[R-sig-ME] metafor: estimate correlation between response variables (meta-analysis)
Viechtbauer Wolfgang (SP)
wolfgang.viechtbauer at maastrichtuniversity.nl
Thu Dec 15 10:29:36 CET 2016
Yes, making blup() and ranef() work with 'rma.mv' objects is on my to-do list.
> -----Original Message-----
> From: Gustaf Granath [mailto:gustaf.granath at gmail.com]
> Sent: Thursday, December 15, 2016 10:03
> To: r-sig-mixed-models at r-project.org
> Cc: Viechtbauer Wolfgang (SP)
> Subject: Re: metafor: estimate correlation between response variables
> Oh, that is embarrassing. Thanks for spotting my mistake. It also
> explains why I got more sensible results when I tried it on my real data.
> Short additional question. Is the plan to implement BLUPs for rma.mv?
> On 2016-12-15 09:50, Viechtbauer Wolfgang (SP) wrote:
> > After you fit the "res.with.covar" model, you have:
> > print(res.no.covar, digits=3)
> > which should be:
> > print(res.with.covar, digits=3)
> > which shows a correlation of -0.967.
> > I haven't really delved into the rest of the code, so I cannot say much
> about that.
> > Best,
> > Wolfgang
More information about the R-sig-mixed-models