Just feed me the sausage: one congressional staffer's view.

Abstract

Policymakers are often faced with many complex policy options and, as a consequence, need tools to distinguish among these options and to understand their effects and costs. The forecasting models that policymakers depend on to estimate these effects are numerous and varied and often produce inconsistent and undependable results. Because we live in a world of uncertainty, imperfect information, and constant change, these models are inherently flawed. Nevertheless, modelers do the best they can to estimate complex effects with limited data as well as limited empirical evidence on which to base a model's assumptions. As Sherry Glied and her colleagues demonstrated, we often lack needed or adequate data to model complex aspects of human behavior. Furthermore, the models are often black boxes, so the users of the results, such as myself, a congressional staffer, have no idea what data the models used or what assumptions they were based on.