This paper studies the relationship between the creative abilities of study peers and academic achievement. We conduct a novel large scale field experiment at university, where students are randomized into work groups based on their score on a creativity test prior to university entry. We first show that the creative abilities of peers matter for a student’s academic achievement. A one standard deviation higher creativity peer group improves study performance by 7.5 to 10 percentage points. Notably, this effect is driven by the average group creativity, there is no special impact of creative superstars. Further analysis suggests that students exposed to creative peers become more creative, but do not adjust their overall study effort. This is in line with the idea that creative approaches and questions of peers help students to master the study material better. Overall, our study highlights the importance of peer effects of creative students in shaping academic outcomes.
Intervention
The intervention is the exposure to peer groups (other students) who are through randomization more or less creative.
Experimental design
Before entering university, students take a Remote Associates Test (RAT) which measures their creativity abilities. Based on the RAT results students are randomized into small scale work groups. In these work groups students spend most of their study time in the first year of their studies. This study aims to measure the impact of being assigned to a high (or low) creativity peer group on study performance (as primarily measured by their first year GPA).
Design details
Randomization method
Randomization is done in an office / by a computer.
The randomization consists of 2 steps.
1) students take the RAT (see above), and obtain a score between 0 - 12. Every students below the mean is labeled as 'low creativity' everyone above the mean is labeled as 'high creativity'.
2) students are allocated to a work group type based on their creativity label. Their are 3 types: only high creativity students, only low creativity students, and an even mix between high and low creativity. All students how didn't get an RAT score (i.e. who didn't participate) are randomly and evenly distributed between all groups.
Randomization is based on blocks. Because of university policy students should be assigned based on their study specialization (there are 7 specializations of different size, and roughly 40 groups per year), and based on their gender (their should be gender balance between groups within the same study specialization).
Randomization unit
The randomization is based on the individual student level.
The treatment is (in a sense) also on the individual level. Students are allocated to a high, low, or mixed creativity group, but the exposure to creative peers comes from the peer group excluding the own creativity. Therefore the distribution of creativity that a student is exposed to varies within a group for students with different creative abilities within this group.
Clusters
The experiments consists of 3 cohorts with on average around 40 work groups per cohort, hence 120 work groups.
Observations
Around 30 students are in one work group, hence the number of students is around 3,600 students.
Arms
There is roughly an equal division of group types between work groups, i.e. around 40 groups consist of high, low, and mixed creativity students.
Primary outcomes
Grade point average (GPA) of all first year courses at university.
Secondary outcomes
Number of courses completed at the end of the first year (with a passing grade) and drop out (whether student has quit the program at the end of the first year).
Cited by
No citation data available for this registration yet.