Generalized joint modeling of mean and dispersion

Quasi-likelihood models are developed to analyze overdispersed data. These models involve the joint modeling of the mean and dispersion. No distributional assumptions are made for either the mean or the dispersion submodel in the joint modeling. Only a variance-mean relationship is assumed.;The performance of this modified function in analyzing overdispersed data is compared to some existing functions.;The joint modeling of the mean and dispersion has been addressed through the extended quasi-likelihood and the generalized extended quasi-likelihood function. These functions are modified and extended requiring less restrictions in the analysis of overdispersed data. The variance function for the dispersion submodel is assumed to have a form other than that of the gamma.