Wednesday, October 31, 2012

Blaming the Messenger

So, this was an interesting development.

All this election season, I've been following Nate Silver's site (fivethirtyeight.com) for his analysis of the polls and what they tell us about the presidential race. It's been his conclusion that beneath the surface noise and media obsession with the 'horse race' it's been a remarkably stable election, with the incumbent maintaining a slight but significant lead throughout.

Now commentators who like to position themselves as thoughtful conservatives, like David Brooks and Joe Scarborough and Politico.com and the National Review, are attacking Silver, claiming that just because his method worked last time around, that doesn't mean it has any validity this election -- e.g.

http://www.politico.com/blogs/media/2012/10/nate-silver-romney-clearly-could-still-win-147618.html?hp=r14

Silver's response ("I'm sorry that Joe is math-challenged") is amusing and his explanation moderate (you add up the states where polls you consider reliable predict candidate A is ahead and compare the total against that for candidate B) -- and of course he's explained his methodology in detail a number of times on his site, including the important caveat that the prediction only covers known facts, not 'October surprises'.*

The most interesting part about all this is that Silver showed Obama ahead for months, with his chance of winning thereby growing larger the closer it got to the election (= less time for the challenger to make up the difference). Then after the first debate that trend reversed, with Obama's lead melting away day-by-day over the next three weeks, stabilizing around the time of the final debate, and climbing steadily back up ever since.

Now, so far as I am aware, none of these conservatives attacked Silver when his poll showed that Romney was rapidly gaining ground on Obama -- this being a message they v. much wanted to hear. But once he reported that Romney's surge had proved ephemeral and was receding, they pounced.

The moral? distrust those who attack the messenger when they don't like the message. And we shd all try to be mindful of our innate tendencies to embrace evidence that supports a conclusion we like and downgrade evidence that supports one we don't like: basic human nature.

--John R., still in election mode


P.S.: Thanks to Janice for pointing out that Paul Krugman, who as a Nobel-Prize winning economist knows a thing or two about numbers and statistics, and who's never shy about offering a pungent comment, has come to Silver's defense:

http://krugman.blogs.nytimes.com/2012/10/28/the-war-on-objectivity/



*Tues. night I heard for the first time the suggestion that Tropical Storm Sam was a classic 'October Surprise'.


1 comment:

grodog said...

Good stuff, thanks John (and Janice :D )!

Allan.