▰▰▰▰▰▰▰▰ 100%

Here's one paper on that, in the context of hiring but you could almost copy and paste peer review. There are, of course other concerns like LLMs forcing science back towards the training data vs. outwards towards brand new understandings. www.pnas.org/doi/10.1073/...

life is so bad when you have a tea-making apparatus that sucks. everything turns for the good when you have a tea-making apparatus that is good.

▰▰▰▰▰▰▱▱ 75% Please, no more fried food. Okay, maybe one more sfenj 🤤

All research is exploratory if you’re confused enough

Julia guest spoke at our Psych department's weekly seminar yesterday (I guess that was her second-to-last talk then?) about causal inference, DAGs and all the beautiful things in between. I think some minds were blown - which was my intention all along! 😈 Slides: juliarohrer.com/wp-content/u...

Students often assume that models create insight. I remind them: models compress what you already understand. If you don’t understand the domain, the model won’t rescue you. Assumptions come first. #AI #DataScience #MachineLearning #RStats

▰▰▰▰▱▱▱▱ 50% Now with latkas! 🤤

One place I never imagined "not being a sports guy" would catch up with me is grading papers... About half the projects in my data science course are about the NBA, and I find myself with Wikipedia ope - trying to make sense of the analyses the students are describing, and... just... 😮‍💨

New preprint 🎉 Psych constructs are complex. Symptoms overlap, people rarely fit neat categories, and patterns are non-linear. Most methods compromise this richness. Self-Organising Maps don't. We provide a step-by-step tutorial with annotated R code to make them accessible. doi.org/10.31234/osf...

Happy holidays! A Menora is really just a Hanukkah progress bar: ▰▰▰▱▱▱▱▱ 37.5%

Grading and googling hallucinated citations, as one does nowadays, and now that LLMs have been around for a while, I've discovered new horrors: hallucinated journals are now appearing in Google Scholar with dozens of citations bc so many people are citing these fake things

I ignored the strip.clip argument in #ggplot2 for way too long 😲 Combined with a small negative margin tweak, you can place facet labels inside each panel. A tiny trick that makes small multiples feel so much cleaner. 🔵 no manual coordinates 🔵 inherits theme styling 🔵 scales nicely when resizing

Seven-parameter drift-diffusion pdfs and cdfs now in Stan statmodeling.stat.columbia.edu/2025/12/11/s...

Empathic disequilibrium could change the way we think about empathy and help recognise and accept diverse empathic experiences, moving us beyond the usual “low” or “high” empathy narrative. #Neurodiversity #Autism #Empathy www.cell.com/trends/cogni...

#rstats Wondering about the status of a package you submitted to CRAN? Worried that some reply went into your spam folder? There's a CRANSAYS dashboard for that r-hub.github.io/cransays/art...

Some closing thoughts for my students this semester on LLMs and learning #rstats datavizf25.classes.andrewheiss.com/news/2025-12...

Every time I do this, I feel like a pervert. mat_list <- matrix(list(), nrow =..., ncol = ...) #rstats

%notin% is coming to Base R! Heck to the yes. We are truly blessed on this day, thank you R Core. 🤩 #rstats

?untidy (after loading emmeans) "Dare to be un-tidy! ... Statistical analysis is not just a workflow; it is a discipline" #rstats

Someone please make "responsibility analysis" a thing!

Power analysis. No responsibility analysis. Only power analysis.

Really? What if people decide to have children exactly at the moment when they anticipate their sleep satisfaction to decline? 😜

The interesting thing about the Table 2 Fallacy is that it works even if you're only looking at regression as a prediction model: standardized coefficients (betas) aren't equal to semi-partial correlations. #stats

The scientist-statistician was one of the most common professions in 21st century universities. In this era, statistics was seldom conducted by trained statisticians. Instead, subject experts, who possessed hard- and software, were responsible for tasks ranging from study design to causal inference

Indeed. It's sometimes (often?) harder to do things the right way in medical/health research because editors/reviewers stop you from doing it. statsepi.substack.com/p/how-bad-st...

I've been exposed to enough peer review that I think we can look at two classes of reviewers: 1. Constructive peer review 2. Adversarial peer review I'm looking at this mostly from a psychology/methodology perspective (but wonder what other fields experience) 🧵 1/

I am going to avoid QTing as the intention is not to dunk. However complaints that stats is "too hard" and there should be tools to automate it miss the mark. Stats is a profession that requires hard work, like most others. They ARE hard. That's why you partner with experts. #statsky #rstats #medsky

RBF for RBF: using support vector machines withto classify resting bitch face

Probably a rhetorical questions, but I'll add the data anyway: We recently estimated—among other things—adoption of Bayesian analyses across different areas of psychology. The plots show the last 20 years and (rather simple) projections for the next 20 years. osf.io/preprints/ps...

Client does ANOVA. Reviewer 2 requests complex multilevel model instead. I help client run the complex model, get interesting results. Reviewer 2 is critical of the original power analysis claiming the study is under powered for the complex multilevel model. Well, yeah, duh 🙄🤦‍♂️

This blog post blog.mc-stan.org/2022/08/03/o... discusses use of external BLAS/LAPACK libraries with @mc-stan.org

Okay, I just did this and it is soooo worth it! #rstats

Logistic regression with 5 million observations. glm() - about 2 hours glm(start = <values found on a tiny subset>) - less than 2 minutes. Almost feels too good... #rstats

Best use of #Positron assistant?

All models are wrong; some are useful. All LLM outputs are hallucinations. Some are useful.

As a teacher, give me someone who wants to learn over basically any other trait

”by = x” generates predictions for the data “as is“ and then averages them for each level of x (eg average of predictions for the men in the data) “variable = x” generates predictions for counterfactual datasets in which x is set to different levels (eg average prediction “if everybody was a man”)

🚨 SynthNet is out 🚨 Researchers propose new constructs and measures faster than anyone can track. We (@anniria.bsky.social @ruben.the100.ci) built a search engine to check what already exists and help identify redundancies; indexing 74,000 scales from ~31,500 instruments in APA PsycTests. 🧵1/3

Follow up #rstats pop quiz: > x <- matrix(0, nrow = 10, ncol = 2) > x[cbind(c(4, 3), c(2, 1))] <- 1:2 > x What is printed out?

I'd say I need to get a hobby... but this is my hobby! 💙🩵

@profanity.accountant alright let's see it..

Not wanting to disappoint you so much that it lies is the last quality I want in a computer.

#rstats It is with profound sadness I heard that my long-time friend and colleague, John Fox passed away this week. He was the author of {car}, {effects}, {Rcmdr}, ... and numerous influential books. I will miss him greatly. www.john-fox.ca

Build a an example for next week's class of using #shiny in @quarto.org with results from a #brms mixture model. I think it turned out pretty cool 😎

In our new tutorial, @meanwhileina.bsky.social and I walk you through meaningful hypothesis testing using a less known form of calculating Bayes Factor. osf.io/preprints/ps... People transitioning to Bayes often wonder how to decide if an effect is meaningful or not. Here you go... 1/🧵

I will give another talk about mediation analysis and I have finally figured out what I want to say about "design-based" solutions: - longitudinal data can help rule out certain types of confounding - if you sequentially intervene on X and M, you need assumptions to piece the estimates together >

I believe eigen fly. I believe eigen touch the sky.

Some thoughts on empirical distributions of z-scores statmodeling.stat.columbia.edu/2025/11/24/s...