Modelling of motorcyclists' risky behaviour at an urban T-junction using generalised linear model: An exploratory study

Motorcyclists represent the greatest share of recorded traffic crashes and fatalities in Malaysia. The association between motorcyclists' behaviour and traffic conflict occurrence was assessed at a typical stop-regulated T-Junction in an urban area of Kuala Lumpur, Malaysia. Traffic activities were filmed over four months and the behaviour of motorcyclists entering the main road from the minor road was observed from recorded video sequences. Situations ending in a traffic conflict were compared to similar interactive situations not ending with a conflict. In total, 447 sets of interactions of motorcyclists and other motorists at the T-Junction were analysed where 242 interactions ended in conflicts (three of them ended with traffic crashes). The generalised linear model with a binomial response and link logit was adopted to assess the association of motorcyclists' behavioural variables with the probability of conflict occurrence. The significant behavioural variables were classified into categories according to the statistical variation of the value they can assume in the dataset. The motorcyclist's entering angle was the most significant contributory factor in the probability of traffic conflict. The findings can be helpful in deciding on road safety countermeasures. The results could feed into the decisions of policymakers to structure the education and licensing process.