The required inputs will vary according to the statistical distribution selected. Most often a small group of experts will work together to determine the appropriate inputs for each risk distribution. This group will include representatives from the owner's team, hired experts, and Partnerships BC staff.
It is common when using a distribution that is bounded like the triangular, or a discrete distribution to complete the process with a distribution that has a narrow range of potential outcomes than would be actually observed. This is often due to an over-confidence of the estimator in the middle point of their range. Some techniques to deal with this central tendency bias are:
i) If using a distribution that is bounded and has a defined mid point (e.g., a triangular distribution) a smaller range will generally result if the most likely result is identified first, then moving out to the points at the extremes. When an expert identifies a most likely estimate they will tend to then want to defend it and a larger range of potential outcomes can seem like they are unconfident in the most likely estimate. This can be overcome by first determining the extreme bounds of the distribution. The maximum value should be determined first. A good opening question to an expert is "What is the worst impact you have ever witnessed or know about for a similar risk in a comparable project"? The second step would be to determine what would be the best impact they have ever witnessed for a similar risk. Having set the bounds now the most likely impact can be determined.
ii) Experts and others providing inputs can be challenged with suggesting comparable bets. For example if someone said they were 90 per cent certain that a certain outcome would be less than a set amount they can then be offered a bet. The bet would be: would they risk $100 for the chance to win $1,000 if the actual outcome is greater then the set limit. If they say yes to that proposition then that indicates they really think the chance of realizing that outcome is greater than 90 per cent (they would rather have the bet). The goal is to reach a point of indifference. They should then be asked what level of outcome they would be indifferent to the bet or the outcome.