A meta-analysis of case reports might include comparison of categorical variables with Chi-square tests and frequencies of means with t-tests. Moreover, meta-regression might be performed to identify predictors of effect estimates. However, is it proper to treat single case reports as studies? If yes, how can we treat studies that include more than one case report as within-study variability occurs under these circumstances?
Hi Vasileios,
Meta-regression, under its classical definition, requires to weigh each effect size data point (each study); therefore, variability values associated to each study is required. If variability within study is not available, based on how the studies were designed an option is to run the regression without weighing each data point: that will be classical regression (not meta-regression), won't take advantage of the variability to get more precise estimations, but might still provide a useful insight if values are not too spread.
Thank you, Diego. You made it clear. I really appreciate your thoughtful answer.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis