re: st: list x matrix

[Kit Baum <Baum@bc.edu>]

<>
Maarten said, in response to my Mata posting,

If we take Richard's question literaly, then this if probably
the best answer. However, I would argue that the fast 
majority of users should *not* use it, this includes users who
program in Stata.

I wholeheartedly agree. If a tested and supported command in Stata will do the job, users should take the time to seek 
it out and use it. The fact that one _can_ do any matrix computation quite easily in Stata/Mata does not mean that it is a good
idea for users to 'roll their own' solution. The recent interchange on this list on computational issues with -margins- is
a good case in point. The way that Stata calculates something as simple as  a linear regression is much more
numerically stable that a brute-force computation of (X'X)^{-1} (X'y). For pedagogical purposes, using the latter is 
fine. For serious research that could involve ill-conditioned regressor matrices, it is decidedly not recommended.

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/