The effect of ride experience on changing opinions toward autonomous vehicle safety

Autonomous vehicles (AVs) are a promising emerging technology that is likely to be widely deployed in the near future. People's perception on AV safety is critical to the pace and success of deploying the AV technology. Existing studies found that people's perceptions on emerging technologies might change as additional information was provided. To investigate this phenomenon in the AV technology context, this paper conducted real-world AV experiments and collected factors that may associate with people's initial opinions without any AV riding experience and opinion change after a successful AV ride. A number of ordered probit and binary probit models considering data heterogeneity were employed to estimate the impact of these factors on people's initial opinions and opinion change. The study found that people's initial opinions toward AV safety are significantly associated with people's age, personal income, monthly fuel cost, education experience, and previous AV experience. Further, the factors dominating people's opinion change after a successful AV ride include people's age, personal income, monthly fuel cost, daily commute time, driving alone indicator, willingness to pay for AV technology, and previous AV experience. These results provide important references for future implementations of the AV technology. Additionally, based on the inconsistent effects for variables across different models, suggestions for future transportation survey designs are provided.

Automated summary

The study explores the impact of various factors on people's initial opinions and opinion change towards AV safety, including age, personal income, monthly fuel cost, education experience, daily commute time, driving alone indicator, willingness to pay for AV technology, and previous AV experience. These factors provide important references for future implementations of the AV technology, as well as suggestions for future transportation survey designs based on inconsistent effects across different models.