Paper Number

1296

Paper Type

Short Paper

Abstract

Polarization, the negativity and lack of constructiveness in discussions between members of opposing groups, is an important issue on online platforms. However, prior literature offers little guidance on viable depolarization strategies for discussion participants. We aim to address this research gap by investigating whether data-supported reasoning – supplementing arguments with relevant data – can depolarize discussions by providing a common ground for discourse across group demarcations. We exploit a novel opportunity to study social phenomena and conduct an experiment in which a large language model (LLM) simulates human discussions. Unexpectedly, our findings indicate that statements supported by data tend to elicit more polarized responses. Interestingly, this effect diminishes when an identity signal is provided alongside the data-supported reasoning. We discuss the theoretical implications through the lens of cognitive dissonance theory and offer practical suggestions for online platforms and their users.

Download

Share

COinS
 
Jun 14th, 12:00 AM

Depolarizing Online Discussions: The Role of Data-Supported Reasoning and Identity Signals

Polarization, the negativity and lack of constructiveness in discussions between members of opposing groups, is an important issue on online platforms. However, prior literature offers little guidance on viable depolarization strategies for discussion participants. We aim to address this research gap by investigating whether data-supported reasoning – supplementing arguments with relevant data – can depolarize discussions by providing a common ground for discourse across group demarcations. We exploit a novel opportunity to study social phenomena and conduct an experiment in which a large language model (LLM) simulates human discussions. Unexpectedly, our findings indicate that statements supported by data tend to elicit more polarized responses. Interestingly, this effect diminishes when an identity signal is provided alongside the data-supported reasoning. We discuss the theoretical implications through the lens of cognitive dissonance theory and offer practical suggestions for online platforms and their users.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.