980908

The Aggressivity of Light Trucks and Vans
in Traffic Crashes

Hampton C. Gabler

William T. Hollowell

U.S. National Highway Traffic Safety Administration

ABSTRACT

Light trucks and vans (LTVs) currently account for over one-third of registered U.S. passenger vehicles. Yet, collisions between cars and LTVs account for over one half of all fatalities in light vehicle-to-vehicle crashes. Nearly 60% of all fatalities in light vehicle side impacts occur when the striking vehicle is an LTV. This paper will examine this apparent incompatibility between cars and LTVs in traffic crashes. An analysis of U.S. crash statistics will be presented to explore the aggressivity of LTVs in impacts with cars and identify those design imbalances between the cars and LTVs, e.g., mass, stiffness, and geometry, which lead to these severe crash incompatibilities.

INTRODUCTION

During the past decade, a profound shift in the composition of the passenger vehicle fleet has been realized in the U.S.. Fueled by the growing popularity of pickup trucks, minivans, and, more recently, by sports utility vehicles, the demographics of the U.S. fleet are characterized by a growing population of light trucks and vans (LTVs). As a group, LTVs are heavier, of more rugged construction, and have higher ground clearance than the passenger cars with which they share the road. The concern is that these design features, introduced to allow specialized functions e.g. off-road driving, may make LTVs fundamentally incompatible with cars in highway crashes, and in some cases dangerous to the occupants of cars struck by LTVs.

NHTSA has initiated a research program to investigate the problem of aggressive vehicles in multi-vehicle crashes. The near term objective of this program is to identify and demonstrate the extent of the problem of incompatible vehicles in vehicle-to-vehicle collisions. The focus of this goal is to identify and characterize compatible vehicle designs with the expectation that improved vehicle compatibility will result in large reductions in crash related injuries. Specifically, the objective is to identify those vehicle structural categories, vehicle models, or vehicle design characteristics which are aggressive based upon crash statistics and crash test data. LTV-to-car collisions are one specific, but growing, aspect of this larger problem.

The goal of this paper is to examine LTV aggressivity in vehicle-to-vehicle crashes. The specific objectives are to define the nature of the problem through examination of crash statistics, and to explore the relationships between crash aggressivity and vehicle design characteristics.

 

The Demographics of LTV Aggressivity

As shown in Figure 1, registrations of LTVs currently account for over 1/3 of all light vehicle registrations (Polk, 1980-1996), and are a growing component of the U.S. fleet. During the period from 1980 to 1996, LTV vehicle registrations increased from 20 percent to 34 percent. Although LTVs only account for 1/3 of all registered vehicles, traffic crashes between an LTV and any other light vehicle now account for the majority of fatalities in vehicle-to-vehicle collisions. As shown in Table 1, in 1996 LTV-car crashes accounted for 5,259 fatalities while car-car crashes led to 4,013 deaths and LTV-LTV crashes resulted in 1,225 fatalities.


Figure 1. U.S. Light Truck and Van Registrations by Model Year from 1980-1996 as a fraction of light vehicle registrations (R.L. Polk Co., 1980-96).

 

 

A disproportionate number of the fatalities in LTV-car crashes are incurred by the car occupants. Of the 5259 fatalities in LTV-car crashes in 1996, 81 percent of the fatally-injured were occupants of the car. As shown in Figures 2 and 3 and tabulated in Table 2, side impacts in which an LTV was the bullet vehicle led to 56.9 percent of all fatalities in side struck vehicles. As illustrated in Figure 4 and tabulated in Table 3, frontal impacts in which an LTV was involved accounted for 59.7 percent of all fatalities in frontal-frontal impacts. These statistics suggest that LTVs and passenger cars are incompatible in traffic crashes, and that LTVs are the more aggressive of the two vehicle classes. In particular, crashes with an LTV cause a disproportionate number of vehicle-to-vehicle fatalities.

Year

All

car-car

All

car-LTV

All

LTV-LTV

Total

1980

6506

3580

510

10596

1981

6510

3292

482

10284

1982

5437

3452

556

9445

1983

5157

3408

505

9070

1984

5340

3540

593

9473

1985

5174

3608

635

9417

1986

5450

3895

660

10005

1987

5489

4277

788

10554

1988

5320

4676

802

10798

1989

5175

4730

861

10766

1990

4726

4719

867

10312

1991

4482

4297

873

9652

1992

4208

4421

804

9433

1993

4364

4451

977

9792

1994

4219

4972

1059

10250

1995

4097

5238

1183

10518

1996

4013

5259

1225

10497


Table 1. Fatalities in Light Vehicle-to-Vehicle Crashes


Figure 2. Side Impact Fatalities resulting from LTVs striking passenger cars and other LTVs as a fraction of total side impact fatalities (FARS 1980-96).


Figure 3. Distribution of Side Impact Fatalities (FARS 1980-96).


Figure 4. Distribution of Frontal-Frontal Impact Fatalities (FARS 1980-96).

Year

Car

into Car

Car

into LTV

LTV

into Car

LTV

into LTV

Total

1980

2071

170

962

78

3281

1981

2077

161

876

87

3201

1982

1881

174

1015

102

3172

1983

1848

190

1134

118

3290

1984

1996

153

1186

122

3457

1985

1943

178

1168

130

3419

1986

2149

192

1285

147

3773

1987

2121

246

1382

216

3965

1988

2026

262

1645

194

4127

1989

2144

231

1697

238

4310

1990

1976

255

1628

234

4093

1991

1812

216

1614

232

3874

1992

1705

252

1698

223

3878

1993

1759

224

1609

256

3848

1994

1667

229

1983

318

4197

1995

1749

237

2049

316

4351

1996

1613

276

2181

314

4384


Table 2. Light Vehicle-to-Vehicle Side Impacts: Fatalities in Side-Struck Vehicle

 

Year

Car-Car

Car-LTV

(LTV Fatals)

Car-LTV

(car Fatals)

LTV

-LTV

Total

1980

3395

521

1254

303

5473

1981

3331

422

1208

278

5239

1982

2731

463

1197

330

4721

1983

2465

400

1115

283

4263

1984

2515

434

1159

351

4459

1985

2420

415

1217

362

4414

1986

2440

461

1289

373

4563

1987

2493

483

1407

404

4787

1988

2398

487

1498

437

4820

1989

2238

481

1478

446

4643

1990

2056

500

1538

450

4544

1991

1964

437

1321

432

4154

1992

1857

448

1282

418

4005

1993

1899

439

1460

543

4341

1994

1956

442

1548

521

4467

1995

1692

510

1658

656

4516

1996

1776

453

1529

651

4409


Table 3. Fatalities in Light Vehicle-to-Vehicle Frontal Impacts

The incompatibility of LTVs and cars is a growing problem. As shown in Figure 5, LTV market share has risen steadily from 1980 to 1996 [1]. LTVs captured over 43 percent of all light vehicle sales in 1996. Comparison of LTV registrations and LTV-caused fatalities over the same period show that LTV impacts have always caused a disproportionate number of vehicle-to-vehicle fatalities. For example in 1980, LTVs accounted for 20 percent of the registered light vehicle fleet, but side impacts in which an LTV was the bullet vehicle led to 31 percent of all fatalities in side struck vehicles. The magnitude of this problem then is not only due to the aggressivity of LTVs in crashes, but also the result of the dramatic growth in the LTV fraction of the U.S. fleet.


Figure 5. U.S. Sales of LTVs from 1980-1996 expressed as a fraction of light vehicle market share (Automotive News Market Data Book).

 

 

 

Problem Definition

The research program examined U.S. crash statistics to determine the characteristics and extent of the vehicle compatibility problem. One obstacle to quantifying the compatibility of a vehicle is the lack of an accepted measure of compatibility. A primary objective of our research effort was to develop a clearly defined a metric for measurement of vehicle aggressivity. To date, the NHTSA aggressivity research program has developed two potential aggressivity metrics.

Option 1:

Option 2:

The first metric was used in our early Aggressivity research as reported at the 15th ESV conference [2]. For each vehicle make / model, this metric determines the number of fatalities in the collision partner resulting from collisions with the subject vehicle. Only two-vehicle crashes in which both vehicles were either a car or an LTV are considered. The fatality count is normalized by the total number of registrations of the subject vehicle so that vehicles with large populations are not unfairly penalized. Using this metric, the U.S. fleet was rank ordered by aggressivity as presented at the 15th ESV conference. This initial study indicated that LTVs as a group were twice as aggressive in crashes as passenger cars -- i.e., per vehicle, LTVs caused more than twice as many fatalities in their collision partners as do cars.

The second, more recent, metric represents a refinement to the earlier definition of aggressivity. The second metric defines aggressivity to be the number of driver fatalities in the collision partner normalized by the number of vehicle-to-vehicle crash involvements of the subject vehicle. Only two-vehicle crashes in which both vehicles were either a car or an LTV are considered in computing the fatality count and the crash involvement count. One of the confounding factors in determining aggressive vehicle designs is aggressive driver behavior. Because aggressive drivers are involved in more crashes than less aggressive drivers, normalizing by the number of crashes rather than vehicle registrations focuses the metric more on vehicle performance and less on driver behavior. Note also that the second metric keys on driver fatalities rather than all fatalities in the struck vehicle. Because all vehicles have only one driver, this refinement avoids any biases accruing from differences in vehicle occupancy rate between, for example, pickup trucks and minivans.

Approach

The analysis for the second metric used statistics from the 1991-94 Fatality Analysis Reporting System (FARS) to determine the number of fatalities, and statistics from the 1991-94 General Estimates System (GES) to determine the number of crash involvements. FARS provides a comprehensive census of all U.S. traffic related fatalities. The GES is a large sample of over 60,000 police reported crashes collected annually. The scope of our analysis was constrained to cars, light trucks, and vans under 10,000 pounds in Gross Vehicle Weight Rating. The focus was further narrowed to two vehicle collisions in which the vehicles were either cars or LTVs. The fatality counts in the struck vehicle were limited to driver fatalities.

Initial analysis of GES revealed that approximately half of the make and model codes in this database were listed as unknown. For those GES cases with valid Vehicle Identification Numbers (VINs), the make and model was obtained by decoding the VIN using a combination of the VINDICATOR code, developed by the Highway Loss Data Institute, and the VINA code, developed by the R.L. Polk Company [3]. However, even after decoding the VINs, approximately 20 percent of all vehicle make and models remained unknown. The number of crash involvements for all vehicles was weighted accordingly in order to preserve total number of crashes. Although this strategy maintains the total count of crash involvements, this approach has the disadvantage of preserving any reporting biases. An improved approach would be to explore the missing data as a function of vehicle body type and model year, and prorate unknown make-models within these categories if biases exist.

Overall Fleet Aggressivity Ranking.

The second metric, hereafter referred to as the aggressivity metric (AM), was used to rank order all passenger vehicles, cars and LTVs, by their relative aggressivity. Only current production vehicles with at least 10,000 police-reported crashes over the period of 1991-94 were included in the ranking. Figure 6 presents the top twenty vehicles from that ranking in descending order by their relative aggressivity. The most striking feature of Figure 6 is that every vehicle in the rank ordering of the top twenty aggressors is either a light truck or van.

Figure 6 suggests that LTVs as a group are more aggressive than passenger cars. To further pursue this observation, the aggressivity ranking was regrouped by vehicle family into five categories of LTVs – sports utility vehicles, full-sized pickups, small pickups, minivans, and full-sized vans – and four categories of passenger cars – large, midsize, compact, and subcompact. The categories assigned to each vehicle are as tabulated in the Automotive News Market Data Book [1]. This study groups luxury, near luxury, and large cars into a single large car category.


Figure 6. Aggressivity Ranking: Top 15 Vehicles (FARS/GES 1991-94).

 


Figure 7. Vehicle Aggressivity by Vehicle Category (FARS/GES 1991-94).

As shown in Figure 7, full-sized vans were found to be the most aggressive vehicle category with an AM = 2.47. This category was closely followed by Full-Size Pickups (AM=2.31), Sports-Utility Vehicles (AM = 1.91), and small pickups (AM = 1.53). Minivans were the least aggressive of all LTV groups with an average AM = 1.46. The AM of passenger cars was significantly lower and ranged from AM= 0.45 for subcompacts to AM=1.15 for large cars.

Vehicle weight is not always the overriding factor dictating aggressivity as clearly demonstrated by Figure 7. Mid-sized cars, e.g., the Ford Taurus, and the small pickups, e.g., the Ford Ranger, both have approximately the same curb weight of 3,000 pounds. However, small pickups (AM = 1.51) are over twice as aggressive as mid-sized cars (AM = 0.70). The higher aggressivity of the small pickup class may be due to its greater structural stiffness and its higher ride height.

Among cars, the Aggressivity Metric is a strong function of vehicle weight. AM for the large car category, e.g., the Ford Crown Victoria, is 1.15. This is two to three times higher than the AM for the subcompact car category, e.g., Geo Metro, which is 0.45. The conservation of momentum in a collision places smaller cars at a fundamental disadvantage when the collision partner is a heavier vehicle. The importance of car size in providing occupant protection has been demonstrated in several studies of the U.S. crash statistics [4,5].

 

Aggressivity by Impact Mode.

Having established that LTVs are incompatible with cars in traffic crashes, the next requirement was to determine the relationship between aggressivity and impact direction. Joksch, et al. [6], in a NHTSA-sponsored study, computed the ratio of driver fatalities in the striking vehicle vs. driver fatalities in the struck vehicle for cars versus each of three LTV categories. In this study, minivans and full-sized vans were aggregated into a single ‘van’ category, and small and full-sized pickups were grouped into a single ‘pickup’ category. The counts of fatalities were obtained from 1991-94 FARS. To eliminate the complication of determining occupant restraint effectiveness, the study excluded airbag-equipped vehicles. It should be noted in the discussion which follows that this analysis was based on small numbers of fatal crashes (on the order of a hundred for each case), and the results should be regarded as preliminary.

The ratio of striking-to-struck driver fatalities resulting from frontal-frontal impacts is presented in Figure 8. In collisions between sports-utility vehicles (SUVs) and cars, 5.6 drivers died in the car for every driver who was killed in the striking SUV. In collisions between vans and cars, 5.4 drivers died in the car for every driver who was killed in the striking van. In collisions between pickup trucks and cars, 3 drivers died in the car for every driver who was killed in the striking pickup. Clearly, the fatality toll in car-LTV frontal crashes is disproportionately shouldered by the drivers of passenger cars.

The ratio of striking-to-struck driver fatalities resulting from side impacts to the left (driver) side of the struck vehicle are presented in Figure 9. As a control configuration, note first that in car-to-car impacts approximately 6 side-struck drivers are fatally injured for every fatally-injured driver in the bullet car. This imbalance is not unexpected as the side structure of passenger vehicles provides little protection for the side-struck occupant when compared with the significantly greater protection afforded by the front structure to the bullet vehicle driver.

The analysis is even more startling for LTVs striking cars in side impact. As shown in Figure 9, 30 side-struck car drivers are fatally injured for every driver who dies in a bullet SUV. For every driver who dies in a striking van, 13 side-struck car drivers are fatally injuried. For every fatally-injured driver of a striking pickup truck, 25 side-struck car drivers are killed. By contrast, when cars strike LTVs in side impact, the ratio of striking car to struck LTV driver fatalities is roughly 1:1.


Figure 8. Ratio of Fatally-Injured Drivers in LTV-to-Car Frontal Collisions. FARS 1991-94. (Joksch, et al., 1997)


Figure 9. Ratio of Fatally-Injured Drivers in LTV-to-Car Left Side Impacts. FARS 1991-94. (Joksch, et al., 1997)

 

Why are LTVs more aggressive?

The preceding analysis of crash statistics has clearly demonstrated the incompatibility between cars and LTVs in highway crashes. Still remaining to be determined however are the design characteristics of LTVs which lead to their incompatibility with cars. In general, crash incompatibility arises due to the three factors:

The following section will examine the relationship between LTV-car compatibility and these sources of incompatibility.

Mass Incompatibility

LTVs are 900 pounds heavier than cars on average [5]. The conservation of momentum in a collision places smaller vehicles at a fundamental disadvantage when the collision partner is a heavier vehicle. As shown in Figure 10, LTVs, as a group, tend to be heavier than passenger cars [7]. Figure 10 crossplots AM as a function of vehicle weight, and demonstrates the strong relationship between mass and aggressivity. The mass incompatibility between cars and LTVs appears to be growing. As shown in Figure 11, during the time frame of 1985-1993, average mass of both cars and LTVs has steadily increased, and the weight mismatch between the two classes of vehicles has slowly grown as well.


Figure 10. Aggressivity as a function of Vehicle Mass.

 


Figure 11. Sales-averaged Mass of LTVs and Passenger Cars from 1985-93 (Kahane, 1997).

Stiffness Incompatibility

As a group, LTV frontal structures are more stiff than passenger cars. LTVs frequently use a stiff frame-rail design as opposed to the softer unibody design favored for cars. Drawing on NHTSA New Car Assessment Program crash test results, the linear stiffness of a selection of LTVs and cars was estimated using the following relationship:

k = (mv2) / x2 (Eqn. 1)

where m is the mass of the vehicle, v is the initial velocity of the vehicle, and x is the maximum dynamic crush of the vehicle. The relationship between linear stiffness and AM is shown in Figure 12. Figure 12 indicates that stiffness is a contributing factor to the aggressivity of a vehicle. Because the stiffness of a vehicle is also somewhat related to its mass, as shown in Figure 13, stiffness may not prove to be as dominant an aggressivity factor as mass. Although stiffness and mass are related in many cases, stiffness is not totally driven by the mass of the vehicle. Figure 13 shows that for any given mass, there is a wide distribution of stiffness values. For example for 1750 kg vehicles, the least stiff vehicles are passenger cars while the most stiff vehicles are LTVs.

Figure 14 compares the frontal stiffness of a Ford Taurus and a Ford Ranger pickup. Both vehicles have approximately the same mass, but note that the Ranger pickup is significantly stiffer than the Taurus. In a frontal collision between the two, the bulk of the crash energy would be absorbed by Taurus and the Taurus occupants. Far less energy would be absorbed by the Ranger. From a compatibility perspective, a more ideal scenario would be for the Taurus and Ranger structures to each share the crash energy rather than forcing one of the collision partners to absorb the bulk of the crash.


Figure 12. Aggressivity as a Function of Linear Stiffness as computed from NCAP crash test results.


Figure 13. Relationship between Frontal Stiffness and Vehicle Mass as determined from NCAP Crash Tests.

 


Figure 14. Frontal Stiffness: Small Pickup (Ford Ranger) vs. Midsize Car (Ford Taurus)

Geometric Incompatibility

LTVs, especially four-wheel drive sport utility vehicles, ride higher than cars. This creates a mismatch in the structural load paths in frontal impacts, and may prevent proper interaction of the two vehicle structures in a collision. In a side impact, this imbalance in ride height allows the LTV structure to override the car door sill, and contributes to the intrusion of the side-impacted vehicle.

Ideally, the ride height used in an analysis of this type would be the height of the forward-most load bearing structural member of the vehicle. The location of this forward-most structural element however has no precise definition, and must be estimated from other measurements. Some analyses have used bumper height as the height of this load bearing member. However, because in the U.S., the bumper must only meet a 2-½ mile/hour bumper impact standard, and LTVs have no bumper standard, our belief is that, with respect to occupant protection, bumpers are largely ornamental, and their location provides little evidence of the location of load bearing members. The rocker panel, on the other hand, is a much more substantial structural member, and because the rocker panel is typically lower than the forward-most structure, serves as a superior lower bound on the location of the frame structure.

Figure 15 shows that ride height is related somewhat with vehicle mass. For this analysis, ride height is defined to be the rocker panel height as measured at the trailing edge of the front wheel well [7]. However note that the rocker panel height across all masses of passenger cars is relatively consistent – perhaps due to the bumper standard with which all passenger cars must comply. On the other hand, LTVs, which have no bumper standard, exhibit a wide variation in ride height and are in general much higher than passenger cars.

Figure 16 presents average ride height by vehicle category. Sport utility vehicles have the highest ride height with an average rocker panel height of 390 mm. Subcompact cars have the lowest-riding height with an average rocker panel height of 175 mm. SUVs ride almost 200 mm higher than mid-sized cars – a geometric incompatibility that would readily permit the SUV to override any side structure in a car and directly strike the car occupant.


Figure 15. Relationship between Vehicle Mass and Ride Height as estimated by Rocker Panel Height. Reference: AAMA Vehicle Specification Sheets (1990-94).


Figure 16. Geometric Compatibility: Average Ride Height vs. Vehicle Category as determined from AAMA Vehicle Specification Sheets (1990-94).

It should be noted that the data for the preceding analysis was drawn from AAMA Vehicle Specification Sheets supplied by vehicle manufacturers, and collected in the NHTSA Vehicle Attributes Database [7]. While geometric data was available for most passenger car models, the Vehicle Specification sheets for LTVs was much more limited. The LTV data presented here was primarily obtained from foreign manufacturers, and contains no data on full-sized pickups or vans.

Discussion

The study presented in this paper based its measure of aggressivity upon fatalities per 1000 police reported crashes. No effort was made to control for the severity of the crashes as this information is not available in the GES files. Some make-model vehicles, such as high performance sports cars, may have more severe crashes more because of the driver than because of the vehicle structure. Normalizing fatalities by number of crash involvements removes much of this driver aggressivity effect but does not completely eliminate this effect. Future work will explore refinements to the aggressivity metric which account of crash severity in addition to crash frequency.

The aggressivity metric used in this study assumes that all make-models strikes the same cross-section of the vehicle population, i.e., the same proportion of small cars, large cars, minivans, pickups, and so forth. The influence of this assumption upon the aggressivity ranking will be explored in future work. Joksch (1997) has noted that the age distribution of struck drivers varies somewhat from striking make-model to make-model. As injury tolerance is a strong function of age, his analysis suggests an additional refinement to the aggressivity metric which accounts for the average age of struck drivers for a given striking make-model.

The crash statistics presented in this paper clearly demonstrate a clear incompatibility between cars and LTVs. A comparison of mass distribution, stiffness distribution, and ride height geometry confirm that these two categories of vehicles are incompatible from a design point-of-view. However, this study has not attempted to assign what proportion of the aggressivity of LTVs is a function of each of these three separate sources of incompatibility. To determine the relationship between LTV design features and crash aggressivity, NHTSA plans to conduct a series of LTV-to-car crash tests in conjunction with a series of finite element simulations of LTV-to-car crash events.

 

Future Work

NHTSA research to date on LTV aggressivity has focused on problem definition based upon U.S. crash statistics and their correlation with vehicle design features. Follow-on work will expand upon these initial analyses as a precursor to potential rulemaking in this area. Specific future tasks include:

Additional tests will be conducted based on results of the first test series.

 

Conclusions

This paper has examined the compatibility of LTVs and cars in vehicle-to-vehicle collisions. Using struck driver fatalties per crash involvment of the subject vehicle as an aggressivity metric, examination of U.S. crash statistics has clearly shown a striking incompatibility between cars and all categories of LTVs. LTVs now account for over one-third of light vehicles on U.S. highways, but collisions between cars and LTVs lead to over 50% of all fatalities in light vehicle-to-vehicle collisions. Furthermore, a disproportionate number of the fatalities in LTV-car crashes are incurred by the car occupants. A comparison of LTVs and cars reveals that the LTVs are more aggressive than cars for a number of reasons including their greater weight, stiffer structure, and higher ride height. This mismatch in design has serious consequences for crash safety as approximately one-half of all passenger vehicles sold in the U.S. are LTVs, and presents a growing source of incompatibility within the fleet.

Acknowledgements

The authors would like to acknowledge the contributions of the Volpe Transportation Systems Center to this study. In particular, we wish to acknowledge Hsi-sheng Hsia and Roger Flanders for their expert and insightful analyses of the FARS and GES databases which made this study possible. The authors also gratefully acknowledge Hans Joksch of UMTRI for his many important contributions to the analysis presented in this paper and for the insights he has shared with us during the preparation of this work.

 

References

  1. ____, Automotive News Market Data Book, Crain Communications, Inc., Detroit, MI., 1980-96.
  2. Hollowell, W.T., and Gabler, H.C., NHTSAs Vehicle Aggressivity and Compatibility Research Program", Proceedings of the Fifteenth International Enhanced Safety Vehicle Conference, Paper No. 96-S4-O-01. Melbourne, Australia (1996).
  3. ____, "VINA: Vehicle Identification Number Analysis Software", The R.L. Polk Company, June 20, 1997.
  4. Evans, L., "Driver and Fatality Risk in Two-Car Crashes Versus Mass Ratio Inferred", Accident Analysis and Preventation, Vol. 26, 1994, pp. 609-616.
  5. Kahane, C. J., "Relationships between Vehicle Size and Fatality Risk in Model Year 1985-93 Passenger Cars and Light Trucks", DOT HS 808 570, January 1997.
  6. Joksch, H., Massie, D., and Pichler, R., "Vehicle Aggressivity: Fleet Characterization Using Accident Data", University of Michigan Transportation Research Institute, Report No. UMTRI-97-16, September 1997
  7. McCullough, C.A., Hollowell, W.T., and Battisti, P. "Applications of NHTSA’s Vehicle Parameter Database", Society of Automotive Engineers Paper No. 950360.