This is a question for StataCorp, so I
have to guess at an answer. I doubt that
there is a strong programming objection to what you
suggest. Rather, the aim of -adjust-
is to tabulate sets of predictions, and
what you suggest would make the default
output somewhat trivial. In other words,
it looks like a design issue, not a syntax one.
The first and last time I wanted predictions
as function of one covariate, all others
being set to their means, it irritated
me that you had to talk your way past
that -by()- requirement. So I wrote a wrapper
that does it for me, but hesitated at
making that public, given the labour of writing
the help and a worry that it might not be
general enough to bear the weight some users
might put on it. That was January 2003,
and I don't recall anyone raising the
matter on Statalist before now, so it's not clear
how often this is needed.
Nick
[email protected]
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Richard
> Williams
> Sent: 08 May 2004 19:54
> To: [email protected]; [email protected]
> Subject: RE: st: plot predicted effects after regression
>
>
> At 06:58 PM 5/8/2004 +0100, Nick Cox wrote:
> >You can use -adjust-: you just need to talk
> >your way past the requirement for a -by()-
> >option (unless that is part of what you want).
>
> As a sidelight, I've never liked or understood the requirement that
> -adjust- include the -by- parameter; why not just default to
> analyzing all
> cases at once instead of requiring that calculations be done for
> subgroups? It can be worked around, but it is a minor
> nuisance so why
> require it in the first place? If dropping the -by-
> requirement would
> create some sort of compatibility problem, perhaps some sort
> of optional
> parameter to analyze all selected cases could be added instead.
*
* 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/
