### Goals of the project

The term “flipped classroom” (“classe inversée” in French) refers to a teaching approach in which students get a first exposure to course content before class, typically through readings or videos, then spend class time deepening their understanding of that content through learning activities with the teacher and/or teaching team (e.g. interactive quizzes, experiments, demonstrations, group problem solving assignments, etc.). This model is usually opposed to a more “traditional” teaching approach in which students get the first exposure in class, mainly through lectures, and then deepen their understanding through homework after class, when the teacher is not present.

After having been developed in K‐12 education in the USA over the past decade, the flipped classroom model has been increasingly adopted internationally in higher education in the last few years. As MOOCs are developed in many institutions, the flipped classroom model is more and more seen as an efficient way to integrate the use of MOOC resources into on‐campus teaching and learning activities. The flipped classroom model builds on several well‐known and well documented teaching approaches. In particular, using in‐class time for more active learning activities as been constantly shown by research to improve learning. A major meta‐analysis published in 2014 by Freeman et al., who compared more than 200 studies, has shown that “active learning leads to increases in examination performance that would raise average grades by a half a letter, and that failure rates under traditional lecturing increase by 55% over the rates observed under active learning.”.

The goals of this project are to **explore the issues involved in implementing flipped class approaches in a large class, and to assess the impact of such an approach on students’ learning approaches and behaviours**.

The project team:

Simone Deparis

Cécile Hardebolle

Himanshu Verma

Roland Tormey

In 2018, Simone Deparis has been awarded the 2018 Credit Suisse Award for Best Teaching for his flipped classroom.

### Project overview

Le Linear Algebra courses (MATH-111) are offered during the propedeutics year to all the EPFL classes but Architecture, Mathematics and Physics. The final exam is common at 80% among all these courses, which are taken by a total of about 1’800 students.

In addition to the already existing courses, **a flipped version of the course, given by Simone Deparis, was opened to volunteers in 2017-2018 and in 2018-2019**. A carefully designed study has been put in place over the successive editions of the course.

#### Design of the course

The flipped part of the course was designed as follows. The weekly course material was learned by students at home using the MOOC Algèbre Linéaire by Prof. Donna Testerman. In class, the students worked on specific proofs and advanced exercises individually or in small groups with the support of Prof. Deparis. A classroom response system (clickers) was used to gather students’ responses. The problems were then corrected collaboratively, with class-wide discussions focusing on conceptual and methodological difficulties.

#### Design of the study

Students attending the flipped class have been selected among volunteers using stratified sampling to create comparable experimental and control groups while controlling for volunteering bias. 500 students volunteered for Edition 1 of the course and 100 were selected for the Experimental group. In Edition 2, the experimental group was scaled up to 180 (over 400 volunteers) to better represent the usual size of EPFL classes.

A crossover design has been used to allow comparisons between the traditional and the flipped teaching approaches while controlling for teacher and class effects. Only 5 weeks out of 14 were flipped in Edition 1 of the course, with the first and last weeks taught traditionally. In Edition 2, 10 weeks were flipped to study effects of scaling up the model in time.

### The 2017-2018 Edition

A first pilot course MATH-111(pi) was run successfully in the Autumn semester of 2017-2018. The “flipped classroom” approach was implemented over **5 weeks of the semester**, with a **group of 100 students selected among the 500 who volunteered for the course**.

This first occurrence yielded interesting results. First, the student performance at the final exam was statistically equivalent to that of the control group (the 400 volunteers not selected for the course). The results tend to show a higher learning gain for the pilot group than for the control group, although it is not statistically significant. In addition, the student satisfaction with the quality of the course was high, as in the previous years with the same teacher and a traditional format. Students highly appreciated the flipped part of the course, so much that they asked for more flipped sessions during the semester.

For these reasons, it was decided together with the SMA section to run a second occurrence of the course in 2018-2019 with more time allowed for the flipped format and a larger group of students.

### The 2018-2019 Edition

For the second occurrence of the pilot course MATH-111(pi) it has been decided to implement the “flipped classroom” approach over **10 weeks of the semester**, with a **group of 200 students**, to better represent the usual size of EPFL classes and to study effects of scaling up the model in time.

In this second edition, the findings suggest a reduced gender gap (both male and female students perform equally well in the flipped condition, whereas males outperform in control condition) and a reduced impact of student background in the final exam. However, the overall performance of the experimental group is statistically comparable with the control group.

We are therefore already capable of answering one essential question: the student performance in the flipped class was statistically equivalent to that of the control students (the non-selected volunteers). In addition, the dataset generated allows to explore a number of questions on the Flipped Classroom model, including long term learning effects, inclusiveness, self-regulated learning effects, and motivational effects.