I think that the desire for warnings is probably inversely related to user statistical knowledge. With that said, I support more warnings :)
----- Original Message ----
From: Martin Weiss <[email protected]>
To: [email protected]
Sent: Wednesday, November 26, 2008 7:30:39 AM
Subject: RE: st: RE: egen & sum()
Line for the server...
We had a presentation at the San Francisco UGM this year that advocated an
increase in the number of warnings in Stata
http://www.stata.com/meeting/fnasug08/epstein_stata_november_2008.ppt
but support for the idea during the discussion was certainly weak...
Warnings like the -xtmixed- output stating that the lrtest is conservative,
with a link to a help file containing a more comprehensive explanation, seem
appropriate to me. Personally, I like Stata`s conservative approach to
warnings.
Also note that Stata prevents you from doing unreasonable things entirely in
some cases.
http://www.stata.com/statalist/archive/2008-11/msg00984.html
is a good recent example. If you interpret Stata`s refusal to do something
as a severe warning, then the number of warnings suddenly increases
dramatically...
HTH
Martin
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Nick Cox
Sent: Wednesday, November 26, 2008 1:56 PM
To: [email protected]
Subject: RE: st: RE: egen & sum()
The trouble with warnings is that you could write them almost anywhere
on all sorts of grounds: how about more statistical warnings in
statistical software, for example?
I would take note of more warnings if and when they existed and I am
happy to believe that they would be widely welcomed.
But the trade-off between yet more work that somebody (i.e. the Stata
developers) has to do in adding warnings, work that would displace
developing other stuff that we want the company to do _soon_, no right
_now_, is very delicate.
I like the Stata philosophy that users take responsibility for finding
out and that we don't want or need little nagging warnings all the time.
Nick
[email protected]
David Kantor
Here's a suggestion to StataCorp and to anyone who writes programs
and then changes the names of programs or options or syntax.
(I've done this on at least one occasion.)
Allow the old names or syntax to continue working, but issue a
warning message alerting users to the change, whenever they are used.
For those who would object to the possible intrusion of these
messages, perhaps there could be an environmental variable to switch
this on and off.
*
* 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/
*
* 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/
*
* 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/
