I am working with 11 leaf-traits. For Manova and other statistical analysis, i have log-normalized the data-set comprising leaf trait values. Then also some of the trait values are still skewed. suggestions in this context will be highly welcomed and appreciated.
Thanks.
Dear Mayank. Pavlos' suggestions will be useful, of course. But my point will be different: are you saying that you only have 11 observations? If that is the case, you are facing a small-sample problem and very few statistical tools are able to deal with this issue. Please clarify the size of your data: I can also interpret that you have a sequence of 11 treatments made to the same 1000 individuals (and then you have problems with 3 of those 11 "panels"), but please try clarify everything about your data. Well, assuming you have enough data, I would argue that the treatment may not only affect the mean of the outcome, but also higher moments of the outcome. Your model should try to account for this possibility. In a sentence: skewed distributions are against the maintained assumptions of some simple models, but not against models with weaker assumptions.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis