I had data with repeated measurement and nested design. Conventional ANOVA requires strict control on homogeneity of variance and repeated measurement ANOVA requires assumption of sphericity. Multi-level modeling allows greater flexibility as the covariance matrix could be modified and thus, I used Linear Mixed Model(LMM) option in SPSS and defined the repeated measurement matrix with AR(1).
I am confuse do I still have to keep the assumption of homogeneity of variance if I use LMM.
My dependent variable is a continuous scale data, can I use a Generalized Linear Mixed Model (GLMM) for analysis if homogeneity of variance is not met? However, my data is not binomial distributed.