[R-meta] Classic fail-safe N

[Michael Dewey]

Dear Naike

Without seeing the code used by CMA one can only speculate but I would 
have thought that small relative differences in the computation of the 
extreme tails of the normal distribution are probably responsible.

The value of z which CMA quotes corresponds to a p of approximately 10^{-88}

On 26/06/2017 13:57, Naike Wang wrote:
> Hi all,
> I was conducting a meta-analysis of single proportions(i.e. without a
> control group) using the metafor package. When I performed a classic
> fail-safe N, I noticed that the result (the number of missing studies that
> would bring p-value to the alpha, to be exact)was different than that I got
> in Comprehensive Meta-Analysis Version 2.0. I wonder why R and CMA got
> different results.
>
> *Below is the R code:*
> dat=read.table("Your working directory\\Example.csv",header=T,sep=",")
> transf.ies=escalc(xi=cases,ni=total,measure="PLO",data=dat) #I transform
> the data using the logit transformation first. In CMA, it also uses the
> logit transformation.
> transf.pes=rma(yi,vi,data=transf.ies,method="DL",weighted=TRUE) #Pooling
> individual effect sizes in the logit scale.
> ranktest(transf.pes) #Performing the fail-safe N.
>
> *Below are the results from R:*
> Fail-safe N Calculation Using the Rosenthal Approach
> Observed Significance Level: <.0001
> Target Significance Level:   0.05
> Fail-safe N: 8446
>
> *Below are the Classic fail-safe N results from CMA:*
> Z-value for observed studies 19.91594
> P-value for observed studies 0.00000
> Alpha 0.05000
> Tails 2.00000
> Z for alpha 1.95996
> Number of observed studies 58.00000
> Number of missing studies that would bring p-value to > alpha 5931.00000
>
> Notice that I got 8446 in R and 5931 in CMA.
>
> Can anyone shed some light on this discrepancy? Thank you!
>
> You can find my data set here: https://drive.google.com/open?id=
> 0B41wTxciaMqtTEJWZE9sX20wOXM
>
> Best,
> Naike
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>
> ---
> This email has been checked for viruses by AVG.
> http://www.avg.com
>
>

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