st: Why does one command "lose" observations compared to another command?

[Suzy <scott_788@wowway.com>]

Dear Stata Listmembers:

I have come across a problem that I can't figure out. I don't know why I 
am "losing" 873 observations with one stata command versus the other -  
for mean age by race, if age>=18. Both age and race are 100% present in 
the total dataset - total (N=21004) and for age>=18 (N=11,441).

This is complex survey data and of course, I would use the survey 
commands for analysis, but it still doesn't make sense why the 
non-survey command would use less observations (even so, the point 
estimates are identical.

Any insight into this would be greatly appreciated.

Best regards,
Suzy

Here is the output for both:

svy, subpop(if age>=18): mean age, over(race)
(running mean on estimation sample)

Survey: Mean estimation

Number of strata =      28          Number of obs    =   21004
Number of PSUs   =      57          Population size  = 2.8e+08
                                   Subpop. no. obs  =   11441
                                   Subpop. size     = 2.1e+08
                                   Design df        =      29

           1: race = 1
           2: race = 2
           3: race = 3
           4: race = 4
           5: race = 5

--------------------------------------------------------------
            |             Linearized
       Over |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
age          |
          1 |   46.87004   .3829944      46.08673    47.65335
          2 |   42.29357   .4316838      41.41068    43.17646
          3 |   36.81336   .5968429      35.59268    38.03404
          4 |   41.05272   .9506344      39.10846    42.99699
          5 |   41.83675   1.234968      39.31096    44.36254
--------------------------------------------------------------

mean age [pweight = mecwt4] if age>=18, over(race)

Mean estimation                     Number of obs    =   10568

           1: race = 1
           2: race = 2
           3: race = 3
           4: race = 4
           5: race = 5

--------------------------------------------------------------
       Over |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
age          |
          1 |   46.87004   .2514365      46.37718     47.3629
          2 |   42.29357    .343518      41.62021    42.96693
          3 |   36.81336   .2793463      36.26579    37.36094
          4 |   41.05272   .9437899      39.20272    42.90273
          5 |   41.83675   .7637752      40.33961    43.33389
--------------------------------------------------------------

.


*
*   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/