Great set of final NYT polls for the Democrats, particularly Michigan, Texas and Kansas.
Also they changed their methodology from previous cycles which actually boosts the Republicans. Image on the right shows what they'd be on the old methodology.
Feeling a bit more confident in my predictions...
Also they changed their methodology from previous cycles which actually boosts the Republicans. Image on the right shows what they'd be on the old methodology.
Feeling a bit more confident in my predictions...
November 2, 2026 at 7:35 AM
Great set of final NYT polls for the Democrats, particularly Michigan, Texas and Kansas.
Also they changed their methodology from previous cycles which actually boosts the Republicans. Image on the right shows what they'd be on the old methodology.
Feeling a bit more confident in my predictions...
Also they changed their methodology from previous cycles which actually boosts the Republicans. Image on the right shows what they'd be on the old methodology.
Feeling a bit more confident in my predictions...
ridiculous map. Russia did not “invade” Cuba. Conversely it *did* invade Germany, but you’re not counting that bc it would lead to people asking some obvious questions about your methodology.
If russia hasn’t attacked your country yet, just wait a little…
November 1, 2026 at 4:51 PM
ridiculous map. Russia did not “invade” Cuba. Conversely it *did* invade Germany, but you’re not counting that bc it would lead to people asking some obvious questions about your methodology.
So anyway here goes, here's my thinking on the midterms:
1) Polling is cooked. The methodology behind modern polling is ridiuclous — incredibly low response rates, batshit weights assigned by pollsters to tiny samples, fradulent polling groups literally using chatGPT to make up fake results.
1) Polling is cooked. The methodology behind modern polling is ridiuclous — incredibly low response rates, batshit weights assigned by pollsters to tiny samples, fradulent polling groups literally using chatGPT to make up fake results.
November 3, 2026 at 5:09 AM
So anyway here goes, here's my thinking on the midterms:
1) Polling is cooked. The methodology behind modern polling is ridiuclous — incredibly low response rates, batshit weights assigned by pollsters to tiny samples, fradulent polling groups literally using chatGPT to make up fake results.
1) Polling is cooked. The methodology behind modern polling is ridiuclous — incredibly low response rates, batshit weights assigned by pollsters to tiny samples, fradulent polling groups literally using chatGPT to make up fake results.
You can talk about aggregation, you can talk about averaging, you can talk about whatever I do not think the methodology behind 2026 polling *can* be accurate. I think pollsters have kludged together reasonable guesses by fucking with demographic weights and making assumptions.
November 3, 2026 at 5:12 AM
You can talk about aggregation, you can talk about averaging, you can talk about whatever I do not think the methodology behind 2026 polling *can* be accurate. I think pollsters have kludged together reasonable guesses by fucking with demographic weights and making assumptions.
"listening is unscientific" is terrrrrrible methodology
November 2, 2026 at 6:25 AM
"listening is unscientific" is terrrrrrible methodology
Is there signal there? Sure. Yes. It is better to be leading polls than losing them. But I think pollsters are (mostly) clustering around 50/50 calls for every close race to cover up the holes in their methodology and that doesn't actually tell us much.
November 3, 2026 at 5:16 AM
Is there signal there? Sure. Yes. It is better to be leading polls than losing them. But I think pollsters are (mostly) clustering around 50/50 calls for every close race to cover up the holes in their methodology and that doesn't actually tell us much.
lol at it already
November 3, 2026 at 3:00 AM
lol at it already
[GIVING TED TALK ON RESEARCH METHODOLOGY] you’ve heard of the grapefruit method
[CROWD SHIFTS IN THEIR SEATS]
[ME, AN EMPATH, SENSING DISCOMFORT OF MY AUDIENCE, SWITCHING TO A TIGHT 5] well let me share with you the sandpaper method
[CROWD, SILENT]
[SWALLOWING LOUDLY INTO MIC] um, so methodology
[CROWD SHIFTS IN THEIR SEATS]
[ME, AN EMPATH, SENSING DISCOMFORT OF MY AUDIENCE, SWITCHING TO A TIGHT 5] well let me share with you the sandpaper method
[CROWD, SILENT]
[SWALLOWING LOUDLY INTO MIC] um, so methodology
November 3, 2026 at 5:13 AM
[GIVING TED TALK ON RESEARCH METHODOLOGY] you’ve heard of the grapefruit method
[CROWD SHIFTS IN THEIR SEATS]
[ME, AN EMPATH, SENSING DISCOMFORT OF MY AUDIENCE, SWITCHING TO A TIGHT 5] well let me share with you the sandpaper method
[CROWD, SILENT]
[SWALLOWING LOUDLY INTO MIC] um, so methodology
[CROWD SHIFTS IN THEIR SEATS]
[ME, AN EMPATH, SENSING DISCOMFORT OF MY AUDIENCE, SWITCHING TO A TIGHT 5] well let me share with you the sandpaper method
[CROWD, SILENT]
[SWALLOWING LOUDLY INTO MIC] um, so methodology
Ah bringing metrics based business methodology to sermon prep, what could go wrong with this?
October 31, 2026 at 12:53 PM
Ah bringing metrics based business methodology to sermon prep, what could go wrong with this?
I'm in the Department of Anesthesiology, Pharmacology, and Therapeutics and the Therapeutics Initiative @drug-evidence.bsky.social
I'll be doing all the causal inference & translational methodology research you love, plus helping the BC government answer pressing questions about medical treatments!
I'll be doing all the causal inference & translational methodology research you love, plus helping the BC government answer pressing questions about medical treatments!
November 2, 2026 at 9:42 PM
I'm in the Department of Anesthesiology, Pharmacology, and Therapeutics and the Therapeutics Initiative @drug-evidence.bsky.social
I'll be doing all the causal inference & translational methodology research you love, plus helping the BC government answer pressing questions about medical treatments!
I'll be doing all the causal inference & translational methodology research you love, plus helping the BC government answer pressing questions about medical treatments!
Give Olivia Reingold some credit, this is a major leap in journalistic methodology from "I read some tweets"
November 2, 2026 at 11:55 PM
Give Olivia Reingold some credit, this is a major leap in journalistic methodology from "I read some tweets"
The sphinx, visibly pissed off: you’re correct but I’d eat you for the methodology if I was allowed to
November 3, 2026 at 5:28 PM
The sphinx, visibly pissed off: you’re correct but I’d eat you for the methodology if I was allowed to
Reminder: Polls don't vote; people vote. So please, get out and vote first! But still, this is an interesting discussion re the methodology of polling and a reminder that polls are snapshots in time and are not static facts.
www.nytimes.com/2026/10/28/i...
www.nytimes.com/2026/10/28/i...
Can You Trust the Midterm Polls? Three Times Journalists Talk It Out. (Gift Article)
Members of the team behind the Times/Siena Poll on what the data show, how they’ve improved their methodology and what it means to be “right.”
www.nytimes.com
November 2, 2026 at 12:28 PM
Reminder: Polls don't vote; people vote. So please, get out and vote first! But still, this is an interesting discussion re the methodology of polling and a reminder that polls are snapshots in time and are not static facts.
www.nytimes.com/2026/10/28/i...
www.nytimes.com/2026/10/28/i...
I actually appreciate that he showed what the old modeling result would be vs. the new methodology. Made me feel better about the current projections
November 2, 2026 at 4:07 AM
I actually appreciate that he showed what the old modeling result would be vs. the new methodology. Made me feel better about the current projections
#Stonewall
This goes for the UK government on human rights too.
If they hone a methodology for destroying the #HumanRights of one group, they create a weapon that can be turned against anyone.
This goes for the UK government on human rights too.
If they hone a methodology for destroying the #HumanRights of one group, they create a weapon that can be turned against anyone.
Hell, the entire queer community. S'onewall has shown they'll abandon one group, we now have to believe they'll abandon any other group too.
November 1, 2026 at 11:24 AM
#Stonewall
This goes for the UK government on human rights too.
If they hone a methodology for destroying the #HumanRights of one group, they create a weapon that can be turned against anyone.
This goes for the UK government on human rights too.
If they hone a methodology for destroying the #HumanRights of one group, they create a weapon that can be turned against anyone.
this place's Andrews+5 is unfortunately a junk methodology if you ask me
Little better than a push poll
static1.squarespace.com/static/54512...
Little better than a push poll
static1.squarespace.com/static/54512...
November 3, 2026 at 12:13 AM
this place's Andrews+5 is unfortunately a junk methodology if you ask me
Little better than a push poll
static1.squarespace.com/static/54512...
Little better than a push poll
static1.squarespace.com/static/54512...
siena is probably the best pollster in the US. not always the most accurate but their methodology is grounded, measured, statistically sound, etc. and NYT poll team is very good.
Yesterday Cohn told us they saw some wacky raw numbers showing insane Dem leads and they decided to "deskew" em bc vibes
Yesterday Cohn told us they saw some wacky raw numbers showing insane Dem leads and they decided to "deskew" em bc vibes
November 3, 2026 at 1:40 AM
siena is probably the best pollster in the US. not always the most accurate but their methodology is grounded, measured, statistically sound, etc. and NYT poll team is very good.
Yesterday Cohn told us they saw some wacky raw numbers showing insane Dem leads and they decided to "deskew" em bc vibes
Yesterday Cohn told us they saw some wacky raw numbers showing insane Dem leads and they decided to "deskew" em bc vibes
I’ve had an unintentional experiment using this methodology on one of my YT channels that got discovered by Lula supporters the day after the Brazilian election.
Result: more views on one video than I’ve ever had in ten years doing news analysis.
Brazil has written the new political guidebook.
Result: more views on one video than I’ve ever had in ten years doing news analysis.
Brazil has written the new political guidebook.
From coordinated social media campaigns to reclaiming national symbols, the online strategy mobilized many to vote for Lula, in an effort that experts say offers a striking example of how the Brazilian left learned to fight the right on its turf. https://cnn.it/48sYbEp
November 2, 2026 at 10:58 AM
I’ve had an unintentional experiment using this methodology on one of my YT channels that got discovered by Lula supporters the day after the Brazilian election.
Result: more views on one video than I’ve ever had in ten years doing news analysis.
Brazil has written the new political guidebook.
Result: more views on one video than I’ve ever had in ten years doing news analysis.
Brazil has written the new political guidebook.
Roger, Roger! Ready to bring your B1 Battle Droids to life? Follow along with our latest AMP Painting Methodology to learn how to paint their iconic armor, complete with highlights and battle damage!
November 3, 2026 at 2:00 PM
Roger, Roger! Ready to bring your B1 Battle Droids to life? Follow along with our latest AMP Painting Methodology to learn how to paint their iconic armor, complete with highlights and battle damage!
Like the polls are the polls. The Times should explain its methodology but Healy is basically admitting that he's sensitive to whether right wing complainers will see it and "not trust it." It is an admission that is consistent with other's complaints that he's a ref being worked.
November 2, 2026 at 3:23 PM
Like the polls are the polls. The Times should explain its methodology but Healy is basically admitting that he's sensitive to whether right wing complainers will see it and "not trust it." It is an admission that is consistent with other's complaints that he's a ref being worked.
An extremely comprehensive guide to the kind of information surveys should be able to share about their methodology, and why it matters:
Transparency is necessary in evaluating polls, but readers must understand what the disclosed information means. With immense help from 2026 W.E.B. Du Bois Summer Scholar Nancy Toure, we are excited to publish a guide to understanding disclosure items: https://bit.ly/4ifsy5F @aapor.bsky.social
November 2, 2026 at 7:40 PM
An extremely comprehensive guide to the kind of information surveys should be able to share about their methodology, and why it matters:
github.com/obra/superpowers and github.com/ovid/paad together work well for me flow is sp:brainstorm paad:pushback sp:write-plan paad:alignment sp:subagent-implementation - both have other useful skills in them but that is main flow
GitHub - obra/superpowers: An agentic skills framework & software development methodology that works.
An agentic skills framework & software development methodology that works. - obra/superpowers
github.com
November 1, 2026 at 9:33 PM
github.com/obra/superpowers and github.com/ovid/paad together work well for me flow is sp:brainstorm paad:pushback sp:write-plan paad:alignment sp:subagent-implementation - both have other useful skills in them but that is main flow
Right, exactly: the most useful issue polling is trendline polling - where you ask an identical question using the same methodology many times over a long period. Presidential approval, economic sentiment - there are a few topics that get this treatment.
But most of it is one-off.
But most of it is one-off.
It is probably possible to do issue polling that's at least somewhat meaningful (shifts in how the exact same wording gets answered over time are probably capturing something real, for instance)
But my *god* the vast majority of it is just air
But my *god* the vast majority of it is just air
November 2, 2026 at 2:45 AM
Right, exactly: the most useful issue polling is trendline polling - where you ask an identical question using the same methodology many times over a long period. Presidential approval, economic sentiment - there are a few topics that get this treatment.
But most of it is one-off.
But most of it is one-off.
STARTS TOMORROW (4th Nov 2026)! 👩🏻💻 🌳Join me as I run a 4-week course with the Society of Genealogists on 'How to be More Confident in your Family Tree'. Four talks presented on four consecutive Wednesdays covering host of methodology topics (see comments)
Book here: portal.sog.org.uk/Event/view/2...
Book here: portal.sog.org.uk/Event/view/2...
November 3, 2026 at 10:32 AM
Is their methodology still ridiculous like it was in 2020?
November 2, 2026 at 9:34 PM
Is their methodology still ridiculous like it was in 2020?