Impacts of Vehicle Design on
Pedestrian Injury Severity
Author: Josh Roll
Unit: ODOT Research Unit
Email: josh.f.roll@odot.oregon.gov
Github Handle: JoshRoll
Unit: ODOT Research Unit
Email: josh.f.roll@odot.oregon.gov
Github Handle: JoshRoll
1 Introduction
Fatal pedestrian traffic injuries in the United States continue to rise with preliminary data for 2022 likely to be the highest count of pedestrian deaths since 1980. While the frequency of pedestrian fatal injuries has increased, so too has their share of total traffic deaths with pedestrian fatal injuries now nearly 18% of all fatal traffic injuries. This report documents reports efforts by the Oregon Department of Transportation (ODOT) to quantify the role of vehicle type, weight, and size as well as roadway and road user characteristics in fatal and serious injury crashes involving pedestrians.
The Oregon Department of Transportation is assessing Vulnerable Road User (VRU) traffic safety as directed by the Federal Highways Administration (FHWA). The US Department of Transportation’s National Roadway Safety Strategy (NRSS) calls for Safer People, Safer Roads, Safer Vehicles, Safer Speeds, and Post-crash care so ODOT’s VRU is assessing bicycle and pedestrian safety through the lens of the NRSS. Using data from three states in the Pacific Northwest, this report develops information on multiple inputs to pedestrian injury severity using descriptive statistics and logistic regression to parse the effects of each input on variations in pedestrian serious and fatal injury.
This report first describes in the background section, what is known at an aggregates level, about the changes in characteristics of the passenger vehicle fleet and changes in injury outcomes. The report then details past literature on pedestrian injury severity including a review of multivariable analysis as well as a brief review of biomechanical mechanisms involved in producing more serious injuries from larger vehicles. Next, a review of the data and methods used in this report is described followed by an analysis of the data using descriptive analysis and logistic regression.
2 Background
As pedestrian injuries have risen over the last two decades research has increasingly highlighted the role in this increase played by the growth in larger passenger vehicles, namely pickup trucks and SUVs also known singularly as light trucks. Average weight of passenger vehicles has grown as well with the average vehicle up 6% in total weight with pickups’ weight up 30% compared to pickups of the 1970s. SUVs and pickups represent an ever-greater percentage of the total passenger vehicles with new sales of these vehicle types at record levels (3). The chart below shows new vehicle sales data from the Bureau of Transportation Statistics, describing the changes over time of the predominate body type being sold each year. Whereas sport utility vehicles (SUVs), crossover utility vehicles (CUVs) and pickups accounted for just 35% of the new vehicles sold in 1990, that figure is now nearly 80% of total sales in 2022.
Figure 2.1: New vehicle Sales by Vehicle Type in U.S.
At an aggregate level, SUVs and pickup trucks are being linked to an increase in overall traffic injury for vehicle occupants and vulnerable road users alike. One research paper estimated that for every million light trucks added to the U.S. fleet, 34 additional road deaths occur per year. The authors also demonstrated that even though larger vehicles might protect their occupants’ larger vehicles put enough external risk on other smaller vehicle occupants and vulnerable road uses to likely cause 4.3 deaths for every large vehicle occupant saved (4). Other research pointed out that the percentage of SUVs in fatal crashes increased from 13.5% in the year 2000 to 21.5% in 2019, corresponding with the average weight of vehicles involved in crashes increasing from 3,850 pounds to 4,265 pounds over the same period (5). This research also concluded the growth in light trucks could be attributed to at least 8,000 excess pedestrian deaths between 2000 and 2019.
Traffic speed is an established risk factor known to increase the probability that a pedestrian crash will be fatal (6, 7). Given that vehicle characteristics such as curb weight proportionately affect force applied to impacted pedestrians, and therefore the severity of the injury, supplementation of crash severity analyses with these data is critical for informing transportation policies and plans. This report aims to fill this gap by leveraging vehicle identification numbers to obtain vehicle characteristics followed by assessment of contribution to pedestrian injury using descriptive statistics and multinomial logistic regression modeling.
3 Literature Review
The literature review first summarizes the published research that relied on analytic modeling techniques to determine the influence of various vehicle, roadway, pedestrian, driver, and environmental characteristics on pedestrian injury severity. The literature review then summarizes the current understanding of biomechanical mechanisms of pedestrian injury with regards to how different vehicle types correspond to injury type and injury severity.
3.1 Pedestrian Injury Severity Analysis
Starting in the early 2000s significant research was performed that highlighted the connection between larger personal vehicles and pedestrian injury and death. In 2003, Ballesteros et al. (8) used data from Massachusetts to understand vehicle type and speed on pedestrian injury using around 2,500 linked police and hospital crash records. They classified vehicles into passenger cars, light trucks and vans (LTVs) which included sport utility vehicles and pickups and found that LTVs were associated with a 1.32 increase in the odds of a fatal pedestrian injury and vans associated with a 1.30 increase in odds of fatal injury. The authors also analyzed curb weight finding that vehicles over 3,200 pounds were also more likely to kill a person walking in the even of a crash with odds ratio of 1.39. Speed too was documented as a factor that increased the odds of serious and fatal injury, but this research did not include other control variables for the crash participants such as age or gender nor did they include information about the roadway like the lighting conditions or whether the crash occurred at an intersection or segment. Henary and Crandall (9) used data from the Pedestrian Crash Data Study (PCDS) to determine the effect of LTV on pedestrian crash injury outcomes. NHTSA funded the PCDS data collection which started in 1994 and ended in 1998 with 521 pedestrian crashes collected at six cities across the U.S. Researchers found that LTVs are associated with a 1.4 increase in the odds of fatal pedestrian injury and a 1.31 increase in the odds of serious pedestrian injury (9). The authors controlled for the age of the pedestrian and also found vehicle speed to be a major continuing factor. PCDS data was used again in another study (10) where the authors found that LTVs were associated with an increase in the odds of pedestrian fatal pedestrian injury by 1.5 times and 2.1 times for serious pedestrian injury. The authors also found vans to increase the odds of fatal and serious injury with odds ratios of 1.4 and 1.6 respectively. While the authors did not control for vehicle speed and factors of the driver, researchers studying pedestrian injuries in New York City found that LTVs were associated with an increased of 4.2 in the odds of children ages 5 to 9 being killed when struck by a vehicle (11).
In 2007 researchers studied 3,200 crash records containing both pedestrian and bicycle crashes and found similar results to those mentioned above. Vehicle type, including when the striking vehicle was an SUV or pickup truck significantly increased the odds of a pedestrian fatal and serious injury (12). Speed was a major contributing factor as was time of day though lighting was not significantly associated, but the authors suggest this may be due to including time of day as an explicit factor. In another study (13) over 9,000 pedestrian fatal crashes from the Fatal Accident Reporting System (FARS) to analyze the factors influencing whether the pedestrian died on the scene, died one day after the incident, died 10 days after the incident or died 10 to 30 days after the incident. In addition to considering other factors such as pedestrian age and gender the authors found that smaller vehicles were negatively associated dying at the scene while lower speeds were also associated with a smaller chance of the pedestrian dying at the scene. Other road user variables used in the analysis included whether the driver was impaired or was male with the former being associated with a higher chance of the pedestrian dying at the scene while the latter, contrary to other analyses summarized in this literature review, was associated with a smaller chance of dying at the scene.
A study of crashes in Colorado (14) used nearly 14,000 crash records to understand the impact of various factors in pedestrian injury severity finding that when an SUV or pickup struck the pedestrian the likelihood of death were significantly higher with odds ratios of 1.59 and 1.96 respectively. The authors also noted vehicle speed (estimated by police at the time of the collision) was associated with higher odds of the pedestrian dying, as was the age of pedestrians, impairment of either the driver or pedestrian, darkness, and crashes occurring on segments compared to intersections. Researchers in Finland documented the effects of vehicle type, speed, and lighting conditions on pedestrian injury in Finland concluding that larger vehicles are associated with higher odds of fatal injury though no odds ratios were reported for this variable (15). High speeds and darker conditions were also associated with higher odds of pedestrian fatality in this study.
A research team used over 10,000 pedestrian crash records from Victoria, Australia to determine contributors to pedestrian serious and fatal injury finding pickup trucks to be associated with higher odds of fatal and serious injury with odds ratios from logistic regression of 1.2 and 1.3 respectively (16). SUV vehicle types were grouped with station wagon vehicle types, and which is why that vehicle type may have been associated with lower odds of serious and fatal pedestrian injury. Speed limit, lighting conditions, and pedestrian age were associated with higher odds of serious and fatal injury in this study. One anomaly of this study was that intersections increased the odds of fatal and serious injury which is contrary to the other studies summarized in this literature review. In 2010 researchers (17) performed a meta-analysis of published research to determine the impact of LTVs on pedestrian injury reviewing 12 studies that met their inclusion criteria. Using the most high-quality studies, the author estimated pooled model odds ratios and concluded that LTVs increase the odds of fatal injury by 1.54, demonstrating the additional risk placed om pedestrians by larger vehicles.
3.2 Biomechanical Pathways to Pedestrian Injury
The pathway that links vehicle size and design to fatal and serious pedestrian injury has been examined by researchers and rests on the effects of higher front ends of SUVs and pickups and the increased likelihood of injuring vital parts of the human body. A number of studies (18, 19, 20, 21) all demonstrated using simulation software how vehicle type influences what vital body locations are injured in the event of a collision. Analysis of empirical data confirms results from simulation data where taller vehicles tend to injury impacted seems to vary by vehicle type (22, 8). The past research on pedestrian injury severity has informed the data and analysis methods used in this paper which will be summarized in the next section.
In the figure below taken from Maki et al. (2002) the differences in likely location of impact when struck by a passenger vehicle (sedan) versus a larger vehicle like an SUV are demonstrated using a simulation. On the right hand panel the pedestrian is struck by the passenger car in the lower extremities mostly avoiding vital organs with the head impact a secondary impact instead of a primary impact. On the right hand panel the graphic shows that when a larger vehicle strikes a pedestrian its more likely to make primary impact on the torso and upper body where vital organs are contained. When the upper parts of the body are struck the chances of vital organs and the person’s head experiencing the primary strike increase thus increasingly the chances of serious injury.
Figure 3.1: Difference in Impact Area by Vehicle Body Type (Maki et al. 2002)
4 Data and Methods
This report relies on crash data from Oregon, Washington, and Idaho including person, vehicle and crash level tables from each state. This report uses those data to describe variations in pedestrian injury severity using descriptive statistics to tell the story at a high level, without controlling for the interaction of various data elements. Following the descriptive statistics summary, multinomial logistic regression is utilized to more precisely analyze the effect of each of the model inputs which include:
- Vehicle characteristics
- Roadway characteristics
- Road user characteristics
- Environmental conditions
The aim of including the various inputs it to frame these factors and their role in pedestrian injury severity around the USDOT’s NRSS. Traffic crashes are complex events and multiple factors play a role and this report attempts to describe each component’s part.
4.1 Crash Data
This report relies on crash data from Oregon, Washington, and Idaho where person, vehicle and crash tables were combined for analysis featured in this report. Information about the crash participants, in this case the pedestrian and the driver, include the participants age, gender, and whether these participants were impaired by drugs or alcohol during the crash. Information about the crash including details like posted speed limit, lighting conditions, road surface condition and weather are added to the pedestrian crash incident. Finally, information about the vehicle involved in the crash is added based on data elements extracted by decoding the vehicle identification number (VIN) of the vehicle involved. Only crashes where a single vehicle was involved are used in this report.
4.2 Vehicle Data
Washington and Idaho crash data include the VIN for nearly every crash while Oregon crash data does not on a systematic basis. Leveraging past work using deterministic data linkage, the VIN for Oregon crash data was acquired by joining its Crash Data System (CDS) data maintained by the Crash Analysis and Reporting (CAR) unit with crash data maintained by Orgon DMV where VIN number is maintained. Since VIN is not collected by DMV in all cases the Oregon crash data used in this analysis has a smaller number of observations compared to Washington and Idaho. This smaller number of observations has some implications described below where parameters from logistic regression have larger standard errors and confidence intervals.
For all crash records where the VIN was available, the VIN was decoded using the NHTSA VIN decoding service using the application programming interface (API) as well as the use of Canadian Vehicle Specifications (CVS) database available through NHTSA (23). The NHTSA VIN decoding service has the potential to provide a wide array of vehicle data elements. Make, model, model year, body type, number of doors, style, and displacement (measured in liters) are mostly complete, however, no requirement exists for vehicle manufacturers to report to this database so data is incomplete for some elements. Curb weight is derived from the CVS database but are not able to be retrieved through a deterministic linkage using the VIN and instead are linked using fuzzy string search using the Jaro-Winkler metric with a cutoff of 0.40 (24). Many of the input variables documented below are self-explanatory such as the posted speed limit, pedestrian and driver age, and weather conditions. Vehicle characteristics may be less straight forward however and are worth some explanation here. Light-duty passenger vehicles are classified by their weight and include all vehicles weighing 10,000 pounds or less. This paper includes only pedestrian crashes where light-duty passenger vehicles are involved. The body types of these passenger vehicles include passenger cars, pickups, sport utility and cross over utility vehicle (SUV/CUV) and vans. Two more detailed vehicle characteristics are included in this paper including curb weight and overall height. Curb weight is the weight of the vehicle in pounds straight from the factory including a full tank of fuel and does not account for any custom add-ons including grill guards, headache racks, or other vehicle features that might change the factory weight. Overall height measures the vertical distance between the ground and the highest point on the vehicle.
5 Descriptive Analysis
This section summarizes the available data with an aim to describe how the various data element relate to variations in pedestrian injury severity. The table below summarizes the pedestrian crash data by injury severity level across the various factors that will be used below in the logistic regression analysis section. Table 1 below details data elements used in this research by summarizing the number of pedestrian injuries by severity for different categories of input data used in the modeling analysis. This table aims to detail the number of observations for each injury severity across the different categories of inputs while also highlighting at a basic level how different factors are associated with fatal pedestrian injury. For instance, in the body type section of Table 1, while passenger cars are involved in more (in nominal terms) fatal injury crashes, only 4.1% of incidents involving passenger cars result in the death of the person walking. Contrast that finding with the percentage of pickup and SUV/CUV involved crashes where 7.5% and 5.6% respectively, of the crashes result in the death of the pedestrian. A similar outcome is present when looking at the difference in the percentage of crashes that are fatal for curb weight category and overall vehicle height where heavier vehicles and taller vehicles have a higher proportion of the crashes resulting in a fatal outcome. For the vehicles included in the data used in this analysis, the average weight and heigh of passenger cars, pickups, SUVs/CUVs, and vans is summarized in the bullets below:
- Passenger Cars: 3,065 lbs. / 57 in.
- Pickups: 4,680 lbs. / 72 in.
- SUV or CUV: 4,123 lbs. / 69 in.
- Van: 4,320 lbs. / 70 in.
The proportion of injuries that are fatal on roads with higher speeds highlights the role of speed in fatal and serious injury pedestrian crashes. Whereas only 2.1% of crashes are fatal when posted speeds are 25 mph or less, that proportion climbs to 15.2 % and 32.9% when the posted speed is 40-50 mph and greater than 50 mph respectively. For posted speed limits, this value does not necessarily indicate the speed of the vehicle at the time of the crash since some vehicles may be slower if the crash occurred at an intersection or driveway, however its somewhat indicative of the speed at the time of impact as can be seen in the data descriptive table. The data descriptive results for lighting conditions highlights the importance of well-illuminated roadways for pedestrian safety 18.2 % of pedestrian crashes are fatal when its dark and no lighting is present compared to 6.3% when lighting is present and 2.4% when its daylight.
Roadway type indicates whether the crash was on a segment in between intersections and driveways, or at an intersection or driveway. The table below highlights when a pedestrian crash occurs on the segment the crash is more likely to be fatal, likely due to higher vehicle speeds on segments versus intersections and driveways where people are likely slowing down to make a turn.
Pedestrian age and driver age are also included below in Table 1 which indicates that older pedestrians are more likely to be fatally injured in the event of a crash. Gender of both the pedestrian and driver are included below with the information showing that when the pedestrian or driver are male the crash is more likely to be fatal. Other roadway conditions are summarized below including lighting condition, weather conditions, and road surface conditions.
As mentioned above, data from three states are included in this analysis with the number of records from each state provided in Table 1. In sum, Table 1 illustrates that many variables can impact pedestrian injury severity. To better understand the influence of vehicle characteristics such as body type, curb weight, and height, as well as how these vehicle parameters can compound with other continuous variables such as speed, pedestrian and driver age and impairment, roadway conditions, and environmental conditions, multinomial logistic regression was performed.
Group | Variable | Possible Injury | Minor Injury | Serious Injury | Fatal Injury |
|---|---|---|---|---|---|
Body Type | |||||
Passenger Car | 3739 (37.1%) | 4024 (40%) | 1920 (19.1%) | 386 (3.8%) | |
Van | 463 (35.3%) | 542 (41.3%) | 240 (18.3%) | 68 (5.2%) | |
SUV or CUV | 1741 (37%) | 1839 (39.1%) | 859 (18.2%) | 268 (5.7%) | |
Pickup | 1043 (32.5%) | 1160 (36.1%) | 755 (23.5%) | 253 (7.9%) | |
Curb Weight (lbs.) | |||||
3,500 lbs or less | 4420 (36.9%) | 4750 (39.7%) | 2296 (19.2%) | 506 (4.2%) | |
Greater than 3,500 lbs. | 2566 (35%) | 2815 (38.4%) | 1478 (20.2%) | 469 (6.4%) | |
Overall Height (in.) | |||||
70 inch tall vehicle or less | 5319 (37%) | 5682 (39.5%) | 2749 (19.1%) | 632 (4.4%) | |
Greater than 70 inch tall vehicle | 1433 (34%) | 1594 (37.8%) | 880 (20.9%) | 309 (7.3%) | |
Posted Speed (mph) | |||||
<25 mph | 2663 (38.5%) | 2961 (42.9%) | 1138 (16.5%) | 146 (2.1%) | |
26 - 30 mph | 1974 (41%) | 1926 (40%) | 772 (16%) | 147 (3.1%) | |
31 - 35 mph | 1872 (34.4%) | 1980 (36.4%) | 1281 (23.6%) | 301 (5.5%) | |
36 - 40 mph | 213 (28%) | 287 (37.7%) | 185 (24.3%) | 76 (10%) | |
40 - 50 mph | 194 (22.9%) | 285 (33.6%) | 243 (28.7%) | 126 (14.9%) | |
50+ mph | 70 (13.2%) | 126 (23.8%) | 155 (29.2%) | 179 (33.8%) | |
Roadway Type | |||||
Intersection or Related | 4988 (40.7%) | 4987 (40.7%) | 1954 (15.9%) | 339 (2.8%) | |
Driveway or Related | 417 (40.2%) | 387 (37.3%) | 185 (17.8%) | 49 (4.7%) | |
Segment Related | 1556 (26.2%) | 2168 (36.5%) | 1622 (27.3%) | 586 (9.9%) | |
Pedestrian Age | |||||
12 Under | 621 (33.1%) | 892 (47.5%) | 325 (17.3%) | 38 (2%) | |
12 to 22 Age | 1634 (35.3%) | 2009 (43.4%) | 902 (19.5%) | 88 (1.9%) | |
23 to 39 Age | 1998 (39.2%) | 1944 (38.1%) | 943 (18.5%) | 213 (4.2%) | |
40 to 64 Age | 2120 (37.2%) | 2053 (36%) | 1154 (20.3%) | 368 (6.5%) | |
65+ Age | 613 (30.7%) | 667 (33.4%) | 450 (22.5%) | 268 (13.4%) | |
Driver Age | |||||
18 Under | 361 (33.3%) | 447 (41.2%) | 230 (21.2%) | 47 (4.3%) | |
19 to 35 Age | 2277 (34.5%) | 2630 (39.8%) | 1312 (19.9%) | 385 (5.8%) | |
36 to 64 Age | 3184 (37.3%) | 3256 (38.2%) | 1661 (19.5%) | 425 (5%) | |
65+ Age | 1164 (37.7%) | 1232 (39.9%) | 571 (18.5%) | 118 (3.8%) | |
Pedestrian Gender | |||||
Female | 3247 (38.5%) | 3342 (39.6%) | 1520 (18%) | 329 (3.9%) | |
Male | 3739 (34.4%) | 4223 (38.9%) | 2254 (20.8%) | 646 (5.9%) | |
Driver Gender | |||||
Female | 3108 (38.1%) | 3284 (40.3%) | 1457 (17.9%) | 307 (3.8%) | |
Male | 3878 (34.8%) | 4281 (38.4%) | 2317 (20.8%) | 668 (6%) | |
Lighting Condition | |||||
Daylight | 3918 (37.7%) | 4514 (43.4%) | 1717 (16.5%) | 252 (2.4%) | |
Dark - Has Lights | 2253 (36.3%) | 2168 (35%) | 1387 (22.4%) | 395 (6.4%) | |
Dark - No Lights | 417 (25.2%) | 489 (29.6%) | 464 (28.1%) | 283 (17.1%) | |
Dawn/Dusk | 378 (39%) | 367 (37.9%) | 183 (18.9%) | 41 (4.2%) | |
Weather Conditions | |||||
Clear | 4216 (34.7%) | 4902 (40.3%) | 2417 (19.9%) | 614 (5.1%) | |
Raining | 1619 (41.7%) | 1411 (36.4%) | 683 (17.6%) | 166 (4.3%) | |
Cloudy or Overcast | 1033 (35.4%) | 1106 (38%) | 609 (20.9%) | 166 (5.7%) | |
Fog or Smoke | 46 (28.9%) | 64 (40.3%) | 31 (19.5%) | 18 (11.3%) | |
Snow or Freezing Rain | 72 (36.2%) | 82 (41.2%) | 34 (17.1%) | 11 (5.5%) | |
Road Surface Conditions | |||||
Dry | 4594 (34.3%) | 5420 (40.4%) | 2695 (20.1%) | 704 (5.2%) | |
Wet | 2243 (40.9%) | 1990 (36.2%) | 1003 (18.3%) | 254 (4.6%) | |
Ice/Snow/Slush | 149 (37.5%) | 155 (39%) | 76 (19.1%) | 17 (4.3%) | |
State | |||||
Idaho | 896 (27.9%) | 1230 (38.4%) | 875 (27.3%) | 205 (6.4%) | |
Oregon | 104 (26.8%) | 158 (40.7%) | 88 (22.7%) | 38 (9.8%) | |
Washington | 5986 (38.1%) | 6177 (39.3%) | 2811 (17.9%) | 732 (4.7%) |
The chart below summarize the percentage of total pedestrian involved crashes that are fatal by vehicle type. This chart shows graphically information presented above and highlights how for larger vehicles types, the percentage of crashes that are fatal is higher compared to passenger vehicles. In Figure 5.1, of all the pedestrian injury crashes with passenger cars only 3% are fatal compared to 7% when a pickup is involved a nearly doubling of the rate of fatal injury. The difference is smaller for SUvs, CUVs, and Vans.
Figure 5.1: Proportion of Pedestrian Injuries Classified as Fatal by Striking Vehicle Body Type
6 Logistic Regression
This report relies on logistic regression to parse the effects on pedestrian injury severity from the various factors available for analysis, but some less complicated analysis is also presented in this report. Multiple analytic approaches have been employed to understand risk factors for pedestrian injury crashes including fixed and mixed effects multivariate logistic regression, ordered logistical regression and ordered probit regression. Authors using fixed effects logistic regression (9, 10, 11,14,15) were common though many also used mixed ordered logistic regression (12, 13, 16, 25).
Some of the previous research on this topic have compared the results of logistic regression versus more sophisticated approaches using random parameters or ordered logistic or probit regression reporting very small differences between the approaches. For instance, when one study (13) compared fixed effects multinomial logistic regression with a mixed effects model and an ordered model and reported only marginal gains in terms of root mean squared error and log-likelihood values. When researchers used both fixed and mixed effects ordered logistic regression and reported nearly identical elasticity effects for the effect of vehicle type though speed effects were greater in the mixed effects model (12). Another analysis compared unordered logistic regression with ordered logistic regression and finding that many of the estimated odds ratios were similar (16). For instance, the effect of pickup trucks striking the pedestrian in the unordered and ordered models were identical (suggesting perhaps a reporting error in the paper) while the odds ratio of speed limit was only marginally different with as 1.5 and 1.6 respectively.
Given that fixed effects multinomial logistic (MNL) models precedence as a sound analytic framework to understand factors associated with serious and fatal pedestrian injury this modeling approach will be used in this paper. Multinomial logit models are estimated in the open-source statistical computing platform R using the nnet package in R (26) and summarized below in Table 2. The multinomial logit model takes the form described in equation 1 below where injury severity S for injury in:
\[ \begin{aligned} S_{in} = X_{in}\beta_{i} + \epsilon_{in} \end{aligned} \] where:
\[ \begin{align} &X_{i} && \text{is a vector of explanatory variables}\\ &\beta_{i} && \text{is a vector of parameters to be estimated}\\ &\epsilon_{in} && \text{is a type 1 extreme value distributed error term} \end{align} \]
Given the error is Type 1 extreme the underlying probabilities are computed based on the logit probability of injury severity ii for crash ni:
\[ \begin{align} P_n(i) = \frac{e^{(X_{in}\beta_i)}}{\sum_{I} e^({X_{I}\beta_{I})}} \end{align} \]
The model’s response variable includes four levels of pedestrian injury severity including possible injury, minor injury, serious injury and fatal injury with possible injury as the reference category. The results in Table 2 are presented as odds ratios which shows how much the odds change when the independent variable is changes by one unit or in the case of factor variables, compared to the reference value. Odds ratios greater than 1.0 indicate an increase probability of the injury type when the given independent variable increases by one unit and an odds ratio of less than 1.0 indicates a decrease probability of the injury type when the independent variable increases by one unit. As mentioned, the results for factor or categorical variables must be interpreted relative to their reference category (27) and also in reference to the reference injury severity which in this research is possible injury. These results are summarized in the next section.
Model Results Following several exploratory analyses and model specifications, five separate models are presented below. The five models below vary by vehicle design characteristic variable inputs and described below in the bullet points:
- Body Type Only – Use just body type for vehicle characteristic input into model.
- Overall Height Only - Use just overall height of vehicle for vehicle characteristic input into model.
- Body Type + Overall Height - Use both body type and overall height for vehicle characteristics input into model.
- Curb Weight Only - Use curb weight for vehicle characteristic input into model.
- Body Type + Curb Weight - Use both body type and curb weight for vehicle characteristics input into model.
The model variables were chosen through a literature review, incorporating factors previously used in pedestrian injury severity models, and also based on inputs from agency staff interested in understanding the relationships with injury severity of various factors. Models were selected after trying other combinations of vehicle design factors but keeping roadway characteristics and driver and pedestrian characteristics present in the models to determine which models performed best based on Akaike Information Criterion (AIC) and Nagelkerke R2 measures. Models with both curb weight and overall height were discarded since these two variables exhibited collinearity with a Pearson’s correlation coefficient of 0.78 (p <.05).
The results presented below summarize the odds ratios for each of the five models as well as the level of statistical significance represented by the number of asterisks and finally the standard error displayed in parentheses.
These results summarize the odds ratios for each injury severity and the discussion here will focus on the fatal injury severity column. Vehicle design characteristics consistently increase the odds of a fatal injury, all else being equal, in each of the models. For body type, compared to a passenger vehicle (reference category), pickup truck body types increase the odds of a fatal pedestrian injury by as much as 2.5 but less when controlling for overall height and curb weight where odds of fatal injury are 1.57 and 1.98 respectively. SUVs and CUVs increase the odds of a fatal pedestrian injury by at least 1.17 but potentially as much as 1.81 while van body types increase the odds of a fatal pedestrian injury by between 1.11 and 1.71. For every additional inch in overall vehicle height, the odds of fatal pedestrian injury increases by 1.035 to 1.04 and for the curb weight parameter estimate, for every 1,000 pounds increase, the odds of a fatal injury increases by 1.14 to 1.37. These factors are all associated with pedestrian fatal injury at the 0.05 level of statistical significance. Posted speed limit parameter results are consistent across models with the odds of a fatal pedestrian injury increasing by about 1.09 for every mile per hour increase. Therefore, an increase in speed limit from 25 mph to 35 mph would increase odds of fatal pedestrian injury by nearly a factor of two. For roadway location type, compared to roadway intersections, pedestrian crashes on driveways increase the odds of a fatal pedestrian injury by 1.48 to 1.60 while pedestrian crashes on segments increase the odds of fatal injury by a factor of three. Lighting conditions also significantly increase the odds of fatal injury with dark conditions without lighting increasing the odds of an injury by a factor of 2 with dark conditions and no lighting even worse with the odds climbing to a factor of nearly four times compared to daylight conditions. Driver and pedestrian characteristics were significant contributors to the odds of pedestrian crashes being fatal with impairment and age notable elements. When the driver was impaired the odds that the pedestrian was fatally injured increased by 1.24 to 1.29 and when a pedestrian was impaired the odds of fatal injury doubled. Compared to drivers 18 years of age or less, older drivers decreased the odds of pedestrian fatal injury. The age of the pedestrian was an important contributor to the odds of being fatal injured with pedestrians under 12 years of age having between 1.24 to 1.35 higher odds then the reference age group (12 to 22 years). Pedestrians over 65 were 10 times more likely to die in the event of a collision with a vehicle, all else being equal. For the serious injury response, it should be noted that most of the vehicle characteristic parameters were associated with increasing the odds of that injury type similar to the fatal injury response with some exceptions. Other variables were included as controls such as weather conditions and road surface conditions as well as a variable for the crash year and state since these variables could affect the variation in the response variable.
| Dependent Variable |
Minor Injury |
Serious Injury |
Fatal Injury |
Minor Injury |
Serious Injury |
Fatal Injury |
Minor Injury |
Serious Injury |
Fatal Injury |
Minor Injury |
Serious Injury |
Fatal Injury |
Minor Injury |
Serious Injury |
Fatal Injury |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Vehicle Characteristics | |||||||||||||||
| Pickup (Ref: Car) | 1.018 (0.013) | 1.335*** (0.01) | 2.458*** (0.001) | 0.799*** (0.01) | 1.092*** (0.007) | 1.575*** (0) | 1.025. (0.013) | 1.215*** (0.009) | 1.982*** (0.009) | ||||||
| SUV or CUV | 1.017 (0.026) | 1.067*** (0.014) | 1.813*** (0.001) | 0.912** (0.029) | 0.906*** (0.017) | 1.17*** (0.001) | 1.019 (0.024) | 1.014 (0.014) | 1.591*** (0.002) | ||||||
| Van | 1.117*** (0.004) | 1.085*** (0.002) | 1.711*** (0) | 1.012 (0.01) | 0.901*** (0.005) | 1.11*** (0) | 1.123*** (0.004) | 1 (0.002) | 1.453*** (0.001) | ||||||
| Striking Vehicle Overall Height (in) | 1.003 (0.002) | 1.011*** (0.003) | 1.044*** (0.005) | 1.009** (0.003) | 1.014*** (0.003) | 1.037*** (0.005) | |||||||||
| Curb Weight ( 000s lbs.) | 1.016 (0.018) | 1.101*** (0.022) | 1.37*** (0.036) | 0.998 (0.017) | 1.063** (0.021) | 1.143*** (0.034) | |||||||||
| Striking Vehicle Model Year | 1 (0.003) | 0.976*** (0.003) | 0.979*** (0.006) | 0.999 (0.003) | 0.973*** (0.003) | 0.97*** (0.006) | 1 (0.003) | 0.971*** (0.004) | 0.976*** (0.007) | 0.999 (0.003) | 0.973*** (0.003) | 0.971*** (0.006) | 1 (0.003) | 0.974*** (0.003) | 0.975*** (0.006) |
| Roadway Characteristics | |||||||||||||||
| Crash at Driveway (Ref: Intersection) | 0.882*** (0.003) | 1.058*** (0.002) | 1.479*** (0) | 0.915*** (0.003) | 1.114*** (0.002) | 1.483*** (0) | 0.908*** (0.004) | 1.036*** (0.002) | 1.598*** (0) | 0.892*** (0.003) | 1.072*** (0.002) | 1.425*** (0.001) | 0.897*** (0.003) | 1.086*** (0.002) | 1.445*** (0.001) |
| Crash at Segment | 1.292*** (0.024) | 2.161*** (0.016) | 3.188*** (0.002) | 1.307*** (0.025) | 2.161*** (0.017) | 2.986*** (0.002) | 1.314*** (0.027) | 2.236*** (0.018) | 3.18*** (0.003) | 1.29*** (0.025) | 2.148*** (0.017) | 3.083*** (0.003) | 1.292*** (0.025) | 2.157*** (0.017) | 3.17*** (0.003) |
| Posted Speed Limit | 1.008*** (0.002) | 1.037*** (0.003) | 1.092*** (0.004) | 1.008** (0.003) | 1.036*** (0.003) | 1.091*** (0.004) | 1.005. (0.003) | 1.035*** (0.003) | 1.092*** (0.004) | 1.008** (0.002) | 1.036*** (0.003) | 1.09*** (0.004) | 1.008*** (0.002) | 1.037*** (0.003) | 1.091*** (0.004) |
| Lighting - Dark w/ Lights (Ref: Daylight) | 0.888*** (0.026) | 1.437*** (0.021) | 2.909*** (0.004) | 0.895*** (0.027) | 1.419*** (0.021) | 2.738*** (0.004) | 0.917** (0.029) | 1.46*** (0.022) | 3.189*** (0.004) | 0.888*** (0.026) | 1.432*** (0.021) | 2.809*** (0.004) | 0.885*** (0.026) | 1.435*** (0.021) | 2.854*** (0.004) |
| Lighting - Dark no Lights | 0.937*** (0.008) | 1.512*** (0.008) | 3.926*** (0.001) | 0.921*** (0.009) | 1.54*** (0.008) | 3.845*** (0.002) | 0.899*** (0.009) | 1.539*** (0.009) | 3.98*** (0.002) | 0.928*** (0.009) | 1.504*** (0.008) | 3.812*** (0.002) | 0.936*** (0.009) | 1.519*** (0.008) | 3.881*** (0.002) |
| Lighting - Dawn or Dusk | 0.852*** (0.003) | 1.127*** (0.002) | 1.841*** (0) | 0.874*** (0.003) | 1.132*** (0.002) | 1.785*** (0) | 0.875*** (0.003) | 1.265*** (0.002) | 2.066*** (0) | 0.856*** (0.003) | 1.114*** (0.002) | 1.806*** (0) | 0.848*** (0.003) | 1.105*** (0.002) | 1.744*** (0) |
| Snow or Slush Road Surface (Ref: Dry) | 0.743*** (0.001) | 0.523*** (0.001) | 0.258*** (0) | 0.787*** (0.001) | 0.53*** (0.001) | 0.264*** (0) | 0.838*** (0.001) | 0.546*** (0.001) | 0.255*** (0) | 0.759*** (0.001) | 0.528*** (0.001) | 0.281*** (0) | 0.709*** (0.001) | 0.519*** (0.001) | 0.253*** (0) |
| Wet Road Surface | 0.827*** (0.021) | 0.84*** (0.022) | 0.679*** (0.005) | 0.836*** (0.021) | 0.83*** (0.022) | 0.672*** (0.005) | 0.889*** (0.023) | 0.865*** (0.025) | 0.659*** (0.005) | 0.83*** (0.021) | 0.841*** (0.022) | 0.688*** (0.005) | 0.838*** (0.021) | 0.86*** (0.022) | 0.738*** (0.005) |
| Driver & Pedestrian Characteristics | |||||||||||||||
| Driver Age 19 to 35 Age (Ref: 18 or younger) | 1.009 (0.021) | 0.965. (0.022) | 1.016* (0.006) | 0.998 (0.022) | 0.959. (0.022) | 1.007 (0.007) | 0.977 (0.024) | 0.964 (0.024) | 0.827*** (0.007) | 1.004 (0.022) | 0.964. (0.022) | 0.992 (0.007) | 1.012 (0.022) | 0.96. (0.022) | 1.008 (0.007) |
| Driver Age 36 to 64 | 0.902*** (0.021) | 0.91*** (0.023) | 0.706*** (0.007) | 0.901*** (0.022) | 0.902*** (0.024) | 0.714*** (0.007) | 0.875*** (0.023) | 0.909*** (0.026) | 0.573*** (0.007) | 0.901*** (0.021) | 0.907*** (0.023) | 0.712*** (0.008) | 0.904*** (0.021) | 0.902*** (0.023) | 0.697*** (0.008) |
| Driver Age 65+ | 0.932** (0.022) | 0.89*** (0.012) | 0.643*** (0.001) | 0.916*** (0.022) | 0.875*** (0.012) | 0.63*** (0.001) | 0.896*** (0.024) | 0.9*** (0.013) | 0.488*** (0.001) | 0.935** (0.022) | 0.892*** (0.012) | 0.64*** (0.001) | 0.935** (0.022) | 0.89*** (0.012) | 0.639*** (0.001) |
| Driver Is Impaired (Ref: Not Impaired) | 1.272*** (0.015) | 1.676*** (0.017) | 1.244*** (0.003) | 1.218*** (0.016) | 1.681*** (0.018) | 1.245*** (0.003) | 1.22*** (0.017) | 1.57*** (0.019) | 1.237*** (0.003) | 1.244*** (0.016) | 1.7*** (0.018) | 1.283*** (0.003) | 1.211*** (0.016) | 1.654*** (0.018) | 1.296*** (0.003) |
| Ped Age 12 or younger (Ref: 23 to 39) | 1.346*** (0.012) | 1.203*** (0.007) | 1.33*** (0) | 1.324*** (0.013) | 1.168*** (0.007) | 1.292*** (0.001) | 1.345*** (0.014) | 1.184*** (0.008) | 1.25*** (0) | 1.349*** (0.013) | 1.209*** (0.007) | 1.33*** (0.001) | 1.368*** (0.013) | 1.211*** (0.007) | 1.355*** (0.001) |
| Ped Age 12 to 22 | 1.228*** (0.024) | 1.195*** (0.014) | 0.657*** (0.001) | 1.214*** (0.025) | 1.173*** (0.015) | 0.686*** (0.001) | 1.218*** (0.027) | 1.177*** (0.016) | 0.558*** (0.001) | 1.233*** (0.024) | 1.203*** (0.015) | 0.702*** (0.001) | 1.242*** (0.024) | 1.195*** (0.015) | 0.665*** (0.001) |
| Ped Age 40 to 64 | 0.998 (0.024) | 1.236*** (0.015) | 2.131*** (0.003) | 0.977 (0.025) | 1.2*** (0.015) | 2.053*** (0.003) | 0.959 (0.027) | 1.191*** (0.016) | 1.919*** (0.003) | 0.998 (0.024) | 1.231*** (0.015) | 2.089*** (0.003) | 1.003 (0.024) | 1.232*** (0.015) | 2.099*** (0.003) |
| Ped Age 65 Plus | 1.138*** (0.013) | 2.012*** (0.008) | 10.827*** (0.002) | 1.133*** (0.013) | 1.977*** (0.008) | 9.787*** (0.002) | 1.159*** (0.015) | 1.998*** (0.009) | 10.073*** (0.003) | 1.14*** (0.013) | 2.017*** (0.008) | 10.37*** (0.002) | 1.15*** (0.013) | 2.025*** (0.008) | 10.681*** (0.002) |
| Pedestrian Is Impaired | 1.036* (0.016) | 1.22*** (0.018) | 2.579*** (0.003) | 1.077*** (0.016) | 1.217*** (0.019) | 2.632*** (0.003) | 1.051** (0.017) | 1.241*** (0.02) | 2.657*** (0.003) | 1.058*** (0.016) | 1.209*** (0.019) | 2.528*** (0.004) | 1.103*** (0.016) | 1.249*** (0.019) | 2.531*** (0.003) |
| Driver is Male (Ref: Female) | 1.064* (0.024) | 1.121*** (0.014) | 1.372*** (0.002) | 1.044. (0.025) | 1.126*** (0.015) | 1.369*** (0.001) | 1.018 (0.028) | 1.093*** (0.016) | 1.329*** (0.002) | 1.059* (0.025) | 1.139*** (0.015) | 1.381*** (0.003) | 1.062* (0.025) | 1.112*** (0.015) | 1.352*** (0.004) |
| Pedestrian Is Male (Ref: Female) | 1.045. (0.025) | 1.07*** (0.015) | 1.127*** (0.002) | 1.042 (0.026) | 1.079*** (0.015) | 1.104*** (0.002) | 1.06* (0.028) | 1.114*** (0.017) | 1.08*** (0.003) | 1.047. (0.025) | 1.071*** (0.015) | 1.124*** (0.002) | 1.045. (0.025) | 1.071*** (0.015) | 1.122*** (0.002) |
| Environmental Characteristics | |||||||||||||||
| Cloudy or Overcast Weather (Ref: Clear) | 1.011 (0.03) | 1.047* (0.019) | 0.988*** (0.001) | 0.984 (0.031) | 1.054** (0.019) | 0.983*** (0.001) | 0.954 (0.034) | 1.071*** (0.021) | 1.003*** (0.001) | 1.007 (0.03) | 1.046* (0.019) | 0.984*** (0.001) | 1.013 (0.03) | 1.043* (0.019) | 0.976*** (0.001) |
| Fog or Smoke Weather | 1.282*** (0) | 1.006*** (0) | 1.332*** (0) | 1.24*** (0) | 0.966*** (0) | 1.322*** (0) | 1.137*** (0) | 0.85*** (0) | 1.251*** (0) | 1.259*** (0) | 1.002*** (0) | 1.344*** (0) | 1.272*** (0) | 1.005*** (0) | 1.317*** (0) |
| Raining Weather | 0.979 (0.018) | 0.852*** (0.018) | 0.824*** (0.004) | 0.967. (0.018) | 0.859*** (0.019) | 0.839*** (0.004) | 0.93*** (0.02) | 0.83*** (0.021) | 0.838*** (0.004) | 0.978 (0.018) | 0.857*** (0.018) | 0.828*** (0.004) | 0.967. (0.018) | 0.836*** (0.018) | 0.759*** (0.004) |
| Freezing Rain or Snow Weather | 1.201*** (0) | 0.857*** (0) | 0.93*** (0) | 1.086*** (0) | 0.868*** (0) | 0.929*** (0) | 1.157*** (0) | 0.858*** (0) | 0.812*** (0) | 1.193*** (0) | 0.873*** (0) | 0.94*** (0) | 1.269*** (0) | 0.889*** (0) | 0.933*** (0) |
| Other Controls | |||||||||||||||
| State is Oregon (Ref: Idaho) | 1.267*** (0.003) | 0.656*** (0.002) | 0.991*** (0) | 1.216*** (0.004) | 0.63*** (0.002) | 0.898*** (0) | 1.198*** (0.005) | 0.641*** (0.003) | 0.939*** (0) | 1.262*** (0.003) | 0.642*** (0.002) | 0.922*** (0.001) | 1.197*** (0.004) | 0.659*** (0.002) | 0.986*** (0.001) |
| State is Washington | 0.844*** (0.027) | 0.517*** (0.02) | 0.736*** (0.002) | 0.84*** (0.028) | 0.507*** (0.021) | 0.729*** (0.002) | 0.825*** (0.031) | 0.524*** (0.023) | 0.772*** (0.002) | 0.85*** (0.028) | 0.514*** (0.021) | 0.724*** (0.002) | 0.837*** (0.028) | 0.525*** (0.021) | 0.744*** (0.002) |
| Crash Year | 0.99*** (0.003) | 1.028*** (0.003) | 1.023*** (0.006) | 0.992** (0.003) | 1.03*** (0.003) | 1.035*** (0.006) | 0.998 (0.003) | 1.029*** (0.004) | 1.022** (0.007) | 0.99*** (0.003) | 1.03*** (0.003) | 1.031*** (0.006) | 0.996 (0.003) | 1.027*** (0.003) | 1.023*** (0.006) |
| Values in Parentheses are Standard Errors | |||||||||||||||
| Significance codes: 0.001 ‘’ 0.01 ’’ 0.05 ’’ 0.1 ‘.’ 1 | |||||||||||||||
6.1 Logistic Regression Results in Charts
The charts below show the identical information of that in the Table 6.1 but just shows it in graphical a chart instead of a table. The first chart highlights how vehicle design elements like body type, specifically pickup trucks, SUVs/CUVs, and vans increase pedestrian fatal injury risk as do increased height and weight. Model year, or the newer the vehicle decreases the odds of fatal injury perhaps due to different materials or technology present in newer vehicles.
The information in the chart below shows the logistic regression results for the roadway characteristics set of variables. From this chart you can observe that posted speed limit consistently relates to higher risk of pedestrian death and that segments present more risk of pedestrian death than intersections. Lastly, conditions where no street lighting is present also significantly increase risk of pedestrain fatal injury.
7 Marginal Effects of Speed and Vehicle Characteristics
The model results presented in the table above highlight the effects of various vehicle design, roadway, and driver and pedestrian characteristics on the odds of pedestrian crashes being fatal, serious, or minor. In addition to the results in Table 2, Figure 1 highlights how select factors come together to affect the probability of pedestrians being fatally injured in a crash. These figures highlight the marginal effects of vehicle characteristics and posted speed by holding other model parameters constant and varying only the vehicle and speed. These marginal effects are partial derivatives of the statistical models estimated above and help to better illustrate the impact of different features the estimated models.
The figures below show the results of the three models that include body type to highlight how the body type, weight, and height impact the probability of fatal pedestrian injury when other inputs are held constant. Each of the charts tell a similar story where increases in the posted speed limit increases the likelihood of fatal injury. In the first chart the probability of the pedestrian being fatally injured at 25 miles per hour is less than 5% but this increases by nearly four-fold as the speed doubles.
The chart below implements the Body Type + Overall Height model in a marginal effect test to highlight how the body type, overall height and posted speed limit all contribute chances of the pedestrian being killed when struck by a vehicle. For all body types, vehicles less than 4.5 feet in overall height have a lower probability of fatally injuring the pedestrian in the event of a crash while vehicles 6.5 ft increases nearly doubles the probability that the pedestrian will be killed.
The chart below utilizes the model that included both the body type and curb weight variables to show that both are important contributors to an increases probability of pedestrian injury fatality. Note that not all combinations of body type are presented since not all combinations were present in the estimation data. For instances there was not a passenger car weighing more than 2,000 pounds involved in any of the pedestrian crashes so there is no line representing this body type and vehicle weight combination.
8 Discussion and Recommendations
The results from the analysis above reveal the importance of considering vehicle characteristics when investigating variations in pedestrian injury severity. Past research has not always controlled specifically for vehicle characteristics that contribute to serious and fatal injury such as body type, weight, and height. With the inclusion of these vehicle characteristics, in addition to the roadway, pedestrian and driver, and environmental characteristics, this research aims to provide a model and method for how other state DOTs can conduct injury severity analysis.
The results of the five specified MNL models demonstrates results that are aligned with past research where vehicle type, namely SUVs, pickups and van increase the odds of pedestrian fatal injury. Results from past research documented the odds ratios of fatal pedestrian injury for LTVs, SUVs, and pickups ranging from 1.2 to 4.2 (9, 10, 11, 14,16). The research presented in this paper documented odds ratios for these vehicle types as ranging from 1.11 to 2.45, slightly lower but for the most part consistent with past research. Additionally, curb weight and overall height are also linked to an increase in odds that a pedestrian will be seriously or fatally injured in the event of a collision. Posted speed limit, as demonstrated in the literature review and in this study, is a powerful explanatory input into injury severity for pedestrians. Road authorities develop the rules they use to set speed limits and practices, such as using the 85th percentile of observed traffic speed, to set speed limits which does not allow for resetting speed limits to improve safety performance. Emerging practices in Oregon (28) that allow for setting the speed limit based on the surrounding built environmental context will hopefully help reduce operational speeds and avoid serious and fatal injuries for all road users.
The results of the models for other variables such as age and road conditions are consistent with past research (12, 13, 14, 16, 25). Lighting conditions are an important feature of the road for pedestrian safety and this research highlighted that when conditions are dark and there is no street lighting, pedestrians are more likely to be fatally injured. Pedestrian characteristics where very young and very old pedestrians are more likely to be fatally injured aligns with past research and should be considered by road authorities since the overall population is aging with many seniors looking to stay active by walking and biking to meet their daily physical activity needs (29). Like past research, this work demonstrated how impairment of either the drivers or pedestrian is more likely to result in a serious or fatal outcome, possibly due to other risky behaviors of the road users such as excessive speed for the driver or unexpected or erratic movements by the pedestrians. Road authorities should work with their state liquor and substance control agencies to find ways to disincentivize using alcohol or drugs while travelling by maintaining taxing and enforcement penalties while incentivizing safe travel options like transit and ride share. Male drivers and pedestrians are more likely to be injured as well potentially indicating excessive risk taking by this demographic group.
This research focused on the role of vehicle characteristics and posted speed limit on pedestrian injury. The results of this analysis implicate speed limit and vehicle as significant contributors to pedestrian fatality and strongly suggest that these parameters must be incorporated into safety intervention strategies. State road authorities do not have the kind of policy mechanisms that the federal government possess but can still intervene by establishing pricing, enforcement, and education policies that impact use and adoption of SUVs and pickups. Many states have a weight-mile tax for freight vehicles and a similar regiment of fees could be applied to passenger vehicles. Since heavier SUV and pickups body types exert more force in the event of a collision, moving violations for drivers of these vehicles could reflect the disparate impact of breaking the rules that govern the road so that speeding and reckless driving tickets cost more depending on the vehicle type. And similar to how fuel efficiency ratings are easily available for vehicles, a similar rating system should be explicit for car buyers so that they understand how the vehicle they are purchasing can increase the risk for other road users, especially people walking. With fuller information, consumers may elect to purchase different vehicles.
This research aimed to document the role of vehicle characteristics and speed in pedestrian injury severity but lacked vehicle information on every crash for all years of data for the three states. In some records VIN was not recorded which was especially true for Oregon data so vehicle characteristics were not able to be gathered for all crashes. Ensuring that all crashes with a minor injury or higher have the VIN for the vehicle involved will make analysis like that presented here easier to accomplish. Additionally, vehicle elements present in the NHTSA VIN decoding service that would be of high interest to pedestrian safety, like whether a vehicle has automatic emergency braking systems, do not appear to be well populated. It is not mandatory for vehicle manufacturers to report all data elements to NHTSA but this should change so the data contains all the vehicle characteristics of interest to traffic safety analysts.
9 References
Pedestrian Traffic Fatalities by State – 2022 Preliminary Data. Governor’s Highway Safety Association (2023). https://www.ghsa.org/sites/default/files/2023-06/GHSA%20-%20Pedestrian%20Traffic%20Fatalities%20by%20State%2C%202022%20Preliminary%20Data%20%28January-December%29.pdf. Accessed July 2023.
National Roadway Safety Strategy. United States Department of Transportation (2022). https://www.transportation.gov/sites/dot.gov/files/2022-02/USDOT-National-Roadway-Safety-Strategy.pdf. Accessed July 2023.
2022 EPA Automotive Trends Report. Greenhouse Gas Emissions, Fuel Economy, and Technology since 1975 (2022). Environmental Protection Agency. https://www.epa.gov/system/files/documents/2022-12/420r22029.pdf . Accessed July 2023.
White, M.J., (2004). The “Arms Race” on American roads: the effect of sport utility vehicles and pickup trucks on traffic safety. J. Law Econ. 47 (2), 333–355.
Tyndall, J. (2021). Pedestrian deaths and large vehicles. Economics of Transportation, 26:100219
Systemic Pedestrian Safety Analysis Report 893. National Cooperative Highway Research Program National (2018). https://highways.dot.gov/safety/data-analysis-tools/rsdp/rsdp-tools/nchrp-report-893-systemic-pedestrian-safety-analysis
The Highway Safety Manual (2010), American Association of State Highway Transportation Professionals, Washington, D.C., http://www.highwaysafetymanual.org
Ballesteros, M.F., Dischinger, P.C., Langenberg, P., Pedestrian injuries and vehicle type in Maryland, 1995–1999, Accident Analysis & Prevention, Volume 36, Issue 1, 2004,Pages 73-81.
Henary, B. Y., Crandall, J., Bhalla, K., Mock, C. N., & Roudsari, B. S. (2003). Child and adult pedestrian impact: the influence of vehicle type on injury severity. Annual proceedings. Association for the Advancement of Automotive Medicine, 47, 105–126.
Roudsari, B. S., Mock, C. N., Kaufman, R., Grossman, D., Henary, B. Y., & Crandall, J. (2004). Pedestrian crashes: higher injury severity and mortality rate for light truck vehicles compared with passenger vehicles. Injury prevention : journal of the International Society for Child and Adolescent Injury Prevention, 10(3), 154–158. https://doi.org/10.1136/ip.2003.003814
DiMaggio, C., Durkin, M., & Richardson, L.D. (2006) The association of light trucks and vans with paediatric pedestrian deaths, International Journal of Injury Control and Safety Promotion, 13:2, 95-99, DOI: 10.1080/17457300500310038
Eluru, N., Bhat, C.R., & Hensher, D.A. (2007).A mixed generalized ordered response model for examining pedestrian and bicyclist injury severity level in traffic crashes. Working Paper ITLS-WP-07-12. Institute of Transport and Logistics Studies.
Dong, C, Khattak, A.J., Shao, C. & Xie, K. (2019) Exploring the factors contribute to the injury severities of vulnerable roadway user involved crashes, International Journal of Injury Control and Safety Promotion, 26:3, 302-314, DOI: 10.1080/17457300.2019.1595665.
Batouli, G., Guo, M., Janson, B., & Marshall, (2020). W. Analysis of pedestrian-vehicle crash injury severity factors in Colorado 2006–2016. Accident Analysis & Prevention, Volume 148, 2020,105782. https://doi.org/10.1016/j.aap.2020.105782.
Malin, F., Silla, A. & Mladenović, M. Prevalence and factors associated with pedestrian fatalities and serious injuries: case Finland. (2020). Eur. Transp. Res. Rev. 12, 29. https://doi.org/10.1186/s12544-020-00411-z
Nasri, M., Aghabayk, K., Esmaili, A., and Shiwakoti, N. (2022). Using ordered and unordered logistic regressions to investigate risk factors associated with pedestrian crash injury severity in Victoria, Australia. Journal of Safety Research
Desapriya, E., Subzwari, S., Sasges, D., Basic, A., Alidina, A., Turcotte, K., & Pike, I. (2010). Do light truck vehicles (LTV) impose greater risk of pedestrian injury than passenger cars? A meta-analysis and systematic review. Traffic injury prevention, 11(1), 48–56. https://doi.org/10.1080/15389580903390623
Crocetta, G., Piantini, S., Pierini, M. & Simms, C. (2015). The influence of vehicle front-end design on pedestrian ground impact, Accident Analysis & Prevention, Volume 79, https://doi.org/10.1016/j.aap.2015.03.009.
Simms CK, Wood DP. (2006). Pedestrian Risk from Cars and Sport Utility Vehicles - A Comparative Analytical Study. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering. 220(8):1085-1100. doi:10.1243/09544070JAUTO319
Yin, S., Li, J., & Xu, J. (2017). Exploring the mechanisms of vehicle front-end shape on pedestrian head injuries caused by ground impact, Accident Analysis & Prevention, Volume 106, https://doi.org/10.1016/j.aap.2017.06.005.
Tamura, A., Nakahira, Y., Iwamoto, M., Watanabe, I., Miki, K., Hayashi, S., Kitagawa, Y., Yasuki, T., (2008). Analysis of traumatic brain injury due to primary head contact during vehicle-to-pedestrian impact. International Journal of Crashworthiness. 13 (4), 375–385.
Longhitano D, Henary B, Bhalla K, Ivarsson J, Crandall J. (2005). Influence of Vehicle Body Type on Pedestrian Injury Distribution. SAE Technical Paper Number 2005-01-1876; Warrendale, PA.
National Highway Transportation Safety Administration Vehicle Identification Number Decoding Service https://www.nhtsa.gov/vin-decoder
Van der Loo, M.P.J. (2014). The stringdist package for approximate string matching The R Journal. https://CRAN.R-project.org/package=stringdist
Phuksuksakul, N., Yasmin, S. Haque, M.M. (2023). A random parameters copula-based binary logit-generalized ordered logit model with parameterized dependency: Application to active traveler injury severity analysis, Analytic Methods in Accident Research, Volume 38,https://doi.org/10.1016/j.amar.2023.100266.
W. N. Venables and B. D. Ripley. (2002). Modern Applied Statistics with S. Springer Publishing. Ney York. https://www.stats.ox.ac.uk/pub/MASS4/
Greene W. H. (2003). Econometric analysis (5th ed.). Prentice Hall.
Oregon Department of Transportation (2022). Speed Zone Manual. https://www.oregon.gov/odot/Engineering/Docs_TrafficEng/Speed-Zone-Manual.pdf
American Association of Retired Persons. Walking is Exercise’s Leader of the Pack. https://www.aarp.org/pri/topics/health/prevention-wellness/walking-attitudes-habits-adults-50-older.html. Accessed July 2023.