Some papers prefer to use median (IQR) instead of Mean (SD). Should I transform the median (IQR) to mean (SD) to include in the meta-analysis? Although some method is available in the literature, that it is not accurately possible to transform these values with each other. Which one is the right approach? Transform and include, or do not include?
Medians are used for skewed/kurtotic distributions, hence any method may be problematic.
Transforming them is reasonably accurate, so you can include them. however, you should also perform a sensitivity analysis and/or meta-regression to investigate whether this had a significant effect on the outcome if you do this.
Medians are used for skewed/kurtotic distributions, hence any method may be problematic.
Dear Ette Etuk ,
It was the first time I heard that Median is nominal, ie categorical. In this respect, it also seems reasonable. Thanks
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis