Modeling Injury Outcomes of Crashes Involving Heavy Vehicles on Texas Highways

Abstract

Concern related to crashes that involve large trucks has increased in Texas recently because of the potential economic impacts and level of injury severity that can be sustained. However, detailed studies on large truck crashes that highlight the contributing factors leading to injury severity have not been conducted in Texas, especially for its Interstate system. The contributing factors related to injury severity were analyzed with Texas crash data based on a discrete outcome-based model that accounts for possible unobserved heterogeneity related to human, vehicle, and road–environment factors. A random parameter logit (i.e., mixed logit) model was estimated to predict the likelihood of five standard injury severity scales commonly used in the Crash Records Information System in Texas: fatal, incapacitating, nonincapacitating, possible, and none (i.e., property damage only). Estimation results indicated that the level of injury severity outcomes was highly influenced by several complex interactions between factors and that the effects of some factors could vary across observations. The contributing factors include driver demographics, traffic flow, roadway geometric features, land use, time characteristics, weather, and lighting conditions.

A comparison of the mixed logit and latent class methods for crash severity analysis

“…These drowsy events would increase a driver's susceptibility to a crash and therefore, would likely increase both the frequency of heavy truck crashes and the severity of heavy truck crashes during these times of the day. Heavy truck crashes occurring between 3 p.m. and 7 p.m. were less likely to result in an injury (Islam and Hernandez, 2013). The same study revealed that the likelihood of non-incapacitating injuries was lower for crashes occurring during the fall season (September-December).…”

“…Past studies (Islam and Hernandez, 2013;Lemp et al, 2011, Qin et al, 2013Bham et al, 2012) found a negative correlation between adverse road conditions, such as rain or snow, and crash severity, suggesting that the decline in severity under adverse roadway and weather conditions is attributable to drivers being more cautious under such conditions. However, Islam et al (2014) found that inclement weather, such as fog or mist, increased the probability of a major injury in urban single vehicle heavy truck crashes.…”

“…Islam and Hernandez (2013) found that male drivers were less likely to sustain possible injuries in heavy truck crashes, while 45-55 year old drivers were more likely to be involved in no injury crashes. While younger truck drivers (under 25 years old) have been associated with higher crash frequency (Cantor et al, 2009;Chang and Mannering, 1999), the effect of age on heavy truck crash severity has not been fully established.…”

“…Islam et al (2014) estimated four separate random parameter logit models of injury severity resulting from single-and multiple vehicle large truck at-fault crashes on rural and urban roadways in Alabama, and reported differences in the impact of variables on truck crash severity across urban and rural areas. Islam and Hernandez (2013) found that heavy truck crashes occurring in rural areas were more likely to result in fatal and major injuries, while crashes occurring in urban areas were less likely to result in incapacitating injuries. Khorashadi et al (2005) also found significant differences between the two area types and their respective models, suggesting that interactions between driver characteristics and environmental factors are likely affecting the demands placed on drivers in urban and rural areas.…”

“…The authors concluded that the partial proportional odds model was the preferred model among the three. Other, less common, discrete outcome crash severity methods employed in recent heavy track safety studies include random-parameter logit models (Chen and Chen, 2011;Islam and Hernandez, 2013;Islam et al, 2014), and non-parametric classification and regression tree models (Chang and Chien, 2013). These models were found promising for analyzing truck driver injury severity, although they were not directly compared to other model specifications in terms of prediction accuracy and inferences.…”