Statalist


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

Re: st: programming with stata


From   "Austin Nichols" <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: programming with stata
Date   Sun, 4 May 2008 14:43:17 -0400

Tunga Kantarci <tungakantarci@hotmail.com>:

The advantage of having a critical mass of colleagues who use the same
program you do is that you can share code more easily, which is a
great boon.  If you are at Tilburg University, you may be sufficiently
near the esteemed Maarten Buis at Vrije Universiteit Amsterdam and Ben
Jann at Eidgenössische Technische Hochschule Zürich to run into them
at seminars etc. and consider them colleagues as well, even if they
happen to work in a merely closely related field!  Of course, the user
community for Stata is not limited by geographic proximity, or lack
thereof, thanks in no small part to Statalist and SSC.

When Maarten talks about "making a command" he means writing an ado
file (automatically loaded program) that others can use as easily as
you can, with the syntax parallel to official Stata commands.  For
example, if you wanted to program a new generalized autoregressive
conditional heteroskedasticity estimator for panel data, you could
work entirely in a matrix language (e.g. Mata) or C++ or what have
you, or you could write an ado file that would call the estimator
(implemented in Mata, say).  So you might write xtgarch.ado and have a
bunch of Mata code in that file that did all the calculations. The
easiest way to get started is probably by reading the code for some
simple existing commands, and working up to more complicated examples.
See e.g.

ssc install qll
which qll
view "`r(adofile)'"

for a nice example with some Matlab code appended, or

ssc install ivpois
which ivpois
view "`r(adofile)'"

for a GMM estimator.  If you write a help file as well, and post it on
SSC, there is always the chance that someone will download it and give
you useful feedback as well...  see
http://repec.org/bocode/s/sscsubmit.html for details.

On Sun, May 4, 2008 at 2:12 PM, Tunga Kantarci
<tungakantarci@hotmail.com> wrote:
> No program is dominating another in my econometrics department: form R to
>  Matlab,
>  from stata to gauss. And nobody is aware of Mata.
>  But I also don't see why I should choose what others choose.
>  One is alone in learning the language anyway.
>  If combining mata with higher level language of stata is very valuable, why
>  should I follow what others do.
>
>  Could you make your following statement more explicit:
>
> "it is very easy to make a command that is indistinguishable from a standard
>  Stata command"
>  In terms of writing a complete time series model, for instance, what would
>  "making a command" mean?
>
>  Thanks for your concern in answering my question...
>  But I think I should figure the answer out myself.
>  Because I just don't know what I will be missing by using mata and not ox.

*
*   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