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
New post! Debunking the Google Gemini/Cloud commercial on the full moon and baseball stats.
marcoshuerta.com/posts/full-m...
marcoshuerta.com/posts/full-m...
The Full Moon Myth Takes the Field – Marcos Huerta
Google’s ad on the full moon effect in baseball is a case study in small number statistics
marcoshuerta.com
October 26, 2026 at 11:38 PM
New post! Debunking the Google Gemini/Cloud commercial on the full moon and baseball stats.
marcoshuerta.com/posts/full-m...
marcoshuerta.com/posts/full-m...
Okay, going to vote. See you on the other side. 🤞
October 27, 2026 at 6:50 AM
Okay, going to vote. See you on the other side. 🤞
Why it's important to consult a statistician before data collection, example 17445:
If you collect data from 1000 patients, nested in 20 therapists, your effective sample size is much (much!) smaller than you think it is... 😏
If you collect data from 1000 patients, nested in 20 therapists, your effective sample size is much (much!) smaller than you think it is... 😏
October 26, 2026 at 6:13 AM
Why it's important to consult a statistician before data collection, example 17445:
If you collect data from 1000 patients, nested in 20 therapists, your effective sample size is much (much!) smaller than you think it is... 😏
If you collect data from 1000 patients, nested in 20 therapists, your effective sample size is much (much!) smaller than you think it is... 😏
People, if you wanna p-hack the shit out of your data, there's no need to pay a statistician - a subscription to the most capable LLM is much cheaper, without any of the pushback.
Guess I'm firing a client today 🤷♂️
Guess I'm firing a client today 🤷♂️
October 26, 2026 at 5:54 AM
People, if you wanna p-hack the shit out of your data, there's no need to pay a statistician - a subscription to the most capable LLM is much cheaper, without any of the pushback.
Guess I'm firing a client today 🤷♂️
Guess I'm firing a client today 🤷♂️
Reposted by Mattan S. Ben-Shachar
Your monthly reminder that if you're using ChatGPT to do your science, you're failing your colleagues, community, and all of us.
Here, ChatGPT makes up data and presents it as the real thing — then apologizes when confronted.
Here, ChatGPT makes up data and presents it as the real thing — then apologizes when confronted.
October 26, 2026 at 5:37 AM
Your monthly reminder that if you're using ChatGPT to do your science, you're failing your colleagues, community, and all of us.
Here, ChatGPT makes up data and presents it as the real thing — then apologizes when confronted.
Here, ChatGPT makes up data and presents it as the real thing — then apologizes when confronted.
Has anyone implemented double-machine-learning with #tidymodels?
October 25, 2026 at 7:34 AM
Has anyone implemented double-machine-learning with #tidymodels?
Three different ways of presenting interactions in this little thread. I think "(non)parallel lines" is the most common, but I like the heatmap, especially when there isn't a focal predictor and a moderator and we're really looking to visualize a joint effect.
There is a very useful graphic that @robin-j-brooks.bsky.social makes repeated use of comparing price elasticity of demand for crude, oil exports from the Gulf and prices. I thought the actual graphic could be improved so had a go at improving it. #rstats code at freerangestats.info/blog/2026/10...
October 25, 2026 at 7:29 AM
Three different ways of presenting interactions in this little thread. I think "(non)parallel lines" is the most common, but I like the heatmap, especially when there isn't a focal predictor and a moderator and we're really looking to visualize a joint effect.
Reposted by Mattan S. Ben-Shachar
You know how the "No brown M&M's" clause in Van Halen's contracts works? (summary: https://en.wikipedia.org/wiki/Van_Halen_test) Well that's what demanding a simulation-based power analysis does for a complicated, e.g. multilevel model: forces you to understand that model properly before showtime.
October 24, 2026 at 9:04 PM
You know how the "No brown M&M's" clause in Van Halen's contracts works? (summary: https://en.wikipedia.org/wiki/Van_Halen_test) Well that's what demanding a simulation-based power analysis does for a complicated, e.g. multilevel model: forces you to understand that model properly before showtime.
Reposted by Mattan S. Ben-Shachar
I actually don’t think you can be a substantive expert without a certain amount of methodological competence. Otherwise you don’t really have a sense of how to interpret your data or the studies of others. It’s just vibes.
October 24, 2026 at 2:27 PM
I actually don’t think you can be a substantive expert without a certain amount of methodological competence. Otherwise you don’t really have a sense of how to interpret your data or the studies of others. It’s just vibes.
As always with AI we need to ask:
1. Is AI good at this?
2. Should we use AI for this?
People are often surprised not only by how hard it is to answer (1) but also by how (2) is a completely separate question than (1)!
1. Is AI good at this?
2. Should we use AI for this?
People are often surprised not only by how hard it is to answer (1) but also by how (2) is a completely separate question than (1)!
Gary King just dropped a study showing that AI peer reviews now dominate human peer reviews on every metric they tested.
gking.harvard.edu/peerless/
gking.harvard.edu/peerless/
AI Article Reviews Dominate Human Peer Reviews at Their Best: Evidence and Policy Choices | Gary King
We show that AI-generated article reviews are now better, by every measure we could construct, than reviews written by well-compensated human experts and judged to be of unusually high quality by jour...
gking.harvard.edu
October 24, 2026 at 6:00 AM
As always with AI we need to ask:
1. Is AI good at this?
2. Should we use AI for this?
People are often surprised not only by how hard it is to answer (1) but also by how (2) is a completely separate question than (1)!
1. Is AI good at this?
2. Should we use AI for this?
People are often surprised not only by how hard it is to answer (1) but also by how (2) is a completely separate question than (1)!
Reposted by Mattan S. Ben-Shachar
Very important for science that you plug a couple of numbers that you made up into the wrong fields in G*Power to get a number that does not change anything, because you’re going to tinker around until it’s the number you want 😌
October 24, 2026 at 3:48 AM
Very important for science that you plug a couple of numbers that you made up into the wrong fields in G*Power to get a number that does not change anything, because you’re going to tinker around until it’s the number you want 😌
Reposted by Mattan S. Ben-Shachar
Which famous movie, book, poem, etc, should Christopher Nolan adapt with no subtext?
Entertaining answers only!
Entertaining answers only!
Would loooove a "Christopher Nolan adapts Citizen Kane" with no subtext.
October 23, 2026 at 11:34 AM
Which famous movie, book, poem, etc, should Christopher Nolan adapt with no subtext?
Entertaining answers only!
Entertaining answers only!
This was a great episode - both very introductory and nuanced.
October 23, 2026 at 8:40 AM
This was a great episode - both very introductory and nuanced.
Reposted by Mattan S. Ben-Shachar
Had this idea in lockdown, half-built it, forgot about it, remembered it again, though might as well finish it.
Local Wikipedia: The Encyclopedia For Local People
Wikipedia, but only for things a half hour walk from you.
edjefferson.com
October 22, 2026 at 12:28 PM
Had this idea in lockdown, half-built it, forgot about it, remembered it again, though might as well finish it.
Reposted by Mattan S. Ben-Shachar
Fully expecting this to be based on a tiny self-selected sample predisposed to aggression, with participants coached on how to act, exaggerated performances for observers, all run by a host with a flair for showmanship but questionable ethical instincts. You know, a direct replication
Netflix has announced a new reality series titled ‘THE NEW STANFORD PRISON EXPERIMENT’.
The series follows 30 participants who enter a real prison and are randomly assigned as guards or prisoners. They must see if they conform to the roles or if morality wins.
The series follows 30 participants who enter a real prison and are randomly assigned as guards or prisoners. They must see if they conform to the roles or if morality wins.
September 16, 2026 at 1:47 AM
Fully expecting this to be based on a tiny self-selected sample predisposed to aggression, with participants coached on how to act, exaggerated performances for observers, all run by a host with a flair for showmanship but questionable ethical instincts. You know, a direct replication
Reposted by Mattan S. Ben-Shachar
#dataviz #stats #rstats
New blog post: Thinking about Penguin Variables
Bill length & depth show Simpson's paradox. Re-expressed as size (L×D) and shape (L/D), it vanishes: same data, rotated 45°, a simpler story.
friendly.github.io/blog/posts/2...
New blog post: Thinking about Penguin Variables
Bill length & depth show Simpson's paradox. Re-expressed as size (L×D) and shape (L/D), it vanishes: same data, rotated 45°, a simpler story.
friendly.github.io/blog/posts/2...
October 20, 2026 at 3:26 PM
#dataviz #stats #rstats
New blog post: Thinking about Penguin Variables
Bill length & depth show Simpson's paradox. Re-expressed as size (L×D) and shape (L/D), it vanishes: same data, rotated 45°, a simpler story.
friendly.github.io/blog/posts/2...
New blog post: Thinking about Penguin Variables
Bill length & depth show Simpson's paradox. Re-expressed as size (L×D) and shape (L/D), it vanishes: same data, rotated 45°, a simpler story.
friendly.github.io/blog/posts/2...
Reposted by Mattan S. Ben-Shachar
Just published open access in 𝘗𝘴𝘺𝘤𝘩𝘰𝘭𝘰𝘨𝘪𝘤𝘢𝘭 𝘔𝘦𝘵𝘩𝘰𝘥𝘴: “𝗪𝗼𝗿𝗸𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗖𝗶𝗿𝗰𝘂𝗹𝗮𝗿 𝗗𝗮𝘁𝗮: 𝗔 𝗧𝘂𝘁𝗼𝗿𝗶𝗮𝗹 𝗳𝗼𝗿 𝗖𝗼𝗴𝗻𝗶𝘁𝗶𝘃𝗲 𝗮𝗻𝗱 𝗕𝗲𝗵𝗮𝘃𝗶𝗼𝗿𝗮𝗹 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵”
🔗 doi.org/10.1037/met0...
🙎 Co-authors: @bayslab.org, @ivntmc.bsky.social, @cjungerius.bsky.social, Xiaolu Wang, Julie de Falco, Zahara Gironés, Adam Triabhall
🔗 doi.org/10.1037/met0...
🙎 Co-authors: @bayslab.org, @ivntmc.bsky.social, @cjungerius.bsky.social, Xiaolu Wang, Julie de Falco, Zahara Gironés, Adam Triabhall
October 19, 2026 at 4:40 PM
Just published open access in 𝘗𝘴𝘺𝘤𝘩𝘰𝘭𝘰𝘨𝘪𝘤𝘢𝘭 𝘔𝘦𝘵𝘩𝘰𝘥𝘴: “𝗪𝗼𝗿𝗸𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗖𝗶𝗿𝗰𝘂𝗹𝗮𝗿 𝗗𝗮𝘁𝗮: 𝗔 𝗧𝘂𝘁𝗼𝗿𝗶𝗮𝗹 𝗳𝗼𝗿 𝗖𝗼𝗴𝗻𝗶𝘁𝗶𝘃𝗲 𝗮𝗻𝗱 𝗕𝗲𝗵𝗮𝘃𝗶𝗼𝗿𝗮𝗹 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵”
🔗 doi.org/10.1037/met0...
🙎 Co-authors: @bayslab.org, @ivntmc.bsky.social, @cjungerius.bsky.social, Xiaolu Wang, Julie de Falco, Zahara Gironés, Adam Triabhall
🔗 doi.org/10.1037/met0...
🙎 Co-authors: @bayslab.org, @ivntmc.bsky.social, @cjungerius.bsky.social, Xiaolu Wang, Julie de Falco, Zahara Gironés, Adam Triabhall
This was a fun one - on mass, people are calibrated predictors (apparently!)
October 20, 2026 at 4:06 PM
This was a fun one - on mass, people are calibrated predictors (apparently!)
Reposted by Mattan S. Ben-Shachar
🤠I'm excited to share our new paper, just published in PNAS, in which we examined more than ONE MILLION forecasts from Polymarket, Kalshi etc, The main finding is striking: Prediction markets and collective forecasting platforms are remarkably well calibrated.
October 20, 2026 at 2:15 PM
🤠I'm excited to share our new paper, just published in PNAS, in which we examined more than ONE MILLION forecasts from Polymarket, Kalshi etc, The main finding is striking: Prediction markets and collective forecasting platforms are remarkably well calibrated.
Reposted by Mattan S. Ben-Shachar
🤫 {shiny} isn't the only "real" way to make an app in #rstats
I've made the same website using {shiny} and {plumber2} for a made-up coffee company and deployed both on @ricochet.rs
ricochet.rs/blog/ssr-in-r/ #databs
I've made the same website using {shiny} and {plumber2} for a made-up coffee company and deployed both on @ricochet.rs
ricochet.rs/blog/ssr-in-r/ #databs
The Shiny myth 🫢 | ricochet
Or, why plumber2 might be all you need.
ricochet.rs
October 19, 2026 at 3:17 PM
🤫 {shiny} isn't the only "real" way to make an app in #rstats
I've made the same website using {shiny} and {plumber2} for a made-up coffee company and deployed both on @ricochet.rs
ricochet.rs/blog/ssr-in-r/ #databs
I've made the same website using {shiny} and {plumber2} for a made-up coffee company and deployed both on @ricochet.rs
ricochet.rs/blog/ssr-in-r/ #databs
Reposted by Mattan S. Ben-Shachar
bmm 1.4.1 is on CRAN!
Signal detection theory & multinomial processing tree models join bmm: hierarchical Bayesian measurement models of cognition, written as brms-style formulas and fit with Stan.
Get the new version now, with:
install.packages("bmm")
🧵 #rstats #brms #PsychMethods
Signal detection theory & multinomial processing tree models join bmm: hierarchical Bayesian measurement models of cognition, written as brms-style formulas and fit with Stan.
Get the new version now, with:
install.packages("bmm")
🧵 #rstats #brms #PsychMethods
October 19, 2026 at 2:16 PM
bmm 1.4.1 is on CRAN!
Signal detection theory & multinomial processing tree models join bmm: hierarchical Bayesian measurement models of cognition, written as brms-style formulas and fit with Stan.
Get the new version now, with:
install.packages("bmm")
🧵 #rstats #brms #PsychMethods
Signal detection theory & multinomial processing tree models join bmm: hierarchical Bayesian measurement models of cognition, written as brms-style formulas and fit with Stan.
Get the new version now, with:
install.packages("bmm")
🧵 #rstats #brms #PsychMethods
Reposted by Mattan S. Ben-Shachar
Why do I want access to all the books & articles? For training, testing & inference. But don't worry, it's 'transformative use'; what comes out will be very different from what I put in. It'll even grow the market for what I'm asking for. No, no, this isn't about LLMs. I'm talking about my students.
October 19, 2026 at 5:24 PM
Why do I want access to all the books & articles? For training, testing & inference. But don't worry, it's 'transformative use'; what comes out will be very different from what I put in. It'll even grow the market for what I'm asking for. No, no, this isn't about LLMs. I'm talking about my students.
Reposted by Mattan S. Ben-Shachar
What about PhD students who confuse correlation and causation
October 16, 2026 at 7:44 PM
What about PhD students who confuse correlation and causation