Source: http://xkcd.com/1131/
On Monday night I tweeted a useful post by on judging forecasts by Dart Throwing Chimp, noting that
forecasts are probabilities, not absolutes, and we really ought to evaluate them as such.
Several of my FB friends who know this well have nonetheless succumbed to the temptation to jump on the “big winner” bandwagon and proclaim that Nate Silver, quants, science, etc. were proven right by the US Presidential Election. D’oh! [bangs his head repeatedly on his desk.]
To quote Mike Ditka: Stop it!
Testing hypotheses based on one forecast (in this case the mean of a prob. distribution) is misguided but aggregating across polls, then looking at the mean, and using all of that information to forecast…seems like a better way of doing things then relying on a single story by a pundit, no?
http://themonkeycage.org/blog/2012/11/07/the-accuracy-of-the-final-national-polls/
Ack! This post suggests that my post is easily misunderstood. That’s not good.
1. Forecasts like those by Silver, Linzer, etc. are awesome.
2. Judging forecasts like those by Silver, Linzer, etc. (or the person who forecasts your weather, hurricanes, etc.) based on whether a single forecast was consistent with a single outcome is BAD. It is very bad. It is nonsense. Don’t do it!