送交者: xj 于 2005-8-20, 00:20:43:
回答: The problem is: 由 cornbug 于 2005-8-19, 23:49:59:
if you want to impress others, you would better try repeated measurement model, as you repeatedly measure the same subject four times. These measurements are likely correlated within subjects.
In SAS, proc MIXED with REPEATED statement will do the trick.
However, I am not aware of how to do multivariate repeated measurement in SAS, as I think SAS proc MiXED only allow for one dependent variable. Stata cannot do it either, and NLME in R is worse. Anybody wants to enlighten me?
If your x1,x2,x3,x4 are not correlated in obvious ways (but everything will be correlated, so don't look at the correlation matrix), just do four separate models.
Talk to someone in your institution to get some real help. This is too much for internet consulting.