Exploring the associated factors for multiple-threats and near-miss incidents at signalized midblock crosswalks

Abstract

This study explores the associated factors for multiple-threats and near-miss incidents resulting from drivers yielding right-of-way at signalized midblock crosswalks. It utilized observational survey data collected from 15 crosswalks located in Las Vegas, Nevada. The study sites are signalized with yellow flash-based signals. The descriptive analysis revealed that 5.61% and 1.56% of all 3976 driver-pedestrian interactions were multiple-threat and near-miss incidents, respectively. Due to the rarity of the events, traditional logistic regression (TLR) and rare events logistic regression (RELR) were applied and compared. The RELR performed slightly better than TLR in modeling near-miss events. The modeling results revealed that the odds of near-miss incident occurrences are about five times higher in the presence of multiple-threat incidents. The common factors associated to a high likelihood of multiple-threat and near-miss incidents are number of lanes, high incoming vehicle speeds, and time taken to yield right-of-way. A narrow-raised median, high vehicle density, and green lights at the next intersection downstream of the crosswalk are associated with an increased likelihood of multiple-threat incidents, whereas near-miss incidents are positively associated with residential land use, inner and outer turning lanes, and night time. These study findings can be used by city engineers and planners.