Driver Behaviour as a Predictor of Near-Miss Occurrence at Unsignalized Intersections in Port Harcourt, Nigeria: A Stratified Traffic Conflict Analysis

Captain Gospel OTTO * and Awarri William AWARRI

Department of Civil Engineering, Faculty of Engineering, Rivers State University, Port Harcourt, Nigeria.
* Corresponding Author; Email: captain.otto@rsu.edu.ng
ORCID Details

 

Research Article
Open Access Research Journal of Engineering and Technology, 2026, 11(02), 001–014.
Article DOI: 10.53022/oarjet.2026.11.2.0070
Publication history: 
Received on 22 August 2026; revised on 04 October 2026; accepted on 06 October 2026
 
Abstract: 
Near-miss events are widely accepted surrogate measures of road safety, yet they remain under-exploited in Nigerian cities where crash registers are incomplete. This study quantifies the association between observed driver behaviour and near-miss involvement at seven unsignalized intersections in Port Harcourt, Rivers State. Using the Traffic Conflict Technique, 9,231 driver–intersection interactions were video-recorded and manually coded into four mutually exclusive behavioural classes (over-speeding, failure to yield, distracted driving and safe driving) and three interaction outcomes (vehicle–vehicle near miss, vehicle–pedestrian near miss and no near miss). Association was tested with Pearson chi-square and likelihood-ratio statistics computed cell-wise on (r−1)(c−1) degrees of freedom, supported by Monte-Carlo permutation p-values, Cramér’s V with bootstrap confidence intervals, adjusted standardized residuals and Benjamini–Hochberg correction. Stratified Mantel–Haenszel odds ratios, Woolf heterogeneity tests and a binomial logistic regression with behaviour and intersection terms were used to separate behavioural from locational effects, and the model was internally validated by 10-fold cross-validation, bootstrap optimism correction, split-half replication and stratified permutation. Overall, 53.8% of interactions involved a near miss. Behaviour and outcome were dependent at six of the seven sites (all adjusted p < 0.001; Cramér’s V 0.21–0.72) but independent at Rumuokoro Junction (p = 0.580). Adjusted odds of near-miss involvement relative to safe driving were 8.26 for distracted driving, 6.41 for over-speeding and 6.19 for failure to yield. A significant behaviour × intersection interaction (p < 0.001; I² = 96–98%) shows that behavioural risk is strongly site-modified, so pooled estimates alone are misleading. The validated model achieved a cross-validated c-statistic of 0.79. Behaviour-based countermeasures should therefore be prioritized site by site rather than applied uniformly.
 
Keywords: 
Near-Miss Analysis, Driver Behaviour, Unsignalized Intersections, Traffic Conflict Technique, Logistic Regression, Road Safety, Port Harcourt.
 
Full text article in PDF: 
 
Copyright information: 

Copyright © 2026 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0