Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: error message after dtobit

From   "Nick Cox" <>
To   <>
Subject   RE: st: error message after dtobit
Date   Mon, 8 May 2006 12:32:35 +0100

Thanks for the mention. 

I ain't so kind as Richard implies, or rather as 
you may infer from this. Rather, 
if a new version of a program I write requires
a later version of Stata, I tend to keep the 
older version in circulation under a different 
name, so that people still using older versions
have the chance of something rather than nothing. 
But I never go back to older versions 
and add new features _even when it is possible_. 

Life is too short and busy for that. 

I am not speaking for the other people named, 
whose practices may differ slightly. But in 
general, Stata moves on and so too do almost
all Stata user-programmers, especially the most
active. Anything that requires 
the new graphics will require Stata 8, and
anything that requires Mata will require Stata 9, 
for example. People who, usually for pressing 
financial reasons, stick with Stata 7 or earlier 
versions are missing out on more and more as time
passes, mostly on official additions but also 
on user-written ones too. 

On the broader issue here, note that although 
official Stata changes have caused some awkwardness
for users of user-written programs, the problem
is soluble and it is not as common as is the case
with some other software. 


Richard Williams
> At 08:15 PM 5/7/2006, Richard Williams wrote:
> >You often need to use version control on the estimation command with 
> >user-written post-estimation commands.  Stata changes something and 
> >these older post-estimation commands get zapped.  Try this:
> To follow up on that a bit:  If you type -help version- , you find 
> that if you set version to less than 9,
> "14.  Commands ologit and oprobit revert to pre-Stata 9 logic in how 
> e(b) and e(V) are stored.  Results were stored in two equations, with 
> all cut points stored in the second.
> "15.  Commands tobit and cnreg revert to pre-Stata 9 logic in how 
> e(b) and e(V) are stored.  Results were stored in a single equation 
> containing both coefficients and the ancillary variance parameter."
> So, my guess is that older user-written post-estimation commands 
> written to work with ologit, oprobit, tobit and cnreg may encounter 
> problems in Stata 9.  There may be others.  Unfortunately, the fact 
> that the user-written command used version control does not help, 
> because it expects the results stored in e() to be in a certain 
> format.  This seems to affect the user-written commands dtobit, 
> margeff, and no doubt many others.
> The solution is to run the estimation command with version 
> control, e.g.
> version 8.2: ologit y x1 x2 x3
> Perhaps the authors will eventually update their programs, or if you 
> are daring you can try to make the modification yourself.  Also, some 
> authors offer different versions of their programs for different 
> versions of Stata (e.g. the spost programs of Long & Freese; Roger 
> Newson and Nick Cox are also among the kindly authors who do this).
> It is also possible sometimes just to write a little code in your 
> program that checks for the version of Stata being used and then 
> adjusts accordingly.  When it works, this saves the hassle of having 
> multiple versions of a program floating around.  Or, you might be 
> able to come up with a more general coding that works regardless of 
> the version of Stata used.

*   For searches and help try:

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