we are dealing with overdispersed data, and thereby the p-values tell us nothing about the relationship of the predictor and response. Therefore, it is imperative to test the equidispersion assumption, meaning the equality between the conditioned mean and variance of claim frequency, when constructing and interpreting a Poisson model. In this context, the method of Cameron and Trivedi and the test of Greene are used. Table 4 shows the results obtained after estimating and testing the parameter