The Impact of a Peer-to-Peer Mentoring Program on University Choices and Performance

We design an intervention facilitating encounters between high school students and successful university students enrolled in a quantitative
field and volunteering as mentors. The one-to-one online meetings were designed to be open-ended, allowing mentors to encourage mentees to pose questions. This approach aimed to personalize the mentorship experience, ensuring it met the individual needs of the
students. Mentor-mentee pairs were encouraged to met two or three times in the months leading up to the selection of a university degree program.

We designed the intervention to study the effect of mentorship on university choices, particularly the choice of the field of study, among last year high school students. We matched motivated and successful undergraduate and graduate students from the host university (mentors) with high school students (mentees) for one-on-one online orientation mentorship sessions. High school students had the opportunity to ask questions during the sessions, and the interactions were unrestricted to ensure that the mentorship was tailored to the needs of the mentees. While we encouraged mentors to prompt their mentees to ask questions, we also provided university students with guidance to facilitate discussions and cover a wide range of topics. Meetings were scheduled via a dedicated platform, where mentors could indicate their availability and mentees could book meetings. One-to-one meetings were mostly conducted via the MS-Teams platform, and we encouraged participants to use only official channels (the dedicated platform, institutional email, and MS-Teams), especially for the first meeting. Each mentor-mentee pair was encouraged to schedule 2 or 3 half-hour meetings.

2022-01-08

2022-09-15

Dummy variable taking value 1 if students choose the same field of study of the matched mentor

We mostly rely on answers to the endline survey. For a subsample, data are validated using administrative data on university performance from the host institution.

Enrollment in STEM/Econ fields; Prospective wages; Academic performance (grades and study credits) by the end of the first year

Enrollment in STEM/Econ fields: variable based on answers to endline questionnaires, pooling together quantitative fields.
Prospective wages: based on AlmaLaurea data.
Academic performance: based on administrative data from the host institution (only for the subsample available in the data)

The mentoring program was mostly advertised during large online orientation events organized by the host institution -- a large public university located in Italy -- with a participation of roughly 20,000 students. To enroll in the mentoring program, students had to complete a baseline questionnaire where we collected background information and their ranked list of fields of study they were interested in. We used this information to match each applicant with a mentor; randomization into Control and Treatment group relied on program oversubscription, and only students in the Treatment group were informed about the assignment. A few months after the intervention, and right before the start of the academic year, we run the endline survey where we collected the enrollment choices of our participants. We complement and validate these self-reported measures with administrative data about student’s enrollment and performance from the host university

High school students applied to the program between January and February 2022, a few months prior to the closure of the first intake. The final intake available for enrollment in a degree program was in September 2022. Personalized meetings were conducted between March and July 2022, with a higher frequency of meetings occurring in the initial months of the intervention while high school was still in session. We contacted all students for an endline survey just before the start of the academic year 2022/23. To account for the medium-term effects of the intervention, we collected administrative data regarding credits and grades at the end of the first year of university (December 2023). These data are only available for students who enrolled at the host university.

Mentors were recruited from second and third-year bachelor’s students and first and second-year master’s students across nine fields: Accounting Business and Management; Agricultural and Food Sciences; Architecture and Industrial Design; Biology and Environmental Sciences; Chemistry Physics and Mathematics; Computer Sciences; Economics and Finance; Engineering; and Statistics. We specifically targeted quantitative fields which are known for having higher returns in the labour market.

Baseline and endline survey:
• Background information: We collected information on gender, year of birth, education level of parents, type of high school, county of the high school, mathematics and Italian grades (in the previous school year).
• Ranking of fields and degree programs: Each prospective participant had to choose two or three fields of interest and rank them from the most preferred to the least preferred. For each chosen field, they could select up to 4 specific degree programs. The list of courses was based on the degrees available at the host university. After selecting all programs of interest, they were asked to rank them from the most to the least preferred.
The survey included also questions about the motives that might drive the choice of a university program and their subjective expectations similar to Boneva and Rauh (2017).

The treatment assigned is stratified at the mentor level using a serial dictatorship mechanism to form mentor-mentee pairs. First, all eligible students were matched with a mentor. Subsequently, students matched with a particular mentor were randomly assigned to either the Treatment or the Control group. Treated students were introduced to their mentors, while control students did not receive any communication about the matching procedure.

Randomization done in office by a computer

Randomization at the individual level, stratified by matched mentor

No

The randomization is stratified at the mentor level. Number of mentors: 82

Number of students: 337

168 students assigned to the Control group and 169 assigned to the Treatment group.

Considering the full sample, the MDE for the primary outcome is 0.18 standard deviation (approximately 9 percentage points) Considering the final sample of respondents, the MDE is 0.28 (approximately 14 percentage points)