Injury Severity Models Research Articles

How bicycle level of traffic stress correlate with reported cyclist accidents injury severities: A geospatial and mixed logit analysis

Transportation agencies need efficient methods to determine how to reduce bicycle accidents while promoting cycling activities and prioritizing safety improvement investments. Many studies have used standalone methods, such as level of traffic stress (LTS) and bicycle level of service (BLOS), to better understand bicycle mode share and network connectivity for a region. However, in most cases, other studies rely on crash severity models to explain what variables contribute to the severity of bicycle related crashes. This research uniquely correlates bicycle LTS with reported bicycle crash locations for four cities in New Hampshire through geospatial mapping. LTS measurements and crash locations are compared visually using a GIS framework. Next, a bicycle injury severity model, that incorporates LTS measurements, is created through a mixed logit modeling framework. Results of the visual analysis show some geospatial correlation between higher LTS roads and “Injury” type bicycle crashes. It was determined, statistically, that LTS has an effect on the severity level of bicycle crashes and high LTS can have varying effects on severity outcome. However, it is recommended that further analyses be conducted to better understand the statistical significance and effect of LTS on injury severity. As such, this research will validate the use of LTS as a proxy for safety risk regardless of the recorded bicycle crash history. This research will help identify the clustering patterns of bicycle crashes on high-risk corridors and, therefore, assist with bicycle route planning and policy making. This paper also suggests low-cost countermeasures or treatments that can be implemented to address high-risk areas. Specifically, with the goal of providing safer routes for cyclists, such countermeasures or treatments have the potential to substantially reduce the number of fatalities and severe injuries.

A comparative injury severity analysis of motorcycle at-fault crashes on rural and urban roadways in Alabama

The research described in this paper explored the factors contributing to the injury severity resulting from the motorcycle at-fault accidents in rural and urban areas in Alabama. Given the occurrence of a motorcycle at-fault crash, random parameter logit models of injury severity (with possible outcomes of fatal, major, minor, and possible or no injury) were estimated. The estimated models identified a variety of statistically significant factors influencing the injury severities resulting from motorcycle at-fault crashes. According to these models, some variables were found to be significant only in one model (rural or urban) but not in the other one. For example, variables such as clear weather, young motorcyclists, and roadway without light were found significant only in the rural model. On the other hand, variables such as older female motorcyclists, horizontal curve and at intersection were found significant only in the urban model. In addition, some variables (such as, motorcyclists under influence of alcohol, non-usage of helmet, high speed roadways, etc.) were found significant in both models. Also, estimation findings showed that two parameters (clear weather and roadway without light) in the rural model and one parameter (on weekend) in the urban model could be modeled as random parameters indicating their varying influences on the injury severity due to unobserved effects. Based on the results obtained, this paper discusses the effects of different variables on injury severities resulting from rural and urban motorcycle at-fault crashes and their possible explanations.

Investigation of factors affecting the injury severity of single-vehicle rollover crashes: A random-effects generalized ordered probit model

Rollover crashes are responsible for a notable number of serious injuries and fatalities; hence, they are of great concern to transportation officials and safety researchers. However, only few published studies have analyzed the factors associated with severity outcomes of rollover crashes. This research has two objectives. The first objective is to investigate the effects of various factors, of which some have been rarely reported in the existing studies, on the injury severities of single-vehicle (SV) rollover crashes based on six-year crash data collected on the Malaysian federal roads. A random-effects generalized ordered probit (REGOP) model is employed in this study to analyze injury severity patterns caused by rollover crashes. The second objective is to examine the performance of the proposed approach, REGOP, for modeling rollover injury severity outcomes. To this end, a mixed logit (MXL) model is also fitted in this study because of its popularity in injury severity modeling. Regarding the effects of the explanatory variables on the injury severity of rollover crashes, the results reveal that factors including dark without supplemental lighting, rainy weather condition, light truck vehicles (e.g., sport utility vehicles, vans), heavy vehicles (e.g., bus, truck), improper overtaking, vehicle age, traffic volume and composition, number of travel lanes, speed limit, undulating terrain, presence of central median, and unsafe roadside conditions are positively associated with more severe SV rollover crashes. On the other hand, unpaved shoulder width, area type, driver occupation, and number of access points are found as the significant variables decreasing the probability of being killed or severely injured (i.e., KSI) in rollover crashes. Land use and side friction are significant and positively associated only with slight injury category. These findings provide valuable insights into the causes and factors affecting the injury severity patterns of rollover crashes, and thus can help develop effective countermeasures to reduce the severity of rollover crashes. The model comparison results show that the REGOP model is found to outperform the MXL model in terms of goodness-of-fit measures, and also is significantly superior to other extensions of ordered probit models, including generalized ordered probit and random-effects ordered probit (REOP) models. As a result, this research introduces REGOP as a promising tool for future research focusing on crash injury severity.

Factors influencing injury severity of crashes involving HAZMAT trucks

Abstract This paper investigates factors affecting injury severity of crashes involving HAZMAT large trucks. It uses the crash data in the state of California from the Highway Safety Information System, from 2005 to 2011. The explanatory factors include the occupant, crash, vehicle, roadway, environmental, and temporal characteristics. Both fixed- and random-parameters ordered probit models of injury severity (where possible outcomes are major, minor, and no injury) were estimated; the random-parameters model captures possible unobserved effects related to factors not present in the data. The model results indicate that the occupants being male, truck drivers, crashes occurring in rural locations, under dark-unlighted, under dark-lighted conditions, and on weekdays were associated with increased probability of major injuries. Conversely, the older occupants (age 60 and over), truck making a turn, rear-end collision, collision with an object, crashes occurring on non-interstate highway, higher speed limit highway (≥65 mph), and flat terrain were associated with decreased probability of major injuries. This study has identified factors that explain injury severities of crashes involving HAZMAT, and as such, it could be used by policy makers and transportation agencies to improve HAZMAT transport, and thus, the overall highway safety.

Modeling single-vehicle run-off-road crash severity in rural areas: Accounting for unobserved heterogeneity and age difference

This study investigates factors that significantly contribute to the severity of driver injuries resulting from single-vehicle run-off-road (SV ROR) crashes. A mixed logit model approach is employed to explore the potential unobserved heterogeneous effects associated with each age group: young (ages 16–24), middle-aged (ages 25–65), and older drivers (ages over 65). Likelihood ratio tests indicated that the development of separate injury severity models for each age group is statistically superior to estimating a single model using all data. Based on the crash data collected from 2009 to 2013 in North Carolina, a series of driver, vehicle, roadway, and environmental characteristics are examined. Both parameter estimates and their elasticities are developed and used to interpret the models. The estimation results show that contributing factors which significantly affect the injury severity of an SV ROR crash differ across three age groups. Use of restraint device and horizontal curves are found to affect crash injuries and fatalities in all age groups. Reckless driving, speeding, distraction, inexperience, drug or alcohol involvement, presence of passengers, and driving an SUV or a van are found to have a more pronounced influence in young and middle-aged drivers than older drivers. Compared to the passenger cars, older drivers are less likely to experience possible injuries in a large-size vehicle (e.g., truck or bus). The average annual daily traffic volume and lighting conditions are also found to influence the resulting injury severity of an SV ROR crash specific to young drivers.

Analysis of occupant injury severity in winter weather crashes: A fully Bayesian multivariate approach

Multivariate injury severity models that consider the cross-group heterogeneity in the crash data where individuals or occupants are nested within vehicles and vehicles are nested within crashes are limited in the literature. Most previous studies on crash injury severity were conducted at the crash level ignoring the potential correlation in severity for the vehicles involved in the same crashes or individuals involved in the same vehicles. Ignoring these correlation and dependence effects might result in underestimation of standard errors and erroneous inferences. The objective of this paper is to correctly determine the factors affecting occupant injury severity in winter seasons by addressing the within-crash and between-crash correlation of injury severity. To achieve this, fully Bayesian hierarchical multinomial logit models were developed for estimating occupant injury severity in weather-related crashes, non weather-related crashes, and all crashes. These models were developed using disaggregate crash data with occupants nested within crashes for four winter seasons in Iowa. Significant factors affecting occupant injury severity included factors related to occupants (gender, seating position, occupant trapped status, ejection status, and occupant protection used), as well as crash-level factors (road junction type, first harmful event and major cause of crash). Weather-related variables, such as visibility, pavement and air temperature, were also significant factors in winter weather crashes. Interaction effects involving crash-level variables and occupant-level variables were also found significant. Overall, the model diagnostics suggested significant within-crash correlation in the study dataset justifying the use of a multivariate model specification that addresses multivariate error term correlation issues.

Quantifying the impact of adaptive traffic control systems on crash frequency and severity: Evidence from Oakland County, Michigan

IntroductionDespite seeing widespread usage worldwide, adaptive traffic control systems have experienced relatively little use in the United States. Of the systems used, the Sydney Coordinated Adaptive Traffic System (SCATS) is the most popular in America. Safety benefits of these systems are not as well understood nor as commonly documented. MethodThis study investigates the safety benefits of adaptive traffic control systems by using the large SCATS-based system in Oakland County, MI known as FAST-TRAC. This study uses data from FAST-TRAC-controlled intersections in Oakland County and compares a wide variety of geometric, traffic, and crash characteristics to similar intersections in metropolitan areas elsewhere in Michigan. Data from 498 signalized intersections are used to conduct a cross-sectional analysis. Negative binomial models are used to estimate models for three dependent crash variables. Multinomial logit models are used to estimate an injury severity model. A variable tracking the presence of FAST-TRAC controllers at intersections is used in all models to determine if a SCATS-based system has an impact on crash occurrences or crash severity. ResultsEstimates show that the presence of SCATS-based controllers at intersections is likely to reduce angle crashes by up to 19.3%. Severity results show a statistically significant increase in non-serious injuries, but not a significant reduction in incapacitating injuries or fatal accidents.

Introducing Latent Psychological Constructs in Injury Severity Modeling: Multivehicle and Multioccupant Approach

This paper presents a comprehensive model of injury severity that accounts for unmeasured driver behavior attributes. The results of the model have important implications for the design of safety interventions and advanced vehicular features and technologies. Engineering designs that accommodate the diminished capabilities of older drivers, that include rear-seat safety features, and that alert drivers to potential frontal collisions before they occur (collision warning systems and automated braking systems) would contribute to substantial reductions in injury severity for occupants of vehicles.

Comprehensive Injury Severity Analysis of SUV and Pickup Truck Rollover Crashes: Alabama Case Study

The research described in this paper explored the factors contributing to injury severity resulting from rollover accidents of light trucks—SUVs and pickups in particular—in Alabama. Given the occurrence of a rollover crash, random parameter logit models of injury severity (with possible outcomes of fatal, major, minor, and possible or no injury) were estimated. The estimated models identified a variety of statistically significant factors influencing the injury severities resulting from SUV and pickup rollover crashes. According to these models, some variables were significant in one model (SUV or pickup) but not the other. For example, variables such as roadway downgrade, female drivers, and daylight were significant only in the SUV model. Variables such as driver fatigue, driver or occupant wearing a seat belt, and freeway were significant only in the pickup model. In addition, some variables (such as not wearing a seat belt, two-lane roadway, horizontal curve, and old driver) were significant in both models. Estimation findings showed that two parameters (horizontal curve and at intersections) in the SUV model and two parameters (horizontal curve and dry roadway surface) in the pickup model could be modeled as random parameters, indicating their varying influence on the injury severity related to unobserved effects. The results obtained are used in this paper for a discussion of the effects of variables on pedestrian and bicyclist injury severities and their possible explanations.

Ordered Fractional Split Approach for Aggregate Injury Severity Modeling

In crash frequency models, frequency by severity level is examined with multivariate count models. In these multivariate approaches the impact of exogenous variables is quantified through the propensity component of count models. The main interaction between variables across severity levels is sought through unobserved effects; that is, there is no interaction of observed effects across the multiple count models. Although this is not necessarily a limitation, it could be beneficial to evaluate the impact of exogenous variables in a framework that directly relates a single exogenous variable to all severity count variables simultaneously. An alternative approach to examining crash frequency by severity is proposed. Specifically, instead of modeling the number of crashes, a fractional split modeling approach is used to study the fraction of crashes by each severity level on a road segment. Given the ordered nature of injury severity, an ordered probit fractional split model is used to study crash proportion by severity levels. The model is estimated for roadway segment data for single-vehicle and multivehicle crashes in Florida for 2009 through 2011. The model estimation results highlight the effect of traffic volume, lane width, shoulder width, proportion of divided segments, and speed limit on crash proportion by severity. The model results are used to predict hot spots for various crash types. The results highlight how the ordered probit fractional split models can be used for highway safety screening.

Comparison of the binary logistic and skewed logistic (Scobit) models of injury severity in motor vehicle collisions

The binary logistic model has been extensively used to analyze traffic collision and injury data where the outcome of interest has two categories. However, the assumption of a symmetric distribution may not be a desirable property in some cases, especially when there is a significant imbalance in the two categories of outcome. This study compares the standard binary logistic model with the skewed logistic model in two cases in which the symmetry assumption is violated in one but not the other case. The differences in the estimates, and thus the marginal effects obtained, are significant when the assumption of symmetry is violated.

Exploring the application of latent class cluster analysis for investigating pedestrian crash injury severities in Switzerland

One of the major challenges in traffic safety analyses is the heterogeneous nature of safety data, due to the sundry factors involved in it. This heterogeneity often leads to difficulties in interpreting results and conclusions due to unrevealed relationships. Understanding the underlying relationship between injury severities and influential factors is critical for the selection of appropriate safety countermeasures. A method commonly employed to address systematic heterogeneity is to focus on any subgroup of data based on the research purpose. However, this need not ensure homogeneity in the data. In this paper, latent class cluster analysis is applied to identify homogenous subgroups for a specific crash type-pedestrian crashes. The manuscript employs data from police reported pedestrian (2009–2012) crashes in Switzerland. The analyses demonstrate that dividing pedestrian severity data into seven clusters helps in reducing the systematic heterogeneity of the data and to understand the hidden relationships between crash severity levels and socio-demographic, environmental, vehicle, temporal, traffic factors, and main reason for the crash. The pedestrian crash injury severity models were developed for the whole data and individual clusters, and were compared using receiver operating characteristics curve, for which results favored clustering. Overall, the study suggests that latent class clustered regression approach is suitable for reducing heterogeneity and revealing important hidden relationships in traffic safety analyses.

Exploratory multinomial logit model–based driver injury severity analyses for teenage and adult drivers in intersection-related crashes

ABSTRACTObjective: Teenage drivers are more likely to be involved in severely incapacitating and fatal crashes compared to adult drivers. Moreover, because two thirds of urban vehicle miles traveled are on signal-controlled roadways, significant research efforts are needed to investigate intersection-related teenage driver injury severities and their contributing factors in terms of driver behavior, vehicle–infrastructure interactions, environmental characteristics, roadway geometric features, and traffic compositions. Therefore, this study aims to explore the characteristic differences between teenage and adult drivers in intersection-related crashes, identify the significant contributing attributes, and analyze their impacts on driver injury severities.Methods: Using crash data collected in New Mexico from 2010 to 2011, 2 multinomial logit regression models were developed to analyze injury severities for teenage and adult drivers, respectively. Elasticity analyses and transferability tests were conducted to better understand the quantitative impacts of these factors and the teenage driver injury severity model's generality.Results: The results showed that although many of the same contributing factors were found to be significant in the both teenage and adult driver models, certain different attributes must be distinguished to specifically develop effective safety solutions for the 2 driver groups.Conclusions: The research findings are helpful to better understand teenage crash uniqueness and develop cost-effective solutions to reduce intersection-related teenage injury severities and facilitate driver injury mitigation research.

The temporal stability of factors affecting driver-injury severities in single-vehicle crashes: Some empirical evidence

This study explores the temporal stability of factors affecting driver-injury severities in single-vehicle crashes. Using data for single-vehicle crashes in Chicago, Illinois from a nine-year period from January 1, 2004 to December 31, 2012, separate annual models of driver-injury severities (with possible outcomes of severe injury, minor injury, and no injury) were estimated using a mixed logit model to capture potential unobserved heterogeneity. Likelihood ratio tests were conducted to examine the overall stability of model estimates across time periods and marginal effects of each explanatory variable were also considered to investigate the temporal stability of the effect of individual parameter estimates on injury-severity probabilities. A wide range of variables potentially affecting injury severities was considered including driver-contributing factors, location and time of day, crash-specific factors, driver attributes, roadway characteristics, environmental conditions, and vehicle characteristics. The results indicated that, although data from different years share some common features, the model specifications and estimated parameters are not temporally stable. In addition, complex temporal stability behaviors were observed for individual parameter estimates such as driver gender, apparent physical condition of driver, type of vehicle, vehicle occupancy, road surface, weather, and light conditions. It is speculated that this temporal instability could be a function of the urban nature of the data, possible variations in police-reporting of crash determinants over time, the impact of continuing improvements in vehicle safety features and drivers’ response to them, and/or the effects of macroeconomic instability that was present over the time period considered in this study. Although the source of temporal stability is not clearly known, the general subject of temporal instability warrants substantial attention in future research. The possible presence of temporal instability in injury-severity models can have significant consequences in highway-safety practice where accurate forecasting of the impacts of alternative safety countermeasures is sought.

A note on generalized ordered outcome models

While there is growing application of generalized ordered outcome model variants (widely known as Generalized Ordered Logit (GOL) model and Partial Proportional Odds Logit (PPO) model) in crash injury severity analysis, there are several aspects of these approaches that are not well documented in extant safety literature. The current research note presents the relationship between these two variants of generalized ordered outcome models and elaborates on model interpretation issues. While these variants arise from different mathematical approaches employed to enhance the traditional ordered outcome model, we establish that these are mathematically identical. We also discuss how one can facilitate estimation and interpretation while building on the ordered outcome model estimates – a useful process for practitioners considering upgrading their existing traditional ordered logit/probit injury severity models. Finally, the note presents the differences within GOL and PPO model frameworks, for accommodating the effect of unobserved heterogeneity, referred to as Mixed Generalized Ordered Logit (MGOL) and Mixed Partial Proportional Odds Logit (MPPO) models while also discussing the computational difficulties that may arise in estimating these models.

Effects of Geodemographic Profiles of Drivers on Their Injury Severity from Traffic Crashes Using Multilevel Mixed-Effects Ordered Logit Model

The purpose of this study was to use a statistical model to examine the effects of various geodemographic factors on levels of driver injury severity. A driver's geodemographic profile with respect to involvement in a traffic crash consists of variables from multiple hierarchical levels such as drivers nested within crashes and crashes clustered within areas. This structure implies that driver-level observations are correlated rather than independent as assumed in many injury severity models. To capture within-group and between-group correlations in observations, a multilevel mixed-effects ordered logit model was employed. The mixed effect allows some variables to vary by observations (i.e., random parameters). The analysis was based on national traffic crash data in the United Kingdom between 2009 and 2011, consisting of 271,654 drivers from 217,523 traffic crashes occurring across 27,773 census areas. Data on area, social deprivation, census, and land use patterns were collected from multiple sources and integrated with a geographic information system framework. Results indicated that the severity of injuries sustained by urban drivers involved in crashes increased if they traveled to rural areas; the level of driver injury severity also increased if traffic crashes occurred in areas with high car ownership per capita; and drivers from more disadvan-taged areas would sustain, if all else is equal, more severe injuries. Findings from this study would be useful to the UK Department for Transport and local authorities in formulating safety policies aimed at enhancing driver education, training, and licensing programs.

Comparative Analysis of Injury Severity Resulting from Pedestrian–Motor Vehicle and Bicycle-Motor Vehicle Crashes on Roadways in Alabama

A comparison is provided between the influence of a variety of variables and the injury severities resulting from pedestrian-motor vehicle and bicycle–motor vehicle crashes in Alabama with the incorporation of the effects of randomness across the observations. Given the occurrence of a crash, random parameter logit models of injury severity with possible outcomes of fatal, major, minor, and possible or no injury were estimated. The estimated models identified a variety of statistically significant factors influencing pedestrian and bicyclist injury severities. According to these models, some variables were found to be significant in only one model (pedestrian or bicycle) but not in the other. For example, variables such as roadway crossing, crashes on Saturdays, and crashes on private roadways were found significant only in the pedestrian model. By contrast, variables such as the use of retroreflective clothing, crashes during the evening peak hour, and crashes in summer were found significant only in the bicyclist model. In addition, some variables—such as older pedestrians and bicyclists, dark roadway, and low-speed roadway—were found significant in both models. Also, estimation findings showed that three parameters in the pedestrian model (between midnight and 7:00 a.m., clear weather, and at intersections) and two parameters in the bicycle model (level grade on tangent and clear weather) could be modeled as random parameters; these findings indicated their varying influences on injury severity. From the results obtained, the effects of different variables on pedestrian and bicyclist injury severities and their possible explanations are discussed.

Differences in Injury Severities Between 2-Vehicle and 3-Vehicle Crashes

Objective and Methods: In traditional injury severity–related studies, 2-vehicle and 3-vehicle crashes are typically considered in a combinatory manner and thus the injury casual factors for these 2 crash types are implicitly assumed to be the same. This article attempts to investigate the potential discrepancy between 2- and 3-vehicle crash severities with the aid of a continuation ratio logit model with the property of partial proportional odds.Results: The modeling results show that there are a number of significant differences between 2- and 3-vehicle crash injury severities in terms of the contributing factors, the magnitude of impact, and even the direction of effects.Conclusions: The research illustrates that a series of environmental and crash factors (e.g., rear-end straight crashes, urban roadways, alcohol usage, and different driving cohorts) are statistically significant in interpreting the disparity of coefficients between 2- and 3-vehicle crash injury severity models. It raises awareness that the combined analysis of 2- and 3-vehicle crashes should be exercised with caution, particularly when safety research targets crashes with less severe injuries.

Motor vehicle drivers’ injuries in train–motor vehicle crashes

The objectives of this research were to: (1) identify a more suitable model for modeling injury severity of motor vehicle drivers involved in train–motor vehicle crashes at highway–rail grade crossings from among three commonly used injury severity models and (2) to investigate factors associated with injury severity levels of motor vehicle drivers involved in train–motor vehicle crashes at such crossings. The 2009–2013 highway–rail grade crossing crash data and the national highway–rail crossing inventory data were combined to produce the analysis dataset. Four-year (2009–2012) data were used for model estimation while 2013 data were used for model validation. The three injury severity levels—fatal, injury and no injury—were based on the reported intensity of motor-vehicle drivers’ injuries at highway–rail grade crossings.The three injury severity models evaluated were: ordered probit, multinomial logit and random parameter logit. A comparison of the three models based on different criteria showed that the random parameter logit model and multinomial logit model were more suitable for injury severity analysis of motor vehicle drivers involved in crashes at highway–rail grade crossings. Some of the factors that increased the likelihood of more severe crashes included higher train and vehicle speeds, freight trains, older drivers, and female drivers. Where feasible, reducing train and motor vehicle speeds and nighttime lighting may help reduce injury severities of motor vehicle drivers.

Pedestrian at-fault crashes on rural and urban roadways in Alabama

The research described in this paper explored the factors contributing to the injury severity resulting from pedestrian at-fault crashes in rural and urban locations in Alabama incorporating the effects of randomness across the observations. Given the occurrence of a crash, random parameter logit models of injury severity (with possible outcomes of major, minor, and possible or no injury) for rural and urban locations were estimated. The estimated models identified statistically significant factors influencing the pedestrian injury severities. The results clearly indicated that there are differences between the influences of a variety of variables on the injury severities resulting from urban versus rural pedestrian at-fault accidents. The results showed that some variables were significant only in one location (urban or rural) but not in the other location. Also, estimation findings showed that several parameters could be modeled as random parameters indicating their varying influences on the injury severity. Based on the results obtained, this paper discusses the effects of different variables on pedestrian injury severities and their possible explanations. From planning and policy perspective, the results of this study justify the need for location specific pedestrian safety research and location specific carefully tailored pedestrian safety campaigns.