A Naturalistic Driving Study of Feedback Timing and Financial Incentives in Promoting Speed Limit Compliance

Inappropriate speed choice increases crash risk. Emerging technologies, such as in-vehicle systems, can provide real-time and post-drive feedback to alert or educate drivers of their unsafe speed choices. Financial incentives, such as insurance reductions, can also be integrated with feedback. Although previous research supports the benefits of these approaches, research is limited in establishing their relative effectiveness and their long-term impact on speed limit compliance. A naturalistic driving study with 58 participants was conducted to investigate the effects of post-drive feedback versus financial incentives, both provided in conjunction with real-time feedback, on mitigating speeding behaviours. The study included a 4-week long baseline data collection period (no feedback), a 10-week intervention period examining three feedback types in a between subjects design (real-time only; real-time and financial incentives; and real-time and post-drive feedback), and a 2-week post-intervention period (feedback removed). Both real-time feedback alone and in conjunction with financial incentives were effective in raising speed limit compliance, but the effects did not sustain after the removal of feedback/incentives. Post-drive feedback, which was provided in the vehicle after a trip and as aggregated feedback on a website, did not appear to provide any benefits towards speed limit compliance; potential factors, including (voluntary) access frequency of post-drive feedback website, are discussed. Future research should continue to explore the feedback/incentive design space—carefully considering individual differences in driver characteristics and motivations—for supporting speed limit compliance and other risky driving behaviours.Many thanks to James Foley, Kazu Ebe, and Chuck Gulash from Toyota CSRC for their feedback on this paper as well as providing directions for our research in general. We would also like to acknowledge the assistance of Ryan Kealey, Jeffrey Alam, Tony Ye, and Isaac Chung in conducting this study