Promote Equity in Selection for a Talent Recognition Program

This RCT tests two versions of a brief message shown to faculty immediately before they begin nominating students for the Talent Recognition Program. Faculty who volunteer to make nominations are randomly assigned to see one of two prompts:
1. Control (Standard Nomination): A simple request to nominate students taking into account their academic performance.
2. Inclusion Prompt: The standard request plus an explicit reminder to actively consider all students they know, particularly those from adverse socioeconomic backgrounds, to ensure no deserving student is overlooked.

2026-03-16

2026-06-30

I will test differences between control and salience:
1. Diversity of nominations: share of low class or female students nominated
2. Quality of nominations: average GPA; also willingness to accept nomination

1. Beliefs on student performance and strategies (this is a pre-treatment measurement): for different groups, share of students with high GPA, share of students who seek professional guidance from faculty, share of students who go beyond grades
2. Reasons for nomination (post treatment): explanation of why they nominated the student they did

I will invite all faculty at UNAB to participate in an incentivized online survey.
- Pre-treatment: Faculty answer demographic questions and elicitations of their beliefs about UNAB student behaviors.
- Evaluation: Faculty view a promotional video for the 2nd edition of the talent recognition program and evaluate its perceived value.
- Intervention & Outcomes: Those who are willing to make nominations are randomly assigned to either the Control or the Salience condition. They can then nominate students to be invited to the talent recognition program. For making high quality nominations they enter a lottery for monetary bonuses..
By comparing nomination patterns between the two arms, I seek to identify whether an salience prompt reminder about socio-economic adversity mitigates biases and improves the diversity of the nominee pool.

Not available

Randomization is conducted via a computer randomizer embedded in the Qualtrics survey flow. I will block randomize/stratify based on variables elicited earlier in the survey. Blocking variables will include: gender, whether they nominated a student in the previous 2025 edition (elicited in Part 3 of the survey), and social class.

Individual faculty member.

No

701 faculty members will be invited

about 50% of the invited sample: 350 faculty members

Expected equal allocation between the two treatments: ~175 in Control, ~175 in the Salience (Treatment) prompt.

- Unit: The unit of randomization and analysis is the individual faculty member. Because the intervention is assigned at the individual level, there is no clustering in the experimental design (design effect = 1). - Standard Deviation: The expected baseline mean for the share of low-SES students nominated is 0.25 (25%), which corresponds to an expected baseline standard deviation of 0.433 (calculated as the standard deviation of a proportion: $\sqrt{0.25 \times 0.75}$). - Percentage: Assuming a balanced total sample size of 350 faculty members (175 per arm), 80% statistical power, and a 5% significance level for a two-sided test, the minimum detectable effect size (MDES) is approximately 0.30 standard deviations. Translated into absolute percentage points (accounting for the variance of two independent proportions), the study is adequately powered to detect an absolute difference of 13.8 percentage points between the control and treatment conditions. This means the study can reliably detect an increase in the share of low-SES students nominated from the expected baseline of 25.0% in the control group to 38.8% in the treatment group.