Dan Zeltzer
@danzeltzer
Followers
661
Following
2K
Statuses
421
Associate Professor of Economics at @TelAvivUni; visiting @Stanford
Tel Aviv, Israel
Joined October 2013
Check out our new paper on falsification tests for IV!
WP🚨 with @danzeltzer @DanielNevo @B_WNSN & Walk. If you used an IV, you likely ran a falsification or placebo test for exogeneity. We show these tests can be *too harsh*. They can flag issues even if the IV is exogenous.#EconTwitter #CausalInference 🧵>>
1
3
20
Nice review of "Deep Research" for math (tldr not there yet)
For this it essentially read the Wikipedia page on the conjecture, supplemented by a couple of other sources. I more or less agreed with it as to the results it listed (and it was a nice ego boost when it mentioned one of my results). The proof sketches were completely wrong.
0
0
2
RT @OlavSindre: Today, I'm super excited to announce that we've finally launched @tana_inc to the world! After four years in stealth mode,…
0
14
0
RT @HannoLustig: To reduce my bilateral trade deficit with Trader Joe's, I am now going to shop at Whole Foods, which is more expensive and…
0
255
0
RT @ben_golub: Modern supply chains don't look like trade theory 101! They involve constant border crossings, each now hit by tariffs. Ta…
0
735
0
RT @RoeeLevyZ: First time writing a handbook chapter! Always fun to work with Guy, Rafa, and Lena :) Hopefully, this guide will be useful,…
0
12
0
RT @ben_golub: BREAKING: OpenAI's o3, a new PhD level math agent, found a contradiction in ZFC, none of your theorems are secure, mathemati…
0
6
0
RT @DanieliOren: We plan to spend a few weeks over the summer in Boston (July and August) and are looking for a place to stay (2 adults and…
0
5
0
RT @aarmlovi: Congestion pricing coverage is peak economics Every person is like, "I didn't used to bear the external costs of my behavior…
0
491
0
RT @NoamGidron: We often define the ideological space in terms of the economic and cultural dimensions; how does this resonate with ordinar…
0
28
0
RT @ZoeMcLaren: Intergenerational Correlation in Mortality: The Role of Chronic Conditions #ASSA2025 Dan Zeltzer, Stanford Univ & Tel Aviv…
0
1
0
Presenting Sunday morning at #ASSA2025 on 1) congestion and strategic interactions among hospital departments, and 2) intergenerational mobility in mortality. DM/email me to meet!
0
5
18
A few years ago we shifted all undergrad teaching to R and it works well!
We should care about students’ career outcomes. Teaching them R will be MUCH more helpful than Stata. Not only is R more highly valued, it also makes learning things like Python much easier.
2
0
18
RT @mkoplow: For anyone unsure of how Netanyahu is responding to Trump's election, appointing Yechiel Leiter as the next Israeli ambassador…
0
269
0
This is a nice paper highlighting the sensitivity of LLMs to framing and irrelevant info, casting doubt on supposed "reasoning" capabilities
1/ Can Large Language Models (LLMs) truly reason? Or are they just sophisticated pattern matchers? In our latest preprint, we explore this key question through a large-scale study of both open-source like Llama, Phi, Gemma, and Mistral and leading closed models, including the recent OpenAI GPT-4o and o1-series. Work done with @i_mirzadeh, @KeivanAlizadeh2, Hooman Shahrokhi, Samy Bengio, @OncelTuzel. #LLM #Reasoning #Mathematics #AGI #Research #Apple
2
0
3
RT @JingLiUW: Co-led by Dr. Lawrence Casalino and myself, our new @JAMAHealthForum study is the first to link experimentally measured #altr…
0
7
0
I can relate. Claude Sonnet 3.5 is my go to model these days for both R code and writing tasks
Using o1-preview makes me appreciate Claude Sonnet 3.5. When o1 does something really impressive, I usually try it in other models and am surprised at how often Sonnet pulls it off. Not all the time, o-1 is definitely a better complex reasoner & handles harder problems, but often
0
0
0