[R] R package for meta-analysis from z scores

[Michael Dewey]

Dear James

Your question really boil down to whether you can estimate tau^2, the 
between study variance of the effect sizes, if you only have p-values. 
As far as I can see the answer has to be no.

Michael

On 16/01/2020 13:10, james poweraid wrote:
> Hello,
> 
> I have a set of Z scores from an N=2 studies and I need to run a
> meta-analysis across the two Z scores for many N variables. I do not have
> effect sizes and SEs. I realize there are many different meta analysis
> packages in R, but I only have Z scores and it seems to me this is
> limiting. I am using the metap package because this very conveniently
> accepts directly p values which I can get from my Z scores. I have applied
> what is called in metap package the sumlog Fisher’s method Chi square (2
> df).
> 
> I have three questions:
> 
>     1. Is this the same as a fixed effect meta-analysis without weights?
>     2. Is there any way to do a random effects meta-analysis starting only
>     with Z scores?
>     3. Is there a way to get the I2 heterogeneity across studies from this
>     or from other packages? It looks like I need the estimates for this. Is
>     there another package that can easily get this using as input Z scores or P
>     values?
> 
> Data
> 
>      library(metap)
> 
>      zscores = cbind(c(4, 2, 0.1),c(5, 2.5, 0.1))
> 
>      pvalues = apply(zscores, 1:2, function(x)  2*(pnorm( abs(x) ,
> lower.tail=F)))
> 
>      meta = apply(pvalues, MARGIN=1, FUN=sumlog)
> 
> 
> 
> Thank you much
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 

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