if the meta-analysis is estimating dichotomous data, can we calculate the odds ratio and also calculate the risk difference?
also, is it normal for the odds ratio to be 0.5 but the risk difference is only 1%?
Suggest you see the appropriate definitions. this should answer your questions. Best wishes David Booth
Yes you can! If you use programs such as RevMan or CMA and insert data on number of study participants and number of events/non-events for each primary study, you can simply choose if your pooled effect results are displayed in absolute terms (risk difference) or relative terms (risk ratio or odds ratio).
I would not use the word normal (as the effect totally depends on what your question is and what the data says), but these figures are possible; a rather large relative effect and quite small absolute effect.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis