Modeling the Dynamics and Determinants of Road Traffic Accidents in Southwest Nigeria: A Dual-Level Analytical Approach

Abstract This study examines the trends, patterns, and determinants of road accidents in Southwest Nigeria using a dual-level analytical framework that integrates macro-level time series modeling and micro-level categorical analysis. Quarterly accident data from 2014 to 2024 were analyzed to identify long-term trends and seasonal fluctuations, while detailed accident records were assessed to determine the factors associated with accident severity. At the macro level, both Auto-Regressive Integrated Moving Average (ARIMA) and Error–Trend–Seasonal (ETS) models were applied for comparative forecasting, with the ARIMA (2,0,0) model demonstrating superior performance based on lower AIC and error metrics (RMSE, MAE, MAPE). The forecast results project a continuation of elevated accident trends through 2026. At the micro level, Chi-square tests of independence revealed significant associations between accident severity and variables such as driver gender, experience, and collision type, highlighting human and behavioral influences as critical determinants. The combined evidence underscores that accident frequency and severity in the region are shaped more by behavioral factors than environmental conditions. The study concludes with evidence-based policy recommendations emphasizing driver reorientation, seasonal safety enforcement, predictive monitoring, and data-driven road safety planning to mitigate future risks.