Deciphering the Impact of AI-Generated Fake Reviews on Consumer Behavior and Willingness to Pay

How to cite

WANG, Ke; ZHANG, Xu. 2024. "Deciphering the Impact of AI-Generated Fake Reviews on Consumer Behavior and Willingness to Pay." AEA RCT Registry. AEARCTR-0014491. https://doi.org/10.1257/rct.14491-1.0

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

Ke WANG* , Xu ZHANG

* Primary investigator

Registry ID
AEARCTR-0014491
DOI
https://doi.org/10.1257/rct.14491-1.0
First registered
Last updated
Status
On going
Trial dates
2024-03-01 – 2024-11-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14491

Abstract

This experimental study investigates the ability of consumers to identify AI-generated fake reviews in online marketplaces and examines the factors influencing this capability and consumers' willingness to pay. The experiment is divided into a preliminary phase and a main experimental phase. In the preliminary phase, participants are exposed to product information along with reviews categorized as either AI-generated or genuine, followed by assessments of their authenticity. The main experiment manipulates the proportion of fake reviews and the insurance variable across different scenarios to measure its effect on WTP. The outcomes are expected to provide insightful contributions to understanding the dynamics of consumer trust and decision-making in the face of evolving digital marketing strategies.

Experimental design

This experiment seeks to understand consumers' ability to detect AI-generated fake reviews and how this affects their economic behavior, specifically their willingness to pay (WTP). The study comprises a preliminary phase where participants evaluate the authenticity of reviews for different products. This phase aims to establish a baseline for consumer discernment abilities. The main experiment then introduces different proportions of fake to genuine reviews and varies the presence of return insurance to observe its impact on WTP. The experiment concludes with a cognitive reflection test to measure participants' reflective thinking capabilities, correlating these with their susceptibility to misinformation.

Design details

Randomization method
Randomization will be conducted on an online survey platform.
Randomization unit
Individual.
Clusters
Preliminary Experiment: 150 adults Main Experiment: 600-1200 adults.
Observations
Preliminary Experiment: 150 adults. Main Experiment: 600-1200 adults.
Arms
Preliminary Experiment: As a baseline setting without varying conditions, all 150 participants will undergo the same treatment, serving as a control group to establish initial discernment abilities. Main Experiment: Low Proportion of Fake Reviews, Insurance Present: 100 participants Low Proportion of Fake Reviews, Insurance Absent: 100 participants Medium Proportion of Fake Reviews, Insurance Present: 100 participants Medium Proportion of Fake Reviews, Insurance Absent: 100 participants High Proportion of Fake Reviews, Insurance Present: 100 participants High Proportion of Fake Reviews, Insurance Absent: 100 participants
Primary outcomes
Ability to Identify AI-generated Reviews, Cognitive Reflection Test (CRT) Score, Willingness to Pay (WTP)

Cited by

No citation data available for this registration yet.

Keywords