Institute for Software Research
School of Computer Science, Carnegie Mellon University
Validating Computational Models
Kathleen M. Carley
Center for the Computational Analysis of Social and Organizational Systems
This report supersedes
The use of computational models in the social sciences has grown quickly in the past decade. For many these models represent a bewildering and possibly intimidating approach to examining data and developing social and organizational theory. Few researchers have had courses or personal experience in the development and building of computational models and even fewer have an understanding of how to validate such models. And while many papers extort the relative advantages and disadvantages of the computational approach, and many call for the validation of such models, few provide insight into how to validate such models and the issues involved in validation. This paper represents an attempt at redressing this oversight. An overview is provided of computational modeling in the social sciences, types of validation, and some of the issues in doing model validation.