Hi Alfonso,

Thanks for the explanation.

I tested again--between the simpler "Patients-pre' "Controls" (contrast -1 1) and "Patients-post" "Controls" (contrast -1 1) analyses (separate computation of residual variance) and the main group effects "Patients-pre' "Controls" by Selecting "Patients-Pre" "Patients-Post" "Controls" "Subject1" "Subject2" ... "Subject10" and enter contrast [-1 0 1 -ones(1,10)/10] while "Patients-post" "Controls" contrast [0 -1 1 -ones(1,10)/10 (pooled computation of residual variance). Indeed, both analyses showed the similar tendencies in results on the one hand but revealed some differences in statistics on the other hand.

In our case, only patients were measured 2 times. Would it be recommendable to use pooled computation of residual variance design then? Or in experience, what would be the advantages and disadvantages to employ the pooled design rather than simpler contrast design?

I also find difficulties to interpret findings by means of 'conjunction of simple main group effects (differences between patients and control during either pre- or post-medication)' design. Normally, in what circumstances, this design could be used and what are its cons and pros?

Thanks in advance again!

Best regards,

Yong