Investigating the Determinants of Driver Trust in Semi-Autonomous Vehicle Systems: A Multidimensional Analysis

This study investigates the determinants of driver trust in semi-autonomous vehicles through an integrated theoretical framework. Building upon established automation trust models and empirical observations of current system deployments, we identify eight critical antecedents: system predictability, automation reliability, interface transparency, human-machine interaction design, anthropomorphism, contextual driving scenarios, inherent user traits, and information accuracy. A structural equation model (SEM) was developed to test 12 hypothesized relationships. Methodologically, we adapted measurement scales from prior literature through iterative pilot testing (N=45), culminating in a validated 28-item questionnaire. Empirical data from 231 valid responses (total collection: 276) were analyzed using SmartPLS 3.0, incorporating: 1) confirmatory factor analysis assessing convergent validity (AVE=0.61-0.79), discriminant validity (HTMT<0.85), and composite reliability (CR=0.83-0.91); 2) bootstrapped mediation analysis (5,000 subsamples); and 3) path coefficient evaluation (β=0.18-0.43, p<0.05). Results reveal that seven determinants significantly predict trust formation, with system transparency demonstrating the strongest total effect (β=0.41, p<0.001). Theoretical implications for human-automation collaboration and practical guidelines for ADAS interface optimization are discussed.