[R] Meta-analysis with zero values for mean and sd

[Viechtbauer Wolfgang (STAT)]

I know nothing about the biological background here, but I also wonder if the standard meta-analytic approach is meaningful here. I also see some confusion in terminology (these are not "count data"!). 

There appear to be only 3 observations per group in each of the studies -- the standard deviations in each of the studies are consequently very inaccurate estimates. A potential solution may be pooling SDs across studies.

While I do not know what was measured in each of the "studies", I suspect that it may have been "the same thing". Therefore, I would suggest that the outcome should be the raw/unstandardized mean difference (i.e., ControlMean - TreatedMean) and not the standardized one (i.e., (ControlMean - TreatedMean) / SD).

And Jim's comment is also spot-on -- does it make sense to compare those zeros with non-zeros?

Best,

Wolfgang

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
> On Behalf Of Jim Lemon
> Sent: Sunday, July 10, 2011 01:21
> To: Rosana Ferrero
> Cc: r-help at r-project.org
> Subject: Re: [R] Meta-analysis with zero values for mean and sd
> 
> On 07/09/2011 08:16 PM, Rosana Ferrero wrote:
> > Hi!
> > I want to do a meta-analysis with count data for treatement/control
> cases.
> > Mi problem is that I need to use zero values (an informative value) for
> the
> > mean and standard deviation for one of the treatement, but R has a
> > problem: "Studies
> > with zero values for sd.e or sd.c get no weight in meta-analysis". I can
> > agroup the case by Family (byvar=Family).
> 
> Hi Rosana,
> Your data indicates that in three studies, whatever was being measured
> had a value of zero for each of the three spaghetti worms in the
> treatment group. Maybe they were dead or something. You may be able to
> get away with substituting a very small value (0.1 or something) for the
> SD in these cases. However, I would look very carefully at what this
> means. If, say, the worms are dead, is the measurement between live and
> dead worms meaningful? Just a guess.
> 
> Jim
> 
> ______________________________________________
> R-help at r-project.org mailing list
> 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.