Actually, I like the idea of sharing replications and discussing
potential problems on this level. I could imagine that this would
ensure high quality discussion as well as liveliness in the forum.
Best,
Artur
2011/5/3 Riccardo (Jack) Lucchetti <r.lucchetti(a)univpm.it>:
On Mon, 2 May 2011, Allin Cottrell wrote:
>> OTOH, I'm not overly enthusiastic to have general econometric
>> discussions on gretl-users. It's true that the traffic here is
>> not high, and this is also due to the fact that in the past
>> off-topic (=not gretl-specific) questions have been discouraged.
>> But this low traffic makes it possible that a relevant number of
>> subscribers are actually following most (if not all) threads. I
>> fear that if general econometric questions are discussed (which
>> I believe will be quite basic most of the time) then some people
>> might pay less attention to the list in general, with
>> detrimental effects on the gretl-specific issues.
>
> I think this point is well made. I'm still mulling the issue over,
> but I tend to agree with you.
I don't know.
On the one hand, Sven's point carries weight. But on the other hand, I
suspect that the mere existence of a "gretl-methods" mailing list would lead
to a wild increase of the "will-you-do-my-homework-for-me"-type we posts we
sometimes see on this list. If that was the case, would you follow that list
as closely as you follow gretl-users today?
Over the years, I've seen several attempts at setting up public discussion
places for debating econometrics at a professional level: mailing lists,
forums, you name it. All have failed, for one reason or another. In my view,
for a methods-related list to have success, you need to have two things
happening: (1) keep it high-profile (2) keep it alive.
In order to achieve (1), the list needs to be moderated, so messages with
subject "What is this cointegration thing people keep talking about?" would
be silently zapped before they see the light of day. But then, no-one
subscribes to a list where a new subject appears once every four months, so
||: no-one subscribes, no-one answers, no-one posts :|| (sorry for the
musical notation).
A slightly different project may be more feasible and, possibly, have more
chances of success (not to mention the fact that it may be a nice selling
point for gretl): a list centred on replication of published methods and
results which, occasionally, may host discussion like the one started by
Talha about the trace test.
Allin and I have used replication as an indirect way to check gretl's code
for ages, and I believe that there's lots to be learned from a systematic
replication activity. Moreover, more and more journals have now adopted the
policy that (I believe) the JAE inaugurated a long time ago, that is, making
datasets and possibly code available. So this is the incentive for READING
the messages.
What would the incentive for WRITING messages be? Well, I can see one, but
maybe there are some more: if you're starting work on something for which
the econometric/computational methods may be not so well established, the
first thing you want to do is replicate some published results to make sure
you're doing things ok. I don't know about you guys, but I do this all the
time. The tricky bit is that in some lucky cases you do get the same results
as in the published article, but in most cases you don't. A specialised
venue where to ask for help and advice would be most helpful. So, when you
do succeed in replicating some published results, do the world a favour and
post instructions (and possibly data+code) to "gretl-replication" (or maybe,
with a musical analogy, we may call it "gretl-covers").
Do you guys think this may work?
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users