Learning from peer performance and peer effort: evidence from a field experiment

We provide students with relative performance feedback. We provide feedback in terms of effort (how many hours they planned to study and some information about the distribution of planned effort of their peers), and in terms of grades.
At the beginning of the academic year students are randomly assigned to groups in which they make assignments. Each group is randomized into one of four treatments. The treatments differ in the quality of the feedback that students receive about the performance of their peers. T1: low quality information about peers’ effort and grade, T2: low quality info effort, high quality info grade, T3: high quality info effort, low quality info grade, and T4: high quality info about both effort and grade.
We will examine heterogeneity of the results by differentiating between students who are new at the university where the intervention takes place, and those who were already students at the institution. Hereby it is possible to learn whether information shapes outcomes differently when students are aware of the "norms".

2016-09-19

2017-10-27

Beliefs (mean and dispersion) on both grades and in-class rank, effort (planned and actual study hours), (absolute) grades.

An explanation for the elicitation and calculation of beliefs can be found in the analysis plan, this document also explains how we ask students about planned and actual effort, and how we deal with `mistakes' that students make when willing in the surveys.

The experiment is conducted during a graduate course which takes eight weeks. The final grade consists out of performance on two assignments and a final exam in the eighth week. The set-up of the experiment is as follows.

Stage 1: In the first public lecture students are requested to provide their beliefs about how they will perform in the course (grade, scale from 1 to 10). After this they are given a mock test with basic questions related to the course, after which we ask their beliefs again. This stage is identical for all students.

Stage 2: Students are randomized into groups of 3 to 5 students and these groups are randomized to one of four possible treatment groups. These are the groups in which students have to do the two assignments.

Stage 3: Students have finished their first assignment and are presented with the results. Grades are given in a survey in which we also provide the students with information on the performance of their group relative to other groups, and their planned study hours versus other individual students. Students are asked to fill in these surveys in private. Again we ask them about their beliefs about their performance for the rest of the course, and their planned effort.

Stage 4: This stage is similar as the third stage, only now students are provided with information about their performance in the second assignment, and we give them feedback regarding the amount of planned study hours as they provided in stage 3. The survey also contains questions about beliefs and study effort.

Stage 5: Students complete the final exam.

Randomization done in office by a computer.

Students are randomized into groups of 3-5 students (within tutorial groups), these groups are randomized into treatments. Treatment status is balanced over the five different tutorial groups.

Yes

The experiment is run for two consecutive years in a course, which each year has approximately 150 participants. These students are randomized into groups of 3-5 students, and treatments are balanced across tutorial groups (ten groups in two years). Over the two years the experiment is run, we will have 95 groups and hence 95 clusters.

About 344 students are randomized into different treatment groups over two academic years.

There are four treatment arms. T1 contains 88 individuals, T2 contains 91 individuals, T3 contains 83 individuals and T4 contains 77 individuals.