I have heard about GEE. I liked it and I used it. Next, I remembered my statistics professor's advice: keep it simple.
After my xtgee [...], family(nbinomial) link(log) corr(uns) robust with a strongly unbalanced panel:
Estimated within-id correlation matrix R:
c1 c2 c3 c4 c5 c6 c7 c8 c9
r1 |1,00
r2 |0,02 1,00
r3 |-0,26 -0,17 1,00
r4 |0,60 0,49 -0,68 1,00
r5 |-0,14 -0,64 -0,53 -1,00 1,00
r6 |0,00 -0,34 -0,19 0,45 0,37 1,00
r7 |0,50 0,10 -0,05 -0,20 0,49 -0,12 1,00
r8 |0,63 0,13 0,01 -0,33 0,49 0,05 0,34 1,00
r9 |-1,00 -0,35 0,30 -1,00 0,06 -0,44 -0,19 -0,26 1,00
Question: do I really need GEE? Or can I use random effects? Is there any test to help me?
Thanks,
Nicola
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
