is it not just a percentage of simulation runs one way or other?
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Re-posting here, because I am curious as to other's opinions on this.https://twitter.com/_HectorTorres/status/757984839454699520 …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The absence of error bars or any kind of statistical context is absurd. Trump's odds right now probably look more like 46.8 ± 40%
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Modeling is fine, but it's not like random sampling (which we can't even do well anymore) where you have precise error. And this early? Meh.
-
Wouldn't mind it with some explanation, presented as a range, no digit after the period FFS, with some info on how model performed in past.
- Show replies
New conversation -
-
-
@benjaminwittes@FiveThirtyEight a Bayesian approach. How would you do it diff?Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Elections too infrequent, polling history too short, for anything like that level of confidence.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@benjaminwittes in quantified society, numerical estimates treated as more valuable than narrative--no matter how contestable. Sad! -
Is it wrong to say that Google is worth 508.93B today? Or that the cubs have a 21% of winning the WS?
- Show replies
New conversation -
-
-
I think they should present not numbers but visual ranges that are blur accordingly for this kind of models.
-
If you scroll down herr they do have ranges for 80% confidence. http://projects.fivethirtyeight.com/2016-election-forecast/ …
- 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.