3

I have a quick question about displaying robust estimates in a forest plot.

I am using rma.mv() from the metafor package to fit a multilevel random-effects model. For robust estimates, I use the conf_int() and coef_test() functions from the clubSandwich package. However, the forest() function from metafor only accepts objects of class rma, which are not produced by the functions for robust estimates.

Is there any way to use the forest() function to create a nice forest plot, but somehow integrate the robust overall estimate from the clubSandwich functions?

Appreciate any advice or alternative solutions!

fuestro
  • 45
  • 5

1 Answers1

1

You can use forest() to plot the individual estimates and then addpoly() to add the polygon. For the latter, you just need to pull out the pooled estimate and corresponding standard error from the object returned by coef_test().

Wolfgang
  • 2,810
  • 2
  • 15
  • 29