[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
FAQs [was: RE: st: RE: RE: Calling a Do-file]
"Nick Cox" <email@example.com>
FAQs [was: RE: st: RE: RE: Calling a Do-file]
Thu, 23 Oct 2008 19:16:05 +0100
I am not sure that this is correct. But it raises a question about what
people think the role of the FAQs is. I often see comments on this made
in passing on the list. They frequently puzzle me because my perception
of what the FAQs are intended to do is often quite different from what
people seem to be implying.
Therefore, although this is a response to Sergiy's post, it is not
directed to him personally.
The issue is starkest when people say something like "I searched the
FAQs and couldn't find anything on this" and that appears to have been
their prior search! Or when people write personally and say "I can't
find an FAQ on this" they really have the wrong end of the stick. (They
are often right, but that is not the main point.)
I mean here the Stata website FAQs. (There are several other sites
offering Stata FAQs but -- on this occasion it is fair to single out
just one other -- the excellent UCLA FAQs are the main alternative. My
comments are not intended to apply to them, but I suspect the point
As usual, this is just a personal view. But I have a perspective on this
through single or joint authorship of various Stata website FAQs and so
have thought a fair amount about what the FAQs can and can't do.
First off, we can presumably take it that all Stata users have access to
the on-line help. There will be all sorts of small questions about
whether people can find what they want in the places they look, but
those are on the side.
Second, it so happens all Stata users using the internet have access to
the Stata website FAQs (or so I presume -- perhaps someone will now
explain that their firewall lets them see some sites but not others).
However, from the point of StataCorp's design it is pretty clear to me
that the hierarchy of documentation runs more or less like this
Stata Press books by StataCorp people FAQs
Stata Press books by others other Stata
The FAQs are definitely to be considered _after_ the manuals! What's
more, the FAQs are not intended to give systematic coverage of anything.
That's the job of the manuals.
The FAQs are intended to fill in some of the gaps that the manuals
leave, or to fill out [sic] the manuals by giving extra coverage of
points people find unclear, etc., etc.
Also, the FAQs have a function of publicising points very, very quickly
(e.g. teething troubles with new releases).
Of course, whether you or your institution can afford the manuals and
whether you can find the time to read the manuals carefully are key
practical questions. But that's what they are: practical questions.
But to repeat I don't think anyone designs the FAQs to give systematic
coverage of anything. On the contrary, most are written ad hoc in
response to repeated queries. Once the same point has arisen repeatedly
in questions to Stata Technical Support, it becomes more efficient to
write it up as an FAQ. Many of the user-written FAQs on the Stata
website arose in the same way, as the same questions on Statalist gave
rise to the same answers and it seemed a good idea to polish up the
answers and make them more accessible. But there's a real capriciousness
about what gets written up. Occasionally I write to people privately and
say "That was a great posting. Why not make it an FAQ?" but if they
can't bothered that will be the end of it.
There are funny little ups and downs in what people ask about on this
list. At the moment I am sensing quite a lot of interest in how to learn
Stata programming. I want to be encouraging -- it's easy!, and as Kit
Baum says a little programming can go a long way -- but if people think
they can learn about Stata programming by reading up some help files,
and some FAQs, and looking at the Statalist archives, they are going to
find that very frustrating.
To learn Stata programming successfully, you need to do most if not all
of the following:
1. Read the manuals comprehensively, carefully and repeatedly. (Not all
of them, but enough!)
2. Take a NetCourse, or some other equivalent book or course.
3. Look at lots of Stata code carefully.
4. Have some experience with programming already.
5. Be prepared to spend a lot of time making mistakes.
This isn't beyond the wit of any smart user and hundreds of people have
done it, but newbies out there who hope that reading a few FAQs or other
short documents will explain how to do it are likely to be disappointed.
In Stata 'variables' are columns in the dataset (rows
are called 'observations') and there is only one dataset that Stata
may work with at a time. 'locals' and 'globals' are equivalents of
what is called 'variables' in other programming languages. This is
explained in the FAQ somewhere.
* For searches and help try: