Sign up FAST! Login

Nate Silver Pundits, Politicians

Stashed in: Politics!, Awesome, Math!, politics, Nate Silver

To save this post, select a stash from drop-down menu or type in a new one:

Because Nate Silver proves pundits provide absolutely zero value to the political system, and thought politicians can't talk their way out of an algebra class.

Nate's Blog has become the only place on the Internet I trust for 2012 Election insights:

He accurately predicted one wave election in favor of his favored candidate. Some of his raw analysis may be sound, but his conclusions are suspect. We shall find out in a week, but imo, his view that 2008 created a persistent, sustained D+~8 is absurd.

I'm looking forward to seeing in a week how accurate he is, too.

If he does succeed in calling it, does that mean anything?

it means the subsequent GOP civil war will be epic. :-)

But here's what we will know:

Ohio polls close at 730 Eastern. IF exits call _anyone_ immediately, the election is likely over, since the exit numbers in the target precincts will match or exceed the pre-election crosstabs by demo. If there's any contention between news orgs over whether or not to call Ohio, then chances are the early call is ignoring or discounting early vote polls and things could get hinky later in the evening.

Polls close in Florida at 7p. However, the Florida Panhandle is in the Central tz. In 2000, CNN and other outfits called Florida for Algore at 7pm Eastern, causing ~10-12k people in the panhandle to not vote (You all prolly didn't know that did you? That whole thing was shady as hell). If any major news outlet calls Florida when the EST polls close, you know the fix is in since Florida is a presume-win state for Romney and Romney voters in those Central tz states which are open later (MO, IA, WI, and MN) are less likely to come out for the final 2-3 hours of voting. If they don't call Florida by 10p or so, then Romney has probably lost the election.

Polls in VA also close at 7p. If it gets called quickly for Obama, then it's likely over for Romney. If it's called quickly for Romney then the D+3 faction among pollsters will be correct over the D+8 faction and it will probably be a good night for Romney. If they call Kaine quickly over Allen, but not POTUS, then VA will likely be a narrowish victory for Obama. If they call Allen over Kaine, then Romney will likely win the state comfortably, but we won't know until later.

Polls close in PA and MI at 8p. If either state gets called early when Ohio is still contentious, then we'll have a never-before Ohio outlyer. MI is in the same situation as Florida, so an 8p EST call would be another shady situation, tho the CST parts of MI are very sparsely populated and tend to lean Democrat. If either PA or MI gets called for Romney, the show's over and Romney wins. If PA doesn't call quickly, but they do call the Senate race for Smith over Casey, then the white working-class demo is leaning hard for Romney and he will likely carry the night.

If none of these states call, then we're in for a long, long night.

WI, MN, and IA only matter if they call for Romney quickly. We will likely know how the night will play out, whether Obama, Romney, or nailbiter, before these states (or any further west) close.

So if I understand correctly, Florida will likely go to Romney, and Pennsylvania and Michigan will likely go to Obama.

This leaves five key toss up states: Ohio, Virginia, Colorado, Iowa, Wisconsin, Nevada, and New Hampshire (which is likely too small to matter).

I didn't realize 10k panhandlers didn't vote in 2000. Super shady!!

Jason, he correctly picked 49/50 states and all 35 senate races. The favored candidate comment is a non-sequitir; what does his "favored" candidate have anything to do with his predictions being 98.8% accurate?

Was there any pundit, blogger, or statistician who has more accurately predicted the outcomes of the 2008 election?

Again... it was a wave election and his analysis largely consisted of applying 'universal swing' (iirc, about D+11) to the demographic crosstabs. Universal Swing has been a common analytical technique in parliamentary democracies but tends to break down here, due to the fact that our culture is far from homogeneous and the electoral college creates discreet battlegrounds of intense political combat.

People forget that Nate Silver did not invent some new analytical technique for political polling, he simply subscribes to the pollsters' more detailed data and blogged about his predictions in an analytical-seeming format with more graphics. Every pollster does the exact same thing.

Example: Zogby was right on the money in 2004 (i think it was 2004 anyway) and he was lauded as the second coming as much as Silver is now.... but Zogby's been way off quite a bit lately.

Well I, for one, can't wait to see how this all turns out.

That's like saying people forget Howard Dean used the internet effectively 4 years before Pres. Obama -- then-Senator Obama -- did; it's not particularly relevant who was *first*, what's relevant who does it, and who does it *best.*

Silver is popular for the same reason Malcolm Gladwell and Michael Lewis are; they help tell simple stories out of all the complex data people are seeing, and as we know, simple is actually incredibly hard.

The average interested politically minded citizen is going to hear results from perhaps a dozen polls a week; Silver makes sense of all of that, and that's an incredibly useful service.

I can clearly see your political view; however, at the end of the day -- who do you think, today, right now, is likely to win the presidency?

Who? I don't know. I think the fundamentals are better for Romney, and the 47% has been a consistent ceiling for Obama, but it's essentially a tie so it's going to come down to the demographic turnout targets. Also, even tho the politicos are kind of ignoring Sandy, I think it's going to play a part.

There's one variable in the polls which is seriously up in the air: cell phone only voters. Cell phones can't be autodialed (well they are, but a human has to be on the other end, not an IVR), but autodialed polls have fewer subjective variables related to voice delivery (If you've ever worked a phone bank, you know how much the person on the dial matters in response).