Nate Silver

Nate Silver @NateSilver538

@mattyglesias That's not quite right because you're not just taking states off the map, you're also informing the model's prior about other states. For example also if you give CO and WA to Ds, it increases their probability quite a bit it removes some "crazy" scenarios from the table. — PolitiTweet.org

Nate Silver @NateSilver538

RT @galendruke: Who wants to talk more about that 32 point rightward shift among independent women in this week's NYT/Siena poll? Great! L… — PolitiTweet.org

Nate Silver @NateSilver538

If this was the stock market, you'd say "we're officially into bear market territory" (for Democrats). https://t.co/phHcIJyYkD https://t.co/WoDb5jCdrT — PolitiTweet.org

Nate Silver @NateSilver538

Ahh yes how dare someone would want to disrupt the neighborhood character of Times Square. https://t.co/JlWpWtxZMY https://t.co/LdZYCOApaU — PolitiTweet.org

Nate Silver @NateSilver538

@SeanTrende Yeah, I feel like RCP/538 figured out how to build a better burger (tastier, healthier, more sustainable and/or cheaper than the competition) and get blamed for not enough people being vegetarian. — PolitiTweet.org

Nate Silver @NateSilver538

These are two interesting articles that I kind of wish had a conversation with one another—educational polarization and polarization around the Q of how much to trust technocratic elites seem like largely the same phenomenon. https://t.co/bmXp2rFSJk https://t.co/vqxEss0B3I — PolitiTweet.org

Nate Silver @NateSilver538

Especially considering the alternative. Campaign coverage 10+ years ago was generally full of cliches, insider groupthink and regurgitated conventional wisdom, with little concern for what voters actually believed. Some of those things have probably improved but idk. — PolitiTweet.org

Nate Silver @NateSilver538

No — PolitiTweet.org

Nate Silver @NateSilver538

Clearer signs of movement back toward the GOP in the Senate than you had a couple of weeks ago. Getting close to tossup territory. https://t.co/GA3xa1GJaG https://t.co/baWOOLoVbJ — PolitiTweet.org

Nate Silver @NateSilver538

Whether the traders are right or not, their estimate is internally consistent. Ds have narrow advantages in 3 key states (PA/GA/AZ). If everything *exactly* goes to form they win the Senate. But it's dicey probabilistically, which traders understand and this journalist doesn't. — PolitiTweet.org

Nate Silver @NateSilver538

The economy is pretty weird, without a lot of good historic precedents, and macroeconomic forecasts have a long history of overconfidence. A recession may be more likely than not, but if your model says 100%, it is a bad model. — PolitiTweet.org

Nate Silver @NateSilver538

Thing about the midterms is that the big picture tends to be slow-moving and doesn't change much from day to day, it's not like a presidential primary or something where you can have wild swings overnight. So, it's usually not worth caring much about any one day's worth of polls. https://t.co/FYOGN8Gcq9 — PolitiTweet.org

Nate Silver @NateSilver538

Oh no. https://t.co/GAIuMRpASf — PolitiTweet.org

Nate Silver @NateSilver538

Yes absolutely no rain anywhere in the NYC area. https://t.co/Y69B34ztBf — PolitiTweet.org

Nate Silver @NateSilver538

It's good pollsters publish demographic crosstabs for transparency reasons. And aggregated over large samples, they can be informative. But most of the crosstab analysis on here is just manufacturing narratives out of random noise, the margins of error are very large usually. — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias There's no shortage of people with proper public health credentials who have expressed similar sentiments. — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias I remember when I'd get yelled at for suggesting some public health elites were motivated by broader political/ideological concerns rather than optimal pandemic policy per se, so it's funny seeing an article that's like "we blew the opportunity to overthrow capitalism!". — PolitiTweet.org

Nate Silver @NateSilver538

A modestly accurate indicator that's uncorrelated with other indicators generally provides more information than a highly accurate indicator that's highly correlated with other indicators, although the Selzer poll has historically been both uncorrelated *and* highly accurate. — PolitiTweet.org

Nate Silver @NateSilver538

The reason this is interesting is not because Grassley is going to lose (probably not) and not even because the Selzer poll has been accurate (though it has been) but because she's a pollster who will publish what her numbers tell her and not herd toward the conventional wisdom. — PolitiTweet.org

Nate Silver @NateSilver538

@mattyglesias IDK Selzer isn't one of those pollsters that's been prone to large errors though! — PolitiTweet.org

Nate Silver @NateSilver538

A lot of overwrought critiques of the media these days. But the way it covers debates is the absolute worst. Any version of "beat expectations" or "cleared a low bar" is about as blatant as putting one's finger on the scale for reductive both-sidesism gets. — PolitiTweet.org

Nate Silver @NateSilver538

@VanessaKade The problem is that polygraphs are borderline junk science to begin. Even under ideal circumstances you'd given them little weight. But this ins't ideal. People can be trained to beat them with relatively little practice, certainly << than you'd get playing poker even for a year. — PolitiTweet.org

Nate Silver @NateSilver538

Some thoughts on the ways in which political betting markets are smart and dumb. In the weeds but some of you will like this column, I think. https://t.co/wT9z1zZONz — PolitiTweet.org

Nate Silver @NateSilver538

C'mon, folks. Polygraph evidence isn't very reliable in general and it's particularly inappropriate for a poker player! The skills you'd need to deceive a polygraph are very similar to the ones you'd need to avoid giving off tells in poker. — PolitiTweet.org

Nate Silver @NateSilver538

RT @SeanTrende: @NateSilver538 I think something that needs to be drilled into young researchers is that sometimes the answer to "it's the… — PolitiTweet.org

Nate Silver @NateSilver538

People aren't direct enough about this: research is hard, and for many/most papers, the research design leaves so much room for error or ambiguity that you can't really conclude anything from it. The peer-review process is not a strong enough filter, especially for newsy topics. — PolitiTweet.org

Nate Silver @NateSilver538

Yeah, I was subtweeting this Scotland COVID study earlier, but it is a Bad Use of Polling. Low response rates, obvious signs of response bias. No real attempt to correct for it, study authors seem unaware of the problems, frankly. Doing good research on this topic is hard. — PolitiTweet.org

Nate Silver @NateSilver538

@SeanTrende @asymmetricinfo I think it's hard to disaggregate (1) sleep, exercise or eating routines often being disrupted during travel days and (2) days before/after travel often being quite busy (if it's a business trip, I have to be more "on" than usual, it's tiring) from the rigors of travel itself. — PolitiTweet.org

Nate Silver @NateSilver538

@TheZvi Yeah, that long COVID survey had all sorts of signs of response bias. ~2/3 of respondents with a symptomatic infection were female. Also older, higher socioeconomic status. Also, people with COVID were much more likely to respond than control group. Bad use of polling. https://t.co/iIiIhny6kr — PolitiTweet.org

Nate Silver @NateSilver538

I'd guess that out of every 10 papers that claim to adjust for confounders/selection effects: 1 actually does a convincing job 2 aren't entirely convincing but do as well as they can 3 go through the motions but leave obvious Qs unanswered 4 should never ever have been published — PolitiTweet.org