Re: [Gretl-devel] Signalling
On Mon, 7 Jan 2013, Talha Yalta wrote:
> This has been discussed before, but I think this criticism is
just mistaken.
> The support for decimal comma in Gretl is just as good as the support in
> Excel. Also, you deliberately changed focus from statistical packages to
> spreadsheets. Which statistical scripting languages support localised
> inputs?
“Let’s not kid ourselves: the most widely used piece of software for
statistics is Excel” –Ripley.
Not only that it is the most popular, but also there are many
textbooks for statistics as well as econometrics with Excel. So this
is a perfectly valid example and we do need to consider what makes
Excel, SPSS, Eviews popular if we are talking about how to make gretl
more popular.
The key question here is "popular among who"? Popularity _per_se_ is not
something I personally care very much about. In my view, people who feel
ok doing statistics on Excel may very well keep doing so: even if they
adopted a modified version of gretl in which they could use their beloved
decimal comma, the project, in my not so humble opinion, would gain little
benefit, if any at all.
The reason why I would like gretl to be more widespread is because I would
like to have a larger pool of people who can give valuable contributions
to the project in terms of features. And the reason I feel this way is not
because I want to brag about the advanced stuff. It's because of what I
think the project's end goal should be.
As far as I'm concerned, the end goal is to provide a stats/econometrics
package which is (1) free; (2) advanced enough so that every conceivable
piece of applied economic research can be done by using it (at least in
principle); (3) friendly enough to be used in the widest possible range of
teaching situations. Please note: (3) would not be possible without (2):
the emphasis on math/stat functionality is a necessary condition to make
your package useable in teaching (past the undergraduate level). The
converse doesn't work.
A million users whose command of statistics and programming is near-zero
(and who, reasonably, find it odd that you can use the comma in excel but
not in gretl) are less important, to the reaching of the end goal, than a
dozen users who can find bugs and write meaningful bug reports, who follow
the recent trends in applied literature and ask the developers "could we
add this estimator/test/whatever?" or, better still, do it themselves and
upload a function package. The situation has improved over the years, but
the pool of contributors we have as of today is very small compared to,
say, the R project or even vintage stuff like RATS or TSP (yes, I'm
deliberately omitting Stata).
Sure, numbers are important, but then: can you provide any example of a
stat package which is *frendlier* than gretl is as of now?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------