Wednesday, February 08, 2017

Nate Silver still doesn't get why he blew 2016

Nate Silver gave President Trump a 2% chance of being nominated -- and then on Election Day, a 28% chance of winning the presidency.

We know why he was wrong. But let's not spoil the fun. Let us look at the reasons he says he blew it.

In a six-part series beginning January 19, Silver told us, “The Real Story Of 2016.”

Yeah, right.

The Prologue:
While data geeks and traditional journalists each made their share of mistakes when assessing Trump’s chances during the campaign, their behavior since the election has been different. After Trump’s victory, the various academics and journalists who’d built models to estimate the election odds engaged in detailed self-assessments of how their forecasts had performed. Not all of these assessments were mea culpas — ours emphatically wasn’t (more about that in a moment) — but they at least grappled with the reality of what the models had said.
By contrast, some traditional reporters and editors have built a revisionist history about how they covered Trump and why he won. Perhaps the biggest myth is when traditional journalists claim they weren’t making predictions about the outcome. That may still largely be true for local reporters, but at the major national news outlets, campaign correspondents rarely stick to just-the-facts reporting (“Hillary Clinton held a rally in Des Moines today”). Instead, it’s increasingly common for articles about the campaign to contain a mix of analysis and reporting and to make plenty of explicit and implicit predictions.
He claims that "some traditional reporters and editors have built a revisionist history about how they covered Trump and why he won" -- but is revising his own history of error compounded by error.

However, he ended that piece:
The cognitive biases reflect more deep-seated problems and have more implications for how Trump’s presidency will be covered; they’re also the root cause of some of the technical errors. But they won’t be easy to correct unless journalists’ incentives or the culture of political journalism change.
Yep, the same biases and self-interest (many corporations including Disney, owner of ESPN, which owns Silver's site) affect adversely coverage of President Trump just as those prejudices adversely affected coverage of his candidacy.

In Part I, Silver wrote:
FiveThirtyEight’s statistical model, for example, saw the Electoral College as a significant advantage for Trump, and projected that he’d be about even money to win the Electoral College even if he lost the popular vote by 1 to 2 percentage points. Overall, it assigned a 10.5 percent chance to Trump’s winning the Electoral College while losing the popular vote, but less than a 1 percent chance of Hillary Clinton’s doing the same.
So his computer model was 89.5% wrong.

It gave Trump's chances of winning without winning the popular vote 10.5%.

Which is what happened.

Imagine a doctor chopping off the correct arm only 10.5% of the time.

And where do these tenths of a percent come from?

Part II:
If the “emerging Democratic majority” was one pillar of the flawed argument that Hillary Clinton had an Electoral College advantage over Donald Trump, the other was the “blue wall,” the claim that Democrats began with a base of 242 electoral votes because they’d won them in each election since 1992.
Yep.

Guess what, in 2000 just about everyone missed West Virginia flipping Republican. Without those five Electoral College votes, it would have been President Gore.

So Silver caught on?

Nope. There is this:
But Clinton faced more headwinds in 2016, trying to win a third consecutive term for her party amid a mediocre economy. Against a “generic” Republican such as John Kasich or Marco Rubio, she might have been in a toss-up race or even a slight underdog, in fact. So she was counting on good economic news — or for Trump to underperform a “generic” Republican because of his unique flaws as a candidate.
But a generic candidate -- whom I derided in the campaign as Jebbio McCruzney -- would not have had the appeal in Michigan, Pennsylvania, and Wisconsin that Trump had.

We know this because even with Paul Ryan of Wisconsin on the ticket, Romney failed to carry any one of those states.

But hey, let Silver keep telling himself the Republican Establishment in 2016 would have done better than it did in any election since Reagan was president.

Remember, Reagan was president in 1988. He was the reason Bush 41 won. Trump's 30 states and 304 Electoral College votes were the best Republican showing since 1988.

In Part III, Silver wrote:
The undecideds were a warning sign that Clinton hadn’t sealed the deal with quite a wide enough coalition of voters, conversely — especially in the Midwest where undecideds were plentiful. In the states that were the biggest upsets relative to the polls — Michigan, Wisconsin and Pennsylvania — Clinton met or slightly exceeded her share of the vote in polls, but Trump beat his by more.
Yes, that was a warning sign. Ignored. I am curious as to how a computer model misses that. I mean, I can see a human being missing it, but if his much vaunted computer model is so darned scientific, how could it miss that? Hindsight may be 20/20 but foresight also can be 20/20 as well, and is expected of any computer model.

Unless that computer model is garbage, which in this case, Silver's is.

In Part IV, Silver wrote: "Ohio Was A Bellwether After All."

In the past century, Ohio missed it but one time. And my readers know that I was among the few who still said So Goes Ohio, So Goes the Country.

In Part V, Silver wrote: "Why Early Voting Was Overhyped."

There did seem to be a last minute swing toward The Donald (as there usually is for the underdog in the presidential race) that overcame that early edge.

Finally, in Part VI, Silver wrote:
Error no. 1: Clinton focused too much on close states rather than tipping-point states. On average from Sept. 1 through Nov. 7, the closest states in our polls-only model were Ohio, Nevada, North Carolina, Iowa and Florida. And yet Clinton spent more time in these states than she should have. A combined 54 percent of Clinton’s events (and 47 percent of Trump’s) were held in these states, whereas there was only a 39 percent chance that one of them would be the tipping-point state.
And later:
Error no. 2: Clinton was overconfident and campaigned in too narrow a range of states.
His distinction between close states and "tipping-point states" is a puzzler.

But it made sense for the candidates to battle it out in battleground states, with Colorado, Florida, Iowa, Nevada, New Hampshire, North Carolina, Ohio, and Pennsylvania being the primary ones.

However, once Trump sewed up Iowa and Ohio, he could campaign in the others. He seemed to write off Nevada and New Hampshire, favoring Michigan and Wisconsin.

I do not really fault Hillary for where she campaigned. Had she taken Florida and North Carolina, she would be president, even without Michigan, Pennsylvania, and Wisconsin.

Her plan was fine. But plans are the first thing lost in a war. She failed to turn her Titanic around when things closed late in Michigan and Wisconsin.

While Silver called the series "The Real Story Of 2016," in the prologue and first five-and-a-half parts, he avoided the true story, which was if not unholy, then unhealthy alliance of the press and the Clinton campaign.

Coverage of her campaign was too soft -- often little more than an in-kind campaign contributions by three major corporations -- Comcast (owner of NBC/MSNBC), Disney (ABC), and Time-Warner (CNN) -- and the lesser ones that own newspapers.

All the positive press on her, and all the negative on her opponent gave her a false sense of security.

Silver finally got around to this:
To some extent, the media’s misconceptions about Electoral College strategy and Clinton’s errors may have reinforced one another. For the most part, decisions about where to allocate resources should be determined by where states line up relative to one another. Typically, news events produce similar changes in lots of states at once, so even a major shock (say, Comey’s letter or the release of Trump’s “Access Hollywood” tape) won’t change the correct Electoral College approach all that much.
It's just a slight acknowledgement that maybe a steady diet of media cotton candy coverage is not the best thing for a dull candidate slogging her way to victory.

But he misses the Real Story of 2016, which simply was a rebellion by voters against the Establishment in Washington. It was a trouble brewing since Reagan left office that had led to the Perot candidacy in 1992, the Gingrich Revolution in 1994, the Obama presidency in 2008, and the Tea Party in 2010.

The press missed three of the four.

What does that tell you about the worthiness of the media?

By refusing to admit his computer model was wrong -- he gave only a 10.5% chance to the scenario that actually occurred (a Clinton plurality, a Trump victory) -- Nate Silver is not helping himself repair his tattered reputation.

@@@

"Trump the Establishment" is an account on how the Establishment used the media to stop the Trump Train in the general election by constantly badgering him on personal issues, often petty. But Trump prevailed based on his abilities as a CEO, and his stances on the real issues of the economy and national security, which I also delve into.

The book is available in paperback Create Space, or if you prefer, or (via Instapundit) Amazon as a paperback.

Kindle will be available March 1.

This is the sequel to "Trump the Press," which covered the nomination. It is available on Kindle, or in paperback on Create Space.

21 comments:

  1. Silver is a Lefty shill and as crooked as a dog's hind leg. He tried to stack the deck and now he's trying to save his credibility.

    ReplyDelete
  2. I'm not as harsh on Silver as a lot of people; he's simply in denial over this election. I was a Cruz guy initially, but when I noticed that many of my blue-collar, traditional democrat family members (union members) were on the Trump train, and on it hard, I knew something was up - and I knew a whole lot of people were missing it (God forbid the pollsters get down in the muck). And Nate needs to realize what everyone knows when it comes to computer models: GIGO.

    ReplyDelete
  3. Silver was in denial before the election, and will still be in denial in 2018 and 2020.

    ReplyDelete
  4. It was crystal clear to me as early as 2013 that the polls were skewed. This from a 2012 Pew report:

    "The percentage of households in a sample that are successfully interviewed – the response rate – has fallen dramatically. At Pew Research, the response rate of a typical telephone survey was 36% in 1997 and is just 9% today."

    http://www.people-press.org/2012/05/15/assessing-the-representativeness-of-public-opinion-surveys/

    People just aren't answering the phone. It is impossible to get a representative sample, for polling purposes, with that kind of response. The big question is the reason that people aren't participating. We still have no clue as to why but we can reasonably say that Trump supporters were less likely to participate.

    The same problem manifested in prior elections where the exit polling didn't come as close to the final tabulations as the pollsters expected. Some people stopped to answer the questions, others didn't.

    The pollsters like to say "well, we came within 2%". This is very misleading. If you blindly took the average of the last four elections you'd likely be within 2%. Given the time, money and hype they didn't miss by 2%. More like 50%.

    ReplyDelete
  5. Cognitive biases, tipping-point states, blah blah blah. Talking jargon while making excuses doesn't impress me. Maybe Silver's model and assumptions are not as good as he thinks they are. Sounds like he is setting himself up for another schlonging in 2020.

    ReplyDelete
  6. "But he misses the Real Story of 2016, which simply was a rebellion by voters against the Establishment in Washington. It was a trouble brewing since Reagan left office that had led to the Perot candidacy in 1992, the Gingrich Revolution in 1994, the Obama presidency in 2008, and the Tea Party in 2010."
    This is what I think is the biggest insight of this column and perhaps the key to the entire Trump campaign. I doubt that this was overlooked by the crew in Trump Tower, either. Historians of the history of ideas are sure to see the continuity you picked up on. And if the dems and the establishment Republicans had been paying attention we wouldn't have this "rube", this " con man" who just happened to be smarter than every last one of them deciding policy.

    ReplyDelete
  7. The Trump people I talked to at CPAC 2016 all basically said the same thing: We Got This. And they weren't talking about the primary! Nate Silver can blah blah blah all he wants, but until he spends a couple hours at Traver's Country Store in Hedgesville, WV, he will continue to misread the mood of the country.

    ReplyDelete
  8. Silver seems to be using a different system of arithmetic to the rest of us ... "Nathematics"? ... but I rather get the impression that he can't count to twenty-one without taking off his shoes and dropping his pants.

    ReplyDelete
    Replies
    1. Nathematics -- I am so stealing it

      Delete
    2. If Nate Silver takes off his shoes and drops his pants, I still don't see how he gets past 20.

      - Mark S.

      Delete
  9. Doctors do post-mortems to learn from Nature and from their mistakes. Pollsters do post-mortems on their mistakes to salvage their reputations and their careers. I've got to hand it to the tarnished Silver: he has good 20/20 hindsight.

    ReplyDelete
  10. Wish I'd said it first, cuz it's classic comedy:

    "The reason he's called Nate Silver is because his candidates always come in second."

    ReplyDelete
  11. From the beginning " .... they at least grappled with the reality .... ". Wha?? They didn't recognize reality, from the start.

    He also seems to hint that the fault is Clinton's, not his models, because she underperformed.

    ReplyDelete
    Replies
    1. I don't care whose fault it is, as long as Clinton lost. For that I am grateful.

      Delete
  12. Just like the sports, people try to model real world conditions and while it might work much of the time, it is, in the end, imperfect. By definition. One big problem with models is they cause us to over-value the things we can measure, because we can measure them, and under-value the things we can't measure, because we can't. It's baked in. It's a form of confirmation bias.

    ReplyDelete
  13. They used the same program to forecast global warming. And it was just as accurate.

    ReplyDelete
  14. Silver is a fraud, simple as that.

    ReplyDelete
  15. Trump wrote off New Hampshire? That's news to the thousands of people who attended his second to last rally before the election. That rally was held, drum roll please, in Manchester NH. And it was a surprise, only being announced the morning of.

    ReplyDelete
  16. Nate Silver gave Trump a far greater chance of winning than anyone else that tried to predict the election based on statistics. The fact that he gave something a 10 percent chance of happening that actually happened only means that all the polls and stats in the world can't pick up late trends, and that's what happened here.

    If you want to rightly mock someone, mock the people that gave Trump a 2 percent or less chance of winning.

    ReplyDelete