Statalist The Stata Listserver


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

st: RE: Stata functions


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Stata functions
Date   Tue, 24 Jan 2006 00:17:28 -0000

I suspect that this was decided as the 
convention a long time ago and Stata 
has stuck to it since. 

There is at least one very good reason, however. 

You are thinking of functions for which the 
arguments in practice are always lists of variables, but 
syntactically there is nothing to stop you typing 
other stuff, say, 

missing(exp(1), _pi, 42)

and in a program one might well calculate 

missing(`somelocal') 

where the macro evaluates to a number not known 
in advance. Now once expressions evaluating to 
numbers are allowed, in particular, the problem
is what is meant by 

missing(4 -7) 

Stata's other rules would treat 

4 - 7 

as an expression to be evaluated on the fly, 
namely to -3. Insisting on 

missing(4, -7) 

fixes this problem. 

In short, allowing lists without comma separators would solve one 
problem at the expense of creating another. 

Of course, you may well say that you want both syntaxes! That 
in turn would be understandable, but objectionable on yet other
grounds. 

-missing()- is in practice fairly extreme in that most applications
probably involve variables. (And -egen- offers functions too.) 
With other functions this would arise all the time and often bite
users: in fact, any time that a negative number was not first in the list.
What's more, this would be badly hidden as it could well happen
in some observations and not others. 

If this arises often a little utility is programmable to 
put the commas in. In fact one such is lurking in your folders. 

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

Fred Wolfe
 
> Apropos of the recent discussion of inlist, I have often 
> wondered why Stata 
> doesn't allow varlists in functions, but instead requires 
> comma separated 
> lists. For example,
> 
> missing(,a,b,c,d,...) is required. Why not a varlist? I can't 
> quite see any 
> difference. However, in terms of programming, a varlist is 
> much easier.
> 
> reg y  a b c if !min(b_*)
> 
> is much easier than typing each of the 10 b_ variables which 
> may be long 
> and easy to misspell.

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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