Validating Simulation Models

This paper discusses aspects of validating simulation models designed to describe, explain and predict realworld phenomena. It starts with a short review of arguments used in the simsoc mailing list discussion on theory, simulation and explanation a few months ago, deals with the use of quantitative and qualitative computational models to make quantitative and qualitative predictions or rather to draw conclusions from complex antecedents, and then discusses different types of explanation and prediction (and the relation between these two), It closes with an overview of topics in validity and validation from the point of view of the structuralist programme in the philosophy of science. INTRODUCTION: THEORY, SIMULATION, EXPLANATION AND OBSERVATION A few months ago, the simsoc mailing list experienced a longish discussion1 which originated from Thomas Kron’s question “about the relation of computer simulation and explanation, especially sociological explanation”. More than fifty contributions to this discussion followed within less than three weeks, and contributors discussed the role of simulation in theory building (mostly, but not only) in the social, economic and management sciences — as well as the relation between observation on one hand and computer-assisted theory building (Hanneman 1988) on the other. Scott Moss came back to his presidential address at the 1st conference of the European Social Simulation Association, Groningen, September 2003, in which he said “that if social simulation with agents is to be anything other than another in the long line of failed approaches to social science, it will be a positive departure only because it facilitates the dominance of observation over theory” and continued that the great successful scientists (outside the social sciences) built their generalisations around observation, developing new theoretical structures based on and validated by new evidence (quoted from his contribution to the simsoc mailing list as of November 14, 2003). Well in the line of this trait of thinking is the role of simulation or computational modeling which can be found in Gilbert and Troitzsch 1999 1The discussion can be found in the November 2003 section of http://www.jiscmail.ac.uk/archives/simsoc.html, topics “simulation and explanation” and “theory and simulation”. which was recently extended by Alexis Drogoul (Drogoul et al. 2003: 5) and can be seen in Figure 1.