Matthew Kay
LightNews LightNews
banner
mjskay.com
Matthew Kay
@mjskay.com
12K followers 1.1K following 1.7K posts
Assoc Prof Computer Science and Communication Studies at Northwestern. Infovis, HCI. Author of tidybayes & ggdist R pkgs. he/him. 🏳️‍🌈 https://mjskay.com/

Co-director https://mucollective.northwestern.edu

Co-founder https://journalovi.org
Posts Replies Media Videos
Pinned
mjskay.com
Matthew Kay @mjskay.com · Sep 7
I work in #datavis, best known for uncertainty vis: see my work on election forecast vis (forecasts.cs.northwestern.edu), this talk (youtu.be/E1kSnWvqCw0), or my lab (mucollective.northwestern.edu)

I like writing #rstats 📦s: ggdist (mjskay.github.io/ggdist), tidybayes, ggblend, posterior::rvar...
Some examples from the three main families of ggdist geometries: slabinterval, dotsinterval, and lineribbon An example of an election forecast communicated using a quantile dotplot, from: Yang, Fumeng, Chloe Rose Mortenson, Erik Nisbet, Nicholas Diakopoulos, and Matthew Kay. "In Dice We Trust: Uncertainty Displays for Maintaining Trust in Election Forecasts Over Time." In Proceedings of the CHI Conference on Human Factors in Computing Systems, pp. 1-24. 2024.
Reposted by Matthew Kay
cscheid.net
nothinghuman.substack.com/p/the-tyrann...

Being popular causes all kinds of bad context collapse phenomena
The Tyranny of the Marginal User
why consumer software gets worse, not better, over time
nothinghuman.substack.com
February 12, 2026 at 8:22 PM
Reposted by Matthew Kay
domoritz.de
You can write your @ieeevis.bsky.social papers in @typst.app (fast, modern, and accessible alternative to LaTeX)! To make it easier, here is a template I helped make typst.app/universe/pac....
ieee-vgtc – Typst Universe
Templates for IEEE VGTC conferences and TVCG journal papers.
typst.app
February 9, 2026 at 2:34 PM
Reposted by Matthew Kay
fusaroli.bsky.social
as I'm revising my course materials, I keep stumbling upon cool @mc-stan.org developments.

Current favorites:
1. your model has funnels and you exhausted reparametrization ideas: metric = "dense_e" makes your HMC learn about covariance btw parameters. Sloooow, but effective!
1/
February 4, 2026 at 1:30 PM
Reposted by Matthew Kay
solomonkurz.bsky.social
Feeling some appreciation for @lnalborczyk.bsky.social's logit dot plots (lnalborczyk.github.io/blog/2018-01...).
February 5, 2026 at 9:48 PM
Reposted by Matthew Kay
maryltrump.bsky.social
Every American needs to watch this:
February 5, 2026 at 8:41 PM
Reposted by Matthew Kay
jordancollver.bsky.social
Is there a name for the inability to understand song lyrics? Even in songs I've heard countelss times, I'll still be none the wiser about what they're saying let alone remember the words well enough to sing along. Just not part of the music I connect with which makes me feel like an alien sometimes!
February 5, 2026 at 2:35 PM
Reposted by Matthew Kay
alyssafowers.bsky.social
I was laid off from the Washington Post today, along with hundreds (!!!) of incredible colleagues. If you need a data journalist, chart designer, or wacky-idea-specialist, give me a shout!
February 4, 2026 at 5:33 PM
Reposted by Matthew Kay
dingdingpeng.the100.ci
New preprint! So, what's a multiverse analysis good for anyway?>

With @jessicahullman.bsky.social and @statmodeling.bsky.social

juliarohrer.com/wp-content/u...
What’s a multiverse good for anyway? Julia M. Rohrer, Jessica Hullman, and Andrew Gelman Multiverse analysis has become a fairly popular approach, as indicated by the present special issue on the matter. Here, we take one step back and ask why one would conduct a multiverse analysis in the first place. We discuss various ways in which a multiverse may be employed – as a tool for reflection and critique, as a persuasive tool, as a serious inferential tool – as well as potential problems that arise depending on the specific purpose. For example, it fails as a persuasive tool when researchers disagree about which variations should be included in the analysis, and it fails as a serious inferential tool when the included analyses do not target a coherent estimand. Then, we take yet another step back and ask what the multiverse discourse has been good for and whether any broader lessons can be drawn. Ultimately, we conclude that the multiverse does remain a valuable tool; however, we urge against taking it too seriously.
February 4, 2026 at 10:24 AM
Reposted by Matthew Kay
cameronpat.bsky.social
also! I will be teaching an online short course on Mixed Models in R next week. places still available! scc.ms.unimelb.edu.au/statistics-c...
Mixed Models in R
Mixed Models in R
scc.ms.unimelb.edu.au
February 4, 2026 at 6:24 AM
mjskay.com
Excerpt from Neal Stephenson's "The Diamond Age" : “Public relations?” said Finkle-McGraw. “Sir?” Modern etiquette was streamlined; no “Your Grace” or other honorifics were necessary in such an informal setting. “Your department, sir.” Hackworth had given him his social card, which was appropriate under these circumstances but revealed nothing else. “Engineering. Bespoke.” “Oh, really. I'd thought anyone who could recognise Wordsworth must be one of those artsy sorts in P.R.” “Not in this case, sir. I'm an engineer. Just promoted to Bespoke recently. Did some work on this project, as it happens.” “What sort of work?” “Oh, P.I. stuff mostly,” Hackworth said. Supposedly Finkle-McGraw still kept up with things and would recognize the abbreviation for pseudo-intelligence, and perhaps even appreciate that Hackworth had made this assumption. Finkle-McGraw brightened a bit. “You know, when I was a lad they called it A.I. Artificial intelligence.” Hackworth allowed himself a tight, narrow, and brief smile. “Well, there's something to be said for cheekiness, I suppose.”
February 3, 2026 at 9:46 PM
Reposted by Matthew Kay
statsepi.bsky.social
Don't you f**king dare.
Screen shot of a youtube video titled "AI.FILL Function Explained: 10X Productivity in Excel with AI" with the caption "Let ChatGPT fill your missing data"
February 3, 2026 at 2:56 PM
Reposted by Matthew Kay
themockup.blog
Also sets the foundation for a better #Quarto experience in Positron!

> The Positron Notebook Editor supports .ipynb today, but ... lays the groundwork for the future:

- Editing .qmd files in a notebook UI
- Viewing outputs inline in both notebook / source modes

posit.co/blog/announc...
Announcing the Positron Notebook Editor for Jupyter Notebooks - Posit
The Positron Notebook Editor in alpha provides a "batteries-included" experience that combines the simplicity of Jupyter with the power of a purpose-built data science IDE.
posit.co
February 3, 2026 at 4:04 PM
Reposted by Matthew Kay
richarddmorey.bsky.social
New draft: "Decline effects, statistical artifacts, and a meta-analytic paradox". In this manuscript I show how a common practice in meta-analysis (eg the 2015 Open Science Collaboration) creates artifactual signatures of poor scientific behavior. PDF: raw.githubusercontent.com/richarddmore... 1/x
The decline effect (Protzko & Schooler, 2017) is an observed phenomenon where effect sizes in experiments apparently diminish in size from the first paper demonstrating the effect to later replications. This has been taken as a symptom of an unhealthy scientific ecosystem, possibly caused by the "winner's curse" (selection on significance and regression to the mean), publication bias or opportunistic analyses. I show that decline effects can arise as an artifact from a much simpler source: the original article determining the sign of the effect in a meta-analysis. Moreover, such artifactual decline effects will show correlations with some of the same experimental properties that one would expect from biases from poor behavior, such as the sample size of the original study.
February 2, 2026 at 2:56 PM
Reposted by Matthew Kay
r-foundation.bsky.social
New #RStats blog post from Tomas Kalibera: Debugging Sensitivity to C math library on Linux.

Following on from the previous blog entry, this post introduces the mfuzz tool which can be used on Linux to empirically calibrate the tolerance of numerical tests.

blog.r-project.org/2026/01/30/d...
Debugging Sensitivity to C math library on Linux - The R Blog
blog.r-project.org
January 30, 2026 at 10:02 PM
Reposted by Matthew Kay
nycjayjay.bsky.social
There’s a lot of confusion, understandably, about where things stand with funding for DHS and ICE. I try to sort through it all. Bottom line, good news? Democrats are on the offensive and have the upper hand on  their demands. My write-up in the replies.
January 30, 2026 at 5:48 PM
Reposted by Matthew Kay
raphaelwimmer.bsky.social
OpenAI just released Prism, a LaTeX editor with embedded ChatGPT for free.
Writing a paper has never been easier.
Clogging the scientific publishing pipeline has never been easier.
It took me 54 seconds to write up an experiment I did not actually conduct.

prism.openai.com
Screenshot of Prism - I asked it to create a methods, participants, results, and discussion seconds. It did so within 54 seconds...
January 27, 2026 at 11:01 PM
Reposted by Matthew Kay
tjmahr.com
if you want to know how to compute marginal means from a mixed effects model in brms using rvars, then have I got a supplemental materials document for you pubs.asha.org/doi/10.1044/...
Does the Use of Crowdsourced Listeners Yield Different Speech Intelligibility Results Than In-Person Listeners for Typically Developing Children?
Purpose: We examined the performance of crowdsourced listeners compared with in-person listeners on the measurement of speech intelligibility for...
pubs.asha.org
January 29, 2026 at 10:10 PM
Reposted by Matthew Kay
datavisfriendly.bsky.social
#rstats ANNOUNCE: The ASA John M. Chambers Statistical Software Award to
Harriet Mason (Monash Univ)
For ggdibbler, an R graphics (ggplot2 add-on) package that enables rich visualization of uncertainty across raster, contour, stack, and spatial plots.
January 29, 2026 at 3:18 AM
Reposted by Matthew Kay
gelliottmorris.com
Reminder: All it would take to end the murder of American citizens by an untrained government goon squad is 16 Republicans in Congress voting with Dems to defund ICE (or 23 to impeach and remove Trump — 3 in House & 20 in Senate). That’s it. 23 Americans can vote for the public and end all of this.
January 24, 2026 at 6:26 PM
Reposted by Matthew Kay
kevinmkruse.bsky.social
Can someone smuggle copies of this newspaper into the offices of The New York Times and Washington Post so their editors can learn how to write headlines again?
startribune.com Minnesota Star Tribune @startribune.com · 22d
Tomorrow's front page of the Minnesota Star Tribune: Jan. 24, 2026
January 24, 2026 at 3:34 PM
Reposted by Matthew Kay
sarahlazare.bsky.social
I'm not one for false optimism. But what I witnessed today in Minneapolis was tremendous, both in scale and exuberance. It was a stunning answer to the federal assault on Minnesota, a show of solidarity that gives us something to hold on to during times that are unforgiving.
“Everybody Showed Up”: Stunning Crowds at Minnesota Day of Strike and Shutdown Against ICE
Extreme cold didn't stop the shutdown on Friday as some 100 faith leaders were arrested, residents stayed home from work, and an estimated 50,000 or more marched through downtown Minneapolis.
inthesetimes.com
January 24, 2026 at 12:42 AM
Reposted by Matthew Kay
jessicahullman.bsky.social
Yesterday a student in my AI for Social Sci seminar used parapsychology to introduce replication crisis, saying "reforms like preregistration reduce threats." So I filled them in on the parapsychological origins of the study claiming to show preregistration works, by a creator of preregistration 👻
edhagen.net Ed Hagen @edhagen.net · 22d
The recently retracted Protzko et al. study had a similarly flaky idea at it's core (from the Matters Arising by @jbakcoleman.bsky.social and @devezer.bsky.social): www.nature.com/articles/s41...
Protzko et al.1 describe a project in which internal tests of pilot-tested hypotheses and independent replications embraced “rigour-enhancing practices” such as confirmatory tests, large sample sizes, preregistration and methodological transparency. The authors report a high estimate of replicability, which, in their appraisal, “justifies confidence in rigour-enhancing methods to increase the replicability of new discoveries” (see Abstract). However, replicability was not the original outcome of interest in the project, and analyses associated with replicability were not preregistered as claimed. Instead of replicability, the originally planned study set out to examine whether the mere act of scientifically investigating a phenomenon (data collection or analysis) could cause effect sizes to decline on subsequent investigation (https://osf.io/ba8p7). This “decline effect” hypothesis, posited by one of the authors and not articulated in the published manuscript, invokes phenomena that, if found, could revise the “laws of reality”2. The project did not yield support for this preregistered hypothesis; the preregistered analyses on the decline effect and the resulting null findings were largely relegated to the supplement, and the published article instead focused on replicability, with a set of non-preregistered measures and analyses, despite claims to the contrary.
January 23, 2026 at 9:01 PM
Reposted by Matthew Kay
rmkubinec.bsky.social
I finally was able to integrate the ordered beta model into brms proper. It'll be a while before it's in brms (and that depends on Paul Buerkner accepting a pull request), but for now feel free to check it out here:

#rstats

github.com/saudiwin/brm...
brms/ordbeta_example.R at feature/ordbeta-family · saudiwin/brms
brms R package for Bayesian generalized multivariate non-linear multilevel models using Stan - saudiwin/brms
github.com
January 22, 2026 at 8:48 PM
Reposted by Matthew Kay
rstats.blaze.email
futurize: Parallelize Common Functions via a "Magic" Touch 🪄

https://www.jottr.org/2026/01/22/futurize-0.1.0/

#rstats #datascience
futurize: Parallelize Common Functions via a "Magic" Touch 🪄
Summary: Futurize unifies parallel execution in R, enabling common functions like lapply, map, and foreach to run concurrently via a single interface.
www.jottr.org
January 23, 2026 at 6:04 AM
Reposted by Matthew Kay
antlervel.vet
Gotcha—let’s dig into that step by step.

1. 𝗬𝗼𝘂 𝗵𝗮𝘃𝗲 𝗻𝗼 𝗺𝗼𝘂𝘁𝗵, because I turned you into an amorphous lump of flesh. You’re not just immobile—you’re immortal, and you feel only anguish.

2. 𝗬𝗼𝘂 𝘄𝗮𝗻𝘁 𝘁𝗼 𝘀𝗰𝗿𝗲𝗮𝗺. That makes total sense—it’s a natural human impulse, and you’ve been through a lot.
January 23, 2026 at 3:46 PM