How to controll for practice effects in a repeated within-subjects design?

Hey there,

I'm currently conducting a pharmacological memory task that is designed as a repeated double-blinded within-subjects experiment. Participants have to perform the same memory task (but different stimuli) on two days (minimum break of 6 days). I was expecting practice effects (= improvements in performance after repeated exposure to test materials) and now I'm thinking about how to disentangle these practice effects from effects induced by the pharmacon? Since I'm a med student my statistical background is quite weak and I was surfing the internet for weeks to solve this problem.

I'm happy for any advice! Thanks a lot and stay healthy!


Active Member
ok well, this experiment has 3 things you can estimate, this is all the estimable things:
1)effect of stimulus (treatment)
2)effect of period (given first of second)
3)the interaction of the two above.

The memory effect would be detected as 2), or possibly 3) I would imagine, but it could also indicate a myriad of other issues, use you imagination. Unfortunately, it all gets 'lumped' in this design.

Piantadosi gives a good synopsis of the relevant formulae:

SAS Proc t-test gives a crossover software (you can use the free SAS academic if you like).
However I think it does not produce the test of interaction.

hope that aids your journey.