Lee Merkhofer Consulting Priority Systems
Implementing project portfolio management

"Creating influence diagrams isn't a required step for building a project selection model, but doing so makes the steps that are required easier"

Creating Influence Diagrams

Once an objectives hierarchy has been constructed for building a project priority system, the factors influencing the achievement of objectives can be identified. An effective way to accomplish this step is to construct influence diagrams. Figure 14 shows how this step relates to the other steps in my process for creating a project selection decision model.


Steps for creating a project selection model

Figure 14:   The influence diagramming step.


Why Construct Influence Diagrams

An influence diagram, also called a relevance map, knowledge map, means-ends diagram and other names, is a useful structuring and modeling tool originally developed in the 1970's at the Stanford Research Institute [1, 2]. An influence diagram displays the factors relevant to a decision as nodes and the relationships among the factors as arrows. Creating influence diagrams isn't a required step for building a project selection model, but doing so makes the steps that are required easier. In my experience, drawing influence diagrams is well worth the effort.

Construct Influence Diagrams as a Group Exercise

Software tools are available for constructing and displaying influence diagrams [3], but you can draw them by hand on a sheet of paper or white board. I routinely work with project portfolio management (PPM) teams and others to construct influence diagrams for the objectives in their objectives hierarchies. "Experts," individuals from the organization with specialized expertise related to a specific objective, are asked to participate, and they typically take a lead role in suggesting influencing factors for diagrams related to their areas of expertise, for example, specialists in health and safety, the environment, the production of the organizations key products and services, regulatory affairs experts, and so forth. Building influence diagrams as a facilitated group exercise enables the participants to gain a shared understanding of the logic by which projects can promote the achievement of objectives [4].

Create an Influence Diagram for Every Lowest-Level Objective Except Financial Objectives

Constructing an influence diagram will prove useful for each of the lowest-level objectives in the objectives hierarchy, with the exception (usually) of financial objectives. The reason that there is typically no need to construct influence diagrams for financial objectives is that project financial metrics and how to calculate them is well-established and understood (at least by individuals in your finance department). What is frequently not so well understood is what metrics should be used for non-financial objectives and how those metrics should be calculated. Constructing influence diagrams helps answer these questions.

How to Build an Influence Diagram

As with the objectives hierarchy, influence diagrams are constructed from the top down, typically as a group exercise [5]. I normally begin by showing the participants one or more influence diagrams created for a similar objective by a team from another organization similar to their organization. Although helpful, you don't need examples to construct an influence diagram. Start by writing the statement for the selected objective near the top of a white board, surround it by a square, and then draw a circle below it with an arrow pointed toward the square. Inside the circle write "Degree to which [insert the objective statement] is achieved." The circle will be the top-level "bubble" in the influence diagram. The goal is to identify, in sequence, the factors that influence it and then the factors that influence the factors that influence it. The idea is to keep adding factors until the factors are outcomes that, in the future, could be measured, and, in the present, can be estimated [6].

Idiosyncratic Nature of Influence Diagrams

Influence diagrams are not unique, the bubbles and arrows in the diagram depend on how the participants decide to decompose the objective, the considerations they view as important, and their view of the logic that connects considerations to the objective. Sometimes, a large influence diagram can be greatly simplified by adopting a different logic at the top of the diagram for connecting and organizing the factors below it.

Example: An Influence Diagram for Public Health Risk from Siting a Hazardous Waste Facility

Figure 15 shows an example influence diagram constructed as part of an exercise sponsored by the U.S. Department of Energy (DOE) wherein members of the public designed and then applied a logic for prioritizing alternative sites for building a facility for long-term storage of hazardous waste [7]. The exercise was conducted after DOE's announcement of a plan to construct the facility at the agency's preferred location was met with considerable resistance from a local citizen's advisory board and the public at large. In response, the DOE asked the citizen's advisory board to work with representatives of the agency to determine whether any acceptable site could be identified for constructing the facility.


Influence diagram for prioritizing waste sites

Figure 15:   An example influence diagram.


The objective for site selection addressed by the diagram is "minimize public health risk." As shown by the first two influencing factors, the participants began by recognizing two types of risk, chronic risk due to low-level exposures over long periods of time via exposure pathways identified in the diagram, and acute risk, which could result if someone violated site security measures (fences) and got onto the site and came in contact with the waste. After creating the diagram, the participants were asked to identify the "drivers;" that is, the factors that, depending on the site chosen, would produce the greatest difference in the amount of public risk. After much deliberation and consulting with the DOE technical specialists, the participants concluded that all of the candidate sites were essentially identical with regard to all of the factors in the diagram except for the distance of the site from community populations. Accordingly, distance from community populations was the metric chosen to evaluate the candidate sites with regard to the presumed level of public health risk.

Influence diagrams were constructed and performance measures selected for each of the other site selection objectives identified by the group, which included worker health and safety, the environment, regulatory appropriateness, cost, and time to complete. Each participant then individually scored and weighted the performance measures. When each participant's weighted scores were summed, the results showed that a remote site was consistently at or very near the top for every citizen participant. That site was not originally favored by DOE in part because the technical specialists viewed all sites as completely safe (zero weight assigned to distance). Also, the citizens' preferred site was relatively uncontaminated and viewed by the technical specialists as likely to obtain certification from regulators as requiring "no further action," an outcome desired by DOE because it would demonstrate progress for dealing with contamination problems. The DOE technical specialists were, however, happy to donate the site in return for public approval to construct the waste facility.

In retrospect picking a remote location for the waste facility may seem like an obvious solution. Prior to the analysis, however, no one had considered the option. Commenting on the fact that citizen participants assigned non-zero weights to distance, one participant explained that his weight represented a "less arrogant level of confidence in the performance of the facility." Participants reported that they liked the emphasis of the process on constructing influence diagrams. Comments included, "I really felt like we did a thorough job," "What we did made common sense," and "This process should be used more often" [7].

An Influence Diagram for Stakeholder Relations

Influence diagrams can be particularly useful for complex objectives. One such objective often included in project priority systems is "protect and enhance the image of the organization." Because the organization's image is determined by how people perceive it, this objective might be expressed as "protect and improve relations with stakeholders." Stakeholder relations can be important for project selection, especially if there are groups whose pleasure or displeasure about the organization's project choices can translate into actions that impact the organization's ability to accomplish its mission.

Figure 16 shows an influence diagram constructed by a PPM team for a stakeholder relations objective. The diagram identifies two main considerations: (1) the specific stakeholder groups that care whether or not the project will be funded and (2) the likely action, if any, that these groups would take depending on whether or not their preferred project decision is chosen. The diagram shows that assessing a project's impact on organizational image requires (a) identifying the stakeholders who have an interest in the project and (b) predicting the likely reactions of these groups. The degree to which knowledge of a threat to stakeholder relations will affect project priorities depends, of course, on the weight that is assigned to the objective.

.
Stackholder relations influence diagram

Figure 16:   An influence diagram for maintaining relations with stakeholders.


Intensity-Quantity Influence Diagrams

The influence diagrams constructed for many complex objectives relevant to project prioritization have a "two-consideration" structure similar to that in the above diagram [8]. One cluster of factors relates to a concept of intensity while the other relates to a concept of quantity. In the organizational image influence diagram above, the "quantity" component specifies the number and types of stakeholders who care about the project decision. The "intensity" component is the intensity of the reaction that those concerned stakeholders will exhibit.

Figure 17 shows an influence diagram developed in an effort to create a priority system for e-learning projects sponsored by a state university system. The influence diagram's objective is "maximize value to degree-completion learners," an objective in the e-learning objectives hierarchy shown on the previous page. Though the diagram identifies a great many factors, the common two-consideration structure is evident. To estimate the value delivered by an e-learning project to students intent on graduating, consideration is given to (1) the number of students whose graduation prospects will be impacted by the project (quantity), and (2) for those students who are impacted, the estimated increase in their likelihood of graduating (intensity).


Degree learners

Figure 17:   An influence diagram for value delivered to degree-completion learners.


Another common influence diagram structure occurs for project prioritization systems for organizations whose assets pose risks to workers or to the public [9]. Figure 18 is an influence diagram developed by a PPM team for an objective to minimize environmental risks. The diagram contains the common two-consideration structure identified above, represented here by the number and sensitivity of the environmental resources at risk (quantity) and the seriousness of the harm that might occur to these resources (intensity). In addition, the diagram shows a common two-consideration structure for accounting for the likelihood of environmental harm: factors impacting (1) the probability that the identified risk scenario will occur and (2) the probability that, if the scenario occurs, the feared level of harm will result to the resources that are at risk.


Stakeholder Attributes

Figure 18:   Influence diagram for environmental risk.


The logic for evaluation implied by this diagram is a simplified version of the standard approach for conducting risk assessment. To evaluate a project impacting environmental risk one or more environmental risk scenarios must first be identified. The specific environmental resources that might be harmed by the scenario are identified (quantity), and the potential consequence (intensity) to those resources is indicated. Finally, the probability of the risk scenario is specified, and then, assuming that the risk scenario occurs, the probability is specified that the feared level of harm will occur to the resources at risk. If serious risk scenarios are identified from conducting this simplified version of risk assessment, then, obviously, the organization will know that consideration should be given to conducting a more rigorous risk assessment.

After creating influence diagrams, the next step in the process of building a project selection model is specifying performance measures.

References

  1. A.C. Miller III, M.W. Merkhofer, R.A. Howard, J.E. Matheson, and T.R. Rice, "Development of Automated Aids for Decision Analysis," Stanford Research Institute, Menlo Park, CA., 1976.
  2. R.A. Howard, J.E. Matheson, M.W. Merkhofer, A.C. Miller, and D.W. North, "Comment on Influence Diagram Retrospective," Decision Analysis 3(2), 117-119, 2006
  3. R. D, Shachter, "Evaluating Influence Diagrams," Operations Research 34(6) 871-882, 1986.
  4. D. L. Keefer and C. W. Kirkwood, "Perspective on Decision Analysis Applications, 1990–2001," Decision Analysis, 1(1), 4-22, January 2004.
  5. M. W. Merkhofer, "Using Influence Diagrams in Multiattribute Utility Analysis—Improving Effectiveness Through Improving Communication," Influence Diagrams, Belief Nets and Decision Analysis 297-317, 1990.
  6. A Detwarasiti and R. D. Shachter, "Influence Diagrams for Team Decision Analysis,"
  7. R. A. Howard and J. E. Matheson,. "Influence Diagrams," Decision Analysis 2(3), 127-143, 2005.
  8. Decision Analysis 2(4), 207–228, December 2005.
  9. R. L. Keeney and R. S. Gregory, "Selecting Attributes To Measure the Achievement of Objectives," Operations Research 53(1) 1-11, 2005.
  10. M. W. Merkhofer, R. Conway, and R. G. Anderson, R.G., "Multiattribute Utility Analysis as a Framework for Public Participation in Siting a Hazardous Waste Management Facility, Environmental Management, 21(6), 831-839, 1997.