Lightnews — Scholar-powered news

alex hayes

@alexpghayes.com

4.6K followers 1.8K following 390 posts

postdoc @ stanford econ + incoming assistant prof @ oregon state statistics. networks, causal inference, contagion, measurement error, #rstats. he/him https://www.alexpghayes.com

www.alexpghayes.com

Posts Media Videos Starter Packs

Pinned

alex hayes @alexpghayes.com · Oct 16

New pre-print!

Keith Levin and I show that the reflection problem is a lot worse than people think it is, and linear-in-means models are degenerate in a very concerning way

arxiv.org/abs/2410.10772

#econsky #statssky

3 14 47

alex hayes @alexpghayes.com · 2d

Just read a referee report clearly generated by AI and I really hope journals start requiring Gen AI disclosures in referee reports

alex hayes @alexpghayes.com · 11d

What's the context for the last one?

1 1

Reposted by alex hayes

sarah @sarahjohnson.io · Aug 21

I was loving Claude Code... until I tried it with #rstats. Constant errors, wouldn't use the tidyverse even when asked, "optimized" functions were slower.

Frustrated, I started a session just to teach R to Claude and summarize what it learned into a CLAUDE.md file gist.github.com/sj-io/3828d6...

Claude R Tidyverse Expert

Claude R Tidyverse Expert. GitHub Gist: instantly share code, notes, and snippets.

gist.github.com

11 25 120

alex hayes @alexpghayes.com · Aug 31

nice guide to making github profiles (especially academic ones) legible to others!

Yani Bellini Saibene @yabellini.bsky.social · Aug 31

🧵 Skytorial 2 – Your profile on @github.com

1/ 🌟 The first thing on GitHub is your profile: your cover letter to the academic and scientific world.
This is how colleagues, collaborators, and potential reviewers of your projects find you.

Let's create it step by step 👇

3 10

alex hayes @alexpghayes.com · Aug 28

Some of Andrew Vickers work on Net Benefit might be relevant, although that's written for a clinical context

e.g., journals.sagepub.com/doi/10.1177/...

Calibration of Risk Prediction Models - Ben Van Calster, Andrew J. Vickers, 2015

Decision-analytic measures to assess clinical utility of prediction models and diagnostic tests incorporate the relative clinical consequences of true and false...

journals.sagepub.com

Reposted by alex hayes

arxiv.stat.ME @arxiv-stat-me.bsky.social · Aug 12

Subhankar Bhadra, Michael Schweinberger
Causal Inference Under Network Interference
https://arxiv.org/abs/2508.06808

2 1

alex hayes @alexpghayes.com · Aug 25

Can you share some pointers to this literature? I'm curious to read more

alex hayes @alexpghayes.com · Aug 22

Once again enjoying these materials. Do you know of any typography resources with Latex/Typst/Quarto example code?

Practical Typography is fascinating but I'm unsure how to implement many of the recommendations

1 2

alex hayes @alexpghayes.com · Aug 13

The caveat being that you always get clusters back so you need ways to distinguish between good and bad clusters

2 7

alex hayes @alexpghayes.com · Aug 13

I think the value of clustering is more often about EDA and discovery rather than confirmatory work or solving a known problem

1 4

Reposted by alex hayes

Richard D. Morey @richarddmorey.bsky.social · Aug 8

Also, we think people should stop using forensic meta-analytic techniques that have not been adequately vetted by experts in statistics. The p curve papers have very little statistical detail, and were published in psych journals. They did not get scrutiny appropriate to their popularity. 15/?

3 12 48

alex hayes @alexpghayes.com · Aug 10

Makes sense! Thanks!

alex hayes @alexpghayes.com · Aug 9

Is there an alternative that does these things?

alex hayes @alexpghayes.com · Aug 9

Is this what ggplot2::geom_dotplot() does?

1 1

alex hayes @alexpghayes.com · Aug 8

Or slightly more specifically: do you use both approaches? Do you mix them? If you use them at different times, when?

alex hayes @alexpghayes.com · Aug 8

I'm curious if you have any thoughts on how this notebook-based approach to data analysis differs/is/isn't compatible with {targets} and build-system style approaches to data analysis

alex hayes @alexpghayes.com · Aug 3

More generally there's a whole industry of accounting for homophily via network embeddings, in which case you can think of the network as a proxy for the latent thing driving homophily

alex hayes @alexpghayes.com · Aug 3

What are the standard CMR reasons?

alex hayes @alexpghayes.com · Aug 3

Not perfect but the meditation data example in my paper on network mediation is essentially proximal inference in a parametric setting

jmlr.org/papers/volum...

1 2

Reposted by alex hayes

Vincent Arel-Bundock @vincentab.bsky.social · Aug 2

This workshop is going to be great fun, and I really look forward to meeting everyone!

It would mean a lot to me if you signed up!

Dariia Mykhailyshyna @dariia.bsky.social · Aug 2

❗️Our next workshop will be on August 14, 6 pm CEST, on marginaleffects package by
@vincentab.bsky.social !
Register or sponsor a student by donating to support Ukraine!
Details: bit.ly/3wBeY4S
Please share! #AcademicSky #EconSky #RStats

2 13 34

alex hayes @alexpghayes.com · Aug 1

I'm presenting in the Business and Economic Statistics Student Paper Awards session:

Peer effects in the linear-in-means model may be inestimable even when identified

Thursday, Aug 7, 8:30 AM - 10:20 AM
Music City Center, Room CC-103C

alex hayes @alexpghayes.com · Aug 1

I'll be #JSM2025 from Sun Aug 3 - Thurs Aug 7

Excited to chat with folks! These days I'm thinking about:

- peer effects
- measurement error in networks
- stochastic blockmodels and friends
- causal inference on networks
- semi-parametric mediation & transportability
- #rstats

1 3 11

alex hayes @alexpghayes.com · Jul 29

@allendowney.bsky.social this seems up your alley given your interest in Simpson's paradox showing up in data

1 2

Reposted by alex hayes

Emily Riederer @emilyriederer.bsky.social · Jul 27

Technical writing is hard bcs "writing is thinking" but we often should tell our story not in the order we worked. Solution? I wrote a quick post on how @quarto.org 's embed shortcodes can reframe technical writing as reproducible evidence curation

www.emilyriederer.com/post/quarto-...

🧵 (1/n)

How Quarto embed fixes data science storytelling | Emily Riederer

Literate programming excels at capturing our stream of conscience. Our stream of conscience does not excel at explaining the impact of our work. Notebooks enable some of data scientists’ worst tendenc...

www.emilyriederer.com

4 24 110

alex hayes @alexpghayes.com · Jul 26

Yeah, building a thing is not enough to get people to use the thing