Driver Risk Factors in High-Altitude Winter Traffic Accidents: A Multivariate Behavioral Analysis of Erzurum Crashes

Creative Commons License

Öztaş S.

7st INTERNATIONAL TRAKYA SCIENTIFIC RESEARCH CONGRESS, Edirne, Türkiye, 28 - 29 Ağustos 2025, ss.227-235, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Edirne
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.227-235
  • Atatürk Üniversitesi Adresli: Evet

Özet

This study comprehensively investigates the impact of driver-related errors on the severity of traffic accidents in Erzurum Province, Turkey, a high-altitude region characterized by prolonged and harsh winter conditions. The province frequently experiences snowfall, icy road surfaces, low temperatures, and reduced visibility, which amplify the consequences of risky driving behaviors. Using an extensive dataset of 16,793 officially recorded traffic accidents from 2014 to 2023, the study focuses on the roles of alcohol test results, seat belt and helmet usage, and primary driver faults in shaping accident outcomes. Unlike previous research primarily centered on environmental and temporal factors, this work adopts a multivariate, behavior-focused approach to reveal the influence of human errors under high-risk winter conditions.

A Multivariate Analysis of Variance (MANOVA) was first conducted with fatalities and injuries as joint dependent variables, demonstrating that alcohol consumption, seat belt compliance, and driver fault categories significantly affect the combined pattern of accident severity (p < 0.001). To further identify and differentiate driver risk profiles, Discriminant Function Analysis (DFA) was employed. The results show that unsafe maneuvers, right-of-way violations, and overspeeding are the most influential factors contributing to high-severity crashes, while helmet use is strongly associated with the low-severity cluster.

These findings highlight the critical role of driver behavior in high-altitude winter traffic safety and underline the importance of targeted law enforcement, driver education programs, and preventive safety interventions. The study suggests that future research should integrate real-time driver monitoring, traffic volume data, and advanced predictive modeling techniques, such as machine learning and ensemble methods, to support proactive accident prevention strategies and enhance safety outcomes in similarly hazardous regions.