have been affected in a similar but smaller way simply because the national polling which is constantly reported is bullshit and completely immaterial to how we run elections.
So, again, I’m not defending Nate, but removing Nate still leaves us with a big problem. And the paper doesn’t address that problem. Democrats still would have gone into 2016 (less) confident in a victory, Â but confident nevertheless, mainly because the national polls were lying to them.
Baud
@Brachiator:
We can’t educate ourselves about the things that influence behavior of banning is off the table?
janesays
@New Deal democrat: Silver’s most accurate prediction was actually in 2012 – he called every single state that year. He was already pretty big because of 2008, but 2012 solidified his status as the gold standard of poll forecasting. FiveThirtyEight was still pretty accurate in 2014, but it was 2016 when everything went to shit. His 2018 predictions were decent – he slightly underestimated the Democrats performance in the House, and slightly overestimated it in the Senate, but he was correct in predicting that the House would go blue and the Senate would stay red.
Darkrose
@randy khan: I find it very amusing that the team most known for their use of analytics over the past 5-10 years has been the Astros. Turns out analytics are less effective than using the replay cameras and banging on a trash can with a baseball bat.
Major Major Major Major
Youâre going in the pie filter (couldâve sworn you were already there) since youâre âapparentlyâ incapable of recognizing a link.
Major Major Major Major
@Brachiator: it is a huge and disingenuous leap to say that media criticism amounts to suggesting we ban the media.
Captain C
I don’t think the message they sent was the one that they thought they intended.
Soprano2
@Martin: Yes, yes, yes! I keep telling the Bernie fans this over and over – it doesn’t matter how much he wins CA by, or how popular he is there – what matters are MI, WI, PA, MN, NC, NH, OH, FL, and AZ. How does he poll in those states? I bet not nearly as well as he does in the meaningless national polls. National polls are meaningless because of the EC – they should quit doing them.
smintheus
@janesays: Part of Silver’s claims to accuracy stem from the fact that he doesn’t make a single prediction. He always gives several competing models, without choosing one. Then his admirers get to refer to the one that was closest to the results, ignoring the others. In elections where all his models are laughably bad, especially for British elections, those just go down the memory hole.
Hoodie
@Martin: Begs the question as to why to do national polling at all in a race that is decided state-by-state. It’s a bad first-order approximation, at best. It would be better if he generated an electoral college vote prediction.
Brachiator
@Baud:
There is usually a robust discussion of polling here, along with people incorrectly invoking polls, or being cautioned about polls. And some folks nailing wonderfully insightful points about polls.
Seems to me this is what you want. And if this article points people in a useful direction, all the better.
New Deal democrat
@janesays: Remember, the humans are also observing back. They learned from 2016, and altered their behavior in 2017, 2018, and 2019.
janesays
So, you were predicting the key economic indicators in Napoleorange’s 4th year would be significantly better than what almost everybody else was predicting at the start of his presidency?
If you were to replace Trump with generic, bland President X but kept the same economic indicators as Trump has in his fourth year in office (setting aside this week’s stock market plummet), you would expect that person to coast to easy re-election with near landslide numbers. If we do wind up taking Trump out in November, we’ll have done so during the strongest economy in which any incumbent president has ever lost re-election.
Barney
@Betty Cracker: It seems to me this is not “all the more reason to ignore them completely”; it’s a reason to not decide whether to vote based on a prediction on who will win. So you were right before. The paper’s authors say “if what you care about is conveying an accurate sense of whether one candidate will win, probabilistic forecasts do this slightly better than vote share”, and “our headline isnât about accuracy, itâs about confusion. And the evidence from this research and past work taken as a whole suggests that probabilistic forecasts confuse peopleâââsomething that came up at the end of segmentâââeven if the result sometimes is technically higher accuracy.”
The problem is not Nate Silver; it’s many people, given the information he provided.
L85NJGT
@Darkrose:
The first order effect was teaching their pitchers how to load a baseball with sticky stuff.
Repatriated
@Martin:
The difference is in inumaracy.
A 3% poll difference looks like every vote counts.
A 40% difference in probability (after aggregation) looks inevitable so individual votes appear to matter less.
Same underlying reality. Different descriptions. Different reactions.
janesays
@smintheus: His most well-known forecast is his U.S. presidential forecast, where he not only gives each candidate a specific percentage of their odds of victory, but he also makes a prediction about how every single state will vote in the presidential election. He reached his pinnacle in this model in the 2012 presidential race, when he accurately predicted the winner of every single state in the nation.
Captain C
Followed by “Why do I have to wait in line at EU customs?!? I thought Brexit would just keep the [derogatory word for non-English people, usually of color] out, not have any material effect on my life!”
Martin
So, I’ll restate what I said in a previous thread that I suspect there are hundreds of cases of Covid in CA right now. There is a suspected case with a UC Davis student. The student and two roommates are under quarantine. They can’t test the student because the student didn’t visit China, per CDC rules.
Gavin Newsoms presser yesterday had real vibes of ‘we have a lot of circumstantial evidence that we are fucked, but are prevented from getting real answers’.
Understand that a LOT of people travel between California and China. Apple alone flies 50-100 people a day there. I have coworkers that go there at least monthly. We have very large Chinese communities here and very strong ties to Chinese businesses and universities. Surely a few people brought the virus into the state undetected.
Yutsano
@piratedan: INCONCEIVABLE!!!
Poor Vizzini. Done in by his arrogance. And by a fundamental misunderstanding of his opponent.
janesays
@New Deal democrat: He was “poblano” at DKos, if I recall correctly. Not sure who Publius is.
Hoodie
@Martin: A colleague is related to one of the kids in quarantine at Davis. You’re probably right about the number of undiagnosed cases, which makes me think the estimated fatality rate will go down some.
smintheus
@janesays: I know that. I also know that most other aggregators have been figure out pretty accurately both the likely national vote and the likely results for all states except those that are nearly toss ups. He’s working with and doing similar things that the others are doing. And no, he was not the first to invent poll aggregation, which was fairly accurate for the 2004 election too.
Repatriated
@Martin:
[Bender]Yep. We’re boned. [/Bender]
CA has resources and is prepared to use them to address the crisis, rather than just manage the optics.
Other states expecting the Feds to take the lead…. yikes.
Martin
@Repatriated: No, I agree on the magnitude effects, but that 3% is still a lie. That’s not how our elections work. That still needs to be directly addressed.
And I contend that in 2020, CA and NY and MA could account for a 5-6% overvote, meaning a Dem candidate at 53/47 in a head to head is probably in a dead heat for the EC.
Yes, drop the probabilistic component, but a map which could take state polling, subtract out the overvote from the national, and forecast the map would be highly beneficial. Nate is still doing that, he just needs to lose the top line probability. And none of the regular national reporting is squaring the national polling to the EC.
Chris Johnson
@janesays: Other than that, Mrs. Lincoln, how was the play?
(i.e. this is not a strong economy, it’s a massive con trying to obscure total desperation on the part of normal everyday voter-type people, and you should not be saying it is a strong economy, especially after the MOTUs continue to collapse the stock market because they, collectively, KNOW it is a con where you seem not to know this)
Major Major Major Major
@Martin: Gotcha; fair.
I was just picking on Silver because I’ve come to despise him. Sam Wang, who I like a lot, has moved away from top-level predictions partly for this reason. He’s focusing instead on how data can help identify good targets for campaign investment and anti-gerrymandering efforts.
He of course has not made “getting clicks from election forecasting” his day job, and can behave dispassionately in the light of results that his hobby may be destructive.
janesays
I don’t believe the latter claim for a second – I think having self-confidence in your ability to achieve an athletic feat absolutely has a positive impact on your ability to achieve that feat. I don’t know how quantifiable it is, but I have no doubt that the greatest athletes in the world have far more confidence in their own abilities than the poorest ones do.
Brachiator
@joel hanes:
Goddam, I hate shit like this. Trump’s right wing stooges are willing to see people sicken and die while they defend the White House and disseminate lies.
smintheus
@Martin: Many US colleges and universities have hundreds of Chinese students on campus. I had a student who couldn’t get back for the start of term because she was caught in the lockdown…then the 2nd week of term she walks into my office and up to my desk to announce “I’m back on campus, I thought you’d want to know!” Lord knows how many students returning from China are asymptomatic carriers, but it would be nuts to assume that none are.
Martin
@Repatriated: In defense of the other states, they don’t have CAs  resources. None of us should have to get by without a federal govt.
Repatriated
@Martin: Agreed.
(I was only addressing the study in the OP about effects, not the issue of accuracy.)
Repatriated
@Martin:
True. That’s what a Federal Government is FOR!
But these days?
smintheus
@janesays: Confidence should come from success. There are lots of athletes who have exaggerated self confidence, and it makes them even worse rather than better.
Darkrose
@Martin: It’s real fun here in Sacramento at the moment. The woman from Solano who’s the first confirmed U.S. case to have no connection to China travel is at UCDMC. Three students in the Los Rios Community College District who work at UCDMC–including one at Sac City, where I work–have tested positive. Two of the three went back to campus after contracting the virus.
So far, nothing’s been cancelled, and there’s no sense of panic, but it’s definitely a concern.
janesays
@Eunicecycle: Ha. As a lifelong Cardinals fan, I took the Cubs winning that World Series as an extremely bad omen heading into the election. I remember a small part of me wondering on the morning of November 8th as I went to vote, “Wait… what if this dude actually wins tonight?” I dismissed the thought quickly because it seemed too absurd in the moment, but later that night as the nightmare was playing out I remembered that exact same sinking feeling that I had just six days earlier when it dawned on me that the Cubs were actually going to pull it off.
New Deal democrat
@janesays: there is an election forecasting model called the âbread and peaceâ model, which offsets the economy by war casualties. Probably wider moral concerns can be included. There is another that overlays the Index of Leading Indicators over poll results from the first quarter of the election year.
Apropos of both, Trumpâs problem is that he is morally anathema to about 50% of voters.
Eighteen months ago I called for a slowdown or recession that would have an epicenter of right about now. We got the slowdown, and coronavirus may well turn it into a recession. Left to its own devices – and coronavirus most certainly means it will not be – the economy most likely would improve before Election Day.
L85NJGT
Not all turnout is driven by individual choice.
Darkrose
@janesays: I had said for many years that the Cubs winning the World Series would be the first sign of the Apocalypse.
I was right.
Brachiator
@janesays:
Well… maybe. Perhaps an athlete has a deep knowledge of his abilities and is able to reach in and quickly use that knowledge for any challenge.
And this is just off the top of my head stuff, mixed with curiosity and amazement at some athletic feats.
This reminds me of the George Foreman/Ali fight, when Foreman was defeated. in interviews, Foreman talks about how he was punching the crap out of Ali. And Ali responded with “Is that all you got?”
At one point, Foreman told himself, yep, that’s all I got. Previously, he had depended on the overwhelming power of his punches, and when this was countered by Ali’s rope-a-dope tactics, George did not have any other resources to bring to the fight.
Ali, on the other hand, had been a long time student of other boxers. He not only borrowed from them to craft his own style, he studied the various ways of defeating even the most fearsome opponent.
Or, who knows, could be magic.
Martin
@J R in WV:
Oh, hell yes it will. Sports psychology is a valid field. I mean, that’s 80% of the task of teaching a kid to catch a ball – the confidence that they won’t miss and have it hit them in the head. Same with riding a bike. And you can see the moment they lose their confidence, their brain goes to shit, and they crash.
The effect is much more subtle in the good amateur category – where most of us live. I played a bit of ball in college, so there’s not much that will affect my performance because I’m past the ‘I lack basic skills’ self doubt, but well into the muscle memory category. I can’t not catch the ball, it’s just a reflex. But it shows up again at the elite level, where the gap from good to great is a percentage or two. Tiger Woods confident he could clear that water hazard was very different from Tiger Woods unsure that he wouldn’t re-injure his back doing it. Or Lewis Hamilton ‘I can take this turn at 130, but not at 131’ compared to his opponents ‘anything over 128 and I’m probably going to lose it’.
Don’t think any of that applies to voting, though.
Major Major Major Major
@L85NJGT: This is not an attempt to explain all turnout.
janesays
I’m betting a huge number of those people weren’t actually undecided, but rather were people who disliked both candidates greatly and clearly leaned to one over the other, but didn’t want to articulate that position out loud because they didn’t want to give even the slightest impression that they might like the person they were leaning towards voting for. My brother was one such person. He voted for Clinton, but really disliked her greatly, and said so frequently. He’s pretty much always voted for Democrat, so it was obvious to me that he would be voting for Clinton, but I remember him going out of his way to avoid saying out loud who he would be voting for. If a pollster had called him a week before the election, he very likely would have told them he was undecided, even though he really wasn’t.
Martin
@Hoodie: Right, and that’s the point. And that’s Nates point as well. The national polls need to go, but without them you put the problem of math back in the hands of the reporters. So, remove the probabilistic component and keep the rest.
Major Major Major Major
@Brachiator:Â @Martin: and let’s not forget about the yips!
Martin
@Major Major Major Major: Nate is suffering at ESPN. NYT should have kept him.
trollhattan
@Martin:
Watch a gymnast or slalom skier previsualize their routine/run just before they perform, and you’ll see living examples of how mindfulness, being in the present and blocking out extraneous stimulus are all critical to performance.
I still don’t know how anybody ever connects with a well-thrown hardball. Voodoo stuff.
janesays
They’re basically identical for Sanders and Biden when you factor in MoE. These are from the RCP averages…
MICHIGAN
Biden: +5.2; Sanders: +5.3
PENNSYLVANIA
Biden: +3.0; Sanders: +3.0
WISCONSIN
Biden: -1.7; Sanders: -1.0
Martin
@smintheus: Our 2nd largest foreign student population? Iran. Our 3rd? South Korea.
Hundreds in the university? Fuck, I’ve got 1000 just in one school. And another 1000 who are first generation Americans to parents who grew up in those countries who enjoy regular access to family in Asia and routinely go there in summer or over holidays.
Brachiator
@Major Major Major Major:
My main point here is that this particular media criticism about polls, and the focus on Silver, is wrong-headed and misguided.
Brachiator
@Major Major Major Major:
Oh yeah! Absolutely.
Martin
@trollhattan: I can’t explain it either. I don’t think about it. My hand just goes there – and well enough to catch it in the webbing and not in the palm, so I’m compensating from a freehand catch to a mitt catch, without thinking.
Played a lot of soccer as well and I can pass a ball to a player that I can’t see. I know they’re there, but I don’t know how I know they’re there. It’s kind of spooky when I stop and think about it. One of the fascinating things in basketball is watching blind passes. I swear 80% of basketball is subconscious sneaker squeak echolocation.
janesays
@smintheus: FiveThirtyEight actually has a list of pollsters that they refuse to use, and they give A through F grades to all of the polling outfits whose data they analyze. Research 2000 was the pollster hired by Daily Kos in 2008, and their figures were so lousy that they were eventually banned at FiveThirtyEight.
https://projects.fivethirtyeight.com/pollster-ratings/
LongHairedWeirdo
@Repatriated: Yes. This is one reason why I like to mention how Silver’s predictions comes out in the roll of a 6 sided die. “Clinton has a 70% chance to win; Trump has a 30% chance to win” sounds like “Clinton in a landslide” while any gamer who can translate that to “7 in 10” or “4 in 6” will realize that’s a really crappy chance, for a “one roll of the dice” scenario. At least, if they’ve done enough tabletop/dice generated gaming to realize “roll 7 or higher on a d20” or “don’t roll a 1 or 2 on a d6” will only work “most of the time” and fail *a lot* of the time.
Usually, when trying not to be disintegrated, impaled, turned to stone, caught in the supervillain’s death trap, etc..
Major Major Major Major
@janesays: Wasn’t Silver the one who figured out that Research 2000 was fabricating results?
Major Major Major Major
ok.
geg6
@Jim, Foolish Literalist:
Still is as far as I can tell. Of course, that is only based on embedded tweets here and other places. No fucking way I’m ever going to use Twitter. I didn’t like the idea from the start and now that it’s been weaponized by the worst people on earth, I wouldn’t go there if you paid me to.
J R in WV
@Major Major Major Major:
One of retired news correspondent wife’s big gripes about reporters is how many confuse per cent with percentage points, which are not the same. She also hates seeing place names misspelled and common grammar errors on CNN and MSNBC streaming news.
Plus opinions as facts !! Calling Andrea Mitchell Greenspan and Chris Matthews out there personally, leaving NYT oppo columnists out of that as they aren’t supposed to be reporters and show it.
janesays
@Hoodie: Silver does generate an electoral college vote prediction. He actually puts more emphasis on that than on any national popular vote prediction, since the latter is significantly less meaningful.
Darkrose
@LongHairedWeirdo: Probability makes much more sense to me after playing Final Fantasy XIV. Recently I ran the latest 24-person raid. Loot drops per 8-person party. I rolled on the music rolls and got a 41 and a 93. I got neither, because one of the other seven people rolled higher than me on both drops.
Then there’s crafting. Using various skills, you can increase the chance that the sword or staff or ring you’re crafting will he “high quality,” giving it better base stats and allowing you to sell it for more game money. However, the check happens once, at the time of creation, so anything less than 100% quality rating means there’s a chance you will get normal quality instead, because it’s a random behind-the-scenes dice roll. It’s annoying as hell to have 99% and still get NQ crafts, but sometimes RNG be like that. It’s definitely helped me understand probability better, though.
janesays
@smintheus: He may not have been the first, but he was the one who made it a huge thing in 2008, namely because of his incredible accuracy.
LongHairedWeirdo
@janesays:
And that is where the Republican hate machine holds such cunning sway.
Stop, and think about this: other than all the negative news stories told about her, almost all of which turned out to be nothing (including her use of a private server), how much do you know about Hillary Clinton?
By all accounts, she was a lifelong public servant, and a good woman. She was a fine senator, a competent SoS, and would have made a fine President.
She wouldn’t have been stupid enough to declare emergencies, to raise tariffs in the worst way possible; she wouldn’t have left North Korea alone because she gets “beautiful letters” (Hey, trumpie! They WERE WRITTEN BY GHOST WRITERS, YA GD IDIOT!!!); she wouldn’t have thought “boy, Iran *just barely* accepted the deal, so if I go it alone, and demand more, I’m *sure* that they’ll gladly give up more!”; she wouldn’t have withdrawn from the Paris Accord (which put *no* restrictions on our activities, and merely gave our word – which Republicans have been breaking with impunity); she wouldn’t have made deep cuts to the people who’d fight a pandemic; she wouldn’t have violated her oath of office, the law, and the Constitution by trying to get trumped-up investigations into political opponents; she wouldn’t have blown the deficit wide open; and she wouldn’t have made America a laughing stock in Russia, North Korea, Turkey, Saudi Arabia, and probably more than a few others; etc..
janesays
@Martin: Illinois actually had a bigger impact on the overvote in 2016 than Massachusetts, even though Clinton’s victory margin was ten points smaller there. But it is the nation’s 5th largest state, while MA is only the 15th largest.
janesays
@Major Major Major Major: Wang was actually a lot more wrong in 2016 than Silver in his top line prediction – he had Hillary with a 93% chance of victory heading into election day. Silver was just over 71%.
Major Major Major Major
A high school math teacher friend of mine does monte carlo simulations with her AP classes every election cycle. 18-year-olds can get roughly the same results.
@janesays: As Wang acknowledged even before election day. Unlike Silver, he doesn’t edit his model as he goes.
Also, he uses only state polls, which are major lagging indicators; they didn’t have time to catch the Comey effect.
janesays
@smintheus: And there are also normally terrific athletes who find themselves in slumps and psych themselves into self-fulfilling prophecies of continued failure. Because their confidence has been shaken.
J R in WV
Yes, but for player John Doe, believing harder today that he will hit that ball than believing it last week will not change John Doe’s batting stats or this year’s stats over last year’s stats.
It is obvious that Willy Mays had confidence in his abilities, and I’m not sure why you would think I was arguing against Willy’s confidence. Or any superstar, like Serena Williams, who should know that in any given match, she is the best player in that match.
But “knowing” that a given politician will win a race appears to prevent that win from happening in some large percentage of cases…
LongHairedWeirdo
@Martin: AGREED!
My favorite example was Rahul of the Great British Baking Show, season 9 (which might not be the “session” on Netflix, if there’s where you catch it – but his season *is* there.)
The poor guy had a bad rise on a rye loaf, and you could just *see* what a horrible case of the “yips” it caused – it was his first real *failure*, and he’d lost all confidence.
I was mourning for him, and trying to mentally send thoughts back in time: “Come ON man! You can KICK ASSÂ AND TAKE NAMES! Okay, okay, not your style. You can USE YOUR FOOT TO GENTLY PROD DONKEYS and REQUEST FORMAL INTRODUCTIONS! You *know* this, man! SEIZE YOUR (flipping) POWER!”
The difference between “I can *do* this” and “what if I *can’t*?” is huge, and, as mentioned, especially high when at the elite level.
geg6
@Martin:
Doesn’t excuse those states from trying to get out in front of the situation. Thankfully, here in PA, our governor and surgeon general have been taking some preliminary steps to prepare and to inform the public. Good government, meaning having Democrats in charge (mostly), isn’t just in California.
But just in case I should catch it, my plan is to go to one of the dozens of MAGA rallies Trump and Pence seem to have here in Western PA, and infect everyone I see. Fuck it. I’m done with these people.
janesays
@Martin: He’s actually with ABC News now (as of 2018), though that was really just a lateral move within the same Disney behemoth that owns both ESPN and ABC.
smintheus
@janesays: That list is way too short. There are pollsters particularly on the right that they know consistently produce polls with a ridiculously large partisan slant. The methodologically correct thing to do with evidence that you know is being disingenuous is to eliminate it from consideration. You don’t try to control for the manipulation by a source of information that is trying to control you; you jettison it. Most aggregators refuse to take that basic step.
Major Major Major Major
@LongHairedWeirdo: Rahul was how I taught my husband what the yips were!
smintheus
@Major Major Major Major: A lot of people suspected it was, long before it came to a public smack down.
smintheus
@janesays: It was accurate in 2004. Silver didn’t make the basic operation more credible. It was already credible, which is why he decided to jump into it.
smintheus
@geg6: Interesting plan. You could just volunteer to canvass for Trump.
janesays
Large percentage of cases? Such as?
I’m sure it’s happened before, but putting aside the 2016 clusterfuck, I’m having trouble thinking of another example of the forecasters ever getting any significant election so completely wrong in the odds of victory they gave to the loser. I definitely don’t think it’s anything remotely resembling a frequent occurrence.
janesays
@smintheus: Maybe so, but very few people were talking about election forecasting in 2004, at least in comparison to 2008. They certainly weren’t getting significant coverage in the mainstream news. Even Sam Wang acknowledged in a 2008 NY Times article that Silver was the person most responsible for making forecasting a big deal.
geg6
@smintheus:
Nah, too much effort. Much easier to get them all in one place, maybe in the line to get in, walk up and down the line, coughing and hacking in their general direction. Seems like we have a rally here at least once a month, so it’s definitely a possible plan.
smintheus
@trollhattan: Visualizing in advance at your leisure is helpful for certain kinds of competitions. With a fast team sport though, you need to empty your mind and rely upon the instincts you’ve already trained up to a high level. The less thinking you can do, the better you can sense where you are in the field of play and the faster you can react to what your opponents are doing. Also (this is easily overlooked) the less front-of-your-brain thinking you’re doing while competing, the less tired you become. Your job ideally is to try to force your opponents to think a lot about what you’re doing, so as to wear them down even if nothing you’re doing initially seems to be leading to scoring.
glory b
@Major Major Major Major: *Cough* Steve Kornacki *Cough
smintheus
@Major Major Major Major: I should add that they were known for offering to do polling at knock down prices. That to me was the first red flag that their polls might be sketchy.
JoeyJoeJoe
I donât think national polls are entirely useless; a candidate who is ahead nationally by a large margin is almost guaranteed to be ahead in enough states to win the electoral college. Â To lose, that candidate would have to over perform in their strongest states and also overperform In the most unfriendly states but not by enough to actually win them while at the same time underperform in the swing states. Â Thatâs super unlikely, which is why predictions that a Democrat could win the popular vote by big margin and still lose are crap. States are more polarized than they used to be, so national polls are less predictive than in the past, but theyâre not meaningless.
janesays
Concur. I think there’s a strong likelihood that if Trump wins he will do so while simultaneously losing the popular vote, but it’s not going to be by a substantial margin. I’m gonna say 4% is the maximum margin we’ll see anytime this decade for a popular vote losing presidential election winner. Guessing what the landscape will look like for 2032 and beyond is pointless.
While I say I don’t think it’s possible for someone to win the EC while losing the popular vote by more than 4 points, if the Democratic candidate only has a 5 point lead in the aggregate polls heading into this year’s election, that is far from being in safe territory when you factor in MoE. If that person is up by 10 points in the aggregate polls on election eve, they’re almost certainly safe.
randy khan
@Martin:
The broader point that we have 51 state elections, not a national election, is right, but there is a certain whiff of the strange about “take out the two biggest states that Clinton won – and which together account for 1/3 of the total necessary to win the Electoral College – and Clinton lost the rest of the country.” Because, you know, that still leaves 49 individual state contests.
And for that matter, if you’re talking about wasted votes, Trump won 10 states plus a Nebraska CD with 60%+ of the vote, and those places had a total margin of 3.66 million in his favor. It’s not reasonable to pull out Clinton’s best states and count Trump’s.
Lymie
Freely available version of the paper:
https://www.dartmouth.edu/~seanjwestwood/papers/aggregator.pdf
Ruckus
When I tutored statistics in college one of the things that I realized was most of the people I worked with thought that you always got the same number when you manipulated the data. They thought that the concept of statistics was that the answer was always consistent, like all math they had seen up till then, 2+2=4. But statistics is about manipulating numbers/data. If done consistently and openly the info can be  useful, 99.9% of light bulbs manufactured last x hours. But the bigger the number of variables and the size of the study has an increasing degree of inaccuracy. Add in that if you are taking the actions of humans, which often defy any rationality and you get a far greater degree of variation.
Uncle Cosmo
Making meaningful contact with a baseball thrown from the mound by a MLB pitcher is arguably the single hardest task in all of sports. The pitching mound is 60.5 feet** from home plate, & the ball is probably released ~58′ from the plate. At 79 mph – a speed so leisurely it’s considered an off-speed pitch – the ball takes 1/2 second to cross the plate. In that half second, the batter must
All of the above in one-half second.
Note, kindly, that a US/Canadian football quarterback who consistently completes less than half his passes usually risks being tossed off the team. Similarly for basketballers hitting less than half their shots. But a batter who can strike a baseball pitched by a MLB pitcher in such a fashion as to attain a base (or score), only has to do that consistently (over a number of seasons) 3 times in 10 at-bats to be a prime candidate for the Hall of Fame.
(My dad was a shortstop on one of the best semiprofessional teams in the US before WW2. One of his nephews played in the major leagues for a few years, which was some compensation for the fact that his older son had the worst hand-eye coordination in recorded history.)
** NB I am fucking well NOT going to translate this into metric for you un-American hangers-on. Find a conversion app & kwitscherbeliaekin.
a swing
@trollhattan:
Uncle Cosmo
Not just great athletes either. One of the earliest adopters of a primitive protostatistical approach to baseball was former Baltimore Orioles manager Earl Weaver. He’d keep index cards for each combination of O’s hitter and opposing pitcher, summarizing the results over seasons across the league, & pull them out when he had to make a decision.
Gary Roenicke, the right-hand hitting half of the O’s left-field platoon (John Lowenstein was the lefty), once said there were two great aspects of Weaver’s House of Cards. First, players not in the starting lineup didn’t take it personally, it was just that someone else had had better results against the opposing starter. (“He went to the cards on you.”) Second – & here’s where we finally get to answering your post – Weaver famously never asked a player to do something he wasn’t capable of. When Earl sent him up to bat, that player knew those cards said he was the best choice on the roster, the most likely to succeed. And knowing that made him a more confident & therefore better batter.
Uncle Cosmo
@Ruckus: For most of my working career I earned my salary either as an applied statistician or an analyst who could write coherently (thereby allowing the engineering department to bypass the customarily technically-challenged tech writing department). I used to get all kinds of pushback from managers who didn’t want to pay me for an analysis they could have a junior grunt whip up with a stat package. I would tell them, “You’re not paying me to run the analysis – you’re paying me to know which of the numerous analyses that might be run might give you results you might be able to use, among all the alternatives that in all probability (SWIDT? ;^D) would be worse than useless.”
I was “workin’ in a dee-fense plant” (Phil Ochs) when the head of engineering, much enamored of statistics, brought in a PhD demographer being let go by another division. After some time as coworkers & cubicle-mates, I told him,
And in fact I shook his hand & wished him well as he was carrying his personal items out of the office for the last time….
When I taught an evening section of Stat 101 at the local community college, I’d tell the students, “No one but a few cranks like me actually enjoys doing statistical work. We do it because we have to. It’s one of the few ways we can understand the workings of an imperfect & incomplete world where the key information is hidden or unobtainable.”
And with that, I’m out…
Bobby Thomson
@New Deal democrat: his handle wasn’t Publius. He went by poblano.
Matt McIrvin
@Major Major Major Major: Wang should never have gotten into probabilistic Election Day predictions. He did so reluctantly in the first place and his model was just bad–way too little assumed correlated spread.
Steve Biodrowski
@James E Powell: “…coast to victory…”
Polls showed a tight race within the margin of error. That’s sort of the point of this article – that people misunderstand probability to mean certainty.
Another Scott
Dead thread, but I don’t see another Open one…
WaMo – Cardiologist weighs in on Bernie:
Reasonable questions…
Cheers,
Scott.
Another Scott
This is my shocked, shocked face.
They’re shameless and think they can do whatever they want, no matter the law. We must vote them out.
Cheers,
Scott.