The various nerd models give us a 3/4 chance of winning the House. How is such a prediction testable? Isn't it a case of everyone able to declare victory, no matter what happens?
-
Show this thread
-
Replying to @Pinboard
It's not testable. You cannot assign probabilities to one-off events. But folks who's paychecks depend on these forecasts will ramble on a lot about "calibration" and "Bayesian stats" until you give up.
3 replies 0 retweets 12 likes -
Replying to @beenwrekt @Pinboard
Except you can decompose that individual prediction into 100s of individual races and test your probability calibration on those
1 reply 0 retweets 0 likes -
-
Replying to @Pinboard @beenwrekt
Really don’t understand the shade being thrown on these kind of predictions
1 reply 0 retweets 0 likes -
Replying to @lesnikow @beenwrekt
My problem is not the predictions, but the way they couple to journalism and fundraising. I think Nate Silver in particular has done a good job communicating uncertainty this year.
1 reply 0 retweets 0 likes -
Replying to @Pinboard @beenwrekt
Well then say that instead of questioning whether the predictions are testable
1 reply 0 retweets 0 likes
Okay, but they're also not testable...
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.