Dear AllStaters -
I am trying to run a GEE in SAS 6.12 on a 7-wave panel dataset which is set
up in univariate format, i.e.,
case1 Time1 y1 x11 x21 x31 x41 x51
case1 Time2 y2 x12 x22 x32 x42 x52
case1 Time3 y3 x13 x23 x33 x43 x53
Etc.
I am new to panel analysis, and am unclear on how one deals with variables
that are only measured in certain waves.
Concretely, I have 7 waves, but am particularly interested in a pair of
independent variables (let's call them A and B) that are measured only in
Waves 5-7. But my response variable, as well as all of the other
independent variables, are measured in all 7 waves.
Adding A and B to my 7-wave GEE results in the dropping of information from
waves 1 to 4. In other words, my 7-wave panel model reverts to a 3-wave
model. This is a disaster!
This issue is presumably pertinent to all manner of panel models, not just
GEE. Is there any way to deal with this problem? What is the reference
term for this kind of problem?
Thanks,
Dave Sorensen
Rutgers University and University of Copenhagen
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|