Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: st: Meta-Analysis of Proportions
From
Trelle Sven <[email protected]>
To
"[email protected]" <[email protected]>
Subject
RE: st: Meta-Analysis of Proportions
Date
Mon, 14 Oct 2013 12:32:39 +0000
An alternative to the Freeman-Tukey transformation would be a logit transformation. You can use the xtmelogit command with the binomial option. Do not forget to exponentiate the constant to get the pooled proportion. R using the meta-package (with the metaprop command) is probably the most extensive non-Bayesian meta-analysis command at the moment (several transformations including Freeman-Tukey and several approaches to estimate between-trial heterogeneity tau-square if metafor-package is also installed).
Best
Sven Trelle
Original message
------------------------------
Date: Fri, 11 Oct 2013 09:14:00 +0000
From: Chamberlain Jonviea Danielle <[email protected]>
Subject: st: Meta-Analysis of Proportions
Hi,
I'm interested in performing a meta-analysis of a proportion/rate (for example, I'd like to come up with an pooled estimate of the case-fatality rate).. I recently read that the Freeman-Tukey transformation should be used when using proportions. Is there a preferable way to meta-analyze proportions instead of using this transformation? If yes, what would that be and how would you perform the command in stata? Or, is it ok to not use a transformation? I attempted to use the Freeman-Tukey transformation and compared results with my meta-analysis of the non-transformed proportion and results were quite similar. Also, does anyone have any recommended texts on advanced analysis in stata that also explains the 'why' of why you'd choose one method over another?
Thanks for your help!
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/