Behavioral Game Intervention for Reducing Bullying and Discrimination in La Paz, Bolivia

Last registered on April 24, 2026

View Trial History

Pre-Trial

Trial Information

General Information

Title

Behavioral Game Intervention for Reducing Bullying and Discrimination in La Paz, Bolivia

RCT ID

AEARCTR-0017837

Initial registration date

April 20, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

April 24, 2026, 8:58 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name

Lorena Heller

Affiliation

Universidad Privada Boliviana

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Joaquin Morales

PI Affiliation

Universidad Privada Boliviana

Contact Investigator

Additional Trial Information

Status

In development

Start date

2026-02-18

End date

2026-08-31

Keywords

Behavior, Education, Gender

Additional Keywords

Bullying, Bystander behavior, Social norms, Prosocial behavior

JEL code(s)

C93, D91, Z13, D71, J16, J15

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

School violence, discrimination, and gender-based bullying remain persistent challenges in Bolivia despite progressive legal frameworks intended to protect children and adolescents. Recent research conceptualizes bullying not as a dyadic conflict but as a collective social process in which bystanders play a decisive role in either sustaining or interrupting aggression. This study evaluates a behavioral intervention designed to activate prosocial bystander behavior by reshaping the social meaning and incentives associated with aggression and defense.

The intervention employs game-based mechanisms to simulate peer interactions in a controlled environment, using visual, emotional, and strategic nudges to alter how social status is earned within the group. Rather than directly discouraging aggression, the games make defensive and empathetic actions visible, rewarded, and socially salient, thereby shifting perceived norms around acceptable behavior. Two thematic treatment arms are implemented. One card-based game addressing racism and classism, assigns players rotating roles of aggressor, target, and bystander, allowing them to accumulate points through either empathetic defense or reinforcement of bullying. Another focused on gender-based violence, is a social deduction game with hidden roles in which defending vulnerable players entails personal costs but confers social recognition.

The study hypothesizes that norm change operates through multiple mechanisms: social learning, as students observe peers engaging in defense; norm reinforcement, as prosocial behavior becomes collectively validated; and belief updating, as defending is reframed as effective and socially valued. To mitigate social desirability bias, the study combines self-reported survey measures with incentivized behavioral outcomes derived from in-game decisions that reflect real-time trade-offs under peer observation. These behavioral measures capture latent preferences and responses that are not directly observable through attitudinal surveys alone. Together, this design allows for a rigorous assessment of both the direct impact of the intervention and its capacity to generate broader social contagion within schools.

External Link(s)

Registration Citation

Citation

Heller, Lorena and Joaquin Morales. 2026. "Behavioral Game Intervention for Reducing Bullying and Discrimination in La Paz, Bolivia." AEA RCT Registry. April 24. https://doi.org/10.1257/rct.17837-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

This study evaluates a school-based behavioral intervention aimed at reducing bullying, racism, classism, and gender-based violence among secondary school students in La Paz, Bolivia. The intervention leverages game-based experiential learning combined with behavioral nudges to shift peer norms and promote prosocial bystander behavior.
The intervention consists of two treatment arms and one control condition:
* Control group: Students play a non-interactive, chance-based game designed to preserve engagement while minimizing social interaction and behavioral influence.
* Treatment 1 (Racism and Classism): Students participate in a card-based game simulating bullying dynamics. Players assume roles and make real-time decisions to attack, defend, or reinforce aggression. The game incorporates empathy and prosocial action cards, and rewards defending behavior through point accumulation.
* Treatment 2 (Gender-Based Violence): Students engage in a role-playing game with hidden identities (bullies, victims, bystanders). The game introduces strategic and moral trade-offs, where bystanders decide whether to intervene at a personal cost under conditions of uncertainty and social pressure.
Across both treatments, the key innovation lies in the use of visual and emotional nudges embedded in game materials. These nudges aim to increase empathy and perspective-taking; make prosocial behavior visible and socially rewarded; shift the perceived social payoff from aggression to cooperation.
The intervention targets bystander behavior as the primary mechanism of change, based on the premise that peer responses determine whether bullying escalates or stops. By altering the social incentives and normative expectations within peer groups, the intervention seeks to move classrooms from antisocial to prosocial equilibria.

Intervention Start Date

2026-03-23

Intervention End Date

2026-06-30

Primary Outcomes

Primary Outcomes (end points)

Prosociality Index
Tolerance to Injustice
Social Distance (Racism and Classism): Measured using an adapted Bogardus Social Distance Scale (BSDS). The outcome is the average social distance score toward different profiles (range: 1–5), where lower values indicate higher prejudice and higher values indicate greater acceptance.
Gender Norms and GBV Attitudes: Measured using an adapted Gender Role Attitudes Scale (GRAS). Outcomes include composite indices and sub-dimensions capturing:

Acceptance of gender equality (egalitarian attitudes)
Endorsement of sexist beliefs and norms that justify gender-based violence

Primary Outcomes (explanation)

Both the Prosociality Index and Tolerance to Injustice (Justice Enforcement) are derived from incentivized allocation decisions in a modified Third-Party Punishment (Dictator) Game. This design allows us to observe costly prosocial behavior in a controlled setting, minimizing social desirability bias.

Game Structure:

Each participant is assigned the role of a third-party decision-maker (observer) in a three-player setting:

An Actor is endowed with 10 tokens and instructed to make an allocation decision toward a passive Recipient. The allocation is pre-determined to be highly unequal (i.e., the Actor keeps most tokens), thereby creating an unfair outcome.
The Participant (third party) observes this allocation and is then endowed with 5 tokens.
The participant is asked to choose how to use their tokens under two possible actions:
Keep tokens: The participant retains their full endowment, accepting the unfair outcome.
Punish the Actor: The participant can spend tokens to reduce the Actor’s payoff. Each token spent by the participant reduces the Actor’s payoff by a fixed multiple (e.g., 1 token spent reduces 3 tokens from the Actor).
This decision is made privately and has real payoff consequences within the experimental setting.
Tolerance to Injustice (Justice Enforcement):
Defined as the number of tokens the participant chooses to spend on punishing the unfair Actor (range: 0–5).
Higher values indicate lower tolerance to unfairness and a greater willingness to engage in costly punishment (active bystander behavior).
A value of zero indicates full tolerance of the unfair allocation.
Prosociality Index:
Constructed from the same decision as the willingness to sacrifice personal resources to improve fairness outcomes. Specifically, it reflects the extent to which participants incur a personal cost to reduce inequality.

Higher values indicate stronger prosocial preferences and fairness concerns.
Lower values (or zero) indicate self-regarding preferences and acceptance of inequity.

Secondary Outcomes

Secondary Outcomes (end points)

In addition to experimental outcomes, the study includes a self-reported measure of overall exposure to bullying dynamics, collected through student surveys at baseline and endline. The measure reflects the student’s overall involvement in or exposure to bullying-related behaviors within the school environment.
Persistence Outcomes

All primary behavioral and attitudinal outcomes are measured both immediately after the intervention and at a one-month follow-up, allowing assessment of the persistence of treatment effects.

Secondary Outcomes (explanation)

Students report the frequency of a range of bullying-related behaviors over a defined recall period. These behaviors include:

Physical aggression (e.g., hitting, pushing)
Verbal harassment (e.g., insults, teasing)
Social exclusion (e.g., ignoring, spreading rumors)
Discriminatory behaviors (e.g., based on ethnicity, socioeconomic status, or gender)
Gender-based harassment (e.g., sexual comments, controlling behaviors)

Responses are recorded using ordered categorical frequency scales (e.g., “never,” “once,” “a few times,” “many times”).

Experimental Design

The study employs a cluster-randomized controlled trial with two-stage randomization. In the first stage, secondary schools in La Paz are randomly assigned to either treatment or control conditions. In the second stage, within treated schools, classrooms are randomly assigned to either receive the intervention (direct exposure) or not (indirect exposure), enabling the identification of spillover effects.

The intervention consists of game-based behavioral activities designed to influence social interactions and decision-making among students. Two treatment arms are implemented, each using structured games that simulate peer dynamics related to cooperation, conflict, and social behavior. The control group participates in a neutral, non-interactive activity designed to maintain engagement without inducing social interaction effects.

Outcomes are measured using a combination of behavioral tasks and student surveys, administered immediately after the intervention and at a one-month follow-up.

Experimental Design Details

Not available

Randomization Method

Randomization was conducted in-office using a computer-based random number generator. Schools were first randomly assigned to treatment or control conditions. Within treated schools, classrooms were then randomly assigned to either direct treatment or spillover groups. Randomization was implemented prior to fieldwork using reproducible code to ensure transparency and replicability.

Randomization Unit

Randomization was conducted at two levels:

First stage: The unit of randomization is the school, which is assigned to either treatment or control.
Second stage: Within treated schools, the unit of randomization is the classroom, which is assigned to either direct treatment (participation in the intervention) or spillover (no direct participation).

The primary unit of analysis is the individual student, with clustering at the school level for inference.

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

40 Schools, 4 courses per school and in average 25 students per classroom.

Sample size: planned number of observations

The planned number of observations is 4,000 observations (students).

Sample size (or number of clusters) by treatment arms

40 schools × 4 classrooms per school × 25 students per classroom = 4,000 students

So, the planned sample size is 4,000 observations (students).

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

The study is powered to detect a minimum effect size of 0.30 standard deviations in the main outcomes. Unit: Individual student-level outcomes Effect size: 0.30 standard deviations Percentage interpretation: Equivalent to detecting a change of approximately 30% of a standard deviation This calculation assumes: Power: 80% Significance level (α): 5% Design: Cluster-randomized with two levels (school and classroom) Intra-cluster correlation (ICC): 0.05 at the school level and 0.10 at the classroom level Average cluster size: 30 students per classroom Simulation results indicate that with the planned sample (e.g., ~40 schools with multiple classrooms per school), the design achieves approximately 80–90% power to detect this effect size.

Supporting Documents and Materials