The obtained values of the likelihood ratio test ( 〖LL〗_(NB2-p)=403.62 and 〖LL 〗_(NB1-P)= 188.68 ) are greater than the theoretical one ( 〖χ^2〗_(2α;1 ) = 2.706 ) for both NB1 and NB2 models in comparison with Poisson regression.The resultsunderline that NB1 and NB2 models give a better fit of the data as opposed to Poisson regression. The remaining comparison between the negative binomial models indicates that NB2 model is preferred here to NB1 as it has higher log-likelihood ( 〖LL〗_NB2=-54942.9562>〖LL 〗_NB1= -55050.4282 ). The validity of these statements can also be confirmed by the information criteria. The lowest values of AIC and BIC comparative-fit tests are obtained for the NB2 model which underlines that this one is chosen to the detriment of both NB1 and Poisson models.