The use of analogues in comparability analysis.

Comparability studies are an engineering procedure that uses similar equipment as a basis for predicting features of equipment still in the design stage. Interviews with engineers who have performed comparability studies, and analysis of the reports prepared, showed patterns of inference that could not fit any existing account of reasoning by analogy. The reasoning patterns seemed to be task-driven rather than comparison-driven. Expertise in the use of vague knowledge, such as rules of thumb, played a key role in applying the analogy. We concluded that naturalistic problems constituted a use source of data for studying inferential processes.