1/ I have a new favorite term: "unadjusted hypothesis-naïve exploratory linear regression"-- and it comes from @Ad_SotoMota, who @nicknorwitz and I are honored to be working with in this new project (news on that soon...)
I don't typically see the term as it's rarely employed...
I don't typically see the term as it's rarely employed...
2/ My layperson explanation is that this is explicitly letting the data speak for itself. Literally starting from scratch and seeing what patterns machine learning tells us are emerging. No adjusting the data at the beginning of the process which can bias the outcome (of course)
3/ I'm much more used to methodology sections describing several steps of adjustment selection by the researchers. This isn't to imply any wrongdoing by any means, but understandably, this can be problematic if one or more underlying assumptions for adjustments are incorrect...
4/ Using unadjusted hypothesis-naïve exploratory linear regression is very powerful as it can better identify the largest patterns worth of attention (or at least, at a minimum, what shouldn't be getting ignored).
5/ I'm thankful @Ad_SotoMota could bring his mathematical wizardry to our project to help meet this standard of transparency we love so much.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
