Hi all,
Based on several simulation studies and opinions, I've prepared a decision table for testing the averages of 3+ independent groups.
Yes, there are "rules of thumb", and these are no good rules.
You should understand the properties of your data and select an appropriate probability model. Withion your statistical model (of which the probability model is one part; the other part is the functional relationship between your predictors and the response) you would identify a meaningful hypothesis to test. You should know why you what to reject that hypothesis and what interpretation it allows you if you can reject it.
This is extremely difficult in the case of KW. Can you clearly state what hypothesis is tested and what rejecting this hypothesis means? And is this something useful to you (scientifically)?
Steve s.
Thank you. I agree with you. Found different opinions and results, so it may be better to plan it ahead.Jochen Wilhelm Danke schoen. I'm building a Work Instruction file for a clinical statistician. So it is not intended for a specific hypothesis testing.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Modeling
- Analyses of Variance