[R-meta] mismatch in outputs from meta-analysis and corresponding forest plot

[Michael Dewey]

Dear Joao

Nobody is going to be able to understand your posts if you keep posting 
in HTML as what we see is not what you sent. Please post in plain text. 
I do not use gmail but I am told you can post in plain text from it.

Michael

On 03/02/2020 17:59, Joao Afonso wrote:
> Dear all,
> 
> I am conducting a meta-analysis and came across an odd issue. I ran the
> model for assessing the moderator effect of a variable as per below
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> *subganal.lcmbi=rma(yi, vi, data=ies.logit.lcmbi, mods=~lcmbi,
> method="DL")pes.logit.lcm=rma(yi, vi, data=ies.logit.lcmbi,
> mods=~lcmbi=="Records", method="DL")pes.logit.records=rma(yi, vi,
> data=ies.logit.lcmbi, mods=~lcmbi=="Locomotion Scoring Method",
> method="DL")pes.subg.lcmbi=predict(subganal.lcmbi,transf=transf.ilogit)dat.samevar=data.frame(estimate=c((pes.logit.lcm$b)[1],
> (pes.logit.records$b)[1]),stderror=c((pes.logit.lcm$se)[1],
> (pes.logit.records$se)[1]),tau2=subganal.lcmbi$tau2)pes.logit.lcm=rma(estimate,
> sei=stderror,method="DL",data=dat.samevar)pes.lcmbi=predict(pes.logit.lcm,
> transf=transf.ilogit)print(pes.logit.lcm, digits=3) #display subgroup 1
> summary effect sizeprint(pes.logit.records, digits=3) #display subgroup 2
> summary effect sizeprint(subganal.lcmbi, digits=3) #display subgroup
> analysis resultsprint(pes.lcmbi, digits=3) #display recomputed summary
> effect size*
> 
> and got the following output
> 
> 
> *  pred ci.lb <http://ci.lb> ci.ub cr.lb <http://cr.lb> cr.ub  0.237 0.149
> 0.355 0.112 0.434*
> 
> However when plotting the corresponding forest plot I get the pooled at
> 0.27 (0.08 - 0.53).
> 
> Any idea of what could be producing this or what could I be doing wrong?
> 
> Many thanks,
> 
> Joao
> 
> 

-- 
Michael
http://www.dewey.myzen.co.uk/home.html