On a related note, how does one calculate within-subject variance, given one baseline measurement? I am afraid that my subjects are small enough that I can't obtain many totally different samples for the variable.