A blog by the magazine's editors and contributors


Nate Silver weighs in on the Crimea vote

Fans will be delighted to know that 538 is back up and running. Leading off is an assessment of what the citizens of the Crimea may really have thought: an interview with leading pollsters on polling in the Crimea and Ukraine. Headline: "Many Signs Pointed to Crimea Independence Vote--But Polls Didn't"



Commenting Guidelines

Didn't this all start over the Ukraine's relationship with the European union? Where is that at now?

The protests started after the then-President Yanukovych decided not to sign a preliminary agreement with the EU. Membership in the EU was not on offer--as far as I can tell. Ukraine had many internal reforms to make before that would happen. So far the EU, like the U.S., has offered loans of some sort to get them over their current situation, which I gather is close to bankruptcy. The other issue, big with Putin, is Ukrain's relationship to NATO. As far as I can tell it was never on offer--only perhaps an aspiration in Ukraine, but a deep suspicion in Russia.

In addition to the loans or loan credits from the U.S./EU, the IMF is inspecting the books and the rules and are expected to come up with an austerity program (CF. Greece) that may or may not straighten out the immediate issues. But the economy! Ukraine seems a long way from membership in the EU, even if current troubles with Russia even make that possible. That's all I know.

Thanks for the Nate Silver site, Ms. S.  The article gives a rather clear explanation of how Bayesian statisticians reach their forecasts -- it's a process in which the statisticians start with a number of old, established probabilties about factors which might determine an event, and then, as new data are gathered, the new data are used to revise the earlier statements of probability.  

The method also explains why the media reports on the cause(s) of the Malaysian airplane disappearance keep shifting -- some of the investigators are using Bayesian methods, so they have to keep changing as the new data come in.  

Anne Olivier, glad you found it helpful. I confess that I am muddled when it comes to mathematics or mathematically inclined explanations of basketball, elections, etc. So I have read Nate Silver mostly for his conclusions about the probabilities of who is going to win (or lose). It's too bad the NYTimes let him go. I thought he was a antidote to some of their political coverage.

I don't understand much of the math either, and I must admit I didn't finish his book, but one thing I really like about him is that he doesn't assume that the general public to too dumb to understand any of it, and he explains so well that even I can get some of the rudiments.   Given how dependent policy is on statistics these days we'd better know which statisticians to trust.

Part of the problem is that probability theory is still in its infancy -- its most basic assumptions haven't been settled by the philosophers of math yet, so even the vocabulary the statisticians use is still pretty shifty.  They don't even agree about what the word "probability" means.  It' a really, really hard philosophical problem involving questions of causality, epistemology and measurement.

In a recent New York interview  Nate talks about his new 538 site devoted to "data journalism" and how it differs from the op-ed journalism of The NYT and other newspapers.  He claims that the pundits don't base their opinions on facts, but on "priors", i.e., assumptions for which they have scant evidence if any.  And he faults them for not changing their opinions when the facts change.  (Sounds like he doesn't read Krugman, but I'll second his opinion of Tommy Friedman.) New York magazine.

He seems to have struck a nerve big time.  Leon Wieseltier answers in kind in The New Republic, faulting him for narrowly focusing on the quantitative,  pointing out that math can't answer questions of value. The New Republic.

My complaint about coverage of politics and political campaigns (above) has to do with the amount of chatter and gossip that passes for information, even by serious political reporters. 538 cuts through that and simply reports what voters are like to do, and what they've done in the past number of elections. It's clarifying.

My main complaint about political coverage is that the pundits don't do what they really are -- or should be -- qualified to do, namely, talk about the *reasons* X, Y, or Z is a good or bad policy and the reasons *why* an event is a significant event.  They mainly describes and evaluates without giving enough reasons to justify the evaluations.  They don't get enough into causes, except the most obvious ones, though The NYT and WaPo sometimes try to do better.  Not to meniton outlets like Fox which mainly engage name-calling.  Nate's right -- they mostly don't know as much as they need to if they're going to pontificate successfully.  Shallow, shallow, shallow.