Matt Parkes
banner
mattyjparkes.bsky.social
Matt Parkes
@mattyjparkes.bsky.social
Statistician. Interests: clinical trials methods, psychology, MSK medicine, rheumatology, oncology, chronic pain, adaptive trials, outcomes, digital health/wearables, Open Research, Stata, R, and code ligatures.

https://orcid.org/0000-0002-1574-9933
(This applies to clinical medicine, and arguably all science, too)
December 3, 2025 at 9:51 AM
Still cracks me up that not long ago everyone was like “Trials are too expensive”, and the solution to this is to do massive AI models with colossal piles of data. More expensive, more assumptions, weaker causality. 🍌🍌
November 27, 2025 at 7:54 AM
Also it’s way cheaper than pouring barrels of cash into building ultra-mega-giga-data-centers
November 27, 2025 at 7:43 AM
The problem is the cliff-edge dichotomous thinking, which I think can be exacerbated by the use of p-values, but you can still do this with CIs - it's not a property of either, more a problem with interpretation.
November 18, 2025 at 9:23 PM
I think Positron might do something similar, called ‘Open Collaboration’, but I’ve not tried that yet!
November 18, 2025 at 9:11 AM
For asynchronous editing, I think it’s a gap. If you’re doing synchronous edits, you can do a kludgey solution by opening the .Qmd in VS Code and firing up a ‘Live Share’ session and sharing the link with collaborators. Not ideal, but not bad.
November 18, 2025 at 9:07 AM
This dot gif?
when i left you i was but the learner but now i am the master!
ALT: when i left you i was but the learner but now i am the master!
media.tenor.com
November 11, 2025 at 8:48 PM