[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: RE: Stata Wish List
Nick Cox wrote (excerpted):
> I'm interested in learning if anyone
> thought that anyone else was claiming that
> statistical program output _was_ typically
> suitable for straight reproduction in reports.
To respond to Nick's request, I never expected that statistical program output
was suitable for direct reproduction in the body of a report, although SAS with
its ODS has gone a long way in that direction.
If there were only a few commands' output to deal with, I've done the
transcription manually, with manual inspection (by me and one other person) for
transcription errors before the report is issued.
For many repetitive analyses, I've written Stata programs ad hoc (-program
define printem-, -program define displayem- up at the top of a do-file to be
called later on in the do-file after each -regress-, -means- or -fsum-, for
example) that open a smcl log file (whose name is one of the program's
arguments), take the results returned in macros, scalars, coefficients (_b,
_se) and statistics derived from coefficients and their p-values, etc., and
-display- them formatted (%X.Xf, whatever) with variable labels and laid out as
I want in smcl in the log file. At the end of the analyses, the do-file
converts all of the smcl log files with log2html. In accordance with clients'
overwhelming preference, I use Microsoft Word for reports; one hurdle that I've
had is delimiting the output in smcl with tab characters in order to be able
eventually to convert text-to-table in Word. Tab characters seem to get
converted into spells of spaces in smcl, which get carried over as such into
html and into Word. When I try to incorporate the output of -table- or
-tabulate- in an analogous way, likewise, I have to deal with a variable series
of space characters instead a tab character in the imported html file should I
wish to convert it into a table in Word.
> Of course, none of this stops users or
> StataCorp giving us tools for the report
> generation stage.
Such a capability is one of the ostensible advantages of SAS, and one reason
(and not necessarily a minor reason) why it maintains dominance in certain
spheres, for example, the pharmaceutical industry. This capability would
certainly be a selling point for Stata to have, but there are other items on my
wish list that are higher up (capabilities in mixed models, double-byte
character data, and perhaps XML), all of which SAS also has over Stata.
* For searches and help try: