Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: complex anova vs. xt regression model

From   "george m hoffman md" <[email protected]>
To   <[email protected]>
Subject   st: complex anova vs. xt regression model
Date   Wed, 31 Jul 2002 18:24:34 -0500

Dear Stata listers:
I have a panel dataset with 48 hourly observations per patient  on a number of
physiologic variables and treatment variables. I am trying to test the effect
of one physiologic variable (a)  and its interaction with one treatment
variable (treat) on another physiologic variable (y).  the 'independent'
physiologic variable is not linearly related to the outcome variabke, so i
constructed categories of it (acat). the basic structure i think would be:

 anova y  acat / acat|treat / treat /patient|treat hour , rep(hour)
bseunit(patient) bse(treat)

alternatively, i have tried xtreg:

xi: xtreg y treat i.acat*treat

the significance of regressors is quite different in the two models.
specifiying xtreg ,be or ,fe is not quite good enough - as far as i can see,
variable "acat" is a within effect, but variable "treat" is a between effect. i
think this is a similar question as to asking what the appropriate error term
for acat is.

general guidance on this topic would be appreciated. I have scoured the manuals
and the faqs regarding repeated-measures tests, xtreg, between estimators, etc.
thank you in advance....
George M Hoffman, MD
Associate Professor, Anesthesiology & Pediatrics, MCW
Milwaukee, WI 53226 (414)266-3560
<[email protected]>

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index