RE: st: Display of missing by groups

[Fred Wolfe <fwolfe@arthritis-research.org>]

I used the summarize program as an example that would be understandable to 
all, but the issue is broader as it affects many programs, including the 
one that follows:

. bys ranada,rc0:fsum age sex haq_ difhaq_,s(mean sd)


-> ranada = 0

                            Variable 
|        N     Mean       SD 
-------------------------------------+---------------------------
                         Age (years) |     4627     2.18     0.65
                        Sex (% male) |     4631    24.29
                           HAQ (0-3) |     4608     1.06     0.67
              HAQ (0-3) (difference) |     4600     0.17     0.41

-----------------------------------------------------------------------------------------------------------> 
ranada = 1

                            Variable 
|        N     Mean       SD 
-------------------------------------+---------------------------
                         Age (years) |     4898     2.21     0.66
                        Sex (% male) |     4907    19.81
                           HAQ (0-3) |     4876     1.08     0.70
              HAQ (0-3) (difference) |     4869     0.18     0.40

-----------------------------------------------------------------------------------------------------------> 
ranada = .

                            Variable 
|        N     Mean       SD 
-------------------------------------+---------------------------
                         Age (years) |       83     2.20     0.64
                        Sex (% male) |       83    31.33
                           HAQ (0-3) |       79     1.22     0.73
              HAQ (0-3) (difference) |        0


-rc0- doesn't fix the problem because there is no error (see above).

-tabstat- is a superb program, but doesn't allow the kind of variable 
formatting I need for publication quality tables without a lot of extra work.

In answer to Nick's question about what missing implies, in this analysis 
of a controlled clinical trial it implies that some subjects were not 
randomized [ranada is randomized to a treatment called ADA]. In the example 
above -ranada- is dummied in, a place holder for when the randomization 
code is available.

So I guess my wish was just to have a simple way to drop the missing group 
in a Stata command option.

For my current use, preserve, drop, and restore work best and are easiest 
because I want to retain all subjects in the data set for other analyses.

Fred


At 08:56 AM 2/8/2006, Nick Cox wrote:

<snip>

> Also, missing on some but not all of the variables
> implies what?
>
> Nick
> n.j.cox@durham.ac.uk


> David Harrison
>
> > I don't see why this would be a problem, as -by- already has
> > the options
> > -sort- and -rc0-.
> >
> > Still, I would go with -tabstat, by()- anyway, as the output is
> > formatted much more attractively.

Fred Wolfe
National Data Bank for Rheumatic Diseases
Wichita, Kansas
Tel (316) 263-2125     Fax (316) 263-0761
fwolfe@arthritis-research.org


*
*   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/