[R-meta] metamean

[Viechtbauer, Wolfgang (NP)]

The message is pretty clear: Studies where the SD is 0 are omitted.

Compare:

metamean(mean=c(2,3,4), sd=c(1,1,1), n=c(50,50,50), comb.random=TRUE, comb.fixed=FALSE)
metamean(mean=c(2,3,4), sd=c(1,1,0), n=c(50,50,50), comb.random=TRUE, comb.fixed=FALSE)

In the second case, the third study is omitted because its SD is 0.

So, I think you need to tripple check your data.

Best,
Wolfgang

>-----Original Message-----
>From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-project.org] On
>Behalf Of mark walton via R-sig-meta-analysis
>Sent: Friday, 10 March, 2023 15:00
>To: r-sig-meta-analysis using r-project.org
>Cc: mark walton
>Subject: [R-meta] metamean
>
>Dear Michael
>
>Apologies, I thought i had started a new thread.
>
>I have checked the SDs and omitted any that are zero, so I don’t think that is
>the problem. I am more worried by the warning “Studies with non-positive standard
>deviation get no weight in meta-analysis” I assume this signifies that means with
>a standard deviations of greater value than the mean are ignored (eg mean ± SD =
>10 ± 14). Is that correct?
>
>Thanks
>Mark
>
>On Friday, 10 March 2023 at 14:24:55 CET, Michael Dewey <lists using dewey.myzen.co.uk>
>wrote:
>
>Dear Mark
>
>First of all please do not hijack an old thread but start a new one as
>your e-mail (and my reply) appear to be about marginal means in the archive.
>
>Have you checked the standard deviations? I assume some are zero which
>explains both your messages. It is likely the package then drops them.
>
>Michael
>
>On 10/03/2023 11:46, mark walton via R-sig-meta-analysis wrote:
>>  Dear All
>> I wonder if some one can help.
>> I am using metamean to determine with of my habitats have the most
>microplastics, I have collected 400+ papers with means, SD and sample
>number.Using the command:-metamean(MP_global, mean=meanMP_per_m2, na.rm=TRUE,
>sd=MP_SD, na.rm=TRUE,n=Sample_no, na.rm=TRUE ,byvar=Ecosystem,
>na.rm=TRUE,comb.random=TRUE,comb.fixed=FALSE,verbose=TRUE,
>control=list(maxiter=1000))
>>
>> I get some repeated warnings: Warning: Studies with non-positive standard
>deviation get no weight in meta-analysis.Warning: Ratio of largest to smallest
>sampling variance extremely large. May not be able to obtain stable results.
>>
>> But the model converges and I get means and confidence intervals, and
>significant differences between subgroups
>> However when I estimate my own mean values for each habitat weighted by sample
>number (below) the values are very different for some, and I am worried that the
>metamean results are not a true reflection of the data.
>> ddply(MP_global, .(Ecosystem),  function(x)
>data.frame(wret=weighted.mean(x$meanMP_per_m2, x$Sample_no)))
>>
>> Can anyone shed some light? Is it because metamean is ignoring some values?
>> Thanks very muchMark