This paper compares the correlative and predictive abilities of three forms of models based on the approach which combines an equation of state (EOS) with a g(E)-model in an EOS framework. In particular, the MHV1 mixing rule with a quadratic mixing rule for the mixture b parameter, the LCVM method and a variation of the LCVM method have been investigated. Estimation of supercritical fluid-liquid (SCFL) phase behavior of mixtures involving carbon dioxide and ethane as the "light" compound and fatty acid esters (carbon number > 10) and hydrocarbons (n-Alkanes) as the "heavy" compounds have been considered. Finally, a generalized expression for the mixture ct-parameter, from which several other mixing rules (including the ones investigated in this paper) can be derived, is presented. The derivation of the MHV1 mixing rule and the LCVM method from the generalized expression is given together with the estimated phase equilibria.