Detecting outliers in Negative Binomial Regression

[steph88]

Hi all!

I am conducting a negative binomial regression for my statistics project. I have been looking online about how to detect outliers but am not finding anything. I know that when using linear regression, the mahalanobis distance is used. However, since negative binomial regression is non-linear, how would I detect outliers in the data? Is there another technique or can I just use mahalanobis distances?

Thanks!
Steph

 

[jrai]

Very interesting question. I'd also like to know answer to this!

 

[BGM]

I have just check the book "Negative Binomial Regression". In chapter 5: Assessment of Count models, it also mention (Pearson) Residuals, hat statistics, and leverage. Maybe you can take a look.