Statalist


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

st: RE: AW: RE: AW: RE: Re: Question about -tabi- and its calculations


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: AW: RE: AW: RE: Re: Question about -tabi- and its calculations
Date   Tue, 24 Mar 2009 17:08:59 -0000

Sorry, but this looks like the same comment to me, and my own view is
still the same. 

Most importantly, -profiler- is not at all the best way to find out how
something is done. It's not a service to urge its use when there are
better ways to tackle the question, notably -viewsource-. At most, it
indicates what programs are called and thus may indirectly suggest where
something may be done. 

Users can naturally have different views on what kind of messages are
helpful. The idea of warnings isn't alien to Stata. In general, however,
StataCorp style is to issue very few, as you say. In particular,
-profiler- does not have this kind of message, and whether that's by
accident (as I guess) or design it seems quite right to me. 

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

Martin Weiss


" The effect of your recommendation would be to add this kind of message
to
every set of -profiler- outputs, as absolutely no Stata program (strict
sense) can act independently of the executable."

Stata has come up with so many sensible decisions when to issue messages
and
when not to (mostly by being conservative), I trust it would manage the
problem you are describing quite well. The case Tiago was interested in
seems one where my proposed message, cleansed of the histrionics, would
make
a lot of sense to most users...

Nick Cox

I don't agree. 

-profiler- does what it says. 

The effect of your recommendation would be to add this kind of message
to
every set of -profiler- outputs, as absolutely no Stata program (strict
sense) can act independently of the executable. 

That might be a useful reminder for a first-time user, but it should
rapidly
become tedious and pointless on repetition. 

At most, what it does not do w.r.t. the executable might be made a
little
clearer in the documentation (if it's unclear; I've not read the manual
entry). 

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

Martin Weiss

True, I overlooked this issue. -profiler- could profit from a note at
the
bottom of its output to the effect of "Dear user, Stata also dipped its
toes
into the executable, so you may want to be careful to read too much into
my
results"...

Nick Cox

This is correct, but does not much illuminate Tiago's question. 

-tabi- is just a wrapper for -tabulate- and all the work that concerns
Tiago
is done off-stage by -tabulate-, which is part of the executable. Thus
there
is no code that users can look at. Precisely what -tabulate- does is
thus a
matter for the manual entry. 

. viewsource tabi.ado 

is a more direct way to see this. 

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


*
*   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