I just finished @gelliottmorris.com's delightful Strength in Numbers and I'm excited to read more about opinion research and polling Let me know if you have recommendations for followup reading!

which is the worse statistical sin, multiplying regression coefficients (mediation) or dividing regression coefficients (instrumental variables)

Keith and I strengthened our lower bounds for randomized experiments in the linear-in-means model to a general minimax result Updated results available at arxiv.org/abs/2410.10772

A new release of the mgcv #RStats 📦 is out on CRAN and Simon Wood (U Edinburgh) has added some significant new features despite the small bump in version number: 🌟 scasm() for estimating GAMs with shape constrained smooths. Can be used with any family & smoothness selection is via the EFS method

What are your favorite books/articles/resources about graphic design for academic posters and presentations?

Now I'm also looking for a research software engineer to implement a pile of research results to R packages loo, posterior, bayesplot, projpred, priorsense, brms or/and Python packages ArviZ, Bambi and Kulprit. Apply by email with no specific deadline (see contact info at users.aalto.fi/~ave/)

this thread nicely describes the epistemology of a decent portion of modern meta-science, and i think also explains why it lost steam

I missed that an expanded version of Philip Stark's "Pay No Attention to the Model Behind the Curtain" was published a few years back Needless to say it's very good and worth a read link.springer.com/10.1007/s000...

This may just be the best CS paper I’ve read this year. Just read the abstract and first para of the intro! The rest of the intro is really wild too, but very very good: dl.acm.org/doi/pdf/10.1...

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

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...

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

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

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/?

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!

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

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)

Do folks know about A/B testing datasets where information about guardrail metrics is released alongside outcomes? #rstats #datascience #kaggle

boosting for the weekend crowd

Any tips and tricks for @typst.app <-> LaTeX translations? Pandoc is struggling with custom macros for notation and I'm weary of LLMs for technical writing (plus context window issues)

~~ making sense of academic statistics ~~ i wrote about the confusing relationship between statistics and data analysis, and also about how statistics relates to science #statistics #rstats #datascience www.alexpghayes.com/post/making-...

I've invested a lot into LaTeX in the last few years and didn't feel a strong desire to pivot, but I just tried @typst.app and I'm an immediate convert The compilation speed is so nice and it's so much easier to understand

The more I do multiverse data analysis, the more I find that conclusions vary dramatically with model specification. This makes me quite hesitant to trust empirical results using network models

Is there a go-to implementation of sparse SVD in #python? scipy.sparse.linalg.svds is much slower than the #rstats counterparts in {irlba} and {RSpectra} and I'm wondering if there are better options (I'm using the same data and CSR representations in both cases)

Hamilton, Tawn, Firth (2023) "The many routes to the ubiquitous Bradley-Terry model": arxiv.org/abs/2312.13619 Very cool! I've considered writing notes on how BT models can be motivated in myriad ways, with diff motivations leading to diff generalizations. This looks like a fantastic crack at that.

Evergreen

I'm looking for methods to visualize signed graphs with ~100 nodes and lots of negative edges So far the best I've come across is in {signnet} but I'm hoping I can find some other options github.com/schochastics...

link 📈🤖 Interpretable sensitivity analysis for the Baron-Kenny approach to mediation with unmeasured confounding (Zhang, Ding) Mediation analysis assesses the extent to which the exposure affects the outcome indirectly through a mediator and the extent to which it operates directly through other

Enjoying this conceptual paper on when/why predictive models can achieve causal goals

#dataviz #rstats Least squares by Springs: Here's a remarkable diagram, from 1932, illustrating a mechanical device to fit an exponential relation, y = A e^{-k x}, to data, here y= the milk yield of cows over x = 11 months. From: Gaines & Palfrey www.jstor.org/stable/1658643 ⬇️

So causalfusion.net and dagitty.net are great to check if controls C are sufficient to estimate the total effect of A on Y without needing to check backdoor paths yourself These tools can also algorithmically check if direct and indirect effects mediated by some M are identified!

Measuring the dynamical evolution of the United States lobbying network arxiv.org/abs/2503.11745

If you want to browse academic work without accidentally doom scrolling I cannot recommend this feed enough

Have anytime-valid methods made their way into empirical metascience yet? I'm thinking of them as a way to look for sampling to a foregone conclusion in published work

Learn how to use grouped tabsets and the Tabby extension in #Quarto to write documentation for both #RStats & #Python users! Read about grouped tabsets and the Tabby extension here: posit.co/blog/creatin...

Sunset across Lake Mendota in Madison

I want to understand recent developments in recsys land It seems like contrastive learning is the big thing now? Does this reflect an evolving class of models or just a slight evolution of negative sampling to address computational burden of large data?

Please apply, this project involves quite a bit of fun work on combining data from observational studies with randomized trials to estimate heterogeneous treatment effects!

HelioCampus is hiring! We have openings for data scientists and data analysts. www.heliocampus.com/company/care... I'm the hiring manager for the data scientist role, and we're specifically looking for someone with experience working with student data in the higher ed industry.

Does anyone have a favorite field experiment (or A/B test) with public data where there is outcome data over time, ideally something like daily data or even timestamped events? This is a common format in industry, but few public data sets look like this.

Enjoying Kino Zhao's recent HDSR paper on a way to think about statistical assumptions I'm also pleased that a stats venue is publishing this kind of work hdsr.mitpress.mit.edu/pub/qasl4fza...

My work on network regression and mediation in latent space models is now published at JMLR! jmlr.org/papers/v26/2...

If I want an AI coding assistant to consider my project/repository to help debug something, what are my options? Does Copilot do this?

PLEASE RT :) Submit to the annual Network Science and Economics Conference, the largest US event for network theory and empirical research in and adjacent to economics Deadline Jan 30 (soon!) Conference Apr 11-13 at Stanford

I cannot recommend interning with Posit, and Max and the tidymodels folks in particular, highly enough! Apply!

latexdiff-vc is a magic way to see what changes coauthors made to a LaTeX manuscript latexdiff-vc --git --pdf -r {my_last_commit} -r HEAD paper.tex www.mankier.com/1/latexdiff-vc

Heads up if you're into data and passionate about the science of science! @crossref.bsky.social seeks TWO Data Scientists for this new team. You'll collaborate to glean insights from and detect gaps in scholarly metadata and help build the research nexus. Join us! www.crossref.org/jobs/2025-01...

@noupside.bsky.social if I want to learn about effective strategies for increasing vaccination rates, what should I read and who should I follow on here?

I finally found a website that regularizes user ratings towards the mean and it's like I've entered the promised land

I've recently been enjoying some excellent books on evidence-based teaching by @jamesmlang.bsky.social and @retrievelearn.bsky.social Are there similar titles targeted at students about how to be evidence-based learners? #highered #education