Hi,
In a quasi experimental design non equivalent Control Group (Pre-treatment-post ), what is the difference of having the unit of analysis as intact classes (e.g. 8 classes: in three different schools and in each school there are experimental and control groups) or individuals (175 students)?
Hi Hoda,
That will depend on the unit you are measuring the outcome. If your outcome is the change of an individual measure, then I think I'd be preferable to model it as individuals nested in classes nested in schools and the treatment variable as a covariate.
Best wishes,
Sebastián
I'd agree with @Sebastian.
I think at the end of the day, you would have only 2 groups of individuals (experimental and control) regardless the school. This will allow you to actually see the treatment effect It would be statistically confusing if you just compared 4 to 4 (hope you get what i meant by this).
Plus, if you do this, you can explain the possible variation effects of the treatment, based on school or individual characteristics.
- Badges
- Science topic
- Similar topics
- Teacher Education
- School