[R-meta] Expanding range paper and follow-up times

Luke Martinez m@rt|nez|ukerm @end|ng |rom gm@||@com
Mon Jan 10 18:19:45 CET 2022

Dear James,

I recently came across your "expanding range" paper. Thank you for
this valuable contribution.

I'm interested in applying a 3-level (CHE) model to a set of
pre-post-control studies I'm meta-analyzing. But I have encountered a
logical problem.

In your paper, you indicate that your demonstration involves *"studies
with a broad variety of ... follow-up times"*.

My logical problem is with applying the CHE model to studies with
"follow-up times".

In such studies, pre-tests effect sizes (before treatment X
introduction) logically don't contain/denote any treatment X effect.
But in your fitted CHE models, it seems that you mix such
non-treatment effect sizes (from pre-tests) with treatment effect
sizes (from post-tests) by not interacting a time indicator with your
focal categorical moderator (e.g., `dv_cat`).

I'd appreciate it if you could help me understand if/when this may or
maynot be OK?

Thank you very much,

More information about the R-sig-meta-analysis mailing list