The purpose of this study was to examined the relationship of juvenile delinquents' multifactorial self concept and probability of re-offense. The sample of research was consisted of juvenile offenders in juvenile protection facilities and juvenile prison which are located in Seoul, Busan, Gwangju, Cheon. The result from the regressions to the future probability of re-offense with multifactorial self concepts shows that moral-ethical self concept was negaively related with the probability of re-offense in the four type of delinquent behaviors except rape, personality-related self concept was negatively related with the robbery, violent behavior and rape, social self concept was negatively related with the theft but positively related with violent behavior, ability-related self concept was positively related with rape, and family self concept was negatively related with probability of all the type of delinquent behaviors. The result from the regressions with the interaction terms shows that the four interaction terms significantly were related with the probability of re-offense.

In this paper, the interactive effects of major factors on juvenile delinquency were examined using multiple regression analysis. Major factors on juvenile delinquency are attachment to the parents, supervision of the parents, school achievement, strain caused by school achievement(achievement strain), strain caused by college(or job) entrance (entrance strain), the degree of contact with delinquent peers(delinquent contact) and the degree of bond with delinquent peers(delinquent bond). Interaction terms were created by multiplying each of the major factors. Juvenile delinquency was measured by three methods; the times of minor delinquency, the times of serious delinquency, and the seriousness measure of the committed delinquent behavior by seriousness scale. Three sets of regressions were then performed. First, minor delinquency measure was regressed on the seven variables of major factors, with and without the twelve interaction terms. Second, serious delinquency measure was regressed on the seven variables of major factors, with and without the twelve interaction terms. Third, seriousness measure was regressed on the seven variables of major factors, with and without the twelve interaction terms. The result from the regressions with the interaction terms shows that the twelve interaction terms significantly increase the amount of explained variance in both serious delinquency and seriousness degree but not in minor delinquency. The amount of explained variance in the serious delinquency is increased from 26.8% to 31.5%, or 17.6%. Explained variance in seriousness of delinquency is increased from 31.2% to 34.0%, or 9.0%. But increasement in minor delinquency is not significant, from 43.9% to 44.3%, only 0.9%. The five terms of twelve interaction terms have a significant effect on the serious delinquency or the seriousness of delinquency; 'attachment × entrance strain', 'attachment × delinquent contact', 'supervision × delinquent contact', 'school achievement × delinquent contact', and 'school achievement × delinquent bond'. Overall, the paper indicates that the addition of the interaction terms substantially improves the prediction of serious delinquency.