Similar acks seem to work in teaching too? Like just casually and nonjudgmentally saying "you've communicated [X] to me in a nonverbal medium" suddenly allows whatever [X] is to be talked about

New advising trick I've been trying: when a student looks skeptical about something, but doesn't *say* they are, I say "the look on your face says you don't believe me"

I think explicitly acknowledging the nonverbal cue gives permission to voice what they're unsure about so we can talk about it

the first rule of dealing with assholes is don't accept their stupid premises

Stratified swarm dotplot with exact x positions and two groups stacked on top of each other. Low resolution of stratification (more regular row-like layout).

Stratified beeswarm with exact x positions and two groups stacked on top of each other. Low resolution of stratification (more regular row-like layout).

Stratified swarm dotplot with exact x positions and two groups stacked on top of each other. High resolution of stratification (less regular rows).

Stratified beeswarm with exact x positions and two groups stacked on top of each other. High resolution of stratification (less regular rows).

Mostly just me trying to make beeswarms / dotplots with exact x positions whose layouts I don't hate. Side benefit is they also allow stacking.

(+ some other minor improvements, including better automatic binwidth and faster layouts)

why do I have to help write VIS papers when I have all this cool new dotplot code to push

(tbf I am very excited about our papers this year)

I think my problem with a debate club is when it suggests it is virtuous to be able to take an arbitrary side of an issue and win. That makes the exercise about learning how to distort facts to make a case not how to judge what is true.

This is what happens when people who think a debate club is a good idea write law

X is boiling journalists’ brains

A reminder that if you’re looking for dataviz content on Bluesky, there’s a lot here 👇📊

I'll admit I'm baffled by this one - I agree Atlanta is Nashville (and vice versa) but many of the others on this list are neither

the worst

Homepage of the R Mailing List Archives showing a search bar, three popular lists (R-help with 398,500 messages, R-devel with 63,425, and R-package-devel with 12,125), and a full directory of core lists and special interest groups.

29 years of #rstats community knowledge was sitting in hard-to-search pipermail archives. So I built a more modern home for it.

Introducing the R Mailing List Archives: 631,000+ messages from 32 lists, fully searchable and available as open data.

r-mailing-lists.thecoatlessprofessor.com

Simulation-based validation of Bayes factor computation We propose and evaluate two methods that validate the computation of Bayes factors: one based on an improved variant of simulation-based calibration checking (SBC) and one based on calibration metrics...

Preprint alert: Simulation-based validation of Bayes Factor computation with @paulbuerkner.com and S. Stroppel. We bring lessons learned in SBC to validation of BFs. arxiv.org/abs/2508.11814 The idea is simple: simulate data from the models, fit and see if the inferences are calibrated. 1/10

I made a tiny tool for quickly sharing small datasets (< ~1000 rows) without uploading any data to a server.

🔗 ziptbl.com

It compresses the data into the link itself, so there’s no account, hosting, or storage layer involved.

Here's Florence Nightingale's famous 📊 data:
ziptbl.com#d=eNpdlE-LGz...

Yeah that tracks. I think in North America skinny jeans started falling off 5-10 years ago and probably hit peak unpopularity 2-4 years ago

Haha same. My trick is (1) I don't care what other people think and (2) I live in a gay neighborhood so no one stopped wearing tight pants

Lol I think they've been out since like 2016

I'm anticipating the return of skinny jeans, it's about that time in the cycle

ggtypst Typst-powered text and math rendering for ggplot2.

There is also the recently-released ggtypst (haven't tried it myself but looks promising): yousa-mirage.github.io/ggtypst/

"No shit Sherlock"

They all read Snow Crash and watched Hackers and thought they were Zerocool but it turns out they're just The Plague

imagine reading*, say, Crime and Punishment and thinking introspection is a modern conception

________
*obviously it's a stretch to imagine any of these dudes reading but bear with me for the sake of the point...

he is white and old and dead and a man, so

this jabroni: "400 years ago it would have never occurred to anybody to be introspective"

sure there definitely isn't some famous dead Greek dude from thousands of years ago who said "the unexamined life is not worth living", nope definitely not

This take resonates.

Like the author below, I kept trying to figure out why I thought OBAA was just "fine". My working theory is that most every attempt at satire in it didn't understand what it was satirizing, so what should have been funny or absurd just sort of... hung there on screen.

Missing the Pynchonian Element A narrow reflection on "One Battle After Another"

Kotsko's is the best take on OBAA adamkotsko.substack.com/p/missing-th...

Yeah that was my impression when I read it. A version that can set multivariate priors would be great.

PriorWeaver: Prior Elicitation via Iterative Dataset Construction In Bayesian analysis, prior elicitation, or the process of facilitating the expression of one's beliefs to inform statistical modeling, is an essential yet challenging step. Analysts often have belief...

There is work in HCI/vis on interactively eliciting priors in data space, like this from @yuweixiao.bsky.social @shuaihci.bsky.social @oulasvirta.bsky.social @eunicemjun.bsky.social

Curious if you had particular cases in mind?

I think it is sometimes a stand-in for "any vector of length 0" which is not always discouraged by R's coercion rules but is perhaps different from a more traditional notion of NULL as something like a "none" value in an option type...

does typing c(3, 3, 15, 15, 20, 20, 23, 23, 20, 20) count as modeling