iiinova — where your digital self keeps evolving

Tim Feeney

@tfeend.bsky.social

1.2K followers 302 following 403 posts

Scientist, Physician, Husband, Dad, Editor @bmj.com, Nerd/Geek since before it was trendy. Epi/Biostat, Causal Inference, Epistemology. #episky thefeeney.netlify.app fediscience.org/@tfeend

Posts Replies Media

Tim Feeney @tfeend.bsky.social · 24/08/2026

Chandelier of stomachs

001

Tim Feeney @tfeend.bsky.social · 26/07/2026

speed without compromise! well, maybe just a little compromise.

000

Tim Feeney @tfeend.bsky.social · 12/03/2026

turns out these are not what you might expect.

120

Tim Feeney @tfeend.bsky.social · 23/11/2025

made me think of this

110

Tim Feeney @tfeend.bsky.social · 03/09/2025

Out this morning, TARGET a guideline to help get target trial emulation studies to report a standard set of items #prc10 @peerreviewcongress.bsky.social @bmj.com @societyforepi.bsky.social

2113

Tim Feeney @tfeend.bsky.social · 21/05/2025

That NEJM published this by these two clowns should make the journal a pariah among scientists. I don’t care what their official position is, they are liars and grifters and the worst of the medical community. Shame on the NEJM.

050

Tim Feeney @tfeend.bsky.social · 26/03/2025

Include information about your DAG somewhere in the publication to let researchers and editors see what you were assuming. If you're feeling cheeky, include code that helped you assemble the DAG (I'm looking at you Dagitty) to make it easier on your readers to recreate the DAG for their use.

100

Tim Feeney @tfeend.bsky.social · 26/03/2025

Next is to identify all the variables that directly cause the exposure, outcome or selection into your cohort of interest. These is ideally led by subject matter experts and this will likely benefit from discussion and iteration. Next identify the relationships between variables.

DAG without arrows but variables that might have an impact on outcome and exposure

DAG with filled in relationships between all the variables

110

Tim Feeney @tfeend.bsky.social · 26/03/2025

DAGs can be useful for identifying all sorts of biases, such as confounding, collider bias, adjustment for mediators (overadjustment), and areas where a sensitivity analysis may be needed. DAGs can help determine what variables may be needed which analyses may be challenging due to intractable bias.