Are we ready to embrace connected and self-driving vehicles? A case study of Texans

Connected, highly automated, and autonomous vehicles (CAVs) represent a significant technological advancement in transportation. The study's central question revolves around public readiness to adopt these technologies. The success of CAVs hinges on public perception, encompassing benefits, concerns, and willingness to adopt. Uncertainty surrounds CAVs' impact on carsharing, land use patterns, and the need for adjusted tolling policies. Existing research, while offering valuable insights through summary statistics, often lacks the depth of multivariate analysis necessary to fully understand the complex interplay of factors influencing public opinion. This Texas-wide survey and subsequent econometric modeling aims to fill this gap by investigating the multivariate relationships between Texans' opinions of CAV technologies and their demographic and locational characteristics. Understanding these relationships is crucial for multiple reasons. First, it provides valuable consumer demand insights essential for accurate fleet forecasting. Second, it equips policymakers and public officials with knowledge to make informed decisions regarding infrastructure investments, address legal and safety concerns, and support the overall development of CAV systems. Third, the study refines existing models of willingness to pay (WTP) for CAV technologies by incorporating individual characteristics, providing a more realistic forecast of WTP evolution and, consequently, CAV adoption rates. This contrasts with previous studies that assumed fixed annual increments in WTP. Finally, the study explores the often-overlooked influence of peer pressure on adoption timing, along with the potential impact of CAVs and shared autonomous vehicles (SAVs) on home location choices and the need for updated tolling policies to mitigate increased vehicle-miles traveled (VMT).

Numerous studies, encompassing academic research, private sector surveys, and reports from automotive-related websites, have explored public opinion on CAV technologies. A common thread across these studies is the public's cautious stance, often citing safety, affordability, and information security as primary concerns. While many studies provide summary statistics, few have delved into the multivariate relationships between individual characteristics and opinions. Krueger et al. (2016) employed a stated choice experiment in Australia, revealing younger travelers and existing carsharing users as more likely to adopt SAVs. Haboucha et al. (2015) conducted similar research in Israel and the U.S., finding a preference for SAVs among Israelis and highlighting the need for public education and increased costs of conventional car use to encourage SAV adoption. In contrast to these studies focusing on SAV preference, this study examines SAV usage frequency at different price points and explores the broader societal implications of CAV technologies. Howard and Dai (2013) used logit and log-linear models to understand the link between public opinion and demographics regarding AVs. Bansal et al. (2016) focused on Austin, Texas, identifying equipment failure as the primary concern and revealing a strong willingness to pay among wealthier, tech-savvy males, but less interest from older drivers. The current study builds upon this foundation by employing a larger, more geographically diverse sample with a more robust set of explanatory variables to provide a more comprehensive understanding of Texan opinions and their underlying drivers.

This research employed a Texas-wide survey administered through Survey Sampling International (SSI) using Qualtrics, a web-based survey tool. The survey, comprising 93 questions across seven sections, gathered data on opinions regarding AVs, crash history, WTP for various technologies (including automation levels and connectivity), adoption rates of carsharing and SAVs, home location shift decisions, and opinions on tolling policies. The initial sample included 1297 respondents; after data cleaning and sanity checks, 1088 respondents were deemed eligible for analysis. Person- and household-level weights were calculated to correct for sample biases, using the 2013 American Community Survey's Public Use Microdata Sample (PUMS) for Texas as a benchmark. Geocoding, utilizing Google Maps API and ArcGIS, linked respondents' locations to relevant built-environment attributes. The willingness to pay (WTP) for connectivity and various automation levels was analyzed using interval regression (IR) models, while the interest in connectivity, adoption timing of AVs, SAV adoption rates, home location shift decisions, and support for congestion pricing policies were analyzed through ordered probit (OP) models. Model specifications involved a stepwise process of variable selection, prioritizing statistical significance (p-values < 0.32) and practical significance. Practical significance was defined as a 0.2 standard deviation change in WTP for IR models and a 40% or greater shift in choice probabilities for OP models. Stata 12 software was employed for all model estimations. The study comprehensively addressed several aspects of public perception of CAVs, going beyond summary statistics to uncover intricate relationships between opinions and various individual, household, location, travel, technology and safety related attributes.

The study revealed several key findings. Regarding willingness to pay (WTP), Texans displayed an average WTP of $127 for connectivity, $2910 for Level 2 automation, $4607 for Level 3, and $7589 for Level 4 automation. A significant portion of respondents (29.3% for connectivity, 54.4% for Level 2, 31.7% for Level 3, and 26.6% for Level 4) were unwilling to pay more than $1500 for these technologies. Regarding SAV adoption, 41% of Texans indicated unwillingness to use SAVs even at $1 per mile, and only 7.3% intended to rely entirely on SAVs at this price point. This reluctance decreased as the price per mile increased to $2 and $3. Regarding home location decisions, 81.5% of Texans intended to remain in their current locations, even with increased SAV availability. Support for various congestion pricing policies showed relatively little variation. The two most popular activities Texans envisioned while riding in self-driving vehicles were talking to passengers (59.5%) and looking out the window (59.4%). Affordability and equipment failure emerged as the top concerns, while learning to use AVs and privacy breaches were the least concerning aspects. Regarding connectivity features, automatic notification of emergencies and vehicle health reporting were highly valued, while in-vehicle email and internet access were less popular. Only about 10% of Texans who had heard of carsharing were members of carsharing programs, primarily motivated by cost savings and environmental friendliness. Approximately 12.2% of Texans familiar with UberX or Lyft had used these services, driven by cost and time savings. Comfort in ride-sharing with strangers was low, at 16.4%. Econometric model results revealed several significant relationships. For instance, more experienced and older drivers demonstrated lower WTP, while higher-income and safety-conscious individuals showed higher WTP. Individuals living farther from transit stops were less likely to adopt SAVs, and those living far from city centers were more likely to support tolls that reduced property taxes. The models also highlighted the influence of various factors on adoption timing, revealing that those with higher WTP tended to be less influenced by friends' adoption decisions.

The findings address the research question by revealing a complex interplay of factors influencing Texans' acceptance of CAVs and SAVs. The significant relationships uncovered between individual characteristics, WTP, and adoption intentions provide valuable insights for transportation planners and policymakers. The relatively low WTP for connectivity and the high percentage of respondents unwilling to pay more than $1500 for automation technologies underscore the need for affordable and reliable technologies to ensure widespread adoption. The reluctance to use SAVs, even at low prices, highlights the need for strategies to overcome potential barriers to adoption, such as concerns about safety, convenience, and personal preferences. The finding that most Texans intend to stay in their current locations, despite the potential of CAVs to reshape urban development, suggests that the impact of CAVs on urban form might be less dramatic than previously anticipated. The relative lack of difference in support for various tolling policies implies a lack of strong public preference for specific revenue allocation mechanisms. These results have broader relevance to the field, highlighting the importance of considering public perception when developing and deploying CAV technologies. The findings call for a more nuanced approach to policy-making, one that acknowledges the heterogeneity of public preferences and the potential challenges in achieving widespread adoption.

This study provides valuable insights into Texan attitudes toward CAVs and SAVs, offering a comprehensive set of summary statistics and econometric model results. The identification of practically significant variables allows policymakers to tailor strategies to specific regions, promoting adoption in areas with lower penetration rates while focusing on infrastructure upgrades in high-adoption areas. The models' outputs can refine forecasts of CAV adoption and SAV usage, informing manufacturers, investors, and fleet operators. Future research could explore the evolution of these opinions over time, investigate the impact of specific policy interventions, and expand the geographical scope to better understand the national or global context.

The study's conclusions are based on a survey of Texans in 2015, and the attitudes toward CAVs and SAVs might have evolved since then. While the sample aimed for broad representation, biases might still exist, and the generalizability of the findings beyond Texas may be limited. The models employed make certain assumptions about error terms and variable relationships that might not fully capture the complexity of human decision-making. Lastly, the survey relies on stated preferences, which may not perfectly reflect actual adoption behavior.