Traditional meta-regression analyses provide the pooling estimates assume independence of the estimates, an assumption that may be not true because the estimates for separate exposure levels depend on the same reference group in each study. Although generalized least square (glst) is an option for dealing with this issue, I'm not sure whether the CMA Software Version 3 supports this function.
No impossible with CMA.
As I know, The STATA software is easy to analysis dose-response MA.
Dear Po-Wen,
I experienced CMA being a easy tool for standard MA (on Windows computers), but as soon as you want to calculate analyses that are not very common you are stuck if you use CMA. I use the software environment R for my meta-analyses, because you can basically calculate whatever you want and there's a very helpful online community. Especially Wolfgang Viechtbauer ("metafor" package) and Guido Schwarzer ("meta" package). I contacted them when I experienced problems or difficulties and always got a helpful and friendly response.
I myself never used STATA, but I heard a lot of good things about it.
Cheers,
Christoph
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis