Josh McCrain
@joshmccrain.bsky.social
Political scientist at the University of Utah. Public policy, political institutions, media and politics
An astounding piece of work, full thread here: x.com/xuyiqing/sta...
Yiqing Xu on X: "1/ ๐ Our gift for the Year of the Horse: An AI-assisted workflow that scales reproducibility in empirical research. w/ @YangYang_Leo Paper: https://t.co/fslLN0zTQO https://t.co/SpX8SzTyNP" / X
1/ ๐ Our gift for the Year of the Horse: An AI-assisted workflow that scales reproducibility in empirical research. w/ @YangYang_Leo Paper: https://t.co/fslLN0zTQO https://t.co/SpX8SzTyNP
x.com
February 19, 2026 at 5:03 PM
An astounding piece of work, full thread here: x.com/xuyiqing/sta...
I have not seen this mentioned here, but this is a great example of how the research world has dramatically changed underneath of us and if you're not keeping up you're going to be left behind quickly:
yiqingxu.org/papers/2026_...
yiqingxu.org/papers/2026_...
February 19, 2026 at 5:03 PM
I have not seen this mentioned here, but this is a great example of how the research world has dramatically changed underneath of us and if you're not keeping up you're going to be left behind quickly:
yiqingxu.org/papers/2026_...
yiqingxu.org/papers/2026_...
Reposted by Josh McCrain
@mattgrossmann.bsky.social, @joshmccrain.bsky.social, and I are putting together a panel at PolMeth in July on using CongressData, CSPP, and SenateData. We are aiming to have up to 5 people (preference to junior faculty using the data or interested in doing so). Please reach out to me if interested!
February 18, 2026 at 6:35 PM
@mattgrossmann.bsky.social, @joshmccrain.bsky.social, and I are putting together a panel at PolMeth in July on using CongressData, CSPP, and SenateData. We are aiming to have up to 5 people (preference to junior faculty using the data or interested in doing so). Please reach out to me if interested!
Hard to convince my daughter not to draw on herself when we both have a bunch of tattoos
February 16, 2026 at 6:15 PM
Hard to convince my daughter not to draw on herself when we both have a bunch of tattoos
Yes I agree with this
February 12, 2026 at 9:05 PM
Yes I agree with this
Scott Cunningham has a great aubstack
February 12, 2026 at 8:55 PM
Scott Cunningham has a great aubstack
Substantively I think the difference is not generated through the programming part per se but through smart use of agents
February 12, 2026 at 5:44 PM
Substantively I think the difference is not generated through the programming part per se but through smart use of agents
Also this is especially true in some other social science disciplines that don't even have appendices for papers
February 12, 2026 at 5:43 PM
Also this is especially true in some other social science disciplines that don't even have appendices for papers
Yes I think this is right. But many many papers are written with no intention of presenting
February 12, 2026 at 5:40 PM
Yes I think this is right. But many many papers are written with no intention of presenting
Would you like to elaborate
February 12, 2026 at 5:27 PM
Would you like to elaborate
If you're just using it for coding that is a very narrow scope
February 12, 2026 at 5:25 PM
If you're just using it for coding that is a very narrow scope
One thing that it is going to amplify is resource divides. The true advantage is having it spin off lots of agents, which gets expensive really fast
February 12, 2026 at 5:25 PM
One thing that it is going to amplify is resource divides. The true advantage is having it spin off lots of agents, which gets expensive really fast
I mean you absolutely have to now. I just can't even fathom whst things will look like in even a year
February 12, 2026 at 5:23 PM
I mean you absolutely have to now. I just can't even fathom whst things will look like in even a year
The standards they're going to be held to; also what happens when the rate of journal submissions increases by an order of magnitude
February 12, 2026 at 5:22 PM
The standards they're going to be held to; also what happens when the rate of journal submissions increases by an order of magnitude
Yes definitely, lots of ppl more qualified than me but I would be down
February 12, 2026 at 5:21 PM
Yes definitely, lots of ppl more qualified than me but I would be down
If your thinking is that Claude code is just like another LLM in terms of work flow impact you are deeply wrong
February 12, 2026 at 5:19 PM
If your thinking is that Claude code is just like another LLM in terms of work flow impact you are deeply wrong
If you've not used Claude code you really have a hard time understanding how quickly things are going to change. I certainly did. Projects that took me 6 months of coding can be done in minutes
February 12, 2026 at 5:16 PM
If you've not used Claude code you really have a hard time understanding how quickly things are going to change. I certainly did. Projects that took me 6 months of coding can be done in minutes
So little discussion of Claude code on here. We are experiencing an unprecedented shock to how empirical research is done and it's going to primarily affect grad students
February 12, 2026 at 5:15 PM
So little discussion of Claude code on here. We are experiencing an unprecedented shock to how empirical research is done and it's going to primarily affect grad students
Anyone else stop reading at "...we used large language models..." ?
What if my abstract said: "We used a bullshit machine to do some science and here's what we found..."
โ๏ธ
youtu.be/uF9kfYaXgoc?...
What if my abstract said: "We used a bullshit machine to do some science and here's what we found..."
โ๏ธ
youtu.be/uF9kfYaXgoc?...
February 11, 2026 at 9:05 PM
this is extremely interesting
We coded our ~100k articles using LLMs. Should you believe them? To answer this, we benchmarked 4 human RAs against 3 LLMs on their ability to recover ground truth article data. Details in the paper and appendices, but the LLMs did well and handily beat the highly trained humans.
February 11, 2026 at 9:04 PM
this is extremely interesting