Efficacy of self-report questionnaire for assessing driver behavior using naturalistic driving data

Abstract: Understanding driver behavior is critical for improving road safety, optimizing traffic management, and developing effective interventions for risky driving. The relationship between self-reported driving behavior and actual driving performance has become a key area of interest in transportation research, particularly with the rise of advanced data collection techniques such as naturalistic driving studies (NDS). While previous research has predominantly focused on speeding to relate to the Driver Behavior Questionnaire (DBQ), this study adopts a novel approach by incorporating four distinct parameters, providing a comprehensive understanding of driver behavior, and enabling a more robust correlation between the questionnaire and NDS data. This study aims to explore the relationship between these two approaches by assessing the extent to which self-reported data can predict or reflect actual driving behavior, as captured by NDS.
The primary objective of this research was to assess whether the self-reported data gathered through this n-item questionnaire could serve as a reliable proxy for actual driving behavior, as measured objectively by NDS. To achieve this, the study involved collecting data, NDS, and questionnaires, from a sample of professional and non-professional drivers and aligning their self-reported behaviors with corresponding metrics derived from NDS data. The questionnaire is structured around four key driving parameters: speeding, overtaking, harsh acceleration, and jerk braking. To ensure the internal consistency of the questionnaire, Cronbach’s alpha scores were calculated.
The NDS data included quantifiable metrics such as speeding percentage, the number of wrong overtaking maneuvers, the percentage of time spent in harsh acceleration, and the frequency of jerk braking events. These parameters were chosen because they represent common risky driving behaviors and are readily measurable through NDS. Each driver’s NDS data was aligned with their questionnaire responses, allowing for a direct comparison of self-perception versus real-world behavior. To evaluate the predictive validity of self-reported measures, the Pearson correlation coefficients were calculated for each parameter, as the data followed a normal distribution.
The findings are expected to shed light on the accuracy of self-perceptions in driving behavior and how well self-reported measures can predict real-world driving patterns. By incorporating multiple driving behaviors and advanced data collection techniques, this study provides a more comprehensive approach to understanding the relationship between subjective and objective measures of driver performance.