Nate Silver

All Photos 8 hours ago
How Amy Coney Barrett could change the Supreme Court:
Retweeted by Nate Silver
The exception is some Midwestern and Great Plains/Rocky Mountain states, where cases are clearly increasing too quickly to be explained by testing alone.
There are lots of reasons why one might expect an increase (stuff moving indoors, schools/colleges starting, people not "focusing" on COVID as much and/or getting fatigued with social distancing, Europe is having a 2nd wave) I'm just not sure if the data is showing one ... yet.
It's not clear whether COVID cases are actually on the rise again in the US or whether this reflects increased testing, especially with there being several different types of tests now that are becoming harder to track. Positivity rates have remained level at ~5% for a few weeks.
It might also help to clarify exactly what scenarios people are talking about. Some of the items on the list below are quite likely to happen, IMO. Some are unlikely but hugely consequential. And this isn't a comprehensive list, by any means.
Politics Twitter can be polemical, prone toward overstatement, not inclined to think probabilistically, etc, plus it tends to lean left. So yeah I think some people here are going too far. But the mainstream media in general, I think, is not taking these threats seriously enough.
I've gotten into some meta-debates on here about whether people are overrating or underrating the chance that Trump could successfully steal the election. I think I've been a bit guilty of mistaking "people I follow on Twitter" for people in general.
One way to look at the problem with Facebook's News Feed is it's trying to optimize for the same qualities you'd want when reading content from friends/family, i.e. content that triggers reflexive emotional responses, when that isn't necessarily what you want from news coverage.
This is shoddy logic from the Facebook executive. Humans design the Facebook algorithm, and "engagement" isn't some intrisic quality but one defined by Facebook engineers.…
Genuine (although slightly leading) question: What are examples of court intervention overturning the result of US elections in races that weren't *extremely* close?
It's not persuasive to cite 2000 because 2000 involved an extraordinarily close vote (within 0.009%) in which it was genuinely ambiguous who won. The odds of having that close a result in a tipping-point state are very low. The Q is what Trump could do in a less-close race.…
No. 3: The perception that Trump will steal the election could lower turnout No. 4: The perception that Trump will steal the election could cause strategic blunders by Democrats No. 5: Either side could have trouble accepting the result even if the election ends up being fair
Issue No. 1: There's also an election to cover, with a pandemic, SCOTUS pick, racial protests, etc. No. 2: Instilling fear or ambiguity is a classic authoritarian tactic, and there is some risk of amplifying Trump's message in a way that undermines trust in the electoral process
No one is debating that. In fact, I have said explicitly that it's important to pay attention to. But there are some issues that can arise from treating it as a high-probability event instead of a low-to-medium probability one.…
But if Democrats return their ballots sooner, the late shift could be mitigated or even reversed. There's maybe the slightest hint of that already in North Carolina, where Democrats represent 49% of absentee ballot *requests* but 54% of ballots *returned* so far.
In most states, mail ballots received by Election Day will be counted with other ballots on election day; in some states they'll even be counted first, in fact. It's ballots postmarked by E-Day but not yet received (allowed in some states but not all) that can cause a late shift.
One thing Democrats could do—with basically no strategic downside—is to hammer home the message that voters should ask for and return their mail ballots as early as possible, which makes it more likely they'll be included in tallies on election night.
We're at a weird moment where—amid understandable concern that Trump could steal or delegitimize a close election—people seem to have forgotten both that the most likely outcome is a fairly emphatic Biden win *and* that Trump has a decent chance (20-25%) to win legitimately.
We’ve identified 12 polls that have asked some version of “Should Ginsburg’s seat be filled this year by Trump, or next year by the winner of the 2020 election?” On average, 52% of respondents said to wait and 39% said Trump should fill the seat now.
Retweeted by Nate Silver
Exactly. It is not up to him. If he loses the election, the message from everyone should be unequivocal that Jan. 20 is his last day. He and his staff have now basically announced that he will not consider any election that he loses fair. Seems fairly unlikely he will concede.…
Retweeted by Nate Silver
Quite a few mediocre polls for Biden in Arizona lately, which has brought his lead there down to 3.3 points.…
More than half of Trump's wins now involve winning the Electoral College but not the popular vote.…
In a sign of how boring and stable our forecast has been this year, only two states have "flipped" at any point... Ohio and North Carolina have gone back and forth a couple of times. Biden is also now favored in Maine's 2nd Congressional District, when Trump was favored before.
Indeed, if you look at our current forecast in each state (treating it as a binary rather than probabilistic) it's VERY close to Obama's 2008 map, with Biden adding AZ but losing IA and IN (and Iowa is very competitive).
It's worth keeping in mind. On the other hand, it's not that hard to imagine a lot of Obama-Trump voters flipping back to Obama's VP. Especially because a lot of them didn't particularly like Trump but weren't fans of Hillary either.…
Yeah Biden now just a tiny tiny bit ahead in our Ohio average. Also a tiny bit ahead in our forecast there, though that's even closer.…
Fox News has consistently shown some of Biden's best state polls. These are three fairly important states, too: PA is the most likely tipping point, NV hasn't been polled much, and Biden will likely pull ahead in our OH polling average with this poll.…
Pod today on Trump & his unwillingness to guarantee a peaceful transfer of power. Think it was a pretty interesting discussion of, well, that^^, but also how this sort of environment (*waves hands all around*) makes things really difficult for journalism.…
Retweeted by Nate Silver
You could also have states where some mail ballots are counted ahead of time, giving Biden a big lead early in the PM, then Election Day votes are counted, swinging the race to Trump later, then late-arriving mail votes are counted, swinging it back to Biden a couple days later.
Then there are some further complications around early in-person votes and provisional ballots. Many states will have a "blue shift" but some could have a "red shift" instead. In general, it will just be pretty weird.
Yeah, there's going to be a lot of variability from state to state, depending on whether 1) mail ballots must be RECEIVED by Election Day or merely postmarked by Election Day and 2) whether states count mail votes before/after/simultaneously with in-person votes.…
With Biden getting some strong polls in the Midwest over the past few days but some mediocre ones in TX/AZ, we're starting to get a clearer signal that his gains relative to Clinton are mostly among white working-class voters.
* I know that's a slightly strawman-ish formulation. Few places in the world are truly "locked down" right now, etc. But my observation is that US-based public health experts slightly overestimate the public's tolerance for refraining from in-person social contact. 4/
It may be waiting longer for a vaccine we're more certain is effective is "worth it", especially if it helps to facilitate public trust. But I wouldn't bank on "we can just count on social distancing for another X months in early/mid 2021*" as necessarily being viable. 3/
The contribution I'd make as an outsider/observer of political behavior is that it seems clear patience with lockdowns and social distancing is wearing thin, in the US and (perhaps even more so) globally, and it may be even thinner after what could be a difficult winter. 2/
Seeing debates about different strategies for testing vaccine efficacy, some of which would result in quicker but less comprehensive readouts than others. (Note: None of the experts are saying we should compromise on safety; this is all about how we measure efficacy.) 1/
@jbarro @NateSilver538 If you absolutely know you’re going to vote and can vote in person you should, but if you otherwise wouldn’t vote then you should vote by mail. It’s hard to message that.
Retweeted by Nate Silver
This is another reason why people need to think carefully when writing about the chance that Trump will steal the election. IMO, there is a legitimate threat. However, the threat itself can be used to create ambiguity and sew doubt about the outcome, and possibly reduce turnout.…
Overall, state polls have been a bit better than national polls for Biden, implying about an 8-point popular vote win. Although, Biden's margin in our national polling average has bounced back up to 7.4 points; there was a bigger gap a week or two ago.
In other news, here's a bit from me on what you can learn about the *national* race from state polls:…
Another question is whether Democrats recommend that their voters vote by mail. You've started to see some pundits recommend against it because of Trump. But political operatives in both parties usually encourage it as it tends to boost turnout.…
One reason it's important to *actually assess the probability* (in broad strokes, hard to be super precise) of Trump successfully stealing the election is that you're starting to see recommendations that might *otherwise hurt Biden's chances* in response to the perceived threat.…
For example, putting resources into Ohio, as this thread advocates, is possibly a bad move for Biden under ordinary circumstances. It's not a disaster; there's an outside chance that things get weird & Ohio is the tipping-point state. But it's less important than, say, Wisconsin.
Very much agree w/this thread. I think what's been annoying me is that it's time for clear-headed, fairly precise thinking about exactly what mechanisms Trump could use to steal the election and what the checks might be. There's been some of that—but also a lot of doom-porn.…
Yeah, I don't think we're disagreeing at all. People should be more worried about 1%, 2%, 5%, 10%, 15% probabilities, etc. when the consequences are very bad, in this case basically the end of our democracy as we know it.…
I'd say "Indiana's doing the full Sweden" except Sweden actually has a lot more restrictions than this.……
2. We've spent more time estimating the relationship between congressional and presidential voting and will now assume a stronger correlation than before. This makes little difference to the topline but will allow us to explore more scenarios, e.g. the chance of a D trifecta.
These versions tend to be slightly worse for Collins, FWIW. We'll also be doing this for the presidential race, but it doesn't make much difference there.
Twiends™ uses the Twitter™ API, displays it's logo & trademarks, and is not endorsed or certified by them. These items remain the property of Twitter. We do not sell followers, we only provide display advertising. Bots & fake accounts are not permitted on twiends. © 2009
Grow Your Twitter Free
Want To Grow Your Twitter?
We help other people find and follow you on Twitter.
Key Info:
Started in 2009
Over 6 million signups
Country targeting provided
We never auto tweet to your timeline
We never auto follow others
We actively moderate our community
Please Share
Please upgrade your browser  chrome