Makes sense. A practical challenge seems to be, that to allow prediction, the spline basis functions need to be known, but this may not be readily available (in SAS)

Currently using SAS, and natural cubic (RCS); wasn't sure if you need to know specifics of the spline functions used in the background for fitting the line

are there examples of what the prediction equation would look like?

I loved this resource, thanks so much! I am currently working on a log-reg model with a var w/ a nonlinear relationship with my outcome... my spline variable has 3 knots (resulting in two beta coefficients in my model equation. I am wanting to use it for future predictions --

This is an excellent resource. Thanks! I had question regarding what the full derived prediction model equation would look like, when incorporating the splined variable with three knots (two coefficients). I will message you directly.

This post is great. I'd definitely consider citing/referencing it in later discussions/guides for research. Thanks for sharing!

Would there be a recording available, or content? Would love to see if able. I appreciate your perspectives for the feather ruffling stuff!

I'd love to get a "Quantitative GC Group" together sometime to meet/informally discuss topics in this area -- would be so cool @j9austin.bsky.social

And our spoken language/vernacular is often challenged by conversations in this area -- so recommendations for clarity or conversations is always welcome

Thank you 😀 -- there is SO much to try to cover, so we missed some interesting topics like causality, definitions of confounders/confounding, multivariable adjustment, bias, and so much more (DAGs!); def share feedback for future iterations @j9austin.bsky.social

👍👍appreciate it!