In the sociological analysis of school inequalities, the effect of school context is a major question. A considerable body of previous research has demonstrated that differences between schools and classes have an impact on students’ achievement. It is now weel recognized that "school makes differences": in certain cases, school context reduces inequalities of achievement, in other cases, it accentuates them. In sociology of education, especially in the quantitative sociology, three levels of context effects are generally distinguished: establishment, class and teacher. The international researches consist to identify the extent of effects and the mechanisms by which each level influences school inequalities. At primary school, the teacher is a very important element to analyse production of inequalities. This communication examines the effect of teachers' conduct on low achievement pupils. Indeed, we know that some teachers are more equitable than others, by the way that they permit a reduction of the inequalities between pupils, to the profit of the lower students. If many results are known about effective classes or schools characteristics, we know less about teachers’good practices. In particular, do teachers really have to be focused on low students, preparing special activities, in spite of stigmatisation risks? Are there differences between the practices of teachers with whom low students differently success? In order to answer to these questions, empirical data were collected, connecting a quantitative and a qualitative approach. In a first time, scores in reading have been measured, in September 2004 and June 2005, among a sample of 1300 pupils in 100 classes of first grade of elementary school in a French academy (Brittany). Teachers filled in a questionnaire characterizing many fields of their practice in the classroom. Multilevel modelling techniques were employed both to estimate pupils’ progress between the beginning and the end of the school year and to link this progress to teachers’ practices. In a second time, five classes from the initial sample were isolated in which pupils who had low achievement at the beginning of the school year had made the best progress; in comparison, five classes were also isolated in which pupils who had low achievement at the beginning of the school year had made the less progress. Ethnographical methods (observations in the ten classrooms and interviews with the ten teachers) were carried out. The results of the first time of the research let us wonder about the impact of specific practices intended for low students, such as ability grouping, differentiation of activities…. More often, the impact is negative, as shown in much previous research. But the second time of the study anables us to moderate and to specify these first results. The specific help given to low students is very present and sustained in the most effective classes but this help is also very different from those provided in ineffective classes. For example, in the most effective classes, teachers keep a certain level of expectations, mainly focus on learning and not only on well-being, offer many types of help in the same time... Moreover, in effective classes, in order to protect low students from stigmatisation effects, teachers pay attention to really involve low achievement students in the whole group of pupils.

Multilevel modelling analyses on a sample of 1300 pupils in 100 classes. Ethnographical methods (observations ans interviews) in 10 classes

A considerable body of previous research has demonstrated that it can be pernicious to systematically focus on low achievement students. Indeed, there are big risks of stigmatising students and reducing expectations. In total, help could be unproductive. A provocative question could be: should teachers help low students? The study anables us to answer positively. Data collected and analyses also give us the oppotunity to specify what type of help can be favourable.