[R-meta] Multi-level model accounting for within-cluster correlation
Dr. Guido Schwarzer
gu|do@@chw@rzer @end|ng |rom un|k||n|k-|re|burg@de
Tue Jun 20 18:40:13 CEST 2023
Thank you for your very helpful comments!
The data structure is pretty basic - multiple estimates within clusters. The variable "study_id" is actually just a running index within clusters like the variable 'school' in the dataset dat.konstantopoulos2011 in R package metadat.
Accordingly, the working model using rho = 0.6 is probably the better one.
More information about the R-sig-meta-analysis