Comparing buffer versus transmission models of bullying victimization, math self-efficacy, and math achievement: A multilevel approach

Abstract

Experiences of bullying victimization (BV) associate with poorer academic performance and academic self-efficacy, with self-efficacy often mediating the relation between BV and academic performance. This transmission model implies that BV can erode the social cognitive and affective self-perceptions related to academic performance (self-efficacy), thereby declining actual academic performance. A plausible alternative but under-investigated model is that academic self-efficacy protects against the impacts of BV on students’ academic functioning by moderating the relation between BV and achievement (the buffer model). In this study, we use the Trends in International Math and Sciences Study (TIMSS) United States data to compare these models of BV, math self-efficacy (MSE), and math achievement and differences in this model across student, classroom, and school contexts. Approximately 8,700 students in 270 schools and 500 schools in Grades 4 and 8 each participated in the study. Using Bayesian multilevel multivariate modeling, we compare the buffer and transmission models, finding that in both Grades 4 and 8, the transmission model evidenced the best relative model performance. We discuss the patterns of relations at the individual student level and also examine classroom and school context effects. We discussion implications research and practice as well as limitations and future directions.