I have a dataset with multiple independent variables (time being one of them, so repeated measured) and multiple ordinal dependent variables that have 3 grades for each (like grade 2 for being diarrhea). I have 5 or 10 technical repeats (mouse model) for an experimental group but the mouse also gradually died out in the later time points so I also have missing data.
Does anyone have any suggestion on which test or model I should use to show whether a specific combination of those independent variables gives an outcome that is significantly different than the other's? I thought about MANOVA, but it seems to work only for continuous dependent variables, not ordinal.
Edit: more details for the dataset below
- Independent variables: topical treatment (no topical & topical A,B,C...), injection treatment (no injection & injection D,E,F...) [there're combination treatment groups of a topical + a injection], treatment time (in days)
- Dependent variables: body weight loss (in g), hair loss (grade I, II, III), phenotype G (grade I, II, III), phenotype H (grade I, II, III), phenotype I (grade I, II, III),...
- Questions I want to answer from the dataset:
The mice of which combination of treatments group have significantly better outcomes (lower bodyweight loss, lower grades for those phenotypes) over time?
Whether there's any interaction between the injection vs topical treatments?
Whether there's any interaction between the time vs injection/topical treatments?
Starting from which day does the outcomes of different treatment groups significantly different from each other?