Bogotá Elite Opinions

Eligible students are invited to paid research-assistant work observing urban spaces and infrastructure. The treatment assignment involves experiencing public goods, including travel by public transport and visits to parks and other public spaces in southern Bogotá. These spaces and neighborhoods are intended to be less familiar to the participants and are often perceived as unsafe. The active control involves fieldwork in private shopping malls in northern Bogotá. The routes span poorer and wealthier areas of the city. Both assignments require the same types of fieldwork deliverables:
1. Visit the assigned locations during daylight, following the route and safety instructions and completing the required check-ins and photo verifications.
2. Complete a structured observation log, recording arrival and departure times and how the spaces function and are used.
3. Submit the observation log and two short texts: a reflection on the visit and a recommendation describing the places to a friend or relative.
4. Complete the endline survey.

2026-09-09

2026-10-11

Two standardized primary outcome indices, comprised of component indices:
1. Safety and exploration
1.1 Mobility and exploration index
1.2 Incentivized safety beliefs index
2. Public provision, state capacity, and taxation
2.1 Support for taxation index
2.2 Valuing public goods index
2.3 State capacity perceptions index

The safety and exploration index combines willingness to explore unfamiliar places and expressed interest in further fieldwork with incentivized safety beliefs. The public provision, state capacity, and taxation index combines support for taxation and parks donations; public-goods valuations, hypothetical spending preferences, and petition-related responses; and ordinary and incentivized assessments of government capacity. Both primary indices include non-incentivized survey questions alongside actual choices and incentivized belief elicitations. The attached question map identifies the measures in each component, as well as what survey measures were excluded from these primary components and why.

Exploratory outcomes include: a solely behavioral and incentivized version of the two primary indices (i.e. excluding non-incentivized survey opinion questions); analysis of the component indices and individual measures of the primary outcomes; and measures developed from the field assignments (e.g. text analysis of written reflections).

All secondary analyses are exploratory. They aim to understand mechanisms, assess robustness to measurement choices, identify which beliefs and choices change, and explore differences across participants.

HETEROGENEITY
Our primary planned heterogeneity dimension is baseline eliteness, using interactions with treatment assignment. As a secondary exercise, we will also conduct exploratory heterogeneity analysis using baseline measures or proxies of primary outcome components (e.g. variation with baseline safety beliefs and baseline use of public goods) to see how belief updating varies with prior beliefs.

RECRUITMENT AND ELIGIBILITY

The university advertised a paid research-assistant opportunity to Universidad de los Andes undergraduates on Aug 12, 2026. Applicants completed screening and a baseline survey before randomization. Screening verified basic student information and collected socioeconomic characteristics used to construct the recruitment sample. We received 1,838 valid screening responses and randomized 1,211 students. To fit the recruitment target, the team randomly subsampled applicants in the (middle class) socioeconomic strata 3 and 4, where applications were most numerous, and included all applicants from the highest socioeconomic strata 5 and 6. After random assignment, the team withdrew invitations from minors and prior-pilot participants (who were erroneously not screened out pre-randomization). Updated exclusions imply 1,161 eligible students: 584 treatment and 577 control.

ASSIGNMENT AND COMPARISON

We assigned 606 students to public-goods fieldwork and 605 to mall fieldwork on Sep 8, 2026. The main comparison is the effect of assignment to public-goods fieldwork relative to the active mall control, regardless of task completion.

MAIN ANALYSIS

Our target is the intention-to-treat effect of assignment among eligible students. We will estimate regressions using available outcomes, accounting for randomization strata and baseline controls selected by double LASSO. We project approximately 800 completed endlines; the observed count will depend on endline and behavioral-measure completion and may differ across outcomes. We plan paid endline follow-up for eligible students who do not complete their field tasks.

We will report standard p-values and p-values adjusted across the two primary outcome tests. For exploratory component analyses, we will report standard and adjusted p-values across the two immediate safety/exploration components, and separately across the three immediate public-provision components.

SPILLOVERS

We may explore spillovers from friends or associates in the experimental sample, using the number assigned to treatment while controlling for the number of named peers in the randomized sample and the total number named. Whether we include peer-assignment terms in the main specification will depend on network coverage, exposure variation, the implemented randomization, and opportunities for interaction. We will make that choice using network and assignment information, without examining outcome associations. The network is still being cleaned and constructed, and its suitability for this analysis remains uncertain.

PILOT DEVELOPMENT

A pilot experiment in the first half of 2026 informed the experimental design and major measurement decisions. It ran with 182 students assigned to public-goods treatment (100) or an active mall control (82). Controls visited Parque La Colina and El Retiro shopping malls; the main study replaces El Retiro with Titán Plaza. This pilot sample is entirely separate from the full experimental sample.

REGISTRATION TIMING

This registration follows random assignment and the start of intervention and endline collection. Randomization occurred on Sep 8, 2026. The team is submitting the plan after confirming launch and correct implementation of randomization. Some students completed their tasks and endlines quickly. As of September 24, 2026, 27 students (2.23% of the 1,211 randomized students) had completed endline. Researchers prepared this registration blinded to endline data.

Not available

Individual random assignment in R, stratified by parental socioeconomic stratum, public-transport use, perceived victimization risk, and pessimism about state capacity. These four characteristics defined 16 strata.

Individual Student

No

N/A

1,211 students

606 students treatment, 605 students active control