Yiqing Xu
@yiqingxu.bsky.social
Stanford researcher. Work on applied stats, AI (?), China.
🧵 Introducing ReproAI, an author-facing agentic AI plugin for replication packages: reproai.org We built this tool to help authors prepare replication packages before submission, w/ @leoyangyang.bsky.social.
ReproAI: the plugin for reproducible replication packages
An author-facing AI plugin that helps you prepare a cleaner, reproducible replication package before submission.
reproai.org
I intentionally chose a red day to post this, to avoid giving the impression that I am encouraging people to buy stocks. yiqingxu.org/guide/person...
My very personal reflection on the use of AI in social science, perhaps the first of many. yiqingxu.substack.com/p/ai-disarray
AI Disarray
I have felt a little disarrayed recently, and I attribute it to AI.
yiqingxu.substack.com
1/ Happy to share that "Factorial Difference-in-Differences" (FDID), with Anqi Zhao and @pengding00 , is out in JASA - ACS. doi.org/10.1080/0162... It has been a truly thrilling experience working with Anqi and Peng.
Factorial Difference-in-Differences
We formulate factorial difference-in-differences (FDID), a research design that extends canonical difference-in-differences (DID) to settings in which an event affects all units. In many panel data...
doi.org
Trying to understand why synthetic control is for a while. Comments are welcome! yiqingxu.substack.com/p/what-is-sy...
What is synthetic control?
What the literature has converged to in the past 20 years.
yiqingxu.substack.com
On panelView, and why we should always look at our data first.. yiqingxu.substack.com/p/plot-your-...
Please look at your (panel) data
Many of the most egregious issues can be avoided by simply looking at the data.
yiqingxu.substack.com
I started a substack... to jot down my thoughts and learning notes on causal inference, panel data, and AI --- and maybe a few other random events along the way.
Notes on Causal Inference, Panel Data, and AI
I decided to start writing notes on causal inference, panel data, and possibly research with and on AI.
yiqingxu.substack.com
Thanks, Anton, for the adverticement. We believe this is just one small attempt of combinine the (minimally) structural apporach with DML -- randomization is doing a lot of work here, too.
@yiqingxu.bsky.social, Jens Hainmueller and Avi Acharya have a cool paper to get political scientists to just fit structural choice models to their choice survey experiments! yiqingxu.org/papers/2026_...
1/ Happy to release StatsClaw — an open-source multi-agent workflow for building statistical software with AI. w/ @tianzhuqin.bsky.social Site: statsclaw.ai Paper: bit.ly/statsclaw GitHub: github.com/statsclaw/
StatsClaw
Collaborative AI for Stats Packages.
statsclaw.ai
1/2 Important caveat: Computational reproducibility does not imply credibility, especially w.r.t. to research design. In the scientific literature, *reproducibility* means reproducing results using authors’ code and data, whereas *replication* goes beyond what was originally done or provided.
1/🧵 A major update to our paper: "Scaling Reproducibility" w/ Leo Yang [Cross-posted from X] We move beyond reanalyzing a single design to (almost) full-paper replication! Paper: bit.ly/repro-ai
1/ Sorry for double-posting from X. Sharing a new working paper for the Year of the Horce 🐎: "An AI-assisted workflow that scales reproducibility in empirical research" (bit.ly/repro-ai) w/ Leo Yang Yang
A blog post giving a more thorough take on survey experiments and the credibility revolution: cyrussamii.com?p=4168
Survey experiments and the credibility revolution – Cyrus Samii
cyrussamii.com
This is an impressive project. My reaction to what it shows though is that survey experiments have gotten out of hand in polisci. I will blog more on this, but I do not think survey experiments are emblematic of the credibility revolution. Some are already interpreting as such, which is a problem.
Preliminary. Comments are welcome!
New paper! @william-dinneen.bsky.social @guygrossman.bsky.social Yiqing Xu and I use GPT to code 91k articles from 174 polisci journals (2003–2023)and track research designs, transparency practices, and citations. How has the credibility revolution reshaped the discipline? doi.org/10.31235/osf... 🧵
Thrilled to share that **fect** has won the 2025 Best Statistical Software Award from the Society of Political Methodology. We're honored! polmeth.org/statistical-... To celebrate, we've just released fect v2.0.5 on CRAN & Github 🎉
Honored by this recognition!
We are pleased to announce the 2025 Editors’ Choice Award for the paper “How Much Should We Trust Instrumental Variable Estimates in Political Science? Practical Advice Based on 67 Replicated Studies” by @apoorvalal.com, @maclockhart.bsky.social, @yiqingxu.bsky.social, and @garyzu.bsky.social.
Glad this paper is finally out in @apsrjournal.bsky.social, five years after we began the project. Hope it proves useful to researchers. www.cambridge.org/core/journal... Software support: yiqingxu.org/packages/fec...
Causal Panel Analysis under Parallel Trends: Lessons from a Large Reanalysis Study | American Political Science Review | Cambridge Core
Causal Panel Analysis under Parallel Trends: Lessons from a Large Reanalysis Study
cambridge.org
As a Chinese immigrant living in the U.S., I have witnessed declines in optimism purely due to the drastic erosion of existing institutions—first around 2015 in my home country, and now, in 2025, in the United States. It's devastating.
Draft “A Practical Guide to Estimating Conditional Marginal Effects: Modern Approaches” is on arXiv: arxiv.org/pdf/2504.01355 w/ two amazing grad students, Jiehan_Liu & Ziyi Liu 🧵
I woke up at 5:30 a.m. today, and the first thing on my mind was Stefan Zweig’s The World of Yesterday: Memoirs of a European, a book I loved as a child. His despair stuck with me and feels especially relevant now.
"The bottom 80% of earners spent 25% more than they did four years earlier, barely outpacing price increases of 21% over that period. The top 10% spent 58% more." www.wsj.com/economy/cons...
We streamlined six new DID-like estimators and created this tutorial for implementation in R. yiqingxu.org/packages/fec... Hope you no longer need to spend months figuring out what these estimators are and how to use them.
Looks like everyone is here 😆 1/ Recently, Professor Uri Simonsohn critiqued Hainmueller, Mummolo & Xu (2019), arguing that the proposed methods fail to recover the conditional marginal effect (CME): datacolada.org/121 We appreciate the critique and offer this response: arxiv.org/pdf/2502.05717 🧵