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

st: RE: xi command

From   "Nick Cox" <>
To   <>
Subject   st: RE: xi command
Date   Wed, 13 Aug 2003 20:19:55 +0100

Åsa Odenbro

> I am a new user of Stata and I am doing a project on
> squamous cell cancer and smoking habits in the construction
> workers cohort in Sweden.
> I have been trying to do Poisson regression with the xi
> command, but it has been giving me some trouble. 2
> questions have arisen:
> 1.	What is wrong with the xi3 command? I saw on stata
> list that someone had noticed in february that there is
> something wrong with it but I never found a good answer. He
> had the same problem as I have had. Command:
> xi: poisson outcome i.variable1*i.varable2*i.variable3, exposure(….
> Stata output: varlist not allowed
> 2.	When I use xi with multiple interaction terms, stata
> tells me he is dropping a bunch of variables due to
> collinearity, but then it looks like he has not dropped
> them at all (please see stata output below). But if I then
> put xi3 in front of the command stata does not say he is
> dropping any variables and the results are exactly the same!!

With regard to (1)

-xi3- is not an official Stata command. It was written
by Michael Mitchell and Phil Ender of UCLA, although
neither the .ado file nor the .hlp file appears to mention
this. If you say _exactly_ what you typed, then they
should be able to help answer this.

With regard to (2)

on the face of it your results look very puzzling.
In particular, the messages about collinearity don't
come from either -xi- or -xi3-, or even -poisson-, but from
-_rmcoll- --- and this is the same official Stata command,
called by both commands.

Perhaps someone else can say what is going on.


*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index