In recent years, perspectives on globalisation suggest that there is a harsher side of globalisation, which Appadurai (2006) described as marked by "increased violence, exclusion, and growing inequality" within and across nation-states (p. ix-x). Indeed, violent conflict appears pervasive in the world today, shown for example in the recent terrorist attack in Mumbai, the wars in Afghanistan and Iraq, and the continued bloodshed in Gaza. Along with increases in violence and conflict worldwide are growing numbers of violent incidences in schools around the world, such as the recent school shootings in Finland and the U.S. These acts of school violence have received international media attention, sparking an increase of thinking and worrying about school safety, the relationship between school and society, and the implications of violence and bullying in schools. As noted by Akiba (2008), there has been increase in violence in schools recently, reflected in a number of Trends in International Mathematics and Science Study (TIMSS) participating countries. This has led to a set of national initiatives in different countries and the sharing/borrowing of policies and practices across countries, which intended to disrupt violence in schools (Akiba et al., 2002; Akiba, 2008). Nonetheless, as recent studies illustrate, many countries still experience substantial violence and incidences of bullying in schools. This paper uses large-scale, cross-national data to examine the nature of the relationship between school violence and mathematics achievement at an international level. Based on TIMSS 2007 student, teacher and principal responses to school violence questions in nearly 50 countries, we model the relationships between violence at the school and individual level and achievement. As smaller-scale studies have shown, both school violence (Schwab-Stone et al., 1995) and national levels of conflict (Saigh, Mroueh, and Bremner, 1997; Schwartz and Gorman, 2003) have deleterious effects on scholastic achievement. Therefore it is reasonable to expect that these relationships will hold at the international level especially given the globalising nature of violence as described by Appadurai (2006). Of particular importance for this analysis are the following research questions: 1) Is student achievement in mathematics associated with inter-personal and school levels of violence across participating TIMSS 2007 countries? 2) Does the relationship between achievement in mathematics and personal and school violence differ across boys and girls internationally?

Given the hierarchical structure of the data, multilevel modelling methods were used to examine the relationship between violence and mathematics achievement. We use a two-level intercepts-and-slopes-as-outcomes model with students at level one and schools at level two. We use as level one predictors a measure of bullying levels and gender. In addition, mean bullying levels, frequency of behaviour problems teacher safety and disruptive student problems are used as level two predictors. A separate model is fit for each country.

Our findings indicate that, in many countries, student perceptions of violence and bullying are key predictors of educational mathematics achievement. In some countries, the effect is high and should be noted as a major predictor of student achievement. For most countries the mean levels of school bullying (an aggregate of individual bullying experiences) did not predict achievement. However, in those countries where school levels of bullying were related to achievement, the size of the effects was extremely large. In our models, after controlling for violence and victimization, there are significant gender differences for about half of the countries surveyed. Interestingly, while levels of bullying and gender were significant predictors of achievement, we found that there is no interaction between gender and bullying. The paper will include specific country examples.