[R-meta] Modeling multi-end point and multi-treatment, longitudinality
Simon Harmel
@|m@h@rme| @end|ng |rom gm@||@com
Fri Jan 8 04:22:48 CET 2021
Dear All,
I'm conducting a meta-analysis where some studies (15 out of 50) have used
multiple outcomes.
Also, most studies have multiple treatments. Finally, most studies contain
2 post-tests (longitudinal designs).
(1) Given the type and design of the studies, I was wondering how much of
the complexity (i.e., multi-end points and multi-treatments,
longitudinality) described here can be modeled using perhaps metafor (I
appreciate suggested syntax)?
What format the data should be in so that your suggested model can be fit?
Thank you all,
Simon
p.s. I have computed a pre-post Cohen's d for each treatment group on each
time interval (e.g., pre to post-test_1), and then subtracted it from a
corresponding control group's Cohen's d computed in the same fashion. So,
the # of effect sizes in each study on each outcome equals the # of
treatment X number of post-tests.
[[alternative HTML version deleted]]
More information about the R-sig-meta-analysis
mailing list