Won't forget how mistaken FiveThirtyEight was just hours before the final results. Unbelievable.
7
18
3
19
Replying to @glazou
How is that mistaken? They're probabilities, and error is expected to correlate between states.
L. David Baron @dbaron@w3c.social @davidbaron
9 Nov 2016
The @FiveThirtyEight forecast was not bad: p(T)>.5: 25T .5<p(C)<.6: 1C 2T .6<p(C)<.7: 1C .7<p(C)<.8: 1C 2T .8<p(C)<.9: 4C 1T p(C)>.9: 14C

Nov 9, 2016 · 9:39 AM UTC

1
4
Replying to @davidbaron
I understand probs but his team apparently completely missed something on the votes, in particular in swing states
1
1
Missed what? That was the final forecast based on polling.
1