completed
ICFES Nudge to Search — Colombia
Pop-up messages on the ICFES Saber 11 results platform linked students to a purpose-built information portal, built with Banco de la República, centralizing earnings by degree level for the nation and each department, employment rates, costs, and financial-aid eligibility. Two nationwide randomized experiments tested which messages move information search. The precursor to IcfesBot — no chatbot was involved.
Target population
All Saber 11 takers — the fall 2017 national cohort and the spring 2018 cohort
Key outcomes
- Pop-ups moved attention at national scale: 15.2% of treated students clicked through in 2017, 24.6% in 2018
- A plain invitation to search (20.9%) beat every belief-targeting message variant (13.1-16.7%) by 4-8 percentage points
- Loss-framed messages did worse than gain-framed equivalents (13.1% vs 16.7%), contrary to prospect-theory predictions
- Enrollment did not move: average effects above 2.6 percentage points are ruled out
- The experimental nulls motivated the personalized WhatsApp guidance that became IcfesBot
Team
- Christopher Neilson — Principal Investigator
- Eric Bettinger — Co-author
- Michael Kremer — Co-author
- Felipe Lizarazo — Co-author
- Christian Posso — Co-author
- Juan Saavedra — Co-author
