Dear SPSSers,
I'm doing a research where I've choose randomnly 20 clusters (districts) and within them were selected about 100 childs (CHILDID) under a systematic sampling. I consider that the observations within the clusters are not independente because children under the same cluster are similar because of the socioeconomic characteristics. I considerar doing an analysis with GEE, on the sintaxis in SPPS I've written:
/REPEATED SUBJECT=Cluster WITHINSUBJECT=CHILDID, but I have doubts if this is correct or not, also I have difficulties choosing which is the right matrix job. I know if I make mistakes using the wrong Working Correlation Matrix this isn't a big problem if I use COVB=ROBUST. However, when I used CORRTYPE=INDEPENDENT X2 it was significative, but when I considered CORRTYPE=EXCHANGEABLE, X2 it was not significant. This makes me think that I'm not making the right decision. I would appreciate if you could guide me about it.
Dependent variable=Y
Categoric independent variable=XCAT
Continuous confounding variable=X2
Thanks in advance
Magi
=====================
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
