Statalist


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

st: RE: RE: RE: RE: clad post-estimation question


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: RE: RE: clad post-estimation question
Date   Thu, 6 Nov 2008 11:11:30 -0000

In principle, yes indeed. In practice I recognise none of the authors as
people active on Statalist. Previous questions on -clad- on this list
have been answered by other people if at all. 

I don't know how much benefit there would be in updating -clad-, but my
strong guess is anyone who wants that will have to do it themselves. 

Let's be clear: this is a simple downside of user-written software.
User-programmers are busy people and their interests can shift like
anybody else's. Sometimes they write something in Stata and then they
get too busy, or their interests change, or they get using other
software. Upshot: they may feel no obligation to maintain their
programs. It's as basic as that. 

Also, user-users have an obligation not to be dopey and to look
carefully at what they are using. Only the other day I was asked offlist
whether a program of mine written in 1998 -- this fact had been
emphasised in an earlier mail -- supported a statistic published in
2001. 

Nick 
n.j.cox@durham.ac.uk 

Martin Weiss

"In this case, it is not reasonable to suggest that the authors of a
Stata 5 program should have made use of commands or concepts not then
introduced."

Maybe the authors benefit from my discussion of their contribution and
edit
their -program- accordingly.  The current situation is confusing, and it
took me some time and a longer session of -viewsource- to figure out the
behavior reported by "New User". 

Nick Cox

As Martin separately indicated, this program was published in STB-58 --
in 2000 for Stata 5. 

As a reminder, all users -- new or not -- are asked in the FAQ to make
clear where user-written programs they refer to come from. Otherwise
quite what they are asking will be unclear to many and they reduce the
chance of a reply that is prompt, clear, complete and correct. 

In this case, it is not reasonable to suggest that the authors of a
Stata 5 program should have made use of commands or concepts not then
introduced. 

-clad- was written for Stata 5 but evidently will work smoothly with
later versions of Stata. The heavy lifting is all done by Stata's own
-qreg- which is now eclass. Thus any stuff shown by -eret li- is
returned not by -clad-, but by -qreg-. 

Users can use 

. viewsource clad.ado

to see what is going on. 

Martin Weiss [expletive deleted] 

Look at -eret li- after estimation and ask yourself: Can this really
come
from -clad-? [...] The results saved there are from the last replication
call to -qreg- as evident from e(cmdline). Specify the -saving- option
to
-clad-, let it run, type -eret li- and then -use- the file specified in
-saving-, -list in l- and you see the result that still resides in e(b)
and e(V). It makes me think the authors should have thrown in something
like -eret clear-.

New User/Bill 

I used clad command for the censored least absolute deviation
estimation. After I ran clad, I intended to get coefficient and
std.error estimates using e(b) and e(V). What surprised me was that
the estimates in the report table and the values retrieved by e(V)
were hugely different. Has any one ever met such problem?

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index