The validity of computational models in organization science: From model realism to purpose of the model

“...even though the assumptions of a model may not literally be exact and complete representation of reality, if they are realistic enough for the purposes of our analysis, we may be able to draw conclusions which can be shown to apply to the world.”

Abstract

Computational models are widely applied to address fundamental and practical issues in organization science. Yet, computational modeling in organization science continues to raise questions of validity. In this paper, we argue that computational validity is a balance of three elements: the question or purpose, the experimental design, and the computational model. Simple models which address the question are preferred. Non-simple, imbalanced computational models are not only inefficient but can lead to poor answers. The validity approach is compared with well-known validity criteria in social science. Finally we apply the approach to a number of computational modeling studies in organization science, beginning with Cyert's. They were pioneering and are examples of well designed computational models.