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

Abstract

Cohen and Richard M. Cyert (1965) 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.

A generic architecture for redesign of organizations triggered by changing environmental circumstances

“…However, it does give a good idea of the possible success of the approach given that the organizational templates and the requirements that are used within the system are suitable. Burton and Obel [34] raise the issue of validity of the organizational models designed from a computational perspective. They state that in order to create a valid model, three elements should be kept in balance: the question of purpose (for what purpose is the model being designed), the experimental design (how can the computational model be manipulated so that the purpose can be met), and the computational model itself (how has one chosen to define the model).…”

Organizational Design and Restructuring in Response to Crises: Lessons from Computational Modeling and Real-World Cases

“…To have a more robust match with the empirical data and to focus on the issues relevant to this study, we have aggregated some aspects of organizational design and task environment in this study. The design of the model has followed the call by Burton and Obel (1995) for a balance of relevancy, realism, and simplicity in computational models; the design has been shown to be both empirically valid and methodologically reliable (Carley 1996, Carley et al 1998. Through virtual experiments, CORP enables us to conduct precise and consistent analyses of organizational performance under crisis situations, while we grasp key organizational features.…”

“…In addition, our real organization cases only cover a small spectrum of the possible area, and many more cases could be examined using the computational model. Finally, because highly generalizable computational models often cannot be easily validated (Burton and Obel 1995), CORP is limited in its generalizability, which has enabled us to engage in the validation study presented. With the further development of computing technology and real-time data-capturing techniques, future studies should be able to consider more aspects of real-world environmental dimensions, organizational features, and individual characteristics building on complex system studies, organizational communications, and ethnography research (Siggelkow and Levinthal 2003, Taylor and Van Every 2000, Vromen 1995.…”

Agent Based Modeling to Inform the Design of Multiuser Systems

“…Agent-based modeling can be used for a wide range of purposes such as description of behaviors and training managers to make better decisions (Burton & Obel, 1995 ), development of theories of the conditions or mechanisms that generate certain behaviors (Davis et al, 2007 ), discovery of unexpected consequences of local interactions, and prescription to suggest better modes of operation or organization (Harrison, Lin, Carroll & Carley, 2007 ). We believe there are at least two important ways in which agent-based modeling can be leveraged in HCI research: to advance theories related to multiuser systems and to inform the design of these systems as well as interventions, policies, and practices surrounding them.…”