Sensitivity Analysis in the Social Sciences: Problems and Prospects

Abstract

Eighteen papers, originally presented at a conference held in Schloss Dagstuhl, Germany, in May 1997, examine theoretical and methodological issues related to the application of simulations in the social sciences; contribute to the literature on the sensitivity analysis of simulations; discuss various multiagent-based simulations and demonstrate their application to the study of social processes; and apply cellular automata-based simulations to the investigation of social evolution processes in various contexts. Papers focus on models, processes, and algorithms in moving toward a simulation toolkit; the methodology of artificial societies; computer assisted interfacing; interactive neural networks as tools for modeling social systems; a serial multiagent system for social simulation (SMASS); a reference model for the simulation of multiagent systems; building multiagent simulation applications; diversity in strategies of partnership formation; cellular automata models as a modeling tool; evolution of learning rules in space; simulating the role of nonlinearity and discreteness in dynamic social impact; representation methods and the emergence of intergroup cooperation; sensitivity analysis in the social sciences; tests of dynamic social models with time-related surveys; a sensitivity analysis of the spreading of respectful reputation; economic development, opinion formation, and technological change in a multilevel simulation model; microsimulating farm business performance; and ordering parameters in the rule space of social systems. Contributors are in the fields of economics, sociology, psychology, and computer science. Suleiman is at the University of Haifa. Troitzsch is at the University of Koblenz-Landau. Gilbert is at the University of Surrey. Subject and author indexes.