OP feels needlessly condescending, but I'd also say both of those tasks are things I'd have expected to be able to solve with a quick google search a decade ago.
It's a bit frustrating that we seem to have reinvented the wheel in this regard, except now it's stochastic and comes with usage limits.
"This decision is too important, complex and urgent to try any of that 'evidence-based' nonsense" certainly is an all-time classic in the public policy world 🥲
Yes, generally speaking, there's a lot of situations where marginalization can provide performance gains or circumvent sampler limitations (e.g. discrete parameters) but the design choice in brms is providing maximum flexibility by procedurally constructing the full joint likelihood when possible
Hah, reading this just as I'm waiting for tinytex to finish installing on my new laptop. As someone who is working from a somewhat unstable internet connection at the moment, every little bit of improvement is very much appreciated!
If you have time/compute you can spare on the problem, perhaps multiple imputation would be an option? (brms supports fitting over mice-imputed datasets via brm_multiple)
Given the correlation between age and income, one probably meets less haired people in general :)
Nice clear presentation of commonplace meta-analysis failure mode: Pooling coefficients that mean different things, because original models had different adjustment sets. A coefficient gets its meaning from the whole model, not just from the predictor variable it multiplies.
🧵1/ Our first meta-science paper (with 350+ coauthors) is published today in Nature. It presents one of the largest-ever reproducibility projects in economics & political science.
Here’s what we found 👇
I blogged about data model and likelihood being different things and importance of using these terms correctly (with some excerpts from the forthcoming Bayesian Workflow book, but we talk about these much more in the book)
New short paper forthcoming in Statistics & Probability Letters:
An objective non-local prior for skew-symmetric models.
arxiv.org/abs/2603.08285
This paper develops a Moment-Objective Minimum-Discrepancy (MOOMIN) Prior for testing symmetry against skew-symmetric alternatives.
Super excited to see that somebody finally wrote about this 🥳
If you do research about research, you really need causal thinking (just like when you do research about anything else really).
I'd say the fact you can immediately question the appropriateness of the visualization speaks for its usefulness :)
And even though it's not quite publication ready, it still seems like a very good first choice during exploratory analyses
Episode 2 of my trilogy on building replication packages for social scientists is now out! i4replication.org/a-researcher... 😉
Interesting take! Feels intuitively correct, but I think it's something that probably could be tested empirically.
So...my undergrad thesis student is doing a quality analysis of studies found in meta-analyses. She identified a few and we contacted the authors to request their effect sizes and other variables for the studies in their papers.
Here's what happened:
scientiapsychiatrica.com/index.php/Sc...
Have you seen this gem of a resource for #DataViz with R?!
The AI Summit is today! We look forward to celebrating the opening of ELLIS Institute Finland and hearing from stellar speakers in AI research, business and government. Tune in to the livestream areena.yle.fi/1-76514739 starting at 10:00 EET. Check out the program here: aisummit.fi
@ellis.eu
First page of Opinion piece: "Conceptual and methodological flaws undermine claims of a link between the gut microbiome and autism"
The link between the gut #microbiome and autism is not backed by science, researchers say.
Read the full opinion piece in @cp-neuron.bsky.social: spkl.io/63322AbxpA
@wiringthebrain.bsky.social, @statsepi.bsky.social, & @deevybee.bsky.social
Did we ever settle whether masks work against COVID19? It seemed like a very important scientific question to determine whether that particular cluster of proteins follows the same laws of physics that others do 🙃
I had a lovely catch up chat with @jamesheathers.bsky.social at the weekend. I am thoroughly inspired by the work he and team are doing at the Medical Evidence Project. They act on tip offs for fraudulent and/or inaccurate science used to shape medical practice. Know any? medicalevidenceproject.org
I definitely think the local experience can vary a lot, certainly across native vs non-native English speaking crowds. No one in my current environment is a native speaker and it's quite normal to see open ChatGPT tabs being used to help translate/rephrase/check grammar/etc.
Very cool (from Ehm-Gneiting-Jordan-Krüger, JRSSB 2016): for mean estimation, all consistent scoring rules can be obtained as conic combinations of 'extremal' consistent scoring rules, with an explicit structure. Similar results hold for quantiles (and perhaps other tasks as well!).
I appreciate @bmj.com follows a formal process, but just how much evidence do they need before adding an Expression of Concern.
Numerous PubPeer comments for stem cell for heart disease paper - which had huge media attention hailing it as a medical breakthrough.
pubpeer.com/publications...
I am excited to share that I am looking for a #postdoc with an interest in causal machine learning to join my lab at TU Dortmund University and RC Trust. #hiring #causality #ML #AI
📅 03.12.25
Group: rc-trust.ai/groups/causa...
Details: tinyurl.com/4622p6at
Multiverse analysis! 👁️👁️
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
The detectCores() apocalypse is creeping up on us 👻🐛
As more people are getting access to 128+ CPU cores, code spinning up parallel cluster with detectCores() workers fails - not enough #RStats connections available
Friends, do *not* default to detectCores(), bc www.jottr.org/2022/12/05/a...
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/)
A "methods primer" article in the journal "BMJ Medicine", titled "Factors associated with: problems of using exploratory multivariable regression to identify causal risk factors"
We wrote an article explaining why you shouldn't put several variables into a regression model and report which are statistically significant - even as exploratory research. bmjmedicine.bmj.com/content/4/1/.... How did we do?
