[R-meta] Questions about Robust Variance Models using Meta-for

Emily Quinn quinnem at ohsu.edu
Tue Sep 26 23:36:07 CEST 2017

Dear Dr. Wolfgang Viechtbauer and other meta-analysists,

I hope that this email finds you well.
I was writing for your advice re: an issue related to the metafor R package.

Currently, I am estimating a robust variance model using metafor rma.uni and robust functions and I noticed that the estimations were different than the estimates when I ran the same model in STATA using the robumeta package and in R using Robumeta See:

STATA Robumeta:  hedges g= 0.3694, SE=0.1326, UL CI= 0.0939 - 0.6450
R Robumeta: hedges g=0.369, SE=0.1333, UL CI = 0.0939- 0.645
R metafor: hedges g = 0.307, SE 0.1149, UL 0.0736- 0.5479.

Specifically, I was wondering if there was any arguments that I can specify that would make the metafor estimates closer to/ more consistent with the STATA and R Robumeta estimates:

This is the syntax that I'm using to estimate the mode in metafor:
Emily.t2 = rma.uni(yi=E$g,vi=E$var_g)
Emily.t3 = robust(Emily.t2,E$author,adjust=T)

Thank you so much for your time and efforts in advance.

Emily D. Quinn

	[[alternative HTML version deleted]]

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