Nevada shows why Nate Silver's approach is so wrong ...

[Alhena]

Silver has Nevada as a Trump-lean state based on polls. But Nevada political experts say that actual early voting info there is so overwhelmingly in favor of Dems that Trump has basically no chance there.

Silver relies way too much on polls, which is necessary early on, but when early voting actually starts sifting through that data is far more reliable.

Trump has little chance without Nevada.

[metroins]

2. It's not wrong

It's his tried and true methodology.

He shouldn't count early voting numbers, because that's not what he does.

Nate won't get everything correct, but his methodology has been proven to be right a lot.

[brush]

4. Nah. He was accurate in 2008 but slipped in 2012. Now he's just wrong a lot.

He needs to adjust his methods to reflect actual early voting.

Not the hard to comprehend. Use early voting results.

He's not looking too sharp. What's he waiting on?

[metroins]

8. I disagree

His model is based on polling and he should not add in early voting. It's not what he uses, he uses polling.

I also do not think he's slipping.

[brush]

24. I disagree. Once early voting results are in he should adjust his model to factor in the information

Some of his recent findings are way off. For instance, he's way off in Nevada. I live in Vegas and we've had a huge surge of Dem early votes that way outnumber the repugs but his results don't reflect that.

Trump has no chance here, even Ralston is reporting that.

[AgadorSparticus]

26. Yep! Silver is off...and Nevada highlights that well

[DDySiegs]

33. Your Attitude is Correct - It's Bayesian

The point is that one should always take into account new or changed data. As John Maynard Keynes responded to someone who chided him for changing his position -- When the facts change I change my mind,what do you do?

[Albertoo]

11. Nope: Silver was correct in 2012 too

[Major Nikon]

13. He was 50 out of 51 in 2008. In 2012 he was 51/51.

There was no slip.

[Statistical]

10. Any good system will adapt from year to year to improve error rates.

If you see your model is good but then totally diverges from early voting and you have early voting data then maybe early voting should be included in future versions of the model.

[metroins]

19. Or, if your model works

You don't change it.

Nate isn't trying to agree with early voting, he's trying to predict winning of races based upon polling.

I don't see why people are making this an issue, Nate Silver has a track record that works. You don't change it 4 days before an election.

[cbdo2007]

6. We won't know if he's wrong until Tuesday.

But also, I don't think you understand what he does, you just aren't happy with his approach. That doesn't make it wrong.

Besides, he's still got Clinton winning so who the fuck cares.

[Peaches999]

9. Voter party registration does not change

So a Democrat can vote for Trump and a Republican can vote for Clinton.

Thus, when it comes to voter preferences, and the short term fluctuation of that based on the Trump tapes or the FBI reopening the email investigation, voter registration won't capture that.

The Democrats have had an advantage in overall voter party registration in Nevada for some time.

What voter registration data does capture is intensity or voter likelihood. However, we don't know if more Democrats are just voting early or if Republican turnout will be depressed overall.

All in all, I wouldn't be so quick to say Nate Silver's methodology is wrong.

[duffyduff]

16. I would. Trump has no shot in Nevada. n/t

[Peaches999]

20. Statistically, that's not true

I don't want to sound like Debbie Downer, but the shift in the polls over the past week is scary. This election is going to be close, in my opinion. It's looking a lot more like 2000 than 2008.

[Motown_Johnny]

14. It depends on the demographics.

He isn't changing what he believes will be the percentage of latino votes simply because there has been a higher turn out among latinos in early voting.

It is a reasonable point of view. That is not to say it will turn out to be correct, but it is still reasonable.

[Ford_Prefect]

21. It also depends on an organization and ground game strategy. Trump has none of either.

The GOP seems to be counting on general disaffection, disinformation and confusion to depress Democratic voting.

The polls do not tell you who will actually show up and vote. They tell you who said they like a particular candidate . Many contemporary polls fail to accurately measure the intentions of those who haven't got land line phones. Silver said this himself.

[NobodyHere]

22. Shouldn't we wait for the results to be in before saying his approach is wrong?

[tymorial]

28. I wish that would happen but it won't

Many of these posts are a reaction to reading something that disagrees with political views. It is therefore wrong, suspicious, or evidence of conservative bias. The emotionalism and lack of objectivity is obvious.

If Silver predicted a sea of blue with only a couple of red states, the same people would call him a visionary genius.

[davidn3600]

23. Republicans typically vote on election day

Many polling experts have cautioned using early voting numbers as measuring sticks. The only real use for such data is perhaps to compare turnout to previous elections. Democrats tend to vote early and Republicans tend to vote late. There are elements of the GOP that philosophically do not believe in early voting.

Also, you don't know for sure exactly how these early voters are voting. These numbers are based only on raw party ID. These are NOT exit polling data that has been crunched. We won't get any exit poll information until Tuesday evening. The networks have agreed to withhold that data until then.

[qazplm]

27. Nate got burned

now he's over-reacting the other way so that no one can accuse him of "missing on Trump" a second time.

[rollin74]

30. a very large majority of Nevadans vote early

early voting has ended in NV

Trump cannot realistically overcome the gop deficit in voter turnout that has happened. The damage is done.

Even with strong indie support for Trump and a huge Republican turnout on election day (doubtful), he will still likely lose

[Orrex]

31. NATE SILVER IS SACROSANCT AND A POX ON ANY WHO DENY HIM!!!!!1!

Frankly, I've had enough of the 538 adoration to last me for several election cycles.