Multivariate meta-analysis

#1
Hi all,

I'm a graduate student in Biology and I'm preparing a PhD application at the department of Ecotoxicology of my university. The title of my subject will be:
'Species richness/density, pollution and climate change: a multivariate meta-analysis for West-European waters'.

I'm very new to the concept meta-analysis, but I have been reading about meta-analysis a few months now, and a had a question. What kind of statistical methods are most appropriate when performing a multivariate meta-analysis?

I'm planning on gathering large datasets for my research: spatiotemporal datasets on sea surface temperatures, pollution gradients and biodiversity (richness, eveness). These datasets will be subjected to statistical analyses to check for any correlations between these variables (they are all gathered for the same locations, and all the data will be mapped with GIS software). I'm assuming a software package such as MetaWin will do the trick.

But so my question is: what kind of statistic is appropriate in my case? How will I be able to calculate effect sizes and estimate correlations (Pearsons, Spearman)?

Most meta-analyses talk about the effect of variable x on variable y, but I'm comparing three or more variables with eachother.

Many thanks in advance,

Detlef
 

Mean Joe

TS Contributor
#2
I'm planning on gathering large datasets for my research
You'll be getting the raw data? If so, I don't think you need to do a meta analysis. Just combine the datasets.
Then, if necessary, you can introduce a (categorical) variable indicating that certain data come from different parts of Western Europe, and adjust for that variable in your analysis. This may not be necessary if you don't think there should be any differences between parts of Western Europe.