All,
We are thinking of using mixed to analyze some data and I have a question
about how it handles certain issues. Let me also acknowledge that I don't
know how any other program handles the issue either and I'd really like both
a general answer and an spss specific answer--if they differ. OK. Model and
data.
The level 1 model is Y(i,j)=B0 + B1*VA + B2*VB + B3*VC + e(i,j).
VC is the interaction of VA and VB. B0 is random; B1, B2, and B3 are
constrained to have no level 2 variance. The level 2 model is an intercepts
only model.
Each level 2 unit consists of either 2 or 4 records. There is no missing
data. Furthermore, any 2, 3, or 4 records in a level 2 unit may have the
same Y value. So:
1) if all records have the same Y value, Y has no variance.
2) if a level 2 unit has just two records, VB and VC will be constants, thus
no variance here either.
Of course, both 1) and 2) may be true.
My question is how mixed handles these issues. I've looked at a couple of
books (Raudenbush and Goldstein) and I didn't see an obvious discussion of
these issues.
Can anybody educate me a bit or suggest a citation that covers these issues?
Thanks, 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
Locked
