How Risky Is Driving a Recalled Toyota? Insights from Data Professor Paul Fischbeck October 2010 Carnegie Mellon University Pittsburgh, PA 15213 412-268-3240 fischbeck@cmu.edu 1
Outline Risk is a ratio Risk varies based on measures and populations Finding answers with and without averages Comparison group Has the risk changed? Compared to what? Risk communication (The Press) Decision context Correcting misconceptions Public reaction to risk stories Searching for an external cause/justification/rationale for a bad outcome 2
Risk Is a Ratio Outcome Measure Exposure Measure Outcome Fatalities Injuries Crashes Exposure Population Registered vehicles Licensed drivers Vehicle mile Passenger mile Number of trips Time on the road 3
Ratio Selection Is Important Per population or per licensed driver measures can hide important driving characteristics Data is relatively easy to acquire Precise (little uncertainty) Cars or drivers that aren t on the road much get over weighted True exposure measures are hard to get Per mile, minute, trip are very limited and uncertain NHTS (National Household Transportation Survey: 2001, 2009) But provide insights On average, women drive shorter trips at slower speeds compared to men Compared to men, older women drivers (70-80) have much lower fatality risk when measured per trip or minute, but are higher per mile. 4
Comparing Apples What is the baseline risk? Risk varies greatly by age, location, time of day, month of year With whom do you compare? All licensed drivers? All drivers of the same make/model/year? All same-aged drivers of the same make/model/year? At the same time of day? Day of week? Gender? Prius fatalities Who drives a Prius? (10 years older) How is a Prius driven? (4,000 fewer miles per year) Does this matter? 5
Number of Drivers by Age 2,000,000 Males Females 1,500,000 1,000,000 500,000 0 6 16 18 20 22 24 30-34 40-44 50-54 60-64 70-74 80-84 Number of drivers per age year Age groups
% of Population with Driver Licenses 100 Males Females Many more men keep their licenses as they age 80 60 40 20 0 16 18 20 22 24 30-34 40-44 50-54 60-64 70-74 80-84 % drivers per age group Age groups 7
2.50 2.00 1.50 1.00 0.50-8 76 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 79 82 85 Fatality Risk per Year Age of Driver Risk (Fatalities per year per 1,000 drivers)
0.0035 0.0030 0.0025 0.0020 0.0015 0.0010 0.0005 0.0000 Males Females 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5-9 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 15 20 25 30 35 40 45 50 55 60 65 70 75 Driver Age Risk (Fatalities per year per licensed driver) Risk (Fatalities per 100 million driver miles) 80 Males Females Driver Age 40 35 30 25 20 15 10 5-15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 15 20 25 30 35 40 45 50 55 60 65 70 75 Risk (Fatalities per 100 million trips) 80 Gender Comparisons for Fatality Risks Using Different Measures Per Year for Licensed Drivers Per 100 mil Miles traveled Note that women exceed men only for this measure for older drivers. More men are licensed and they drive more. 0.0025 Per Year for Population Per 100 Trips taken 0.0020 0.0015 0.0010 0.0005 0.0000 Males Females Males Females Driver Age Driver Age Risk (Fatalities per year per capita
Measures Matter 75 year old men either have 20% less risk (per mile) or 80% greater risk (per trip) 10
Risk per Mile and Fatalities 6,000 Fatalities (1999-2005) 5,000 4,000 3,000 2,000 24 23 2726 28 25 29 30's & 40's 20 21 22 18 19 17 16 With our aging population, if the number of older drivers on the road increases, then the risk per mile may not change, but the number of fatalities would increase (shifting these points up) 50's 1,000 60's 70 72 71 74 73 75 77 78 76 79 80 81 84 83 82 85 0-1.00 2.00 3.00 4.00 5.00 6.00 Risk (Fatalties per 100 m illion driver m iles) 11
Risk per Mile and Miles/Year 20,000 18,000 40 28 Drivers with the most miles have the lowest risks per mile Miles per year per driver 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000-30's, 40's, 50's & 60's 29 27 26 70 72 24 23 71 25 74 73 75 20 21 22 77 As risks increase, people drive less, or As people drive less, risk increases? 76 78 18 79-1.00 2.00 3.00 4.00 5.00 6.00 19 17 16 80 81 Older drivers drive the least by far, so even though they are high risk per mile, they are less risky when looked at per year. 83 84 82 85 Risk (Fatalities/100 m illion driver m iles) 12
Risk per Year and Fatalities 6,000 5,000 20 21 19 18 Fatalites (1999-2005) 4,000 3,000 2,000 30's & 40's 50's 25 27 29 26 28 24 23 22 When compared using risk per year, 80-year olds are like 24-year olds. 17 16 1,000 60's 72 73 70 71 74 75 76 77 78 79 80 81 83 84 82 85 0-0.50 1.00 1.50 2.00 Risk (Fatalities per year per 1,000 drivers) 13
Risk per Mile and Risk per Year 2.50 Risk (Fatalites per year per 1,000 drivers) 2.00 1.50 1.00 0.50 When compared using risk per year, 80-year olds are like 24-year olds 28 29 24 27 26 25 72 71 70 74 73 75 20 21 23 22 78 77 76 18 79 19 17 80 81 16 When compared using risk per mile, 80-year olds are like 17-year olds 84 83 82 85-30's, 40's, 50's & 60's - 1.00 2.00 3.00 4.00 5.00 6.00 Risk (Fatalities per 100 million driver miles) 14
Types of Fatal Crashes Varies by Age 0.9 Percent of All Crash Types 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Multi-vehicle Single vehicle Rollover Other 0 Ages 16-20 Ages 21-24 Ages 25-34 Ages 35-44 Ages 45-54 Ages 55-64 Ages 65-74 Ages 75-84 Ages 85+ 15
Driving Risk and Time of Day 14,000 Vehicle occupant fatalities (1999-2004) 13,000 12,000 11,000 10,000 9,000 8,000 7,000 6,000 5,000 16:00 17:00 15:00 14:00 18:00 20:00 13:00 21:00 19:00 12:00 22:00 11:00 10:00 7:00 6:00 8:00 9:00 1:00 23:00 2:00 0:00 20 X Riskier 3:00 5:00 4:00 4,000 0 1 10 100 Fatalities/100 million passsenger miles 16
Age 75+ Driver Risk 10,000 1,000 9:00-17:00 Fatalities (1995-2005) 100 18:00 8:00 7:00 19:00 20:00 21:00 6:00 22:00 5:00 23:00 0:00 4:00 1:00 3:00 2:00 10 1 1 10 100 1,000 Driver risk (Fatalities per 100 million miles) 17
Comparison across Time of Day: When are miles driven 10% 9% 17:00 % of age 16-64 miles driven 8% 7% 6% 5% 4% 3% 2% 1% 7:00 8:00 6:00 20:00 21:00 22:00 0:00 23:00 5:00 18:00 19:00 16:00 15:00 12:00 13:00 14:00 9:00 10:00 11:00 0% 1:00-4:00 0% 2% 4% 6% 8% 10% % of age 65+ miles driven 18
Comparison across Time of Day: When do fatalities occur? 10% 9% 8% % of age 16-64 fatalities 7% 6% 5% 4% 3% 23:00-2:00 22:00 20:00 21:00 19:00 3:00 6:00 7:00 5:00 4:00 18:00 8:00 17:00 9:00 16:00 12:00 10:00 15:00 14:00 13:00 11:00 2% 1% 0% 0% 2% 4% 6% 8% 10% % of age 65+ fatalities 19
Linear Relationship between Fatalities and Miles 12% 10% 8% 11:00 12:00 10:00 13:00 16:00 14:00 15:00 % of Miles 6% 19:00 18:00 17:00 9:00 4% 20:00 8:00 21:00 2% 22:00 7:00 23:00-5:00 6:00 0% 0% 2% 4% 6% 8% 10% 12% % of Fatalities 20
Younger Drivers Have Different Distributions 12% 10% Drivers 16-64 Drivers 16-64 65+ 14:00-18:00 8% % of Miles 6% 7:00-13:00 19:00 9:00-16:00 4% 20:00 6:00 21:00 21:00 2% 22:00 23:00-5:00 22:00-5:00 0% 0% 2% 4% 6% 8% 10% 12% % of Fatalities 21
400% 350% 300% 250% 200% 150% 100% 50% 0% -50% -100% Differences in Risk 22 0:00 1:00 19:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 20:00 21:00 22:00 23:00 Time of Day Relative to 16-64 year old drivers, risk of 65+ drivers
The Toyota Risk Problem appears to be (have been) mechanical, related to car mats and/or sticky pedals Toyota denies (has not found) any relationship to electronic equipment The short timeline of events 19 fatalities Over the past 3 years 2.3 million vehicles Toyotas recalled The longer timeline of events (NYTimes, 2 March 2010) 59 fatalities Over the past 10 years 8 million vehicles involved Which interpretation looks riskier? Even though the longer-term perspective has more fatalities, because it involves more vehicles over more years, the exposure is greater making the risk of fatality less per mile driven 23
Facts about Driving in the US 250 million vehicles on the road 210 million drivers (~1.20 vehicles per driver) Average vehicle drives 12,000 miles per year About 38,000 traffic-related fatalities in 2009 20,600 drivers of vehicles 9,200 passengers in vehicles 3,200 on motorcycles 4,200 pedestrians 700 on bicycles Fatalities associated with recalled vehicles adding 6 more deaths per year 24
Fatality Risk per Mile Average risk is 1.05 fatalities per 100 million miles driven, but the risk varies across types of vehicles Minivans 0.44 per 100 million miles SUVs 0.85 per 100 million miles Passenger cars 1.02 per 100 million miles Pick-up trucks 1.12 per 100 million miles Motorcycles 33.82 per 100 million miles Pick-up trucks are over 2 times more risky than minivans Risk is varies by vehicle type and the drivers who drive those vehicles and their behavior behind the wheel Teasing out which factors are important can be tough Minivans are big cars driven by mothers carrying children in the middle of the day 25
Additional Risk for an Average Driver from Driving a Recalled Toyota Using the short timeline values, (19 fatalities over 3 years and 2.3 million vehicles) Annually 32 billion miles driven by recalled Toyotas 315 expected deaths from non-recall-related crashes Additional 6 expected deaths from recall-related crashes 2% increase in risk measured by risk-per-mile From 1.05 to 1.07 fatalities per 100 million miles driven One recall-related fatality for every 5 billion miles driven If all cars in the country had this problem, there would be an additional 580 fatalities per year Additional annual risk from the problem is 2 in a million Annual risk of dying in a crash increases from 126 in a million to 128 in a million 26
To get a risk equal to the additional risk of driving a recalled Toyota for a year (2 in a million), you would have to travel Miles of Travel Equal to Fatality Risk of Driving a Recalled Toyota for a Year 10,000 1,000 100 10 1 0 0.5 6 10 15 250 2,000 Canoeing Motorcycle Walking Biking Driving Commercial Flying Relevant comparison? 27
The Additional Risk of Driving a Recalled Toyota for a Year is Equal to 2 in a Million To reduce your annual risk by 2 in a million, drive 250 fewer miles each year Do not decide to park your recalled Toyota and walk. Your risk will increase 19 times per mile Same annual risk as a US adult dying from meningitis Compared to the risk of recalled Toyota, the annual risk of dying from a heart attack is 90x greater for a 45-year old woman 900x greater for a 65-year old woman 300x greater for a 45-year old man 2,200x greater for a 65-year old man 28
Other Comparisons of Equal Risk Using a gambling perspective, 2 in a million is equivalent to getting 19 heads on 19 fair coin flips One parachute jump or one flight off/on an aircraft carrier is equivalent to the additional risk of driving a recalled Toyota for 25 years Being a cop for 2.5 days has the same risk Lightning strikes, living x years near a nuclear power plant, eating y jars of peanut butter.. How are these relevant? But they make good press Put risk into a decision context 29
Baseline Driving Risk Varies across Gender and Age Groups Fatalities per 100 million miles Ages 16-20 2.78 Ages 21-24 2.07 Ages 25-34 0.98 Ages 35-44 0.80 Ages 45-54 0.82 Ages 55-64 0.87 Ages 65-74 1.15 Ages 75-84 2.69 Ages 85+ 8.99 Males 1.31 Females 0.73 Average for the country is 1.05 fatalities per 100 million miles Younger drivers and older drivers have much higher risks of dying for each mile driven Men have a much higher risk than woman The recalled Toyota risk adds about 0.02 to these risks. 30
When Comparing Populations, Baseline Driving Risks Vary Considerably Geographic regions (fatalities per 100 million miles) New England (CT MA ME RI VT) 0.64 E South Central (AL KY MS TN) 1.71 Mountain (AZ CO ID MT NM UT WY) 1.38 Pacific (AK CA HI OR WA) 0.79 Combinations yield very large differences 30-year old female in New England 0.28 17-year old male in E South Central 16.40 The recalled Toyota risk adds about 0.02 to these risks. This additional 0.02 is a large relative risk increase for safe drivers, and small for risky drivers. 30-year old female in New England 7.0% increase 17-year old male in E South Central 0.1% increase 31
Additional Risk for an Average Driver from Driving a Recalled Toyota Using the longer timeline values, (59 fatalities over 10 years and 8 million vehicles) Annually 104 billion miles driven by recalled Toyotas 1,100 expected deaths from non-recall-related crashes 6 expected deaths from recall-related crashes ½ % increase in risk: From 1.05 to 1.06 fatalities per 100 million miles driven One recall-related fatality for every 12 billion miles driven If all cars in the country had this problem, there would be an additional 170 fatalities per year Additional annual fatality risk is less than 1 in a million Annual risk of dying in a crash increases from 126 in a million to just under 127 in a million 32
The Audi 5000 Problem (1985-6) Reported runaway acceleration leading to 7 fatalities and 700 accidents from 250,000 vehicles from 8 model years 60 Minutes rigged a news segment to highlight the problem Problem eventually determined to be a combination of pedal placement and operator error The accelerator and brake pedals on automatics were closer together than most vehicles. Used the pedal design found on manual clutch vehicles that facilitated heal & toe operation. Drivers pressed on the accelerator thinking that they were stepping on the brake pedal The faster the vehicle went, the harder the driver pressed on the brake pedal Problem fixed by redesign of pedals and the installation of an interlock requiring firm pressure on the brake pedal before shifting out of park for automatics. Sales dropped over 80% because of the negative publicity 33
Impact on Toyota Sales have dropped in January and February 2010, but have rebounded strongly in March with sales incentives Small change when compared to Audi Company is adding new incentives Average price of a new Prius declined several hundred dollars, but also affected by the lower gas prices Have become the punch line of many jokes and cartoons 34
3.50 2.50 1.50 0.50-0.50-1.50-2.50 Sales Trends: Deviation from two-year trend Honda Toyota Nissan 35 Jul-08 Jul-08 Aug-08 Aug-08 Sep-08 Oct-08 Nov-08 Dec-08 Jan-09 Feb-09 Mar-09 Apr-09 May-09 Jun-09 Jul-09 Aug-09 Sep-09 Oct-09 Nov-09 Dec-09 Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Jul-10 Aug-10 Sep-10 Oct-10 Sep-08 Oct-08 Nov-08 Dec-08 Jan-09 Feb-09 Mar-09 Apr-09 May-09 Jun-09 Jul-09 Aug-09 Sep-09 Oct-09 Nov-09 Dec-09 Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Jul-10 Aug-10 Sep-10 Oct-10 3.50 2.50 1.50 0.50-0.50-1.50-2.50 Cash for Clunkers Ford GM Toyota Toyota Recall Ford GM Honda Chrysler Toyota Nissan Ford 1.00 GM 0.81 1.00 Honda 0.21 0.04 1.00 Chrysler 0.54 0.54 0.03 1.00 Toyota 0.82 0.84 0.06 0.53 1.00 Nissan 0.02 0.09 0.71 0.19 0.11 1.00
Compared to Other Recalls Ford Pinto and the exploding gas tank (1978) 27 (500) deaths, 2.5 million vehicles recalled Audi 5000 and runaway acceleration (1986) 7 deaths, 250,000 vehicles recalled GM full-size pickups side impact fire (1993) ~6 deaths per year in 4.7 million vehicles (no recall) Firestone tires and rollovers in SUVs (2000) 148 deaths, 6.5 million tires Toyota acceleration (2010) 19-59 deaths, 2.5-8.0 million vehicles recalled 36
Age of Prius Deaths Typical driver on the road Average age 43.8 Median age 38 Typical vehicle fatality Average age 40.2 Median age 32 Age of Prius sudden acceleration fatalities Average age 58.1 Median age 61 80% greater than the median driver age 88% greater than the median vehicle fatality The last runaway Prius in San Diego that lasted 50 miles and needed a police escort was most likely a publicity stunt. 37
Decision Context: If It Does Happen Don t Forget the Brakes! The brakes on your vehicle are much more powerful than its engine To accelerate from 0-60 mph using the engine takes a typical car 8-9 seconds To decelerate from 60-0 mph using the brakes takes a typical car 3-4 seconds (much quicker) If you step on the accelerator and brake pedals at the same time, you won t move (the brakes win every time) Car and Driver did a test For a Camry going from 70 to 0 mph, the stopping distance went from 174 feet for a typical car to 190 feet for a car with a stuck accelerator (16 feet longer or one car length). For an Infiniti going from 70 to 0 mph, the stopping distance went from 161 feet for a typical car to 170 feet for a car with a stuck accelerator (9 feet longer), but the vehicle had a throttle cut-off once the brake was depressed. http://www.caranddriver.com/features/09q4/how_to_deal_with_unintended_acceleration-tech_dept 38
Stopping Distance with Runaway Acceleration 268 horsepower engine Throttle cut-off installed Manual transmission http://www.caranddriver.com/features/09q4/how_to_deal_with_unintended_acceleration-tech_dept 39
What to Do Steady constant brake pressure will stop the vehicle Don t panic; keep steering Don t pump the brakes In addition to braking, put the transmission in Neutral If you have an unrepaired recalled Toyota, spend a few minutes thinking through what you would do, just in case the very rare event happens. Do not turn the engine off at high speed because you could lose power steering and brakes Besides cars with push-button starts are not easy to turn off when trying to control a runaway accelerator 40
In the Press Hyping the possible electronic problem The brakes failed! It all started when 41
On the Blogs Detailed LA Times article with bio sketches of each fatality Suggesting that human factors are a major factor Age of drivers Locations of incidents 42
Solution? Fix the floor mats Fix one potential reason for sticky accelerator pedals Change software to force throttle to idle when brake applied 43
50 45 40 35 30 25 20 15 10 5 0 NHTSA Reporting Reporting Incident 44 Jan-02 Mar-02 May-02 Jul-02 Sep-02 Nov-02 Jan-03 Mar-03 May-03 Jul-03 Sep-03 Nov-03 Jan-04 Mar-04 May-04 160 140 120 100 80 60 40 20 0 Jan-02 May-02 Sep-02 Jan-03 May-03 Sep-03 Jan-04 May-04 Sep-04 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Reporting Incident What happened to the incidents? Have they stopped? Has Toyota found and fixed the problem? What about 2004?
Complaints to NHTSA Camry, MY 2002-2010, Acceleration Problems January-11 July-10 January-10 July-09 January-09 July-08 January-08 July-07 January-07 July-06 January-06 July-05 January-05 July-04 January-04 July-03 January-03 July-02 January-02 August-01 45 July-02 January-03 July-03 January-04 July-04 January-05 July-05 January-06 July-06 January-07 July-07 January-08 July-08 January-09 July-09 January-10 July-10 January-11 Reporting Date Incident Date
July-10 January-11 January-10 July-09 January-09 July-08 January-08 July-07 January-07 July-06 January-06 July-05 January-05 July-04 January-04 July-03 January-03 July-02 January-02 August-01 Complaints By Model Year 2002 MY 2003 MY July-10 January-11 January-10 July-09 January-09 July-08 January-08 July-07 January-07 July-06 January-06 July-05 January-05 July-04 January-04 July-03 January-03 July-02 January-02 August-01 46 July-02 January-03 July-03 January-04 July-04 January-05 July-05 January-06 July-06 January-07 July-07 January-08 July-08 January-09 July-09 January-10 July-10 January-11 July-02 January-03 July-03 January-04 July-04 January-05 July-05 January-06 July-06 January-07 July-07 January-08 July-08 Reporting Date Incident Date Incident Date January-09 July-09 January-10 July-10 January-11 Reporting Date
What Happened between December 2009 and March 2010? Percent since November 2009 100% 90% 80% 70% 60% 50% 40% 30% 20% Time on road Incidents Reports % of complaints speed related 10% 0% MY 02 MY 03 MY 04 MY 05 MY 06 MY 07 MY 08 MY 09 47
Delay in Reporting Historical Events 300 250 200 150 Within 10 days Within 200 days 100 50 0 48 Reports per month Jan 00 Jun 00 Nov 00 Apr 01 Sep 01 Feb 02 Jul 02 Dec 02 May 03 Oct 03 Mar 04 Aug 04 Jan 05 Jun 05 Nov 05 Apr 06 Sep 06 Feb 07 Jul 07 Dec 07 May 08 Oct 08 Mar 09 Aug 09 Jan 10 Jun 10 Date report files
What Follow What? 1.00 Correlation Press Hits and Reported Problems 0.80 0.60 0.40 0.20 0.00-12 -11-10 -9-8 -7-6 -5-4 -3-2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12-0.20 Reports months before press hits Reports months after press hits Month Lag Incidents drive press coverage or press coverage drives incident reporting? 49
Back to Normal? Why have the reporting of problems essentially returned to pre-recalled levels? If electronic, then the problem has not been fixed Lock out of throttle when brake is pressed, does not fix the problem, just makes the problem controllable If human factors was a dominate cause, then older drivers are still pushing the wrong pedal Does greater awareness of potential problems reduce human error this much? 50
New Data Arriving Updates to traffic risk web site in January 2011 Fatalities: FARS 1999-2009 Exposure: NHTS 2001 and 2009 Injuries: GES 1999-2009 Vehicle registration data merged with driver license data and Census Make/model/year and possibly VIN Location through zip code Age/gender of driver Miles per year Average household income and population density Longitudinal (across varying fuel prices and economies) 51
52