For my study, I will use a data set that includes 22 schools, 2 to 20 teachers in each school, and 4 students from each teachers' classroom. I would like to investigate the treatment effect at the student level where the schools randomly assigned to either treatment or control groups. Should I use HLM or MANOVA/MANCOVA for this analysis?

If I should use HLM, Is the number of students (4 in each classroom) sufficient for HLM?

If I should use MANOVA/MANCOVA, how can I justify not using HLM for my data which is nested?