Incorporating stakeholder input for assessing alternatives

Abstract

Large projects often assess the feasibility of multiple alternatives during the initiation phase. These assessments typically build a business case using financial measures to determine which alternative aligns best with the project objective(s) and organizational strategy. Organizations that are socially responsible engage stakeholders in the initiation phase of a project to solicit their feedback on the alternatives under investigation to create a shared understanding and to increase acceptance. However, the assessment is often criticized, as the decision-making is distant and lacks public support and stakeholders are left to wonder if their contributions were considered, and may feel that their concerns were not heard. Reasons for this include the lack of systematically integrating stakeholder knowledge to assess how project alternatives under discussion are perceived. Second, there is no systematic approach to investigate how stakeholder perceptions differ across stakeholder groups and in comparison to experts. Third, there is limited capability for investigating the tradeoffs, interdependencies, and indirect effects of project decisions from a holistic point of view that takes organizational objectives and stakeholders into account. This leaves decision makers at risk to underuse the stakeholder inputs that were gathered at great expense to make project decisions that have unexpected and unintended consequences that can harm the success of the project, and to miss out on opportunities to make improvements to proposed project alternatives during the planning stage to better fulfill stakeholder and organizational needs. To address these problems, this paper proposes a novel approach for bridging stakeholder engagement and decision making using Fuzzy Cognitive Mapping (FCM) modeling. The results from applying the FCM method for a large transmission upgrade project by Bonneville Power Administration (BPA), a U.S. federal agency are illustrated and discussed.

Keywords: assessment; decision support; fuzzy cognitive maps; social responsibility; stakeholder

Introduction

In the initiation phase of large and important projects, socially responsible organizations engage with stakeholders to determine the feasibility of project alternatives under investigation, to create a shared understanding of the project, and of associated stakeholder concerns, and to increase project acceptance (Project Management Institute, 2013). Moreover, stakeholder engagement is frequently mandated by law (Glasson, Therivel, & Chadwick, 1994) and recommended by voluntary corporate social responsibility standards (AccountAbility Institute, 2008; ISO, 2010). Accordingly, numerous methods exist for engaging with stakeholders and understanding their environmental and socio-economic concerns, such as soliciting feedback during public hearings and structured workshops, involving citizens on review boards, planning commissions, and advisory committees, or collaborating through joint planning sessions (AccountAbility Institute, 2005a, 2005b; Fisher, 2005; Genus, 2006; Glasson, et al., 1994; Schot & Rip, 1997). All of these methods provide solutions for identifying stakeholders, analyzing stakeholder interests, developing a plan for engagement, and for managing and controlling the engagement. However, these methods do not discuss how to systematically link these inputs to project decision making; instead, decision-makers are expected to become aware of and sensitive to stakeholder issues and concerns by reading summary reports, or by interacting with the organizational units that have engaged with the stakeholders (Fisher, 2005; Glasson, et al., 1994). As a result, decisionmaking is criticized as distant and lacking public support while stakeholders are left to wonder if their contributions were considered, and may feel that their concerns were not heard (Brooks & Harris, 2008; Dresner & Gilbert, 1999).

Current practices raise several problems for sponsors, project managers, and team members. First, it is difficult in systematically integrating stakeholder knowledge, to assess how project alternatives under discussion are perceived. Second, there is no systematic approach to investigate how stakeholder perceptions differ across stakeholder groups and in comparison to experts. Third, there is limited capability for investigating the tradeoffs, interdependencies, and indirect effects of project decisions from a holistic point of view that takes organizational objectives and stakeholders into account. This leaves decision makers at risk to underuse the stakeholder inputs that were gathered at great expense, to make project decisions that have unexpected and unintended consequences that can harm the success of the project., Decision makers therefore may miss out on opportunities to make improvements to proposed project alternatives during the planning stage to better fulfill stakeholder and organizational needs.

Project conflicts have been analyzed using cognitive mapping (Al-Tabtabai, 2001). Fuzzy Cognitive Mapping (FCM) makes cognitive maps computable (Kosko, 1986, 1988, 1993) and has the capability to capture inputs from stakeholder and expert knowledge sources, and to systematically integrate them to understand the effects of decision alternatives (Mouratiadou & Moran, 2007; Uygar, Ozesmi, 1999; Uygar, Ozesmi, & Ozesmi, 2003). However, current methods are limited to a relatively small dataset, and provide little guidance for capturing a wide range of stakeholder and expert input. They also focus on understanding diverse stakeholder views or on supporting decision making, not integrating the two.

This paper addresses these gaps by applying FCM modeling to the planning stages of a large transmission upgrade project by Bonneville Power Administration (BPA), a U.S. federal agency that provides electric energy in the Pacific Northwest. The research captured a wide range of stakeholder input using 70 documents and from experts using the draft and final EIS documents in 16 cognitive maps. It then translates the cognitive maps into FCM models to understand how the proposed eight project alternatives—different locations and transmission capacities—affect different groups of stakeholders, both from the stakeholder's and expert's perspectives. Finally, FCM simulation is used for analyzing each alternative with regard to impacts on stakeholders and organizational objectives.

Including this introduction, this paper is organized into six sections. The second section discusses the practical challenges with project planning processes when engaging stakeholders using existing methods. The third describes the theory and practice of FCM modeling. The case study background is then described, as well as the methods employed. The fifth section presents the results of applying FCM simulation during project planning for a large infrastructure project, and the conclusion shows the research limitations and proposes future research directions.

Stakeholder Engagement: Challenges and Practices

Engaging stakeholders in the initiation phase of a project creates a shared understanding, and it increases acceptance of the proposed solution and/or product alternatives (Project Management Institute, 2013). Stakeholders are people and organizations who have a stake in the project, because they have an ethically legitimate claim (Carroll, 1979; ISO, 2010; Wartick & Cochran, 1985) or because they are powerful and have a strategic interest in the organization operations, products, or services (Ackermann & Eden, 2011; Freeman, 2004; Harrison & Freeman, 1999; King, 2007; Mitchell, Agle, & Wood, 1997). Stakeholder engagement is commonly used in environmental impact assessment, technology assessment, and stakeholder management.

Environmental Impact Assessment (EIA) is a methodology for assessing ecological, social, and economical with proposed build alternatives (e.g., roads, bridges, etc.). The process encourages public stakeholder participation in its early phases using public hearings, joint planning sessions with advisory committees and stakeholders, delegated authorities conducting structured workshops that include citizens review boards, and planning commissions (Glasson, et al., 1994; Hildebrand & Cannon, 1993; Wilkins, 2003). However, long-standing criticisms are the boundaries, and are too narrow, there is missing information, assumptions are incorrect or too simple, and the assessment and decision making typically occur in different organizational units and as a result, the decision making is not transparent (Brooks & Harris, 2008; Dresner & Gilbert, 1999; Kvaerner, Swensen, & Erikstad, 2006).

Technology Assessment is another methodology that uses stakeholder engagement to analyze the ecological, social, and economical impacts of scientific technology research and development (Fisher, 2005; Guston, 2001; Schot & Rip, 1997). The assessment methodology includes internal and external reviews with experts and discursive elicitation methods with stakeholders such as, socio-technical mapping that combines stakeholder analysis and plotting of recent technical dynamics, anticipatory agenda building, and dialogue between innovators and the public using consensus conferences, citizen panels, and workshops (Guston & Sarewitz, 2002; Schot & Rip, 1997). The goal is to assure the assessment considers a wider range of stakeholder perspectives before making policy decisions to govern future developments and applications used by society (Schot & Rip, 1997; Van Eijndhoven, 1997). However, these assessments had little or no influence on policy decision (Hildebrand & Cannon, 1993; Schot & Rip, 1997; Van Eijndhoven, 1997) because the decision-making process was not integrated or aligned with the technology assessment process (Fisher, 2005; Genus, 2006; Guston, 1999).

Stakeholder management is a process for gaining an empathetic understanding of stakeholder issues (AccountAbility Institute, 2005a, 2005b; ISO, 2010). The intention is to understand stakeholder values, which are based on stakeholders' judgments and assumptions about a project's impact for their community and the environment (Wilkins, 2003). Stakeholder management commonly classifies stakeholders as primary and secondary. Primary stakeholders exchange resources with the organization, for example, as customers, suppliers, and employees. Secondary stakeholders are not directly involved in the organizational activities, but can still influence or affect the organization, for example, as consumer organizations, governmental agencies, nongovernmental agencies (NGO), and environmental groups (Clarkson, 1995). Primary and secondary stakeholders can collaborate. Stakeholder engagement requires understanding the interconnections among all the stakeholders, especially those stakeholders who have direct and indirect relationships (Ackermann & Eden, 2011).

A commonality of all stakeholder engagement approaches is the problem of who to include in the stakeholder analysis. Research shows that organizations tend to focus their strategies and decision making on the more powerful stakeholders, while ignoring the impacts on less visible and powerful stakeholders. This may occur as an oversight (Hart & Sharma, 2004) or as the result of pressure to complete the stakeholder study quickly and restrict the process not to consider the effects on downstream or secondary stakeholders (Clarkson, 1995; Wilkins, 2003). Such practices can create unintended consequences for stakeholders and for the project (Aaltonen, 2011). For example, even though Monsanto's seed sterilization technology was approved by governmental authorities, and Monsanto had engaged with salient stakeholders and had upheld all laws; it came under attack by consumer groups, retailers, and NGOs in Europe. Protests eventually spread to developing countries where small farmers in India protested (Hart & Sharma, 2004). Broadening stakeholder participation to include less salient stakeholders may improve the understanding of indirect and far-reaching impacts of alternatives; however, it comes at a cost. As the U.S. Office of Technology Assessment (OTA) and the European equivalent Constructive Technology Assessment (CTA) found out, broadening public participation can impede the exchange of values that are held closely by stakeholders (Genus, 2006). This is because experts tend to adopt an instrumental role when rationalizing their views, whereas the public tend to play a more cooperative role when it comes to rationality. The public also possesses the need to feel a positive self-worth when participating in the discourse (Dresner & Gilbert, 1999; Genus, 2006). Moreover, trying to unify experts and lay people on one particular issue runs the danger of closing the issue too early (Palm & Hansson, 2006). Methods for stakeholder engagement need to take these challenges into account and take particular care to adequately represent the views of stakeholders with little power and expert knowledge.

A second commonality of stakeholder engagement approaches is the need to make sense of the collected data. At the completion of stakeholder assessments, the findings are generally passed on to the decision makers who reside in a different organization, and are not directly involved in the assessment (Glasson, et al., 1994). Therefore, they are expected to become aware of stakeholder issues and concerns by reading summary reports and interacting with the organizational units that engaged with stakeholders and experts so that they may be sensitized to the long-term and far-reaching impacts of their actions. Consequently, the stakeholder management methods do not provide a systematic approach for capturing a wide range of stakeholder and expert inputs from multiple and disparate knowledge sources. Furthermore, stakeholder management methods do not integrate their input into the technology assessment and decision-making process. As a result, it makes it difficult for decision makers to understand the value or harm associated with the alternatives under investigation. The far-reaching and indirect effects of their decisions thereby, increase the risk of creating unintended consequences for the decisions made. Consequently, the methods did not show how new or changing stakeholder input would change their decision, which also contributed to unintended consequences with their decisions. There is limited ability for conducting tradeoff analysis on the project objectives, and the positive and negative impacts from these alternatives on stakeholders to understand the interdependencies among stakeholders values and how they affect the decisions they are about to make. As a result, there is a lack of transparency as to how stakeholder input influenced the decision, leaving stakeholders to wonder if their concerns were ever heard, and thus questioning the value of these stakeholder engagement activities.

Theoretical Foundations and Practicality of Using FCM

An evolving stream of literature proposes a novel methodology for capturing input from a wide range of stakeholders and experts and integrating the input into the assessment and decision making process using fuzzy cognitive mapping (FCM). Uygar Ozesmi (1999) interviewed stakeholders from four stakeholder groups to study the harvest of aquatic vegetation in the Kizilirmark Delta wetlands and how human practices are an integral part of the ecosystem. In another study, U. Ozesmi & S. L. Ozesmi (2003) interviewed stakeholders to develop a participatory ecosystem management plan for the Uluabat Lake in Turkey. They used FCM simulation to run “what-if” questions to assess how policy decisions affect the stakeholder groups using a combined social map of all stakeholders. Mouratiadou and Moran (2007) interviewed stakeholders from five stakeholder groups to understand the current state of and pressures on water resources to simulate the acceptance of different water management policies in the Pimos River Basin in Greece, and explore the potential effects on the water resources and the economy of the area. They too used FCM simulation to run “what-if” questions to assess how policy decisions affect the stakeholder groups using a combined social map of all stakeholders. Soler, Kok, Camara, and Veldkamp (2012) used data from literature and interviewed several experts to compare and contrast the determinants of land cover change in the Brazilian Amazon. Giordano, Passarella, Uricchio, & Vurro (2007; 2005) used FCM to identify quality demand issues in water management of the Candelaro River Basin in Italy with the purpose of creating an integrated decision support system.

FCM is a systems thinking approach to assess real world problems and the potential actions on all or portions of the problem by evaluating the actions in a microworld that virtually represents microcosms of the real world (Checkland, 2000; Salomon & Seegers, 1996; Schroder & Jetter, 2003; Senge & Sterman, 1992; Sterman, 2000; Voinov & Bousquet, 2010). FCM make qualitative causal cognitive maps, which had originated in social science, (Axelrod, 1976; Eden, 1988; Huff, 1990) computable to understand the dynamic behavior of the system they represent (Bart Kosko, 1986). Causal cognitive mapping is a technique to capture the mental models of experts and stakeholders (F. Ackermann, & Eden, C., 2005; Axelrod, 1976; Bryson, Ackermann, Eden, & Finn, 2004; K. Carley & Palmquist, 1992; Nakamura, Iwai, & Sawaragi, 1982).

The starting point of any FCM is a causal cognitive map, like the map depicted in Figure 1: Concepts (“nodes” or “circles") are linked through arrows that represent causality. Concepts are described verbally and can represent hard-to-quantify phenomena such as “construction disturbances,” “displacement of wildlife,” and “future use.” The arrows are denoted with “+” or “- “, depending on what type of causality exists. Positive arrows between two concepts (e.g. C₁ and C3) imply that an increase in one concept causes an increase in the other concept. Negative arrows (e.g. between C₂ and C₃) reflect a decrease of the second concept, when the first concept increases.

Figure 1: A causal cognitive map

FCMs are regarded as a simple form of recursive neural networks, with concepts being the equivalent to neurons. Other than neurons in a neural networks, concepts in FCMs are not either “on” (= 1) or “off"(= 0), but can take states in between and are therefore “fuzzy” (Tsadiras, 2008) . Fuzzy concepts are non-linear functions that transform the path-weighted activations directed towards them (their “causes") into a value in [0, 1] or [-1:1]. When a neuron “fires” (i.e., when a concept changes its state), it affects all concepts that are causally dependent upon it. Depending on the direction and size of this effect, and on the threshold levels of the dependent concepts, the affected concepts subsequently may change their state as well, thus activating further concepts within the network. FCMs allow feedback loops; therefore, it is possible that the newly activated concepts influence concepts that have already been activated before. This resulting activation spreads in a non-linear fashion through the FCM net until the system reaches a new stable state-fixed state vector called fixed-point attractor or cycles between a number of fixed state values called a limit cycle (B. Kosko, 1988).

FCM calculation models the spreading activation through the network by multiplying a state vector of causal activation with the square connection matrix derived from the FCM graph and by thresholding the result in accordance with the concept's squashing function, as shown in EQ(1). W is an n x n matrix of concepts that represent alternatives, causes, consequences, utilities, and objectives. The causal relationship between two concepts is the arc edge weights (positive or negative) W_ij. Each concept is represented by a number at an activation level for concept at time step t. An input vector A_i^t = [ A₁^t, A₂^t ... .. A_n^t] activates the concepts in W as described in EQ(1). The result is the summation of all arc edge weights (positive or negative) in Wij, where j is not equal to i because FCM does not allow directions between a concept and itself (Tsadiras, 2008). Furthermore, hyperbolic tangent sigmoid threshold squashing function f assures the result of summation belongs the interval [-1,0]. The activation level at step f results in a new state A_i^t+1 = [ A₁^t+1, A₂^t+1 ... .. A_n^t+1].

The following example will illustrate the activation sequence. If concept C₁ (highlighted in grey) in Figure 1 is activated, while all other concepts are turned off, the initial state vector is:

S¹ = [1 0 0 0]

It is then multiplied with the square connection matrix, which is equivalent to the signed digraph in Figure 1.

Matrix multiplication and the application of a threshold squashing function lead to a new state vector:

(In this particular example, a binary threshold squashing function converts inputs of < 0 to 0 and inputs of > 0 to 1 is used). The resulting new state vector is again multiplied with the connection matrix. The process is repeated until stability is reached (in this case after S4), or a stop criterion is met:

S² = [0 0 1 0]

S³ = [0 0 0 1]

S⁴ = [0 0 0 0]

S⁵ = [0 0 0 0]

The calculation is slightly different, if the activation of concept C₁ is not a one-time impulse (e.g. a natural disaster), but a change that lasts over extended periods of time (e.g. regulation laws). In this case, the concept is “clamped” and always set back to its initial activation level, as the following example, which already reaches a stable state after three cycles, will show:

S'¹ = [1 0 0 0]

S'² = [1 0 1 0]

S'³ = [1 0 1 1]

S'⁴ = [1 0 1 1]

S'⁵ = [1 0 1 1]

The system's behavior depends on the structure of the causal map, the input vector, and the choice of threshold squashing functions. Different threshold squashing functions can be applied to each concept, though typically FCMs are calculated with the same squashing function for all nodes (Tsadiras, 2008). Commonly used squashing functions, such as bivalent, trivalent, or logistic function restrict the concept states to discrete final states, such as [0; 1] or [-1; 0; 1] or to intervals [-1;1] or [0;1]. Bivalent squashing functions can represent an increase of a concept, trivalent functions can represent an increase or a decrease of a concept and logistic functions can represent the degree of an increase of decrease of a concept (Tsadiras, 2008).

All FCMs have “meta-rules.” Several input vectors—so-called input regions—lead to the same final system state. The meta-rules of a FCM can be identified experimentally through simulation (Dickerson & Kosko, 1994) and, if strict restrictions are met, analytically (Miao, Liu, Tao, Shen, & Li, 2002). FCMs with discrete concept states, so-called “finite state machines” result in either a fixed state vector or in a limit cycle between a number of fixed state vectors (Stach, Kurgan, Pedrycz, & Reformat, 2005). The stable fixed point attractor or limited cycle is typically reached in less than 30 cycles (Uygar, Ozesmi, & Ozesmi, 2004) and oftentimes much sooner (Stach, et al., 2005). “Continuous state machines"—FCM with squashing functions that output values on an interval—can result in chaotic behavior (Dickerson & Kosko, 1993; Stach, et al., 2005).

FCM demonstrates the systematic integration of stakeholder and expert input into assessment and decision-making processes (Jetter & Schweinfort, 2011; Jetter & Sperry, 2013; Mouratiadou & Moran, 2007; Uygar, Özesmi, & Özesmi, 2004; Schroder & Jetter, 2003), but they do not investigate how perceptions differ among and between stakeholder and expert groups. Consequently, there is limited guidance for using FCM to understand the far-reaching and indirect effects of decisions in order to avoid unintended consequences for stakeholders, while—if possible—also achieving organizational objectives.

Libby to Troy – A Case Study

The objective of this research is to develop a novel proactive methodology for evaluating technology alternatives using stakeholder and expert input, and integrating the assessment with the decision making process to make decisions that are sociably responsible, and align with the organizational objectives. This is achieved by capturing stakeholder and expert knowledge in cognitive maps, then translating the cognitive maps into FCM to simulate and analyze how stakeholders perceive the value or harm associated with the alternatives; which stakeholders share the same perceptions; and how strongly are their perceptions of value or harm and regard to what aspects. FCM simulation is also used to show how experts perceive the value or harm of the alternatives; how experts assess the value or harm associated with the alternatives on stakeholders; and how stakeholder value affects the project objectives.

The following section describes a case study on the use of the novel FCM method for assessment and decision support. The case study is a large transmission upgrade project by Bonneville Power Administration (BPA). BPA provides about a third of the electric power and seventy-five percent of the high-voltage electric transmission in the Pacific Northwest. In 2005, it needed to rebuild or reinforce a section of the transmission line from Flathead Electric Cooperative (FEC) substation near the town of Libby, Montana to a BPA substation near Troy, Montana. This section is an integral part of larger transmission loop, making it impossible to keep the existing line unchanged. The project had to comply with the environmental policies as set forth by the National Environmental Policy Act (NEPA). As a result, BPA had to prepare an environmental impact study (EIS) for the project. This occurred in two stages: in 2005 a notice of intent informed the public about the upcoming project and invited comments that were considered during the preparation of a draft EIS. In 2007, the draft EIS was again put out for public comments, which were considered in the final EIS prepared by BPA experts. Throughout the project, BPA's objectives were to minimize environmental impact, as well as project costs. BPA considered rebuilding the existing transmission line, using either a 115-kV single-circuit, similar to the one that is currently in place or a 230-kV double-circuit. Furthermore, BPA considered three realignments to the existing route: Pipe Creek, Quartz Creek, and Kootenai River. In all, there were eight alternatives considered. Over 300 stakeholders were identified and asked for comments in town hall meetings and through public notices. They could provide their comments in person or by mail, fax, and through a website. Stakeholders included residents/landowners, businesses, local, State and Federal government agencies, and tribal communities. All stakeholder comments are publically available documents. In all, there 58 scoping documents containing stakeholder comments and 22 additional stakeholder documents after the draft EIS was issued. They were analyzed for the purpose of this study. Moreover, the draft and the final EIS, which show BPA's organizational perspective on stakeholder issues, are publically available and were also used for this study.

The case study used an exploratory approach (K. Carley & Palmquist, 1992) to draw out the concepts from the publically available documented stakeholder comments, and the draft EIS and final EIS produced by BPA experts. Mapping from these secondary sources required identifying the cause and effect concepts, which are the subject or object in a statement that can take on different values, and the relationship (positive or negative) between the two concepts as indicated by the verb/adverb (K. Carley & Palmquist, 1992; K. M. Carley, 1997; Nakamura, et al., 1982; Roberts, 1989; Wrightson, 1966). For example, encoding of the text for “Construction disturbances could result in displacement and elevated stress levels of wildlife species in or near the construction area” is shown in Figure 1. The cause concept “construction disturbances” results in two effect concepts, “displacement of wildlife” and “elevated stress levels of wildlife,” both of which are negative. More often than not, concepts are not described using the same terminology; therefore, a common ontology of concepts was used to simplify the coding and to maintain continuity.

Figure 1: Encoding text

The final step in the process was developing stakeholder and the BPA expert causal cognitive models from using the comments they submitted to BPA. Models were developed at the stakeholder group level, not at the individual stakeholder level. Stakeholders were grouped according to their areas of interests and/or geographical location. During the scoping phase, six stakeholder groups voiced their concerns and needs: Pipe Creek Residents, Bighorn Terrace Residents, Residents at Large, State of Montana, and Tribal Communities. After the draft EIS was released, stakeholders were again allowed to voice their concerns. As a result, two new stakeholder groups identified City of Libby and U.S. Federal Government. In total, eight stakeholder groups were identified as depicted in Figure 2. The larger oval represents the entire project area and the smaller ovals represent specific areas of interest.

Figure 2: Stakeholder group's interests

Stakeholder group cognitive maps were developed from scoping comments. These cognitive maps were updated with new or changing concerns. This happened after the draft EIS was released. New cognitive maps were developed for stakeholder groups who did not voice their concerns during the scoping phase (City of Libby and U.S. Federal Government). Figure 3 provides an example of a cognitive map and shows the issues and concerns of Bighorn Terrace residents. On the left-hand side, it shows the eight alternatives: Existing 115kV, Existing 230kV, Pipe Creek 115kv, Pipe Creek 230kV, Quartz Creek 115kV, Quartz Creek 230kV, Kootenai River 115kV, and Kootenai River 230kV. All alternatives were structured as transmitter concepts (only out arrows). They are casually connected to concepts that are of concern to the stakeholders, such as construction activities and the granting of right-of-ways (ROW) for the transmission line. Blue arrows and positive signs show positive relationships, orange arrows and negative signs show negative causal relationships. Edge weights were assigned to be positive (+1) or a negative (-1) (B. Kosko, 1988) because the research could not infer any degree of influence from the stakeholder comments. The cognitive maps generated for other stakeholder groups follow the same overall structure.

Figure 3: Bighorn Terrace residential stakeholder cognitive map

Expert cognitive maps were developed from the draft EIS and final EIS. These documents provided a Likert scale of high, medium to high, medium, low to medium, and low to describe the harmful or beneficial effect to the environment/society in the draft and final EIS documents. These values were translated in edge weights using values 0.9, 0.7, 0.5, 0.3, 0.1, respectively. No impact is always a 0.

In all, 16 cognitive maps were derived from the comments made by the eight stakeholder groups and the draft and final EIS documents. They contained a total of 97 cause-and-effect concepts: 94 were mentioned in the scoping/draft EIS phases and an additional three concepts were identified in final EIS. In addition, there were seventeen impact concepts that represent the EIS areas under study. They include, for example, soil disturbance, cultural resources, and residential views. Finally, there were two concepts to represent objectives: minimize environmental impact and minimize cost, these were modeled as receiver concepts (only in arrows).

FCM Model Results: Assessment and Analysis

FCM modeling shows that some of the stakeholder interests are asymmetrical and conflict with one another, while others are symmetrical and coincide with one another. This presents a challenge when trying to determine those alternative(s) that are beneficial and those that are not. The left hand side of Table 1 depicts the FCM output for how stakeholders (see columns 2–6) perceive the value or harm associated with the eight alternatives (see column 1), ranging from a “-1” to “1”. A positive value is viewed as an improvement over the current situation (the existing, outdated transmission line) and negative values are viewed as a harmful impact. The lower the number the more harm. Conversely, the higher the number is, the less the harm. Not surprisingly, none of the stakeholders gain any benefits from the technical upgrade of the transmission line—for most, it goes hand-in-hand with some disturbances, such as construction noise. However, some stakeholders see no harm in some alternatives that others perceive as very harmful. For example, Pipe Creek (column 2) feels very strongly about the harmful impacts of the Pipe Creek alternative; however, they are indifferent to all other alternatives. On the other hand Bighorn Terrace (column 2), associate the Quartz realignment with only very little harm and even less so, if the 230kV option is chosen. Residents at large were concerned about the harmful impacts associated 230kV alternatives and were indifferent to any of the 115kV alternatives. Local businesses and Libby government were indifferent to all alternatives. State of Montana viewed Quartz Creek realignment as least impact, but felt strongly about the harmful impacts with the 230kV and Kootenai River alternatives. The federal government felt very strongly with harmful impacts from the Kootenai River alternative no matter what voltage and was indifferent to all other alternatives. Finally, the Tribal Communities viewed Kootenai River as no impact, but felt strongly about the harmful impacts associated with the Pipe Creek realignment.

The right hand side of Table 1 shows the same assessment but after the draft EIS was issued. It shows a shift in some opinions: For example, Pipe Creek Residents, Residents at Large and State Government of Montana viewed the impacts were greater after the Draft EIS was release. Bighorn Terrace Residents changed their opinion and were actually even positive about the 230kV Quartz Creek alternative.

Table 1: Stakeholder perceptions on impacts of alternatives

Table 2 depicts how the BPA experts viewed the impacts from the alternatives on stakeholders. Replacing the existing 115kV has the least impact on all stakeholders. The Pipe Creek alternative severely impacted Pipe Creek residents and the tribal communities. The Quartz Creek alternative is seen favorable for Bighorn Terrace, but does impact the tribal communities. Kootenai River realignment is seen as the least impact to the federal government and tribal communities, but impacts Pipe Creek, Bighorn, and residents at large. Replacing the existing 115kVv with 230kV severely impacts all stakeholders with the exception of Kootenai River on tribal communities.

Table 2: How the experts view impacts on stakeholders

It is clear the expert's view of impacts on the stakeholders differed from how the stakeholders viewed the impacts. These differences are reasons to suspect conflict and contention with meeting the project objectives. To understand the difference in perceptions, central concepts provide the insight into what concepts influence the downstream effects the most (Mouratiadou & Moran, 2007; Uygar, Özesmi, & Özesmi, 2004). Upon closer inspection, stakeholders viewed the right of way, the construction of new structures, and roads as the most central concepts. This makes sense given any alternative considered requires acquiring land to widen right of way and disturbance associated with removal and the construction new power line and access roads required for maintenance of the power lines. Furthermore, because 115kV uses the same right of way for the most part, it also makes sense why stakeholders primarily viewed 115kV alternatives over 203kV.

Limitations, Discussions, and Outlook

This paper addresses the gaps in current stakeholders and expert engagement practices, which fail to integrate stakeholder-based assessment and decision making. FCM was proposed as a method for capturing stakeholder and expert input in causal cognitive maps, integrate them, and translate them into FCM simulation models that are used to understand the impact of decision alternatives. To illustrate the method, the paper presents a case study of a transmission upgrade project. Based on the analysis of 80 stakeholder documents and the draft and the final version of the EIS, 16 FCM models were generated. They cover the perceptions of stakeholders and project experts at different times in the project, namely during scoping, after the draft EIS was issued and at the time of the final EIS. The study shows that the eight project alternatives are assessed differently by different stakeholders at different points in time and highlights areas of agreement and disagreement. This enables decision makers to understand the impacts of project decisions, identify areas for improvement to minimize harmful impacts, and anticipate potential conflicts during project execution. Moreover, the study shows were expert opinions about stakeholder impacts and stakeholder perceptions about future impacts are not aligned. Combined, these analyses enable decision makers to understand the direct and indirect positive and negative impacts of alternatives and their interdependencies in order to improve the decisions they are about to make.

A limitation of this research was the use of secondary data, rather than eliciting stakeholder knowledge directly from stakeholders. While these approaches are common in the literature (K. Carley & Palmquist, 1992; Nakamura, et al., 1982; Roberts, 1989), they place limitations on the validation of the stakeholder cognitive maps (Axelrod, 1976). However, provisions were made to for BPA personnel who participated on the project to validate the results. Another limitation is that the project team is not the actual decision maker at BPA; therefore, without understanding how a true decision maker gives importance to estimates or assumptions, the applicability to results of the analysis is limited (Baird, 1989). Furthermore, the research did not address coalition among multiple stakeholders would affect the analysis and—consequently—the decision making process. Moreover, the case study data—though captured in real-time as the project unfolded—was analyzed after the transmission project was completed, which introduces a potential for hindsight bias by BPA personnel. The researchers, however, deliberately chose to remain ignorant about project outcomes until their analysis was completed. Future research should address these limitations and apply the proposed FCM methodology to an ongoing project in real time. This will allow us to assess to what extent it improves project decision maker's understanding of stakeholder issues, triggers ideas for better—less harmful— alternatives, helps to anticipate stakeholder coalitions, and improves decision-making.

PMI endorses the concept of stakeholder management to understand stakeholder interests in relationship to the project objectives. The research contribution bridges stakeholder management theory with project management for integrating stakeholder and expert input into the assessment and decision making process on a large infrastructure project. It incorporates systems thinking into the stakeholder management project planning process to capture a wide range of stakeholder and expert inputs using disparate sources to analyze the effects of alternative actions on stakeholders and to conduct analysis among project objectives and impacts to stakeholders. It also accounted for new or changing stakeholder perceptions and incorporated them into the existing FCMs to assess change in behavior. Finally, it introduces transparency on how stakeholder and expert input influences the decision.