Eventually, to determine whether the data is better modeled using NB2, we considered the likelihood ratio test discussed in Section 2. The log likelihood for the full model is 〖LL〗_(NB2(7))=-54942.9562 and for the null model is 〖LL〗_(NB2(0))=-62236.1707. The likelihood ratio value obtained is LR=2(-54942.9562+62236.1707) = 14586.429 and since the full model includes seven predictor variables, the statistics theoretical value is 〖χ^2〗_(0.05;7 )=14.067. This yields a p-value < 0.0001, highlighting once more that the NB2 is the best model to adjust the basic Poisson algorithm in order to estimate our insurance data.