Hi everyone,
I have a "tricky" dataset with two factors but is unbalanced and sample size small than 5. And this data is not a normal distribution but a bimodal. So I think ANOVA is not very correct in this. I searched and found the Scheirer Ray Hare test can be used to test two-way nonparametric data. But I also see" It has been suggested that the observations should be balanced and that each cell in the interaction should have at least five observations ".So is my dataset strictly forbidden to use the Scheirer Ray Hare test？
If so, what analysis method should I use instead? Looking for a long time... so sad...
Thanks in advance!
Mengying
At the time of writing, if you are using the the rcompanion:: scheirerRayHare function in R: If the design is unbalanced, the order of the variables in the model will matter. This is because the function calls lm() and anova() that defaults to type-I sum of squares. The function could be modified to use car::Anova instead for type II sum of squares.
I don't know the sums of squares method the original authors of the test suggested.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size