Non-parametric testing for linear mixed models (LMM)
I recently discovered that the longitudinal data that I want to use linear mixed modeling for is not normally distributed. Therefore, I need to use a non-parametric test with a robust maximum likelihood estimation.
Is it possible in JASP to conduct such a non-parametric test in LMM?
Comments
An LMM is a parametric analysis. So I think if you're going to do a non-parametric test, there's no sense in which it will be an LMM analysis. It will be an alternative to an LMM. Which alternative you use (if any) will depend on the details of your data.
R
For a non-normal dependent variable, depending on the distribution of the variable you can also use generalized LMM. This is parametric, but it assumes the outcome follows some distribution other than normal (e.g. there's Poisson for counts, inverse Gaussian and gamma distributions for skewed positive values, binomial for binary dependent variables, etc.)