## Data Gathering and Representation Techniques

• Interviewing. Interviewing techniques are used to quantify the probability and impact of risks on project objectives. The information needed depends upon the type of probability distributions that will be used. For instance, information would be gathered on the optimistic (low), pessimistic (high), and most likely scenarios for some commonly used distributions, and the mean and standard deviation for others. Examples of three-point estimates for cost arc shown in Figure 11-8. Documenting the rationale of the risk ranges and the assumptions behind them arc important components of the risk interview, bccause they can provide insight on the reliability and credibility of the analysis.

 WBS Element Low Most Likely High Design \$4 M \$6 M \$10 M Build \$16M \$20 M \$35 M Test \$11M \$15 M \$23 M Total Project \$31M \$41M \$68M

Interviewing relevant stakeholders helps determine the three-point estimates for each WBS element for triangular, beta or other distributions. In this example, the likelihood of completing the project at or below the most likely estimate of \$41 million is relatively small as shown in the simulation results in Figure appearing in 11.4,2.2 (Modeling and simulation

Interviewing relevant stakeholders helps determine the three-point estimates for each WBS element for triangular, beta or other distributions. In this example, the likelihood of completing the project at or below the most likely estimate of \$41 million is relatively small as shown in the simulation results in Figure appearing in 11.4,2.2 (Modeling and simulation

Figure 11-8. Range of Project Cost Estimates Collected During the Risk Interview

• Probability distributions. Continuous probability distributions represent the uncertainty in values, such as durations of schedule activities and costs of projcct components. Discrete distributions can be used to represent uncertain events, such as the outcome of a test or a possible scenario in a decision tree. Two examples of widely used continuous distributions arc shown in Figure 11-9. These asymmetrical distributions depict shapes that are compatible with the data typically developed during the quantitative risk analysis. Uniform distributions can be used if there is no obvious value that is more likely than any other between specified high and low bounds, such as in the early concept stage of design.

Beta Distribution Triangular Distribution  Seta and triangular distributions are frequently used in quantitative risk analysis. The data shown here is one example of a family of such distributions determined by two "shape parameters". Other commonly used distributions include the uniform, normal and lognormal. In these charts the horizontal (X) axes represent possible values of time or cost and the vertical (Y) axes represent relative likelihood,

Figure 11-9. Examples of Commonly Used Probability Distributions

• Expert judgment. Subject matter experts internal or external to the organization, such as engineering or statistical experts, validate data and techniques. 