Understanding reporting behavior on gang extortion: A survey experiment in Medellín

How to cite

Tobon, Santiago; Blattman, Christopher; Mesa-Mejia, Juan-Pablo. 2024. "Understanding reporting behavior on gang extortion: A survey experiment in Medellín." AEA RCT Registry. AEARCTR-0014803. https://doi.org/10.1257/rct.14803-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Santiago Tobon* , Christopher Blattman , Juan-Pablo Mesa-Mejia

* Primary investigator

Registry ID
AEARCTR-0014803
DOI
https://doi.org/10.1257/rct.14803-1.0
First registered
Last updated
Status
On going
Trial dates
2024-09-28 – 2024-12-31
Countries
Private
JEL codes
E26, K42
Registry record
http://www.socialscienceregistry.org/trials/14803

Abstract

This experiment is part of a yearly survey conducted in Medellin asking its citizens questions about the governance, legitimacy, and taxation of three main actors: gangs, the municipal government, and the police. As part of the survey, we are interested in knowing how often and how much gangs extort (tax) households and businesses. We are unsure whether subjects would be willing to report extortion to the authorities. In addition to collecting self-reported willingness and actual responses to a WhatsApp inquiry, we conduct survey experiments where individuals are randomly assigned to one of the three treatment groups that might increase likelihood pf reporting: a video that increases the salience of the issue to the government (and their likelihood of acting on the information); a video that articulates the size and social cost of the problem; and a control group.

Intervention

This study aims to examine the willingness of individuals to exert costly effort in reporting extortion to the authorities. Participants are divided into three distinct groups—two treatment groups and one control group—to assess the impact of varying informational interventions on their reporting behavior.

Group assignments and treatment conditions:

Control group: Participants in the control group receive no informational intervention.

Treatment group 1: Participants in the first treatment group view a video featuring the Secretary of Security. In this video, the Secretary explicitly expresses concern about extortion, emphasizes the government's commitment to addressing it, and directly encourages citizens to report instances of extortion.

Treatment group 2: Participants in the second treatment group view a video presenting quantitative data on the economic scope of extortion in the area. This video details the estimated amount of money extorted weekly, monthly, and yearly by local gangs, thus illustrating the scale and economic impact of extortion.

Experimental design

We expect our primary outcome regression to take the following form:

Y = b1 T1 + b2 T2 + b3 X + C + e

where
Y is our primary outcome
T1 is assignment to the first information treatment (commitment)
T2 is assignment to the second information treatment (fairness and magnitude)
X is a set of pre-information experiment control variables taken from survey data, selected using the double lasso method on primary outcomes
C is a set of comuna or barrio fixed effects
e is an error term (not clustered)

Design details

Randomization method
Randomization was done by the tablet of each surveyed individual
Randomization unit
Survey respondent
Clusters
Roughly 8,000 survey respondents (treatment is not clustered)
Observations
Roughly 8,000 survey respondents
Arms
One-third of the entire sample size
Primary outcomes
Our primary outcome is an index that measures respondents’ willingness to take costly actions in reporting extortion.
Secondary outcomes
We will analyze the subcomponents of the primary outcome as secondary outcomes. In addition, we will analyze answers to a survey question asking about willingness to stop paying the gang and a WhatsApp survey question asking whether the gang has collected extortion in the household or business (a measure of reporting to an external party). If available, we will analyze actual extortion reports (matched with respondents via exact coordinates).

Cited by

No citation data available for this registration yet.

Keywords