RE: st: Multinomial Logit Model - Cramer Ridder Test

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 08:47 AM 11/14/2011, Nick Cox wrote:
> This is not quite correct. Rather, installing -crtest- from SSC 
> implies installing the current version of -_pecats- from SSC, 
> which  presumably remains the responsibility of its authors.

Ahh. I see your point. But Long and Freese abandoned SSC a while 
back, and Long maintains his programs on his own site. A consequence 
is that SSC does not have the most current version of _pecats and 
other popular programs like -fitstat-, which seems to be a frequent 
cause of grief to people who don't realize they aren't running the 
latest and greatest versions of those programs. In any event, I'd be 
reluctant to rely on someone else's user-written program because, as 
you say, they can make changes that would zap my code. I'd rather 
copy the program and give it a unique name, so long as the original 
authors did not object to that. Also, in the present case, having 
_pecats on SSC prevents me from installing -crtest- unless I am 
willing to overwrite the newer version of _pecats that I got from 
Long's site. The current version of _pecats is

. which _pecats.ado, all

c:\ado\plus\_\_pecats.ado
*! version 1.7.3 2007-06-29 Stata 10 fix for categories

Perhaps it would be a good idea to hunt down all these abandoned Long 
and Freese programs on SSC and add a note to their descriptions that 
the most current versions are on Long's site. When in doubt, I 
usually assume SSC has the most current version of a program, but in 
the case of Long & Freese's programs that is not true.

> If -crtest- is installed right now, then -_pecats- 1.6.8 would be 
> installed, setting aside other possible complications:
>
> . ssc type _pecats.ado
> *! version 1.6.8 12/17/00
>
> I don't think there is any objection in principle to using other 
> people's user-written programs. The main problem in practice is that 
> changes in those other programs might zap the using programs.
>
> In particular, anybody interested in this kind of thing is likely to 
> install -spost- from the authors' website (-findit spost-), which in 
> turn may well overwrite the -_pecats- above.
>
> Neither program here is mine, but speaking as a program author I 
> can't possibly know about everything else that uses my programs. I 
> am reluctant to change syntax in a way that would have major effects 
> on others, but Stata moves on, and maintaining multiple versions or 
> having multiple version-dependent switches inside programs is not 
> attractive to me, so sooner or later dependence on a program that no 
> longer exists may become a problem. Many program authors have 
> similar attitudes.
>
> Nick
> n.j.cox@durham.ac.uk
>
> Richard Williams
>
> As a sidelight, -crtest- includes its own copy of _pecats.ado. I
> consider this bad form, because _pecats.ado is a Long and Freese
> command that is part of their suites, and the version included with
> -crtests- is several years behind the current version. It may have
> been included with permission, but if it was me I would have renamed
> it to avoid conflicts with other programs. I tried to install
> -crtests- and got these warning messages about overwriting existing
> files, which I do not want to do.

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/