I am running a mixed model in SPSS. I have one fixed effect and 3 random effects. I am testing this model on various traits to determine if the fixed effect is significant.
For most of the traits I am testing everything works fine. But for some traits, I get the warning:
The final Hessian matrix is not positive definite although all convergence criteria are satisfied. The MIXED procedure continues despite this warning. Validity of subsequent results cannot be ascertained.
One random effect (which is the interaction of the other 2 random effects), says that it is a redundant parameter, and thus sets the variance to 0.
- I use the same model for all the traits I examine, so I am trying to understand why it only happens for some of them, and what exactly it means?
- I am new to mixed models. What makes a parameter redundant, and can I still trust the model?
Also, I wanted to make sure that I am interpreting the output correctly.
- Under "Estimates of Covariance Parameters", is the significance value associated with Wald's Z providing me with information about the significant contribution of the random effect on the trait examined?