Chapter Preview
TopBackground
Objects in social science consist of many elements, and the relationships between them are generally complicated. It is natural and reasonable to employ dynamic simulation modelling in order to solve social science problems. Nevertheless, social science models have been expressed in static ways or have been unrealistically simplified, mainly because of difficulties in handling dynamics equations. However, it is still important to make social science models as dynamic systems. The reason is that elements in societies are mutually connected in complicated way, and the nature of social systems is dynamic; besides, people fail to understand dynamics (Booth Sweeney & Sterman, 2000).
In this situation, J. Forrester developed interface and procedures to make dynamic models in social science fields. This is “system dynamics.”
The first appearance of SD in academic papers was in the Harvard Business Review in 1958 (Forrester, 1958). Since SD is a method of simulation, both theoretical enquiries and practical applications are shown in various journals. The journal of SD, “System Dynamics Review” started in 19851, and System Dynamics Society holds “International Conference of System Dynamics Society” every year2.
There are seminal books of SD fundamentals. Sterman (2000) contains almost all of knowledge to build and test SD models. This book shows many detailed SCM models. Because many people can easily imagine material flow, SCM models would be good examples not only for SCM researchers but also for all practitioners and learners to introduce ideas of a stock-flow structure and feedback loops. This book also has detailed explanations about validation procedures. Validation is also discussed in Qudrat-Ullah (2005) which leads to agent-based simulation models’ validation. Ford and Flynn (2005) present how to find key inputs.
Coyle (1996) describes important functions with mathematical background. This is useful to know precise numerical backgrounds.
Ford (1999) has many model examples from environment study field. This also explains SD modelling process concisely. Maani and Cavana (2000) show many business and economics models using SD. Included models are relatively simple so that this would be a good start point to learn. The models’ parameters and equations are fully clarified; therefore, learners can follow the modelling processes. More complex but practical models are shown in Morecroft (2007) and Warren (2008). Implementations of SD elements described in these books are helpful for practitioners and researchers.
Key Terms in this Chapter
Causality: The relationship between cause and effect (result). In system dynamics, there are two types of causality: positive and negative. Positive causality means a result element changes to the same direction as its cause element changes to. Negative causality indicates a result element changes to the opposite direction to its cause element’s change. This is also called a causal relationship.
Calibration: The activity to fit parameters in a model to specific data. Sometimes it is used as a synonym of optimisation.
Balancing Loop: A feedback loop with an odd number of negative causal relationships. Elements in this kind of loop perform goal seeking or oscillation.
Reinforcing Loop: A feedback loop with an even number of causal relationships. Elements in this kind of loop perform exponential growth or decay.
Feedback Loop: Circulations which consist of causalities between elements in subject systems. Feedback loops are classified into reinforcing loops and balancing loops.
Causal Loop Analysis: The examination focusing on feedback loop structures. Using this, systems’ behaviour can be roughly predicted. This is also called qualitative analysis of system dynamics.
Optimisation: To set parameters in order that target variable reach the goal value set by modellers. This term is sometimes used as a synonym of calibration.
Systems Thinking: In a system dynamics context, a way of thinking based on system dynamics. It is also used to mean system dynamics analyses without quantitative definitions. It focuses on feedback loop structure in order to forecast the direction of performance and find pertinent elements for controlling systems. This is also called qualitative system dynamics.