If NYT/Siena is making you panic, take note that Harris is +12 among women and age 18-29, but NYT fiddles with the numbers to suppress them.
NYT assumed electorate:
52% women
13% age 18-29
Actual electorate 2008, 2012, 2016, 2020:
54-55% women
15-16% age 18-29
NYT assumed electorate:
52% women
13% age 18-29
Actual electorate 2008, 2012, 2016, 2020:
54-55% women
15-16% age 18-29
Comments
https://bsky.app/profile/brianklaas.bsky.social/post/3l76czzghuz27
Long answer is a bit more complicated, and revolves around "recalled vote" and other data proxies pollsters use.
But short answer is yeah, a lot of this comes from trying to avoid underestimating Trump support.
It's like knowing a hand has to fit inside a glove but not quite being able to figure it out so just lopping off a few fingers.
I've called out the issues with sampling pools in NYT/Siena before https://threadreaderapp.com/thread/1845207232125993011.html
So I'd not put it past them to TRY something
"We win if we properly count people for whom voting is hard" is bad news.
But, sure, stick to Pew's numbers for 2016. The NYT/Siena composition is still off:
NYT vs Pew 2016
Women 52% vs 55%
Dems 31% vs 35%
However we slice it, NYT's assumed 2024 electorate is more favorable to Trump than 2016 or 2020 were.
Dont want to put too much cold water on this but we just don’t know from EV data.
Much of the Dobbs Effect is functionally unpollable (between the serious undersampling of women of reproductive age & other factors)
I can't even.
If this group turns out to be 17% of the electorate vs. 13%, that would make a 0.5% difference in Harris's popular vote margin.
She needs a 2.0% margin or so to win the electoral college.