UNANTICIPATED CHANGES OCCUR WHEN AN ATTEMPT IS MADE TO CREATE COMPARABLE TREATMENT AND CONTROL GROUPS. THERE IS DANGER ASSOCIATED WITH GOOD INTENTIONS, AND BIASES EXIST WHICH FAVOR THE SUCCESS OF EXPERIMENTAL PROGRAMS. PROBLEMS ARE ENCOUNTERED IN THE INTERPRETATION OF RESEARCH RESULTS. AN ARGUMENT IS MADE AGAINST INDEPENDENT OUTSIDE RESEARCHERS AND FOR LESS SOPHISTICATED RESEARCH TECHNIQUES. BECAUSE RESEARCHERS ARE FREQUENTLY CRITICAL OF THE QUALITY AND INTERPRETATION OF OFFICIAL STATISTICS, CAREFUL RECORDKEEPING IS ESSENTIAL. THE USE OF GENERAL RECORDS FOR EXAMINING ONGOING PROGRESS HAS ADVANTAGES OVER OTHER EVALUATION TECHNIQUES. WITH CAREFUL RECORDKEEPING, STUDIES OF COHORTS ARE POSSIBLE. MODEST STATISTICAL TOOLS CAN PROVIDE SIGNIFICANT INFORMATION AND, ONCE COMMITTED TO ASKING QUESTIONS, THE JUVENILE JUSTICE SYSTEM MAY BE RECEPTIVE TO ASKING MORE THEORETICAL QUESTIONS. REFERENCES ARE CITED. (DEP)