Statalist The Stata Listserver


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

st: RE: Suggestions for Stata


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Suggestions for Stata
Date   Thu, 6 Apr 2006 21:39:07 +0100

Reactions below. 

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

Bromiley, Philip
 
> Although I'm sure the more sophisticated users don't have these issues
> (and fixes may exist I haven't found), as one who moves 
> across packages, it would help me if:
> 
> 1.  Stata made full variable names available for the output of all
> procedures.

You reckon that 32-character names are consistent with the other output
of all procedures? That is, there is enough real estate in 
terms of monitor size and printer paper to show 
all that is reasonably wanted without SPSS-like tree consumption? 

> 2.  Stata made missing data consistently treated as missing, 
> not just as
> a big number in logical statements.  

The thing that bites most people, including, some times, long-time users 
like myself, is the fact that 

... if x > # 

includes missing values of x. But all other solutions have
big disadvantages too. I believe that StataCorp have played
internally with three-way logics and the idea of another operator 
-- which ended up as 

if inrange(x,#,.) 

-- but they all offer major pain or least awkwardness too. 
I am not sure where you think the precise major issue is. 
In broad terms, missing values 
disappear temporarily for statistics and graphics but are 
evident in data management. (After all, if observations are
not considered part of the data, then just leave them out.) 
Isn't that how it should be? 

The history of missings and very big numbers has been recounted
many times by Bill Gould. Consider any -sort-. The missings have 
to go somewhere, and that somewhere must be at one end. Some 
Alternative Software put them at the beginning, but Bill didn't 
like wading through dreck before he got to interesting data, 
so at the end it was. 

> 3.  Stata used the same single = sign for both assignment and logical
> statements.  

On the contrary, it is mathematics that is wrong here! 

> 4.  Stata offered more flexible table output.  My journals want
> coefficients, stars for significance level, and standard errors in
> parentheses under the coefficients.  It would be nice to have a simple
> way to do this.  Allowing users to set up a template might be helpful.

This has been a long-term issue. My wild guess is that this will happen
around Stata 11 or 12! 

> 5.  Color codes in editors are nice even for experienced users.  For
> example, in SAS some errors jump out because the colors go wrong -- a
> forgotten end of comment makes everything below a comment color, etc.

There's just been a thread on this. You should read the postings.  

*
*   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