Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Students’ Choice of Major: An Information Experiment
Last registered on April 17, 2019


Trial Information
General Information
Students’ Choice of Major: An Information Experiment
Initial registration date
April 10, 2019
Last updated
April 17, 2019 8:14 PM EDT
Primary Investigator
Other Primary Investigator(s)
Additional Trial Information
On going
Start date
End date
Secondary IDs
This paper provides evidence that a lack of accurate information among undergraduate students contributes to the gender gap in student enrolment in Science, Technology, Engineering, and Mathematics (STEM) programs in universities. Specifically, undergraduates in STEM courses tend to underestimate their relative course ranking, suggesting that they are underconfident. Female students are more underconfident than male students.

Conducting a survey at a leading Canadian university, I find that majority of the undergraduate students taking a required first-year calculus course for STEM majors underestimate their relative course ranking, with female students being more likely to underestimate their ranking. This suggests that women are less confident about their relative performance. Additionally, I find that while students are aware of the existence of a wage gap between graduates in STEM and non-STEM majors, both female and male students typically overestimate this gap. In a follow-up randomized experiment, I provide treated participants with information about their relative course rankings, about the fact that female students tend to be less confident according to high-quality academic research, and about expected future incomes for STEM and non-STEM majors. I find that this information treatment led 92% of treated students to update their beliefs. Furthermore, the treated students became more likely to choose a STEM major. These effects are largely driven by female students: treated female students are 17.3% more likely to choose a STEM major and the results are statistically significant at the 1% level, whereas no statistically significant treatment effect is found on the STEM enrolment for male students.
External Link(s)
Registration Citation
Wang, Linda. 2019. "Students’ Choice of Major: An Information Experiment." AEA RCT Registry. April 17. https://doi.org/10.1257/rct.4021-1.0.
Former Citation
Wang, Linda. 2019. "Students’ Choice of Major: An Information Experiment." AEA RCT Registry. April 17. http://www.socialscienceregistry.org/trials/4021/history/45070.
Experimental Details
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
The key outcome variable of interest is the student's choice of major.
The key independent variable is whether they received our treatment email. Other independent variables include non-identifiable demographic and background controls (specifically, gender, age, admission records, GPA, full-time status, mother tongue language, major, year of study, international/domestic student) commonly used in this area of study.
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
The target population consists of first-year students who wish to study a STEM subject, which means, they take a STEM mathematics course.

First, we will send all students a survey to learn about their beliefs regarding their relative ranking in their mathematics courses and about their beliefs regarding STEM.

After that, we plan to randomly divide them into two groups: treatment and control. We will send the treatment group emails about their percentile rank in terms of their final grades, that is, based on their own academic achievement, whether they are in the top x% students (information A). The transcript shows students' final percentage mark and the letter grade of the course average. Students can learn whether they are above average or below average, but they may not know their precise relative ranking, the reason why we want to send them their precise ranking. The treatment email also contains the fact that women are less confident about their relative ranking than men from previous research (information B).

There will be three versions of treatment emails: information A only, information B only, information A+B.

After the experiment, we plan to track students’ choices of major at the beginning of their second year. As control and treatment students are chosen randomly, we would be able to test whether the information treatment is effective by comparing the proportions of students staying in STEM programs between treatment and control.
Experimental Design Details
Randomization Method
Randomization done in office by a computer using STATA
Randomization Unit
The randomization is at individual-level for each STEM mathematics course.
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
1968 students
Sample size: planned number of observations
1968 students
Sample size (or number of clusters) by treatment arms
492 students control, 492 students receive information A, 492 students receive information B, 492 students receive information A+B
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
University of Toronto Research Ethics Boards
IRB Approval Date
IRB Approval Number
Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)