[R-meta] Dear Wolfgang

[Lee, Ju]

Dear Wolfgang,

I had additional question about using glmulti for selecting best meta-regression models.

A dataset I am running a model selection has 2 continuous and 2 categorical variables , for example.
I’ve been running the following code formats:

rma.glmulti <- function(formula, data, random, ...)
  rma.mv(formula, VCV, data=data,  random=random, method="REML", ...)

best.mod <- glmulti(LRR ~ var 1 + var 2+ var 3 + var 4
                               random=list(~ 1|study_ID, ~ 1|ID),
                               struct="DIAG”,
                               data=lf,
                               level=1, fitfunction=rma.glmulti, crit="aicc")

where VCV is the variance-covariance matrix.
Var 1 &2 are continuous and var 3 &4 are categorical.
Study_ID is the unique pulication ID.
ID is the unique effect size ID.

It was my understanding that you need to specify the inner structure of your random effect list (ex. random=list(~ var3|study_ID, ~ var3|ID)) when your moderator is categorical.

My questions are:

  1.  How do you specify inner random effect when you have multiple categorical moderators in your models? (only testing the main effect)
  2.  How do you incorporate this to your model selection procedure using glmulti?
  3.  OR would the random effect structure specified above (random=list(~ 1|study_ID, ~ 1|ID)) suffice?


Thank you very much.
Best,
Juhyung

Juhyung Lee
Postdoctoral Fellow
Marine Science Center, Northeastern University
430 Nahant Rd, Nahant, MA 01908, USA
Phone: +1(650)285-7614


	[[alternative HTML version deleted]]