[R-meta] Applying Hedges' nesting corrections under 3 scenarios

[Yuhang Hu]

Hello All,

I'm following up on the following thread on applying Hedges' corrections (
https://doi.org/10.3102/1076998606298043) to effect sizes and their SEs
from nested designs (
https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2023-November/004959.html
).

I have encountered 3 situations in applying Hedges' corrections where
nested studies are:

1- Partially used in meta-analysis (e.g., one treatment including 2 classes
[out of the total two treatments and 8 classes] used in meta-analysis).
2- Longitudinal (i.e., subjects were pre- and post-tested).
3- Hit with attrition across the time points (e.g., Nt and/or Nc dropped
from pre- to post-tests).

Correspondingly:
Q1: For situation #1, what Ns (total, as well as Nt) and what m (number of
classes) from such studies must be used?

Q2: For situation #2, should we increase or decrease an initial ICC
guesstimate if a nested study is additionally longitudinal?

Q3: For situation #3, what Ns (total, as well as Nt and Nc) from such
studies must be used?

Thanks,
Yuhang

	[[alternative HTML version deleted]]