The Use of Judgement Analysis and A Modified Canonical JAN in Evaluation Methodology

Abstract

Judgment Analysis is presented as a technique for capturing and clustering unidimensional policies among a group of judges or evaluators. JAN utilizes a multiple linear regression model to represent each policy and then cluster evaluators together who are expressing similar policies. JAN is extended to a multidimensional situation in which a modified and simplified Canonical JAN (C-JAN) procedure for capturing policies on more than two criteria is described. Both unidimensional and multidimensional JAN procedures should be of general interest to the evaluation methodologist.