Guy who hates AIPAC because they assume it’s Analyst Institute

NERV IRB Coordinator@professormusgrave.bsky.social · 4d ago

Guy who hates AIPAC because they assume it's AI PAC

Glad Mexico lost and their fans will be out of the World Cup :)

Eric Michael Garcia@ericmgarcia.bsky.social · last mo.

And there goes the “Puto” chant again. Mexican fans know that FIFA is not gonna pause this game. So they are going to do it as much as they want.

Can highly recommend this pairing, I’m very much on my bullshit this week

This implies the existence of the LPM Group, yes? If so, I’m sure they’re sworn enemies, competing for the same work.

Thinking of this basic point from Ansolabehere and Fraga today for no particular reason.

They posted the raw data for this, fascinating: Notes so far: 1. It looks like these responses exist to make the weights converge, given the synth respondents have NA for non-weighting Qs? 2. The MoE excludes the synthetic respondents. I guess that's better than treating them as 1:1 for humans?

The Public Sentiment Institute - National Poll (March 20th) - March 21 Release

docs.google.com

Jacob Rubashkin@rubashkin.bsky.social · 5mo ago

Imagine trust in polling being at an all time low and then doing “polls” where a quarter of respondents are AI agents hallucinating answers based on exit polling.

Bayesian friends- if you’re curious to try out the blazingly fast nutpie sampler in R, I just put together a pretty lightweight package that’ll compile your existing Stan models!

GitHub - andytimm/nutpieR

Contribute to andytimm/nutpieR development by creating an account on GitHub.

github.com

I have a joke about bayes factors but nobody really thinks I should tell it

Thom Volker@thomvolker.bsky.social · 5mo ago

I have a joke about p-values but nobody gets it

A big part of the way I think about survey weighting is through connections to causal inference. One thing that emerges from that perspective: regularized weights are pretty, pretty good! I really like @shiraamitchell.bsky.social ‘s discussion of this connection here, esp the Johansson reference.

Andy Timm@andytimm.bsky.social · 5mo ago

New R package release day, woo :) Regularized raking makes it easier to build complex survey weights that reduce bias without paying as heavy a high variance price. regrake makes building these weights convenient in R. Check it out!

New R package release day, woo :) Regularized raking makes it easier to build complex survey weights that reduce bias without paying as heavy a high variance price. regrake makes building these weights convenient in R. Check it out!

regrake: Regularized Raking in R – Andy Timm

New year, new package release

andytimm.github.io

This is also the face I make when doing bayes, very cool mr. coypu! Looks like a cool, short bayes intro book!

Olivier Gimenez 🦦@oaggimenez.bsky.social · 5mo ago

📘 Nouveau livre ! Introduction à la statistique bayésienne avec le logiciel R Découvrez pas à pas la statistique bayésienne et sa mise en œuvre dans R : théorème de Bayes, MCMC, priors, régression, GLM/GLMM et comparaison/validation de modèles. Aux éditions Quae 🙏 📷 Crédit photo : Yann Raulet

AAPOR program’s up. So much cool work this year- really stoked to hear about all of it! I’ll be sharing some of GP’s work on bot/LLM detection, and also our work on survey experiment designs that manipulate attention in-survey to understand how noisier environments modify ad effects.

I finished CS336 (Language Modeling from Scratch) from @stanfordnlp.bsky.social It was a fantastic way to get back into the weeds with recent LLM training/arch advances and strengthen my systems engineering skills. I wrote up a short review + some practical tips for other auditors-

More like the tech wrong

Could. Not. Resist.

That new Bonica et al paper that's being passed around with claims that moderating hurts turnout doesn't have much turnout analysis in it at all, and certainty not enough for that conclusion open.substack.com/pub/cauldron...

So I was thinking about MIT’s “missing semester” course (missing.csail.mit.edu) on skills needed but often not explicitly taught. What are some stats/survey stats equivalents y’all find super important- that is, important for research/practice, not commonly classroom taught?

This is an absolute scandal. Trump and Marco Rubio abandoned American workers and their families abroad, delaying waivers to provide for their safe return to the United States. As these civil servants lost their possessions, housing, and money, Elon Musk was slandering them with bogus accusations.

Aaron Fritschner@fritschner.bsky.social · last yr.

Via Robbie Gramer at Politico, in a legal filing a USAID employee details trying to leave Kinshasa, Congo with their family and leaving their possessions behidn amid panic created by Trump and Musk's sudden shutdown of the agency

Don’t love the guy to put it mildly, but I have a ton of respect for him here.

Barb McQuade@barbmcquade.bsky.social · last yr.

Sen. Josh Hawley rebukes Vance and allies for suggesting that court orders should be disobeyed. Great to see a GOP senator speaking out against this anti-American idea.

Reminder that ipums.org has all the acs, census, CPS, NHIS, MEPS, and so much more. Bonus that it’s cleaned and harmonized in ways that make these data really easy to use in practice.

Catherine Rampell@crampell.bsky.social · last yr.

The American Community Survey microdata have been taken offline. Questionnaires are down too.

AMFAR has published a report on how the foreign aid freeze affects PEPFAR, which actually still hasn't resumed. They estimate that hundreds of thousands of people lose access to HIV antivirals and 1500 babies are expected to contract HIV — *each day* 🧵

Thanks very much to everybody who chimed in on this thread about attention --I wrote up some notes! My conclusion from the discussion is that polisci should coordinate on an info board / faux feed design to study (probs v. heterogeneous effects on attention) alexandercoppock.com/attention.pdf

Alex Coppock@aecoppock.bsky.social · 2y ago

So far persuasion work in polisci has mostly centered on message -> opinion (main result imo: messages have small fx in the right direction that don't differ too much by message or person or issue) with less work on message -> attention need to figure out experimental paradigm for attention

🤡

Marisa Kabas@marisakabas.bsky.social · 2y ago

🚨 BREAKING: OMB has just issued a memo *rescinding* the previous memo freezing all federal financial assistance programs. Full text, per government source:

Like Bayes, Jaynes, thinking about prior specification, or Kangaroos drinking Fosters? New post: andytimm.github.io/posts/kangar...

Kangaroos, Foster’s, and E.T. Jaynes – Andy Timm

Also maybe some thoughts on priors

andytimm.github.io

I’ve been thinking about this tension lately- Trump gets credit for grabbing attention well, but a key story of last cycle is his strength with low-engagement folks. One way to resolve this is that making news might sort of sieve-like; you need to be constantly high viz everywhere to get through.

Dan Hopkins@dhopkins1776.bsky.social · 2y ago

I agree w/ @ezrakleinbot.bsky.social that attention is a critical political currency, but also think it's noteworthy that Trump's GOP is especially strong with lower-turnout voters, who pay less attention to politics. The attention Trump draws may not always help him.

Our product and engineering teams are both hiring! Grow Progress is a wonderful place to work, and our platform supports highly impactful candidates and organizations on the left. Come joins us! Shoot me a DM if you’re interested and have questions. Links in 🧵

Another incredible essay. Some striking passages to get you interested:

Rafe Meager (they/them)@economeager.bsky.social · 2y ago

Sometimes you have to write “Truth and Probability” just because Margaret won’t fuck with you. open.substack.com/pub/rottenan...

Gottheimer keeps using the "hardworking families" line but apparently does not apply that term to hardworking families who rely on the bus or the train. The only people who matter to him are the small number of people who drive into Manhattan.

Second Ave. Sagas@2avesag.as · 2y ago

Josh Gottheimer, get a freaking grip already. Enough.

Unfortunate news for those of hoping to create a healthy “niche election maps guy” ecosystem over here

Nathaniel Pendleton@npendleton.bsky.social · 2y ago

Update: #BlueSky apparently down samples maps, lowering resolution & readability, (phone image,) making my map mostly unreadable, while Twitter apparently does NOT make maps unreadable. (tablet image) Requires a rethink, eg asking DKElections / DonnerMaps for darker blank cartogram map template.

Bunch of great papers here. Also, not my circus and not my monkeys, what does it say about academic publishing that something as obviously meritorious and practically useful as PSIS took that long to get published? 9 years!?

Progress in 2024 (Aki) statmodeling.stat.columbia.edu/2025/01/07/p...