Re: st: tabstatmat question

[Nick Cox <njcoxstata@gmail.com>]

-tabstatmat- (SSC) is a user-written program. You are expected to explain that.

It just feeds on the last set of r-class results. It cannot look back
in your past to see previous results. The effect of -by:- as a prefix
command is to run -tabstat- repeatedly and to overwrite saved results
each time.

If you issue -tabstat- with a -by()- option you should have more success.

Nick

On Thu, Sep 1, 2011 at 8:30 PM, Alvarez,Sergio <sergioal@ufl.edu> wrote:
> Dear Statalisters,
>
> I'm trying to create a matrix of the grouped means of a variable (num_typ3).
>  The observations in my dataset are grouped by location (zone) and time
> period (wave).
>
> Ideally the matrix would have the zones as the rows and the waves as the
> columns (or vice-versa), something like this:
>
>         WAVE
> ZONE      1          2       ...
> 1      mean(1,1)  mean(1,2)
> 2      mean(2,1)  mean(2,2)
> ...
>
> I would use the matrix to generate a variable that takes the mean value
> depending on the zone and wave, using something like:
>
> gen meancatch = TABLE[zone,wave]
>
> I believe I need to go this route, rather than directly creating the
> variable, because I'm creating a set of 'fake' alternatives for each 'real'
> observation, corresponding to each location.  Each observation corresponds
> to a fishing trip, so the alternatives are the possible fishing trips that
> didn't happen.  I'll do that using -reshape-.
>
> Anyway, I thought -tabstatmat- would do the job, so I tried:
>
> by zone wave, sort: tabstat num_typ3, stat(mean) save
>
> Which gives me the elements of the ideal matrix one by one.  Then I try:
>
> tabstatmat TABLE
>
> But this returns a 1x1 matrix:
>
> symmetric table[1,1]
>       num_typ3
> mean  .94892632
>
> So my question is, where did I mess up?  Or did I just misunderstand what
> -tabstatmat- does?  Is there a convenient way to do this?
>
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/