Deleted tweet detection is currently running at reduced
capacity due to changes to the Twitter API. Some tweets that have been
deleted by the tweet author may not be labeled as deleted in the PolitiTweet
interface.
Showing page 237 of 729.
Nate Cohn @Nate_Cohn
@gelliottmorris I think you're overfitting models on a small dataset and using regularization and out-of-sample cross-validation to delude yourself into thinking that's not what your doing. It just doesn't come close to approximating real world uncertainty — PolitiTweet.org
Nate Cohn @Nate_Cohn
sorry, to clarify--the *omission* of approval may *increase* the weight of H2H polls. — PolitiTweet.org
Nate Cohn @Nate_Cohn
The NYT thing is hard for me to take very seriously, but I would guess/hope that it is not a major factor in adding uncertainty. Historically, it doesn't really do a great job of predicting volatility, after all — PolitiTweet.org
Nate Cohn @Nate_Cohn
Well there are two separate issues here. The omission of approval certainly moves their fundamentals-based forecast to the right, creating more mean revision. Just how much, IDK? But it may also reduce the weight of H2H polls, which are largely collinear https://t.co/cV0K3iypZV — PolitiTweet.org
Tentin Quarantino @agraybee
@Nate_Cohn I get that you have to be deferential before the master but a model that takes the physical dimensions o… https://t.co/8Ckivw1Vx0
Nate Cohn @Nate_Cohn
Looking back through these notes, most are reasons why there might be somewhat less uncertainty, so I guess I take the over on p(Biden). But fundamentally, this forecast thinks about this stuff a lot more like I do than the competition and it'll be fun to watch — PolitiTweet.org
Nate Cohn @Nate_Cohn
Looking back through these notes, most reasons why there might be somewhat less uncertainty than the forecast, so I guess I take the over on p(Biden). But fundamentally, this forecast thinks about this stuff a lot more like I do than the competition and it'll be fun to watch — PolitiTweet.org
Nate Cohn @Nate_Cohn
FiveThirtyEight starts off where it left off, at 71%. I like a lot of the changes, but given that I've poked at some other forecasters (and so has Nate!) I'll offer some critical thoughts as well: https://t.co/0K6hTI3LMf — PolitiTweet.org
Nate Cohn @Nate_Cohn
@NateSilver538 but where is the nowcast?? — PolitiTweet.org
Nate Cohn @Nate_Cohn
GSG/Navigator poll shows a notable uptick in Trump approval, up to -13 from -21 last month https://t.co/y24Bsv6Jg7 — PolitiTweet.org
Nate Cohn @Nate_Cohn
RT @jimgeraghty: You know, whatever else you think of Kamala Harris, it’s pretty amazing and awesome that immigrants can come to the U.S.,… — PolitiTweet.org
Nate Cohn @Nate_Cohn
RT @DLeonhardt: If they win, Biden and Harris will be the first President/Vice President in 40 years not to have an Ivy League degree. — PolitiTweet.org
Nate Cohn @Nate_Cohn
Obviously, those aren't the kind of numbers that I would expect to really move the race much either way, though I'd say that about most vice presidential selections — PolitiTweet.org
Nate Cohn @Nate_Cohn
So it's set: Biden/Harris v. Trump/Pence. In our six-state battleground poll, Harris favorability rating was at +5, 35 to 30 percent, with a 62-8 rating among Biden voters an 8-57 rating among Trump voters. — PolitiTweet.org
Nate Cohn @Nate_Cohn
RT @jmartNYT: It’s not Whitmer, Dems tell @katieglueck and I — PolitiTweet.org
Nate Cohn @Nate_Cohn
And looking at the fivethirtyeight page, looks like they have it too https://t.co/E0Cs0LIH11 — PolitiTweet.org
Nate Cohn @Nate_Cohn
Some of the confusion may be an erroneous tweet from the Marquette account, and there may also be some additional confusion about whether to use the results with or without leaners (I use leaners) — PolitiTweet.org
Nate Cohn @Nate_Cohn
As an aside, it seems that I have a bit of a disagreement with many of you about the trendline on the Marquette poll. As far as I'm concerned, RV/LV shifted 4 points, from +9/+8 to +5/+4 — PolitiTweet.org
Nate Cohn @Nate_Cohn
@byelin nope? — PolitiTweet.org
Nate Cohn @Nate_Cohn
Both Monmouth and Marquette are about 2-3 points to the right of their last polls in late June/early July. Still a clear Biden lead but consistent with 'tightening' over the last month and heading into the convention. — PolitiTweet.org
Nate Cohn @Nate_Cohn
Which would you rather get right now: all of today's poll results, or Biden's VP pick — PolitiTweet.org
Nate Cohn @Nate_Cohn
@PollsAndVotes @MULawPoll maybe you should spread it out a bit more, you know, so like a tweet with the full questions, and the one tweet per item response and frequencies — PolitiTweet.org
Nate Cohn @Nate_Cohn
I'm concerned that we won't get our results until tomorrow since @PollsAndVotes hasn't started the thread yet https://t.co/N5J4vhbGW2 — PolitiTweet.org
former work from home enthusiast @akaashkolluri
@Nate_Cohn What do you think of Marquette rolling out its results tweet by tweet? Frustrating, or extremely funny i… https://t.co/yOYbpoYKmE
Nate Cohn @Nate_Cohn
I'm not really paying attention to the race for agriculture secretary, either https://t.co/JJ2AnDltPh — PolitiTweet.org
Neil C Spencer @NeilSpencerd
@Nate_Cohn Plus Emerson has four polls coming out today, if you’re keeping track of Emerson I suppose…
Nate Cohn @Nate_Cohn
I wouldn't think much, though I do have a pet theory that a Biden pick could hurt him in the polls, given his current standing, if you think there will be controversy and if you buy that in general the candidate in the news fares worse in the polls https://t.co/ISVJ4iWcqk — PolitiTweet.org
Nate Cohn @Nate_Cohn
Well it's a suspenseful day. I know everyone is eager for the big news to finally drop: the results of the Marquette and Monmouth polls. — PolitiTweet.org
Nate Cohn @Nate_Cohn
RT @albrgr: 🤔 🤔 🤔 https://t.co/kMeJh3MVSV https://t.co/XTYcApL4Zj https://t.co/aE2VqqzPJ7 — PolitiTweet.org
Nate Cohn @Nate_Cohn
RT @MonmouthPoll: COMING TOMORROW: National 2020 #GeneralElection. Confidence in @realDonaldTrump / @JoeBiden to handle #COVID recovery. PL… — PolitiTweet.org
Nate Cohn @Nate_Cohn
I'm not exactly sure which of these theories explains the Battleground poll. But it really might not represent the kind of expert-driven methodological decision one might assume — PolitiTweet.org
Nate Cohn @Nate_Cohn
Others are just following long-established norms: that's why the RDD pollsters have been doing it for years, while the RBS pollsters didn't. This rarely represented a considered methodological conclusion on the part of the RBS pollsters. It's just how you did RBS. — PolitiTweet.org
Nate Cohn @Nate_Cohn
I'm also sad to say that this isn't exactly what's happening here. While there are the few exceptions I mentioned earlier, many pollsters just don't even know about these debates at all. https://t.co/bDlSutrGV2 — PolitiTweet.org
Nate Silver @NateSilver538
So if 20% of expert pollsters have thought carefully about these problems, and concluded education weighting causes… https://t.co/LbzNUZrH9O