Purpose: Up until this point, theory and research literatures on school violence have been quite theoretically directional—examining how those personal, family, and school factors affect school violence. Few studies have examined how school variables mediate between family factors and school violence (Benbenishty & Astor, 2005). Furthermore, these kinds of relationships have, to date, never been explored in Asian cultures. School experiences, such as a caring climate or good peer group, may act as inhibitors or mediators of family experiences making some students less likely to be violent school members. In order to clarify what role the schools plays in reducing school violence, the current study aims to examine if teacher-student relationships and students' peer group could be mediators between family factors and school violence perpetration.

Methods: The data based in this secondary analysis is a cross-sectional and national probability sample. The data was collected by the National Survey of Current School Violence Incidence in Taiwan and funded by Taiwan National Science Council. Data was collected in 2001. The total sample included over 14,000 students from elementary to high school (Grade 4 to Grade 12). The analysis reported in this proposal focuses on only elementary school students (Grade 4 to 6, N=3120). Students were given a structured and anonymous questionnaire and were asked information about family background, family conflict, parent monitoring, the interaction between school teachers and students, peer group quality, and students' self-report violent behaviors toward peers and teachers.

Results: The results of the Structural Equation Model (using AMOS) analysis based on the total sample provided a good fit to the data with chi-square(df=156, N=3120)= 551.99, p<.001, and with NFI= .976, RFI= .968, IFI= .983, CFI= .983, and RMSEA= .029. This means the model was a good one. The overall model explained 18% of the variance in both student-toward-student and student-toward-teacher aggressive behaviors. Student-toward-student aggressive behaviors were negatively associated with supportive parental monitoring (&Beta= -.20, p<0.01) and positively associated with family conflict (&Beta= .15, p<0.01); student-toward-teacher aggressive behaviors were negatively associated with supportive parent monitoring (&Beta= -.10, p<0.01), however, no direct association was found between family conflict and student-toward-teacher aggressive behaviors (&Beta= -.007, p>0.01). In addition to the direct effect, supportive parent monitoring had an indirect effect, through risky teacher-student relationship and risky peer quality, on student-toward-student (total indirect effect= -.04) and student-toward-teacher aggressive behaviors (total indirect effect= -.07). Family conflict also had indirect effect on student-toward-student aggressive behaviors (total indirect effect= .07) and student-toward-teacher aggressive behaviors (total indirect effect= .16), through risky teacher-student relationship and risky peer quality.

Implication: The findings implicate that the teacher-student relationship and students' peer group could mediated the effect between family risky factors and school violence. This has important implication for intervention and prevention efforts; enhancing the quality of student-teacher relationship and students' peer group may be a key intervention element for treatment/policy as well as preventing the school violence.