There is probability when you don’t know

Cookie monster like probabilities between 0 and 1.                                                                  

Nate Silver

People have a lot of trouble getting their head around probability.

I would include myself in that. Is probability an objective feature of world, as suggested by considering how a fair dice will land?  Or is it just a way of talking about a subjective condition: our own ignorance of a world fully determined? There either is or is not intelligent life out in space somewhere.  Assigning odds to that may seem silly. And geeze do the reported odds vary.

I have trouble keeping that one straight too, and I doubt I am alone. For example, here is a guy on Slate pretending he has it all figured out, to humorous effect. He opens strongly with:

Ever get the feeling that the presidential polls are fluctuating because the storyline of this election needs them to?


What he is trying to say is that the “true” odds of Hillary Clinton winning the presidency are varying much less than the polling data suggest.  Like some liberal partisans, he urges us not to be bedwetters.

In support of this objective, he trots out the thesis that the news of the day, most recently emails, affects people’s willingness to answer pollsters, but not how they plan to vote. I guess that is plausible. It would be cool to see some evidence of it. Maybe do a poll?

But then he says that the polling fluctuations

may be more noticeable this year, but they still fall within a range. Since the summer, Clinton has almost always polled between the low 40s and mid-40s, Trump between the mid-30s and low 40s. Where they fall within those ranges can vary, give or take a debate, a party convention, an FBI investigation, or a Billy Bush tape.

This would be compelling if those ranges did not overlap. Given that they do, his unsupported hypothesis about people demurring on certain types of news had better be true.

The main inference I draw from this is that this guy does not like probability.  Most polls show Hillary ahead but some show her tied. So the perceived odds of Hillary winning are typically high but sometimes dip to around even. Liberal serene with even odds are made of sterner stuff than I am!

What drives these fluctuations are changes in the polls. This is inherent in the idea of probability and not evidence that pollsters, the people who crunch their results, or those who follow the number crunchers are bed wetters.

I suspect one issue here is that some people really don’t like thinking of this stuff ex ante. Their intuition is that at some point the issue will be resolved, i.e we will have president, as the binary choice is resolved, and that all this hemming and hawing will be revealed as having been pointless.

Ex post, there either is intelligent life in the White House or there is not.  Moreover, whatever polls were misleading will especially not have mattered. Why wait?

Guys with this intuition, which I don’t share, think it is somehow more sober to think as though things are already ex post. I would not say these folks don’t just dislike Nate Silver. I would say they also dislike the whole idea of probability.

That’s fine. I am not sure the philosophers and physicists have that one fully figured out.  And it is beyond my pay grade.  But I think it is fun to call a spade a spade. I don’t get or like probability might be hard to get by the editor. Slagging the Nate Silvers of the world as bed wetters has a nicer ring to it.

I see Brad Delong disagrees with me and has given up on Nate Silver because his odds move around too much. The way he expresses this is more complex and somewhat beyond me, but along the way he reveals that what probability even means is not fully resolved.