THE SURVEY WAS ADMINISTERED TO STUDENTS IN SELECTED SCHOOLS IN A MID-ATLANTIC CITY OF 250,000. WHILE ALL THE STUDENTS PARTICIPATED, THIS ANALYSIS IS RESTRICTED TO THE 914 MALE RESPONDENTS; 191 FROM A SCHOOL IN A PREDOMINATELY BLACK AREA, 386 FROM A WHITE LOW-INCOME NEIGHBORHOOD, AND 337 FROM A MIDDLE-INCOME COMMUNITY. EACH PUPIL WAS ASKED TO INDICATE WHEN OR WHETHER HE COMMITTED ANY OF 19 ITEMS INCLUDED IN THE GENERAL SELF-REPORTING SCALE. FOURTEEN ITEMS DEALT WITH DELINQUENT OR DISAPPROVED ACTIVITIES, TWO OTHERS WERE DESIGNED AS 'GOOD BOY' ITEMS TO HELP BREAK THE TENDENCY TO A PARTICULAR RESPONSE SET, AND THE OTHER THREE DEALT WITH REPORTS OF CONTACT WITH THE POLICE OR COURTS. THE FOLLOWING FOUR CHECKS WERE USED IN A POSTSURVEY APPRAISAL OF THE QUALITY OF THE SELF-REPORT DATA: (1) CHECKING SELF-REPORTS OF EVENTS WITH OTHER REPORTS, UTILIZING EITHER RELIABLE RECORDS OR OTHER OBSERVERS; (2) USE OF DISTINCTIVE CRITERION GROUPS, THE 'KNOWN GROUP' APPROACH; (3) USE OF A 'LIE SCALE' OR 'SOCIAL DESIRABILITY SCALE' TO MEASURE INDIVIDUAL TENDENCIES TO PRESENT AN OVERLY FAVORABLE SELF-IMAGE; AND (4) MEASURES OF INTERNAL CONSISTENCY, OR INTERLOCKING ITEMS. OVERALL THE FINDINGS SUGGEST THAT THE SELF-REPORT INSTRUMENT YIELDS ACCURATE RESPONSES AND CAN MAKE AN IMPORTANT CONTRIBUTION TO ETIOLOGICAL STUDIES OF DELINQUENT BEHAVIOR. EVIDENCE ALSO SUGGESTS THAT THE ITEMS ARE SUFFICIENTLY SENSITIVE TO REFLECT BEHAVIORAL CHANGE. FOR EXAMPLE, FOR SEVERAL YEARS THE CITY HAD BEEN PLAGUED BY A LOSS OF PARKING METER REVENUES THROUGH THEFT. IN CERTAIN SECTIONS OF THE CITY, THE TECHNOLOGY FOR 'CRACKING METERS' WAS BECOMING WIDESPREAD AMONG THE YOUTH. HOWEVER, THE OLD METERS WERE REPLACED BY NEWER THEFT-PROOF MODELS A YEAR BEFORE THE SURVEY AND METER THEFTS WERE REDUCED. THE EFFICACY OF THIS STEP AS ONE DELINQUENCY CONTROL DEVISE IS ATTESTED TO BY THE REPORTS OF THE BOYS THEMSELVES. THE QUESTIONNAIRE, TABULAR DATA, AND REFERENCES ARE INCLUDED. (KBL)