Psychological Structure of Human Trust towards Autonomous Vehicles - The Relations of Interpersonal, System Feature, Risk Perception and Behavioral Intention with Trust

Abstract

The objectives of study are to confirm the effect of the latent variables on human trust, which has been revealed in previous studies and to identify a psychological structure of the variables towards autonomous vehicles. A total of 114 queries were prepared for a survey study. The queries were collected primarily from previous studies trying to identify the latent variables and underlying factors of human trust in autonomous vehicles. A total of 195 survey responses were collected through an online survey. A series of statistical analyses, including correlation analysis and factor analysis, revealed that 51 queries in the entire query set are significantly related and affective to human trust. In order to examine structures of the factors and variables, a structural equation model was developed using the 51 variables. The modeling results revealed 5 higher-level constructs in the psychological structure, including "Interpersonal", "System Feature," "Risk Perception," "Behavioral Intention" and "Trust"". Each construct includes sub factors consisting of associated query variables. The construct of "Interpersonal" represents human intrinsic traits such as attitude, expectation and personality. The "System Feature" refers to the knowledge or thoughts gained from recognition of a particular technology or systems. The modeling outcomes confirmed that: the "Interpersonal" is the most affective to "Behavioral Intention" and "Trust" among other construct; the 'System Feature'' is also affective but one of its subfactor, "brand of vehicle", is not significantly effective in the model; and "Risk Perception" is negatively related with the "Behavioral Intention" and "Trust" as well as other constructs. In general, the results may imply that understanding the user's interpersonal characteristics is important to improve the level of trust of autonomous vehicles. However, while previous studies demonstrated the direct effect of each variable on human trust, this study revealed a comprehensive psychological structure of human trust towards autonomous vehicles by categorizing variables, factors, and constructs in hierarchical manner. Consequently, it is expected that the model could be used for understanding, predicting, and improving user trust towards autonomous vehicles as well as other automated systems.