To help outline my question I will start with how this would normally be undertaken. Say this is animal study looking at the effects of a drug on withdrawal. The outcome in this case will be time spent asleep with the expectation that the drug improves sleep in animals undergoing withdrawal.
There would be four groups:
- No drug 'healthy' group
- Withdrawal + no treatment group
- Withdrawal + treatment group
- Withdrawal + sham treatment group
Ideally, you would want to see that the treatment causes improved sleep compared to untreated animals are no different than the no drug animals. Here the analysis could simply be a one way ANOVA.
With animals you can, however, take baseline measurements before the experiment replacing the no drug group. This seems preferred design wise for several reasons like reducing total number of animals. But the question becomes how would one analyze the data because the baseline is now within group while the other groups would be between. Presumably, you could just do this descriptively (and maybe I'm over thinking this), but I was curious if any formal statistics would be appropriate.
