Nate Cohn

Nate Cohn @Nate_Cohn

On some of the issues most hotly debated on this website the findings are more equivocal. To take the phrase Latinx, 18% say they're offended by it. That's not nothing, but it's not exactly a huge number either — PolitiTweet.org

Nate Cohn @Nate_Cohn

There's been a lot of debate about the Hispanic vote, often without a ton of evidence. In general, the poll suggests that economic issues are a more important factor than cultural issues -- though there's weakness for Democrats on guns, crime, illegal immigration, etc. — PolitiTweet.org

Nate Cohn @Nate_Cohn

Democrats lead by 24 points among Hispanic voters, according to an NYT/Siena oversample of Hispanic voters. That's very similar to 2020. On the one hand, there's no GOP breakthrough; on the other hand, Democrats struggled among Hispanics in '20. https://t.co/gTGAP2P9dJ — PolitiTweet.org

Nate Cohn @Nate_Cohn

@doug_rivers most of the deff is on the oversample, not the rake weighting — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob it also includes data from some polls that simply can't be used for subgroup-level analysis — PolitiTweet.org

Nate Cohn @Nate_Cohn

On a wonkier note -- it gave us a home for a more detailed methodology than we've written in the past — PolitiTweet.org

Nate Cohn @Nate_Cohn

maybe not quite the headline to our poll yesterday, but @williampd succeeded where many had failed before: create publishable crosstabs for an NYT poll. I think they look great. https://t.co/1lUlNR3EdJ — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @ksadjadpour: The rumors of the poor health of Iran’s Supreme Leader @khamenei_ir appear to be true, as reported by the NYT. His death w… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@pollhannes that's right. we've been mulling whether to weight on it, though i remain reluctant given that it would have made nearly all of our '20 polls worse. there are also hallmarks of old fashioned measurement error -- the non-biden/trump vote is GOP in both of our polls so far — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @KSoltisAnderson: “Republicans would lead by six percentage points in the race for Congress, if they could merely win over voters who sa… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Democrats lead by 2 among registered voters on the generic ballot in the new NYT/Siena poll. Joe Biden's approval rating is up 9 points since July https://t.co/MCHglkWFVI — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @MULawPoll: In new Marquette Law School Poll, 49% of likely voters in WI support Republican incumbent Johnson in the Senate race and 48%… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob @Jlrube agreed — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob wut — PolitiTweet.org

Nate Cohn @Nate_Cohn

This is the second post of our newsletter heading into the midterm campaign, and I won't always be tweeting this stuff out--so sign up if you're interested! https://t.co/rkT5rQjnAt — PolitiTweet.org

Nate Cohn @Nate_Cohn

There are a couple of reasons why this midterm might really be 'different' (Or at the very least, there's something different about the reasons this time might be different) https://t.co/AHr72DoGlH — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @Neil_Irwin: CPI comes in hot. +0.1% headline, +0.6% core. — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @billbarnwell: Letting 39 seconds run, taking a timeout, and attempting a 64-yard field goal (and then insisting on using your two timeo… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Now I should note that there is also a relationship between Democratic overperformance in '22 and the poll error in 2018! The '18 polls did overestimate Democrats in a lot of these same places, like OH, PA, FL. It's not as strong of a relationship as '20, though. — PolitiTweet.org

Nate Cohn @Nate_Cohn

In 2018, there wasn't really a relationship between Democratic standing and the '20 poll error. Indeed, Democrats were often underperforming expectations in the Senate--even in places where polls were wrong in '16 — PolitiTweet.org

Nate Cohn @Nate_Cohn

In principle, I'm totally agnostic about whether 2018 or 2020 ought to be the best comparison for 2022. The warning sign, though, is that current Democratic overperformance is fairly well-correlated with 2020 error (just as '20 was with '16) https://t.co/UTBFs41eLl — PolitiTweet.org

Nate Cohn @Nate_Cohn

I do want to reiterate what I said in the article -- IDK if the polls will be as bad as they were in '20, or more like '18, or even better. But if Dems show strength in these areas, I'd guess that this geographic pattern of error will probably be back to one extent or another — PolitiTweet.org

Nate Cohn @Nate_Cohn

This is an old ppt slide from May 2020, showing the apparent shift in early '20 polling v. the poll bias in '16. The warning signs were there from the start (inspiring the 'if polls are wrong' table). Also notice the same pattern in '18, tho the polls weren't biased on *average* https://t.co/zMyqGTaSVD — PolitiTweet.org

Nate Cohn @Nate_Cohn

@simon_bazelon come on, surely you 'understand the line.' there was ~ no national bias in '18 and that's an incredibly significant difference. if there's no national bias rn, democrats really are in a competitive fight for the house and favored in the senate — PolitiTweet.org

Nate Cohn @Nate_Cohn

You can sign up for our newsletter on elections, politics and polling here: https://t.co/rkT5rQiPKV — PolitiTweet.org

Nate Cohn @Nate_Cohn

"Expectations" are the FiveThirtyEight fundamentals. In Wisconsin, for ex., the fundamentals are R+2, the poll average is D+4, so that's Democrats +6 v. expectations. That happens to be fairly similar to the D+8 polling error in 2020 -- a pattern in the Senate battlegrounds — PolitiTweet.org

Nate Cohn @Nate_Cohn

First newsletter post: the polling warning signs are flashing again. Democrats are outpacing expectations in exactly the states where the polls overestimated Joe Biden in '20 https://t.co/Kasyy2m0zc https://t.co/Qa8MCMdg0Z — PolitiTweet.org

Nate Cohn @Nate_Cohn

@RyanDEnos same pattern existed in '18, fwiw, just with an intercept shift v 16/20. still a tough sign for Democrats in a lot of these Senate races https://t.co/GQvzBVoSp2 https://t.co/RfAZsz4drz — PolitiTweet.org

Nate Cohn @Nate_Cohn

If it sounds interesting to you, I hope you'll sign up! https://t.co/vfLG2cOZH5 — PolitiTweet.org

Nate Cohn @Nate_Cohn

It'll also be a natural home content that doesn't have a spot on the Times homepage--the sort of stuff that I've just tweeted out in the past: --announcements about polls --wonkier stuff, like the findings from our 2020 autopsy We can also have some fun. — PolitiTweet.org