Why did the networks take so long to call Wisconsin?

[emailking]

But Wisconsin should have been called MUCH sooner.

All networks and news organizations should have called Wisconsin by 12:30AM. Then the networks should have called Pennsylvania all before 2AM like print media organizations did.

You say that, but unless you were doing a hardcore statistical analysis on it, complete with computer modeling, analyzing all of the precinct totals and exit polling, etc, then you really have no idea whether they "should" have been called earlier. The decision desks are doing all that and don't make calls until there is virtually no doubt.

Georgia was called by all the news organizations from 11:35-11:45PM.

It was bizarre. It was like they forgot about it. Should've been called at around 9:30PM. Atlanta was a little bit slow but it was clear Trump would win it.

"Clear" to you is probably a lower threshold than the statistical threshold they're looking for before making a projection they can stand behind. There's a reason the networks have had only one misfire on projections since 2000, and it's not because they make the calls once the result are clear to Atlas armchair pollsters.

If Clinton was ahead they would have called it ASAP.

Possibly. It would depend on where the vote was coming from that was putting her ahead. Regardless, if she was ahead it would be more likely she would win because it would align with the exit polls. It's basically was Gustaf said. Less evidence would be needed from raw votes to pass their statistical threshold for calling the race.

[emailking]

This ignores the fact Fox News was able to call it (Wisconsin) so quickly.

It only ignores it in the sense that I did not directly address it. But it does not affect anything I said. Fox News has their own model, and that model allowed them to call it at 11:30. Maybe Fox News has the best model, and if so good for them. But it doesn't mean the CNN and NBC models were flawed, or that partisan leanings led the decision desk analysts to hold back on a conclusion that should have been evident. Maybe that did happen, but the evidence for it is circumstantial. And for all we know, partisan leanings could have caused Fox News to jump the gun a bit.

Remember, it was Fox News that first called Florida in 2000, which they then retracted. There are lots of reasons for why the results were so strange. But it illustrates the point that just because they were able to call the race in WI quickly doesn't mean the conclusion was justified. It was justified to them using their model and desired level of confidence. And it wasn't justified for the other networks until later. That's all we really know.

[emailking]

Except the NYT Upshot also had Trump at a 95%< of winning at 11:30 and CBS gave it the "Edge" at that time.

So what? Maybe CNN and NBC had it at >95% too. That's not enough to call it. They don't report their confidence levels, until they call it.

The networks tried to hold off calling Wisconsin until Milwaukee absentee ballots came in, and when they did and Trump's lead was just as large they should have called it.

You don't know why they held off calling it though. That's kind of my whole point. You're just making assumptions.

Remember, it was Fox News that first called Florida in 2000, which they then retracted.

CBS was the first network to retract Gore's projected win, not Fox.

To clarify, what I meant was that Fox News was the first to call Florida for Bush, not that they were the first to retract a call for Gore. It doesn't really matter who it was. I'm just making the point that because a network calls a race before another network doesn't mean the model or analysis of the former is better or more correct. Florida 2000 provides an unambiguous example of that.

[emailking]

95%< = 95.1%+ btw. The Upshot never had a number high than 95% (thus the greater than) so by 95%< the state can be called accurately (if it's been that way for like 20 minutes at least).

No? >95% for 20 minutes doesn't mean you can call a state. You can call a state when the probability reaches a certain threshold (typically 99.5% or higher). It doesn't matter how long it sits at a high threshold. It just doesn't work that way.

A lot of pundits on Twitter thought Wisconsin was over by 12:30AM.

Again, so what? They're just as armchair as you are. None of you were performing hardcore mathematical models using the exit polls and raw results.

Obama believed the call enough to call Hillary to concede after Fox called Wisconsin (this from the "Shattered" book).

Concede before any network had actually called the whole race though? Ok, whatever. It still doesn't matter. I can concede it was extremely likely that he would win Wisconsin once a network (any network) called it. That's probably what Obama based his decision on. That still doesn't mean the result was virtually certain according to the other network analysts, or should have been. They're in a bubble. They don't know that Fox News has called the state.  They're just looking at the results and running they're models.

[emailking]

I think you're either missing my points entirely or aren't discussing this is good faith. I was trying to give the benefit of the doubt before, but after the last post I fear it may be the latter. In any case, I think I'm out. We're just going in circles at this point.