BASED ON THE PREMISE THAT DISCRETION SERVES LARGELY TO DISCRIMINATE JUVENILES IN TERMS OF RISK OF FUTURE MISBEHAVIOR, THIS ARTICLE DESCRIBES HOW THIS JUDGMENT PROCESS CAN BE DESCRIBED AND EVALUATED. IN THE LENS MODEL, THE ISSUE IS THE ACCURACY WITH WHICH ONE JUDGES THE OCCURRENCE OF AN ECOLOGICAL EVENT, WITH ACCURACY REPRESENTING A CORRELATION BETWEEN THE DECISIONMAKER'S USE OF AVAILABLE INFORMATION IN ARRIVING AT JUDGMENTS ABOUT THE EVENT AND THE EMPIRICAL RELATIONSHIP OF THE INFORMATION AVAILABLE TO THE DECISIONMAKER AND THE EVENT. JUDGMENTAL ACCURACY DEPENDS ON THE CONSISTENCY OF A DECISIONMAKER'S JUDGMENT, THE PREDICTABILITY OF THE EMPIRICAL EVENT, AND THE SIMILARITY BETWEEN THE WAYS IN WHICH THE CUES ARE USED AND ARE RELATED TO THE EMPIRICAL EVENT. TO APPLY THE MODEL, RESEARCHERS MUST FIRST IDENTIFY CUES (CASE RECORDS ARE A GOOD SOURCE), QUANTIFY THOSE CUES, AND ASSEMBLE THE TEST CASE BY RELATING THE CUE VALUES TO ONE'S JUDGMENTS OF RISK AND TO AN EMPIRICAL MEASURE OF DELINQUENCY. IF THE FIRST TWO STAGES ARE PROPERLY COMPLETED, AN INVESTIGATOR CAN PROCEED TO ASSESS THE RELATIONSHIPS BETWEEN INDIVIDUAL PROFESSIONALS' JUDGMENT OVER A SERIES OF CASES AND THE VALUES OF THE CUES OVER THOSE SAME CASES, AFTER WHICH MULTIPLE REGRESSION ANALYSIS CAN PROCEED. NEXT, THE EMPIRICAL RELATIONSHIP BETWEEN THE CUE AND VARIABLES AND JUVENILE MISBEHAVIOR MUST BE ASSESSED, AND FINALLY, JUDGMENT POLICIES CAN BE EVALUATED BY DETERMINING THE EXTENT TO WHICH MISBEHAVIOR CAN BE PREDICTED, THE CONSISTENCY WITH WHICH THAT INFORMATION IS USED, AND THE MATCH BETWEEN THE WAYS IN WHICH THE CUES ARE USED AND IN WHICH THEY ARE EMPIRICALLY RELATED TO FUTURE MISBEHAVIOR. THIS SOCIAL JUDGMENT THEORY FRAMEWORK HAS SEVERAL ADVANTAGES: IT LEADS TO THE STUDY OF DISCRETION AT THE LEVEL OF THE INDIVIDUAL DECISIONMAKERS, EMPLOYS MULTIVARIATE RATHER THAN UNIVARIATE TECHNIQUES, AND PROVIDES FOR THE SIMULTANEOUS DESCRIPTION AND EMPIRICAL EVALUATION OF DISCRETIONARY POLICIES. HOWEVER, THE FRAMEWORK IS LIKELY TO BE DISTORTED BY INCOMPLETE CUE SPECIFICATION AND IS WEAKENED BY THE NONUNIFORM DESCRIPTIONS OF THE EMPIRICAL EVENT AND OF PROBLEMS IN INTERPRETING DATA. THE EQUATION AND REFERENCES ARE INCLUDED. (DAG)