I know the standard approach is to enter the effect size in a meta-analysis (e.g: standard error) and let R do its thing. However, I am pooling standardized incidence ratios, and the event in question is rare. This means that just entering effect size actually leads to small inaccuracies in calculating confidence intervals. I would much prefer to calculate my own confidence intervals (letting me use Byar's assumption, which would get around the issue), and then enter these into a meta-analysis, weighting the studies with the inverse variance method. Is there any way of doing this in R?
Well R is open source so the answer is YES. However I would investigate the literature on sampling for rare events first before even thinking about your question. Google search will help you with this. Best wishes David Booth
Isaac,
you may write your own function that is based on parameters from other options. learn how to extract those parameters. This will allow you incorporate these extractions in your own function. Then you can apply the function for all of your meta-analyses/variables.
HTH
Holger
- Badges
- Science topic