I have a dataset from a panel (observational) study.
The data record all subjects' buying behavior in specific stores (for example: 3 ) across 3 equal and continued time-periods.
Some behavioral variables are highly or moderately correlated.
A few new variables are created in need based on some observed variables.
I want to confirm: Comparatively, to what extent, which behavioral variables and/or created variables are significantly time-affected?
What procedure would be better for the above question?
1. One-way Repeated ANOVA /MANOVA test
2. PROC GLM;
3. PROC MIXED (treatment: store & time);
4. Other....
Any comment is welcome.
Thanks.
