Mattan S. Ben-Shachar
@mattansb.msbstats.info
Statistics lecturer | Freelance statistical consultant & research analyst | #rstats dev @easystats.github.io
home.msbstats.info
(He/Him)
home.msbstats.info
(He/Him)
Pinned
1k followers! Calls for a re-introduction:
Statistics lecturer, freelance stats consultant, & #rstats dev @easystats.bsky.social 📊
I try to help social scientists make better inferences from their data & communicate their findings 👨🏫
Hope to bring the #stats twitter/R community vibes over to bsky!
Statistics lecturer, freelance stats consultant, & #rstats dev @easystats.bsky.social 📊
I try to help social scientists make better inferences from their data & communicate their findings 👨🏫
Hope to bring the #stats twitter/R community vibes over to bsky!
Reposted by Mattan S. Ben-Shachar
For sale: randomized controlled trial, never conducted
November 11, 2025 at 9:47 AM
For sale: randomized controlled trial, never conducted
Reposted by Mattan S. Ben-Shachar
Great teaching case! On the Bad Science Bingo Card 8, 9, 10, 11, 13, 31, and 36 are strong contenders for a prize for this article detectingbadscience.wordpress.com/bad-science-...
Bad Science Bingo
Here are 36 potential flaws in research reports. Can you find them all? For each flaw, a short description is below. Follow the link for an elaborate description and visualization. Conflicts of Int…
detectingbadscience.wordpress.com
November 9, 2025 at 2:11 PM
Great teaching case! On the Bad Science Bingo Card 8, 9, 10, 11, 13, 31, and 36 are strong contenders for a prize for this article detectingbadscience.wordpress.com/bad-science-...
Random* YouTubing led me to this very cool website about testing RNGs!
*Maybe?
#stats
random.tastemaker.design
*Maybe?
#stats
random.tastemaker.design
Randomness Testing Guide
Test the randomness of random number generators.
random.tastemaker.design
November 8, 2025 at 5:48 PM
Random* YouTubing led me to this very cool website about testing RNGs!
*Maybe?
#stats
random.tastemaker.design
*Maybe?
#stats
random.tastemaker.design
Reposted by Mattan S. Ben-Shachar
You can now find a recording of my course "Introduction to Bayesian Statistics in R & brms" on youtube.
Slides & code available here: github.com/benjamin-ros...
#Rstats
Slides & code available here: github.com/benjamin-ros...
#Rstats
Introduction to Bayesian Statistics in R & brms - YouTube
The course offers a straightforward and practical approach to applied statistics using Bayesian inference for ecologists.
www.youtube.com
November 7, 2025 at 2:18 PM
You can now find a recording of my course "Introduction to Bayesian Statistics in R & brms" on youtube.
Slides & code available here: github.com/benjamin-ros...
#Rstats
Slides & code available here: github.com/benjamin-ros...
#Rstats
Reposted by Mattan S. Ben-Shachar
The advances we've made in statistics, experimental study design, and causal inference over the past century are remarkably useful for understanding our world. But there is never been a push to make people use them like we are seeing with generative AI. Perhaps take a moment to consider why.
November 7, 2025 at 9:07 AM
The advances we've made in statistics, experimental study design, and causal inference over the past century are remarkably useful for understanding our world. But there is never been a push to make people use them like we are seeing with generative AI. Perhaps take a moment to consider why.
Reposted by Mattan S. Ben-Shachar
Do you teach #rstats? Do your students complain about how lame and old-fashioned dplyr is? Don't worry: I have the solution for you: github.com/hadley/genzp....
genzplyr is dplyr, but bussin fr fr no cap.
genzplyr is dplyr, but bussin fr fr no cap.
GitHub - hadley/genzplyr: dplyr but make it bussin fr fr no cap
dplyr but make it bussin fr fr no cap. Contribute to hadley/genzplyr development by creating an account on GitHub.
github.com
November 6, 2025 at 11:25 PM
Do you teach #rstats? Do your students complain about how lame and old-fashioned dplyr is? Don't worry: I have the solution for you: github.com/hadley/genzp....
genzplyr is dplyr, but bussin fr fr no cap.
genzplyr is dplyr, but bussin fr fr no cap.
The AI revolution is here..?
In a show my kid is watching:
Cute cat is gifted with a dancing robot to dance with.
Cockatoo steals the robot because it wants someone to dance with.
How I thought the story would resolve: cockatoo & cat dance together.
How the story resolved: the share the robot.
In a show my kid is watching:
Cute cat is gifted with a dancing robot to dance with.
Cockatoo steals the robot because it wants someone to dance with.
How I thought the story would resolve: cockatoo & cat dance together.
How the story resolved: the share the robot.
November 7, 2025 at 11:46 AM
The AI revolution is here..?
In a show my kid is watching:
Cute cat is gifted with a dancing robot to dance with.
Cockatoo steals the robot because it wants someone to dance with.
How I thought the story would resolve: cockatoo & cat dance together.
How the story resolved: the share the robot.
In a show my kid is watching:
Cute cat is gifted with a dancing robot to dance with.
Cockatoo steals the robot because it wants someone to dance with.
How I thought the story would resolve: cockatoo & cat dance together.
How the story resolved: the share the robot.
Today's challenge: build a model with fixed parameters (that do not need to be estimated) that can still be used with #tidymodels functionality.
I found a solution (that didn't require setting up a new type of model) - will post some code later - but I wonder how @topepo.bsky.social would do this?
I found a solution (that didn't require setting up a new type of model) - will post some code later - but I wonder how @topepo.bsky.social would do this?
November 4, 2025 at 3:33 PM
Today's challenge: build a model with fixed parameters (that do not need to be estimated) that can still be used with #tidymodels functionality.
I found a solution (that didn't require setting up a new type of model) - will post some code later - but I wonder how @topepo.bsky.social would do this?
I found a solution (that didn't require setting up a new type of model) - will post some code later - but I wonder how @topepo.bsky.social would do this?
Reposted by Mattan S. Ben-Shachar
time travel assassination targets
normal people: Hitler, Stalin, Leopold II
data analysts: whoever invented dates
normal people: Hitler, Stalin, Leopold II
data analysts: whoever invented dates
November 3, 2025 at 3:12 PM
time travel assassination targets
normal people: Hitler, Stalin, Leopold II
data analysts: whoever invented dates
normal people: Hitler, Stalin, Leopold II
data analysts: whoever invented dates
Reposted by Mattan S. Ben-Shachar
This is irresponsible, short-minded and unwise. Knowledge is not something that is injected or attached to our brains. It is acquired with practice and reasoning. Retrieving information from a source is not understanding. Accumulating data is not knowledge. You are promoting a generation of dumb.
November 3, 2025 at 4:18 PM
This is irresponsible, short-minded and unwise. Knowledge is not something that is injected or attached to our brains. It is acquired with practice and reasoning. Retrieving information from a source is not understanding. Accumulating data is not knowledge. You are promoting a generation of dumb.
Reposted by Mattan S. Ben-Shachar
Hell NO!
Stop this nonsense. Doctoral training did not change in nature because we got computers instead of typewriters and calculators.
"As machines get better at data analysis and writing tasks, doctoral training must evolve to make the most of artificial-intelligence outputs."
shorturl.at/fUQcm
Stop this nonsense. Doctoral training did not change in nature because we got computers instead of typewriters and calculators.
"As machines get better at data analysis and writing tasks, doctoral training must evolve to make the most of artificial-intelligence outputs."
shorturl.at/fUQcm
PhD training needs a reboot in an AI world
As machines get better at data analysis and writing tasks, doctoral training must evolve to make the most of artificial-intelligence outputs.
shorturl.at
November 3, 2025 at 4:17 PM
Hell NO!
Stop this nonsense. Doctoral training did not change in nature because we got computers instead of typewriters and calculators.
"As machines get better at data analysis and writing tasks, doctoral training must evolve to make the most of artificial-intelligence outputs."
shorturl.at/fUQcm
Stop this nonsense. Doctoral training did not change in nature because we got computers instead of typewriters and calculators.
"As machines get better at data analysis and writing tasks, doctoral training must evolve to make the most of artificial-intelligence outputs."
shorturl.at/fUQcm
Reposted by Mattan S. Ben-Shachar
Reposted by Mattan S. Ben-Shachar
Actually, we recently spooked away some ghosts with warnings and other scary things (with @semihaktepe.bsky.social)
October 30, 2025 at 7:00 AM
Actually, we recently spooked away some ghosts with warnings and other scary things (with @semihaktepe.bsky.social)
Reposted by Mattan S. Ben-Shachar
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/)
I'm now also looking for a postdoc with strong Bayesian background and interest in developing Bayesian cross-validation theory, methods and software. Apply by email with no specific deadline (see contact information at users.aalto.fi/~ave/).
Others, please share
Others, please share
I'm looking for a doctoral student with Bayesian background to work on Bayesian workflow and cross-validation (see my publication list users.aalto.fi/~ave/publica... for my recent work) at Aalto University.
Apply through the ELLIS PhD program (dl October 31) ellis.eu/news/ellis-p...
Apply through the ELLIS PhD program (dl October 31) ellis.eu/news/ellis-p...
November 3, 2025 at 11:13 AM
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/)
Reposted by Mattan S. Ben-Shachar
If only there was a good explanation as to why 🤔
Pre-post effect sizes neatly rise with baseline severity in this evaluation of inpatient/day treatment outcomes…
www.frontiersin.org/journals/psy...
www.frontiersin.org/journals/psy...
November 3, 2025 at 11:44 AM
If only there was a good explanation as to why 🤔
The university has announced it will be using materials uploaded to the course Moodle and recorded lecture transcripts to train LLM "TA bots".
And thus the circle is complete: AI teachers teaching AI students.
And thus the circle is complete: AI teachers teaching AI students.
November 2, 2025 at 3:49 PM
The university has announced it will be using materials uploaded to the course Moodle and recorded lecture transcripts to train LLM "TA bots".
And thus the circle is complete: AI teachers teaching AI students.
And thus the circle is complete: AI teachers teaching AI students.
Reposted by Mattan S. Ben-Shachar
Version 2.0.0 of emmeans is out! Notable changes: new look to the plots, qdrg() now an S3 method, and new maintainer (me, Russ still highly involved). cran.r-project.org/package=emme...
emmeans: Estimated Marginal Means, aka Least-Squares Means
Obtain estimated marginal means (EMMs) for many linear, generalized linear, and mixed models. Compute contrasts or linear functions of EMMs, trends, and comparisons of slopes. Plots and other displays...
cran.r-project.org
October 29, 2025 at 4:05 PM
Version 2.0.0 of emmeans is out! Notable changes: new look to the plots, qdrg() now an S3 method, and new maintainer (me, Russ still highly involved). cran.r-project.org/package=emme...
Reposted by Mattan S. Ben-Shachar
I completely agree! But if you realise that you can’t specify a SESOI for the life of you, it usually means that you can’t specify your research hypothesis enough to make it statistically testable. In that case, the best decision may to do something else: journals.sagepub.com/doi/10.1177/...
Why Hypothesis Testers Should Spend Less Time Testing Hypotheses - Anne M. Scheel, Leonid Tiokhin, Peder M. Isager, Daniël Lakens, 2021
For almost half a century, Paul Meehl educated psychologists about how the mindless use of null-hypothesis significance tests made research on theories in the s...
journals.sagepub.com
October 31, 2025 at 10:23 AM
I completely agree! But if you realise that you can’t specify a SESOI for the life of you, it usually means that you can’t specify your research hypothesis enough to make it statistically testable. In that case, the best decision may to do something else: journals.sagepub.com/doi/10.1177/...
Reposted by Mattan S. Ben-Shachar
I'm increasingly proud to have written and published things before LLMs existed.
November 1, 2025 at 11:55 AM
I'm increasingly proud to have written and published things before LLMs existed.
Reposted by Mattan S. Ben-Shachar
There still seems to be a lot of confusion about significance testing in psych. No, p-values *don’t* become useless at large N. This flawed point also used to be framed as "too much power". But power isn't the problem – it's 1) unbalanced error rates and 2) the (lack of a) SESOI. 1/ >
But here's, the thing, p values and significance become useless at such large sample sizes. When you're dividing the coefficient by the SE and the sample size is in the tens of thousands, EVERYTHING IS SIGNIFICANT. All you're testing is whether the coefficient is different than zero.
October 31, 2025 at 8:13 AM
There still seems to be a lot of confusion about significance testing in psych. No, p-values *don’t* become useless at large N. This flawed point also used to be framed as "too much power". But power isn't the problem – it's 1) unbalanced error rates and 2) the (lack of a) SESOI. 1/ >
Sometimes the DAGs implied by the statistical analysis people do is just so bonkers.
The worst part?
If you drew these DAGs and presented them to those same people - they would 100% agree that those DAGs make no sense.
always 👏 draw 👏 your 👏 DAGs!
The worst part?
If you drew these DAGs and presented them to those same people - they would 100% agree that those DAGs make no sense.
always 👏 draw 👏 your 👏 DAGs!
October 30, 2025 at 7:39 PM
Sometimes the DAGs implied by the statistical analysis people do is just so bonkers.
The worst part?
If you drew these DAGs and presented them to those same people - they would 100% agree that those DAGs make no sense.
always 👏 draw 👏 your 👏 DAGs!
The worst part?
If you drew these DAGs and presented them to those same people - they would 100% agree that those DAGs make no sense.
always 👏 draw 👏 your 👏 DAGs!
Reposted by Mattan S. Ben-Shachar
I built a DAG diagram with garden hoses for teaching.
Pictured: a collider bias diagram, inspired by a blocked pipe situation I experienced (which I credit with giving me the intuition though it also ruined my belongings in the flooded cellar).
Pictured: a collider bias diagram, inspired by a blocked pipe situation I experienced (which I credit with giving me the intuition though it also ruined my belongings in the flooded cellar).
October 28, 2025 at 5:50 PM
I built a DAG diagram with garden hoses for teaching.
Pictured: a collider bias diagram, inspired by a blocked pipe situation I experienced (which I credit with giving me the intuition though it also ruined my belongings in the flooded cellar).
Pictured: a collider bias diagram, inspired by a blocked pipe situation I experienced (which I credit with giving me the intuition though it also ruined my belongings in the flooded cellar).
Reposted by Mattan S. Ben-Shachar
New release of PowerLMM.js! Browser-based power analysis for longitudinal models with dropout.
Now includes:
- Power analysis summary report
- Reproducible & shareable configs (URL/JSON)
- Calculations validated against R
- Hypothesis region visualization
powerlmmjs.rpsychologist.com
Now includes:
- Power analysis summary report
- Reproducible & shareable configs (URL/JSON)
- Calculations validated against R
- Hypothesis region visualization
powerlmmjs.rpsychologist.com
October 28, 2025 at 2:02 PM
New release of PowerLMM.js! Browser-based power analysis for longitudinal models with dropout.
Now includes:
- Power analysis summary report
- Reproducible & shareable configs (URL/JSON)
- Calculations validated against R
- Hypothesis region visualization
powerlmmjs.rpsychologist.com
Now includes:
- Power analysis summary report
- Reproducible & shareable configs (URL/JSON)
- Calculations validated against R
- Hypothesis region visualization
powerlmmjs.rpsychologist.com
Reposted by Mattan S. Ben-Shachar
in my post-academic era i have become extremely scathing of the way academia considers papers to the primary outcome of intellectual work, and refuses to give credit to academics who write software or build other tooling. in industry we very rarely read your papers; but we always use your software
October 28, 2025 at 2:15 AM
in my post-academic era i have become extremely scathing of the way academia considers papers to the primary outcome of intellectual work, and refuses to give credit to academics who write software or build other tooling. in industry we very rarely read your papers; but we always use your software