Negative Binomial Models with Offsets

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
DuncanCM
Posts: 2
Joined: Tue Jun 05, 2018 2:02 pm

Negative Binomial Models with Offsets

Post by DuncanCM » Tue Jun 05, 2018 2:28 pm

Hi

I'm trying to run two-level negative binomial models. While there is no problem without offsets, when I include offsets MLwiN falls over immediately with the following error message:

“error while obeying batch file C:\Program Files (x86)\MLwiNv2.22\discrete\pre at line number 65:
calc ‘F~(H)’=expo(‘H’)
23171 numeric errors in calculate command, first 20 affected entries listed.
Affected entries replaced with system missing."

I'm working in a secure data lab and was running negative binomial models with offsets there two weeks ago without any problem.
This makes me think the issue could be the environment and the owners of the lab, who contract their secure service from someone else, are checking with their supplier if there have been any hardware/software changes.
In the meantime, I thought I would just check with yourselves in case you can think of anything obvious, though please don't spend too long on it at this stage as it could be the environment.

Many thanks
Craig

ChrisCharlton
Posts: 1018
Joined: Mon Oct 19, 2009 10:34 am

Re: Negative Binomial Models with Offsets

Post by ChrisCharlton » Wed Jun 06, 2018 9:06 am

The version that you are using (v2.22) is fairly old (see http://www.bristol.ac.uk/cmm/software/m ... fixes.html for a version history), so it seems unlikely that if the software in your environment had changed recently this version would have been installed. Are the models.data that failed exactly the same as the ones that previously worked? If not then it could be an issue with the the newer data, such as the offset not being logged.

DuncanCM
Posts: 2
Joined: Tue Jun 05, 2018 2:02 pm

Re: Negative Binomial Models with Offsets

Post by DuncanCM » Wed Jun 06, 2018 12:36 pm

Thanks for your reply Chris.

The model I'm trying to run is really the same, though the offset variable is calculated slightly differently (i.e some individuals have slightly different values) but it was logged.

Another puzzling thing is that I did retrieve a previously stored worksheet with a model using the original offset that had run fine and it fell over with the same error message when I used 'more' to do an extra iteration with no other changes.

Craig

ChrisCharlton
Posts: 1018
Joined: Mon Oct 19, 2009 10:34 am

Re: Negative Binomial Models with Offsets

Post by ChrisCharlton » Wed Jun 06, 2018 12:47 pm

That is odd then. I can only suggest that you check the data and model in case you accidentally changed something before saving, or something wasn't saved correctly. Otherwise it's back to the theory that the software version has changed since you last ran it (maybe there is more than one version installed on the computer?).

Post Reply