Multi-criteria Decision Analysis: a Framework for Structuring Remedial Decisions at Contaminated Sites

Decision-making in environmental projects is typically a complex and confusing exercise, characterized by trade-offs between socio-political, environmental, and economic impacts. Cost-benefit analyses are often used, occasionally in concert with comparative risk assessment, to choose between competing project alternatives. The selection of appropriate remedial and abatement policies for contaminated sites, landuse planning and other regulatory decision-making problems for contaminated sites involves multiple criteria such as cost, benefit, environmental impact, safety, and risk. Some of these criteria cannot easily be condensed into a monetary value, which complicates the integration problem inherent to making comparisons and trade-offs. Even if it were possible to convert criteria rankings into a common unit this approach would not always be desirable since stakeholder preferences may be lost in the process. Furthermore, environmental concerns often involve ethical and moral principles that may not be related to any economic use or value. Considerable research in the area of multi criteria decision analysis (MCDA) has made available practical methods for applying scientific decision theoretical approaches to multi-criteria problems. However, these methods have not been formalized into a framework readily applicable to environmental projects dealing with contaminated and disturbed sites where risk assessment and stakeholder participation are of crucial concern. This paper presents a review of available literature on the application of MCDA in environmental projects. Based on this review, the paper develops a decision analytic framework specifically tailored to deal with decision making at contaminated sites. in Linkov, I. and Ramadan, A. eds “Comparative Risk Assessment and Environmental Decision Making” Kluwer, 2004, p. 15-54 2 1. Current and Evolving Decision-Analysis Methodologies Environmental decisions are often complex, multi-faceted, and involve many different stakeholders with different priorities or objectives – presenting exactly the type of problem that behavioral decision research shows humans are typically quite bad at solving, unaided. Most people, when confronted with such a problem will attempt to use intuitive or heuristic approaches to simplify complexity until the problem seems more manageable. In the process, important information may be lost, opposing points of view may be discarded, elements of uncertainty may be ignored -in short, there are many reasons to expect that, on their own, individuals (either lay or expert) will often experience difficulty making informed, thoughtful choices about complex issues involving uncertainties and value tradeoffs (McDaniels et al., 1999). Moreover, environmental decisions typically draw upon multidisciplinary knowledge bases, incorporating natural, physical, and social sciences, medicine, politics, and ethics. This fact, and the tendency of environmental issues to involve shared resources and broad constituencies, means that group decision processes are called for. These may have some advantages over individual processes: more perspectives may be put forward for consideration, the chances of having natural systematic thinkers involved is higher, and groups may be able to rely upon the more deliberative, well-informed members. However, groups are also susceptible to the tendency to establish entrenched positions (defeating compromise initiatives) or to prematurely adopt a common perspective that excludes contrary information -a tendency termed “group think.” (McDaniels et al., 1999). For environmental management projects, decision makers may currently receive four types of technical input: modeling/monitoring, risk analysis, cost or costbenefit analysis, and stakeholders’ preferences (Figure 1a). However, current decision processes typically offer little guidance on how to integrate or judge the relative importance of information from each source. Also, information comes in different forms. While modeling and monitoring results are usually presented as quantitative estimates, risk assessment and cost-benefit analyses incorporate a higher degree of qualitative judgment by the project team. Only recently have environmental modeling (such as fate and transport models) and formalized risk assessment been coupled to present partially integrated analyses to the decision-maker (e.g., Army Risk Assessment Modeling project, ARAMS (Dortch, 2000). Structured information about stakeholder preferences may not be presented to the decision-maker at all, and may be handled in an ad hoc or subjective manner that exacerbates the difficulty of defending the decision process as reliable and fair. Moreover, where structured approaches are employed, they may be perceived as lacking the flexibility to adapt to localized concerns or faithfully represent minority viewpoints. A systematic methodology to combine these inputs with cost/benefit information and stakeholder views to rank project alternatives has not yet been developed. As a result, the decision maker may not be able to utilize all available and necessary information in choosing between identified remedial and abatement alternatives. In response to current decision-making challenges, this paper develops a systematic framework for synthesizing quantitative and qualitative information that builds on the recent efforts of several government agencies and individual scientists to in Linkov, I. and Ramadan, A. eds “Comparative Risk Assessment and Environmental Decision Making” Kluwer, 2004, p. 15-54 3 implement new concepts in decision analysis and operations research. This will help to both facilitate analysis and provide for more robust treatment of stakeholder concerns. The general trends in the field are reflected in Figure 1b. Decision analytical frameworks may be tailored to the needs of the individual decision maker or relate to multiple stakeholders. For individual decision-makers, risk-based decision analysis quantifies value judgments, scores different project alternatives on the criteria of interest, and facilitates selection of a preferred course of action. For group problems, the process of quantifying stakeholder preferences may be more intensive, often incorporating aspects of group decision-making. One of the advantages of an MCDA approach in group decisions is the capacity for calling attention to similarities or potential areas of conflict between stakeholders with different views, which results in a more complete understanding of the values held by others. In developing this framework, the paper will draw from existing literature on environmental applications of multi criteria decision theory and regulatory guidance developed by the US and international agencies. 2. MCDA Methods and Tools MCDA methods evolved as a response to the observed inability of people to effectively analyze multiple streams of dissimilar information. There are many different MCDA methods. They are based on different theoretical foundations such as optimization, goal aspiration, or outranking, or a combination of these: • Optimization models employ numerical scores to communicate the merit of one option in comparison to others on a single scale. Scores are developed from the performance of alternatives with respect to an individual criterion and then aggregated into an overall score. Individual scores may be simply added up or averaged, or a weighting mechanism can be used to favor some criteria more heavily than others. Typically, (but not always, depending upon the sophistication of the objective function) good performance on some criteria can compensate for poor performance on others. Normalizing to an appropriate single scale may be problematic. Consequently, optimization models are best applied when objectives are narrow, clearly defined, and easily measured and aggregated. Considerable research and methods development has been done on multiobjective optimization. This work has mostly involved finding the “Pareto frontier”, along which no further improvements can be made in any of the objectives without making at least one of the other objectives worse (Diwekar and Small, 2002). • Goal aspiration, reference level, or threshold models rely on establishing desirable or satisfactory levels of achievement for each criterion. These processes seek to discover options that are closest to achieving, but not always surpassing, these goals. When it is impossible to achieve all stated goals, a goal model can be cast in the form of an optimization problem in which the decision maker attempts to minimize the shortfalls, ignoring exceedances. To this extent, overperformance on one criterion may not compensate for underperformance on in Linkov, I. and Ramadan, A. eds “Comparative Risk Assessment and Environmental Decision Making” Kluwer, 2004, p. 15-54 4 Individual Decision Maker