I'd rather discuss the philosophical underpinnings of running 20K simulations of a rare event & taking output distribution as probabilities.
-
-
-
wow. You said that better than I did after 6 days of trying lol
End of conversation
New conversation -
-
-
11 is probably pretty small, esp if the 11 existing data points "look" normal.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Quanta took a crack at a similar question, ended up concluding the probabilities aren't defensiblehttps://www.quantamagazine.org/20161026-election-forecasting-puzzle-solution/ …
-
assumption seems to be that large phenomena are irreducible.
End of conversation
New conversation -
-
-
Might be argued that's it's actually 550 previous cases (50 states/11 elections). (Caveat: I may not know what I'm talking about.)
-
No, the presidential election a very specific case. It really is tough to model, not sure we can do it like this.
- Show replies
New conversation -
-
-
YES! my core discomfort with the 538 model is it is v. complex w/ tons of free variables, not much data. Vast danger of overfitting.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yep, if you have physics. We have no laws of nature for societies.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
How is n=11 different for Upshot, Pollster, PredictWise, Princeton, & Daily Kos, & why doesn't that justify 538's lower confidence?
-
I don't think we have a means to judge which presidential election predictor model is better. They all lean Clinton.
- Show replies
New conversation -
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.