How to decide between quasi-poisson and negative binomial?

Asked Mar 13 '16 at 19:40

Active Mar 13 '16 at 19:40

Viewed 317 times

I tried quasi-poisson and negative binomial glms on my counted data in R. The estimates are pretty much the same but p-value are different. Quasi-poisson gives insignificant result. NB give significant result. I am not sure how to decide between the two models. I computed QAIC for quasi but I read a paper saying that it is created to compare within quasi family. Another one says we should compute to compare with AIC. Is there any other way to compare the two models?

                        Quasi-poisson   NB
Deviance/DF              994.93       3.0014

Person chi-square test   995.64       2.997

QAIC/AIC                 5.041        68.47

asked Mar 13 '16 at 19:40

user220055

Did you make some residual plots? – kjetil b halvorsen Aug 27 '17 at 12:37
See the links at https://stats.stackexchange.com/questions/478209/quasi-poisson-vs-negative-binomial – kjetil b halvorsen Apr 20 '22 at 02:39

How to decide between quasi-poisson and negative binomial?

0 Answers0