I'm testing the effect of denervation on muscle fibers' area and due to skewed distribution I understand the median is a better estimate for this parameter in each subject. Any suggestion on how can I then move to obtain group data and to compare between groups using the median muscle fiber's area of each subject?
If you are using medians, that indicates your data are not normally distributed. In this situation, you should use a Kruskal-Wallis test; it is the nonparametric equivalent of a 1-way ANOVA. A Bonferroni-adjusted Mann-Whitney test is the appropriate post-hoc test if one is needed.
An alternative to using the Kruskal-Wallis test is to see if you can transform the non-normally distributed data to data that is normally distributed. Log transformations are good to try but you might try square roots, et cetera. If after transformation the data are normally distributed, you can use a 1-way ANOVA on the transformed data.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts