All the above forms of diagnosis require attention as the subset of issues grows. Effective and efficient choices cannot be left until all the issues have been jointly evaluated. Careful attention to partial results as the evaluation process proceeds will suggest changes to those looking for such changes. In particular, conditions or scope assumptions associated with partial aggregation of quantified issues may need careful attention. Such conditions or assumptions can become crucial at the stage when all quantified issues have been aggregated (e.g., when defining a budget for a delivery date). When all quantified issues have been combined, interpretation needs to address the importance of all assumptions, including conditions that have not been quantified. Chapman and Ward (2002) develop a 'cube factor' approach to acknowledging identified conditions, unidentified conditions, and bias, briefly illustrated in Chapter 15. The essence of the matter is the need to interpret any quantification of uncertainty as conditional on a range of assumptions, some unidentified, including assumptions leading to bias.
Was this article helpful?