Applying the Systems Approach Learning to Act Effectively in a Digital World1 Geoff McDonnell, Adaptive Care Systems and UNSW. April 2012 Contact: [email protected] Contents Synopsis .............................................................................................................................. 1 Background: Real World Decision Making ........................................................................ 2 The Systems Approach ....................................................................................................... 2 Systemic Frameworks ..................................................................................................... 3 A Systemic Perspective............................................................................................... 3 Sociotechnical Systems ............................................................................................... 4 Socio-Ecological Systems ........................................................................................... 4 Structure Agency Framework ..................................................................................... 5 Critical Systems Thinking............................................................................................... 5 Systems Methods in Practice .............................................................................................. 6 Qualitative inquiry and reflection using a Systems Approach .................................... 8 Model Formulation and Simulation .......................................................................... 11 Simulation Testing and Evaluation ........................................................................... 12 Virtual Intervention Experiments ............................................................................. 14 Conclusion ........................................................................................................................ 18 References ......................................................................................................................... 21 Synopsis Systems modelling is the application of a systemic perspective to conceptual mapping and computer simulation to produce effective action. The most important component is the initial qualitative reflection to select the scope, framing and relevant context to satisfy the concerns of those involved and those affected This reflection produces a concept map representation of these concerns and a clear statement of the variables of interest in the language and mental models of the participants. Using both theory and data, this static map can then be converted into a testable computational model of behavior of time. Computer simulation is now a mature and powerful tool for modeling systems to test how different factors may improve efficiency, effectiveness and equity in situations where it is not possible to conduct real-world experiments. We can test whether a policy will actually result in improvement in the right place and at the right time. Faced with future uncertainties we need to perform in silico experiments to design and test policies that cover a range of possible futures. These tools ensure policies are robust enough to 1 See http://insightmaker.com/insight/1121 for its application to eHealth 1 Real World Decision Making Information Individual & Collective withstand these future key uncertainties and help shape the future towards those options Decisions Processes that Task deliverinteraction more desirable and sustainableActions ends for&citizens of the world. Outcomes Communication Background: CoordinationReal World Decision Making Evidence Theory Decisions are the method by which managers turn information into action. These actions Data produce outcomes, which change the information, resulting in further decisions. In joint Experience activities of informing, deciding and acting, these collective processes require interaction, Literature communication and coordination. Information can exist in many forms, including data, Shared experience, empirical evidence, theory and relevant literature. Again, in joint activities people need to share or communicate information. Decision makers use simplified mental models to make sense of the world. They match perceived situations with these models and play out consequences of planned actions in their heads before acting (Klein). Mental models include selective use of key information, filtering of noise, and simplified heuristics or rules of thumb. For joint decisions, we need to share views of the world, especially consensus about objectives and the consequences of actions. All decisions occur in a context, which includes physical and social features. This context is the structure of the system, including the institutions and organisations, policies, social norms, values and goals. This context provides physical resource and social constraints on individual and group behaviours. The Systems Approach We adopt Systemic Frameworks and apply Critical Systems Thinking to real world projects and programs using System Methods. The systems approach seeks to understand interactions among peoples’ viewpoints, values, perceptions and beliefs, in order to learn to take effective collective action. It is an approach to problem solving, by viewing "problems" as parts of an overall system, rather than reacting to specific parts, outcomes or events and potentially contributing to further development of unintended consequences. “The ultimate meaning of the systems approach lies in the creation of a theory of deception and in a fuller understanding of the ways in which the human being can be deceived about his world and in an interaction between these different (biased and deceived) viewpoints. The systems approach begins when first you see the world through the eyes of another 2 1. The systems approach goes on to discovering that every world view is terribly restricted 2. There are no experts in the systems approach 3. The systems approach is not a bad idea.” -C West Churchman The Systems Approach (p231) Systemic Frameworks: A Systemic Perspective A Systemic Perspective represents an iterative unfolding of understanding intended to provide the basis for developing a strategy which, when implemented, is highly likely to address the situation of interest as intended while minimizing the likelihood of unintended consequences. Key features of this way of thinking are: Situation. A situation of interest considered to warrant attention, along with an assessment of the implications of not acting, and a definition of the preferred alternative situation, forms the basis for developing understanding. Behavior. The patterns of behavior represent an unfolding of some aspects of a network of interactions. As such we endeavor to understand the network of interactions responsible for creating the patterns of behavior. Interactions. The network of interactions is the result of some set of actions by one or more stakeholders. As such we endeavor to understand the mental models and motivations of the stakeholders responsible for the situation. 3 Stakeholders. We seek to understand the motivations and the mental models of the stakeholders and the motivations and mental models of those stakeholders who are influenced by the network of interactions. Boundary. Based on an understanding of the network of interactions and stakeholders boundaries are established to keep track of which stakeholders are responsible for which aspects of the network of interactions and which set of interactions are considered to be part of the addressable network of interaction. Challenge Assumptions. It is essential that we challenge those assumptions because decisions made on invalid assumptions are unlikely to support the intended results. Leverage Points. It is essential to identify those leverage points which are likely to transform the current situation into the desired alternative situation. Strategy. The strategy is developed with the intention of migrating the situation of interest in the direction of the desired alternative situation Unintended Consequences. Unintended consequences are typically the result of actions taken without appropriate due systemic consideration. -Gene Bellinger2 Sociotechnical Systems This is an approach to complex organizational work design that recognizes the interaction between people and technology in workplaces. The term also refers to the interaction between society's complex infrastructures and human behavior.3 This broader approach, based on the work of Emery and Trist at the Tavistock Institute, moves from an Input Process Output model of a single system to internal and external interactions among a system of systems. Sociotechnical systems focus on the interactions among people processes and technology and the surrounding organization and external environment. It focusses on managing ongoing technology transitions due to successive adoption of new, sometimes disruptive technologies. Formal models are used particularly in designing and changing processes.4 Socio-Ecological Systems A socio-ecological system consists of 'a bio-geo-physical' unit and its associated social actors and institutions. Socio-ecological systems are complex and adaptive and delimited by spatial or functional boundaries surrounding particular ecosystems and their problem 2 Systemswiki http://bit.ly/zfxzfx 3 http://en.wikipedia.org/wiki/Sociotechnical_systems 4 http://insightmaker.com/insight/1168 4 context. Change in these systems are driven by interacting nested fast and slow adaptive cycles Adaptive management techniques acknowledge doubt and limits of control. This learning while doing approach is referred to as reflexive governance.5 Structure Agency Framework Agency refers to the capacity of individuals to act independently and to make their own free choices. Structure, in contrast, refers to the recurrent patterned arrangements which influence or limit the choices and opportunities available. Social structures enable and constrain human agents, who produce and reproduce social structure.6 Critical Systems Thinking This system of systems (or creative holism) approach adopts a Critical Realist stance, using the scientific method to test assumptions. 1. Critical Realism highlights a mind dependent aspect of the world, which reaches to understand (and comes to understanding of) the mind independent world. A systemic, dynamic and realist evaluation perspective in the pragmatic tradition.7 2. A science is a method of fixing belief, a persistent disinterested pursuit of truth. 3. It is a co-operative social venture, not an individual affair. 4. Its data must be obtained by some form of observation. 5. Its method of dealing with these data is that of rational or logical thought. 6. This includes deductive and synthetic inferences. Synthetic inferences are induction (which relies on relevant representative sampling and self correction) and abduction (the process of arriving at a scientific hypothesis. A surprising fact is observed. But if a hypothesis were true, the fact would be a matter of course. Hence, there is reason to suspect that the hypothesis is true). 7. Its conclusions must be verifiable by observation, experiment or both. 8. Its conclusions are intrinsically provisional and susceptible to further refinement or correction as inquiry is continued.8 5 (see http://insightmaker.com/insight/1169 Panarchy Book http://bit.ly/zqWBwR and the http://www.resalliance.org/ Website) 6 See http://insightmaker.com/insight/1163 and http://en.wikipedia.org/wiki/Structure_and_agency 7 http://bit.ly/wDzynG 8 See http://insightmaker.com/insight/463 5 The key critical systems thinking competencies are Selecting Context, Clarifying Values and Checking Reality. This way of thinking admits that stakeholders, those involved and those affected, may have different ways of making sense of the world, different values and therefore different views of what facts are relevant, important and useful. It provides a formal, pluralistic approach to deciding the scope and level of detail, which includes a broad range of influences. It critically assesses the scope of inquiry (boundary critique) through surfacing and reviewing boundary judgements, observations and concerns from multiple perspectives. It offers useful ways to help persuade stakeholders to start and continue to take collective action.9 This attitude is also referred to as Creative holism (Jackson) Critical systems thinking steps that are useful in argument include problem framing, situation analysis, ideal mapping, choice of methods and evaluation, reflection on methods and challenging of claims using valid argument structures (described by Toulmin,10 Systems Methods in Practice Applied systems methods combine qualitative and quantitative methods for implementing a systems approach, including virtual intervention experiments using computer simulation models. In silico experiments can replace or extend studies that are impossible, too long, too late or too expensive in the real world. They focus on improving real world decisions, extending usual study methods and combining inductive, deductive and exploratory theory development approaches (Axelrod). 9 See http://insightmaker.com/insight/1204 10 see http://insightmaker.com/insight/1175 ) 6 Real World Decision Making Simulation Experiments in silico Experiments and Studies in vivo and in vitro Real world problems are perceived, studied and experimented on, and improvements are designed and external world, the mental world and the social world.11 Virtual experiments inform these real world approaches, especially where real world experiments are impossible to implemented. In this way we learn to make sense of the world together. The world consists of the perform due to cost, time, ethical or other reasons. In this virtual world we make our mental models explicit, shareable and improvable by using concept maps and computational model representations.12 Systems Simulation Method for Virtual Experiments Based on commercial teaching and research experience with health systems simulation projects over the past two decades, we have evolved a joint iterative development approach using hybrid modeling, particularly System Dynamics and Agent based modeling, available in a single commercial software package, AnyLogic or separate specialized toolsets. The method consists of 4 main stages of joint interactive development: 1. Qualitative inquiry and reflection 2. Model Formulation and Simulation 3. Simulation Testing & Evaluation 11 (Historical thesaurus http://bit.ly/yb4q4a ) 12 See http://insightmaker.com/insight/806 7 4. Simulation Policy and Interaction Experiments Systems Simulation Method Simulation Policy & Interaction Experiments Simulation Testing & Evaluation Model Formulation & Simulation Qualitative Reflection Qualitative inquiry and reflection This stage of the systems approach consists of identifying and defining the problem as a series of key questions and conceptualising the subsequent model needed to answer these key questions. In complex problems the key questions may be refined progressively as the project proceeds and a deeper understanding emerges during subsequent analysis. The aim is to share individual mental models through conversation, storytelling and argument within communities of research and practice (Kim 1993). These ideas are expressed in visual maps, images, recordings and narratives. Useful representations include system maps, rich pictures, cognitive maps and concept maps with embedded pointers, images and video clips. Much of this annotated discussion is now available online through websites, blogs, discussion forums and wikis. The steps within the conceptual modelling stage are: 1. Identify and define the opportunity for improvement 2. Conceptualise the Problem Situation 3. Elicit Mental Models The output is expressed as annotated visual maps of linked concepts. 8 Identify and define the opportunity for improvement This step makes the case for an improvement program, taking into account the level of certainty, control and agreement among those involved and those affected. Here we use the Critical Systems Heuristics approach.13 This is a System of Systems approach. It acknowledges that stakeholders, those involved and those affected, have different ways of making sense of the world, different values and therefore different views of what facts are relevant, important and useful. It provides a formal pluralistic approach to deciding the scope and level of detail, which includes a broad range of influences on how people are persuaded to take collective action. Mapping the Boundaries of concern (from Boundary critique in CSH 12x2 questions) About needs: What is the purpose? Who will benefit? How will improvement be measured? About resourcing decisions What resources will be required to make the improvement? Who will decide to allocate and use the resources? Where and when will the decisions be made? About knowledge sources What expertise will be required? Who will provide the required expertise? Who will verify the case? About legitimacy (approval and formal agreement by those affected) How will the case be made legitimate to those involved and affected? How will those affected be involved (voice and power)? Who will verify the proposal for those affected? What viewpoints (worldviews) will be considered? Conceptualise the Problem Situation This step uses Ackoff’s Interactive Planning Steps adapted for the context and audience. Formulate the mess: What are the causes of performance? Ends planning: What are the goals of improvement? Means planning: What are the improvement interventions? Resource planning: What resource decisions are required? Implementation planning: How and when will the interventions proceed? 13 see http://insightmaker.com/insight/1218 9 Each phase includes the governing controls of the processes of agreeing, choosing, cooperating, valuing the common good and ensuring competence.14 Visual metaphors, like Photovoice, are used to collect images that capture the meaning of the situation for the participants An example is available at http://insightmaker.com/insight/782 Typical policy questions about the future are: What are current and future challenges? What are the key policy levers? What are the key measures of performance? What are the probable alternative futures? What are the important key uncertainties? What are the likely scenarios? In the classic system dynamics method, the entry point is a “puzzling dynamic” e.g. why is cross city travel time worse despite billions spent on roads? Why do real estate prices boom and bust (Sterman 2000)? In other problems we may start with describing the current structure to explore and develop theories of behaviour; or to find ways to identify and avoid unintended consequences of proposed policy changes. Elicit Mental Models This step aims to make the participants mental models explicit, shareable and improvable. It answers the following questions. What are the differences in mental models/beliefs that are contributing to the lack of effective action or inability to learn? What do the target audiences already know? What do they need to know and when? How will they learn it? For risk messages the technique is well described by Morgan (2002). 1. Create an Expert model Create a balanced legitimate focused and relevant expert model about the risk and its management, summarized as a concept map showing factors that influence risk management actions 2. Conduct mental model interviews(up to 1 hr each) 14 See http://insightmaker.com/insight/1263 10 3. Conduct 20-30 open ended interviews progressively shaped to cover the topics of the expert model to elicit current beliefs. Include sorting pictures a mix of relevant and irrelevant images/clips 4. Sort into misconceptions, peripheral (irrelevant) beliefs, over-general beliefs, basic background beliefs not in expert models 5. Conduct Confirmatory surveys Design, test, conduct and analyze 6. Estimate the relevant population prevalence of the beliefs among stakeholders 7. Design and test learning/communication strategy to correct/align the beliefs15 Output of qualitative reflection as Concept map: The interim deliverable from the qualitative reflection phase is a concept map which is used as the input to the next phase of computational model building.16 Model Formulation and Simulation In this phase we build the computational dynamic model from the qualitative static concept map. To do this we use a software package with appropriate visual representations to explicitly model the time dimension. System Dynamics representations include stocks (states), flows(transitions) and feedback loops. Agent based and analytical models often use statecharts or mathematical formulae.17 Initially we adopt an aggregated top down view using SD methods, then bottom up agent based if required for more detailed conceptualisation or if specific detailed data sets are available. This combination of multilevel, multimethod approaches is possible due to the object architecture of the simulation software (Borshchev 2004). Discrete event simulation may be an appropriate starting point if most participants view the world as “constrained randomness” or there are critical timing or scheduling events (Morecroft 2005). The SD top down methods involve mapping the context using stocks and flows of key items of interest in a system using a process view, together with the relevant connecting information feedbacks and delays. This produces a graphical, logical structure (wiring) diagram. The mathematical relationships among each of the components in the logical structure are added or estimated from available data or best opinion. The behaviour of the 15 See MG Morgan 2002 ch 2-5 and p 77 for sample size and number of concepts 16 See http://bit.ly/zqJsNQ See general article http://bit.ly/zbaIqT Novak and dynamic feedback Cyclic concept maps at http://bit.ly/zBL8Dq 17 See http://insightmaker.com/insight/1248 11 system over time is then displayed graphically by the computer simulation engine solving a set of difference equations, using integration approximation and other numerical analysis techniques. The agent based simulation method uses object based analysis to define and describe the detail of classes of interest and related functions, UML statecharts to describe their dynamic behaviour and timers and messaging to specific coordination and interaction. Our current practice is to translate our top down SD model into interacting sector objects and, where required, replace some sector objects with more detailed agent based representation. An example of detailed simulation tasks and interim deliverables for a consulting project are in the Appendix (p 21) Simulation Testing and Evaluation George Box remarked, “All models are wrong, some are useful.” Therefore the simulation model needs to be evaluated by assessing whether it is fit for its intended purpose. During this phase the model is progressively refined over multiple iterations with domain experts and differences between the predicted and observed historical behaviours are detected and reconciled. Sterman introduces Validation and Model Testing as Truth and Beauty (Sterman 2000 Ch21). He describes modelling as…” a process of communication and persuasion among modellers, clients and other affected parties. The real test is whether the model helps make better decisions. Therefore we must assess the overall suitability of the model for its purpose, its conformance to fundamental formulation principles, the sensitivity of results to uncertainty in assumptions, and the integrity of the modelling process.” He goes on to list as series of questions that need to be asked of any simulation model (Sterman 2000, table 21-1 p 852) These address the following conceptual validity (can the model answer the questions asked?), structural and behavioural verification, simulation verification and pragmatics and politics of model use. Conceptual validity questions include: What is the purpose of the model? What is the boundary of the model? Are the issues important to the purpose treated endogenously? What important variables and issues are exogenous or excluded? Are important variables excluded because there are no numerical data to quantify them? What is the time horizon relevant to the problem? Does the model include the factors that may change 12 significantly over the relevant time horizon as endogenous elements? Is the level of aggregation consistent with the purpose? Structural and behavioural verification questions are: Does the model conform to basic physical laws e.g. conservation of matter Are the equations dimensionally consistent without fudge factors. Is the stock and flow structure explicit and consistent with the model purpose? Does the model represent disequilibrium dynamics or does it assume the system is in or near equilibrium all the time? Are appropriate time delays constraints and possible bottlenecks taken into account? Are people assumed to act rationally and to optimise their performance? Does the model account for cognitive limitations, organisational realities non-economic motives and political factors Are the simulated decisions based on information the real decisionmakers actually have? Does the model account for delays distortions and noise in information flows? Simulation verification questions: Is the model robust in the face of extreme variations in input conditions or policies? Are the policy recommendations sensitive to plausible variations in assumptions including assumptions about parameters aggregation and model boundary? Questions on the Pragmatics and Politics of Model Use Is the model documented? Is it publically available? Can you run your model on your own computer? What types of data were used to develop and test the model (e.g. aggregate statistics collected by 3rd parties, primary data sources, observational and field- based qualitative data, archival materials, interviews)? How do the modelers describe the process they used to test and build confidence in the model? (See Oliva, 2003). Did critics and independent third parties review the model? Are the results of the model reproducible? Are the results “add-factored” or otherwise fudged by the modeler? How much does it cost to run the model? Does the budget permit adequate sensitivity testing? How long does it take to revise and update the model? Is the model being operated by its designers or third parties? 13 What are the biases, ideologies and political agendas of the modelers and clients? How might these biases affect the results both deliberately and inadvertently? John Morecroft summarises these model tests as Tests of Behaviour Tests of Structure Tests of Learning Virtual Intervention Experiments In this phase we design and perform virtual experiments to test the effects of our planned interventions. These experiments may range from formal sensitivity analysis using Monte Carlo methods or Optimizers, Laboratory or Web-based Learning Environments, or computer and traditional board games. Policy design ranges from changing parameter values to creating entirely new strategies, structures and decision rules. These include changing time delays and the flow and quality of information available at key decision points or fundamentally re-inventing the 14 decision processes of the actors in the system (Sterman). Lyneis(1988) describes common mistakes in corporate policy as . . . policy isolated by functional area. This is like having three pilots simultaneously trying to fly a single plane using three sets of controls. When these problems are fixed in systemic structures, three different cockpits are built into a single plane! policy isolated from establishing goals (akin to no agreed flight destination, so pilots are continually re-setting the flight destination) policy design isolated from context (like ignoring the weather outside when planning the flight). Other problems with policy can be related to complexity, due to interactions among parts and the environment, interactions being often more important than components and because actions taken to correct an immediate problem may make matters worse in the future. The policy design approach in service industries involves the following: Represent the common service demand pattern; Describe how information about demand and capacity are used, how resourcing goals and staffing levels are set and how discrepancies between goals and actual are corrected. Identify problem behaviour Construct a computer model of the relationships believed to cause the behaviour Develop an understanding between structure and behaviour Design policies that improve behaviour Test the policies under a wide range of market demand patterns and varying other key assumptions and scenarios Testing policy can also use standard engineering variations in the system input pattern of demand, such as: Step, Noise, Seasonal, Cyclical, Growth, Decline. Scenario Planning and Robust Policy Design In any simulation of the future there will always be important key uncertainties. Therefore, we need to perform in silico experiments to design and test policies that cover a range of possible futures. These tests ensure policies are robust enough to withstand these future key uncertainties and help shape the future towards those options that deliver more desirable and sustainable health care futures for the widest range of citizens. 15 Policy testing involves systematically varying interactions especially among structures, policies, system constraints and individual behaviours (Rouwette). Behaviour sensitivity is the policy outcome sensitivity to unknown parameter values. Policy sensitivity analysis tests the assumptions of the method, the influence of the level of aggregation of the model, and the assumptions of non-linear economic relationships. Stochastic optimisation of policy space is performed if an agreed objective function exists (Moxnes). If not, we must represent the various tradeoffs of policies among key performance dimensions, such as cost, efficiency, equity and effectiveness. Simulation as a Collaboration Artefact to share and improve mental model Group model building and policy design and testing can help build consensus about goals, relevant context, and policy impacts (Zagonel 2003). These group interactions can take place as computer games or board games based on the structural system constraints and policy rules of the game. Significant improvements in user interfaces and dynamic graphic animations allow a wider range of non-technical participants in these group activities, including patients families, interested and citizens and political interests. Policy Flight Simulator Experiments to Evaluate the Impact of Systems Simulation on Decision makers Widely accepted game and policy simulators have been used to provide a standard test environment for more controlled experiments about decision-making influences on participants. The most common of these games is the classical supply chain Beer Game, which explores ordering behaviour associated with delivery delays (Croson et al). 16 Planning the Policy Experiments using the Simulator This follows the approach to more classical behavioral experiments in controlled environments. These experiments are being focussed on the key research questions in simulator use: How can we improve the mental models of decision-makers and will this result in better policies? Study designs include baseline and repeat measures pre-simulator and early and late postsimulator flight training, with controlled variations in the influences under test, to test specific hypotheses (Gary). Decision-making influence areas for policy flight simulator experiments There is an opportunity for many observational and controlled experiments with hypothesised influence variables. Some design questions are: What are the observed roles and decision patterns? Who are the agents playing the model? What are the key decision patterns about overall flows? Who makes them? How? Why? What are the local social and cultural influences, performance metrics, information feedbacks and ideologies in use? What are the effects of service availability and policies, resource constraints and incentives especially financial? What are the relevant political dimensions (ownership, knowledge, collective action resources (access to formal power, informal influence, identity, cohesion and trust, social networks)? Performing Policy Experiments/Serious Play (mixed in vitro and in silico tests) Some future research questions about the interaction between individual decision-making and policies in complex systems: What are the key decisions about individuals (which patient bed services etc)? Who makes them? How? Why? What are the heuristics and objective functions used? What are the hypotheses about agents' roles in playing the simulation/game? What are the behavioural influences such as lack of knowledge and lack of trust? What are the differences between novices and experts? How do Computer simulation and board game interactions differ? What are the effects of differences in individual learning styles? How and when are mental models malleable ? What are the quantitative impacts of known confounders such as cognitive ability and complexity of task? How effective are group simulations in conflict resolution and building consensus? 17 Progressively expanding and testing the scope of the model to include individual behaviours Agent based representation make it possible to incorporate the subject behaviours observed in simulation and game experiments into successive versions to provide more flexible and complete decision-making and learning environments. This concept is similar to having the simulator play the roles of individual decision-making participants (like a chess game) or having the audience “step into” a play. Conclusion Systems simulation, using multilevel multimethod approaches, is being applied to improve the understanding of complex systems in science, biology, engineering, business and health (Kitano). In silico policy experiments extend the scientific method used for in vitro and in vivo testing, particularly when rigorous experiments are used in designing and testing policy flight simulators. These simulated learning environments can be progressively refined to provide more realistic decision-making challenges. We need to test how these broader simulations can improve mental models in various contexts and improve both systems design and individual contexts. Benefits of these system simulation tools include: ·Helping to define the relevant system and its boundaries; ·Combining high-level system structures with detailed individual behaviours; ·Demonstrating dynamic behaviour of a system by playing out the logic, relationships and feedback loops that are built into a system; ·Playing out long-time spans quickly and predicting the performance of the system over time; ·Promoting cross-functional understanding, team building & organisational learning; ·Helping identify potential unintended consequences and developing strategies so that they can be avoided or their impact mitigated; ·Enabling search for high leverage points in a risk free environment; and ·Guiding additional data collection by showing where estimates need to be most precise. We are beginning to use these tools for sustainable health system design by solving relevant, focussed, health policy problems. In the future we hope to collaborate with teams that can integrate several system and discipline areas in more compelling and realistic versions. Current application areas include the acute aged care interface, the health workforce, technology diffusion and the use of information and communications 18 technology. Ongoing research and development work with health policy makers and clinical decision-makers should turn health systems simulation into a useful, practical toolkit for health systems improvement. Philosophical Note (Olaya 2009) This explicit representation of our mental models in SD has been described as presentationalism, idealism, a search for positive knowledge within an idealist epistemology, instrumentalism "not claims about the world but instruments for systematizing observations and for boosting learning processes using experimentation via simulation", and relativism rather than positivism. To summarise Systems Methods and Models focus on better policies and explanations of problems explain causal structure of a problem or situation explain how the problem is created by this structure explain why one policy has high impact while others do not explain how established and defended policies are the underlying cause of the problem behaviour argue in favour of better policies Formal computer models are constructed following the scientific method reference mode of problem behaviour, dynamic hypothesis, formal model, testing of the hypothesis against data, extensive analysis, and policy design. A full page table of activities, representations at each stage and information sources follows. Diagram examples are taken from http://insightmaker.com/insight/1400 19 20 Appendix Example of Detailed Simulation Tasks and Interim Deliverables for a Consulting Project Qualitative Reflection Phase Confirm Scope, Context and Level of Detail What questions does the model need to answer? Deliverable: Issue Diagram What Data Sources are available? Deliverable: Documented Data Sources What are the key System Performance measures and techniques? Deliverable: The Case for Improvement including Key Performance Indicators (KPIs) and the Boundary of concerns Deliverable: Concept map Develop Initial SD System Simulation using Ithink software and method (Richmond) Convert to Object Sectors in AnyLogic Add agent-based AB detail to selected sectors Define Object Classes and Variables using class diagrams Define Dynamic Behaviour of Objects using UMLstatecharts Develop AB Simulation Logic using logic functions, agents and actions Develop Initial Animation Display using animation objects Calibrate Model against Existing System Performance Deliverable: Calibrated simulation and Documented Data Gaps (Calibration is from Existing Data Sources, Delphi best estimates, and special surveys if indicated by sensitivity analysis) Refine Model using Joint Interactive Group Modelling. Develop and Simulate Future Case Scenarios Deliverable: Model with what-if Simulations Select Scenarios (workshop) Extend model to handle scenarios Refine Animation and Interactive User Controls Refine Calibration Data Sources for Parameter Estimates There is a wide range of sources for estimating parameters in simulation models; these estimates can be progressively refined, where possible, over the life of the model. Sources we have used in previous projects include: Published literature in journals and conference papers; Publically available data from ABS and AIHW; Internal data from organisations; Focussed studies conducted by special interest groups; Commonwealth and State Health Data Sets; and Estimates from other relevant modelling and simulation studies performed by Government departments, academics and modelling organisations such as NATSEM, Access Economics, IPART, Productivity Commission, CSIRO, HILDA and the like. 21 References Axelrod, Robert. (2003) Advancing the Art of Simulation in the Social Sciences UMICH Aug 2003 http://www-personal.umich.edu/~axe/research/AdvancingArtSim2003.pdf Axelrod Robert and Tesfatsion Leigh. A GUIDE FOR NEWCOMERS TO AGENTBASED MODELING IN THE SOCIAL SCIENCES. To appear in the Handbook of Computational Economics, Vol. 2: Agent-Based Computational Economics, edited by Leigh Tesfatsion and Kenneth L. Judd, Handbooks in Economics Series, North-Holland, Amsterdam, the Netherlands. available at http://www.econ.iastate.edu/tesfatsi/abmread.htm Borshchev, Andrei & Filippov, Alexei. From System Dynamics and Discrete Event to Practical Agent Based Modeling: Reasons, Techniques, Tools www.xjtek.com/download/papers Churchman CW (1979) The Systems Approach. Delacorte Press, New York. Paperback edition Dell Publishing, New York, 1969, second, revised ed. Croson, Rachael, Donohue, Karen, Katok, Elena and Sterman, John. Order Stability in Supply Chains: Coordination Risk and the Role of Coordination Stock http://web.mit.edu/jsterman/www/Order_stability.html Eddy, David. Archimedes Model http://www.archimedesmodel.com/archimedes.htm Forrester, Jay W. (1980). Information Sources for Modeling the National Economy. Journal of the American Statistical Association 75(371): 555-574 Forrester, Jay W. (1974) Understanding Social and Economic Change in the US. Luncheon Address Summer Simulation Conference 1974 MIT D-memo 2103 Gary, Shayne and Wood, Robert.(2005) Mental Models, Decision Making, and Performance in Complex Tasks, Abstract Paper ISDC Boston July 2005 Goude, Thomas A (1950) The Thought of C S Peirce University of Toronto Press, Toronto Gundersen, Lance and Holling C S (2001) Panarchy: Understanding Transformations in Systems of Humans and Nature. Island Press Homer, Jack B and Hirsch Gary B. (2006) System Dynamics Modeling for Public Health: Background and Opportunities. Am J Public Health. March; 96(3): 452–458. Jackson MC (2003) Systems Thinking: Creative holism for managers. Wiley Kim, Daniel H (1993) The Link Between Individual and Organizational Learning MIT Sloan Management Review; Fall 1993; 35, 1 p37-50 Kitano, Hiroaki. (2002) Systems Biology: A Brief Overview. Science 295, 5560, 1662-4 Ison, R.L., (2008). Systems thinking and practice for action research, in Reason, P., Bradbury, H. (Eds.), The Sage Handbook of Action Research Participative Inquiry and Practice (2nd ed.), London, Sage Publications, 139-158. Klein, Gary (1999). Sources of Power: How people make decisions MIT Press, ISBN 0262611465 www.decisionmaking.com Lane, David C. (1999) Social theory and system dynamics practice. European Journal of Operational Research. 113 501-527. 22 Lyneis, James M (1988) Corporate Planning and Policy Design A SD Approach PughRoberts Assoc Cambridge MA ISBN 0-262-12083-6 Meadows, Donella. The Unavoidable a Priori. D-4880 MIT D-memo Meadows, Donella H (2008) Thinking in Systems a primer. Chelsea Green Publishing, Vermont Morecroft, John with Robinson, Stewart. (2005) Explaining Puzzling Dynamics: Comparing the Use of System Dynamics and Discrete-Event Simulation. ISDC Boston July 2005 Morecroft, John (2007) Strategic Modelling and Business Dynamics. Wiley Morgan, M Granger, Fischoff B Bostrom A and Atman CJ (2002) Risk communication: a mental models approach Cambridge University Press Cambridge UK Moxnes, Erling (2005) Policy sensitivity analysis: simple versus complex fishery models Syst. Dyn. Rev. 21, 123-145, Novak JD and Gowin DB (1984). Learning How to Learn. Cambridge University Press Cambridge Olaya, Camilo (2009) System Dynamics: Philosophical Background and Underpinnings Encyclopedia of Complexity and Systems Science Springer Volume 9 p9057-9078 Oliver, Adam and Mossialos, Elias. European Health Systems Reforms: Looking (2005) Backward to See Forward? Journal of Health Politics, Policy and Law, Vol. 30, Nos. 1–2, February–April. p7-28 Oliva R. (2003). Model calibration as a testing strategy for system dynamics models. European Journal of Operational Research 151(3):552-568. http://iops.tamu.edu/faculty/roliva/research/sd/calibration.html Reid, Proctor P. Compton, W. Dale Jerome H. Grossman, and Gary Fanjiang, Editors, (2005) Committee on Engineering and the Health Care System, National Academy of Engineering, Institute of Medicine Building a Better Delivery System: A New Engineering/Health Care Partnership. National Academies Press http://www.nap.edu/catalog/11378.html Richmond, Barry. ithink User’s Manual. High Performance Systems, Hanover NH www.hps-inc.com Rittberger, Berthold. (2003) Endogenizing institutional change: Moving beyond the institutionalist ‘holy trinity’ Paper prepared for panel 12-9 ‘Beyond Institutionalism’ at the 2nd General Conference of the European Consortium for Political Research, 19-21 September 2003, Marburg. Rouwette E, Grossler M and Vennix J. (2004) Exploring Influencing Factors on Rationality: A Literature Review of Dynamic Decision-Making Studies in System Dynamics. Systems Research and Behavioral Science 21, 351-370 Sterman, John D (2001) System Dynamics Modelling: Tools for Learning in a Complex World California Mgt Review 2001 43(4) 8-25 and 1994 Sterman, John D Business Dynamics: Systems Thinking and Modeling for a Complex World. McGraw-Hill 2000 ISBN 0-07-231135-5 www.mhhe.com Toulmin, S. (1958) The Uses of Argument Cambridge University Press, 2nd edition 2003 23 Towill, Denis R. (1996) "Industrial dynamics modelling of supply chains", Logistics Information Management, Vol. 9 Iss: 4, pp.43 - 56 Ulrich W (2012) Operational research and critical systems thinking—an integrated perspective Part 1: OR as applied systems thinking Journal of the Operational Research Society advance online publication, 14 December 2011 doi:10.1057/jors.2011.141 Ulrich W (2012) Operational research and critical systems thinking—an integrated perspective Part 2: OR as argumentative practice Journal of the Operational Research Society advance online publication, 14 December 2011 doi:10.1057/jors.2011.145 Zagonel, Aldo. (2002) Conceptualization in Group Model Building…Tension between Representing Reality and Negotiating a Social Order. ISDC Palermo July 2002 http://www.systemdynamics.org/conf2002/proceedings/308Zagonel.pdf 24
© Copyright 2026 Paperzz