STUMP » Articles » Vindication: IT'S NOT HARD » 20 March 2014, 08:36

Where Stu & MP spout off about everything.

Vindication: IT'S NOT HARD  


20 March 2014, 08:36

So I have this nemesis, Carl Bialik, who had been the WSJ Numbers Guy.

To give you an idea of the flavor, here is my one-sided feud on livejournal: GET A NEW SHTICK, PLEASE MAKE IT STOP, JUST STOP IT ALREADY, SOMEBODY STOP THIS GUY BEFORE HE KILLS AGAIN. You can tell I was very calm about the whole thing.

Well, he’s not at the WSJ any more, but is at Nate Silver’s new place. So I decided to check out if I may have unfairly maligned Mr. Bialik, and that all of the crap was due to WSJ editorial decisions. Here are his recent post titles:

A March Madness Bracket For Stat-Heads

Dayton Does Matter

Buffett’s Billion Won’t Lead to a Perfect Bracket

Another Explanation For Crimea Referendum Landslide

How Statisticians Could Help Find That Missing Plane

You Just Had Sex, So How Many Calories Did You Burn?

Wichita State’s Strength of Schedule Isn’t the Problem

Tracking Health One Step (And Clap, And Wave, And Fist Pump) at a Time

Many Signs Pointed to Crimea Independence Vote — But Polls Didn’t

Reviewing the Peer-Review Process

Okay, the headlines don’t go on about how hard things are to figure out. Good, good.

Most of the stuff is fluff, but that’s okay. Let’s look at the serious stories, though. The one on the missing plane:

Keller said Metron isn’t involved in the Malaysia Airlines hunt. If it were, the same principles would apply: Start with all data, such as radar, visual or acoustic measurements, transmissions from the plane and so on. Then update to account for unsuccessful searches, and keep updating as new information comes in. “Bayesian search theory allows flexibility in this way and even accommodates conflicting information,” Keller said. “Nothing is discounted.”

The extra layers of complexity in the Malaysia Airlines search — the new estimates of the plane’s location, mounting evidence that a deliberate act caused the disappearance — complicate the Bayesian calculations and estimates.

Bradley Efron, a Stanford University statistician, said the complications make Bayes a bad fit for the Malaysia Airlines hunt. “Bayes’ Rule is good for refining reasonable (or at least not unreasonable) prior experience on the basis of new evidence,” Efron, who also expressed skepticism to Al Jazeera America, wrote in an email. “It is not good when new evidence changes the situation drastically.”

But Tony O’Hagan, professor emeritus of probability and statistics at the University of Sheffield in the U.K., said that’s the perfect situation for Bayesian techniques, which should make searchers most effective in adapting to changing information, so long as they properly assume from the get-go that the plane might not be in the initial search area.

There’s a lot more there — I think Bialik would have done better to split that one up into a series of multiple posts, because he covers a lot of different topics and really long articles don’t make for easy reading online, even if they work better in a newspaper column. (Yes, I know I write long posts. But I’m not being paid to do this, so I will please only myself tyvm).

But still, it seemed like a reasonable take. Yes, there are difficulties, but in this case he focuses on the techniques that could be used to help, and less on the IT’S SO HAAAAAAAARD whining.

Oh, what’s this? A couple posts on Crimea. Well, that’s pretty important, too. So let’s see.

First post on Crimea:

On Sunday, voters in the Crimea region of Ukraine overwhelmingly chose to secede and become a part of Russia. Crimean officials said nearly 97 percent of voters backed the move, casting ballots as the peninsula remained occupied by thousands of Russian troops.

One big advantage for pollsters: working in the country, with a per-capita GDP below Iraq’s, is cheap, costing less than one-fifth as much to add a question to a Ukrainian poll as it does for a U.S. survey with the same number of respondents, according to Steven Kull, director of “There’s a lot of worthwhile polling that could be done right now,” Kull said. “I’m tempted to jump into it.”

When the Soviet Union dissolved in 1991 and Ukraine gained independence, pollsters found a country full of people eager to tell anyone, including strangers, their opinions about their country and its governance — an ideal scenario for survey researchers.

The “attitude of the population was very, very positive,” Paniotto recalled. He had to work to build polling infrastructure and train interviewers in a country with little independent polling during the Soviet era, and with two major languages (Ukrainian and Russian). Once his staff went into the field, respondents often invited them in for tea; they had to learn how to leave politely and move on to the next home.

Okay, that’s somewhat interesting. Not an angle I knew about, in terms of widespread polling in poorer places. Again, he did a super-long post. But that really doesn’t have to do much with a 97% result.

But looks like he had some followup:

I wrote earlier on Monday about the overwhelming vote by Crimeans to leave Ukraine and join Russia. The latest reputable polls in Crimea showed that just 40 percent of Crimeans wanted Ukraine to integrate with Russia, yet 97 percent of Crimeans on Sunday voted to reunite with Russia. I offered several possible explanations for the discrepancy, including the ballot question wording, the presence of thousands of Russian troops in Crimea and the possibility that some voters wanted only Crimea, and not the rest of Ukraine, to join Russia.

In the comments of my post, several readers offered an additional explanation I didn’t include: The last polls preceded the overthrow of Ukraine’s pro-Russia government and its replacement by a government friendlier to the European Union than to the Kremlin. The Kiev International Institute of Sociology poll – which found about 40 percent of Crimeans backed unification — was conducted from Feb. 8-18. President Viktor Yanukovych, who favored closer ties with Moscow and received 78 percent of the Crimean vote in 2010, fled the country three days after the poll ended and was voted out of office by Ukraine’s parliament on Feb. 22.

I asked Volodymyr Paniotto, general director of KIIS, about this explanation. Paniotto acknowledged the possibility that Yanukovych’s ouster affected the vote, and expanded the explanation to include the Ukrainian parliament’s “foolish steps” — such ascanceling and later reinstating a law granting equal status to minority languages in Ukraine. He still thinks the context of the referendum, including the Russian troops and Russian propaganda present in Crimea, were the dominant factors in the vote.


Is there really any question why there was 97% result in one way?

(and why couldn’t they get to 100%? I bet Kim Jong-Un is laughing at Putin.)

To help you, I added emphasis in the above — as one of Althouse’s commenters put it, this post should’ve stopped after the first paragraph. Indeed, that particular commenter made the exact point I was thinking:

Numbers are supposed to help you say smart things, not cover up stupid things.

Well, you might think so, but you’ve obviously not been following public pension accounting.

More to the point: sometimes people try too hard to come up with a unique take on a situation, in order to get attention. If your business is attracting eyeballs, as opposed to being correct about stuff, then sure. Posit that aliens grabbed the Malaysian flight.

However, you end up with what I call the Malcolm Gladwell effect — you come up with engaging stories you’d like to be true and seem to be counterintuitive, but what really happened is you missed some really big obvious facts that undermine everything else you have to say.



Sometimes it doesn’t even require math.

Related Posts
Foolery for 2019: I Hate the NY State Legislature (and more)
Meep in 2020: A Few Notes on Where to Find Me
Labor force participation rates, part 4: Old v. Young