I have been running a set of quasi-Poisson generalized linear models. As there is no AIC computed with quasi-distributions I was wondering how I could compare my models. It seems that I can use QAIC but I am really confused by all the reading I have done. Is there a simple function that I can use for that?

  • $\begingroup$ The R package AICcmodavg has a QAIC function & references Burnham & Anderson (2002), Model Selection & Multi-Model Inference. $\endgroup$ – Scortchi Aug 12 '13 at 16:47

Hy, Elena, jast look on deviance (& rdf), the smaller Dev. is the better! Gama pdf has additional parameter and may be useful. In my experience (hardest case: exess of zeros & overdispersion!) the PGAM package (in R) was a better choice ;-) NegBin & QPoisson + hurle models ... all are failed :-(

  • 1
    $\begingroup$ Where's the penalty for bigger models if you just look at deviance? $\endgroup$ – Scortchi Aug 12 '13 at 15:18

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.