Reporting overdispersion for NB regre... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Blazej Mrozinski posted on Saturday, May 11, 2019 - 5:09 am
Good morning Professors Muthen, and good mornign Mplus community.

Trying to comfort reviewers i'm trying to explain that my outcome variable is overdispersed and NB regression is a better choice than Poisson.

Could you please tell me how can I report the Dispersion result found in NB regression?

Estimate 10.774 SE 1.396 Est / SE 7.718 two tailed p. 0.000

Also is it appropriate to report AIC/BIC model fits from NB and Poisson?
Is there a (syntax) way to compare two models and numerically show one fits better than the other?

I'd be very helpful for any hints on this.
Thank you
 Bengt O. Muthen posted on Saturday, May 11, 2019 - 6:54 am
When the dispersion is significantly different from zero, it is needed.

Yes, AIC/BIC is fine - as with all ML analysis.

If you want a test that one model is significantly better, the two models should be nested, e.g. by one model having several of the other model's slope parameters fixed at zero - you can then do a regular LRT chi-2 test.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: