In this project, we will implement an experiment on a sample of around 3000 middle school students (11 to 14 years old) enrolled across 14 schools in 7 Italian cities. The experiment is designed to evaluate the cost of gender stereotypes, including self-stereotypes of students, stereotypes induced by peers' and parents' pressure. Students will perform two tasks: the first task involves math multiple-choice questions and the second task literature multiple-choice questions. For each correct question they will get one point and the student with the highest score in the class will get a 10 euro gift card. Before performing the two tasks, each student can choose one task to get double points and therefore increase their likelihood of winning. We will randomly assign the type of information given to students when making this choice to evaluate if students are more likely to choose the gender conforming field under different circumstances. The information shared is designed to create variation in students' beliefs on who will have knowledge of their decisions and on parents' recommendations/ beliefs. The final goal is to evaluate whether there are significant gender differences on the efficiency of this decision making process and how these differences vary considering different information frameworks.
Choice of the subject for children's experiment: math or literature.
Choice of high-school track for parents' experiment.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
CHILDREN EXPERIMENT: We will randomly assign students at individual level to the six different interventions. The enumerator will provide the same information to all students. However, while completing the survey in the tablets, students will receive the different information for the different treatment groups. With this experiment, we want to understand, first, whether girls (boys) are systematically more likely to pick the gender conforming tasks compared to what would be efficient given their performance in both tasks in the control condition. If so, we would provide evidence of self-stereotypes of students. Second, we want to understand whether social image concerns may push students to behave in a direction closer to the one predicted by gender stereotypes. Third, we want to understand whether girls avoid specific fields, such as math, because they do not want to be the only girls in the group interaction and whether boys boys avoid humanistic field because they don't want to be the only boy in the group interaction. Fourth, we want to see whether prompting children to think about the recommendation from their parents induces them to choose more or less gender stereotypical than children's choice in the control group. Finally, we will study whether informing children that their parents will observe their choice affects their decision.
Experimental Design Details
Randomization Method
Randomization done with Qualtrics
Randomization Unit
The unit of randomization is the student.
Was the treatment clustered?
No
Sample size: planned number of clusters
3000 pupils (15 schools)
Sample size: planned number of observations
3000 pupils
Sample size (or number of clusters) by treatment arms
500 students
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)