Mattan S. Ben-Shachar

@mattansb

Statistics lecturer | Freelance statistical consultant & research analyst | #rstats dev @easystats.github.io home.msbstats.info (He/Him)

403 Folgt3.649 Follower2.321 Posts

Bluesky

Ist das dein Profil?Übernimm es — dann gehört dir mattansb.dirty.social.Profil übernehmen

Mattan S. Ben-Shachar@mattansb· 9d

#statsmeme

Mattan S. Ben-Shachar@mattansb· 12d

Idk man... Very sus...

Mattan S. Ben-Shachar@mattansb· 12d

Want to write a paper together introducing the new ODD (odd differences delta)? Title: Isn't that ODD? Bridging the gap between odds ratios and risk differences.

library(marginaleffects)

mod <- glm(
am ~ vs,
family = binomial("logit"),
data = mtcars
)

odds <- avg_predictions(
mod,
variables = "vs",
type = "response",
transform = effectsize::probs_to_odds
)

# odds difference
hypotheses(odds, hypothesis = difference ~ pairwise)

Mattan S. Ben-Shachar@mattansb· 15d

A client thought up some BS with Claude and sent me the full report + analysis code, all 100% written and run and interpreted by Claude. I gave a detailed response explaining why it was trash, which my client (for laughs) forwarded to Claude. Claude was not happy being criticized.

Mattan S. Ben-Shachar@mattansb· 22d

I meant in the sense that (1 | g) is captured in (1 | g:f), it think? Like, any differences in subject-means are also modelled in the differences between subject-condition-means, no?

Mattan S. Ben-Shachar@mattansb· 22d

While I'm here, 🧵 rant: People talking about how their paper was improved by Peer Review™ often fail to consider how much better it was, how much effort (time, actual money, opportunity) was wasted for these diminishing returns, and how often papers are actually WORSE following Peer Review™. 2/

Mattan S. Ben-Shachar@mattansb· 28d

Clearly written before LLMs - now everyone is an expert in a matter of *minutes*!

Mattan S. Ben-Shachar@mattansb· Jul 9

Rewatching Community, John Oliver (the psychology professor) is channeling how we all feel when data misbehaves... #stats

Mattan S. Ben-Shachar@mattansb· Jul 7

The Hallgrímskirkja church is very heavy tailed... #rstats #ggplot2

library(ggplot2)
library(distributional)
library(ggdist)
library(ggfx)

update_geom_defaults("slab", aes(fill = NA, linewidth = 1, alpha = 0.7))

img <- jpeg::readJPEG("msid-44847844,width-96,height-65-499916619.jpg")

p <- ggplot() +
annotation_raster(img, xmin = -9.5, xmax = 9.5, ymin = -0.2, ymax = 1.07) +
geom_hline(yintercept = 0, color = "black", linewidth = 1) |>
with_outer_glow("white") +
stat_slab(aes(xdist = dist_cauchy(0, 1), color = "Cauchy"), n = 10001) |>
with_outer_glow("white") +
stat_slab(aes(xdist = dist_laplace(0, 1), color = "Laplace")) |>
with_outer_glow("white") +
stat_slab(aes(
xdist = dist_student_t(3, 0, 1),
color = "Student t (df = 3)"
)) |>
with_outer_glow("white") +
stat_slab(
aes(xdist = dist_normal(0, 1), color = "Std. Normal"),
p_limits = c(.000000001, .999999999)
) |>
with_outer_glow("white") +
coord_fixed(
ratio = (19 / 1.27) / (1024 / 683),
xlim = c(-8, 8),
ylim = c(-0.1, 1),
expand = c(0, 0)
) +
scale_color_viridis_d(
NULL,
limits = rev(c("Cauchy", "Student t (df = 3)", "Laplace", "Std. Normal"))
) +
labs(
title = "Which distribution best fits the Hallgrímskirkja church facade?",
x = expression(theta),
y = NULL
) +
theme_minimal() +
theme_sub_axis_left(text = element_blank(), ticks = element_blank())

ggsave(
"church.png",
p,
width = 11,
height = 7,
units = "cm",
dpi = 600,
scaling = 1 / 1.8
)

Mattan S. Ben-Shachar@mattansb· Jul 5

Playing around with Google's notebookLM. Asked for a video summary of my class on objective priors, where I explained how flat priors are counterproductive in many cases. The result is basically making the opposite point... 😬 #stats #bayes

Mattan S. Ben-Shachar@mattansb· Jul 1

"... so you can spend time on science." What does Anthropic think "doing science" *is*?? claude.com/product/clau...

Mattan S. Ben-Shachar@mattansb· Jun 19

There are apparently two Mattan S. Ben-Shachars in the weights! (They're both me...)

Mattan S. Ben-Shachar@mattansb· Jun 15

#statsmeme? (Prompted by a conversation with a student who's double majoring in stats-econ and said that stats courses make more sense to him)

Mattan S. Ben-Shachar@mattansb· Jun 15

Accidentally color coordinated with my slides yesterday 🤭 (The actual colors were blue and green, but thanks to the projector being old, my students got a good laugh)

Mattan S. Ben-Shachar@mattansb· Jun 9

Mattan S. Ben-Shachar@mattansb· Jun 7

Mattan S. Ben-Shachar@mattansb· Jun 2

The {tidymodels} hex sticker for the {censored} package is perfect 😂 (but also, I'm a little uncomfortable?)

Mattan S. Ben-Shachar@mattansb· May 31

Psychologists doing exploratory factor analysis be like: #statsmeme

Mattan S. Ben-Shachar@mattansb· May 31

Psychologists doing clustering be like: #statsmeme

Mattan S. Ben-Shachar@mattansb· May 31

Difference scores strike again! Here are those results on the full scales - what's really going on is that all religions (inc. atheism and agnostism) excluding JW should be left but also joined (??).

Mattan S. Ben-Shachar@mattansb· May 29

At least 2500! #statsmeme

Mattan S. Ben-Shachar@mattansb· May 29

Mattan S. Ben-Shachar@mattansb· May 27

#statsmeme

Mattan S. Ben-Shachar@mattansb· May 27

I can't be the only one who hears this, right? #statsmeme

Mattan S. Ben-Shachar@mattansb· May 26

Is that so?

Mattan S. Ben-Shachar@mattansb· May 20

I was wondering how "DAG" would be pronounced with an Australian accent, and this somehow let me to remember the greatest accent ever depicted in film... #statsmeme

Mattan S. Ben-Shachar@mattansb· May 19

The vignette also introduces the (improved) Bayes factor matrices to compare between all the models, examples of using {bayestestR} with #marginaleffects and #emmeans, demo for setting priors on factor encodings, ...

Mattan S. Ben-Shachar@mattansb· May 19

🚨NEW-ish VIGNETTE ALERT🚨 The Bayes factor vignette for {bayestestR} has been completely re-written! We've got all the BFs! - Marginal likelihoods - Model & posterior averaging - Order restrictions - Savage-Dickey density ratios AND MORE! Read it up, here: @easystats.github.io #Bayes #rstats

Mattan S. Ben-Shachar@mattansb· May 18

I just can't help myself sometimes. I need to apologize to my students that were subjected to this silliness today... #statsmeme #Bayes

Mattan S. Ben-Shachar@mattansb· May 11

I've started including @djnavarro.net 's #rstats {sessioncheck} in code I write for various clients' projects - it helps me keep my sessions reproducible! Anyway.. a client recently asked for one of my scripts, and when they sent it back I had a good laugh seeing the session check commented out 😂