… is from page 215 of Maxwell Stearns’s 2000 book, Constitutional Process: A Social Choice Analysis of Supreme Court Decision Making (footnote deleted):
In evaluating the merit of an economic model, it is misleading to use terms such as correct or incorrect. Economic models are neither true nor false. Instead, the value of a positive economic analysis depends upon the extent to which the model generates meaningful insights about enigmatic phenomena in the real world. Because economic models are intended to be manageable images of reality, a tension necessarily arises between a model’s descriptive accuracy and its manageability. Developing an economic model, therefore, differs from developing a theory in one of the natural, or “hard,” sciences. While even a single contrary datum can raise questions about the accuracy of a genuinely scientific theory, even the most robust economic models are unable to account for a considerable amount of data. As a result, locating one or more data unaccounted for in an economic model may be of no more value in refuting the model than observing that the model and reality are different.