I believe you're right, thank you for your work!

My thinking is that with raw data from included studies I should model directly in multi-level meta. E.g., time (pre- & post-) and condition (A, B) as fixed effects and an RE structure that accounts for study and participant. Apologies for the vagueness, I think this is what you did with the OR data

Hi Solomon, thank you for the awesome post. Would this approach also apply to multi-level meta-analysis? If I'm understanding correctly, instead of calculating a standardized mean change with correlation, you could meta-analyse the raw pre/post data in a multilevel structure?

Hi James, thanks for the post. Very cool visualization. Is it okay if I take your code and adapt it to some data I'm working with?