shit, if there were any certainty, it wouldn’t be statistics 😂

This is fabulous.

We've all been there... 🫂

Data is available upon request … but only in the form of one hundred non sequential 3.5” floppy disks containing a segmented zip file, inside of which are confusingly named files that can only be opened using the 1996 application StatView for Apple Mackintosh. Enclose a cashiers check for postage.

🔥

goodnight, “tweeps” (“bleeps”? “bluseeps”? “skeeps”?) #statistics #AI

In defense of ORs!

I mean, yeah. This pretty much sums it up.

i was today years old when i learned that i'm amongst the minority of biostats folks who don't want to see ORs wiped from the face of the earth. but at least i don't advocate presenting data on the logit scale, cause that would be PURE MADNESS.

It’s all the algorithm this, the algorithm that. But honestly, the algorithm cannot and does not serve me the content I want. It’s really no better than the flip of a coin.

Just spending a Friday night exploring the intricacies of various smooths. Found this bit from Simon wood which gives a succinct overview of tensor product vs thin plate smooths. webhomes.maths.ed.ac.uk/~swood34/mgc...

I am beyond excited to announce that ggplot2 4.0.0 has just landed on CRAN. It's not every day we have a new major #ggplot2 release but it is a fitting 18 year birthday present for the package. Get an overview of the release in this blog post and be on the lookout for more in-depth posts #rstats

my judgement is clouded by frustration (see my previous post as to why). based on this color scale, are values <0 possible? this is a brms conditional_smooth plot from a gamma distribution, so < 0 shouldn't be possible. and all pp_check values are >0. Am i losing my mind? please send help, ASAP.

My r session just abruptly quit, and my workspace wasn’t saved. So *poof* went 3 days worth of brms objects. So yeah, that’s the kind of day I’m having.

Simon Wood, the GOAT of generalized additive models & creator of the mgcv #rstats package, has an Annual Review of Statistics essay on GAMs, available open access #statssky #mlsky www.annualreviews.org/content/jour...

Episode 10,371 in academic publishing’a bottomless chum bucket. Paper submitted in June. No way to contact journal directly, only generic publisher email. Response is that it’s been assigned to an editor and they are waiting for a response. And there’s no way on the site to withdraw the paper. 🤬

#rstats sorry 😅

This is a super post that touches on many cool topics including GAMs, location/scale models, and even splines (😍). 10/10 would recommend reading. blog.djnavarro.net/posts/2025-0...

Like this one, @mattansb.msbstats.info? 🤣

This is absolutely true. And also, as a statistician, sometimes I’m afraid of to tell myself *the very same thing*.

🎯 TARGET Guideline published 🎉 TARGET is a reporting guideline for observational studies of interventions that use the target trial framework. Over 3 years the @TARGETGuideline was rigorously developed and was co-published today in @jama.com & @bmj.com doi.org/10.1001/jama.2025.13350 #episky

Although maybe not @stephenjwild.bsky.social This is for sure *not* brain rot

Amazing set of educational resources (with references!!) for learning Bayesian analytic thinking. The one I wish I had when I was learning (and I’m still learning).

This could make #marginaleffects package even better than it already is!

I'd love feedback on this #marginaleffects WIP. In GLMs with many parameters, the marginaleffects 📦 can get a 5–20x speedup (and better SEs) by calling JAX for automatic differentiation. Try the instructions at this link and let me know if you run into trouble. Thanks! github.com/vincentarelb...

I dont know which is more disheartening—that there are doctors who peddle such mendacities, or that there are people who eagerly pay for it. www.trudiagnostic.com

priorsense 1.1.1, for prior and likelihood sensitivity analysis, is now in CRAN (n-kall.github.io/priorsense/) There is now an easier way to select which priors are power-scaled using the new prior tags feature in brms (2.22.11+), which allows focusing the analysis on specific priors. 1/3

Also, let’s teach more clinical researchers about the importance and inter-relatedness of standard errors, critical ratios, p-values, and CIs. Cause frankly, if I have to explain again why there’s a column entitled ‘SE’ in a regression table, I’m gonna… complaint to my stats friends again.

P-values, in and of themselves, are not problematic. In fact, often we really need thresholds to support decision making. This is not the same as worship of an arbitrary line as a substitute for a thoughtful, principled approach to an analytic question.

Do mildly irascible loaners with interest in A, B, C want to hang out with other irascible loaners with the same interests? Asking for a friend.

Reviewing a paper in which the authors use a Bayesian MLM on an -omics data set with 40k observations. They report ~48hrs needed to fit the models. The patience required for such an undertaking is *inconceivable* to me.

Ever stared at a table of regression coefficients & wondered what you're doing with your life? Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social) Preprint: doi.org/10.31234/osf... Website: j-rohrer.github.io/marginal-psy...

Preprint and the website by @dingdingpeng.the100.ci and @vincentab.bsky.social are great, and I got inspired to illustrate Bayesian workflow for model checking and comparison *before* model interpretation following their friendship importance example users.aalto.fi/~ave/casestu...

Data availability statement: suck it, do your own search. What fresh hell is this?

Even 10 years in, doing stats can still feel like a Sisyphean task. Except the hill is really steep. And it’s more of a U than a hill, so I can repeatedly get flattened from both sides.

Stayed up real late last night trying to figure out how I’m going to explain to a group of clinical researchers that their DV, although ‘numeric’ is probably best modeled as beta distributed. And I decided that I’m just gonna fake my own death, as that would involve less hassle/frustration.

The paper about splines in R that I wish I had known about when I was learning splines in R. bmcmedresmethodol.biomedcentral.com/articles/10....

Why are we not pushing these AI data centers to build clean energy and water infrastructure to support themselves??! You want a data center? Build the resources it needs plus 15% extra to support the community’s needs.

Once, just once, even for 5 minutes, I want the confidence of a reviewer who denigrates the MLM choices I’ve made but based on the comments is likely a clinician who has literally no idea what an MLM is or does.

Any Bayesians out there have favorite references/learning materials for regularizing priors in a multilevel framework? @paulbuerkner.com @f2harrell.bsky.social @chelseaparlett.bsky.social

@quantitude.bsky.social my math-inclined teenager has requested an episode on splines (theory, implementation, interpretation) as apparently my explanation is inadequate (surely it’s her, not me). Pretty please? It’ll either be a big victory for me or a public humiliation (which she would enjoy).