Hi,
I have some gas-production data that I want to analyse, the go-to method is a repeated-measures ANOVA, except that my data is nonparametric (although it meets sphericity conditions).
I was thinking Friedmans test, but since my data is based on a replicated block design this isn't an option.
I've tried log10 and sqrt normalisations with no improvement, so I'm at a bit of a loss as to what to do.
Thanks in advance.
Hi Amanda,
I tried the cube transformation, no luck (p=0.002), only marginally higher than sqrt transformation. Using the 'car' package isn't viable as apparently it's been deprecated, but I'm sure there must be another package that will do it.
If your data has a non-normal random component and/or heteroscedasticity, you should take a look at Generalized Linear multilevel Models. For example, if your data is continuous, has a lower boundary of zero and residual variance increases with the mean, a Gamma linear model would be a good choice. Including random effects to take care of repeated measures is straight forward.
https://www.researchgate.net/project/Book-New-statistics-for-the-design-researcher
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA