I have a question about a sample size calculation that i am sure must be
simple but i have got confused over. I want to do a study that has the
absolute change from baseline as the endpoint. To analyse the results I
will use a paired t-test. If a change of 10 points is considered a relevant
result, how does setting up these two hypotheses below affect the sample
size calculation? ( To calculate the sample i use significance as 0.05 (2
sided as i'm not absolutely certain that the change will be positive) and
power = 90% and the estimated s.d of the difference.
1) you set your hypothesis up as Ho: d=10 vs H1: d is not equal to 10
2) you set up your hypothesis as H0: d=0 vs H1 d is not equal to 0
What are the implications of using each of these tests ( i.e exactly what
are they testing)? Which is best/wrong? Aren't they testing the same thing
but in opposite ways? Perhaps someone could explain the difference to me?
Regards
K Tilson
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|