Nate Silver acknowledges he could be wrong
He recommends the linked critique to explain possible flaws in his model. Of course, he disputes the analysis.
Silver takes state polls at face value. This could be a problem:
In a nutshell 1) Polls historically have underestimated Republican performance; 2) Polls historically have overestimated incumbent support; and 3) state polls indicate a turnout advantage for Democrats even larger than the one they enjoyed in 2008 and there are several reasons why this is unlikely to materialize.
The author believes Obama is still the favorite, but the odds are 60/40, not the 83/17 Silver calculates.
|by Concern Troll||reply 89||11/07/2012|
I'm glad you labeled yourself as a concern troll but please go the fuck away. Everyone that's not an idiot knows that if the state polls are wrong that Silver's (and everyone else's model) will be wrong. No shit. The idea that every pollster has been so consistently wrong is small, and guess what Silver acknowledges that possibility and that is where the Romney chances of winning come from.
God I can't wait for this election to be over, the amount of trolls it has brought to the board is insane.
|by Concern Troll||reply 2||11/04/2012|
Agreed r2. It's not quite as bad as the Hillary/Obama wars, but it's pretty fucking awful.
|by Concern Troll||reply 4||11/04/2012|
You couldn't post this on the Poll thread, OP?
I wonder why?
You know R2 is right about you.
|by Concern Troll||reply 5||11/04/2012|
This morning George Will predicted that Mittens will carry Minnesota -- which would make it the first time the GOP carries that state since 1972 -- because the same-sex marriage amendment will bring out all the evangelicals.
|by Concern Troll||reply 6||11/04/2012|
1) George Will is an ass.
2) There is no poll in existence showing Romney even close in MN.
|by Concern Troll||reply 7||11/04/2012|
You mean the 80/20 Silver used to calculate, or you mean Silver is playing his own devil's advocate?
|by Concern Troll||reply 8||11/04/2012|
You're not a "worrier" you're a shit-stirrer.
You don't "upset" anyone, just annoy--like a fly or a bad smell.
|by Concern Troll||reply 13||11/04/2012|
Freeper heads are exploding early.
Must have come to the realization that their boy is toast.
|by Concern Troll||reply 14||11/04/2012|
OP, I share your deep concern!!! (wrings hands in agony) Oh, what is to be done???
|by Concern Troll||reply 16||11/04/2012|
I wonder if Silver's model accounts for the "Bradley Effect". That's the uncertain thing that has me the most worried about this election.
|by Concern Troll||reply 17||11/04/2012|
For Nate Silver to be "wrong" a number of state polls would have to be wrong. State polls have historically been accurate 97% of the time.
|by Concern Troll||reply 18||11/04/2012|
First you freepers try to paint Nate as a hack then spend the 11th hour saying he might be wrong, so Romney might win.
If Nate is so useless, surely you have other evidence of Romney's "chances."
If so, please share.
|by Concern Troll||reply 19||11/04/2012|
George Will is delusional.
MN hasn't gone repug in my lifetime and we sure aren't going to start with Romney.
|by Concern Troll||reply 20||11/04/2012|
Also, remember his track record. He was the closest of anyone to calling both the 2004 and 2008 elections.
The last I heard, George F. Will was not especially renowned for his statistical abilities.
|by Concern Troll||reply 21||11/04/2012|
R17, the country isn't the same place it was during Bradley's 1982 campaign. I don't think there will be any significant effect.
|by Concern Troll||reply 23||11/04/2012|
I've yet to meet a white person who wasn't voting against Obama to shout it from the rooftops. And all those white people who are voting for Obama, haven't exactly been shy about it either. I think the Bradley Effect is dead in this election.
|by Concern Troll||reply 24||11/04/2012|
Those who think that gay marriage might tip the scales in Mitt's favor in MN should remember California in 2008: Obama won the state handily in spite of Prop 8 being approved.
I hope that marriage equality becomes a reality for MN, but if it doesn't I don't think that will lead to a Romnney win there.
|by Concern Troll||reply 25||11/04/2012|
R6, if this were 2004 I guess he might have some point, but... it's now 2012. Same-sex marriage isn't quite the lightning rod issue it used to be (even in past two years support has gone up dramatically). A majority of Americans now support it. In Minnesota, polling shows the anti-marriage amendment is headed towards defeat (and Obama towards a comfortable win).
I would not be surprised if there have been shifts in evangelical attitudes and motivation about the issue as well - certainly it's hard to believe they're as fired up about the idea as they were a decade ago. In any event, the clearest sign that there won't be some surprise groundswell of turnout for it is that it hasn't been a focus this election cycle - Republicans have let his vocal support of same-sex marriage go largely unremarked upon. They don't want to alienate the moderates and the majority position.
So yeah, maybe George Will needs to refresh his assumptions about what gets people out to vote (the same way Karl Rove apparently has). And just as there may still be evangelicals motivated to vote on it, there will still be plenty of young people, liberals, and LGBT-friendly groups motivated to get to the polls to defeat the amendment as well.
It's kind of like people still being stuck in the mindset that the polls are unreliable because of the "Bradley Effect" and voters not being comfortable with a black president - I mean, go look at the 2008 polls and the actual results already to see if any of that played out (it didn't).
|by Concern Troll||reply 26||11/04/2012|
R26, R6 is just trying in insult/upset the gays.
|by Concern Troll||reply 27||11/04/2012|
Oops, saw that others remarked on the "Bradley Effect" - I don't even know what the reasoning is for that still being a concern for this election.
Like, in 2008 the argument was "white voters will tell pollsters they're supporting the black candidate because they want to appear socially liberal, but in the voting booth their real attitudes will come out."
At this point, President Obama is no longer "the black candidate" - he's been president for four years. There are plenty of reasons to form an opinion about his desirability as president, and trying to appear cool by supporting "the black candidate" really isn't on anyone's mind.
If anything, it seems there's more chatter about "the Bradley effect" for Romney. That Republicans will publicly express their support for him in polls, but private concerns about his Mormonism (especially among evangelicals) will come into play for voter turnout and at the voting booth.
|by Concern Troll||reply 28||11/04/2012|
Uhh, if you understood anything at all about statistics you'd know that Nate Silver implicitly acknowledges he could be wrong in everything he does, because that's the nature of what he does. But his best critics (that is, those who DO understand his work well enough to point out flaws in it, like Sam Wang) give Obama better odds than Silver, anyway.
Now go away.
|by Concern Troll||reply 29||11/04/2012|
Because he knows the Rethugs fucking cheat. Someone is going door to door in Boulder CO right now collecting ballots, but it's a scam.
Motherfuckers should go to jail for 50 years for election fraud. Maybe then they would think twice.
|by Concern Troll||reply 30||11/04/2012|
Thank you r29. The lack of statistics literacy is truly astounding. I mean, at least some meatheads still follow baseball stats, right? That's how the Freakanomics guys and whole slew of other economists and statisticians got "hooked".
|by Concern Troll||reply 31||11/04/2012|
Hey you condescending morons: This is not a question of the inherent uncertainty in statistics. It's a question of a defect in Silver's model. Which is why Silver himself referred us to the article. Assholes.
|by Concern Troll||reply 32||11/04/2012|
From the author's bio:
[quote]The Wall Street Journal has called him a "leading tort-reform advocate."
[quote]serves on the Executive Committee of the Federalist Society Litigation Practice Group
[quote]an attorney on the McCain-Palin 2008 campaign
How shocking that a die-hard Republican would dispute Nate Silver's analysis.
|by Concern Troll||reply 33||11/04/2012|
It is not shocking at all. But that's not the point. Silver disagrees with the author's conclusions but respects his arguments enough to refer his reader to them. So regardless of the motivations of the author, Silver believes his points are worth considering.
|by Concern Troll||reply 35||11/04/2012|
As to Nate Silver, nobody who works heavily with statistics is unaware that they can be misleading at times. Silver is very good at this game, and he constantly backchecks on possible sources of error in his models, even though he believes them to be correct. He has been very full in explaining why he has come to the conclusions he has, but also quite honest in acknowledging that polls are no substitute for actual votes. 538.com is pretty clearly the best political website going (at any rate, for popular audiences), but it cannot promise more than it does, which is an educated surmise about the overwhelmingly likely outcome of the election.
|by Concern Troll||reply 36||11/04/2012|
R32, is there really anything wrong with Silver's basic methodology? He put it pretty plainly the other day - there's certainly a possibility (around 15%) that Mitt Romney could win the election - this is about the chance that the vast majority of polls coming out of swing states are all incorrect and statistically biased against Romney.
Whether you think Silver has some "secret sauce" that he's getting wrong, it's hard to dismiss the numbers: out of 22 battleground state polls that came out Friday (covering 10 different states), 19 showed Obama ahead of Romney, two showed a tie, and one showed Romney ahead in Florida.
If Silver ends up being "wrong", it'll be because the source of his data - state and national polls - are wrong, not because his "secret sauce" didn't work.
There's really nothing confusing or earth-shattering about looking at a lot of polls that show Obama is ahead in several key states, and saying Obama will win.
Sure, you can get into more granular detail (did Silver get the exact percentage right? The exact number of Electoral Votes?) but those are more academic betting questions than of major concern as to who will be elected.
|by Concern Troll||reply 37||11/04/2012|
So your point, OP, in yet another obsessive post of "concern" about Nate Silver is that the polls still say Obama will win.
You do know if both the Silver critic and Silver are correct - i.e., Obama wins -- the difference of an 80/20 chance that he would win and a 60/40 chance are ... um... irrelevant? nonexistent?
|by Concern Troll||reply 38||11/04/2012|
[quote][R32], is there really anything wrong with Silver's basic methodology?
Time will tell. The key issue is whether there should be any adjustment to the state polls based on historical patterns or should they be taken at face value.
|by Concern Troll||reply 39||11/04/2012|
[quote]respects his arguments enough to refer his reader to them.
I can't find this - would you link to it?
|by Concern Troll||reply 40||11/04/2012|
R39, if we take the state polls at face value, Obama clearly wins (I think the Princeton Election Consortium just aggregates them, and has Obama's probability at 99%).
If we take Silver's more conservative weighting, Obama still wins.
It's only if you completely discount the state polls and argue they should all be going the other direction (an Obama +3 should really be a Romney +3), that you get a different result.
THAT sounds much more academically dishonest and likely of being wrong.
I really don't understand the Fox News criticism of Silver - he's just reporting and aggregating the data provided by polling organizations, and spitting out a probability. Take it up with Gallup / Rasmussen / PPP if you think their polls are wildly off.
|by Concern Troll||reply 41||11/04/2012|
Nate Silver @fivethirtyeight
If you want a good case from a conservative about why polling sites have Obama's odds too high, @tedfrank has one:
|by Concern Troll||reply 42||11/04/2012|
Someone who predicts what might happen in the future admits he "could" be wrong? Really? Shocking!
|by Concern Troll||reply 43||11/04/2012|
And P.S. to R32, be careful who you call condescending on the one hand and moron in the same set of keystrokes.
Silver did NOT say there's a flaw in his model. He only acknowledged that there are SOME arguments that the polls are wrong.
Silver's methodology is to take a collective look across polls with multiple methodologies to reduce their error. On that basis, Silver explained that UNLESS the polls are SUBSTANTIALLY biased, they collectively point to an Obama win, and to reverse the findings in enough those polls to get there, the odds are significantly in Obama's favor. OP and R32 can't seem to follow that.
The OP and condescending jerk at R32 is obsessed with Silver, AGAIN I say for irrelevant reasons if both predictions are true.
And who wants to bet OP will come here and tell us there was never an 80 percent chance Obama would win and Silver was wrong.
You are obsessed with Silver, OP. Get help.
|by Concern Troll||reply 44||11/04/2012|
Oh yes, R42, that reads like Silver crying that his own model is flawed.
|by Concern Troll||reply 45||11/04/2012|
He is not saying his prediction might be wrong. He is saying the model that he uses to reach his prediction may be wrong. Not the same thing.
|by Concern Troll||reply 46||11/04/2012|
R46, not even that his model is "wrong", but acnkowledging there are legitimate academic criticisms and other viewponts of his particular methodology.
He doesn't necessarily agree with them, but acknowledges there's a rational counter-argument.
|by Concern Troll||reply 47||11/04/2012|
R45, just giving R40 the info he was looking for. I accidentally pasted over a bit of into text to make that clear. I have no problem with Nate's methods or conclusion.
|by Concern Troll||reply 48||11/04/2012|
[quote]If you want a good case from a conservative about why polling sites have Obama's odds too high, @tedfrank has one:
I guess I'll have to take your word for it. Try as I might, using Ctrl-F I can't find that quote on Five Thirty Eight.
|by Concern Troll||reply 49||11/04/2012|
[quote]Silver did NOT say there's a flaw in his model. He only acknowledged that there are SOME arguments that the polls are wrong.
What part of "could be wrong" are you having trouble with?
[quote]The OP and condescending jerk at R32 is obsessed with Silver, AGAIN I say for irrelevant reasons if both predictions are true.
What is your basis for this ridiculous conclusion? I started one thread about Silver. What I am obsessed with is defending myself against accusations of being a troll and accusations that I do not understand the nature of statistics.
[quote]And who wants to bet OP will come here and tell us there was never an 80 percent chance Obama would win and Silver was wrong.
Regardless of the outcome, after the election, no one will be in a position to argue that the odds were never 80% .
I'm pretty comfortable with the "moron" characterization, as that word is commonly used on DL. Rest assured, I don't not allege that you are literally a moron.
|by Concern Troll||reply 50||11/04/2012|
you just proved the accusation you called ridiculous.
again, if the author you are using over and over to build your case and Silver are correct, you won't be able to prove Silver's model was wrong.
which means the whole THREAD you started was ridiculous. if you had a reason to post this - to encourage democrats to vote - it would make sense.
the ONLY text in your post is that you are attacking Silver. you yourself said you mockingly called yourself concerned.
Silver has said there are well reasoned criticisms but does not accept them nor has he admitted his model is flawed. you saying otherwise is absurd.
they both say Obama will win. if Obama wins the whole disagreement about chances of winning are moot.
your post is absurd. case closed.
|by Concern Troll||reply 53||11/04/2012|
R53--That post is truly astounding in its failures of comprehension and logic. Really, quite impressive. I hope you wrote it as a parody.
|by Concern Troll||reply 54||11/04/2012|
[quote]This thread is typical of how faggots argue. With straights, the woman is always right. ALWAYS.
Someone needs to get laid.
|by Concern Troll||reply 55||11/04/2012|
no it isn't
you are quibbling over chances of winning when both say Obama will win and if they're both right this point will be moot in days. and yet you can't let it go.
and you are exaggerating silver's acknowledgement that a critic makes good points into a claim that Silver admits his model could be flawed.
if you can't get that you're as dense as people are saying.
|by Concern Troll||reply 56||11/04/2012|
[quote]you are quibbling over chances of winning when both say Obama will win and if they're both right this point will be moot in days. and yet you can't let it go.
No, I am not. This post is about the methodology of Silver's analysis. I have not discussed the chances of Obama winning except to note that even the author questioning Silver's methods agrees that Obama is the favorite.
The fact that you think this thread has been an argument over the odds demonstrates that you have not understood it at all and that you are using your own imagination, not my words, to make your points.
Good day, sir.
|by Concern Troll||reply 57||11/04/2012|
Jeesh R42, give it a rest already.
|by Concern Troll||reply 58||11/04/2012|
Of course Silver "could" be wrong. Any prediction, by definition, may be wrong.
Only freeper nutbags like OP would gleefully post that as if she just discovered sliced bread.
|by Concern Troll||reply 59||11/04/2012|
R58, please read my post at R48.
|by Concern Troll||reply 60||11/04/2012|
OP, Nate Silver is scheduled to appear on The Daily Show on Wednesday. Why would he appear on Jon Stewart's show the day after if he was wrong?
|by Concern Troll||reply 61||11/04/2012|
"the same-sex marriage amendment will bring out all the evangelicals."
George Will is a ("Can't spell 'Mormon' without it") MORON. Evangelicals---both Grahams and their ilk notwithstanding---will sit home before they vote for a MORMON.
|by Concern Troll||reply 63||11/04/2012|
Stop calling it a "PREDICTION"
It's a PROBABILITY! Whether it's an 8 in 10 chance or a 6 in 10 chance, Obama's odds are better.
|by Concern Troll||reply 64||11/04/2012|
Um R64, calm down. It IS a prediction. Silver's prediction is that Obama will win, with an 83% probability.
I run statistical models for a living. You should brush up on your high school math before yelling at people on the internet.
|by Concern Troll||reply 65||11/04/2012|
Fuck you R65. Nate Silver is not predicting that Obama will win. He is predicting that if the election were held 10 times, Obama would win 8 times and Romney 2 times.
He's model says Obama is more likely to win than Romney.
|by Concern Troll||reply 66||11/04/2012|
[quote]He's model says Obama is more likely to win than Romney.
THAT is precisely the prediction.
By the way, brush up in your English and anger management.
|by Concern Troll||reply 67||11/04/2012|
Right now at Bookmaker.eu Barack Hussin Obama and the Democrats are the odds on favorite at -325 which means if you think the left will retain the White House you need to bet $325 to win $100 plus the return of your initial investment. The Republicans with Willard Mitt Romney are a +275 underdog which means if you think the right will take back 1600 Pennsylvania Avenue you need to wager $100 to win $275 plus and the return of your original wager
|by Concern Troll||reply 68||11/04/2012|
[quote] brush up in your English
|by Concern Troll||reply 69||11/04/2012|
Wired magazine (November 2, 2012): Probability Versus Prediction in Volcanic Hazards (and Elections)
excerpt: Anyone can make a prediction, and that is the real danger. [bold]Predictions and probability should not be confused[/bold], which is why something like 538 or an event tree is fundamentally different than the talking heads on television. An 80% chance of something happening is NOT the same as saying that the something WILL happen, and many people inside and outside the mainstream media don’t seem to understand (or want to understand) this – both for volcanic unrest and elections.
|by Concern Troll||reply 70||11/04/2012|
Nate Silver seemed so "dead" on TV. He does not seem to have emotion or facial expressions!
He surely does not have a sparkling personality.
|by Concern Troll||reply 71||11/04/2012|
R68, "The Left" hasn't been in the White House for a long time. Obama is at most a centrist.
|by Concern Troll||reply 72||11/04/2012|
Nate is exhausted, he needs a bath and some milk with brandy
|by Concern Troll||reply 73||11/04/2012|
Nate Silver is king of the nerds. Math geeks unite!
|by Concern Troll||reply 74||11/04/2012|
[quote]Regardless of the outcome, after the election, no one will be in a position to argue that the odds were never 80%.
Actually, yes, they will, since there will be quite a lot of new data available (vote counts and exit polls) that will be compared against the various pre-election polls to determine whether Silver's model was accurate or whether Frank has a point. We will certainly know whether Frank's points 1, 2, and 3 have any validity.
If the polls are wrong in just the way that Frank thinks they are, then Silver was almost certainly wrong in assigning an 83% probability. If they are not, then Silver looks pretty good, assuming, of course, that Obama wins by close to the margin predicted.
|by Concern Troll||reply 75||11/04/2012|
New Republic: To ‘Predict’ Nate Silver’s Future, Look to the More Enlightened Sports World
But Silver and his defenders have run aground on the same problem sports statisticians used to face: [bold]the failure of laymen to grasp the difference between predictions and probabilities.[/bold] “The criticism of Nate is that he’s predicting something, when he’s trying to explain that’s not what he’s doing at all,” said Dave Cameron, a baseball statistician at FanGraphs who briefly worked with Silver at BP. “He’s putting the odds on something.” Cameron added, “It is kind of like what we do in baseball.[bold] We recognize there are multiple outcomes. [/bold]
|by Concern Troll||reply 76||11/04/2012|
The new GOP hates smart people.
With a passion.
|by Concern Troll||reply 77||11/05/2012|
Nate has Aspergers and horrible breath, but he's a genius who has made statistics cool. That's no small feat.
|by Concern Troll||reply 78||11/05/2012|
Stop with the aspergers bullshit.
|by Concern Troll||reply 79||11/05/2012|
It's not bullshit, asshole.
|by Concern Troll||reply 80||11/05/2012|
Proof of non-bullshitedness please.
|by Concern Troll||reply 81||11/05/2012|
[quote]If the polls are wrong in just the way that Frank thinks they are, then Silver was almost certainly wrong in assigning an 83% probability. If they are not, then Silver looks pretty good, assuming, of course, that Obama wins by close to the margin predicted.
That would be just guesswork. Silver sees a 1 in 6 chance that Romney would win and Franks see a 2 in 5 chance. If Romney wins you might argue that Frank had a better model, but you could never go back and retroactively say precisely what the odds were today.
|by Concern Troll||reply 82||11/05/2012|
Who's the breath troll? I've read about Chuck Todd's and now Nate Silver's and I believe others too.
|by Concern Troll||reply 83||11/05/2012|
Should I give up on waiting for proof of Nate Silver's Asperger diagnoses and just assume that R80 is spewing bullshit?
|by Concern Troll||reply 84||11/05/2012|
[quote]That would be just guesswork. Silver sees a 1 in 6 chance that Romney would win and Franks see a 2 in 5 chance.
It would be more than guesswork. Frank has made some very specific points that will either be supported or contradicted by the data that will be available on election day. For example, his first point is that "polls have historically had a bias against Republicans relative to how election day turns out."
After tomorrow, we will know for certain whether this point is valid or not. If it is not, Frank doesn't look so good. The same thing applies to points two and three on that blog post. Either they will be upheld or they will be contradicted by the data that comes out on election day.
If they are upheld, then Silver was clearly incorrect in his assessment, since he took none of those into consideration in his model. If they are not, then Frank was equally clearly wrong and Silver's model looks significantly better.
[quote]If Romney wins you might argue that Frank had a better model, but you could never go back and retroactively say precisely what the odds were today.
Yes and no. No, I can't definitively say what the odds are today but I can assess whether Frank's concerns are valid or not. If they are not, I can retroactively say that the probability he assigned was incorrect. If they are, I can retroactively say that the probability that Silver assigned was incorrect.
|by Concern Troll||reply 85||11/05/2012|
Nate Silver is a GOLDEN GOD.
|by Concern Troll||reply 86||11/06/2012|
Nate is 45-0 atm, something like that?
|by Concern Troll||reply 87||11/06/2012|
where is the stupid OP now?
let's see him come and explain why Nate's model was wrong.
|by Concern Troll||reply 88||11/06/2012|