Driver feedback systems have demonstrated short-term effectiveness in mitigating risky behaviors such as speeding, harsh events, and mobile phone use. However, the long-term sustainability of these improvements, especially after feedback is withdrawn, remains a critical concern for transport safety research. This study investigates the relapse patterns of risky driving behaviors following the removal of feedback interventions, aiming to quantify behavior persistence and inform the design of long-term safety strategies. Naturalistic driving data from 31 car drivers were collected over a 21-month period encompassing baseline, feedback, and post-feedback phases. A total of 24,904 trips were analyzed. Survival analysis techniques, namely Kaplan-Meier estimates, were applied to four key risk indicators: harsh accelerations, harsh braking, speeding percentage, and mobile phone use percentage. The temporal evolution of each behavior was examined based on trip counts rather than calendar time to reflect driving exposure. Results show that relapse was most rapid for harsh braking and acceleration, while phone use exhibited the slowest decline in behavior persistence. Trip-based survival analysis offers valuable insights into feedback durability and the need for reinforcement mechanisms to maintain safer driving behavior over time.
| ID | pc648 |
| Presentation | |
| Full Text | |
| Tags |
