To test theory, a policy analysis must be performed, supplemented by an explanation of misconduct the evaluator believes can be tested by the evaluation data. The degree to which the experimental variables was applied to the intended population must be determined as well as the degree to which the intervening variables in the causal process has been changed. The criteria for misconduct must be carefully specified and the theory-testing associations carried out. This kind of theory-testing is illustrated and a widely applicable method of computing Time Unit Offense Experience Scores for evaluating crime and delinquency control programs proposed. Study data and about 30 references are included. (Author summary modified)