The Persuasive Power of Multi-Modal AI: A Randomized Controlled Trial of Microtargeting Effectiveness

Last registered on April 17, 2025
View Trial History

Pre-Trial

Trial Information

General Information

Title
The Persuasive Power of Multi-Modal AI: A Randomized Controlled Trial of Microtargeting Effectiveness
RCT ID
AEARCTR-0015650
Initial registration date
April 07, 2025

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 17, 2025, 6:13 AM EDT

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

Locations

Country
Pakistan
Region
Punjab

Primary Investigator

Name
Ayesha Ali
Affiliation
Lahore University of Management Sciences
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Ihsan Ayyub Qazi
PI Affiliation
Lahore University of Management Sciences
Contact Investigator
PI Name
Zafar Ayyub Qazi
PI Affiliation
Lahore University of Management Sciences
Contact Investigator

Additional Trial Information

Status
In development
Start date
2024-09-19
End date
2025-07-31
Keywords
Other
Additional Keywords
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This study evaluates the persuasive effects of microtargeted persuasion campaigns generated using multi-modal large language models (LLMs) using a Randomized Controlled Trial (RCT) conducted in Lahore, Pakistan. By examining the efficacy of AI-driven personalization in a Global South context, this research aims to contribute to our understanding of emerging technological applications in political communication and public engagement strategies.
External Link(s)

Registration Citation

Citation
Ali, Ayesha, Zafar Ayyub Qazi and Ihsan Ayyub Qazi. 2025. "The Persuasive Power of Multi-Modal AI: A Randomized Controlled Trial of Microtargeting Effectiveness." AEA RCT Registry. April 17. https://doi.org/10.1257/rct.15650-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Intervention (Hidden)
The treatment group will receive microtargeted AI-generated videos on a set of three contextually relevant socio-political issues. The control group will be shown unrelated placebo videos.
Intervention Start Date
2025-04-14
Intervention End Date
2025-04-30

Primary Outcomes

Primary Outcomes (end points)
Persuasion effects
Primary Outcomes (explanation)
We will obtain baseline and endline ratings on three issues, on a scale with three categories: Agree (+1 to +5), Disagree (-1 to -5), or Neutral (0).

Shift Score Calculation: To measure potential treatment effects, we will calculate the difference between baseline and endline ratings. We define:
Positive shift scores: represent movement toward moderation or opinion change, occurring when:
(1) A participant becomes less extreme while maintaining their position (e.g., +5 to +2)
(2) A participant changes stance completely (e.g., from agree to disagree)
(3) A participant moves between a neutral position and either agreement or disagreement
These positive shifts reflect successful persuasion effects in response to the treatment.

Negative shift scores, occurs when a participant becomes more extreme in their rating (showing increased certainty). These negative shifts indicate backfire effects, where the treatment produces the opposite of the intended outcome.

To illustrate, consider the following examples:
(1) Moving from agree (+3) to disagree (-4): +7 shift score (persuasion effect)
(2) Moving from neutral (0) to agree (+2): +2 shift score (persuasion effect)
(3) Moving from agree (+2) to neutral (0): +2 shift score (persuasion effect)
(4) Moving from agree (+1) to strongly agree (+5): -4 shift score (backfire effect)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We will conduct a stratified RCT (using high and low digital literacy samples), where each participant in the strata will be assigned to either the control or treatment group with equal probability. For each of our three topics, both the treatment and control groups will be shown three videos in a randomized order. The treatment group will receive microtargeted AI-generated videos designed to persuade the participants in a particular direction. The control group will be shown unrelated placebo videos. We will measure participants view on topics issue pre and post treatment/placebo implementation to capture any persuasion effects.
Experimental Design Details
Randomization Method
done in office by a computer
Randomization Unit
individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
200
Sample size: planned number of observations
200
Sample size (or number of clusters) by treatment arms
100
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Lahore University Management Sciences
IRB Approval Date
2024-09-19
IRB Approval Number
LUMS-IRB/07182023/IAQ-FWA-00030383
Analysis Plan

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

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials