Dear RG members,
I am conducting a meta-analysis to estimate pooled prevalence. However, the variance for the study moves towards zero and its weight is overestimated. To avoid this, I read that logit transformation of prevalence estimate is important. But, I couldn't get a command to perform this. Please provide me the steps and command to do this on STATA software.
Your answers are highly appreciated!
You have mentioned log and logit, and they are different transformations. The logit seems more useful. It's log(p/(1-p)) where p is the prevalence and the log is by convention the natural log.
the logit transformation is the best way to estimate the non-over weighted variance.
for statistical advanced the advanced on STATA software please contact me inbox.
Dear Daniel Bekele ,
Could you explain how this works. I am doing a metanalysis of incidence risk(prevalence) and would like to present my findings as logit transformation.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis