In this paper a nonstatistical framework for assessing behaviorally based policy models is presented. Such a framework is needed because statistical criteria do not explicitly indicate how well various aspects of a disaggregated model fit together. The framework assumes models have five elements: behavior specificity, motivation or goal specificity, decision-heuristic specificity, belief specificity, and exogenous variable specificity. After the elements have been defined, three models are assessed using the framework. It is found that the framework works well across diverse models but requires considerable refinement and empirical verification.