I am using SPSS to model count data with a Poisson distribution. My initial Poisson model and the default Negative binomial model showed over dispersion and under dispersion respectively. I am fitting a third model with custom negative binomial with log link function to estimate dispersion more accurately. However, I get this message after running the model: "There are no valid cases for the log link function. Only the iteration history is displayed. Execution of this command stops". Is it okay for me to go with the best of the first two models or there is a way around the problem with the third model (in which case what should be done)?
When running a custom negative binomial regression with an estimated dispersion parameter in SPSS, you may encounter iteration problems due to the complexity of the model. Here are some steps you can take to resolve these issues:
By following these steps, you can help to resolve iteration problems when running a custom negative binomial regression with an estimated dispersion parameter in SPSS.
Eman F. Mohamed Hassan Al-Zoubi Asad Freihat
When running a custom negative binomial regression with an estimated dispersion parameter in SPSS, you may encounter iteration problems due to the complexity of the model. Here are some steps you can take to resolve these issues:
By following these steps, you can help to resolve iteration problems when running a custom negative binomial regression with an estimated dispersion parameter in SPSS.
Thank you Hammad. Your answer was really helpful. I realized the problem might be from the number of predictor variables as you pointed out. I could't do anything about that so I rather fitted a quasi-poisson model in R which worked perfectly with very good fit indices. Thank you once again.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Inorganic Materials
- Ceramics
- Glass