I will test whether there is a statistically significant difference between Negative Binomial Regression (GLMs) and Geographically Weighted Negative Binomial Regression (GWNBR).
However, the value of the effective number of parameters in GWNBR cannot be calculated. This results in not being able to calculate the degree of freedom (df) from GWNBR.
Do you know an alternative to the F test? or do you think there is no alternative test that does not involve degrees of freedom?