[R] using alternative models in glmulti

Midgley, Meghan Grace mmidgley at indiana.edu
Tue Sep 11 23:41:18 CEST 2012


All,

I am working on a multiple-regression meta-analysis and have too many 
alternative models to fit by hand.  I am using the "metafor" package in 
R, which generates AIC scores among other metrics.  I'm using a simple 
formula to define these models.  For example,
rma(Effect_size,variance, mods=~Myco_type + N.type +total, 
method="ML")->mod  where Effect_size is the mean response for each 
experiment, variance is the SE around the mean, mods are the variables, 
and the method used to fit the AIC score is maximum likelihood 
estimation.  I thought I might be able to use the function glmulti(mod, 
level=1) to rank all alternative models using AICc scores.  If I fit a 
glm in in this way (e.g. glm(Effect_size~Myco_type+N.type+total)->mod), 
the glmulti function works beautifully.  Is it possible to extract AIC 
scores from the rma models as well?  Is there another package that 
might assimilate AIC scores from many alternative models more easily?

Thank you for any insight,

Meghan



More information about the R-help mailing list