Referrals to motivate the take-up of talent recognition programs

Participants in the referral experiment are drawn from the population of 2,459 students who are eligible to participate in the Academic Excellence Awards selection process. Eligibility requires a GPA of 4 or higher for the awards. In a first phase, these students are invited to complete an incentivized online survey, at the end of which they can nominate up to four peers to be invited to apply for the awards.

The experimental sample consists of students who were nominated by a peer and who meet the eligibility criteria for participation in the award selection process. Students who were also nominated through a separate faculty referral process are excluded from this randomized sample. Eligible nominated students are randomly assigned to one of two groups.

Students in the treatment group (T1) receive an invitation email from the university to participate in the award selection process and are informed that a peer recommended them. For students nominated by more than one peer, the invitation identifies one referring peer. Priority is given to the peer who nominated the student earlier in the referral sequence; ties are broken at random.

Students in the nominated control group (T2) receive the same invitation to participate in the award selection process, but the email does not mention that a peer nominated them. In addition, eligible students who were not nominated by any peer also receive the standard invitation email from the university to participate in the award selection process.

Registration for the award selection process is completed online and coincides with the first assessment module. Registration remains open for one week. The selection process consists of three online assessment modules. Access to each subsequent module depends on completing the previous one, and students have one week to complete each step.

2026-09-07

2026-10-05

The primary outcomes of interest are:

- Take-up of the award selection process, defined as completing the first step of the award selection process during the enrollment window.

- Completion of the award selection process, defined as completing all three online assessment modules by the final deadline for the selection process, unconditional on take-up.

Take-up is a binary indicator equal to one if a student completes the first step of the awards selection process within the one-week enrollment window, and zero otherwise. This step consists of an online survey that includes the first assessment module.

Completion is a binary indicator equal to one if a student completes all three assessment modules by the end of the four-week selection period, and zero otherwise.

Completion conditional on take-up, restricting the sample to students who completed the first step of the awards selection process, which includes the first assessment module.

Perceived reasons for being invited to participate, measured using an open-ended question asked during the first step about why the student believes they were invited to apply.

Perceived academic ability and expected performance, measured during the first step using students’ reported expected percentile in the GPA distribution of the student population and their expected relative ranking in the awards selection process, as well as expected relative performance in the assessment tasks.

Composition of registrants and completers in terms of prior academic achievement, measured using administrative GPA.

Assessment performance of registrants and completers, measured using scores on the assessment modules.

We will code the open-ended responses on why students believe they were invited into pre-specified categories that capture whether respondents mention academic merit, peer recognition or recommendation, or broader non-academic skills or qualities. We will also construct a measure of the number of distinct broad dimensions mentioned in the response. Text coding will be conducted blind to treatment assignment. These measures are intended to capture whether referrals affect students’ beliefs about the basis on which they were invited to participate.

Perceived academic ability will be measured as the student’s expected own percentile in the GPA distribution of the student population. Expected task performance will be measured using students’ reported expected ranking relative to a comparison group of students completing the same task. These measures are intended to capture whether referrals affect confidence or perceived fit with the opportunity.

Administrative GPA will be used as a pre-treatment measure of prior academic achievement. Because GPA is fixed before treatment, it will be used to characterize the composition of students who register for, or complete, the selection process in each experimental arm. Scores on the assessment modules will be used in the same spirit to study whether the students induced to participate by referrals appear weaker, stronger, or similar in measured performance to students in the control group.

This study examines whether learning about a peer referral affects take-up of a talent recognition program. The study population consists of students at a private university in Colombia who are eligible to participate in the Academic Excellence Awards selection process. Eligibility requires a GPA of 4.0 or higher on the university’s 5-point grading scale.

In a first phase, all eligible students are invited to complete a short incentivized online survey for high-achieving students. The survey collects baseline information on socio-demographic characteristics and selected attitudes and preferences. Respondents are then introduced to the Academic Excellence Awards and informed that, although eligibility is based on GPA, final selection depends on performance in a broader multi-step assessment process. At the end of the survey, respondents can nominate up to four peers whom they believe should be invited to participate in the awards. For the first nominated peer, respondents are also asked to explain why they made that nomination.

The randomized experimental sample consists of students who were nominated by a peer, satisfy the GPA eligibility criterion, were not nominated through a separate faculty referral process, and are still enrolled at the time of implementation (September 2026). These students are individually randomized into one of two groups.

Students in the treatment group (T1) receive an invitation email from the university to participate in the award selection process and are informed that a peer recommended them. Students in the nominated control group (T2) receive the same invitation, but the email does not mention the peer referral. For students nominated by more than one peer, the treatment email identifies one referring peer. Priority is given to the peer who nominated the student earlier in the referral sequence; ties are broken at random.

Randomization is stratified by gender, participation in the first edition of the awards, participation in the first-phase survey, receipt of one versus multiple referrals, and selection first versus later by at least one referring peer. In addition, eligible students who were not nominated by any peer also receive the standard invitation email to participate in the award selection process.

Registration for the award selection process is completed online and coincides with the first assessment module, as part of the same survey instrument. Registration and the first module are open for one week. Students who complete the first module receive access to the second module; access to each subsequent module is contingent on completing the previous one. The selection process consists of three online modules designed to assess a range of skills beyond academic performance, including cognitive, social, and associative reasoning skills. Each module is open for one week.

Our primary comparison is between T1 and T2. This contrast identifies whether learning about a peer referral affects registration for and completion of the award selection process, holding the initial selection of nominated students fixed. We also use survey and administrative data to study whether referrals affect beliefs about why students were invited, confidence about expected performance, and the composition of registrants and completers.

Not available

Randomization done by statistical software.

Individual

No

N/A

N = 977 (number of eligible students nominated in the first-phase survey and enrolled in September 2026)

Of the N = 977 students nominated by a seed (and not nominated by a faculty member), N_T1 = 492 were randomly assigned to the treatment group (i.e., informed about the peer referrals), and N_T2 = 485 to the “nominated” control group (i.e., no information).