[R-meta] Question about meta-regression with multiple treatment studies

[Viechtbauer, Wolfgang (SP)]

Hi André,

1) You need to install the most recent version of metafor (2.1-0).

2) I haven't tried this. I assume the issue/problem is how to avoid models that only include the second-degree term but not the linear one.

Best,
Wolfgang

-----Original Message-----
From: Mensching, Andre [mailto:andre.mensching using uni-goettingen.de] 
Sent: Monday, 20 May, 2019 22:45
To: Viechtbauer, Wolfgang (SP); r-sig-meta-analysis using r-project.org
Subject: AW: Question about meta-regression with multiple treatment studies

Good evening!

@ Wolfgang: First of all many thanks for your great help! You have already helped me a lot.

@ all: But I have two more questions:

1.)
I just tried to use rma.mv in combination with MuMIn for some automated model selection. I used the hints from Wolfang's homepage 
(http://www.metafor-project.org/doku.php/tips:model_selection_with_glmulti_and_mumin). Unfortunately it fails when executing the eval(metafor:::.MuMIn) command. There I get the following error message:

> library("metafor")
> library("MuMIn")
> eval(metafor:::.MuMIn)
Error in get(name, envir = asNamespace(pkg), inherits = FALSE) : 
  Object '.MuMIn' not found

I assume that .MuMIn should actually be included in the metafor package, but apparently it's not. Or what could I have done wrong?

2.)
Does anyone have experience with using second degree polynomials in automated variable selection and an idea how to do this with rma.mv?
So far I have only managed glmulti with the consideration of simple interactions.

Again thank you very much in advance! I am looking forward to an answer!

Best regards,
André