Friday, January 20, 2017

What Went Wrong?

Over at the 538 blog wunderkind Nate Silver is reassessing his election forecasts. Silver did not get it as wrong as many other statisticians, but still his 30% number for Trump’s chances was off the mark. His and many other pollsters’ predictions of the aggregate vote were close to correct. Since the election was not decided on the basis of raw popular vote, the numbers were irrelevant.

Now, Silver is going to publish a series of articles explaining what went wrong… not just for him, but also for other political commentators. The Trump/Clinton election stands out as a shining instance of confirmation bias. See my post two days ago..

In this paragraph he offers a fine synopsis, or, if you prefer, a diagnosis of media confirmation bias:

Why, then, had so many people who covered the campaign been so confident of Clinton’s chances? This is the question I’ve spent the past two to three months thinking about. It turns out to have some complicated answers, which is why it’s taken some time to put this article together (and this is actually the introduction to a long series of articles on this question that we’ll publish over the next few weeks). But the answers are potentially a lot more instructive for how to cover Trump’s White House and future elections than the ones you’d get by simply blaming the polls for the failure to foresee the outcome. They also suggest there are real shortcomings in how American politics are covered, including pervasive groupthink among media elites, an unhealthy obsession with the insider’s view of politics, a lack of analytical rigor, a failure to appreciate uncertainty, a sluggishness to self-correct when new evidence contradicts pre-existing beliefs, and a narrow viewpoint that lacks perspective from the longer arc of American history.


Ignatius Acton Chesterton OCD said...

"Now, Silver is going to publish a series of articles explaining what went wrong… not just for him, but also for other political commentators."

Wow, a "series of articles" to say something very simple: mainstream Big Media news organizations are the broadcast arm of the Democrat Party, and their journalists are wannabe Soviet political officers.

Not too hard to 'splain, Nate.

Sam L. said...

IAC, the media does confirm the Dem bias. Also, they live in Dem cities, and pretty much stay there. I suspect many in Flyoverlandia won't answer a call from a pollster, and if they do, might well lie to them.

Ignatius Acton Chesterton OCD said...

Sam L. @January 20, 2017 at 8:59 AM:

Bias? What bias?

Anonymous said...

Trigger Warning said...

While I don't disagree with anything Silver said, I think he's foolish to be soul-searching over his numbers. A 0.30 probability is just that. One out of 3 times, the low probability event will occur. What's the BFD? The weatherman predicted a 30% chance of rain, and it rained. So? Anybody who thinks that 30% means it won't rain doesn't understand probability.

A more relevant question is whether he can do what he claims to be able to do with the statistical tools he uses. And the answer is no. And no amount of hand wringing, garment rending, or invoking the malign spirit of "confirmation bias" is going to fix that. Of course, he'll still be right now and again. Everybody is.

If we could predict the future of ridiculously complex systems we'd be like gods and Genesis would have turned out differently. For Pete's sake, Poincaré showed there's no general solution for the Three Body Problem in physics.

Steve D said...

'but still his 30% number for Trump’s chances was off the mark'

I disagree. That number means if you ran the election 100 times, Trump would win 30 times. We only ran the election once. We need to go back in time and run it another 99 times to test Nate Silver's model.

It was a very close election. 100,000 votes over three states swung it. So Nate may have gotten it right.

Ignatius Acton Chesterton OCD said...

Trigger Warning @January 20, 2017 at 12:17 PM:

Absolutely spot on.

Statistically, the 1980 USA Olympic Hockey Team should never have beat the Soviets. But they did.

Treating everyone as a statistic is inhuman. Treating every unique human event as a statistical probability is stupid. And looking at the world that way gives you a false confidence in your fellow man and demeans his achievement. So it only happens once every x times? The point is that it happened.

And the elite media types imagine themselves as being so smart and polished because they've read a few Malcolm Gladerll and Michael Lewis books. No wonder they are self-congratulatory and morally magnificent. They think life is a big game. Well, perhaps it is... if you play the odds. And by definition, we all have to play.

Nate got it wrong. Writing about it won't change anything. He's just another chattering soul impressed with his own past success. Well, we live in the present. And president Trumpnis our president. And no amount of analysis or statistical deconstructing or multiple multiple regression analyses are going to change that.

It would be nice if everyone got more aligned with the reality of the world in which we live. Right now.

I await Ares Olympus' doubling down on the relevance of his imagination, predictive faculties or his... clairvoyance. After all, he called so many of us fools for months, weeks and days.

Ares Olympus said...

IAC: I await Ares Olympus' doubling down on the relevance of his imagination, predictive faculties or his... clairvoyance. After all, he called so many of us fools for months, weeks and days.

Oh, looks like I'm late to this party, but I'm game.

I did bet against Trump's election, but it was always clear that new voter turnout in keys states could flip the election, and here Trump ended on the up side of that with a net 71,000 vote victory in 3 key states that Obama won, and Jill Stein from the Left broke the margin in all 3 states, although Johnson controlled the margin even more, and surely stolen more Republican votes than Stein stole Democratic votes.

I was worried through out October that Trump could win, and there were a good number of midwestern state scenarios that looked terrifying, and we found out that Michael Moore's Trumpland had it right.

My "what if" game wonders how many people had regrets, either Stein voters who disliked Trump more than Clinton, or citizens who would have preferred Clinton over Trump but didn't bother to vote. But I'm willing to say many of those people can "rationalize" that they disliked BOTH, and don't feel responsible for either. And perhaps if we had an imaginary revote, perhaps Trump would win a bigger margin, because more people see a path forward, change at any cost?

As I've said, I really hated the idea of another 4 years of a democratic president and Republican House and Senate, so it does look better to give one party complete power and responsibility for results, and let them collapse under their own hubris. And although the Democrats picked up a few House and Senate seats, the American people have picked Republicans to express their hubris, just in case they know what they're doing.

Myself, I pity the Republicans with Donald Trump at the helm, and would consider him an independent as much as anything, except for the fact that he's over his head, and he's handed most of the power to the proper ideological conservatives so far.

Yet, it's not clear what comes out of this. And it still looks like Trump will act badly, and pick a fight he can't win with his own party, and they'll pull out their impeachment folders they're carefully preparing, and start the proceedings. I mean that's what I would do in this situation - document careful evidence of Trump's illegal activities, and even provoke him to do rash but harmless acts, like lying under oath, to get the technicalities needed.

But Trump's a fighter if he's anything, and given we now have armed men going into pizzarias looking for Hillary's children prostitutes, you have to think there's dozens more who would start a revolution to protect their King Trump from his own party's attempt to control him.

And somehow 63 million Americans are willing to bet their lives will be better off because of these confrontations.

It's better show when you can get out of the way, and let everyone else pay the price. But I have to think it is resentment winning the day, and a poorer America is now considered superior to a more unequal America which is the direction ordinary politics sends us which ever party is in charge.

I guess what bugs me is how people can claim they voted for Trump because it'll help us "win", while all evidence suggests at best it'll expose our vulnerabilities to division, and test the limits of our institutions to keep order.

So we've voted for chaos, and we've got 4 years. And it looks like a great time for a food fight for all the bullies in the world, to dig into their own populations and their weaker neighbors. America is no longer the police man of the world because we can't even police ourselves.