A flexible discrete density random parameters model for count data: Embracing unobserved heterogeneity in highway safety analysis

Abstract

In traffic safety studies, there are almost inevitable concerns about unobserved heterogeneity. As a feasible alternative to current methods, this article proposes a novel crash count model that can address asymmetry and multimodality in the data. Specifically, a Bayesian random parameters model with flexible discrete densities for the regression coefficients is developed, employing a Dirichlet process prior. The approach is illustrated on the Ontario Highway 401, which is one of the busiest North American highways. The results indicate that the proposed model better captures the underlying structure of the data compared to conventional models, improving predictive power examined based on pseudo Bayes factors. Interestingly, the model can identify sites (highway segments, intersections, etc.) with similar site characteristic (risk factor) profiles, those that manifest similarity in the heterogeneous effects of their risk factors (e.g., traffic flow) on traffic safety, providing useful insight towards designing effective countermeasures.

Effect of the COVID-19 pandemic on bike-sharing demand and hire time: Evidence from Santander Cycles in London

“…Where u t depends on values of u at times different than t ; v e is the associated variance and depends on the number of temporal neighbours (up to two neighbours in the second-order random walk); the more the neighbours, the more the information and thus the smaller the variance v e . Such a relatively complex temporal structure allows the model to borrow strength from the second order neighbours in time, thereby addressing unobserved heterogeneity [ 46 ], specifically temporally-structured heterogeneity, more fully.…”

Road Safety in Low-Income Countries: State of Knowledge and Future Directions

“…are often missing (being unknown or unmeasured) in crash databases, causing the unobserved heterogeneity problem. A large body of literature discusses how to overcome this problem in order to obtain reliable estimates [2,125]. However, most studies that address the abovementioned issues are conducted in the developed world.…”

Further compelling evidence for safety-in-numbers: It is more than meets the eye

“…In fact, such traditional models are unable to account for endogeneity of explanatory variables (Elvik, 2021;Mannering et al, 2020). This may be exacerbated by unobserved heterogeneity (Heydari, 2018;Mannering et al, 2016), which if not addressed properly, would result in bias in estimation.…”