Current behavior genetic models can only employ limited types of phenotypic data, such as Likert scale data or continuous data, as variables. Here, a new method employing paired comparison data is presented within the framework of behavior genetic models. This model facilitates the estimation of genetic, shared environmental, and non-shared environmental contributions to paired comparison variables. Paired comparison methods are sensitive to differences in preferences between items, even when the true preferences are markedly similar. An extended model that combines Likert variables, and which enables the estimation of genetic, shared environmental, and non-shared environmental correlations, is also presented. Simulations are then performed to demonstrate the characteristics of these models. Finally, a real data example is introduced. Mplus script is included in the appendix.