Study Practices in Higher Education

Our intervention consists of providing students with social information about previous problem set completion.

Participating students receive a standard message about the availability of the first problem set (control group) or that same message with an additional line stating previous year problem set completion rate (treated group).

2025-08-27

2025-08-28

Problem set completion, Course engagement, Course grade

For Problem set completion we use detailed information from the course online platform (problem set opening, completion, and delivery). We compare the dates with the course recommended due date.

For engagement (and disengagement) we use detailed data from the course activity logs (including course videos).

For course grade we use the student's best grade from final exam and possible retakes, in accordance to the university's guidelines.

Exam performance, Student experience, Rewarded ECTs and GPA in following years

We exploit detailed information on exam performance (including participation in the first exam and in retakes, total points scored, number of answers given, number of wrong/right answers).

Student experience is assessed through anonymous survey items about the course (covering enjoyment, perceived usefulness, perceived burden, and induced stress).

Rewarded ECTs and GPA are captured using university registers for the following three years. These outcomes will only be studied if we find significant results in any of the course outcomes.

We conduct a randomized controlled trial with two groups: a control group receiving a standard message and a treatment group receiving the same message with an additional line containing social information.

We will execute a survey prior to the course to collect student baseline characteristics. We will examine the heterogeneity of the treatment based on measures of students' self-control problems, loss aversion, norm following propensity, prior knowledge of the course concepts (by using a set of multiple choice questions about economic concepts), and prior beliefs about problem set completion rates. From each of these measures we will construct binary variables to categorize students' prior characteristics. For prior beliefs, we will categorize students taking into accout the differences between the social information provided and prior beliefs (negative, close to zero, and positive). Additionally, within those whose prior beliefs are below the information provided, our main focus will be on those whose priors are not too far from the information provided (based on theoretical insights on the impacts of social information).

We will randomly assign our intervention so that the treatment and control groups are balanced on these characteristics.

We will estimate the treatment effects across each categorization and compare the treatment effects by each binary category.

In addition, we aim to study the treatment effect separately on background information gathered from university registry: gender and degree of study. Due to lack of power, for the latter we will focus on the largest groups of students within the course (such as those in the Business Studies Unit and Open University students, which are non-degree students).

Randomization done in office by a computer.

Individual.
We will stratify our randomization based on baseline questionnaire response: creating control and treatment groups within those that responded the questionnaire and those that didn't responded the questionnaire.

No

The design is not clustered.

We expect 600 students to enroll to the course and expect about 300 students to respond the baseline survey before the intervention. First strata includes the students who have responded to the baseline survey prior to intervention, the randomized groups are balanced on the baseline survey characteristics and background information gathered from university registry. The second strata includes the students who have not responded to the baseline survey before the intervention, the randomized groups are balanced on the background information gathered from university registry.

We expect to have 150 treated students and 150 control for those who have answered the baseline survey and 150 treated and 150 control for those who have not answered the baseline survey before the intervention.