two images of the human body's circulatory system. One of them with good cable management

The human circulatory system, before and after proper cable management.

If you're still dichotomising continuous variables, stop.

Dichotomising your exposure &/or outcome wastes power.

Dichotomising confounders introduces residual confounding.

If a variable has non-linear effects, use splines!

If you want 'clinically relevant' threshold effects, use marginalisation.

It used to be that you knew you were getting old when the registrars (residents in US) started to look like interns. Now it’s the consultants (attendings) that look like interns!!

I think policies that support editors to desk reject more papers could free up capacity

Dichotomise continuous variable - reject

Stepwise variable selection - reject

Table 2 fallacy - reject

Etc

If widely applied, this could drive up standards

Sir, I like the cut of your jib! Bravo!

"No, I am your father"

The Bayesian Workflow book is coming!
statmodeling.stat.columbia.edu/2026/04/16/t...

Congratulations to @jdwilko.bsky.social for the citation of INSPECT-SR in testimony to the US House science oversight subcommittee and its mainstreaming of the scientific value of sleuthing.

a man in a suit and tie is applauding . ALT: a man in a suit and tie is applauding .

May I congratulate you for using “behoove” in a sentence!!

many such cases

My motivation apparatus from a pending deadline varies- it ranges from a feather brushing my cheek >24 hours out, to a sledgehammer <24 hours out. There is no in between.

It’s all lies, Jack! Don’t listen to them!

Can you explain your rationale for this? I’d like to understand your reasoning behind a preference for causality in observational vs RCT design. I don’t use the DiD estimator, so may be missing something!

I may miss the econ nuances with methodology, but estimators don’t have anything to do with causal identification. That occurs prior to the statistical estimand, which is followed by the estimator.

My vote would be on whether identification assumptions are tested through simulations/other methods.

More than Sharknado!?! It’s essentially a documentary…

Do you have a recipe? My son LOVES them so much after a trip to München.

This is one of my neurospicy screening questions! If you get a 15 minute lecture on “insert favourite dinosaur here”, welcome to the club!

Here’s mine, because I like maths and physics and science and rubiks cubes and coding and staying inside and…

NCHDs look at me with two heads when I suggest they shouldn't litter their paper with the term "real world", mainly because their bosses train them to do exactly that.

I don’t know. Finishers tend to be bad…

And we ALWAYS like our matrices positive definite!

a shirtless man in a cowboy hat and a fur coat is dancing in a hallway with other people . ALT: a shirtless man in a cowboy hat and a fur coat is dancing in a hallway with other people .

The Pitt and Scrubs are still the most accurate medical shows! And maybe Greys Anatomy cause we all SEXY!!!!

But in the era of U/S, this isn’t a big hurdle to overcome. Or are they worried about limb ischaemia?

Pretty good correlation bw PiCCO & PAC (~90%). Less invasive.

Agree with most that PACs aren’t used outside the CICU and even then, their use is decreasing with low risk surgery (PUMA pilot).

In the era of echo & transpulm thermodilution, can the risk/cost be justified?

Very interesting distributions from a RCT of antibiotic. Bimodal peaking at 0 and ~7, while also truncated at 0

What distribution would fit this best?

Added a statistical analysis tab to explore Bayesian partial pooling visualization (cc @rmcelreath.bsky.social). Added some model explanations too.

Data viz heavily inspired by @benmoran.bsky.social's beautiful bayesfoRest package