st: In Search of the Perfect STATA

["Steven A. Melnyk" <melnyk@msu.edu>]

I have been reading the comments (pro and con) regarding the changes that 
have taken place in the recent version of STATA.  I have read the defense 
of the changes as presented by several members of the list.  I will throw 
in my two cents worth. First of all, it is important that you understand my 
role and my position on STATA.  I am a researcher in Operations and Supply 
Chain Management in the Business School at Michigan State University. I am 
a user of STATA; I am not a statistician.  I see STATA as simply a part of 
my statistical tools kit -- a kit that currently consists of SPSS, JMP, and 
in the past GAUSS and SYSTAT.  To me, STATA is a tool -- no more, no 
less.  I use STATA because it offers many regression features that are not 
found in JMP, SYSTAT, and SPSS.  I use STATA because of the ease with which 
it can and is updated and maintained (something that is hard to do with 
SPSS).  I use STATA because it is fast and it does not get in the way of my 
analysis.

I loved the old list command because I get peaks into the data very quickly 
-- the response was instantaneous.  Even with large databases (one of my 
databases consists of 1510 observations and 352 variables), STATA was 
quick.  Now, I am confused with the changes because STATA is no longer as 
fast.  I know that there is a way of changing the list command but it does 
not seem to be evident to me when I look at the help file in STATA V8.0.  I 
have found that with some of the other commands that I have used 
previously, STATA seems to have slowed down. To me, STATA is no longer as fast.

I realize that I can build a shell so that I can have clean output, but I 
ask myself the question of "why?"  In other programs, when a new structure 
is introduced (such as in Framemaker), I had the option of turning the new 
features on or off when the new upgraded version was first opened.  This is 
something that I cannot do with the new STATA.

The reason that I bring these issues up is because there are users like 
myself.  When we upgrade we know that there will be changes.  However, my 
expectations are that the new version will at a minimum perform like the 
old one and that it will add to the capabilities of the old one.  I find 
myself perplexed by the response that there is always some upgrade 
reaction.  I have upgraded numerous programs.  Yet, it must be expected 
that corporations such as STATA, when they upgrade programs, do maintain 
some degree of continuity with the past.  This applies to features, data 
structures, and commands.

As a user, I crack the manual open only when I need to (e.g., when to look 
up how to do a procedure such as q-factor analysis).  I do not expect to 
crack up the manual so that I can relearn how to do a list command fast.  I 
expect STATA to be fast.  When I see evidence that this is no longer the 
case, it brothers me.

I realize that these comments might seem out of place. However, I hope that 
you can understand my position as a user.  The tool must meet my 
expectations.  When it does not or when it gets in my way, that bothers me. 
When the tool demands too much attention, then I am given an incentive to 
look elsewhere.

As I said before, my two cents worth.  I hope that I have made my concerns 
about the new upgrade clear.  I realize that it offers more statistical 
power.  However, the trade-offs offered by this new upgrade bothers 
me.  Ultimately, that is the major point of this note.

Steven A. Melnyk
Michigan State University

*
*   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/