Dear colleagues,
We test two treatments (A and B) on a population of about 50 subjects.
Each subject was administered both the treatments (separately). The
outcome of each treatment was measured using five different parameters
(all of them - between 0 and 100, with an average resolution of 5). My
goal is to show that the two treatments are similar.
Following are my questions:
- Had I need to show difference (H0: A and B are similar), I could
have used paired t-test on each of the five parameters. Is paired
t-test valid in this case (H0: A and B are different) and how must I
interpret its results?
- What is the best way to deal with the fact that there are several
independent parameters? How wrong is it to concatenate all the data,
so that 50 patients with 5 parameters become 250 data entries (per
each treatment, A and B)?
Thank you very much
|