Hi, I am trying to do a meta-analysis. I want to investigate the moderating role of the country culture (measured by national cultural scores) on the effect sizes found in these countries. How many effect sizes will I need for each country at least?
To be more precise, I want to know if effect size of X on Y varies across countries. So from each country, how many effect sizes do I need at least?
I assume you want to run a subgroup analysis (or as some people call it, subgroup meta-analysis). There is not hard and fast rule for the smallest subgroup. Ideally, I like my smallest subgroup to have at least 5 studies but I have used as few as 3 before. Regardless, with that few studies, you will need to use a pooled variance component across subgroups.
Thanks Gordon. To be more precise, I want to know if effect size of X on Y varies across countries. So from each country, the least needed and legitimate effect size is three?
In meta analytic studies, the effect size becomes the dependent variable. Thus, the sample size (in this case, the number of studies compared) is dependent on the type of statistics or analyses you use for your analyses. The rules of thumb would be the same as any other study.
- Badges
- Science topic
- Similar topics
- Medicine
- Epi
- Age Groups
- Children