st: calculating statistics for sub-groups with missing data, etc.

[Ani <anirudh.ruhil@gmail.com>]

Hi All;

I have a dataset of supervisors and line-personnel (snapshot follows)

supervisor   line-personnel    Q1    Q2    Q3
1                     0                  3       3      4
1                     1                  2       2      5
1                     1                  5       5      5
1                     1                  4       .       3
1                     1                  1       3      .
2                     0                  3       3      3
2                     1                  2       4      4
2                     1                  2       1      1
:

Q1 -- Q3  are items about the supervisor, with responses on a
five-point Likert scale.
Q1 and Q3 combine to reflect behavioral construct "A", Q2 and Q3
combine to reflect behavioral construct "B"

The goal is to see (1) what supervisors say about themselves, (2) what
their line-personnel say about their supervisors, and then allow each
supervisor to compare their responses to those of (3) all "other"
supervisors and (4) all "other" teachers. I am working through the FAQ
and various postings to the listserv to do this cleanly and
accurately. However, I am getting tripped up by the fact that not all
questions are necessarily answered so the average score on construct A
for line-personnel is not necessarily summing across Q1 and Q3, then
dividing by 8 because there is at least one missing response, and
likewise for construct B there are two missing responses.

I'll keep poking around in the archives and experimenting with
solutions but if anything comes to mind, do share; it is always
sincerely appreciated!

best

Ani

*
*   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/