Identification of factors associated with various types of impaired driving

Motor vehicle crashes are a leading cause of death and injury worldwide, with human factors playing a crucial role. While driving under the influence (DUI) receives considerable attention, other forms of impairment, such as fatigue, distraction, and emotional distress, contribute significantly to fatalities. This study aims to address the definitional ambiguity surrounding impaired driving by investigating the factors associated with various types of impairment in a unified framework. The study's importance lies in its comprehensive approach, considering different types of impaired driving and their associated factors simultaneously, which is crucial for developing effective countermeasures to improve road safety. The study was conducted using data from Wyoming, a state characterized by mountainous terrain, offering a unique context for understanding impaired driving in varied environments. The study focuses on identifying associations, not necessarily causal relationships, between factors and impaired driving behaviors.

The literature review examines existing research on DUI, fatigue, distraction, and emotional conditions as contributors to impaired driving. Studies on DUI highlight the substantial economic and social costs associated with alcohol-impaired crashes and the influence of driver characteristics such as age and gender. Research on fatigued driving emphasizes its similarity to alcohol impairment, affecting reaction time and judgment. Distraction, encompassing various attention-diverting activities, is identified as a significant risk factor with comparable danger to DUI. Finally, emotional states such as anger, sadness, and depression are linked to risky driving behaviors, influencing risk perception and attitudes. However, the literature lacks comprehensive studies examining these various forms of impairment within a single analytical framework, which motivates this study to bridge this gap.

The study employs an extended multinomial logit model (MNL) using data from the Wyoming Department of Transportation (WYDOT) for the years 2015-2019. The MNL model accounts for the categorical nature of the response variable (impaired driving types: DUI, fatigue, distraction, emotional, and normal). The model was extended to allow for robust parameter estimation by incorporating constraints to exclude or equate coefficients across categories. This approach uses the vectorized generalized linear model (VGAM) with iteratively reweighted least squares (IRLS) for solving the algorithm, offering advantages over traditional MNL methods reliant on maximum likelihood estimation. The model includes various predictor variables categorized as: vehicle type (passenger car, pickup truck, SUV), environmental/temporal characteristics (road condition, lighting, weather, AADT, day of the week), driver actions (speeding, lane keeping, following distance, negotiating curves), and driver characteristics (seatbelt use, license validity, citation record, gender, age, residency). The inclusion of vehicle type and environmental factors aims to account for indirect effects of unseen driver attributes, such as personality and attitude. The methodological framework acknowledges the focus is on associations, not causation, due to the complexity of the factors contributing to impaired driving. The model was implemented using the VGLM function in R.

The analysis reveals significant associations between various factors and different types of impaired driving. SUV drivers were negatively associated with all types of impaired driving, suggesting potentially safer driving habits compared to drivers of passenger cars and pickup trucks. Adverse weather conditions were negatively associated with all impaired driving categories except for emotional driving, potentially because impaired drivers may avoid driving in such conditions. Nighttime driving was positively associated with DUI and emotional driving, possibly due to increased opportunities for alcohol consumption or reduced visibility. Work zones showed negative associations with DUI and fatigued driving. Higher average annual daily traffic (AADT) was negatively associated with most impaired driving types, possibly because drivers may perceive higher traffic as riskier. Weekend driving was positively associated with DUI but negatively associated with other impaired driving types. Driver actions such as failure to keep the proper lane were significantly associated with all forms of impaired driving. DUI drivers displayed shorter-term behavioral impairment reflected in actions like lane deviations. Fatigued drivers were associated with over-speeding, potentially to reach a destination quicker. Distracted drivers were associated with failure to maintain proper lanes and reduced speeds. Non-belted drivers and drivers with invalid licenses had higher likelihood of being impaired by any of the factors. Female drivers and older drivers were less likely to be involved in impaired driving, except for emotional states. A modified MNL model, by adjusting three variables, showed significant improvement in model fit compared to a standard MNL model, justifying the chosen method.

This study expands the understanding of factors associated with various types of impaired driving, moving beyond the traditional focus on DUI. The findings highlight the need for a more nuanced approach to policy and public awareness campaigns. The inconsistent definition of impaired driving in previous literature has hindered the implementation of effective safety measures. Drivers often compensate for impairment by avoiding challenging driving conditions. Differences between DUI and other forms of impaired driving suggest distinct underlying mechanisms. Variables such as license validity, seatbelt use, and lane-keeping are strong indicators of potential impairment across all driving categories. The study's focus on associations, rather than causation, needs to be emphasized because the factors involved in impaired driving are complex and interrelated.

This research underscores the need to broaden the scope of impaired driving beyond DUI, incorporating fatigue, distraction, and emotional distress. The findings highlight crucial indicators of impaired driving (license validity, seatbelt use, lane deviations), which can inform policy and interventions. Future research should investigate these aspects through surveys and driving simulations to gain deeper insights into drivers' behaviors and beliefs under various conditions of impairment. More comprehensive strategies are needed to address all types of impaired driving to enhance road safety. Autonomous vehicles hold promise, but a thorough understanding of driver behavior is essential before assessing their full potential.

The study relies on crash data, potentially overlooking instances of impaired driving without crashes. The study does not use detailed driving simulation or survey data to fully explore drivers' perceptions and behaviors under various impaired states. The findings are specific to Wyoming, a state with mountainous terrain, potentially limiting generalizability to other regions.