jcpolls (@jcpolls) header
jcpolls (@jcpolls) profile picture

Founder & CEO https://xitter.cf/t.co/V7AWfO6lSh

Palo Alto, CA
Joined October 2011
Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)Photo by jcpolls (@jcpolls)
Anthropic's 5,300 founders and employees are now worth more than the bottom third of U.S. households (45 million families).
403
1,915
473
9,027
1,243,408
BREAKING: U.S. nationwide ground stop lifted as of 8:15 PM ET.
Multiple airline sources tell @theaircurrent that the U.S. nationwide ground stop is set to be imminently lifted. Awaiting official word from FAA and the White House.
28
190
29
523
119,510
The Mississippi story: economic discontent, overwhelming Black support, Trump crossover, and a strong showing among moderates. The education and gender findings remain disputed across voter polls. opinionstandard.substack.com…
2
162
Why California’s billionaire tax failed // And why it might not next time opinionstandard.substack.com…
2
3
208
Jetting west after having a blast with @ChrisCillizza and @chucktodd on the #Projection #DDHQ
1
1
5
4,247
Here's a refreshed view (430p ET Nov. 4) of the three election-day estimates of group voting, now featuring an average!
125
Dreamy to have THREE estimates of voting by group -- here's how they compare (at 10a ET Nov. 4)
5
7
613
Pattern #1 from scoring 629 final polls: the 2026 error was mostly one shared miss, not 58 separate ones. 88% of pollsters understated Democrats, and a poll's error correlated 0.67 with the average error of the other pollsters in the same race.
2
1
1
3
177
Pattern #6: mode mattered. Live-caller phone polls were 2026's most accurate: +2.8 bias, 4.5 average error across 55 final polls — and better still in the final week (2.7). Online panels: +4.1/5.2. Mixed modes: +4.4/6.1. Caveat: the phone group leans heavily on Siena and Suffolk.(drop those and mode effect goes away).
1
28
And what didn't matter: sample size. Polls with 1,000+ interviews missed by as much as polls under 600, and likely-voter vs registered-voter screens showed no difference in bias. Bigger and more filtered didn't mean righter.
21
Final scorecard on 2026 polling: across 629 final Senate and Governor polls, the average miss was 5.8 points — and it wasn't random. Polls understated Democrats by 4.1 points on average. 85% still picked the right winner.
1
4
193
Breaking it down: Senate final polls (350 of them) missed by 5.7 points on average and understated Democrats by 3.7. Governor finals (279) missed by 6.0 points and understated Democrats by 4.6. Same direction of error in both — just a bit worse in the governor races.
1
1
90
Texas reinforced the youth story. Talarico was D+56 among voters under 30, up 48 points from the 2024 Senate exit and the highest in the Texas series going back to 2000.
6
242
North Carolina and South Carolina show how broad the gains were. Roy Cooper’s exit poll margin was D+19, up 23 points from the 2022 Senate exit. Democrats went from D+8 to D+59 among voters under 30 and from D+6 to D+30 among independents, both highs in the state’s Senate exit poll series going back to 2004. In South Carolina, Annie Andrews went from a tied independent vote in the 2020 Senate exit to D+32, also a series high. With about 94% and 93% of expected votes counted, respectively, the returns had Cooper up 19 and Andrews up 4.
2
182