Consider the manuals
====================
I sense in some contributions to this thread
a tacit desire for a site or sites that will
act as a substitute for the manuals. In contrast,
for example, the StataCorp FAQs are mostly designed
to go _beyond_ the manuals.
Whether that reading of this thread is correct,
I've also sensed a sharp upturn over the last year or so
in the number of questions on Statalist that seek to
by-pass the manuals. How often have you seen postings
of the form "I've searched the FAQs and the archives,
and I can't work out..."?
Such comments are rather staggering to anyone who
grew up on (1) ask the local Stata expert "down
the corridor" (2) read the help (3) read the manual
before any of those, but they presumably reflect the way
in which many users are operating. (It is of course
commendable to do some searching.)
Now the manuals pose a variety of issues which are
themselves the subject of intermittent discussions,
and I'm fully aware that they are very long,
not cheap and in practice not easily accessible to many
users. But I really doubt that even keen Stata users acting
collectively will want to re-create the manuals.
Here is some avuncular advice, which like
most avuncular advice will be widely unwelcome. There is really
is no substitute for reading the manuals if you want to
become truly competent in Stata.
Human factors
=============
This thread has seen several posts from technological optimists
who know of nice software alternatives and appear convinced that
they offer Statalist much better ways of proceeding. I have got to
be sceptical because, as I've already said in other ways,
it is the human factors rather than the technological
solutions which are vital here, namely, who is going to
lead and how easy will it be to follow?
The question of who is going to do the work needs no
extra emphasis.
What about the members, namely you?
Statalist has been going for over a decade now. Judged by
those who get quick, clear, smart solutions to their problems,
and all for free, it is spectacularly helpful. Judged by
those who don't get any answer at all, or not the answer they
wanted, it is not helpful, but those people often get
good answers to other questions. As an ex-economist, I ask
what more could one expect from something that is free?
As statistically-minded people, we ought to be looking more
closely at the data before judging Statalist and its alternatives.
The data are a mix of quantitative and qualitative, as usual.
1. The people using Statalist -- that's you, naturally -- are
a pretty smart and well-qualified bunch of people. A wild guess
is a median of a Master's degree with much quantitative content,
and lots of doctorates everywhere. These people are not going
to change much except that as Stata continues to grow the
average level is likely to go down a bit.
2. Learning how to use Statalist is really pretty easy, by
technical standards:
* You need to send an email to join.
* You should read a single document packed with advice.
That's the FAQ, and although you may not believe it
we struggle to keep it concise as well as informative.
(Occasionally I ponder highlighting ten or so points
that you really must consider, but that could do as much
harm as good.)
* You need to send emails to contribute.
As the FAQ is hardly more complex than most of the
technical material members address in their work, and as small
children these days can email each other, there
really shouldn't be any difficulty there.
3. Nevertheless, there is evidence just about every day
that even point 2 poses severe challenges to many members.
That is, many of the small problems that Statalist shows
arise because users just aren't reading -- and as Kit Baum
often says, heeding -- the Statalist FAQ. By point 1, you can't be
stupid. It is less surprising that a fraction of Statalist
members don't understand how their mailers work or the
differences between plain text and HTML. More to the point,
most of the problems on Statalist arise because people are
too busy to check all the things they should be checking.
(Occasionally I have to guess that laziness plays a part
too, but then I too can be very stubborn about not mastering
details that don't interest me.)
In short: there is plenty of evidence that Statalist is already
complicated enough for many of its members, which makes any
suggestion of something more complicated unattractive to me.
Summary
=======
Let me summarise the situation as I see it:
1. I would be very surprised if StataCorp were willing,
even in principle, to take over Statalist. Statalist was born
as a user initiative and StataCorp have always been positive
about it -- and would be the first to admit that they benefit
mightily. And some StataCorp postings on Statalist give information
available in no other form. But they really don't want to run
the show at all.
For one, that would set up a very awkward set of questions
on whether StataCorp should support those who do not have
a licence, who want to ask questions about user-written software,
who want to be told what next to do in their project, who want to
be told what is statistically (or even econometrically)
correct, and so on.... If _you_ were StataCorp, you
really wouldn't want those responsibilities.
(Hosting the archives and the FAQ is mostly a question of
convenience and practicality, as already said.)
2. Statalist isn't perfect, but what are your criteria
of success? That every question gets an answer? Do feel free
to answer the otherwise unanswered questions yourself! That
searchable archives think the way you do? I'd like that too.
(I often fail to find my own postings in the archives.)
That people always heed good advice? I'd really like
that too.
3. A move away from plain text would disenfranchise or
inconvenience many existing members.
4. Wikis for Stata could work very well. I think this is
really quite distinct from the question of futures for Statalist.
Nick
[email protected]
*
* 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/
