Autoplay: ON Autoplay: OFF
The significance of Nate Silver's numbers: Inside Scoop
Inside Scoop: The significance of Nate Silver's numbers5:06 /
CNET's Sumi Das talks to Senior Writer Daniel Terdiman about how the statistician comes up with his presidential forecast.
-Hello and welcome to this Election Day Special of Inside Scoop, as if you can't tell from the stickers. I'm Sumi Das and with me is Daniel Terdiman, Senior Writer for CNET, also supporting a sticker. Well done, you voted. So, we're gonna talk today about Nate Silver, who has a blog, 538 blog, which is also featured on New York Times' site. Tell us about Nate Silver and what he does. -So Nate, he's been doing this since 2008 before he worked for the New York Times. What he does is he essentially takes all the available polling that's out there and he aggregates that income. This is a pretty sophisticated, you know, algorithm to wait the polls and to figure out which ones are kind of, you know, have more relevance or more accurate and then, and sort of jumps them off together and spend more or less-- -Pretty complex formula, right? -Yeah, and then what he does-- this is sort of the most important part of what he does is he runs 100,000 daily simulations of the election based on all his data, and then spits out probabilities of what's going to happen. -Okay. And he takes polls from I would imagine the [unk] from left to right, from liberal thing takes to conservative outlets, right? -Right. Well, yeah, and like I say, he takes any available polling but he's a [unk] and he takes each poll on kind of a science that weight based on its recency, based on that polling firms, historical accuracy, you know, and then, you know, [unk] things like, you know, what the demographics are in a particular state and so you can't say that one poll is automatically, you know, weighed the same as another poll. And he's kind of built the system around, figuring out the sort of variances and that, and he's proven in the past, it'd be pretty accurate. -Okay. So he's got a good track record. -He does have-- -And what does he predict for this election? -Well, I mean-- -'Cause we're all too impatient to just wait 'til the end of the day. -Right. So, here we are in Election Day, the last I checked he was predicting 91% chance that President Obama would be re-elected with about 313 electro votes and you need 270 electro votes to win. So, but the thing that's gotten him a lot of attention is that, you know, a couple of weeks ago when you remember that after the first presidential debate, President Obama's poll numbers started to drop. -It did. -Governor Romney's numbers started to go up. -Right. -And then a lot of the polling was actually showing Governor Romney ahead or you know, very close and so that's sort of the media narrative at that point was this race is too close to call, there's no way to tell. Maybe even Romney's ahead 'cause Obama had been ahead all the way along. -Right. -Suddenly, Romney might have been ahead after that and Nate Silver at FiveThirtyEight was saying, "No, my numbers still show that Obama is safely ahead." And the reason is it goes down to the way that our presidential elections work is because it's actually not a national election. It's 50 state elections plus a District of Columbia, that's how the electoral college works. -Right. -And he has shown that Obama has been ahead in Ohio all along and Ohio is the state that's gonna make the difference this year. -All blows down to Ohio. -It all blows down to Ohio and so-- -Into a lesser degree of Florida? -Not-- Florida is not sort of key this year. -Up in-- okay. -Yeah. He has this thing called the tipping point states, that's a graph and you know, it shows which state is most likely to be the tipping point in the election and Ohio is like an order of magnitude above than the other state. -So, does that explain what you just said, that explains why we see all these other polls that are saying, you know, oh it's a dead heat. It's 47% and 48%-- -Exactly. -and so closely tied and it's too close to call. -Right. And the reason that those polls are those-- that narrative has happened is because those people are focusing really on the national polling only and they're not looking at the state polling. -Okay. Does this represent sort of a shift away from traditional polls? The attention that Nate Silver is getting over the past four years now, the last two elections, away from you know, the Gallup polls, a few research center polls, that's sort of-- -I think what it does is it changes the narrative, so you can't focus on anyone poll and it used to be that Gallup was the standard, right? -Right. -And then you-- oh, Gallup, you know, desperate candidates ahead in the Gallup poll therefore that's probably what's gonna happen. Now, I think what Nate Silver and other people should be noted he's not the only one who has-- -Sure. -sort of an aggregator model. What they're trying to say is you can't focus on any one poll because any one poll can be an out wire. And sort of the theory there is the five polls say this, and one poll says this, even if it's Gallup, you can't necessarily trust it. You gotta-- you can only kinda look at where the other ones are all grouping together. -Okay. -But I think at the same time I think it's actually put more attention on polling. I think people pay more attention to the idea of polling and I think that's kind of exciting. It's a horse race kind of thing. -Yeah. People are definitely glued to their computers. -Yeah. -Refresh, refresh. -Absolutely. -Daniel Terdiman, thank you so much. -Thanks a lot. -And you haven't done so already, go get your sticker. For Inside Scoop, I'm Sumi Das. Thanks for watching.