The project does not have a simple design. It is planned as follows:
- It is a longitudinal observation during 120 days in 48 subjects (with repeated measures using multiple tests);
- The subjects were initially divided by genotype (24 WT and 24 KO) and their baseline behavior was further analyzed (timepoint I);
- Later on, the subjects were further distributed in 4 different treatments (placebo, A, B, C) and staying, therefore, 6 mice of each genotype under each treatment;
- The behavioral and physiological analyses are performed on days 30, 60, 90 and 120 after baseline tests (timepoints II, III, IV, V, respectively);
It is essential to note:
- At the deadline of this study, the 8 groups (WT + KO) * (Placebo + A + B + C) will vary in the number of subjects included, coexisting between 3 to 6 subjects, to be more precise;
- Not all subjects entered the study at the same time, which means that within a treatment I will have complete sets, for example, at timepoints I, II and III, and only with subjects per group (who entered earlier) for the following timepoints;
- Some subjects were withdrawn from the study during the 4 month period, which means that they only gave data at some initial timepoints;
The aim of the study is to clarify 1) whether A and B treatments have a significant positive effect compared to the placebo group, within the same genotype and over time, and 2) if C treatment offers even more broad differences compared with treatments A and/or B (and therefore the placebo) also within the same genotype and over time. As a less important analysis, it would be interesting to see 3) whether the differences between the groups with WT subjects are similar to the differences between the groups with KO subjects.
In other words, it would be ideal that the statistical test could detail the significant differences (or not) regarding the: (in order of importance)
a. 4 treatments within the same genotype at each timepoint;
b. 5 timepoints of each unique group;
c. 2 genotypes within treatments at each timepoint;
The general opinion of my colleagues is that I should try Repeated Measures ANOVA, but my impression is this design should be subjected to a Nested ANOVA. Moreover, what causes me confusion is whether perhaps the analysis of the all the timepoints won't respect the chronological sequence. Still, any test that I use, will be limited by the issues explained in points 5., 6. and 7..