Order this book
P.O. Box 591, Tanunda, South Australia 5352.
This book joins a growing list of publications which advocate the wider adoption of computer simulation as an important new research methodology for the social sciences. Important prior contributors in this area include Gilbert and Troitzsch (1999), Conte, Hegselman and Terna (1997), Epstein and Axtell (1996), Gilbert and Conte (1995) and, in the application of simulation specifically to organisational science, Prietula, Carley and Gasser (1998) among others. The growth of interest in simulation as a methodology mirrors an important development in the study of social phenomena. This development is the growing realisation that social phenomena including economics (Arthur et al. 1997, Ormerod 1995, 1998), organisations (Marion 1999) and other social systems (Eve et al. 1997) frequently demonstrate the characteristics typical of complex systems containing significant non-linearity. Such systems display emergent behaviour, that is, behaviour not inherent in or predictable from knowledge of their constituent parts (Holland 1998). This approach presents a major challenge for much contemporary theory and method. Ilgen and Hulin state, for example "...our methods and theories remain far better suited for the deterministic and linear corners of [organisation science] than for the well populated chaotic regions of it." (p. xv) They advocate computational methods as a 'third scientific discipline' to complement established method.
Organisation science has become divided in many fundamental ways leading to a pluralism (Reed and Hughes 1992) which has both its advocates (Aldrich 1992) and detractors (Hughes 1992). In the study of organisational behaviour, the subject of this book, frequent recourse will be made to psychology - concerned to understand personal behaviour (micro phenomena) and to sociology, concerned with the behaviour of social aggregates (macro phenomena). Each of these disciplines contains an inherent set of theoretical precepts, some of which are incommensurable with those of the other disciplines. This has led to a so-called macro-micro problem (Coleman 1994) within social science, where the relative interplay between micro and macro phenomena is poorly understood and difficult to study. The editors address this issue in several ways. In Chapter two, Seitz notes that there are few attempts to reintegrate disparate and ad hoc approaches derivative of alternative perspectives. He argues that computational modelling can facilitate testing the respective claims of alternative theories against each other and empirical referents. Indeed, several examples of applications for computational methods contained in this text do just that. In addition, the text progresses through a series of applications from micro level to more macro perspectives thereby illustrating the broad potential application of computational approaches. The growing cross-disciplinary science of complex systems goes a good way in helping us to understand the micro-macro interplay better. Unfortunately, no examples are provided of such an approach and (by and large) the examples provided operate within the bounds of established theoretical assumptions in psychology and sociology.
A second major source of division in organisational science pivots on issues of ontology and epistemology (Cooper and Burrell 1988). In particular, we must be aware of the relative influence of modernist and post modernist thinking. Again, we are confronted with several alternative positions that are inherently incommensurable (Burrell and Morgan 1994). As I have argued elsewhere (Goldspink 2000), considering social systems as complex and non-linear challenges all established theoretical positions. Ilgen and Hulin do not pick up on this fundamental challenge to established approaches. Rather they advocate the need for computational methods as a complement to existing approaches. They place computational methods between 'experimental and correlational' methods typical within behavioural study of organisations. They note "Computational modelling has strengths orthogonal to the strengths of either traditional research discipline" (p. 7). In their introduction, they draw attention to the differing epistemological assumptions inherent in such methods. They state for example "Computer modelling and simulation are consistent with modern views of epistemology and the need to use an encompassing view of the total theoretical and organisational space to link limited and ad hoc theories" (p. xx). In drawing attention to the potential which computational methods have to link ad hoc theories, they point to its value for challenging the extant pluralism and providing a more integrated understanding of diverse organisational phenomena. In my view, however, they do not fully come to terms with the epistemological implications of their approach, either for what they are studying or for the methods they are advocating.
The editors rightly assert that computational methods, born of a more cross-disciplinary approach, have the potential to challenge contemporary divides or to plug important gaps. However, such an outcome is not a necessary consequence. Computational methods have the potential for application in many ways and some of those ways are within the constraints of contemporary disciplines. It may take a sustained effort to encourage the use of computational methods in cross-disciplinary contexts if the benefits they identify are to be realised. While the editors note the particular value of simulation for studying non-linear aspects of organisation, this is not the only way in which computer modelling is used. Indeed there is growing criticism of the 'toy' (and essentially linear) models used by neo-classical economists to model what are increasingly regarded as non-linear economic phenomena (Arthur et al. 1997). So, these tools can clearly be misapplied. There are computer models that incorporate a wide range of systems methods and thinking - from simple cybernetics to compound simulations based around correlational methods - so the computational approach does not necessarily compel a rethink or challenge established research norms and assumptions. There is a need then to be clearer as to what approach is specifically being advocated and under what conditions its anticipated benefits may best be realised.
If simulation is to take up a place among alternative methodologies, there will need to be established principles which guide its application. In Chapter six, Hollenbeck notes difficulties in gaining acceptance for simulation based approaches within established academic communities. Some of these problems are no doubt down to conservatism but some scepticism is well founded. Problems exist in fundamental areas with the further development and application of computational methods, particularly those incorporating non-linear models. These challenges include the following:
Replication: Axelrod (1997) has identified the following problems that complicate the task of replication for simulations:
Generalisability: The ability to infer that the results of a specific simulation or even a set of runs of a simulation reveal results that would apply in alternative contexts is problematic. Because of sensitivity to initial conditions and path dependence, a simulation may yield little that is generalisable, at least in terms of specific outcomes. It may be possible to infer generality within bounds, or to identify patterns of behaviour typical of a class of systems of which the simulation is one example. However, as Kollman et al. (1997, p. 462) note "The risk of any one computational model being "a mere example" unfortunately exists".
Prediction: The contingent, path dependent nature of many complex system dynamics implies that using simulation as a way of predicting real world phenomena is fraught with difficulties.
Data Availability: Development of models, if they are to say anything meaningful in terms of past, current or possible states of real systems, requires the identification of relevant parameters and variables. Values have to be assigned to these parameters and variables and this may involve calibrating those values to real world equivalents. The availability of suitable data is or can be a significant problem.
Validation: Testing a model for validity implies seeking confirmation of functional equivalence at least within the range of parameters characteristic of the system being modelled. Two aspects of simulation work make this difficult. Firstly, simulations are often used as exploratory vehicles, to examine behaviour of a real system. As the simulation is used to understand the system being modelled, it is difficult to directly calibrate or validate against the real system, which is, as yet, not understood. Secondly, as the systems being modelled are frequently very complex (hence the choice of simulation in the first place) making complete comparisons between model and real world behaviour is commonly not possible. This leads to the need to choose a mixed methodology, one where validation and experimentation can take place iteratively, with a methodology directed at seeking verification of modelled behaviour taking place concurrently with the simulation modelling. Leik and Meeker (1995, pp. 465-466) suggest a set of rules for building simulations which have some hope of being validated. These are:
These are demanding requirements. The various examples of simulations provided in this text illustrate many of these problems and how these were (or were not) addressed by the modellers. Illustrating these problems in the context of specific applications is likely to be more meaningful to those not familiar with the more abstract debate about simulation. In this regard the text is well focused on the needs of the likely audience.
In addition to the above it has been noted that minimum standards for conduct, documentation and publication of computational based studies will need to be addressed if fundamentals of scientific methods are to be preserved. As Axelrod notes (1997) this is difficult using existing vehicles for communicating research results such as journals and compilation publications (such as the one under review). Communicating the results of a simulation without the details of the model and its implementation is of little value. There is often a need for a lengthy report to do justice to the complex design and results. This problem may be overcome with the broader acceptance of electronic journals (such as JASSS) which make it feasible to, for example, either provide detailed attachments including source code, or to provide access by other researchers to the fully operational model on-line. This can be important as it eliminates complications such as different hardware and software environments from the equation, thus simplifying replication.
To help their readers appreciate the nature and results of the specific simulation based research presented in the work, and to begin to appreciate some of the above issues, Ilgen and Hulin follow each Chapter which describes computational research with critical analysis by a reviewer. In several instances, there are important lessons highlighted which serve to improve knowledge of the potential and pitfalls of simulation methodology. In this sense Ilgen and Hulin have taken the above issues seriously and demonstrated one approach to obviating them within the constraints of more traditional published media. The approach adopted still falls short of the more rigorous demands set out above however.
Reading through the material presented in this book, it becomes readily apparent that one of the practical challenges for those who wish to apply computational methods is just how flexible the method is. Simulations can be applied to a wide range of subject mater, implementing models incorporating a wide range of assumptions and theoretical foundations. Of course, this is one of the strong points of the approach from a methodological standpoint. However, there is an increasing trend to use pre-built simulation platforms rather than reinventing each model anew using a high level programming language. These platforms are growing in number and commonly incorporate assumptions about the nature and type of phenomena being modelled. They can also be founded on a particular theory. For the intending modeller then, there is a need to be aware of the possible modelling environments and the constraints they might impose. This is important, as once built, the underlying assumptions incorporated in the platform design may not be evident. There are pitfalls aplenty in this regard, as one of the challenges with any computational modelling is to separate data from noise. In computer simulation, noise or artefacts of the model may well appear and interfere with patterning of the data becoming difficult to distinguish from data. If the specifics of the model and model platform (even down to hardware implementation) are not fully understood, then there is a possibility that the results will be misinterpreted. Therefore, while these packaged models offer the possibility of extending the reach of the technique, even where the modeller does not have a sophisticated understanding of computing, such an approach carries traps for the unwary.
In conclusion, this book advances the general case for wider use of computational methods in organisation science. It does this by providing a showcase for a variety of applications of simulation methods to behavioural aspects of organisational research. The book examines issues of specific interest to organisational behaviourists including withdrawal behaviour of employees, faking in personality tests, effectiveness of performance pay and other reward systems, group decision making dynamics, conformance behaviour, group formation and organisational adaptation. In this, it presents some distinctive research in existing areas of organisational study. Hence, in addition to its potential interest to would be simulators, the book provides valuable material to complement existing research in organisational behaviour. Its audience is therefore anyone conducting research in organisation behaviour as well as those interested in better understanding the potential and pitfalls of simulation as a methodology for the social and organisation sciences. The book advocates computational methods as an important complementary research method with the potential to help integrate this increasingly diverse and fragmented field of inquiry. It holds short of exploring the more fundamental epistemic and ontological challenges posed by the method and associated theory of non-linear systems. The editors take seriously the criticisms of the method and use the work to highlight not only possibilities but also pitfalls covering many of the more important concerns. This is done in a way that is likely to be meaningful and readily appreciated by the wide potential audience of the book - by illustration in the context of a particular application. Its coverage of such issues is not however complete and nor are these issues made particularly explicit in some instances. The editors also provide us with an example the way in which conventional media for communicating research can be used to draw out some of the assumptions and limitations of simulation by providing a broader peer review of the published work.
ALDRICH H. E. 1992. Incommensurable Paradigms? Vital Signs from Three Perspectives. In M. Reed and M. Hughes, editors, Rethinking Organization: New Directions in Organization Theory and Analysis, Sage Publications, London.
ARTHUR W. B., S. N. Durlauf and D. A. Lane 1997, editors, 1997. The Economy as an Evolving Complex System II, Addison-Wesley, Reading, MA.
AXELROD R. 1997. Advancing the Art of Simulation in the Social Sciences. Complexity, 3:16-22.
BURRELL G. and G. Morgan 1994. Sociological Paradigms and Organisational Analysis, Virago, London.
COLEMAN J. S. 1994. Foundations of Social Theory, Belknap/Harvard University Press, Cambridge, MA.
CONTE R., R. Hegselmann and P. Terna, editors, 1997. Simulating Social Phenomena, Springer-Verlag, Berlin.
COOPER R. and G. Burrell 1988. Modernism, Postmodernism and Organizational Analysis: An Introduction. Organization Studies, 9:91-112.
EPSTEIN J. and R. Axtell 1996. Growing Artificial Societies: Social Science From the Bottom Up. Brookings Institution Press, Washington, DC and The M. I. T. Press, Cambridge, MA.
EVE R. A., S. Horsfall and M. E. Lee, editors, 1988. Chaos, Complexity and Sociology: Myths, Models and Theories . Sage Publications, London.
GILBERT N. and R. Conte, editors, 1995. Artificial Societies. UCL Press, London.
GILBERT N. and K. G. Troitzsch 1999. Simulation for the Social Scientist. Open University Press, Buckingham.
GOLDSPINK C. 2000. Contrasting Linear and Non-Linear Perspectives in Contemporary Social Research. Emergence, 2:72-101.
HOLLAND J. H. 1998. Emergence: From Chaos to Order. Addison-Wesley, Reading, MA.
HUGHES M. 1992. Decluding Organizations. In M. Reed and M. Hughes, editors, Rethinking Organization: New Directions in Organization Theory and Analysis, Sage Publications, London.
MARION R., 1999. The Edge of Organization: Chaos and Complexity Theories of Formal Social Systems. Sage Publications, London.
KOLLMAN K., J. H. Miller and S. Page 1997. Computational Political Economy. In W. B Arthur, S. N. Durlauf and D. A. Lane, editors, The Economy as an Evolving Complex System II, Addison-Welsey, Reading, MA.
LEIK R. K. and B. F. Meeker 1995. Computer Simulation for Exploring Theories: Models of Interpersonal Co-operation and Competition. Sociological Perspectives, 38:463-482.
ORMEROD P. 1995. The Death of Economics. Faber and Faber, London.
ORMEROD P. 1988. Butterfly Economics. Faber and Faber, London.
PRIETULA M. J., K. M. Carley and L. Gasser, editors, 1998. Simulating Organizations: Computational Models of Institutions and Groupsx. The M. I. T. Press, Cambridge, MA.
REED M. I. and M. Hughes, editors, 1992. Rethinking Organization: New Directions in Organization Theory and Analysis, Sage Publications, London.
Return to Contents of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2001