Return
Blame attribution mechanisms in AV accidents: A multi-dimensional exploration integrating social media and survey data
H
Y
DOI:10.1016/j.trf.2026.103601.png)
Abstract
En 中文
Before the full autonomous vehicle (AV) era, we are in an extraordinary era of shared control between humans and AV systems. In the current era, AV accidents are increasingly inevitable, and the occurrence raises complex ethical and legal challenges, resulting in the difficulty in determine the responsibility. Previous research primarily employed qualitative or quantitative analyses based on hypothetical scenarios to identify public blame targets in AV accidents and reached conflicting conclusions. Importantly, the research merely confirmed the existence of attribution bias, it has not sufficiently explained the underlying mechanisms. By integrating multiple analytical methods, this study developed a comprehensive attribution theory framework to explain the reasons for attribution bias in AV accidents. The research collected 90,426 valid comments from the Chinese social media platforms about AV accidents from April 1, 2021, to January 20, 2025. The three dimensions of concern topic, comment volume, and sentimental polarity identified the AV system as the main blame target. Based on the qualitative analysis results, Heider's naive attribution theory was adopted as the foundational theoretical framework to construct a comprehensive model of responsibility attribution in AV accidents. We found that trust and affective tagging are core factors in blame attribution. There is a gender tradeoff between knowledge and sentiment in AV accidents. Additionally, media exposure revealed a selective activation mechanism of media effects in the death scenario among males.
Keywords:
Blame attribution
Accident
Gender
Structural equation model
Latent Dirichlet allocation
Journal
IF:
4.4
Papers:
3.3K
Citations:
1.3W
