Re: st: Upcoming NetCourses

[Stas Kolenikov <skolenik@gmail.com>]

On Tue, Sep 7, 2010 at 7:24 PM, Mark Nichols <nichols.mark0@gmail.com> wrote:
> You don't need them if you are capable of taking econometircs at San
> Diego. I have examined the material and in my opinion not really worth
> your time or money. That was about 3 years ago. If you can't or don't
> like to figure out things on your own, then the netcourses might be
> just the thing for you.

When I taught in an economics summer school, I told my students that
the most useful class they would have in that summer is the research
seminar. You can read books on micro and macro, and there are long
sequences of either subject in grad schools, but there is no telling
if you will become a micro or a macro economist. But one thing is
certain: you will be reading other people's articles, and maintaining
a database of references; and you will be writing your own articles.
So you need to learn how to find research on a given topic, and how to
use LaTeX (may be coupled with something like citeulike.org). And,
since the summer project involved data analysis and empirical work,
there was one other certain thing: you need to learn Stata. And that's
what the research seminar was about.

I kinda doubt that Granger or White are going to teach anybody how to
run -forvalues- loops. Rather, they would find the finer aspects of
asymptotic theory to be more fascinating material that every economist
must know. The NetCourses are complementary to this sort of training.
They address what's implicitly assumed already perfected by the
standard econometrics class (especially in the top school): that your
data are perfectly well organized, and you just need to run one
regression to get your results. Preparing stable data and running a
stable analysis requires an extensive set of tricks that are next to
impossible to figure out on one's own. You need to look over
somebody's shoulder to get certain "aha" moments. Of course studying
the manuals, reading all the tips and Mata matters and Speaking Stata
articles in Stata Journal will get you there, and probably further
along, than NetCourses... but whether it will be cheaper and less time
consuming is a tough call :))

There's another aspect that's somewhat harder to explain. Stata coding
has certain conventions that most "mainstream" programmers (those who
have taken NetCourses and/or watch statalist and/or appear at user
group meetings) follow. Stata code of people outside this mainstream
is almost inevitably odd-looking, which means, more likely to be
error-prone, and more difficult to document and maintain. A simple
reason for this is efficiency of the code: Stata Corp. writes very
efficient code (subject to the version control of course: some v6 ado
files may still be found among the official ado files, and they do
look odd, too, given what's available in the current version), and
NetCourses will teach you how to approach their efficiency bounds :)).

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
*   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/