a merging problem

[MKA-4]

Hello, I have a merging problem. I administered several scales (emotional intelligenge scale, organizational commitment scale, job satisfaction scale, organizational citizenship scale) to the teachers (N=900) and school managers (N=100). All the scales were administered to all teachers and all managers. One of my research questions is "Do managers' emotional intelligence levels have an effect on teachers' organizational commitment levels?". Data pertainig to teachers and managers are at the different SPSS files. To execute a regression analysis (in SPSS or LISREL) I must merge the managers' emotional intelligence scores in the same file that teachers' organizational commitment scores exist. Because of the sample sizes were different, when I try to add managers' emotional intelligence scores to teachers' SPSS file, many (900-100=800) missing cases emerged. Would it yield correct results if I fill this missing cases with the series' mean. Also another problem is that in the row of a teacher there will be a managers' emotional intelligence score (under the column of managerEI). How I can tackle with this problem and are there alternative methods that I can use in this case.
I will be grateful if someone help me, Mehmet Karakus

====================To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Maguin, Eugene]

Mehmet,

I think you need to think of the data structure problem differently. The
most important questions are:

1) In the organizations from which you collected your data, does a manger
supervise groups of teachers?
2) Is every teacher in your sample sample supervised by exactly one manager
who is also in your sample?

Background. I am assuming the organization is structured so that groups of
teachers are supervised by exactly one manager. You came into the
organization and recruited teachers and recruited their managers to do your
study. This gives you a multilevel structure.

Assuming the just described multilevel structure, you are asking whether the
within group average teacher commitment is related to manager emotional
intelligence. To analyze this correctly, you may be able to use the Mixed
procedure in spss (other alternatives are EQS, Lisrel, Mplus, and HLM, among
others). The question I have with respect to mixed is whether you can treat
this as a multilevel regression problem where the level 2 variable is
continuous rather than categorical. Perhaps others can comment.

Again assuming the multilevel structure, the data management problem is that
you have to have a variable linking manager and his/her teachers. If you
have that you can use the table subcommand (of the match files command) and
create a 'long' file.

To start, what are the answers to questions 1) and 2) above.

Gene Maguin

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD