Statistical modeling of the attitudes toward mathematics from a school survey data of Mayotte

Résumé 0

Motivated by two survey data samples (student and teacher) obtained from institutional records in some school establishments of Mayotte, this paper has two main objectives. Firstly, a global application of statistical models for the analysis of students' attitude toward mathematics (difficulty, interest or perception, value of the level in mathematics and others attitudes such as social factors, school environment) together with the teachers' attitude toward mathematics (difficult courses, professional training experience, teaching preparation time, etc). Some didactic analysis have helped to reach conclusions to ensure the consistency of certain characterization. Secondly, we develop a theoretical methodology for parameter estimation of the density of the students' level in mathematics and teachers' attitude in the framework of stable and extreme values distributions which are suitable for studying non Gaussian and heavy-tailed distributions. More precisely, we described a new estimation method based on the tail regression estimation method for stable distributions. For the extreme family distributions, we consider a mixture estimation using the EM algorithm. Finaly, we end with a brief discussion on a current experimentation of the statistical and didactic analysis results in the form of Lesson study involving several teachers from the island of Mayotte.