Dynamics of Organizations: Computational Modeling and Organization Theories.
In this book, Lomi and Larsen present original chapters that collectively demonstrate the value of using computer simulation and other computational methodologies for building theories of organizations. These methods constitute an approach to formalizing theories that stands as an alternative to mathematical modeling. There are two signal virtues of this approach. First, by formalizing theories in computer programming code, one can trace out the implications of abstract propositions as revealed in the operating processes and patterns of change the theory implies. Results that are inconsistent with what is known about organizations, or with what researchers think they know about organizations, cry out for explanation. Either the assumptions are wrong, and that is why the model performs badly, or the presumed empirical "facts" are wrong. The second virtue is that the method forces the theorist to confront issues that had previously been ignored. That is, the method pressures the theorist to develop more complete theories.
Researchers have been building computer simulation models of organizations for many years. Cohen, March, and Olsen's classic 1971 ASQ paper, "A Garbage Can Model of Organizational Choice," showed how simulation can be used to spin out the implications of original theorizing, but as March notes in his foreword to this book, this mode of theorizing has not been heavily used, even by the 1971 paper's many fans. In the ensuing decades, as computers have gotten cheaper and more powerful, software has been written to make model building easier. Consequently, the method has become more accessible. The chapters in this volume seek to demonstrate that interesting, sometimes surprising results flow from models built on widely accepted and believed bodies of theory. They do this by keeping the computational details to a minimum, telling the reader just enough to understand the argument. Many of the chapters in this book contain no mathematics at all. Generally, the strategy is to cite supporting works in which the technical details are worked through in detail. For the most part, proselytizing for the method is kept to a minimum, although many of the chapters begin with claims about the method. This becomes tiresome when one reads the book from start to finish.
Two properties are common to the theories developed in this book. First, most of them are dynamic. The computational strategy encourages researchers to think in processual terms, and equilibria stand out as results that need to be explained, rather than as expedient assumptions necessary to make mathematical formalizations tractable. So, for the most part, these are theories about change, including the factors that slow down or accelerate change. A second common property is that these chapters are mostly multilevel treatments. The method lends itself admirably to theories of action, in which individual people behave in ways that are aggregated into organizational patterns. Similarly, the behavior of individual organizations is aggregated to form population phenomena. Descending levels of analysis, some chapters are about network constraints and the effects they have on people or organizational subunits. Lomi and Larsen do an admirable job of laying out these themes in their introduction.
Most of the chapters in this book conform to this set of themes and address one or more of the issues outlined above. The chapter by Matthew Bothner and Harrison White, "Market Orientation and Monopoly Power," extends previous work by White that conceives of economic relationships as a three-level social structure linking suppliers, producers, and buyers. This work is not based on simulation studies. Its mode of formalization is mathematical. The theory pertains to any economic actor, and thus is only organizational by inference, but the chapter does deal in a very interesting way with cross-level relationships. The chapter by Laszlo Polos and Michael T. Hannan, "Nonmonotonicity in Theory Building with Applications to Organizational Mortality," is not a simulation-based study either. It constitutes a demonstration of the uses of symbolic logic as a third way to advance theory through formalization. Like simulation, recent software advances have made it easier to address the complexity of modern organizational theory and has similar payoffs. In using these techniques, theorists have to clarify their assumptions and definitions. Novel and interesting inferences emerge in much the same way that one observes in the simulation-based chapters that fill the rest of this volume.
The other chapters fall roughly into four groups. Of course, any attempt to categorize them will involve some arbitrary choices, as many of the chapters could appear in more than one group. Space will not permit a mini-review of each of the fourteen chapters, so I will just offer a few examples. The first group includes chapters based on ideas from social network analysis. The chapter by Kathleen Carley and Vanessa Hill, "Structural Change and Learning within Organizations," is based on a concept of learning in which relations between people ("who") and knowledge ("what") constitutes a two-dimensional structure. David M. Krackhardt ("Viscosity Models and the Diffusion of Controversial Innovations") builds on ideas Carley has previously developed to treat the diffusion of innovation as it is affected by the movement of people between organizational units. Similarly, John H. Miller ("Evolving Information Processing Organizations") examines organizational learning by considering organizations as information processing systems.
A second set of chapters examines populations of organizations. J. Richard Harrison and Glenn R. Carroll ("Modeling Culture in Organizations") extend earlier efforts to consider the diversity of people hired and the rate at which they adopt organizational cultures. They show how such processes have implications for organizational mortality. David Barron's paper, "Simulating the Dynamics of Organizational Populations: A Comparison of Three Models of Organization Entry, Exit and Growth," is extraordinary both for the sophistication of its analysis and the clarity of its exposition. The paper carefully compares the results from a large number of empirical studies and then shows how a simulation model can produce insights about the various theories that have been advanced to explain these results, and it can produce simulated data that quite accurately reproduce the results of the empirical studies. This paper stands out because it so nicely shows the theoretical payoff from simulations, while remaining faithful to the empirical studies that underlie the overall effort.
A third group includes chapters on feedback models of innovation. In these models, organizations or groups of individuals pursue changes based on efficacious outcomes, such as on the acceptance or rejection of others (Michael Macy and David Strang, "A Computational Model of Fashionable Innovation"). Franco Malerba, Richard Nelson, Sidney Winter, and Luigi Orsenigo build their models in the context of a stream of innovation in the computer industry This gives their chapter the very interesting quality of presenting an abstract theoretical formulation while still addressing observations of concrete phenomena in their institutional context. Daniel A. Levinthal extends his previous work on "rugged landscapes" to show how initial differentiation in the context of local search produces inertia aside from the factors that might impose costs on organizations' efforts to change.
Finally, a number of chapters concern social status as it emerges from social interaction. While the chapters by Christoph H. Loch, Benardo A. Huberman, and Sezer Ulku are explicitly about this, Michael Prietula's "Advice, Trust and Gossip among Artificial Agents" (which builds on previous work with Kathleen Carley) allows agents to remember each other's performance over time. In this sense, a form of social status is being modeled.
I think this volume succeeds in providing an introduction to the use of computational modeling as a theory-building tool for organizations researchers. It does so by presenting a rich variety of subjects in an accessible form. Readers should not be left with the impression that this is arcane stuff for people with too much time on their hands. Rather, they should see how formal methods can be used to advance theoretical imagination. As such, this book would make very good reading for doctoral courses on organizational theory.
Haas School of Business University of California Berkeley, CA 94720
|Printer friendly Cite/link Email Feedback|
|Publication:||Administrative Science Quarterly|
|Article Type:||Book Review|
|Date:||Mar 1, 2003|
|Previous Article:||Silent Spill: The Organization of an Industrial Crisis.|
|Next Article:||The Triumph of Ethernet: Technological Communities and the Battle for the LAN Standard.|