@petermacp @stephenjwild I can't help with R coding but, in STATA, I think I would use an IVhet model on double arcsine transformed data. The inputs would be the point estimate and bounds of confidence interval for each study (adjusted or not) https://t.c
1,739 followers
2,536 followers
RT @tomayates: For meta-analysis of prevalence, I have tended to advocate double-arcsine transformations, as per Barendregt (https://t.co/L…
1,739 followers
For meta-analysis of prevalence, I have tended to advocate double-arcsine transformations, as per Barendregt (https://t.co/LF9auURHFy) Newer papers suggest the back transformation might be unstable if skewed distribution of sample sizes, e.g. https://t.co
1,739 followers
@h_esdaile @Stata Thanks! What approach are you taking to meta analysing prevalence? I've previously used a Freeman Tukey transformation, as per - https://t.co/LF9auURHFy. Looks like Epigear have a package, but not sure what version of STATA is needed -