Algebraic notation for variance in negative binomial model

Welcome to the forum for MLwiN users. Feel free to post your question about MLwiN software here. The Centre for Multilevel Modelling take no responsibility for the accuracy of these posts, we are unable to monitor them closely. Do go ahead and post your question and thank you in advance if you find the time to post any answers!

Remember to check out our extensive software FAQs which may answer your question: http://www.bristol.ac.uk/cmm/software/s ... port-faqs/
Post Reply
LiamMS90
Posts: 3
Joined: Sat Jul 28, 2018 1:56 pm

Algebraic notation for variance in negative binomial model

Post by LiamMS90 » Fri Oct 05, 2018 11:37 am

Dear Colleagues,

I'm using version 3.02 to run a two level negative binomial model. There seems to be a difference between the algebraic notation for the level 1 variance displayed in the equations window and the notation in teaching materials referencing earlier versions of mlwin and more generally in the literature.

In 3.02 I see..... pi + pi^2*a

Elsewhere I see.... pi + pi^2/v

Is this a bug or is there another explanation?

Many thanks,

Liam

ChrisCharlton
Posts: 1090
Joined: Mon Oct 19, 2009 10:34 am

Re: Algebraic notation for variance in negative binomial model

Post by ChrisCharlton » Mon Oct 08, 2018 9:14 am

Hi Checked this with Professor Kelvyn Jones who initially requested the change in notation and he provided the following explanation:
The equation is now correct- there are two ways of doing this and we have always estimated it this way but the specification used to be incorrect

It is what Hilbe calls NB2

https://www.encyclopediaofmath.org/imag ... t_data.pdf

“This form of negative binomial has been termed NB2, due to the quadratic nature of its variance function “
image001.png
image001.png (13.87 KiB) Viewed 1541 times
NBD2 has 2 parameters for the level-1 variance; that is quadratic level-1 variance,

α is the over-dispersion parameter; if α > 0 then NBD; if α=0, NBD reduces to Poisson;

in stored model results α estimate is associated with Bcons2.1 label

Lips data
image002.png
image002.png (14.88 KiB) Viewed 1541 times
And results
image003.png
image003.png (14.97 KiB) Viewed 1541 times

Post Reply