Available Languages: English, French

This paper reports on a project to introduce interactive learning strategies (ILS) to physics classes at the Université Pierre et Marie Curie, one of the leading science universities in France. In Spring 2012, instructors in two large introductory classes, first-year, second-semester mechanics, and second-year introductory electricity and magnetism, enrolling approximately 500 and 250 students, respectively, introduced ILS into some, but not all, of the sections of each class. The specific ILS utilized were think-pair-share questions and Peer Instruction in the main lecture classrooms, and University of Washington Tutorials for Introductory Physics in recitation sections. Pre- and postinstruction assessments [Force Concept Inventory (FCI) and Conceptual Survey of Electricity and Magnetism (CSEM), respectively] were given, along with a series of demographic questions. Since not all lecture or recitation sections in these classes used ILS, we were able to compare the results of the FCI and CSEM between interactive and noninteractive classes taught simultaneously with the same curriculum. We also analyzed final exam results, as well as the results of student and instructor attitude surveys between classes. In our analysis, we argue that multiple linear regression modeling is superior to other common analysis tools, including normalized gain. Our results show that ILS are effective at improving student learning by all measures used: research-validated concept inventories and final exam scores, on both conceptual and traditional problem-solving questions. Multiple linear regression analysis reveals that interactivity in the classroom is a significant predictor of student learning, showing a similar or stronger relationship with student learning than such ascribed characteristics as parents' education, and achieved characteristics such as grade point average and hours studied per week.