Copyright Sociological Research Online, 1999

Computer Modeling of Social Processes

Wim Liebrand: Andrzej Nowak and Rainer Hegelsmann (editors)
Sage Publications: London
7619-5424-4 (pb); 7619-5423-6 (hb)
15.99 (pb); 47.50 (hb)
xi+208 pp.

Order this book

There are two dimensions along which different social sciences can usefully be categorised: subject matter and methods. Some established disciplines, like economics, are often accused of being led by methods to the detriment of subject matter. By contrast, newer fields like management science draw together existing methods and attempt to apply them in new (or perhaps better defined) subject areas.

The title of this edited volume Computational Modelling of Social Processes suggests the latter approach. There has been a rapid growth in new computational techniques, developed mainly in the fields of Artificial Life and Artificial Intelligence and spurred on by the extraordinary fall in the cost of computation in the last twenty years. Even allowing for the existence of other edited volumes in this area (Conte et al 1997, Gilbert and Conte 1995, Gilbert and Doran 1994, Hegselmann et al 1996, Troitzsch et al 1996), the relative novelty of the approach means that there is plenty of room for another, covering new applications, new disciplines and perhaps new techniques.

Unfortunately, the title of the book is misleading. Of the nine chapters (not counting the introduction), only four (by Troitzsch, Hegselmann, de Vries and Nowak, Vallacher and Burnstein) can legitimately claim to study social processes using computer models. Furthermore, the chapters by de Vries and Nowak et al are more in the nature of proposals for computational models than actual simulations. (Despite this, the Nowak et al chapter seems extremely well thought out and interesting, though the de Vries chapter is somewhat less so.) The remaining chapters attempt a different, and to this reviewer less interesting task, the use of computers to assist social scientists in the collection, treatment and presentation of data.

In some cases, like Klovdahl's chapter on dynamic graphical representations of social networks, it is very clear that the computer allows us to do what we could not possibly do before. It is highly likely that this liberation will facilitate novel theories and perhaps new substantive methods. By contrast, the chapters by Yung et al (on simulations for bootstrap sampling) and Gueze et al (on integrating video data with other kinds of monitoring), while clear and useful, merely allow us to do what we can do already, perhaps faster or with more accuracy. The book is thus awkward in trying to cover both descriptive techniques (where the criteria of success are generation of novel ideas for testing, explanation, understanding and perhaps prediction) and instrumental techniques (where the criteria are speed, cheapness, ease of use and accuracy). Both approaches are important, but rather far apart (and broad) to be contained effectively in a single book. The resulting volume is neither representative nor comprehensive.

To their credit, the editors have gone to considerable lengths to situate the contributions with respect to each other and the broader literature, both through an introductory chapter and shorter preambles to each section. Matters are further improved by the relatively high standard of the individual contributions, all of which have something to say and are well written. Despite this, the book does not quite escape from edited volume status on two counts. Firstly, the sections of the book seem rather arbitrary. (The chapter by de Vries ought really to be in the second section on Computer Modelling and Neural Networks - though one of its weaknesses is that the type of simulation it proposes is not as clear as it could be - and the chapter by Adèr and Bramsen is more a "translation" of an existing technique - structural modelling - into neural network terminology. It could thus could equally well have been moved to the final section on Computer Modelling and Data Analysis.) Secondly, perhaps because it includes chapters on instrumental techniques without specific applications, the book is far less representative of different techniques and subject areas than several cited above. There is definite social psychology bias, with no distinctively economic applications - it is arguable whether the mutual assistance and Prisoner's Dilemma simulations in Hegselmann's chapter could be seen as economic - and nothing from fields like anthropology or political science. Similarly, two important simulation techniques commonly applied to social systems - agent-based modelling and evolutionary computation - are conspicuous by their absence. As such, this book must be treated with caution as a guide to the field.

From the perspective of the imaginative sociologist, the content is quite satisfying. The chapter by Troitzsch discusses computational techniques for multi-level modelling where there are interactions between levels: students forming classes, classes forming schools and schools forming an educational system. He confronts his models with data and discusses their possibilities. Hegselmann, in a particularly clear and thoughtful chapter, discusses the possibilities of cellular automata (regular grids of cells that interact with their neighbours according to fixed rules) for modelling basic processes of social interaction like group formation and stratification. The chapter is an effective mixture of "how to" and "why". Gernert's chapter provides a useful overview of the strengths and weaknesses of neural networks as instrumental systems for prediction. By contrast Nowak et al provide very interesting arguments for representing interacting social systems descriptively as neural networks. The other chapters, already discussed, will only be of interest to advocates of specific techniques.

To sum up, the attempt to produce something more than an edited volume of separate papers has resulted in an edited volume of a relatively high standard that is well worth looking through, particularly the chapters that actually fit the title. However, the introduction of instrumental techniques has made the scope of the book too broad, with the result that very few readers will benefit from reading right through. Furthermore, this broadness means the book must be treated with caution as a representative sample of work in social simulation. Nonetheless, it can be recommended as a more than usually accessible starting point.

Edmund Chattoe
University of Surrey


CONTE, Rosaria, HEGSELMANN, Rainer and TERNA, Pietro (editors) (1997) Simulating Social Phenomena, Lecture Notes in Economics and Mathematical Systems 456. Berlin: Springer-Verlag.

GILBERT, Nigel and CONTE, Rosaria (editors) (1995) Artificial Societies: The Computer Simulation of Social Life. London: UCL Press.

GILBERT, Nigel and DORAN, Jim (editors) (1994) Simulating Societies: The Computer Simulation of Social Phenomena. London: UCL Press.

HEGSELMANN, Rainer, MUELLER, Ulrich and TROITZSCH, Klaus G. (editors) (1996) Modelling and Simulation in the Social Sciences from the Philosophy of Science Point of View, Theory and Decision Library, Series A: Philosophy and Methodology of the Social Sciences. Dordrecht: Kluwer Academic Publishers.

TROITZSCH, Klaus G., MUELLER, Ulrich, GILBERT, G. Nigel and DORAN, Jim E. (editors) (1996) Social Science Microsimulation. Berlin: Springer-Verlag.

Copyright Sociological Research Online, 1999