Joffredo-Lebrun, Sophie, Meyer, Samantha, Vilette, Bruno, Sander, Emmanuel, Sensevy, Gérard, Hild, Marie, Centre de Recherche sur l'Education, les apprentissages et la didactique (CREAD EA 3875), Université de Brest (UBO)-Université de Rennes 2 (UR2), Université de Rennes (UNIV-RENNES)-Université de Rennes (UNIV-RENNES)-Institut Brestois des Sciences de l'Homme et de la Société (IBSHS), and Université de Brest (UBO)
International audience; This study fits in a more global project that aims to elaborate a mathematical program for the first year of elementary school (ACE Project supported by the Ministère de l’Education Nationale). The objective is to emphasize the role of a regular and systematic practice of estimation during all the 1rst year of school mathematical learning. This role is to map the acquisition of numerical symbol with the «sense » of number and calculation. In this approach, it is the interaction of both representational system of number and calculation (exact and symbolic versus approximative and analogical) that is crucial.In the present study, 667 children from North of France participate. Theses students have been divided into an experimental group (N=365) and a control group (N=302).During the 1st school year, the experimental group followed the ACE program composed of different activities (mental calculation, problem solving, calculation) and a weekly training (45 minutes) of numerical estimation (program on computer). In this program, children have to mark, on a borned continuous line, the position of a number or the result of a calculation, which allows a mapping between symbolic number and analogical magnitude. The control group followed their teacher habitual mathematical program. All students realized a Number Knowledge Test (NKT) composed of several arithmetical tests (mental calculation, problems resolution, written additions, numeration) at the beginning and the end of school year.Mainly, we found no significant difference on pre-test between both groups on all NKT-measures. In the post-test, results are significantly higher for the experimental group compared to the control group on all NKT measures and on the estimation task (p