Can GEE handle a 2-level and 3-level hierarchical clustering, i.e. repeated measures per individual nested in cities, and cities nested in provinces? If yes, how can should I write the GEE for these stages? What should be the working correlation?

  • $\begingroup$ An interesting aspect of the GEE is that, since it effectively marginalizes over correlation structures, you can ignore them completely and apply an independent working covariance structure to obtain correct standard error estimates. This is usually exploited when correlation structures are unknown, like relatedness is small community studies. However, working independence structures are also used many times when correlation structures are known. $\endgroup$ – AdamO May 1 '17 at 17:46
  • $\begingroup$ thanks please tell me how i can construct correlation matrix for 3 level? $\endgroup$ – fatemeh May 2 '17 at 17:07
  • $\begingroup$ can i construct correlation matrix for level 1 and then to level 2 and then to level 3 and sum these correlation structures? $\endgroup$ – fatemeh May 2 '17 at 17:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.