[R-meta] Metafor for Multivariate Analysis Questions - help please :)

[Michael Dewey]

Dear Ayesha

Comment in-line. I am not going to attempt to clarify how to code for 
the multivariate model, there are people on this list who are much more 
expert than me. You might help by telling us what you think might be the 
answer. Note too that your mailer is set to post in HTML which mangles 
post on this plain text list.

See more below.

On 22/09/2021 10:13, Ayesha Sadozai wrote:
> Hi All,
> 
> I hope this email finds you well. My name is Ayesha Sadozai and am a PhD student at the University of Sydney, Australia. I am reaching out as I am hoping to get more guidance/assistance for my PhD project for which I am hoping to use Metafor for. My research interest is in the area of paediatric neuropsychology. I am in the midst of conducting a meta-analysis.
> 
> 
> 
> My research looks at seven paediatric executive function domains (working memory, response inhibition, attention, set switching, etc.) across a range of neurodevelopmental disorders in 78 studies. These disorders include ADHD, Autism, Tourettes, FASD and Learning disorders. My data looks at executive function (EF) measures that load on the 7 areas and these measures have a varied quantitative quality i.e., for one study I may have 5 EF outcomes that are in milliseconds, reaction time and a general number out of 100. (I must add some papers have 12 to 20 executive function variables I have captured). These studies also have 1 control group and this control is compared to more than one neurodevelopmental disorder so for example I may have a study looking at EF domains in children with ADHD and ASD with typical controls.
> 
> 
> 
> I started off using escalc(measure="SMD"….etc) then used another function suggested by Wolfgang such as rma.mv(effectsize, var, etc… but am unsure how I organise the factors into my coding. Any suggestions?
> 
> 
> 
> In CMA, I can outline effect direction to conceptualise the direction the means are taking (lower numbers in some EF measure do not indicate a bad performance in some cases) – how can the metafor package account for this for my data? Is there a transform function that need to be written or if I use effect direction how do I incorporate it into my coding?
> 

The simplest thing to do here is for you to reverse them. So make the 
positive SMD negative or vice versa. You do not need to modify the 
standard errors. Best to document what you did in case you come back and 
forget what you did and flip them back again, a mistake it is easy to make.

Michael

> 
> 
> Apologies if my questions are simple, I am new to this world and attempting to learn it all during a lockdown. 😊
> 
> Kind Regards,
> Ayesha Sadozai | PhD Candidate
> Autism Clinic for Translational Research, Brain and Mind Centre
> THE UNIVERSITY OF SYDNEY
> 94 Mallett St<https://protect-au.mimecast.com/s/n7WuC0YKPvipGkj0SwhSEc?domain=google.com> | The University of Sydney | NSW | 2006
> E asad6836 using uni.sydney.edu.au<mailto:marilena.demayo using sydney.edu.au>  | W sydney.edu.au<https://protect-au.mimecast.com/s/8qnyCgZ0N1iDAq85toeIeb?domain=sydney.edu.au/>   | P 02) 9351 0507 | Wed-Fri only
> 
> 
> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
> 
> 

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