OK. Now that is just funny. I have read a couple of liberal sites where everytime someone tries to mention a poll that is not favorable to Obama — the almost automatic retort has been Nate Silver said…
It is great to see the truth behind the smoke and mirrors.

I definitely agree that Nate Silver is keeping the liberal panic at bay. I mean who can feel panicked when Nate’s giving the prez a 70% chance of winning?!? He is going to look stupid on Nov. 7th.

The problem that I have Silvers hocus pocus is that he or even some right pundits never take into count that the incumbent very rarely wins when his or her numbers are below 50%. For this reason alone I believe that Nate is wrong. The polling organizations polls this week show nothing more than hope on their end to depress the conservative turn out on election day. look at the internals and they tell all that needs to be told. Romney will be the 45th President.

For the polls in Ohio to be held out as a reliable reflection of the voting population is laughable, unless you think turnout is going to be substantially lower than in 2008.

Most of the Ohio polls are dramatically overstating the percentage of voters who have voted early if 2008 level turnout is presumed. Some are doing so by as little as 8%, others by over 20%. Almost every poll gives Obama a 15-20% margin among this subset of voters. If you take these polls and weight the number of early voters today (as reported by the Sec’y of State) as a percentage of the total 2008 vote, and do the same for the Election Day vote, the polls show the race is at worst tied, and very often Romney leads.

To illustrate, Purple Strategies found that 26% of Ohio likely voters had already voted. The data from the Ohio Sec’y of State during the period of this poll, as a percentage of 2008 turnout, was about 16%. Thus if turnout matches ’08, this is an oversample of early voters, and a significant one given that Purple Strategies found that early voters favored Obama by 26%! If you weight the poll to show early voters making up 16% of the total vote, and yet-to-vote voters making up 84% of the total vote, the reported result of a 2% Obama lead becomes a 1% Romney lead.

Another example is PPP. If the already-voted/yet-to-vote subsets are weighted as set out above, Romney’s performance morphs from a 4% defeat to a 3% victory.

Same story with CNN, if the reported 40% already-voted finding is correct.

These numbers are more in line with what one would expect to see if Romney had a small lead nationally. The internals continue to tell a different story than the toplines, and the toplines tell a different story than the hard vote count.

Count me as cautiously optimistic.

At the risk of repeating a cliche….The article proves once again, using Mr. Silver’s case as an example, the fundamental truth addressed in the statement, “garbage in, garbage out”.

Personally, I’ll go with the University of Colorado predictions. They’ve been incredibly accurate on a state by state prediction of electoral college results for the past 30 years – 8 presidential elections. Their model isn’t based on polling, but primarily on economics. They update their model with current state by state economic indicators regularly and release their predictions each time.

For the past several months, they’ve predicted a Romney landslide by the electoral college. Obviously it’s possible this could be the one they get wrong, but they’ve certainly got an excellent track record that’s far better than any polling.

(speaking of the upcoming Romney landslide, you might enjoy my humorous post Uninstalling the Obama Trojan)

Actually, both Romney and Obama each have a 50% chance of winning. After all, there are only two of them.

Valid point. He definitely uses too many significant figures.

More often than not, models will tell you want you want to hear.

If Romney wins, no doubt we’ll start hearing nefarious Deibold-like stories galore, because Nate Silver’s predictions “prove” that Obama won the election fair and square.

That, JKnight, is what we call convoluted logic. Independents are leaving Obama, therefore they HAVE to be fringe Republicans. Yeah…RIGHT…it can’t possibly be that Independents have absolutely no use for the guy. As I said above, reality is in store for these clods in a few short days.

“If you weight the poll to show early voters making up 16% of the total
vote, and yet-to-vote voters making up 84% of the total vote, the
reported result of a 2% Obama lead becomes a 1% Romney lead.”

Brilliant point … it’s even worse than that for Obama, because if he is leading in this poll by much more than rest of poll, AND the “I voted already” is being overstated, that means that the vote liars might not end up voting at all, so it shifts even more for Romney.

These polls of Obama winning the early vote were a concern, but national ones (Gallup) show Romney winning the ‘already voted’ vote.

I care not just about Romney winning, but also important if not MORE IMPORTANT THAN A ROMNEY WIN IS AN OUTRIGHT WIN IN THE US SENATE TO GO ALONG WITH IT, without which, Romney will be unable to do much without making deal with the devil (aka Harry Reid).

Nate has simply ignored polls he doesnt like, which is a big part of why he is wrong. He also is abusing statistical science, but that’s a separate issue from his absurd bias and slant via poll cherry-picking.

An incumbent winning only 47% of the popular vote five days before Election Day is in big trouble. What’s amazing is that 47% of people think the Butcher of Benghazi deserves a second term. God help America.

Steve, you and I must have had math teachers in common . . .

47% of our voters couldn’t even show you Libya on a globe. I really wonder if they even know that it exists.

If the Republicans left the party in 2008 by simply staying home and not voting that would be a reason the Dems had a +7 edge in 08. If those who stayed home in 08 return in 12 (which they will…) it doesn’t matter what they call themselves..a vote for Romney is a vote for Romney…and that +7 edge disappears and becomes a +3 D to +3 R on Tuesday….so how do you get a +7 or +8 Dem with millions of Republicans realizing they have very big interest in voting this year–this wipes out any gains in the population increase of hispanics….and how many Dems are doing what Republicans did in 08–staying home?

There’s a simpler formula than all of this math. If you want to know who will win any Presidential election, ask one question: are people wanting change, or more of the same. We saw how people responded in 2010 to the “change”. That was a vote against Obama. Another word that matters, especially now is “chance”–as in, “let’s give me a chance”. That would be Mitt Romney. Many of the people who gave Obama a “chance” in 2008 regretted it soon after, and were just waiting to see who might replace him. You have to have an intuitive feel for this, from listening to middle American voices, rather than by over processing statistics from whatever sources, right or left. Remember, the polls are just an imperfect reflection of what average people are actually thinking. I think Romney will do a little better than Bush did in 2004, and electorally be around 300.

No, their world will continue on its delusional way. There is no cure for liberal fantasies except reality, and they refuse to see it or accept it. They will blame everyone but Obummer. They will blame Bush, Romney, racism, transfats, global warming, and rabid squirrels. But, they will not accept reality. So, expect them to find a nut case for ’16 and expect the lap dog press to lie about everything Romney does or says.

The problem with people like Silver is that he will still be around no matter the outcome of the election. If he is right, he will be touting that forever. If he is wrong, there will be the endless interviews and such where he will explain why he failed, and why he will be right the next time. The libs keep their failures around and treat them like winners. And, the lame stream media goes along because they are libs at heart. I wouldn’t be surprised to keep seeing Obummer ads even after they lose this one big time. And, even worse, we will have to listen to Bite-me on the View and other trash shows, and Moochelle will be telling us what to eat every chance we get. Of course, she will no longer be “proud” of America if it dumps her current husband. I wonder who his replacement will be.

That makes more sense that listening to pollsters to me. I think, for whatever reason, people are not always honest with pollsters. The proof is in the voting booth pudding. And until then everything is merely speculation. Remember, there are three kinds of lies: Lies, damned lies, and statistics.

The Colorado model is interesting, but you are not 100% accurate with the way you describe it.

They do not (yet) have a predictive model – meaning, the model (well, models) they have developed have been developed after the fact. This is the first election they will be tested as events happen.

In most any situation with lots of raw data you can correlate enough data points to create a model which tracks historical precedents and predicts historical events correctly. With enough data points there are often hundreds, thousands, many times millions of models that will correctly predict the outcomes.The trick in making predictions is separating things that influence the results, and things that are noise.

So to put it this way, after the baseball season is over, you can take the whole stats book, and create a formula that solves the win/loss record for the World Series Championship. In fact, there are probably hundreds of formulas that will get you there. And you can keep adding statistics that make it seem more impressive – the 2nd basemen’s slugging percentage is always higher than the 3rd basemen’s slugging percentage, when the team is home, on Tuesday’s.

It will be interesting to see how Colorado model pans out, but right now, it’s not a prediction model, it’s an economic model.

If Romney is ahead by 2, why is going to PA? That’s a solid lead, he should be moving on to another target.

I think one problem with what are seeing is that all pollsters are having problems with determining the voter mix, self-identification is a very problematic thing right now. Others have pointed out that many solid Republican voters identify as Independent now, and many solid Democratic voters do not identify as Democratic.

Forcing Obama to defend PA means Obama has less time to shore up Ohio, Wisconsin, etc. Also, a 2pt lead is not a solid lead.

“particularly when it comes to things like climate change” – given that my profession directly deals with this: conservatives have been on the wrong side of science here for quite a while. One could quibble about the models and their predictions – someone could convince me that some given factor isn’t being weighted properly or that another one isn’t yet addressed well, and that our uncertainty margins should be adjusted for *future* predictions….

But the record of warming so far (faster than we predicted across the board), as well as massive biotic changes already underway, ocean acidification, etc. these aren’t debatable. They’re just happening, and there’s loads of scientific data to show it. Pretending global warming isn’t happening is just flat out ignorant at this point, from a scientific perspective. I don’t mean that offensively, just as a straight statement of reality. The GOP will do itself a great service to quit trying to die on this hill.

Yes, you’re correct – they backcasted the previous elections using the appropriate data for each. I should have said projection rather than prediction. I’ll still bet on their model over the currently schizophrenic polls which in recent years haven’t been very accurate over time.

I pretty much agree with your general gist – but have to say that it’s not an easy thing to create a model that accurate backcasts complicated systems. That’s why the Univ. of Colorado model is so promising. Time will tell, however! I just hope their model actually pans out, and we see a decisive win for Romney, contrary to what so many polls suggest. The xkcd cartoon is a good one, but isn’t really applicable to this issue.