 |
|
|
|
|
Transforming Conditions
Steps in SD Modeling
Previous |
Next
SD modeling supports a pragmatic, navigational view by pursuing four general lines of inquiry (Figure 15): - Why are certain aspects of the system changing?
- Where is the system headed if no new action is taken?
- How else can the system behave, if different decisions are made?
- Who has the power to move the system in a safer, healthier direction?
Figure 15: System Dynamics Modeling
Addresses Navigational Questions
Enlarge picture
To answer these questions, SD modeling proceeds iteratively through the following general steps (Homer, 1996) (Figure 16):
- Identify a persistent problem that exists, in part, because of dynamic complexity. The emphasis on
dynamic complexity does not refer to problems that have many parts (i.e., combinatorial complexity), but rather to problems that involve mutually reinforcing factor (e.g., behavioral feedback), accumulations over time, significant delays between actions and effects, or non-linear patterns of change (e.g., better-before-worse or vice versa);
- Develop a preliminary dynamic hypothesis (causal map) by identifying which causal forces are at work and how they relate to one another;
- Convert the hypothesis into a formal computer model. This is done by writing a system of differential equations, calibrating them based on available data, and noting any areas of uncertainty, which then become the focus for sensitivity analysis. In other words, uncertainty or lack of previously collected data is not a fatal flaw for SD modeling, as it can be for statistical techniques such as regression modeling or structural equation modeling (Randers, 1980);
- Use the computer model to conduct controlled simulation studies, with the goal of learning how the system behaves and how to govern its evolution over time;
- Choose among the set of plausible futures those that best reflect stakeholder values and that strike an acceptable balance among inevitable trade-offs.
- Keep repeating the process, creating better hypotheses, models, policy insights, and more effective action with each iteration.
Figure 16: Iterative Steps in System
Dynamics Modeling
Enlarge picture
The next two sections illustrate several of these steps in action. The first example summarizes the results of a real-world, well-funded, high-stakes application of simulation modeling to study diabetes dynamics in the wake of the obesity epidemic. The second, by contrast, is a hypothetical scenario in which causal mapping and simulation modeling are used to improve long-term grantmaking strategy in a neighborhood struggling against an entrenched syndemic.
Previous |
Next Back to top
Page last reviewed: January 30, 2008
Page last modified: January 30, 2008
Content source: Division of Adult
and Community Health,
National Center for Chronic Disease Prevention and Health Promotion
|
|
