Helping decision makers frame, analyze, and implement decisions
Managers, policy makers, and decision makers with responsibility for environmental decisions have an extraordinarily difficult job. The systems they manage are complex (coupled human-natural systems), with many dimensions and complicated dynamics. Our knowledge of how those systems respond to management actions is often limited, so many of the decisions have to be made in the face of uncertainty.
All decisions have the same recognizable elements. Context, objectives, alternatives, consequences and deliberation. Decision makers and analysts familiar with these elements can quickly see the underlying structure of a decision.
There are only a small number of classes of decisions. These classes differ in the cognitive and scientific challenge they present to the decision maker; the ability to recognize the class of decision leads a decision maker to tools to aid in the analysis.
Sometimes it is worth collecting more information, sometimes it isn’t. The role of science in a decision-making process is to provide the predictions that link the alternative actions to the desired outcomes. Investing in more science is only valuable to a decision maker if it helps to choose a better action.
Implementation. The successful integration of decision analysis into environmental decisions requires careful attention to the decision, the people, and the institutions involved.
The field of decision analysis provides a comprehensive set of tools for structuring, analyzing, and making decisions. Arising initially as a way to understand and manage risk, the field has expanded in the last 80 years to cover such topics as multipleobjective trade-offs, time-dependent linked decisions, the value of information, and competition among multiple decision makers. At the same time, cognitive psychologists have studied human decisions to understand when our innate processes work and when they fail.
Formal decision analysis is increasingly being used in many sectors, including economic, industrial, manufacturing, agricultural, transportation, and medical sectors, by individuals, corporations, non-profits, and government agencies. Occasional application of decision analysis to environmental decisions began in the mid-1970s, but the environmental science and management world was largely unaware of decision analysis until the late 1990s and early 2000s. (A major aim of the ARC Centre of Excellence for Environmental Decisions – CEED – was to better connect decision analysis and environmental management).
Today, the field of environmental decision analysis is coming of age with the maturation of a rich set of tools to help decision makers frame, analyze, and implement decisions. Sometimes the wealth of tools is daunting, but there are core principles that tie decision analysis together. To help the reader navigate this complexity, we summarize the field of decision analysis in four key messages.
1. All decisions have the same recognizable elements
The decisions we have to make are normally surrounded by a fog of complexity. One of the most valuable contributions of the field of decision analysis is the recognition that all decisions have a common set of elements: context, objectives, alternatives, consequences, and deliberation. These elements form a useful guide for the decision maker.
Context: Albert Einstein once said that if he had 1 hour to save the world he would spend 55 minutes defining the problem and 5 minutes finding the solution. Clarity about the decision context leads to an efficient search for solutions.
Understanding the decision context begins with some simple questions (simple to ask, anyway). What is the problem that is being faced and what triggered it? Who will make the decision and what authority do they have to act? Who else cares about and can influence the decision? What is the geographic and temporal scope of the problem? What ecological, social, and legal background is relevant?
Objectives: The decisions that we make are driven not by science but by the values we hold. Understanding the value sets of the decision makers, the organisations responsible for decisions, and the key stakeholders is a crucial step in a structured approach to decision-making. The values at the heart of any decision are the fundamental objectives—the long-term outcomes we are trying to achieve by implementing a course of action.
There are often many of these objectives, and they might conflict. Thus, we might care about the long-term conservation of wetland birds at the same time we care about maximizing agricultural production and minimizing flooding risk.
We often lose track of fundamental objectives by focusing instead on means objectives. These are potential ways by which we might achieve our fundamental objectives. For example, we might focus on target levels for irrigation withdrawal and argue over those, forgetting that what we really care about are the birds, the crops, and the flood protection.
Alternatives: Once we know what we want to achieve with a decision, it’s possible to start thinking about what we could do to achieve these outcomes. Unfortunately, most people reverse this process, assessing what they could do before understanding why they are doing it.
While working out what actions we can take to achieve our objectives seems simple, the process is hampered by psychological biases. Often we anchor on the decisions we have made in the past or small variations on those decisions (see Decision Point #93). But to achieve a good decision, we need to be creative, to think beyond our current set of actions and our means objectives, and instead focus on developing actions that may help us achieve our fundamental objectives and expose trade-offs within these objectives. Indeed, given the challenges we face in environmental management, creative solutions are needed more than ever, and sticking with business as usual is unlikely to solve the problems we have.
A creative search for alternative actions should challenge constraints. Many constraints to a decision are only perceived, not real. We may be told, for example, that a decision has a firm budget constraint, but creative consideration of leveraged funding might open up new alternatives.
Prediction of consequences: The role of science in decision making is to make predictions. There are no decisions without prediction. If you decide what bus to take, you think about what time you need to get somewhere and predict which bus will get you there in time. We make predictions daily for every decision we make.
What are these predictions? They are a mapping of the actions we are choosing among (the alternatives) to the things we want to achieve (the objectives). The challenge for the scientist is to use the best available information to make these predictions, while comprehensively accounting for uncertainty.
Selection of preferred alternatives: By this point of the process we have evaluated each action or set of actions against each objective. But no decision has been made, yet, and indeed, the best choice might not be obvious. In the deliberations to find a preferred alternative, the decision maker might need to weigh the multiple objectives to find the right balance, or consider the risk posed by uncertainty, or consider how this decision will affect future decisions. These are challenges at the interface of science and values, and the field of decision analysis provides a large set of tools to help.
2. There are only a small number of classes of decisions
Many decisions resemble each other, and this recognition is one of the ways that people become better decision makers and analysts, because they realize they have made a similar decision in the past. For a decision analyst, each class of decisions has a common structure, poses a similar challenge to the decision maker, and benefits from a similar set of analytical tools. Although the circumstances and details of each environmental decision are unique, recognizing the class of decision accelerates the process of analysis. Here we discuss five classes of decisions that arise commonly in environmental contexts.
1. Risk analysis: Coupled socio-ecological systems are complex, our knowledge of how they respond to management actions is limited, and some aspects of their dynamics are out of our control. Thus, many environmental decisions are made in the face of uncertainty and are open to risk. A grassland manager, for example, in an effort to hold back succession, can choose between mowing and burning. The effects of mowing are better understood and more controllable, but the disturbance does not return the same nutrients to the soil nor does it induce germination. On the other hand, even with careful preparation, it is hard to fully control a burn. What are the risk trade-offs in choosing among these actions?
Risk decisions are ones in which the precise outcomes of the alternative actions are not known prior to implementation. The challenge for the scientist is to estimate the probabilities of the various outcomes under the different alternatives. The challenge for the decision makers is to articulate their degree of risk tolerance. The methods for risk analysis are well developed and have been applied in a number of environmental contexts, including endangered-species management, biosecurity, and pesticide regulation.
2. Project prioritisation: Consider an agency that has $300,000 to spend to restore instream, riparian, and upland habitat for the benefit of a suite of threatened endemic species. The agency has received several dozen proposed projects, ranging in cost from $10,000 to $175,000. Taken together the projects cost $1.4 million, far in excess of the available funds. Which of the projects should be funded to assure the most benefit for the species in question? This common class of problem goes by many names including portfolio allocation, the knapsack problem, or combinatorial optimization. In the environmental world it is frequently called project prioritization (see Decision Point #29, p8-10; and Decision Point #103). The challenge is to select a subset of a large number of possible projects, subject to a budget or resource constraint, that maximizes the combined benefit.
The number of possible portfolios to select is typically very large (for example, with 24 projects to select from, the possible number of combinations is 224 = 16.8 million). The benefit of adding a particular project may depend on what other projects are in the portfolio, because there could be synergies, conflicts, or redundancies among projects. The analytical challenges are to calculate the net benefit of a candidate set of projects, then search among the large set of possible portfolios for the optimal one.
Note that a broad set of decisions is included in this class. The ‘projects’ could be species to save, invasive species to combat, research projects to fund, land parcels to acquire, varieties of actions to implement, etc. The resource constraints could be money, staff time, the total number of projects to undertake, or even the number of individuals of a threatened species available for translocation.
3. Spatial planning: Spatial planning is a special case of project prioritisation. The individual ‘projects’ are spatial elements (eg, land parcels, grid squares), and the challenge is to identify the best set of spatial units to include in the reserve to maximize some measure of environmental benefit, while either staying within budget or minimizing the area required. The nature of the interactions among the units can be important, reflecting the benefits of connectivity but possibly also the value of spatial independence.
In the past two decades, a set of tools has been developed to aid governments or other management agencies in spatial planning: identifying where to locate and how to configure areas that should receive some sort of protection to conserve their natural processes. Related tools are used in contexts like urban planning, electoral districting, and school zoning.
4. Recurrent decisions: Many natural resource management decisions are made repeatedly; the same, or a similar, decision is revisited on a recurrent basis. For example, hunting seasons, fishing catch limits, or quotas for subsistence take are often set on an annual basis. Likewise, timber management decisions within a large forest are made recurrently. In some parts of the world, the inundation of wetlands is managed on a seasonal or annual basis to provide habitat for migrating species, encourage primary productivity, or set back succession.
Recurrent decisions are challenging because the systems are dynamic. An action taken in one cycle, which might generate a short-term benefit, changes the trajectory of the system, affecting the actions that can be taken subsequently as well as the benefits that will arise from them. To calculate the long-term benefit of taking an action today, we have to anticipate all the subsequent actions that will be taken over the timeframe of management.
5. Multiple objective trade-off decisions: One of the emblematic environmental issues of the late 20th Century in the United States was the development of the Northwest Forest Plan, which exposed conflicts between conservation of the northern spotted owl and maintenance of an old-growth timber industry. These challenges are ubiquitous in environmental decisions—humans have multiple objectives they would like to achieve from an ecological system, like biodiversity conservation, economic growth, recreation, provision of ecosystem services, and maintenance of subsistence harvest, but it might not be possible to achieve them all to their highest degree. How should these trade-offs be managed?
Multi-criteria decision analysis provides a set of tools for structuring, analyzing, and negotiating multiple-objective decisions. Coupled with the growing field of non-market valuation, these tools can help decision makers grapple with the core values question at the heart of these decisions—what is the relative value of the multiple objectives?
3. Sometimes we need more information, sometimes we don’t
Uncertainty surrounds all the decisions we make. It’s something we must capture in our predictions and consider in our final decision. Sometimes, reducing that uncertainty can improve our ability to make a decision. However, sometimes additional learning won’t change our choice of action. Recent advances provide tools for understanding when new science is needed.
Monitoring: Monitoring is generally viewed as a ‘smart’ activity in the pursuit of improved conservation outcomes. By explicitly asking the question, “is spending money on monitoring justified?”, we must be prepared to not monitor in some cases. Good monitoring rests fundamentally on a clear justification for acquiring information in the first place. That is, what we strive to know should be driven by what we need to know (see Decision Point #52, p4-6, and see the box on analysis and monitoring).
Value of information: Most information gathered about the natural world does not help inform decision-making. This is the crux of value-of-information analysis (see Decision Point #67). It is about asking whether money and time spent collecting data might change our decision. It recognizes that gathering information costs money, delays actions, and takes resources away from management. Formal value-of-information analysis calculates how much a decision maker’s expected outcomes will improve if uncertainty can be resolved prior to committing to a course of action.
Adaptive management: Adaptive management, sometimes called ‘learning whilst doing’, is a formal approach to making linked decisions in the face of uncertainty. More than just trial and error, adaptive management embeds the idea of value of information into monitoring design and experimental actions, so that management actions can improve over time (see Decision Point #102).
4. Implementation: requires careful attention to the decision, the people, and the institutions involved
Environmental decisions are hard, not only because their elements are complex and multi-faceted, but because the political and institutional settings in which they are made are filled with passionate, contentious and diverse people. Faced with this complexity, decision-analysis tools are very attractive. They give us the hope of structured, rational deliberation. But these tools don’t actually make the decisions. They are only aids for the decision makers, not prescriptions. Further, we are as distrustful of other people’s processes as we are of their policies, so the proposal to even undertake decision analysis is itself often contentious.
If these reservations can be overcome, however, decision analysis does offer a rich set of tools to enhance the deliberation behind environmental decisions. Overcoming those reservations requires attention to the interpersonal, institutional, and political dynamics surrounding the decision.
Here are three suggestions that may help overcome some of these reservations?
(1) Use rapid prototyping to sketch an initial decision analysis. Minimal initial investment with the opportunity for substantive input may invite decision makers and stakeholders into the process.
(2) Pay attention to who the decision maker is. In complex institutions (like government agencies), it may not be immediately clear who will end up making the decision.
(3) Recognise the importance of stakeholders and the possibility of multiple decision makers. Deeply political settings involve multiple decision makers, who may not agree to cooperate. Recognition of this dynamic can lead to strategies for negotiation, compromise, and forward progress.
Making good decisions about and for the environment is enormously challenging. Rather than throwing our hands up and saying it’s all too hard, decision makers stand to make considerable gains if they can better engage with the rich and growing field of decision analysis.
More info: Eve McDonald-Madden email@example.com
Decisions about willows in the high country
Here’s an example of how structured decision making (SDM) was used to help develop a long-term management strategy for the invasive gray sallow willow up on the Bogong High Plains of Victoria (Moore & Runge, 2012). The stakes were high with the prospect of this very invasive willow taking over an endangered alpine ecosystem. There is great uncertainty surrounding the available options and, looking into the future, this is compounded by a changing climate and shifting fire regimes.
Structured decision making involves working with key stakeholders involved in a problem to create an agreed framework around the decisions they need to make. The process used involved setting a context, agreeing on objectives, listing the various available options to meet these objectives and devising ways to compare the costs and benefits of those options.
During a three-day structured decision making workshop it emerged during this time that the major decision faced by the willow managers was where to focus their control effort. The removal of willows from EPBC listed alpine bogs was identified as the main objective of management so the control of willows in bogs seemed a good place to start.
However, willow establishment is facilitated by wildfires which are predicted to increase with climate change. Should some effort be allocated to removing mature seed-producing willows in nearby waterways to prevent invasions in the future? If so, how much effort? In the absence of a strategy, managers had been allocating effort to bogs and waterways guided by a combination of intuition and available resources.
The scientists used structured decision making to formally describe this decision. They then built a stochastic model of the spread of willow onto the Bogong High Plains and used it to calculate how different management strategies would affect the amount of willow in bogs over the next 200 years. The model contained approximately 40 factors or parameters; most of these had never been measured and were highly uncertain.
They incorporated this uncertainty into the model by choosing parameters from probability distributions (that represented our uncertainty about the parameter values) and re-running the model 10,000 times. Using the model, they identified where control effort should be focused to minimise the abundance of willows in bogs over the longer term. This was done by calculating which management alternative worked best on average across the 10,000 possible scenarios.
The optimal strategy was to allocate all available effort to the bogs until the budget exceeded 2,000 work days per year. It needs to be noted that 2,000 work days per year is four times the current budget levels. Beyond this point it was optimal to allocate some effort (20-60% total budget) to eliminate populations of seed-producing willows in nearby rivers. Effort was allocated to the closest populations first and then to more distant waterways as the budget increased.
Moore JL & MC Runge (2012). Combining Structured Decision Making and Value-of-Information Analyses to Identify Robust Management Strategies. Conservation Biology 2: 810–820. And see the story on this paper in Decision Point #67
Decision analysis and monitoring
Decision analysis is a procedure for discriminating between suitable courses of action; in the case of monitoring, it helps to select the most appropriate regime for monitoring (see Fig. 1: Decision 6, Question 13) or management (see Fig. 1: Decisions 10, 14, 15). Decision analysis involves a structured enquiry into the different options available to manage or monitor, along with their costs, benefits and constraints. A simple form of decision analysis ranks options according to their expected cost effectiveness, or expected benefit divided by expected cost.
When deciding which management option is best, one must consider the benefit of each possible action in terms of reaching the overall program objective, the probability of success of that action, and of course the cost of implementing that action.
To select the best monitoring regime we include the same components, but the benefits now relate to the quality of information needed to make a decision based on the reason for monitoring (eg, track system state to guide state-dependent management, or track performance to guide adaptive management). Furthermore, adequate monitoring must consider the ability of the strategy to detect changes in the system.
Acquiring information on benefits and costs for a decision analysis can be achieved through expert elicitation or through more detailed scenario modelling. Options for implementing decision analysis range from a simple calculation of the combined benefits relative to the total costs incurred (eg, Benefit x Probability of success / Cost) to a more complex optimisation (eg, stochastic dynamic programming or reinforcement learning). Methods of obtaining data and implementing decision analysis vary in their cost and their ability to provide rigorous results.
McDonald-Madden E, PWJ Baxter, RA Fuller, TG Martin, ET Game, J Montambault & HP Possingham (2010). Monitoring does not always count. Trends in Ecology and Evolution 25: 547-550. And see Decision Point #52, p4-6.
Management thresholds & seaweed
In order to adaptively manage protected areas, conservation managers need to know when to implement management actions to prevent ecosystems trending towards an unfavourable condition. Whilst ecological research and monitoring can help define unfavourable ecosystem conditions; the question of when to implement a management action requires value judgements by decision-makers. Such judgements require decision-makers to subjectively trade-off competing objectives (eg, there is a trade-off between environmental (eg, biodiversity benefits), social (eg, visitor satisfaction) and economic (eg, the cost of management actions) objectives).
Decision scientists worked with Parks Victoria to trial a structured decision making (SDM) process to explore where to set management thresholds for the intertidal brown alga, Hormosira banksii, at Port Phillip Heads Marine National Park (Addison et al, 2015). Hormosira (also known as Neptune’s necklace) is an indicator of the condition of invertebrate and algal communities on Victoria’s rocky intertidal reefs. Parks Victoria has identified that a key threat to intertidal reef communities is trampling by humans.
The challenge for Parks Victoria is this: If the condition of Hormosira starts to decline in the future, at what point should a more intensive management strategy be implemented to minimise the impact of trampling?
The SDM process involved Parks Victoria staff (decision-makers and on-the-ground rangers) and marine scientists with expertise in intertidal ecology. All participants could contribute valuable experience and knowledge of the management of marine national parks and the effectiveness of biodiversity protection.
Participants developed a series of management objectives and alternative management actions relevant to the decision context. The management objectives represented environmental factors (eg, to improve the condition of Hormosira), social factors (eg, to improve visitor satisfaction) and economic factors (eg, to minimise resources spent), all of which were considered fundamentally important to the decision context by participants.
Participants were asked to consider the current condition of Hormosira and three future scenarios of reduced condition of Hormosira (Figure 1). Under each of these scenarios, they elicited participants’ estimates of the consequences of management alternatives on the management objectives. The SDM process is particularly useful at this stage, as the researchers can incorporate participants’ uncertainty in scientific knowledge, particularly when it comes to predicting the effectiveness of management alternatives under future scenarios.
Addison PFE, K de Bie & L Rumpff (2015). Setting conservation management thresholds using a novel participatory modelling approach. Conservation Biology. DOI: 10.1111/cobi.12544. And see the story on this paper in Decision Point #74
An annotated bibliography
A guide for anyone wanting to delve further into the field of decision analysis.
Risk analysis in environmental decisions
Burgman MA (2005). Risks and decisions for conservation and environmental management. Cambridge University Press, Cambridge, UK.
This is the definitive text for risk analysis in environmental decisions. Mark Burgman describes all the tools a decision maker needs to make decisions in the face of uncertainty, including methods for characterizing, eliciting, estimating, and describing uncertainty; methods for evaluating risk tolerance; and methods for managing risk.
Rapid prototyping for environmental decisions
Garrard GE, L Rumpff, MC Runge & SJ Converse ( 2017). Rapid prototyping for decision structuring: an efficient approach to conservation decision analysis. Pages 46-64 in Bunnefeld N, Nicholson E, Milner-Gulland EJ, eds. Decision-Making in Conservation and Natural Resource Management: Models for Interdisciplinary Approaches. Cambridge University Press.
Just as a car manufacturer designs a new model through prototypes, so, too, a decision maker can analyze a decision through a series of successive prototypes. The first prototype is simple, but captures the rough outline of the decision; subsequence prototypes add complexity as needed. Georgia Garrard and colleagues describe the process of rapid prototyping for environmental decisions.
Structured decision making
Gregory R, L Failing, M Harstone, G Long, T McDaniels & D Ohlson (2012). Structured decision making: a practical guide for environmental management choices. Wiley-Blackwell, West Sussex, UK.
Robin Gregory and colleagues have been facilitating difficult environmental decision processes for decades. In this text, they synthesize their insights, presenting an outline of decision analysis from an environmental management perspective. Their emphasis is on the foundational tools to achieve a proper framing of the decision problem.
The genesis of decision analysis
Howard RA (1988). Decision analysis: practice and promise. Management. Science 34:679-695.
In the early 1960s, Ron Howard and Howard Raiffa established the field of decision analysis, pulling together tools and insights from many other fields of study. In this paper, Ron Howard describes the genesis and tenets of the field and reflects on the accomplishments in its first quarter century.
The project prioritization protocol
Joseph LN, RF Maloney & HP Possingham (2009). Optimal allocation of resources among threatened species: a project prioritization protocol. Conservation Biology 23: 328-338.
Economists and mathematicians have studied the nature of allocation decisions for decades. The underlying structure of such decisions involves understanding the collective benefits from allocating resources to a portfolio of actions, and developing algorithms to find the optimal portfolio for investment. Liana Joseph and colleagues bring these insights into the realm of environmental decisions. The project prioritization protocol (PPP) is an easy-to-use algorithm that provides an approximately optimal solution for allocating resources.
How values are at the centre of all decisions
Keeney RL (1992). Value-focused thinking: A path to creative decision making. Harvard University Press, Cambridge, Massachusetts, USA.
In this seminal text, Ralph Keeney describes how values are at the centre of all decisions, and how embracing this fact can lead to better decisions. The recognition that there are often many fundamental objectives at play in decisions leads to the need for multi-criteria decision analysis. Although Keeney describes a broader set of decisions than just environmental decision, the tools and insights are particularly apt for the types of decisions environmental managers face.
Keith DA, TG Martin, E McDonald-Madden & C Walters (2011). Uncertainty and adaptive management for biodiversity conservation. Biological Conservation 144:1175-1178.
In 1986, Carl Walters wrote a seminal text on the adaptive management of natural resources. Since then, adaptive management has become a guiding principle, mantra, and buzzword throughout environmental management institutions. In this paper, David Keith and colleagues review the principles of adaptive management and reflect on the lessons learned in the quarter century since its first description.
A framework for the role of monitoring in environmental decisions
McDonald-Madden E, PW Baxter, RA Fuller, TG Martin, ET Game, J Montambault & HP Possingham (2010). Monitoring does not always count. Trends in Ecology & Evolution 25: 547-550.
In the 1990s and 2000s, the need for monitoring to accompany environmental management became a dogmatic prescription, but this prescription was not always founded in a full understanding of how the monitoring would affect the desired management outcomes. In this review, Eve McDonald-Madden and colleagues provide a framework for the role of monitoring in environmental decisions, along with a useful decision tree for evaluating when to implement monitoring.
Expert elicitation and value of information in adaptive management
Runge MC, SJ Converse & JE Lyons (2011). Which uncertainty? Using expert elicitation and expected value of information to design an adaptive program. Biological Conservation 144:1214-1223.
The value of information is a decision analytic tool to evaluate how much the expected outcomes of a decision could improve if uncertainty could be resolved first. In this paper, Michael Runge and colleagues describe the calculation of value of information, and how it can be coupled with a process of formal expert elicitation. They then show how it was used to evaluate management and research options for an iconic environmental decision.
Decision theory and game theory
von Neumann J & O Morgenstern (1944). Theory of Games and Economic Behaviour. Princeton University Press. Princeton, New Jersey, USA.
During World War II, an extraordinary amount of thought and research was dedicated to how competing actors make decisions. Out of this work was born both the fields of decision theory and game theory. In this seminal and still vibrant text, John von Neumann and Oskar Morgenstern describe the mathematical foundation of utility theory (to characterize the risk tolerance of a decision maker) and game theory (to characterize decisions made by competing decision makers). For a more recent (and delightfully engaging) introduction to game theory, see Len Fisher’s Rock, Paper, Scissors: Game Theory in Everyday Life (2008), Basic Books, NY.