Hello
I'm hoping someone could please point me in the right direction to assess for overdispersion in binary data - I'm using the melogit command currently (mixed effects logistic regression). I found some literature suggesting (from what I can understand!) that a beta binomial model could be used to test the difference but from here I can't seem to work out what I should be doing.
I have evaluated for overdispersion in a mixed effects Poission distribution model (mepoisson) using the menbreg command and interpreted the /alpha as being non-significant as the CI passes through 0 (IRR = -0.17, ci = -0.3 - 0.033). Therefore I am reading this as the data isn't over dispersed so I should keep with the Poisson distribution?
Would really really appreciate any advice as I'm on struggle street here!
Thank you!
Fitzroy
I'm hoping someone could please point me in the right direction to assess for overdispersion in binary data - I'm using the melogit command currently (mixed effects logistic regression). I found some literature suggesting (from what I can understand!) that a beta binomial model could be used to test the difference but from here I can't seem to work out what I should be doing.
I have evaluated for overdispersion in a mixed effects Poission distribution model (mepoisson) using the menbreg command and interpreted the /alpha as being non-significant as the CI passes through 0 (IRR = -0.17, ci = -0.3 - 0.033). Therefore I am reading this as the data isn't over dispersed so I should keep with the Poisson distribution?
Would really really appreciate any advice as I'm on struggle street here!
Thank you!
Fitzroy
