THE FOUR MEASURES ARE FREQUENCY OF DELINQUENT ACTIVITY, DIVERSITY OF ACTIVITY, SERIOUSNESS OF ACTIVITY, AND PROGRESSION INTO DELINQUENT BEHAVIOR. A SAMPLE OF 161 PUBLIC SCHOOL CHILDREN IN OMAHA, NEBR., RESPONDED TO A 28-ITEM QUESTIONNAIRE RATING EACH ITEM FOR SERIOUSNESS AT THE BEGINNING AND END OF A 30-WEEK PERIOD. THEIR RESPONSES WERE USED TO DETERMINE THE TEST-RETEST RELIABILITY OF THE INSTRUMENT, WHICH WAS THEN ADMINISTERED ONCE TO A SAMPLE OF 1,030 PUBLIC SCHOOL CHILDREN AND 655 ADJUDICATED DELINQUENTS IN BALTIMORE, MD., INDIANAPOLIS, IND., AND PHOENIX, ARIZ. THE FOUR MEASURES OF DELINQUENCY ARE SHOWN TO BE HIGHLY RELATED; ALL HAVE HIGH TEST-RETEST RELIABILITIES. THE MEASURES TEND TO BE REDUNDANT, WITH NO ONE MEASURE CLEARLY BETTER THAN THE OTHER THREE. THE SERIOUSNESS MEASURE HAS THE GREATEST SENSITIVITY IN DISTINGUISHING BETWEEN PUBLIC SCHOOL CHILDREN AND DELINQUENTS IN THE THREE-CITY SAMPLE, BUT FAILS TO SHOW SIGNIFICANT DIFFERENCES WITHIN THE OMAHA SAMPLE IN AN ANALYSIS OF OF BIRTH ORDER (SHOWN IN OTHER STUDIES TO BE A SIGNIFICANT PREDICTOR OF DELINQUENCY). ALL FOUR SCALES SHOW SOME DIFFERENCES WITHIN THE OMAHA SAMPLE AS A FUNCTION OF SEX, AGE (ALSO PREDICTORS OF DELINQUENCY), AND BIRTH ORDER, BUT THE DIFFERENCES ARE NOT ALWAYS IN THE SAME DIRECTIONS FOR ALL FOUR MEASURES. THERE IS A STRONG NEGATIVE CORRELATION BETWEEN THE RATED SERIOUSNESS OF AN OFFENSE AND ITS REPORTED FREQUENCY OF OCCURRENCE. THIS FINDING LENDS SUPPORT TO THE USE OF FREQUENCY MEASURES IN BUILDING DELINQUENCY SCALES. THE ANALYSIS ALSO CONFIRMS THE DIFFICULTY OF USING SELF-REPORT ITEMS TO BUILD UNIDIMENSIONAL PROGRESSIONS. THE USE OF GUTTMAN SCALES AND OTHER FREQUENCY-BASED APPROACHES AS THE BASIS FOR ARGUMENTS ABOUT THE DIMENSIONALITY OF DELINQUENCY IS QUESTIONED. SUPPORTING DATA AND A COPY OF THE STUDY INSTRUMENT ARE INCLUDED. (AUTHOR ABSTRACT MODIFIED--LKM)