Generically, decision rules determine when data meet the criteria for inclusion in a particular category. Decision rules are the final steps in polygraph numerical scoring, producing categorical classifications. Optimal decision rules require the following: tracing feature selection; development of best scoring rules, consideration given for base rates; assessing and weighting collateral or countervailing information, and; performance of a cost and benefit analysis to determine the achievable level of accuracy and errors that meet the needs of the consumer. In polygraphy, feature selection and scoring rules have been thoroughly investigated. There are also decision rules in some polygraph analysis systems that include extra-polygraphic information as part of the decision process, though there is no validated method yet published. However, few published decision-scoring procedures allow for consideration of the base rate issue. Also, few models publish a sufficient level of detail to allow a formal cost-benefit analysis to identify the appropriate cutting scores for a set of conditions. See Swets, Dawes & Monahan (2000).
