Bug on AIC/BIC calculation?

Dear All,

I noticed something strange when doing a covariate search in Phoenix NLME (v 1.3).

When looking at the results, I noticed that the number of parameters is the same, altought it is pretty clear that in one case there are more parameters as on the other case (more covariates). Is this situation known to you? If the nParameters is not well calculated, I cannot trust the BICs and AICs.

I cannot send you the project as the data is confidential.

Best Regards,
Joao

Dear Joao,

it appears that irrespective of the scenario name the model seems to be the identical.

You could send your project file to support@certara.com where we can treat the data as strictly confidential.

Best regards,

Bernd

Dear Berndt,

Thank your for the quick reply. It is really not possible for me to send you the project file.

I tought that maybe this would be a well-know problem for you guys. I will try to figure this out by myself.

99% of the times it is not a software, but a user problem :slight_smile:

Best,
Joao

Hi Joao,
Certara has a confidentiality agreement with many parties chances are we already have one with you ?
What is the official release build you are using we need more than 1.3 ( in the about screen you should have more information)

As Berndt mentioned all the -2LL are identical so it is not just a nparam AIC issue.
You can maybe just post the model code for two models that you think should have a different number of parameters but they are showing identical ? or send us the project after you remove the data from it keeping all models and model comparers objects

.