I apologize to any Statalist reader who may have thought that I plan to send
any "dirt" I hear about to Stata's competitors. Quite to the contrary, I
suppose that I am one of the most vocal advocates of Stata to the general
statistical community. I use Stata almost exclusively to demonstrate statistical
topics that I write about in books, articles, and courses. I also frequently
refer to its many capabilities in my Section Editor's Notes column for The
American Statistician (TAS).
During the 8 years that I have been software reviews editor for TAS, we have
not published a separate review of Stata. We've had many reviews in which
Stata has been compared with other packages on some type of capability; e.g.
GLM, GEE, survey stats, exact statistics, and so forth. But Stata has never
been reviewed as a standalone package. So I'm taking on the task of writing such
a review. I know the good points - and they are many. But I do not
intend for the review to simply be an advertisement for Stata - it would
have little worth, and would undermmine my credibility.
I know of some capabilities that I'd like to see added to Stata, and I'm
aware of a few little problem areas. But I may have missed something. If there
are some statistical issues that may need to be addressed, I'd like to point
them out. As it is, I so far have some 19 typewritten pages listing
capabilities -- with examples of its power, comprehensiveness, and excellent support. I
have two paragraphs discussing of a couple of items that are not part of
"official" Stata, but that are nevertheless available through user written
programs. There are also two other statistical capabilities that are not yet in
Stata that I mention. As it is, few packages have either of them.
There may be other points that I discuss, but the end result is that Stata
is a superb package that is continually providing users with better
statistical, programming, graphic, and data management tools. It's support system is
second to none.
==========
As I mentioned, no one package can do everything. If it could, it would
likely be too unwieldy to use. But Stata Corp is making the attempt to provide
its users with the tools required to address their specific needs. There is
absolutely no question about its desire to respond to the desires of its users
-- and has continually made new versions and incremental
enhancements available to its user base. Just think about some of the
positive features of being a Stata user:
1. Free tech support
2. Ability to correspond to other users and to Stata staff via the StataList
3. Can take very reasonably priced web-based courses on programming, etc
4. Can now read full fledged textbooks specifically devoted to explaining
some aspect of statistics, including graphics, data management, and
ML programming. Stata Press books provide Stata users with excellent
support. Take a look at Sophia Rabe-Hesketh's newly released book on
"Multilevel and Longitudinal Modeling Using Stata". It is one of the
best texts I've ever read on the topic. Truly excellent. A host of other
books are being prepared.
5. Annual Stata Users conferences in North America, Europe, and
Australia/NZ.
6. A relatively easy-to-learn programming language which can be used to
program nearly any statistical procedure conceivable. Stata's ML and matrix
(Mata) capabilities are superior. I know of no other higher level
statistical
progamming language easier to use, or with the comprehensiveness of scope.
7. A refereed professional journal (Stata Journal) that provides excellent
support
on using Stata. Previously, beginning in 1991, Stata sponsored the STB that
provided users with a number of useful programs, and provided overall
support.
Like the STB, users are encouraged to contribute articles and programs to
the SJ.
8. Web-based access and links to other sites -- excellent web support. For
example,
the UCLA support site provides programs and tutorials on using Stata - for
free.
9. And many other features...
Well, enough for now. I encourage you to read the review if interested. It
is to
be published in the November issue of The American Statistician.
=========
As a side comment: there were several comments that I received about
capabilities that
were lacking. Well, in some cases Stata did have the procedures. I pointed
them out.
I was also able to show another reader how to program in Stata in order to
obtain their
desired statistical tool. And I also intend to send the (short) list of
desired enhancements to Bill Gould for his consideration.
Joe Hilbe
*
* 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/
