Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

AW: st: RE: RE: RE: RE: error messages while using the stgest command

From   "Martin Weiss" <>
To   <>
Subject   AW: st: RE: RE: RE: RE: error messages while using the stgest command
Date   Fri, 4 Jun 2010 15:40:03 +0200


What happens if you run the examples in the do-files shipped with the
package? Use -findit stgest- to find them. I know it sounds trite, but if
they do run, and yours don`t, you have to find a difference in the setup
that leads to the problems...


-----Ursprüngliche Nachricht-----
[] Im Auftrag von Victor Mauricio
Gesendet: Freitag, 4. Juni 2010 15:32
Betreff: Re: st: RE: RE: RE: RE: error messages while using the stgest

Regarding the first error, I included no graph options in my instructions.
This is exactly what I typed:

stgest diabetes cursmoke agebase sex hyperten obese, lagconf(hyperten obese
diabetes) baseconf(hyperten obese diabetes) /*
*/ visit(visit) firstvis(2) (lasttime(endfudate) range(-50 50) step(1)
saveres(caergestsmoknocens) replace

I tried with two different datasets and got the same error.

Regarding the second error, I did run a Weibull model before running
gesttowb. This is what I typed:

weibull _t diabetes B* L* if visit>=2, dead(_d) t0(_t0) hr


On 06/03/10, Nick Cox  <> wrote:

> That won't fix the error I am guessing at, for two reasons:
> 1. Whatever -version- Victor declares in a .do file is moot given the
> internals of -stgest-. 
> 2. Using -version- will not make -version 7: graph- understand the
> options of -version 8: graph-. That the syntaxes are in part identical
> is just a linguistic coincidence. 
> Nick 
> Martin Weiss
> Could Victor set -vers 7.0- at the top of his do-file to make this work,
> then?
> Nick Cox
> It's indeed difficult to add to Martin's diagnosis of the first error
> because, although meticulous about other details, Victor omitted to
> specify exactly what he typed. But the graph call in question 
> graph z psi if z<2.5&z>-2.5, yline(-1.96, 0, 1.96) xlab ylab
> xline(`gestl', `gest', `gestu') `options'
> is under Stata 7 syntax (in Stata 8 up documented under -graph7-). Here
> `options' is, or should be, whatever graph options Victor typed on his
> command line. Any attempt to add options for the graphics introduced in
> Stata 8 is likely to produce an error message. 
> Nick 
> Martin Weiss
> The second error occurs because -gesttowb- is looking for results left
> behind by a Weibull estimation. If you have not run this estimation
> previously, the macros that -gesttowb- is trying to multiply do not
> exist.
> The example code "" delivered with the package runs without a
> hitch for me, btw, and executes -gesttowb- repeatedly.
> The first error does have something to do with the -graph- command on
> line
> 304 of -stgest.ado-, though the problem is difficult to track down. 
> Victor Mauricio Herrera
> I wonder if anyone could shed light on the following. I consistently get
> two
> error messages while using the stgest command (Stata Journal 2002,
> 2(2):164-182) in Stata/MP 11.0.  The first one happens after the causal
> survival ratio is reported:  
>                 "Note - these estimates should be confirmed with an
>                   interval bisection search by omitting the step option
>                   option ( not allowed
>                   r(198);"
> I guess this has something to do with the instructions to graph the
> psi-by-z
> graph.
> The second error occurs when using gesttowb:
>                  gesttowb
>                  g-estimated hazard ratio *:  operator invalid
>                  r(198);
> My versions of these programs are:
>                    which gesttowb
>                    c:\ado\plus\g\gesttowb.ado
>                    *! Version 1.0         (SJ2-2: st0014)
>                    which stgest
>                    c:\ado\plus\s\stgest.ado
>                    *! version 5.5 JACS/KT g-estimation, June 2002
> (SJ2-2:
> st0014)
> *
> *   For searches and help try:
> *
> *
> *
Research Assistant - Wisconsin Diabetes Registry
Population Health Sciences Department
University of Wisconsin
610 Walnut St. 626 WARF
Madison, WI 53726
(608) 265-3686

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index