2016 Postmortem
Related: About this forumNate Silver is getting hammered on twitter right now
This was sparked by a snide comment he made to Matthew Dowd who questioned why shitty polls were affecting his model so much. Check out the thread for good times.
https://twitter.com/NateSilver538/status/794659577115635716
geek tragedy
(68,868 posts)because he made shit up and ignored the data.
Silver's model doesn't account for early voting. It's a huge weakness.
Nate Cohn's work at the NY Times is light years more interesting and sophisticated.
Thrill
(19,178 posts)He was one of the good ones.
budkin
(6,703 posts)He's like shell-shocked over it. SO AFRAID he'll get it wrong, so he builds in a hedge.
Silent3
(15,219 posts)11/4: "Polling Still Shows Hillary Clinton With a Lead, but Not a Safe One"
http://www.nytimes.com/2016/11/05/upshot/polling-still-shows-hillary-clinton-with-a-lead-but-not-a-safe-one.html
I guess it comes down to how safe you consider "safe". Cohn's last odds are certainly much higher than Silver's at 84%, but not as optimistic as Sam Wang's 98%.
84% isn't bad, but that's also roughly the odds of surviving one round of Russian Roulette. Both make me more than a bit anxious.
geek tragedy
(68,868 posts)1) doing his own damn research instead of piggybacking off the work of other people;
2) actually tying polling data to voter rolls and early vote turnout.
Silver's model is pretty much garbage in, garbage out.
gto
(24 posts)There are quite a few other models, particularly ones that have a more accurate track record than Nate *cough*Wang*cough* , and they're all very good news.
kevin881
(465 posts)regnaD kciN
(26,044 posts)
I would guess he's been getting hammered on a regular basis for some time now.
shenmue
(38,506 posts)Doodley
(9,092 posts)If Trump wins New Hampshire (and four of the last eight polls show him ahead),
http://projects.fivethirtyeight.com/2016-election-forecast/new-hampshire/
and Trump wins Nevada, which is about as 50/50 as you can get,
http://projects.fivethirtyeight.com/2016-election-forecast/new-hampshire/
and he wins all the other states that are leaning his way, he wins the presidency.
Nate Silver is being very realistic. A dangerous fascist could win this and we need to be scared.
budkin
(6,703 posts)when you factor in the early voting. That's another point he's getting destroyed for, his model doesn't factor early voting in at all.
Demsrule86
(68,582 posts)Nate Silver is not being realistic...Trump has to run the board and he wont' ...he won't get every toss-up state ...won't happen Silver is interjecting his personal opinion.
TrekLuver
(2,573 posts)geek tragedy
(68,868 posts)He's also not favored to win North Carolina, or Florida (check out the Hispanic NPA early vote down there if you want some real data, not extrapolations from partisan junk polls).
duffyduff
(3,251 posts)Demsrule86
(68,582 posts)he should have shut his polling down...right now it is nothing but bullshit.
Buckeye_Democrat
(14,855 posts)@NateSilver538 and by the way you made stuff up in GOP primary which is why your call was so off. Some of us called it by data.
In Nate's defense, though, I think he was generally correct shortly before the primaries/caucuses took place.
He just should have stuck to polls and not discounted Trump early on.
helpisontheway
(5,008 posts)Then talking about how Hillary's lead is not stable. Today he is saying that this is a 3 point race.
Johnny2X2X
(19,066 posts)Nate is making the mistake of getting emotional about his model. He should not be engaging in this back and forth, he's embarrassing himself. Really losing credibility because he is emotionally involved.
molova
(543 posts)April Lee
(14 posts)Until Silver gets one really wrong, they can say all they like, but it doesn't mean squat.
budkin
(6,703 posts)But something doesn't seem right.
uponit7771
(90,346 posts)Foggyhill
(1,060 posts)He was also off in the primaries
msanthrope
(37,549 posts)November 9th can't come quick enough.
wisteria
(19,581 posts)Joe Scarborough never, ever mentions 538. I think he was trying to surpress the Democrat vote.
musicblind
(4,484 posts)His model has some flaws and it has some upsides. I like that he has three different versions of his model and I like that he is honest when explaining the flaws in his model.
If he gets it wrong, I don't think it will be because he is biased. I think is truly trying his best