Title not available in English

Abstract

Usually, studies of academic achievement consider two broad classes of factors: student factors (gender, socio-economics status, etc.) and school factors (sector, pupil-teacher ratio, teacher training, etc.). In addition to these factors it is important to consider the effect of the whole class upon the achievement of a specific student, known as compositional, contextual or peer effects. Although the peer effect have been represented by the group means, it is not obvious that it is the best way of introducing it. The article evaluates alternatives forms of inclusion of the peer effect in a model of academic achievement. Using the results of the Mathematics test of SAEB-99, it is showed that group means are not enough to capture all the effects of the peers.