Dear ALLSTAT users,
I have recently faced an interesting problem, which I
don't know how to handle.
Suppose we have 2 groups: placebo and drug.
Vector X=(X_1,...,X_t) corresponds to responses of a
patient on a drug, at time points 1,...,t. It is
assumed to be multivariate normal with mean \mu_D and
covariance matrix \Sigma (having compound symmetry
structure).
Similarly, Y=(Y_1,...,Y_t) is a multivariate normal
vector with mean \mu_P and the same covariance matrix
\Sigma, represents responses on placebo.
Assume that X and Y are independent (parallel arm
design)
Parameter of interest is M = maximum component of
(mu_D-mu_P). I need to obtain a 95% CI for M.
Are there any results about the distribution of sample
maximum for dependent sample?
Thanks in advance,
Alex.
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
|