Leonard Evans President, Science Serving Society President-Emeritus, International Traffic Medicine Association Bloomfield Hills, Michigan USA Devoted to adding reason and knowledge to public policy ScienceServingSociety.com School of Transportation Science and Engineering International Res. Inst. for Multidisciplinary Science Beihang University, Beijing 1 September 2015 Traffic safety: Key vehicle and driver behavior factors

 1.2 million people killed each year on the world’s roads Harm from traffic - almost incomprehensible magnitude  Injures vastly outnumber fatalities  Victims mainly healthy young people

 Key is knowledge gained from science What can be done to reduce this enormous harm?  A solid edifice of knowledge exists  Scientific literature since 1930s has papers on traffic safety  Is too often ignored – with tragic results

Bow of Titanic – 3.8 km under Atlantic Ocean. Photographed 2 Sept by Leonard Evans

Scene from 2000 Chinese movie The Road Home, (director Zhang Yimou)

Summarizes what science has taught about traffic safety

Captain Edward J. Smith 27JAN1850–15APR1912 Safety lessons from sinking of the Titanic Age when Titanic sailed 62 years 2.6 months Titanic leaving Belfast 2 April 1912

Titanic illustrates central distinction between:- 1. Crash Protection Given crash occurs, to reduce consequent harm 2. Crash Avoidance To prevent crash occurring Did Titanic’s superior crash protection  save 705 lives, or  cause over 1500 deaths? Did Titanic’s superior crash protection  save 705 lives, or  cause over 1500 deaths?

Early recognition of distinction between 1. Technology 2. How it is used – driver behavior

Now look at peer-reviewed technical literature All previous anecdotal Paper in Amercan Journal of Psychology

“More efficient brakes on an automobile will not in themselves make driving the automobile any safer.” A Theoretical Field-Analysis of Automobile Driving, American Journal of Psychology 1938

Claim that better brakes do not increase safety based on reason alone But is it possible to address empirically? Did not happen in 1940s 1950s 1960s 1970s 1980s then opportunity occurred!

Dramatic advance in brake technology Increases rollover risk by (39+16)% Large numbers of GM popular vehicles: 1991 MY – NONE had ABS 1992 MY – ALL had ABS Otherwise, no important differences ABS (antilock brakes)

Engineering factors affecting survival in crashes Biomechanics - the science of relationships between injuries and the mechanical forces that produce them:  Penetrating trauma  Blunt trauma

To reduce severity of blunt trauma – reduce force by reducing deceleration of body Spread stopping over as much distance, and time, as feasible through  occupant protection devices  vehicle characteristics and design

Scientific study needs data By far, most reliable data is for fatalities Even though death just tip of iceberg in numbers, much attention is devoted to it

Fatal Accident Reporting System Name changed Feb  Census of all US traffic crashes in which anyone is killed on a public road  Run by US National Highway Traffic Safety Administration (NHTSA)  Data from 1 January 1975  Over 1.6 million fatalities documented through 2013 Fatality Analysis Reporting System ( FARS )

Occupant protection device effectiveness Occupant protection device Occupant Effectiveness in preventing fatalities Airbag plus lap/shoulder belt Car driver (47  4)% Lap/shoulder beltCar driver ( 42  4)% Airbag onlyCar driver (13  4)% Lap beltRear-seat passenger (18  9)% Helmet Motorcycle riders (28  8)%

Occupant protection device effectiveness Occupant protection device Occupant Effectiveness in preventing fatalities Airbag plus lap/shoulder belt Car driver (47  4)% Lap/shoulder beltCar driver ( 42  4)% Airbag onlyCar driver (13  4)% Lap beltRear-seat passenger (18  9)% Helmet Motorcycle riders (28  8)%

Occupant protection device effectiveness Occupant protection device Occupant Effectiveness in preventing fatalities Airbag plus lap/shoulder belt Car driver (47  4)% Lap/shoulder beltCar driver ( 42  4)% Airbag onlyCar driver (13  4)% Lap beltRear-seat passenger (18  9)% Helmet Motorcycle riders (28  8)%

Occupant protection device effectiveness Occupant protection device Occupant Effectiveness in preventing fatalities Airbag plus lap/shoulder belt Car driver (47  4)% Lap/shoulder beltCar driver ( 42  4)% Airbag onlyCar driver (13  4)% Lap beltRear-seat passenger (18  9)% Helmet Motorcycle riders (28  8)%

Occupant protection device effectiveness Occupant protection device Occupant Effectiveness in preventing fatalities Airbag plus lap/shoulder belt Car driver (47  4)% Lap/shoulder beltCar driver ( 42  4)% Airbag onlyCar driver (13  4)% Lap beltRear-seat passenger (18  9)% Helmet Motorcycle riders (28  8)%

The vehicle factor having largest influence on occupant risk is: Vehicle mass (weight) and size

Large Truck 12.7 Large Truck Medium Truck Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Large Truck Medium Truck Van Pickup Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Motorcycle Moped Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Medium Truck Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Motorcycle Moped Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Medium Truck Van Pickup Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Motorcycle Moped Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Motorcycle Moped Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Motorcycle Moped Large Truck Medium Truck Heavy Car Medium Car Light Car Van Pickup Motorcycle Moped Relative Driver Fatality Risk in One Vehicle Crashing into Another Vehicle

How are R and μ related? Two key measures: 1. mass ratio, μ = mass of car 2 mass of car 1 = m2m1m2m1 car 2 mass m 2 car 1 mass m 1 R = probability of driver fatality in car 1 probability of driver fatality in car 2 2. driver fatality risk ratio, R,

 v of 900 kg car 6% greater than  v of 1800 kg car Risk in 900 kg car 30% greater than risk in 1800 kg lb car due to 6% higher  v 900 kg car 14,000 kg truck 1800 kg car risk ratio = 44 risk ratio = 22

 What is the casual role of mass and size?  Am I safer if I put bricks in my trunk? Does NOT answer  How does change in RELATIVE risk affect each driver individually?

If cars differ only in mass, R = μ u If cars differ in another attribute, say one carries additional cargo R = probability of accompanied driver fatality in car 1 probability of lone driver fatality in car 2 R = A μ u A estimates effect passenger’s presence has on driver risk when car masses are the same car 2 car 1 driver + passenger lone driver

CONCLUSION Addition of passenger changes driver fatality ratio by -(14.5 ± 2.3)% BUT We do not know how much of this change is a risk reduction to the accompanied driver and a risk increase to the lone driver To answer, we model process

r 1,2 = x m2m1m2m1 ( ) 1.79 net effect = intrinsic size x intrinsic mass L 1 + L c

A picture is worth 100 equations

Car and Driver, April 1998

When it comes to vehicle safety, putting on weight is good for your health

Driver performance Driver behavior

Drivers with  Best visual acuity  Fastest reaction times  Most knowledge/interest in vehicles Driver Performance -- have highest crash rates! (What the driver can do)

What is crucial is not what the driver can do -- (Driver Performance) Driver Behavior but What the driver does do

Legal limit in all US States

Named for town in Michigan Performed by Robert Borkenstein and colleagues

Effects of speed 1. Risk of crashing increases with increasing travel speed 2. If crash occurs, risk of injury increases with increasing speed 3. If injury occurs, risk that it is fatal increases with travel speed 1% increase in travel speed increases the risk of a fatal crash by at least 4%

Speed Belt wearing Drunk driving Benefits only relevant occupant Affects all road usersAffects all road users, although victim is usually the drunk driver (plus similarly drunk passengers) Affects ALL road users – is central

International comparisons offer insights into best ways to reduce harm in traffic

Because fatalities reach a maximum number, we can define a simple ratio:  Dimensionless (no units)  No arbitrary assumptions  Depends ONLY on fatalities Fatalities in a given year Maximum yearly fatalities  Plot* how this ratio varies in time after maximum is reached *on logarithm scale, and x100 to express as %

2013 value is 40% below 1972 maximum (32,719 down from 54,589)

This is 83.7% below The Netherlands’ all time high of 3,506 in 1972 The Netherlands had 570 traffic deaths in 2013 If US total had fallen by 83.7% from its high, US total in 2013 would be 8,875 (instead of 32,719 observed) More than 20,000 additional Americans are killed annually because US fails to match decline in Netherlands (+ 10 other countries)

USA disaster lies in ignoring basic scientific finding - which is:

 Vehicle factors are important USA disaster lies in ignoring basic scientific finding - which is:

 Vehicle factors are important, but less important than roadway factors USA disaster lies in ignoring basic scientific finding - which is:

 Vehicle factors are important, but less important than roadway factors  Roadway factors are important USA disaster lies in ignoring basic scientific finding - which is:

 Vehicle factors are important, but less important than roadway factors  Roadway factors are important, but less important than driver performance factors USA disaster lies in ignoring basic scientific finding - which is:

 Vehicle factors are important, but less important than roadway factors  Roadway factors are important, but less important than driver performance factors USA disaster lies in ignoring basic scientific finding - which is:  Driver performance factors are importantctors

 Vehicle factors are important, but less important than roadway factors  Roadway factors are important, but less important than driver performance factors USA disaster lies in ignoring basic scientific finding - which is:  Driver performance factors are important, but less important than driver behavior factors factors

Predictions Never make predictions — especially about the future Never make predictions

15 May 2015

Are already safer than human- controlled cars Cannot become universal for many decades System consisting exclusively of self-driving cars can approach perfect safety

This study, in agreement with earlier research, implies that there are over 200,000 annual traffic deaths in China. How can this number be reduced? Two recommendations Conclusion

1 Create data-collecting agency  USA data set FARS is good model 2 Create traffic-safety research organization  Goal - accuracy and completeness  Supported by government, but largely independent of government  Best model is SWOV in the Netherlands  Make solicited and unsolicited science-based recommendations  Evaluate safety programs (implementation and evaluation must be kept separate)  Goal – highest quality objective scientific research  40+ years of ongoing improvement  Free, comprehensive, and simple on web

Concluding comments  Traffic causes enormous numbers of injuries and deaths  Traffic-safety science can sharply reduced these numbers  Reliable data  “Disinterested” quality analysis  This requires

Summarizes what science has taught about traffic safety