Research is needed to identify differences in outcomes caused by specific interventions with particular types of offenders. Also needed is the development, implementation, and continuing validation of a viable offender classification system. Research also must give greater attention to the specifics of the intervention itself so that results can be replicated and failures avoided. Finally, future studies must be methodologically rigorous. There must be adequate definition of the program or techniques being tested. There should be a preferably random division of a given population of offenders into an experimental and a control group. There must be evidence that treatment is received only by the experimental group. There should be before and after measures and operational definitions of program success and failure, as well as some followup measure. Finally, cooperation is needed between researchers and administrators to ensure that research results will have practical applicability. 3 references.