Nate Silver

If only media had used real big data for the US Elections

98433914051b48b09377bdca4b8f91414e5b1f5beb4018137e8248a004248f44

The chart above was highlighted to me by a friend (who is brilliant at statistics) two days ahead of the election. She (and her husband) have been saying for months Donald Trump would win.

But, everyone (including me) ignored really big data, instead preferring to focus on single polls, and the data expertise of Nate Silver. My gut feel told me it was harder to pick than a broken nose, but these two were adamant, Trump was going to win. BTW they are the same two who in 2014 on election night called the election 3 hours before media did…no one will know that except the WO team, but they did.

Now, Nate Silver does use big data, but his data is just the polls all across the country, plus some demographic information and his proprietary algorithms and assumptions. It is all well and good, but when it fails people start questioning polls, so-called big data and ignoring data sets that are right in front of you. Instead of the wisdom of thousands of polls and dozens of pollsters, you are using the wisdom of search engines and millions upon millions of searches.   Read more »

Every poll still gives it to Hillary, so why is Nate Silver more cautious?

I’ve always said this election is harder to pick than a broken nose and it looks like Nate’s nose is well broken.

Polling averages and forecast models are supposed to bring order to the chaos, put outlier polls in proper perspective and provide a sober, unbiased picture of the state of the presidential race.

So why are they all over the place in the final days, with some models asserting a Hillary Clinton victory is a near-certainty, and others giving Donald Trump a real chance at winning?

The dissonance has resulted in fundamental disagreements over how close the race really is in a presidential election that both parties have described, in no uncertain terms, as the most important election in our lifetimes.

The latest RealClearPolitics projections have the race teetering on a knife’s edge. But other, more complex forecast models — based on the same polls — give Clinton a 98 percent or 99 percent chance of defeating Trump next Tuesday.

This is why you need people to trust.  People who have delivered extraordinary accuracy in the past.   Read more »

Tagged:

Live Polls And non-Live Polls yield different results

With a backdrop of companies struggling to get both consistent and credible polling, it is interesting to analyse how different polling methods can skew the answers.

FiveThirtyEight generally takes an inclusive attitude towards polls. Our forecast models include polls from pollsters who use traditional methods, i.e., live interviewers. And we include surveys conducted with less tested techniques, such as interactive voice response (or “robopolls”) and online panels. We don’t treat all polls equally — our models account for the methodological quality and past accuracy of each pollster — but we’ll take all the data we can get.

This split, however, between live-interview polls and everything else, is something we keep our eye on. When we launched our general election forecasts in late June, there wasn’t a big difference in the results we were getting from polls using traditional methodologies and polls using newer techniques. Now, it’s pretty clear that Hillary Clinton’s lead over Donald Trump is wider in live-telephone surveys than it is in nonlive surveys.

We don’t know exactly why live-interview polls are getting different results than other types of surveys; there are a lot of potential causes and it’s something we’ll be digging into.

Online polling will become preferred due to the relatively low cost.  But unless the results are of a reasonable quality, live interview polls will remain the more credible.  We saw this recently when some rag commissioned a poll by a never-before-heard-of US company who had never polled on politics New Zealand previously.  The results were ridiculous.  Read more »

Which pollsters can you trust, and which are bought and paid for?

We should have something like this here.

CC001b20160704

In the process of being an analyser of polls, Nate Silver has had to figure out which are actually delivering reliable and honest results, and which are essentially fronts that get you the answer you’re paying for.

We know that Colmar Brunton are one of the most variable and have a clear bias towards Labour, for example.

As for Horizon, they had the Conservative party in government as part of the coalition with National right now.

Whenever the Fraser House spy tells me that Labour are polling in the low to mid-thirties, and then paid-for polls come out, you can reliably subtract 4-6 points, every time.

 

– 538

Traditional political models predict Trump will absolutely tank

As an analyst of US politics, Nate Silver has been uncannily accurate and able to embarrass the pros.

CC001820160630 CC001c20160630

CC002c20160630

Read more »

Nate Silver on why Republicans backed Trump

Nate Silver writes about the Trump phenomenon.

It is a great article from a writer who is the best predictor in politics.

If you’d told me a year ago that Trump would be the nominee, I’d have thought you were nuts. Don’t just take my word for it: Read what I wrote about Trump in July or August or even in November. Those pieces variously treated Trump’s nomination as being somewhere between improbable and extremely unlikely. You can also read pieces from October, December orJanuary that were less skeptical of Trump’s chances and show how our opinion of him evolved over time. Still, other than being early skeptics of Jeb Bush, we basically got the Republican race wrong.

Nice to see a pundit admit he was wrong. But how did he go wrong?

For a candidate like Trump to win the nomination, it means that several things have gone wrong — both for the Republican Party and in the assumptions we made about how party nominations work. The other day, I summed up the three most important such factors as follows:

Screen Shot 2016-05-05 at 2.15.25 PM Read more »

Trump now has a path to victory ahead of the convention

Yesterday Donald Trump cleaned out 5 state primaries and pretty much hoovered up all available delegates.

He now sits on 950 delegates, just 287 short of the required 1237.

Nate Silver now has him ahead of predictions and now in with a shot of making the 1237 threshold:

Tuesday night went about as well as possible for Donald Trump.

Two weeks ago, after a rough stretch of states for Trump, we issued a series of delegate projections that included something called a “path-to-1,237” projection, a set of targets that would allow Trump to clinch a delegate majority without having to rely on uncommitted delegates. With Trump’s terrific results in New York last week and even better ones in the five states that voted on Tuesday, Trump is actually running a little ahead of the path-to-1,237.

Based on provisional results, it looks as though Trump will sweep every pledged delegate in Maryland (as a result of winning every congressional district), Connecticut (as a result of winning every congressional district and getting more than 50 percent of the vote statewide), Pennsylvania (where statewide delegates are awarded winner-take-all) and Delaware (ditto), along with 11 of 19 delegates in Rhode Island (which is highly proportional). Combined with the New York results,2 that gives Trump 200 delegates since we issued the path-to-1,237 projections, five delegates ahead of his original targets.

Read more »

Trump will need to win on first ballot

If the Republican nomination fails to reach the threshold for delegates before the convention then it will become a contested convention.

If that happens, and there is a good chance that it will then Donald Trump will need to win on the first ballot.

Nate Silver at FiveThirtyEight explains:

If you really think the chance of a multi-ballot convention is 63 percent, but also still have Trump with a 56 percent chance of winning the nomination, that implies there’s a fairly good chance that Trump will win if voting goes beyond the first ballot. That’s probably wrong. If Trump doesn’t win on the first ballot, he’s probably screwed.

The basic reason is simple. Most of the 2,472 delegates with a vote in Cleveland probably aren’t going to like Trump.

Let’s back up a bit. In most of our discussions about delegates here at FiveThirtyEight, we treat them as though they’re some sort of statistical unit. We might say a candidate “racked up 44 delegates” in the same way we’d say Steph Curry scored 44 points. But those delegates aren’t just a scoring mechanism: Delegates are people, my friends. Delegates are people!

And as I said, they’re mostly people who aren’t going to like Trump, at least if the excellent reporting from Politico and other news organizations is right. (If Trump turns out to have more support among GOP delegates than this reporting suggests, even marginally, that could end up mattering a great deal.) How can that be? In most states, the process to select the men and women who will serve as delegates is separate from presidential balloting. In Massachusetts, for instance, Trump won 49 percent of the GOP vote on March 1 — his highest share in any state to date — to earn 22 of the state’s 42 delegates. But the people who will serve as delegates haven’t been chosen yet. That will happen at a series of congressional district conventions later this month and then a Republican state meeting in May or June. According to Politico, most of those delegates are liable to favor Ted Cruz or John Kasich rather than Trump. Twenty-two of them will still be bound to Trump on the first ballot, but they can switch after that. The same story holds in a lot of other states: in Georgia, Louisiana and South Carolina, for instance — also states that Trump won.   Read more »

Hacking the system: how Trump took the lead

Nate Silver writes at FiveThirtryEight about how Trump hacked the system.

As is usual with Nate Silver there is plenty of data and explanations before he gets to his conclusions:

Most of the media’s self-criticism of its Trump coverage has focused on whether Trump’s dominance of the news cycle reflects a craven desire for higher TV ratings or Web traffic numbers. It’s fine to debate that — although these criticisms are sometimes being evinced through crocodile tears given the record ratings and traffic Trump is bringing to news organizations of all kinds.12

But this critique avoids some thornier questions. For instance, with his ability to make news any time he wants with a tweet, news conference or conveniently placed leak, Trump has challenged news organizations’ editorial prerogative. Should the press cover a candidate differently when he makes trolling the media an explicit part of his strategy, on the theory that some coverage is almost always better than none?    Read more »

Trump’s path to victory

There are a great many people speculating on how Donald Trump can win.

I prefer data, which is why I read FiveThirtyEight for my information. Subscribers to INCITE: Politics will know that I recommend this site to everyone interested in knowing what is going on in US politics.

They explore the path to victory for Donald Trump, or at least get close enough to make a claim for him to be the candidate.

Some of their key points are:

harry (Harry Enten, senior political writer): My biggest takeaway is how on the knife’s edge this is going to be. We probably won’t know whether Trump is going to clear the 1,237 threshold until at least June. It’s March 21 — we have a long way to go.

dave (David Wasserman, House editor at the Cook Political Report and FiveThirtyEight contributor): First off, while we don’t know whether Trump will hit 1,237 or not, we should all be able to agree on one takeaway: For the first time in a very long time, every state will matter — and yeah, this thing’s going all the way to June. I don’t see any way for Trump to attain 1,237 untilJune 7, and I don’t see any realistic way for him to be mathematically eliminated from 1,237 before June 7.

natesilver (Nate Silver, editor in chief): When I was filling out my projections, I was a little bit surprised that I had Trump coming up short of 1,237 since I thought I was being fairly optimistic for him in individual states. However, once you account for the facts that (1) there are quite a few uncommitted delegates and (2) Trump isn’t likely to do so well in less populous states west of the Mississippi, which is a fair bit of what’s out there — well then, Trump has to do really well everywhere else.

Read more »