Skip to content
Prev 3746 / 5636 Next

[R-meta] Does QE statistic in rma.mv() account for random effects as well?

Lukasz Stasielowicz 
Dear Simon,

if you compare the output of a simple fixed effect model to a simple 
random effects model then you can see that the QE value is the same.

A reproducible example:

library(metafor)
rma(data=dat.bangertdrowns2004,yi,vi,mods=~length,method="FE")
rma(data=dat.bangertdrowns2004,yi,vi,mods=~length,method="REML")

"Test for Residual Heterogeneity:
QE(df = 44) = 96.2810, p-val < .0001"

In other words. Random effects do not influence the QE value.

You can also see it in the code that Wolfgang provides via github:
https://github.com/cran/metafor/blob/master/R/rma.mv.r (line 1500)


For some conceptual clarifications see also this paper:
Pastor, D. A., & Lazowski, R. A. (2017). On the multilevel nature of 
meta-analysis: A tutorial, comparison of software programs, and 
discussion of analytic choices. Multivariate Behavioral Research, 52(6), 
789?804. https://doi.org/10.1080/00273171.2017.1365684



Best,
Lukasz

Thread (2 messages)

Lukasz Stasielowicz Does QE statistic in rma.mv() account for random effects as well? Jan 21 Simon Harmel Does QE statistic in rma.mv() account for random effects as well? Jan 21