Monograph Menu Report Home Director's Foreword Table of Contents Tables and Figures Acknowledgements Abstract Prologue Introduction Navigating Health Futures Valuing Conditions Crafting Conditions Perceiving Dynamic Conditions Reorienting Public Health Work Transforming Conditions Reflecting on Public. Health. Work. Glossary References About the Author Download Full Report (PDF-8.9Mb)

Transforming Conditions

The System Dynamics Approach

Previous | Next

SD modeling is a way of mapping and then modeling the forces of change in a dynamic system so that their influences on one another can be better understood and the overall direction of the system can be better governed (Milstein and Homer, 2005; Sterman, 2000). The methodology enables planners to assemble their knowledge of a problematic situation into a single, visible dynamic hypothesis and then, using computer simulations, to formally compare various scenarios for how to navigate change. The emphasis is not on forecasting the future, but rather on learning how our actions in the present can trigger plausible reactions both far away and over time (Sterman, 2006). With its ingenious use of simulation games as a virtual world for interacting with an SD model, the learning that occurs is often visceral and emotional rather than purely cognitive or conceptual (Foresight and Governance Project, 2002). As such, SD is a powerful tool for discovering how to move more effectively and ethically in a dynamic and democratic world (Forrester, 1971; Meadows, Richardson, Bruckmann, 1982; Meadows and Robinson, 1985; Sterman, 2002).

With a nearly 50-year history since its development by Jay W. Forrester at the Massachusetts Institute of Technology (Forrester, 1991), SD modeling today is used productively in many fields of human endeavor (Roberts, 1999a; Sterman, 2000). Influential applications encompass projects in human service delivery (Levin and Roberts, 1976), urban development (Forrester, 1969), corporate management (Forrester, 1961); (Pidd, 1996), energy and global ecology (Ford, 1999; Meadows, Randers, Meadows, 2004), K-12 education (Forrester, 1994; Saposnick, 2004), and dozens more.

There are also numerous applications in the health area specifically (Hargrove, 1998; Milstein and Homer, 2005; Taylor and Lane, 1998). Some examples include studies of

• Cardiovascular disease (Hirsch and Wils, 1984; Homer, Hirsch, Minniti, et.al., 2004; Luginbuhl and Hirsch, 1981; Homer, Milstein, Wile, et.al., in press);
• Cervical cancer (Royston, Dost, Townshend, et.al., 1999);
• Chlamydia (Royston, Dost, Townshend, et.al., 1999; Townshend and Turner, 2000);
• Cocaine (Homer, 1993);
• Dengue fever (Ritchie-Dunham and Mendez Galvan, 1999);
• Diabetes (Homer, Hirsch, Minniti, et.al., 2004; Homer, Jones, Seville, et.al., 2004; Jones, Homer, Murphy, et.al., 2006; Milstein, Jones, Homer, et.al., 2007);
• Dental care (Hirsch and Killingsworth, 1975; Levin and Roberts, 1976);
• Drug-resistant pneumococcal infections (Homer, Ritchie-Dunham, Rabbino, et.al., 2000);
• Heroin (Levin, Roberts, Hirsch, 1975);
• HIV/AIDS (Dangerfield, Fang, Roberts, 2001; Homer and St. Clair, 1991; Roberts and Dangerfield, 1990);
• HMO planning (Hirsch and Miller, 1974);
• Mammography (Fett, 2001);
• Mental health (Levin and Roberts, 1976; Smith, Wolstenholme, McKelvie, et.al., 2004);
• Obesity (Abdel-Hamid, 2002, 2003; Homer, Milstein, Dietz, et.al., 2006);
• Patient flows (Lane, Monefeldt, Rosenhead, 2000; Wolstenholme, 1996, 1999);
• Performance assessment (McDonnell, Heffernan, Faulkner, 2004);
• Public health emergencies (Hirsch, 2004; Hoard, Homer, Manley, et.al., 2005);
• Public health planning (Hirsch and Immediato, 1999; Hirsch and Immediato, 1998; Homer and Milstein, 2004; Homer, Hirsch, Milstein, 2007; Innovation Associates and New England Health Care Assembly, 1997);
• Tobacco (National Cancer Institute, 2005; Roberts, Homer, Kasabian, et.al., 1982; Tengs, Ahmad, Savage, et.al., 2005; Tengs, Osgood, Chen, 2001; Tengs, Osgood, Lin, 2001); and
• Syndemics (Homer and Milstein, 2002a, 2003b, 2004)

Still, the SD methodology is not routinely taught in schools of public health despite its tremendous potential for illuminating some of the most challenging phenomena that confront the field (Homer and Milstein, 2003a; Milstein, 2003b). For instance, public health scholars could use SD in innovative ways to study

• Individual diseases and risk factors (e.g., by examining momentum and setting justifiable goals);
• Mutually reinforcing afflictions (syndemics) (e.g., by exploring interactions among related afflictions, adverse living conditions, and the public's capacity to address them both);
• Program dynamics (e.g., by analyzing the system-wide impacts of comprehensive programs with interacting components);
• Regional dynamics (e.g., by incorporating the mediating effects of local conditions, histories, capabilities, and constraints);
• Life course dynamics (e.g., by following health trajectories across life stages)
• Capacities and cost-effectiveness (e.g., by understanding how ambitious health ventures may be configured without overwhelming/depleting capacity—perhaps even strengthening it);
• Value trade-offs (e.g., by developing a deeper analysis of phenomena like the imbalance of upstream-downstream effort, growth of the uninsured, rising costs, declining quality, and entrenched inequalities);
• Organizational management (e.g., by linking balanced scorecards to a dynamic understanding of processes and goals);
• Public deliberation and scenario planning (e.g., by bringing more structure, evidence, and insight to public dialogue and judgment).

Part of what makes SD modeling so well-suited for public health work is that it adheres to a feedback view of causal processes (Richardson, 1991). This perspective stands in contrast to the variable-as-cause orientation that typically frames most population health problems and policies (Susser, 1973, 1991, 2001). The variable-as-cause approach is event-oriented in that it tends to examine one event in relation to another without necessarily understanding the patterns or the structural dynamics out of which those events emerge. To take an extreme example, if it takes six drops of reagent to achieve crystallization in a chemical experiment, a strict event- or variable-oriented causal analysis might erroneously conclude that the first five were ineffective and that only the last drop caused the change. Dose response is, of course, a major factor in conventional causal reasoning. But just imagine if those “drops” corresponded to changes in scores of health-related exposures moving through different pathways and spread out over days, years, or decades. Would we even notice their accumulation, much less their combined influence?

The opposite view, well articulated by the early 20^th century social reformer Jacob Riis, becomes a source of optimism in a world filled with long delays and incremental movement toward goals.

When nothing seems to help, I go and look at a stonecutter hammering away at his rock perhaps a hundred times without so much as a crack showing in it. Yet at the hundred and first blow, it will split in two, and I know it was not that blow that did it—but all that had gone before (Loeb, 1999).

There is much more to the feedback or design view of causality than mere patience and hope (Argyris, 1996; Dent, 2003). As Richardson noted, paraphrasing Jay Forrester, it is all about finding the right vantage point.

The feedback perspective stems from viewing the system from ‘a very particular distance,’ not so close as to be concerned with the action of a single individual, but not so far away as to be ignorant of the internal pressures in the system. (Richardson, 1991)

When a system’s whole structure is understood as the source of observed events—rather than just one or several external variables—there is an incentive to stand back far enough away from potentially misleading or disorienting details and get a fuller picture of the terrain. Some scholars refer to this special point of view as “the overview effect” (White, 1998) or “10,000 meter thinking” (Richmond, 1993, 2000). SD modelers have found that a broad scope is generally needed for finding effective solutions to dynamically complex problems (Homer and Hirsch, 2006; Sterman, 1998). This wide-angle, macroscopic perspective also avoids blaming or scapegoating individuals for seemingly unproductive actions, recognizing that if other people were put in the same position and exposed to the same pressures, they too might behave in similar ways. According to MIT management professor John Sterman, this tendency to blame other people rather than system structure is so strong that psychologists refer to it as the “fundamental attribution error.” Instead, Sterman recommends that we concentrate on “designing organizations in which ordinary people can achieve extraordinary results” (Sterman, 2000:17).

Another benefit to the system-as-cause point of view is that it paves the way for a complementary form of experimental inquiry through simulation. Heightened awareness of our vulnerability to a vast array of terrorist attacks has introduced many public health workers to wonders of scenario planning or simulated event exercises. For example, in May 2004, “more than 100 CDC personnel from all levels of the organization participated in the agency’s first ever full-scale internal emergency management exercise” (Nellis and Birch, 2004). Reflecting on the experience, Duane Smith observed that

It is important to exercise your plan before an event really happens…You need to validate your procedures. You need to see that the plan and reality are the same. It’s also a federal regulation to hold exercises. And it’s a good idea to get people thinking about their roles and how they will support the agency well before a crisis occurs.

If the benefits of rehearsal and simulated response are so great, then why aren’t these techniques used more commonly in other areas of public health work? Why are there now federal regulations to exercise our plans for counter-terrorism, but no such mandate to play out policies for responding to the long list of other risks and diseases that threaten population health to a far greater degree?

Modern computing power has removed many barriers to rehearsing even very complicated scenarios in a compressed time frame. This technology enables planners to evaluate the results of their decisions under controlled conditions rather than relying only on observations in the messy and slow real world. The prospect of developing better health policies through simulation studies offers numerous advantages for public health work. As Sterman puts it,

Even the best conceptual models can only be tested and improved by relying on the learning feedback through the real world...This feedback is very slow and often rendered ineffective by dynamic complexity, time delays, inadequate and ambiguous feedback, poor reasoning skills, defensive reactions, and the costs of experimentation. In these circumstances, simulation becomes the only reliable way to test a hypothesis and evaluate the likely effects of policies (Sterman, 2000).

Another reason for turning to simulation is the potential for SD analyses to yield insight into the ethical dilemmas associated with different decisions. Unintended effects resulting in health inequities, for example, or trade-offs pitting the afflicted against the vulnerable, or people in the present against our children in the future, are examples of the sort of ethical implications that are unlikely to surface when using conventional approaches for policy analysis (Daniels, 2006). Or, if they do surface, to be divisive or to generate short-term actions that eventually worsen problems in the long-term.

Magnussen et. al. provide the example of macroeconomic pro-growth policies in developing countries, pursued by international aid agencies and governments in these countries. These policies, they argue, have the tendency to “provide better opportunities to those with resources and high levels of education, while large segments of the population without these assets are unlikely to benefit and may in fact become casualties of economic transition. Thus, it is the duty of health policymakers to signal when other policies may undermine efforts to promote health equity” (Magnussen, Ehiri, Jolly, 2004:174).

SD modeling is designed to capture the dynamic complexity inherent in feedback systems, large or small, but the technique itself is not necessarily complex.⁴¹ In fact, its use is growing in K-12 schools nationwide precisely because it provides a flexible framework for eliciting students’ mental models and for integrating knowledge across the curriculum. Educators say that it helps young people develop a serious understanding of the complex world in which we live along with a pragmatic outlook better suited to navigating the changes in store (Creative Learning Exchange, 2002; Forrester, 1994; Richmond, 1993). With its broad, endogenous point of view, SD modeling also highlights forces that are under people’s control, instead of purely exogenous influences (if any truly exist). It rightfully endows each of us with insider status in the systems that affect our lives, positioning us as “systems citizens” with all of the responsibilities, powers, and freedoms that full citizenship bestows (Meadows, 1991; Richmond, 2002; Richmond, 2003; Ulrich, 2000).

This notion of systems citizenship explains why a methodology that seems, on its surface, to require proficiency in esoteric mathematics and sophisticated computer skills is not only accessible to us all, but in fact benefits from the inclusion of diverse perspectives. Individuals with special modeling expertise are, of course, critical in any SD project. But their role is best understood as part of a larger group model building enterprise, involving stakeholders with varying points of view and widely diverse talents (Andersen, Richardson, Vennix, 1997; Richardson and Andersen, 1995; Vennix, 1996; Vennix, Andersen, Richardson, 1997). An expert modeler working alone at a computer is unlikely to develop a sound model of real world social dynamics; and even if he or she did, it would be unlikely to be used (Roberts, 1999b). Technical expertise alone is insufficient, in part, because all models require boundary judgments that, in turn, affect which facts are considered relevant as well as normative values about their merit, worth, or significance. Ulrich explains that this essential task of drawing boundaries cannot be justified as the domain of experts alone.

Professional expertise does not protect against the need for making boundary judgments; on the contrary, it depends on them just like everyday knowledge. Nor does it provide an objective basis for defining boundary judgments. It’s exactly the other way round: boundary judgments stand for the inevitable selectivity and thus partiality of our propositions. It follows that experts cannot justify their boundary judgments (as against those of ordinary citizens) by referring to an advantage of theoretical knowledge and expertise. When it comes to the problem of boundary judgments, experts have no natural advantage of competence over lay people. (Ulrich, 2000:8)

Instead, modeling experts and non-experts alike must continually engage in an open dialogue about our problems, how we frame them, and the subsequent implications for change. The inclusive ethic and the emancipatory spirit that animates such critical systems thinking makes it a powerful adjunct to participatory action research (Scholl, 2004) and citizen-centered public health work.

Ulrich also contends that efforts to recognize and critique boundary judgments create the conditions for authentic communication, even in circumstances where there is no consensus or even agreement about facts and values.

Once we understand the role of boundary judgments and know how to deal with them in an open and reflective way, we can grant one another the right to having different rationalities; we can begin to understand, and agree upon, the sources of dissent. Thus we can learn to understand one another even though we cannot agree, as our needs and interests are genuinely different. (Ulrich, 2000:7)

If systems thinking relies on—and indeed facilitates—boundary judgments, which in turn, reveal a plurality of equally legitimate ways of knowing and valuing the world, then we may better appreciate the profound claim, best articulated by public opinion researcher Daniel Yankelovich in his chapter “You Can Argue with Einstein,” that,

For certain purposes, public judgment should carry more weight than expert opinion — and not simply because the majority may have more political power than the individual expert but because the public’s claim to know is actually stronger than the experts’...the judgment of the general public can, under some conditions, be equal or superior in quality to the judgment of experts and elites who possess far more information, education, and ability to articulate their views. (Yankelovich, 1991:220)
 

41. Specialized SD modeling software, such as Vensim (Ventana Systems, 2004) and Stella/iThink (Isee Systems, 2004), makes it possible to develop and use dynamic maps and models without having to do the complex computer programming upon which they are based.

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