Gregory S. Parnell
Department of Industrial Engineering, University of Arkansas, Fayetteville, AR, USA
William D. Miller
Innovative Decisions, Inc., Vienna, VA, USA; School of Systems and Enterprises, Stevens Institute of Technology, Hoboken, NJ, USA
* This chapter draws on material from Chapters 3 and 7 of Parnell, G. S., Bresnick, T. A., Tani, S.N., and Johnson, E. R., Handbook of Decision Analysis, Wiley Operations Research/Management Science Handbook Series, Wiley & Sons, 2013.
If you don't know where you are going, any road will get you there.
(Lewis Carroll)
The decision opportunity and our values determine the decision objectives. The objectives define the goals that we are trying to achieve. In systems engineering, the objective space includes business/mission objectives, stakeholder objectives, and system objectives. Business/mission objectives are derived from organizational and customer needs. Stakeholder objectives include the goals of other important stakeholders in the system life cycle. Finally, system objectives include the technical objectives necessary for the system to meet business/mission and stakeholder objectives throughout the system life cycle. The systems engineering objectives space spans the life cycle of the system's products and services in both commercial and government enterprises including the business/mission need, concept, requirements, architecture, design, integration, verification, validation, production, deployment, operations, support, and retirement phases of the life cycle.
Defining the value of the system's products and services is one of the most critical tasks of systems engineering. In the previous chapter, we discuss the critical role of understanding the decision opportunity or problem. With this understanding, which may include a partial list of objectives, we next strive to identify the full list of objectives for the system. Of course, the identification of a significant new objective can change our understanding of the opportunity. In addition to improving the opportunity space, systems engineers use objectives as the foundation for developing value measure(s) used to evaluate alternatives and select the best alternative(s).
We believe it is good practice in any trade-off study to consider values and objectives before determining the full set of alternatives. Our experience is that studies focusing on alternative development first, with measures used only to distinguish between those alternatives, often miss several important objectives. This results in a missed opportunity to create a broader and more comprehensive set of alternatives with the potential to create more value.
To identify objectives, it is generally not sufficient to interact with only the decision-maker(s) because, for complex decisions, they may not have a complete and well-articulated list of objectives. Instead, significant effort may be required to identify, define, and perhaps even carefully craft the objectives based on many interactions with multiple decision-makers, diverse stakeholders, and recognized subject matter experts.
The chapter is organized as follows. In Section 7.2, we introduce the concept of Value-Focused Thinking. In Section 7.3, we describe shareholder and stakeholder value as the basis for the decision objectives. In Section 7.4, we describe why the identification of objectives is challenging. In Section 7.5, we list some key questions we can use to identify decision objectives and four major techniques for identifying decision objectives: research, interviews, focus groups, and surveys. In Section 7.6, we discuss key considerations for the financial objective of private companies and cost objectives of public organizations. In Section 7.7, we discuss key principles for developing value measures to measure, at the time of the decision, how well an alternative could achieve an objective. In Section 7.8, we describe the structuring of multiple objectives including objective and functional value hierarchies; four techniques for structuring objectives through the use of platinum, gold, silver, and combined standards; best practices; and provide cautions about risk and cost objectives. In Section 7.9, we describe the diverse approaches used to craft objectives for two illustrative problems. We conclude with a summary of the chapter in Section 7.10.
An important philosophical approach to creating value was introduced by Ralph Keeney in his book entitled “Value-Focused Thinking” (VFT) (Keeney, 1992). To be successful, VFT must interactively involve decision-maker(s), stakeholders, subject matter experts, and analysts to create higher shareholder and/or stakeholder value.
VFT has four major ideas: start first with our values, use our values to generate better alternatives, create decision opportunities, and use our values to evaluate the alternatives.
VFT starts with values and objectives before identifying alternatives. Keeney describes the contrasting approach as Alternative-Focused Thinking (AFT), which starts with the alternatives and seeks to differentiate them. He lists three disadvantages of AFT. First, if we start with known alternatives, this will limit our decision frame (our understanding of the opportunity or problem, see Chapter 6). Second, if we try to evaluate only known alternatives, we may not identify important new values and objectives that are relevant for future solutions. Finally, since we may not have a full understanding of the opportunity and the decision frame, we may not have a sound basis for the generation of alternatives.
Our decision can only be as good as the best alternative that we identify. If we have several poor alternatives, the best analysis will only identify a poor alternative! Once we have identified the values and objectives for the decision problem, we can use them to generate better alternatives. We do this by qualitatively and quantitatively defining the value gaps. See Chapter 8.
Many of us have been taught to look for potential problems, carefully define the problems, and then look for solutions. Whereas problem definition is reactive, opportunity definition is proactive. Keeney encourages us not to wait for problems to appear, but instead to focus our energy and creativity on identifying decision opportunities that may result in added value for an organization's shareholders and stakeholders. See Chapter 6.
Finally, we should use our values to qualitatively and quantitatively evaluate our alternatives. The mathematics of single and multiple objective decision analysis can be used to evaluate the alternatives, and once completed, we can use this information to improve existing alternatives or develop new alternatives. See Chapter 8.
Keeney has identified several benefits of VFT:
Decision objectives should be based on shareholder and stakeholder value. Value can be created at multiple levels within an organization. An organization creates value for shareholders and stakeholders by performing its mission (which defines the customers); providing products and services to customers; improving the effectiveness of its products and services to provide better value for customers; and improving the efficiency of its operations to reduce the resources required to provide the value to customers. Defining value is challenging for both private and public organizations. Private companies must balance shareholder value with being good corporate citizens (sometimes called stakeholder value). Management literature includes discussion of stakeholder value versus shareholder value for private companies (Charreaux & Desbrieres, 2001). Public organizations, which operate without profit incentives, focus on stakeholder value, with management and employees being key stakeholders.
Next, we consider shareholder and stakeholder objectives in a private company example and stakeholder objectives in a public organization.
Consider a large publicly owned communications company operating a cellular network that provides voice and data services for customers. Stakeholders include the shareholders, the board of directors, the leadership team, the employees (managers, designers, developers, operators, maintainers, and business process personnel), the customers, the cell phone manufacturers, the companies that sell the cell phone services, and the communities in which the company operates. Competition includes other cellular communications companies and companies that provide similar products and services using other technologies (satellite, cable, etc.).
Since the company is publicly owned, the board of directors' primary objective is to increase shareholder value. Stakeholders have many complementary and conflicting objectives. For example, the board may want to increase revenues and profits; the leadership may want to increase executive compensation; the sales department may want to increase the number of subscribers; the network operators want to increase availability and reduce dropped calls; the safety office may want to decrease accidents; the technologists may want to develop and deploy the latest generation of network communications; the cell phone manufacturers may want to sell improved technology cell phones; companies that sell the cell phones and services may want to increase their profit margins; operations managers want to reduce the cost of operations; the human resources department may want to increase diversity; and the employees may want to increase their pay and benefits.
Next, consider a government agency that operates a large military communications network involving many organizations. There are no shareholders in this public system that provides communications to support military operations. However, similarly to the private company example, there are many stakeholders. Some of the key stakeholders are the Department of Defense (DoD) office that establishes the communications requirements and submits the budget; Congress that approves the annual budgets for the network; the agency that acquires the network; the agency that manages the network; contractor personnel who manufacture and assemble the network; contractor, civilian, and/or military personnel who operate the network; information assurance personnel who maintain the security of the network; mission commanders who need the network to command and control their forces; and military personnel whose lives may depend on the availability of the network during a conflict. Instead of business competitors, the network operators face determined adversaries who would like to penetrate the network to gain intelligence data in peacetime or to disrupt the network during a conflict.
The stakeholders have many complementary and conflicting objectives. For example, the DoD network management office wants the best network for the budget; Congress wants an affordable communications network to support national security; the acquiring agency wants to deliver a network that meets the requirements, on time and on budget; the network management agency wants to ensure an adequate budget; contractors want to maximize their profits and obtain future work; network operators want to increase network capabilities and maximize availability of the network; information assurance personnel want to maximize network security; mission commanders want to maximize the probability of mission success; and military personnel who use the network want to maximize availability and bandwidth.
The identification of objectives is more art than science. The four major challenges are (i) identifying a full set of values and objectives, (ii) obtaining access to key decision-makers and stakeholders, (iii) differentiating fundamental and means objectives, and (iv) structuring a comprehensive set of fundamental objectives for validation by the decision-maker(s) and stakeholders.
In a complex decision, especially if it is a new opportunity for the organization, the identification of objectives can be challenging. In a research paper (Bond et al., 2008), the authors concluded that “in three empirical studies, participants consistently omitted nearly half of the objectives that they later identified as personally important. More surprisingly, omitted objectives were as important as the objectives generated by the participants on their own. These empirical results were replicated in a real-world case study of decision-making at a high-tech firm. Decision-makers are considerably deficient in utilizing personal knowledge and values to form objectives for the decisions they face.” To meet this challenge, we must have good techniques to obtain the decision objectives.
A second challenge is obtaining access to a diverse set of decision-makers (DMs), stakeholders (SHs), and subject matter experts (SMEs). Sometimes, clients are reluctant to provide the trade-off analysis team access to senior decision-makers and diverse stakeholders who have the responsibilities and breadth of experience essential to providing a full understanding of decision objectives. In addition, it can be difficult to obtain access to recognized experts instead of individuals who have more limited experience. Many times, the best experts resist meetings that take their focus away from their primary responsibilities. In addition, even if we have access, we may not have the time in our analysis schedule to access all the key individuals. To be successful, the analysis team must obtain access to as many of these key individuals as possible in the time allocated for the study.
The third challenge is the differentiation of fundamental and means objectives (Keeney, 1992). Fundamental objectives are what we ultimately care about in the decision. Means objectives describe how we achieve our fundamental objectives. An automobile safety example helps to clarify the difference. The fundamental objectives may be to reduce the number of casualties due to highway accidents and to minimize cost. The means objectives may include to increase safety features in the automobile (e.g., airbags and seat belts), to improve automobile performance in adverse weather (e.g., antilock brakes), and to reduce the number of alcohol-impaired drivers (e.g., stricter enforcement). The mathematical considerations of multiple objective decision analysis discussed in Chapter 2 require the use of fundamental objectives in the value model.
The fourth challenge is structuring the knowledge about fundamental objectives and value measures. This structure should enable decision-makers, stakeholders, and experts to validate that the set of objectives and value measures are both necessary and sufficient to evaluate the alternatives.
The key to identifying decision objectives is asking the right questions, to the right people, in the right setting. Keeney identified 10 categories of questions that can be asked to help identify decision objectives (Keeney, 1994). These questions should be tailored to the problem and to the individual being interviewed, the group being facilitated, or the survey being designed.1 For example, the strategic objectives question may be posed to the senior decision-maker in an interview while the consequences question may be posed to key stakeholders in a facilitated group.
Four techniques to obtain answers to these questions and help identify objectives and value measures are research, interviews, focus groups, and surveys. See Chapter 6 for more details on interviews, focus groups, and surveys. We discuss the use of these techniques to identify functions, objectives, and value measures. The amount of research, the number of interviews, the number and size of focus groups, and the number of surveys we use depend on the scope and importance of the problem, the number of decision levels, the diversity of the stakeholders, the number of experts, and the time allocated to defining objectives and identifying value measures. Best practices for these techniques are described in (Parnell et al., 2011) and (Parnell et al., 2013).
Research is an important technique to understand the problem domain; to identify potential objectives to discuss with decision-makers, stakeholders, and experts; and to understand suggested objectives. See Chapter 6 for discussion on knowledge. The amount of research depends on the analysts' prior understanding of the problem domain, knowledge of key terminology, and amount of domain knowledge expected of the analyst in the decision process. The primary research sources include the problem domain (including work done on functional and requirements analysis) and the analysis literature. Research should be done throughout the objective identification process. Many times, information obtained with the other three techniques requires research to fully understand or validate the objective recommendation.
Senior leaders, stakeholders, and “world class” experts can identify important value and objectives. Interviews are the best technique for obtaining objectives from senior decision-maker(s), senior stakeholders, and “world class” experts since they typically do not have the time to attend a longer focus group nor the interest to complete a survey. However, interviews are time-consuming for the interviewer due to preparation, execution, and analysis.
Focus groups are another useful technique for identifying decision objectives. We usually think of focus groups for problem/opportunity identification and product market research; however, they can also be useful for identifying decision objectives. While interviews typically generate a two-way flow of information, focus groups create information through discussion and interaction between all the group members. As a general rule, focus groups should comprise between 6 and 12 individuals. Too few may lead to too narrow a perspective, while too many may not provide all attendees the opportunity to provide meaningful input. It is very important to have a good facilitator to keep the focus group on track and to make sure that a few members do not dominate the discussion.
In our experience, surveys are not used as frequently to identify potential decision objectives as interviews and focus groups. However, surveys are a useful technique for collecting decision objectives from a large group of individuals in different locations. Surveys are especially good for obtaining general public values. Surveys are more appropriate for junior- to mid-level stakeholders and dispersed experts. We can use surveys to gather qualitative and quantitative data on the decision objectives. A great deal of research exists on techniques and best practices for designing effective surveys.
The financial or cost objective may be the only objective, or it may be one of the multiple objectives. Shareholder value is an important objective in any firm. In many decisions of private companies, the financial objective may be the only objective or the primary objective. Firms employ three fundamental financial statements to track value: balance sheet, income statement, and cash flow statement. In addition, firms commonly use discounted cash flow (net present value) to analyze the potential financial benefits of their alternatives. For public company decisions, the cost objective is usually a major consideration. The cost can be the full life cycle cost or only a portion of the costs of the alternative.
In order to understand the financial objectives for private companies, we begin with the three financial statements. Next, we consider the conversion of cash flows to net present value.
A balance sheet is developed using standard accounting procedures to report an approximation of the value of a firm (called the net book value) at a point in time. The approximation comes from adding up the assets and then subtracting out the liabilities. We must bear in mind that the valuation in a balance sheet is calculated using generally accepted accounting principles, which make it reproducible and verifiable, but it will differ from the valuation we would develop taking account of the future prospects of the firm.
The income statement describes the changes to the net book value through time. As such, it shares the strengths (use of generally accepted accounting procedures and widespread acceptance) and weaknesses (inability to address future prospects) of the balance sheet.
A cash flow statement describes the changes in the net cash position through time. One of the value measures reported in a cash flow statement is “free cash flow” (which considers investment and operating costs and revenues, but not financial actions such as issuing or retiring debt or equity). Many clients are comfortable with this viewpoint; hence, a projected cash flow statement is the cornerstone of many financial analyses.
To boil a cash flow time pattern down to a one-dimensional value measure, companies usually discount the free cash flow to a “present value” using a discount rate. The result is called the net present value, or NPV, of the cash flow. We usually contemplate various possible alternatives, each with its own cash flow stream and NPV cash flow. In many decision situations, clients are comfortable with using this as their fundamental value measure. See Chapters 4 and 14.
For many public organizations, minimizing the cost is a major decision objective. Depending on the decision, different cost objectives may be appropriate. The most general cost objective (subject to acceptable performance levels) is to minimize life cycle cost, the full cost over all stages of the life cycle: concept development, design, production, operations, and retirement. However, costs that have already been spent (sunk costs) should never be included in the analysis. Since sunk costs should not be considered and some costs may be approximately the same across alternatives, in practice, many analysts consider the delta life cycle costs among the alternatives. In public organizations, the budget is specified by year and may or may not be fungible between years. When multiple years are analyzed, a government inflation rate is used to calculate net present cost.
See Chapter 4 for further discussion on cost analysis and affordability analysis.
In order to quantitatively use the decision objectives in the evaluation of the alternatives, we must develop value measures for each objective that measure the a priori potential value, that is, before the alternative is selected. The identification of the value measures can be as challenging as the identification of the decision objectives. We can identify value measures by research, interviews, and group meetings with decision-makers, stakeholders, and subject matter experts. Access to stakeholders and experts with detailed knowledge of the problem domain is key to developing good value measures.
Kirkwood (1997) identifies two useful dimensions for value measures: alignment with the objective and type of measure. Alignment with the objective can be direct or proxy. A direct measure focuses on attaining the full objective, such as net present value for shareholder value. A proxy measure focuses on attaining an associated objective that is only partially related to the objective (e.g., reduce production costs for shareholder value). The type of measure can be natural or constructed. A natural measure is in general use and commonly interpreted, such as dollars. We have to develop a constructed measure, such as a five-star scale for automobile safety. Constructed measures are very useful but require careful definition of the measurement scales to ensure that we have an interval or ratio scale (see Chapter 2). In our view, the use of an undefined scale, for example, 1–7, is not appropriate for trade-off analysis since the measures do not define value and scoring is not repeatable.
Table 7.1 reflects our preferences for types of value measures. Our first preference is to select a value measure that directly aligns with the objective and has a natural scale. Our second preference is direct and constructed. We prefer direct and constructed to proxy and natural for two reasons. First, alignment with the objective is more important than the type of scale. Second, one direct and constructed measure can replace many natural and proxy measures. When value models grow too large, the source is usually the overuse of natural and proxy measures.
Table 7.1 Preference for Types of Value Measure
Type of Scale | Direct Alignment | Proxy Alignment |
Natural | 1 | 3 |
Constructed | 2 | 4 |
Not all problems have only the financial or the cost objective. In many public and business decisions, there are multiple objectives and many value measures. Once we have a list of the preliminary objectives and value measures, our next step is to organize the objectives and value measures (typically called structuring) to remove overlaps and identify gaps. For complex decisions, this can be quite challenging. In this section, we introduce the techniques for identifying and structuring, using hierarchies. The decision analysis literature uses several names: value hierarchies, objectives hierarchies, value trees, objective trees, functional value hierarchy, and qualitative value model.
The primary purpose of the objectives hierarchy is to identify the objectives and the value measures so we understand what is important in the problem and can do a much better job of qualitatively and quantitatively evaluating the alternatives. This may be the result of an enterprise-level business case analysis. Most decision analysis books recommend beginning with identifying the objectives and using the objectives to develop the value measures. For complex systems, decisions we have found that it is very useful to first identify the system functions that create value (Parnell et al., 2011). Functional analysis, central to both enterprise business processes and systems engineering, is the starting point for modeling the functional value hierarchy. For each function, we identify the fundamental objectives we want to achieve for that function. For each objective, we identify the value measures that can be used to assess the potential to achieve the objectives. In each application, we use the client's preferred terminology. For example, functions can be called missions, capabilities, activities, services, tasks, or other terms appropriate to the level of the decision analysis. Similarly, objectives can be called criteria, evaluation considerations, or other terms. Value measures can be called any of the previously mentioned terms (see Chapter 2).
The terms objectives hierarchy and functional value hierarchy are used in this book to make a distinction between the two approaches. The functional value hierarchy is a combination of the functional hierarchy from systems engineering and the value hierarchy from decision analysis (Parnell et al., 2011). In decisions where the functions of the alternatives are the same, or are not relevant, it may be useful to group the objectives by categories to help in structuring the objectives.
Both hierarchies begin with a statement of the primary decision objective as the first node in the hierarchy. An objectives hierarchy begins with the objectives in the first tier of the hierarchy, (sometimes) subobjectives as the second tier, and value measures as the final tier of the hierarchy. A functional value hierarchy uses functions as the first tier, (sometimes) subfunctions as the second tier, objectives as the next tier, and values measures as the final tier of the hierarchy. The value hierarchy integrates the enterprise business case analysis objectives hierarchy with the business process or systems engineering functional analysis.
Consider Figure 7.1. When the randomly ordered objectives hierarchy is logically organized by functions, the objectives and measures make more sense to the decision-makers and, many times, we identify the missing objectives (and value measures). The benefit of identifying the functions is threefold. First, a logical structure of the functional objectives hierarchy is easier for the decision analyst to develop. Second, it may help identify additional objectives (and value measures) that might be missed in the objectives hierarchy. Third, the logical order helps the decision-makers and stakeholders understand the hierarchy and provide suggestions for improvement.
In either the objectives hierarchy or functional hierarchy, we can create the structure from the top-down or from the bottom-up. Top-down structuring starts with listing the fundamental objectives on top and then “decomposing” into subobjectives until we are at a point where value measures can be defined. Top-down structuring has the advantage of being more closely focused on the fundamental objectives, but often, we initially overlook important subobjectives. Bottom-up structuring starts by discussing at the subobjective or subfunctional level, grouping similar things together, and defining the titles at a higher level for the grouped categories. Value measures are then added at the bottom of the hierarchy. Bottom-up structuring has the advantage of discussing issues at a more concrete and understandable level, but if we are not careful, we may drift from the fundamental objectives in an attempt to be comprehensive. In theory, both approaches should produce the same hierarchy. In practice, this rarely happens.
The credibility of the qualitative value model is very important since it is the basis of multiple decision-maker and stakeholder reviews. If decision-makers do not accept the qualitative value model, they will not (and should not!) accept the quantitative analysis. We discuss here four techniques for developing objectives: the platinum, gold, silver, and combined standards (Parnell et al., 2013).
A platinum standard value model is based primarily on information from interviews with senior decision-makers and key stakeholders. Decision analysts should always strive to interview the senior leaders (decision-makers and stakeholders) who make and influence the decisions. As preparation for these interviews, they should research potential key problem domain documents and talk to decision-maker and stakeholder representatives. Affinity diagrams (Parnell et al., 2011) can be used to group similar functions and objectives into logical, mutually exclusive, and collectively exhaustive categories. For example, interviews with senior decision-makers and stakeholders were used to develop a value model for the Army's 2005 Base Realignment and Closure value model (Ewing et al., 2006).
When we cannot get direct access to senior decision-makers and stakeholders, we look for other approaches. One approach is to use a “gold standard” document approved by senior decision-makers. A gold standard value model is developed based on an approved policy, strategy, or planning document. Many military acquisition programs use capability documents as a gold standard since the documents define system missions, functions, and key performance parameters (Parnell et al., 2001). A systems engineering functional analysis serves as a “gold standard” artifact in the development of systems products and services. Many times, the gold standard document has many of the functions, objectives, and some of the value measures. If the value measures are missing, we work with stakeholder representatives to identify appropriate value measures for each objective. It is important to remember that changes in the environment and leadership may cause a gold standard document to no longer reflect leadership values. Before using a gold standard document, confirm that the document still reflects leadership values.
Sometimes, the gold standard documents are not adequate (not current or not complete) and we are not able to interview a significant number of senior decision-makers and key stakeholders. As an alternative, the silver standard value model uses data from the many stakeholder representatives. Again, we use affinity diagrams to group the functions and objectives into mutually exclusive and collectively exhaustive categories. For example, inputs from about 200 stakeholders' representatives were used to develop the Air Force 2025 value model (Parnell et al., 1998). This technique has the advantage of developing new functions and objectives that are not included in the existing gold standard documents. For example, at the time of the study, the Air Force Vision was Global Reach, Global Power. The Air Force 2025 value model identified the function, Global Awareness (later changed to Global Vigilance), which was subsequently added to the new Air Force Vision of Global Vigilance, Reach, and Power.
Since it is sometimes difficult to obtain access to interview senior leaders, and many times, key documents are not sufficient to completely specify a value model, the most common technique is the combined standard. First, we research the key gold standard documents. Second, we conduct as many interviews with senior leaders as we can. Third, we meet with stakeholder representatives, in groups or individually, to obtain additional perspectives. Finally, we combine the results of our review of several documents with findings from interviews with some senior decision-makers and key stakeholders, and data from multiple meetings with stakeholder representatives. This technique was used to develop a space technology value model for the Air Force Research Laboratory Space Technology R&D Portfolio (Parnell et al., 2004).
The following are some recommended best practices for developing value hierarchies for systems engineering trade-off analyses.
Use the language of the decision-maker and key stakeholders. A clear problem statement is a very important tool to communicate the purpose of the system to the decision-maker(s), senior stakeholders, and the trade-off study team.
Select terms (e.g., functions, objectives, and value measures) used in the problem domain and from functional analysis in the engineering domain. This improves understanding by the users of the model.
Functional analysis encompasses structure, behavior, and performance of products and services. Structure includes the functional decomposition. Behavioral includes the control flow and the functional interfaces, that is, data, material, and energy flows. Performance includes the timeline analysis as well as throughputs and latencies.
Avoid buzzwords. This improves the understanding of the function or objective. Data, material, and energy are defined with nouns. Physical entities are also defined as nouns.
This provides a framework for helping decision-makers and stakeholders understand the value hierarchy and identify missing functions. In the architecture phase, functions are initially allocated to the generic physical entities of the system or service. (Software is considered a physical entity.) The instantiated physical architecture results from decisions as to the selection of specific physical entities and their spatial distribution.
Fundamental objectives are about why, what, when, where, and how well. These go in the functional/value hierarchy. Means objectives are about how. Means objectives are related to the alternatives.
Use value measures that are direct measures of the objectives and not proxy measures. If no natural measure exists, consider decomposing the objective or developing a constructed scale for attainment of the objective. Proxy measures result in more measures and increased data collection for measures that are only partially related to the objectives.
The hierarchy needs to include business/mission objectives, stakeholders' objectives, and system technical objectives. Developing a good functional value hierarchy for a major system development is very difficult.
Two commonly used objectives require special consideration: cost and risk objectives.
Mathematically, minimizing cost can be one of the objectives in the value hierarchy and cost can be a value measure in the value model. However, for many multiple objective trade-off studies, it is useful to treat cost separately and show the amount of value per unit cost. In our experience, this is the approach that is the most useful for decision-makers who have a budget that they might be able to increase or might have to accept a decrease.
Risk is a common decision-making concern, and it is tempting to add minimization of risk to the set of objectives in the value hierarchy. However, this is not a sound practice. A common example is helpful to explain why we do not recommend this approach. Suppose there are three objectives: maximize performance, minimize cost, and minimize time to complete the schedule. It may be tempting to add minimize risk as a fourth objective, but what type of risk are we minimizing and what is causing the risk? The risk could be performance risk; cost risk; schedule risk, performance and cost risk; cost and schedule risk; performance and schedule risk; or performance, cost, and schedule risk. In addition, there could be one or more uncertainties that drive these risks. In Chapter 8, we use probabilistic modeling to model the sources of risk and their impact on the value measures and the objectives. We believe this is a much more sound approach than the use of a vague risk objective in the value hierarchy. Furthermore, if utility (see Chapter 2) is used, we can model the risk preference directly.
In this section, we provide military and homeland security examples of value models for trade-off analyses.
The following functional value hierarchy was developed for an illustrative trade-off analysis study. The example is also used in Chapter 9. The Army is interested in the Squad of the future. The purpose of the study is to have the future Army squad achieve overmatch against enemies in complex environments. Five functions are identified and presented in a logical time sequence in Figure 7.2. Two subfunctions are identified for the Command and Control Squad.
Figure 7.2 also identifies the objectives for each function or subfunction if the function is divided into subfunctions. For each objective, at least one value measure is identified. Two types of value measures are used: direct natural measures on an interval scale (e.g., speed in mph) and constructed interval scales (e.g., secured connectivity).
The Domestic Nuclear Detection Office (DNDO) of the Department of Homeland Security was created to increase the United States' ability to detect radiological and nuclear (RN) material that could be obtained and then used by terrorists. The office coordinates the Global Nuclear Detection Architecture (GNDA), an international and interagency strategy for detecting, analyzing, and reporting RN materials outside of regulatory control. In 2012, the Government Accountability Office expressed concern about the prioritization of GNDA resources as well as the documentation of GNDA improvements over time. As a result, the DNDO asked the National Research Council (NRC) for advice on how to develop performance measures and metrics to quantitatively assess the GNDA's effectiveness. The result of the NRC study was a report titled “Performance Metrics for the Global Nuclear Detection Architecture.” In the report, the committee created a notional strategic planning framework for evaluating the performance of the GNDA. Using the data from the public report, multiobjective decision analysis techniques, and notional data from research, the NRC framework was expanded to a complete value model. The value model was used to demonstrate that it is possible to evaluate the potential performance of the GNDA over time and to evaluate the cost-effectiveness of potential improvements (Hilliard et al., 2015).
The vision of the study was “For U.S. citizens to live free from the fear of nuclear or radiological terrorism.” The mission of the GNDA was to “Protect the nations from terrorist attacks that use radiological or nuclear materials.” Table 7.2 shows the goals, objectives, and value measures that were used in the analysis. The objectives were numbered to align with the goals and the value measures were numbered to align with the objectives. Twenty-five value measures were used in the study.
Table 7.2 Value Model Structure
Goals | Objective | Measure |
1. Reduce the threat | O 1.1 Deter terrorists' RN attacks by demonstrating high likelihood of failure | M 1.1 Probability of deterrence by denial (1) |
O 1.2 Identify terrorist plans for RN attacks | M 1.2 Probability of identifying plans (2) | |
2. Reduce the vulnerability | O 2.1 Detect RN materials out of regulatory control outside of the United States O 2.1.1 At foreign ports of departure |
M 2.1.1 Probability of detection (PD) at foreign seaports (3) |
O 2.1.2 In route to the United States | M 2.1.2.1 PD at foreign airports (4) | |
M 2.1.2.2 Percent of vessels in US coastal regions that are searched (5) | ||
O 2.2 Detect at US Borders O 2.2.1 Detect at US POEs |
M 2.2.1.1 PD major seaports (6) | |
M 2.2.1.2 PD at minor seaports (7) | ||
M 2.2.1.3 PD at airports (8) | ||
M 2.2.1.4 PD at vehicle entries (9) | ||
M 2.2.1.5 PD at airports (10) | ||
O 2.2.2 Detect between US POEs | M 2.2.2.1 PD on land between POEs (11) | |
M 2.2.2.2 PD on coasts between seaports (12) | ||
O 2.3 Detect inside the US O 2.3.1 PD inside US |
M 2.3.1.1 Train (13) | |
M 2.3.1.2 Truck (14) | ||
M 2.3.1.3 Inland waterways (15) | ||
M 2.3.1.4 Primary airports (16) | ||
M 2.3.1.5 Other airports (17) | ||
O 2.3.2 PD around target vicinity | M 2.3.2.1 Number of major urban areas (18) | |
M 2.3.2.2 Number of critical infrastructure (19) | ||
3. Reduce the consequences of a successful attack | O 3.1 Divert nuclear attacks to lower consequence targets O 3.1.1 Urban areas |
M 3.1.1 Percent of population covered by Urban Security Initiative (20) |
O 3.1.2 Critical infrastructure (CI) | M 3.1.2 Percent of CI targets protected (21) | |
O 3.2 Provide early warning of RN attacks | M 3.2 Detection alert times (22) | |
4. Reduce unintended side effects | O 4.1 Minimize impacts on privacy and civil liberties | M 4.1 Number of complaints from civil liberties groups (23) |
O 4.2 Minimize impacts on flow of commerce and the economy | M 4.2 Level of impact (24) | |
O 4.3 Avoid transfer of RN risks to other nations | M 4.3 Number of countries spending money on GNDA efforts (25) |
The bold items are in the NRC report. (RN = radiological and nuclear, PD = probability of detection, POE = point of entry/embarkation, CI = Critical infrastructure).
Decision objectives are based on shareholder and stakeholder value. Crafting objectives and value measures is a critical step in the trade-off analyst's support to the decision-maker and helps qualitatively define the value we hope to achieve with the decision. We described the differences between Value-Focused Thinking and Alternative-Focused Thinking. We recommend the use of Value-Focused Thinking for trade-off analyses. It is not easy to identify a comprehensive set of objectives and value measures for a complex decision. We describe the four techniques for identifying decision objectives: research, interviews, focus groups, and surveys. Research is essential to understand the problem domain and to determine applicability of different decision analysis modeling techniques. Interviews are especially useful for senior leaders. Focus groups work well for stakeholder representatives. Surveys are especially useful to obtain public opinion. We note that the financial or cost objective is almost always an important objective. Next, we describe the important role of hierarchies in structuring objectives and providing a format that is easy for decision-makers and stakeholders to review and provide feedback. We present objectives and functional hierarchies. We recommend functional value hierarchies for complex system decisions. We define four standard techniques for structuring objectives: Platinum (senior leader interviews), Gold (documents), Silver (meetings with stakeholder representatives), and Combined (using all three). The combined standard is the most common. Two examples are provided illustrating approaches to identifying objectives and value measures.
Objective | Value Measure(s) | Category |
Maximize fuel efficiency | Miles per gallon | |
Minimize impact on environment | Miles per gallon | |
Maximize safety in crash | National Highway Traffic Safety Administration (NHTSA) 5-star crash test rating | |
Maximize automobile safety | National Highway Traffic Safety Administration (NHTSA) 5-star crash test rating | |
Maximize vehicle safety | Number of seat belts Vehicle stopping distance Depth of tire tread remaining |