How can I analyze a design where I want to determine the effect of a treatment (loading or not loading) within subjects and genotype (B6 or IL6 mice) and loading duration (1 or 2 weeks) between subjects? The design is not balanced so I was thinking of using proc glm or mixed model? Any suggestions on how I write the code?
As I read this, each subject gets both loading and not loading treatments, but one subject only has one loading duration. This seems like and example of a Crossover Design. So the data might look like this
Load NotLoad Genotype LoadDur
Load and NotLoad are assessments of the same outcome. Genotype and LoadDur are binary (scored 0/1). There are two ways to analyze this. One is to take the difference between Load and NotLoad (e.g. LoadDiff=Load - NotLoad; ) and use that difference as the outcome with Genotype and LoadDur as independent variables. The other is to use an analysis of covariance framework and treat Load as the outcome and NotLoad as a covariate. The basic GLM model statements would be
MODEL LoadDiff = Genotype LoadDur;
MODEL Load = Genotype LoadDur NotLoad;
You may need interaction terms and/or to specify some options as well.
So there are B6 and IL6 mice (genotype of mouse)
Within each mouse, the left tibia is loaded and the right tibia is not loaded (loading)
We also had a group of B6 and IL6 mice that were underwent this loading treatment for 1wk and another group of B6 and IL6 mice that underwent the treatment for 2wks (loading duration).
I think I should use proc glm to take into consideration the within-subject effect of loading (loaded, control limbs) and between-subject effects of genotype (KO, WT mice) and loading duration (1, 2 wks), as well as interactions between these terms.
Maybe the repeated measures term can account for the paired aspect of loading?
Is this correct and how do I write this code?