Mechanisms of Driver Decision‐Making in Informal Mixed Urban Traffic: A Naturalistic Study in Kinshasa (DRC)

ABSTRACT

Traffic modelling and simulation in informal urban environments in Sub‐Saharan Africa (SSA) remain challenging, as observed driving behaviours often diverge from the assumptions embedded in models developed for structured, high‐income country contexts. This pilot study, conducted in Kinshasa, Democratic Republic of the Congo (DRC) – one of Africa's largest and most congested cities, where naturalistic driving data remain scarce – aims to document and formalise driver decision‐making mechanisms in informal urban traffic. Data were collected through onboard video recordings and self‐confrontation interviews with 14 drivers, selected via purposive sampling to represent the artisanal transport sector, which accounts for a substantial share of local urban mobility services. Using the Context–Motivation–Decision–Action framework, 54 distinct traffic situations were coded, resulting in the identification of five decision classes: short‐term adaptation, avoidance, long‐term adaptation, tactical yielding, and rule compliance. Fisher's Exact Test through Monte Carlo simulation (100,000 replicates, p < 0.0001) confirmed a highly significant association between traffic context and driver decision‐making; the Pearson χ ²  = 95.47 and bias‐corrected Cramér's V  = 0.586 are retained as descriptive indicators only, given the sparse contingency structure of the corpus. A large proportion (57.4%) of the observed situations reflects context‐specific features of Kinshasa's informal urban traffic. To bridge field observations and microsimulation, a conditional probability table P(Decision|Context) was derived to quantify the likelihood of a driver selecting a given decision class under specific traffic conditions, thereby providing preliminary behavioural parameters – treated as empirical priors pending replication on larger corpora – that can inform traffic microsimulation models and context‐adaptive driving assistance systems. All findings derive exclusively from Kinshasa (DRC); broader SSA applicability requires systematic cross‐city replication.