Lightnews — Scholar-powered news

Matthew Kay

@mjskay.com

11K followers 1K following 1.5K 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

mjskay.com

Posts Media Videos Starter Packs

Pinned

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.

3 12 81

Matthew Kay @mjskay.com · 6h

The Confounders would 100% have started as a brother/sister twee-ish aughts indie folk band (think Iron & Wine meets Fiery Furnaces) and over decades shifted through baroque pop, post-rock, and 80s new wave revival to land on some strange edm mashup of like Kraftwerk and Godspeed You! Black Emperor

Matthew Kay @mjskay.com · 21h

good call! or even "insquariance"?

hmm I see definite pedagogical advantages to squariance / rootsquariance / insquariance...

Matthew Kay @mjskay.com · 22h

missed opportunity to rename all dispersion measures following the same scheme

OG variance = squariance
SD = stairiance (also known as rootsquariance)
Gini = fairiance
IQR = iquariance (pronounced with a schwa, like "aquarius")
MAD = marriance
precision = prairiance
etc

1 12

Matthew Kay @mjskay.com · 2d

ah yes, pull request check, perfect 👍

1 4

Reposted by Matthew Kay

Evan Peck @peck.phd · 5d

omg - this alt.vis paper about NASA-TLX was styled like the original NASA-TLX paper 😍

arxiv.org/pdf/2509.24643

arxiv.org

1 2 16

Matthew Kay @mjskay.com · 5d

don't sell yourself short! I see lots of small touches that elevate it: pleasing (and colorblind-safe) colors, densities are lighter shades of corresponding intervals, interval lines are just-thick-enough-but-not-too-thick, y axis labels are stacked to match subgroups (+ all text horizontal), ...

1 9

Matthew Kay @mjskay.com · 5d

nice blorbs

1 1

Matthew Kay @mjskay.com · 5d

Yeah who knows, plus even if it has plateaued those numbers are still very high. I'd be surprised if it dropped in any meaningful way without some kind of widescale movement against it.

Matthew Kay @mjskay.com · 5d

lolwut

1 2

Matthew Kay @mjskay.com · 5d

dunno about you but between the death of twitter and *waves hands* everything fucking else I've been consciously setting my phone down out of arms' reach and (for the first time in years) listening to an honest-to-god physical radio in the morning

Tony Tassell @tonytassell.bsky.social · 5d

Have we passed peak social media? @jburnmurdoch.ft.com here www.ft.com/content/a072...

4 1 21

Matthew Kay @mjskay.com · 5d

that's just German for "the p value, the"

2 2 25

Matthew Kay @mjskay.com · 6d

In either case it doesn't so much have a quibble as generate text that expresses a quibble

1 6

Matthew Kay @mjskay.com · 8d

fun! are we allowed to improve the chart?

that density estimator doesn't account for the boundedness of group C and it's usually possible to do better than random jitter

density + 50% and 95% intervals + mean + compressed dotplot with exact dot positions on three groups, including group C which is automatically estimated using a bounded KDE due to a lower bound.

library(ggplot2)
library(ggdist)

spreads |>
ggplot(aes(Group, Value, fill = Group)) +
geom_weave(
side = "left",
scale = 0.75,
alpha = 0.35,
binwidth = unit(1, "mm"),
overlaps = "keep",
color = NA
) +
stat_slab(
scale = 0.75,
alpha = 0.65,
color = "black",
linewidth = 0.25
) +
stat_pointinterval(
fill = "white",
shape = 23,
point_interval = "mean_qi",
.width = c(.5, .95)
) +
guides(fill = "none") +
labs(x = NULL) +
theme_minimal(base_size = 13) +
theme(panel.grid.minor = element_blank())

1 1 6

Reposted by Matthew Kay

Gregg Gonsalves @gregggonsalves.bsky.social · 13d

I've been sitting with this piece of @olufemiotaiwo.bsky.social for two days, reading it, re-reading it. You should too. #EzraKlein is the avatar of a new kind of liberal indulgence for the worst of the far-right, embracing bad-faith as a practice. www.bostonreview.net/articles/how...

How Can We Live Together? - Boston Review

Ezra Klein is wrong: shame is essential.

www.bostonreview.net

24 150 540

Reposted by Matthew Kay

Dr. Julie Kientz @juliekientz.bsky.social · 28d

At @hcde.uw.edu, we will be hiring an Assistant or Associate Teaching Professor this coming year! Timing wise, we are hiring for a Sept. 2026 start date, but could be sooner.

We are looking forward to welcoming a new colleague!

Learn more and apply at apply.interfolio.com/171881

Apply - Interfolio {{$ctrl.$state.data.pageTitle}} - Apply - Interfolio

apply.interfolio.com

8 11

Matthew Kay @mjskay.com · 13d

this opinion brought to you by the year 1995

Matthew Kay @mjskay.com · 13d

I guess another parallel to R is that many people think poorly of both for (possibly) shallow reasons, and miss out on what makes them interesting (and in some ways quite special)

Reposted by Matthew Kay

Luis Apiolaza @ojala.bsky.social · 14d

I need an 'Excel for R users' book. I _have_ to use Excel in a course (not for stats) and it feels very strange.

7 5 36

Matthew Kay @mjskay.com · 13d

There are interesting parallels between them that could make for a conceptually deep treatment - e.g. the reactive nature of computation in Excel versus lazy evaluation in R (or more explicitly reactivity in Shiny), or mappings between Excel's cell/row/col model and vectors and data frames in R

2 4

Matthew Kay @mjskay.com · 16d

...unless you need cumulative probabilities, in which case a CDF will be best (naturally).

Not an empirical study, but @birdbassador.jorts.horse.ap.brid.gy has a very nice analytical take on variations of raincloud plots and when to use them doi.org/10.1111/cgf....

Teru Teru Bōzu: Defensive Raincloud Plots

Univariate visualizations like histograms, rug plots, or box plots provide concise visual summaries of distributions. However, each individual visualization may fail to robustly distinguish important....

doi.org

1 3

Matthew Kay @mjskay.com · 16d

Off the top of my head I don't know of studies of *combos* of KDE + interval + dot / scatter, but there are studies of them in isolation. I think KDE + dots in rainclouds are often redundant... point + interval + dotplot gets you precise estimates of most everything (including density)...

1 3

Matthew Kay @mjskay.com · 19d

Haha fine fine. Seems to really depend on where/how you learned it. I grew up on PostgreSQL / MySQL (their docs say "S-Q-L") and my databases prof also pronounced it that way. Oracle and MS SQL Server folks seem to be more on the "sequel" side.

1 5

Matthew Kay @mjskay.com · 19d

S-Q-L (the pronunciation according to one of its creators) vs "sequel"

1 6

Matthew Kay @mjskay.com · 20d

And also, as you gain little bits of influence and decision making power, change the incentives!

Reposted by Matthew Kay

MC Stan @mc-stan.org · 21d

MC Stan is here! Follow for the latest Stan news, and tag if you want us to repost your posts about new papers, packages, courses, etc. about Stan

1 33 38