[g1313]

Yeah, it is looking worse this morning for their predictions, and even though I think Nate Silver generally does a good job acknowledging the weaknesses of the data, I am seeing some defensiveness on their blog this morning (Acknowledging misses in Florida and Southeast, but "hey the votes are not all in" in midwest and Nevada which were also big misses on margin which are unlikely to be made up in outstanding votes). They also acknowledged that there might be a systemic polling underestimate of Trump like in 2016, but everyone was trying to correct for that, with apparently little success.

I'm a little more hopeful that with Trump gone, as appears still likely, even with no clear mandate in the Senate, COVID policies will be more rational, even if there is little progress on other liberal goals.

What exactly has 538 gotten wrong so far? The major trump win

has been Florida, which 538 gave him a 1 in 3 chance of doing.

Missed on Florida, as you noted.
Then, like most, 538 had projections of Biden winning Wisconsin, Michigan and Pennsylvania by margins beyond what is posted at the moment.   The result will be outside of their 80% of the results window.   A number of senate projections and odds were off too.

The problem is that each election is a discrete event, so the quick and visually appealing probability is just not the way to go.    Would Biden win 90% of the time if Biden and Trump could hit a magical mulligan button.   Tend to doubt it.

Back to COVID, if Trump wins by a few thousands votes in NV and/or ME#2 or if Biden wins by taking ME #2, NV, MI and WI the net result will be that many in DC will rightfully conclude THEIR constituents don't actually want compromise. 

COVID19 will thrive with the dysfunctional federal government I expect we will see.    

Sure hope the vaccines work and are available as soon as POTUS45 promised.

[g1313]

Yeah, it is looking worse this morning for their predictions, and even though I think Nate Silver generally does a good job acknowledging the weaknesses of the data, I am seeing some defensiveness on their blog this morning (Acknowledging misses in Florida and Southeast, but "hey the votes are not all in" in midwest and Nevada which were also big misses on margin which are unlikely to be made up in outstanding votes). They also acknowledged that there might be a systemic polling underestimate of Trump like in 2016, but everyone was trying to correct for that, with apparently little success.

I'm a little more hopeful that with Trump gone, as appears still likely, even with no clear mandate in the Senate, COVID policies will be more rational, even if there is little progress on other liberal goals.

What exactly has 538 gotten wrong so far? The major trump win

has been Florida, which 538 gave him a 1 in 3 chance of doing.

Missed on Florida, as you noted.
Then, like most, 538 had projections of Biden winning Wisconsin, Michigan and Pennsylvania by margins beyond what is posted at the moment.   The result will be outside of their 80% of the results window.   A number of senate projections and odds were off too.

The problem is that each election is a discrete event, so the quick and visually appealing probability is just not the way to go.    Would Biden win 90% of the time if Biden and Trump could hit a magical mulligan button.   Tend to doubt it.

Back to COVID, if Trump wins by a few thousands votes in NV and/or ME#2 or if Biden wins by taking ME #2, NV, MI and WI the net result will be that many in DC will rightfully conclude THEIR constituents don't actually want compromise. 

COVID19 will thrive with the dysfunctional federal government I expect we will see.    

Sure hope the vaccines work and are available as soon as POTUS45 promised.