The Tools of System Dynamics

From SystemsWiki

Jump to: navigation, search

The system dynamics method produces formal computer models of real world problems which represent a theory of how system structure produces change. This theory is a synthesis of several hypotheses or causal process propositions. Indeed James Hansen the climate change modeler states that “model is a word for people who can’t spell hypothesis.” These models are refined using the scientific method with best practice tools and techniques which focus on building user confidence and acceptance of the structure as a useful explanation to guide future actions (Lyneis 2008). These computable models are related to the implicit informal mental models we use to think, learn and make sense of the world. The method often uses tools of feedback systems thinking including causal loop diagrams to help develop, use and explain the results obtained from formal models. There is a certain viewpoint adopted when applying the system dynamics method, where we zoom out on a problem, focus on structure and consider past and possible future trends and patterns rather than events. From this viewpoint decisions are represented as patterns of how information is translated into action. There is a balance between theory, represented by the model structure, and data (represented by parameter values and calibration) to achieve the right mix of rigour and relevance in order to produce effective action. Unfortunately this approach leads us into the world of disturbing, surprising non-linear relationships that can require unfamiliar techniques for careful analysis and testing. There are excellent examples of moving from linear event oriented to pattern oriented feedback thinking in the general, business and environmental literature (e.g. Morecroft Sterman Richmond Meadows Ford). These texts are an excellent introduction to system dynamics modeling. Here we will offer a brief introduction to the tools so health domain readers can get started with understanding and refining simple models, using the STELLA/ithink package and other emerging online tools.

System dynamics modelling comes as a package that includes a language to represent structure and change over time, a modelling process, a software toolkit and technical tips and tricks required for successful computational modeling One tool we have already mentioned, which can be useful for clarifying concepts and their relationships is the concept map, or a simpler tree version, the mind map. This is a popular way to make our mental models explicit so the concepts themselves, the language used to describe them, and their relationships can be clarified. In system dynamics "the abstractions from our experience are arranged in mental models which form knowledge that we want to improve in order to make better decisions." We believe that representing these mental models as formal models of causation to explain observations using abstractions of real world mechanisms leads to better policies and actions, particularly avoiding unintended consequences. Confidence in these models is based more on the structure of the model than the accuracy of the parameters in the model. Our mental models guide the way we think and talk about a problem. Formal computer models provide a logical consistent framework including explicit handling of time, which helps clarify concepts and mechanisms. We can use these formal computer models with other tools to clarify our thinking, including causal loop diagrams and concept maps. In this way we move from discourse to conceptual maps to computational models. Online environments for supporting this transition are now becoming available (Systemswiki, Insightmaker, Simgua).

Causal loops are a special form of influence diagram, representing cause and effect. They can be considered as special type of concept map with causal links and cyclic structures to show feedback effects. The link represents the direction of change between the two concepts, all other things being equal. This is the “ceteris paribus” assumption used by economists.

We can apply the accepted system dynamics theory of structure (endogenous behaviour, feedback loops, accumulations and delays, representation of decision-making), using stock flow diagrams with converters and connectors (also referred to as Forrester diagrams);


Generic Archetypes

Questions & Comments to Geoff McDonnell
Personal tools