-
-
-
Replying to @moneyreflux
yes and no but also mostly this is kind of just inadvertently shitting on his own model imo
3 replies 0 retweets 16 likes -
Replying to @eigenrobot
say more! should the model include more than just the distribution of polling error? should it assume bigger than historical error?
1 reply 0 retweets 3 likes -
Replying to @moneyreflux
my objection is not to the conceptual structure of the model itself which i dont care about rather if its this insensitive to new information i question its usefulness
1 reply 0 retweets 7 likes -
Replying to @eigenrobot
Fair enough. when you think of new information in that sense, is it things like the Biden laptop or Supreme Court rulings? Or softer stuff?
1 reply 0 retweets 3 likes -
-
Replying to @eigenrobot
This is the point I understand least. The 538 model is actually *more* sensitive to poll movements in the last week of the election, to capture late trends. The recent polls in the last 48 hours average each other out, roughly.
1 reply 0 retweets 2 likes
if they're cumulative runs, maybe
hmm
lol funny that order should matter for this question
-
-
Replying to @eigenrobot
I think that’s what he means - as each polls has been added, the re-run has only nudged the model up and down slightly, as the polls roughly average out. This is in contrast to 2016, where (for example) 538’s model had Clinton move from 85% on 26 Oct to 64% on 5 Nov.
0 replies 0 retweets 1 likeThanks. Twitter will use this to make your timeline better. UndoUndo
-
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.
