Read on Twitter
No. The 90% probability implies that there is a lot of scenarios where the election is not close. This election was close
It does not imply that running a model based on garbage inputs is robust when it is shown that the inputs are in fact garbage
It does not imply that running a model based on garbage inputs is robust when it is shown that the inputs are in fact garbage
This logic is non falsifiable> We can never know if those unrealized scenarios that made up a lot of the 90% probability didn't happen, but could have, in a different polling universe, and therefore were justified
Put simply it was a close election. A model saying before hand, "it is basically 50/50, or 55/45" more captures what happened, than one saying it is 90/10!
And with this thread, which says polls suck so models based on polls suck, I have been confined to the Un-serious Thinker camp!
Since this is getting retweeted again, just re- upping this thread on why I have never trusted election forecasting... https://twitter.com/Chris_arnade/status/1318933057978748929?s=20
