🚨 🚨 A new Maine Senate general election poll from @ZenithPolls — on behalf of @MorePerfectUS — finds Susan Collins in a tough position in her bid for a sixth term. After voters read bios of both Dem candidates, Graham Platner outperforms Janet Mills by 6 pts. Let’s dive in 🧵

How demoralizing for our Asian American communities to be excluded again and again. If your pollster doesn’t prioritize you, it’s time to break up. We can do better. Eg. @ZenithPolls

(Verasight ended up getting more than n=1200 completes, this was sent before we were fully out of field)

Hi there! We provided an extraordinary amount of detail in how we obtained sample for this poll. Radical transparency, especially around sampling and models and weighting, is at the core of everything we do. Please see below, and the full deck here: https://t.co/hMLKO7ueqs

New - Mayor poll - New York City 🔵 Mamdani 50% 🟡 Cuomo 22% 🔴 Sliwa 13% 🟡 Adams 7% Zenith research #C - 1000 LV - 7/25

🎙️ Polls, Power & 2025 🗳️ I sit down with veteran pollster ⁦@admcrlsn⁩ of Zenith Research to break down what the NYC polls got right—and where they missed. Methodology, messaging & trust in the data. 🔗 https://t.co/HTeSssgRob

Zohran Mamdani is set to win the majority of Jews under 45 in the NYC mayoral election with 67% support, according to a Zenith Research and Public Progress Solutions poll of 152 likely voters. Mamdani leads with 43% of the Jewish vote. Cuomo has 26% overall and 33% among

🇺🇸#USA, New York City mayoral election poll: Mamdani (D): 50 % Cuomo (I): 22 % Sliwa (R): 13 % Adams (I-inc): 7 % Walden (I): 1 % Verasight/Zenith Research/PPS, 24/07/25

Also in Zenith Research/Public Progress Solutions poll: Mamdani has 43% of the Jewish vote, including 67% among younger Jewish voters aged 18-44. Cuomo at 26%, and 33% among Conservative/Orthodox Jews. Adams at 15% with Jews, at 27% with Con/Ortho. (Sample 152 LV)

That tracks with the crosstabs of my and @amitsinghbagga’s recent poll, which had Mamdani leading with 35% of Jewish registered voters. Though always better to get a larger sample size just polling one population (n=800) than a standalone crosstab subgroup (n=186)!

Fun fact: I played a small role in offering feedback for the framing and sampling of this poll. Special thanks to Adam for doing his due diligence to first call those of us who work with local communities on the ground.

Our poll’s sample size wasn’t large enough to break out NYC Hindus by age (even though it’s the largest of the cycle thus far), but I wouldn’t be surprised if there was a stark generational difference among Hindus re: Mamdani, like we are seeing across nearly every other subgroup

Here’s the full segment of me and @amitsinghbagga on Fox 5 that aired last night talking about our new, uniquely conducted poll of the New York City mayoral general election. Thank you so much to @morganfmckay for having us on!

Check out our full conversation with @ZenithPolls founder @admcrlsn, where we reflect on polling in the NYC mayoral primary, how Democrats are responding to #ZohranMamdani, and his litmus test for candidates in the future. https://t.co/begMcNUsec

Will happily answer any questions about methodology or the crosstabs! Leave them below.

Boom there you go. Pew ends the debate over weighting on past vote

I agree! It makes everyone smarter Pollsters can be very territorial re: their “secret sauce” / sensitive about criticism We’re all just doing our best to capture the moment accurately & inevitably fall flat sometimes But it’s critical to understand why so we can learn from it

I’d love to see other pollsters join in public post mortem discussions. There is a lot folks can learn by engaging in conversation with each other instead of consultants feeling like they need to have all the answers and discouraging questions about assumptions and methodologies.

6) Prioritizing respondent experience, ensuring answering polls are not too onerous for any demographic group 7) Pairing quantitative polls with qualitative focus groups to get a fuller picture of a particular race, issue, or topic — and releasing results to public when possible

4) Exploring new methodologies (e.g., interactive AI-based interfaces) to try to combat the ever-growing challenge of non-response 5) Conducting polls in multiple languages in areas where relevant, maximizing respondent comfort in answering in their preferred language