Collection duration of driving tracking data of older drivers

Driving tracking, or the observation of actual and naturalistic driving, is an effective approach for understanding and assessing the driving behaviors of older drivers. However, limited information is available regarding the effects of data collection duration on the characteristics of driving behavior. This study examined how different data collection durations (2 weeks and 1, 3, 6, and 12 months) influence older drivers' long-term driving behavior, specifically rapid deceleration events (RDEs). Analysis of the varying durations revealed common tendencies related to low-mileage bias (LMB) as well as differences in the likelihood of RDE occurrence. These factors were incorporated into predictive models, with values estimated using negative binomial regression across the different data collection durations. The results indicated that the characteristics of driving behavior differ between short-term (2 weeks and 1 month) and long-term (3, 6, and 12 months) data collection. Finally, this study provides insights into establishing a methodology for tracking driving behavior in older adults. • Although the RDEs for low- and high-mileage groups were compared across data collection durations, LMB trends were unclear. • The predicted RDEs from the negative binomial regression models showed a tendency consistent with LMB across all durations. • Negative binomial regression indicated that the 3-month model was comparable to models derived from 6- and 12-month models. • The long-term data (3, 6, and 12 months) suggests the presence of individual or other differences in the data. • To obtain stable driving tracking data for older driver’s analyses, at least 3 months of data are required. • The study is important for methodology establishment for driving tracking and naturalistic driving study in older drivers.