1 of 49

Distraction Impact Analysis: A Targeted Approach to Crash Causation

JEFF MUTTART, PH.D., ACTAR ACCREDITED #96

DRIVER RESEARCH INSTITUTE

1

Jan 2024

2 of 49

MEET TEACHING OUR TEAM

JEFF MUTTART, PH.D.

President

SWAROOP DINAKAR, MS

Director of Consulting

TIM MALONEY, MS

Scientist: Retired Police Sgt.

2

3 of 49

Classes Taught and Licensed Users of RESPONSE®

4 of 49

x

1. Limitations of Current Approaches

5 of 49

The Problem

5

EXPERT WITNESSES:

NEED “SUBSTANTIALLY SIMILAR” COMPARISONS

CANNOT ASSUMEAVERAGE DRIVER

6 of 49

The Problem

6

6

MANY SIMULATOR STUDIES: DELAYED RESPONSE TIMES

MOST NATURALISTIC STUDIES: REDUCED CRASH RISK

7 of 49

The Problem

7

7

8 of 49

Performance costs:

0.01 s. or 2 s.?

For all crash types & All Usages?

9 of 49

FARS:�Front-to-Rear��86% of “distracted” events were not from direct evidence��We need a better �cause - effect

9

Task

Number

Percent Objectively categorized

Occupant

151

14%

Moving object

13

Talk or Listen

108

Mobile

133

Adjust radio

43

Components / Controls

75

Use/Reach for device

117

Outside person / object

267

Eating/Drink

42

Smoking

7

Another mobile device

143

No driver

401

Distraction / Inattention

193

86%

Distraction / Careless

2

Careless / Inattention

6

Distraction Unknown

658

Inattention Unknown

1284

Not Reported

3406

Lost in thought

5

Other distraction

22

Unknown if distracted

723

N = drivers (7370)

10 of 49

x

2. Crash Types Distinctly Different Events

11 of 49

11

1 bit = log2 1.0

12 of 49

CRASH SAFETY SOLUTIONS, LLC 2021

12

Comparative Probability

Source: 2nd Strategic Highway Research Program

13 of 49

More Probable = Less Uncertain = �Faster Response Times

13

More info to process

More uncertainty

Less info to process

Less uncertainty

Faster response

Shannon Bits of Information(x) = -log( prob(x) )

14 of 49

14

Source: Muttart and Dinakar, 2022

15 of 49

x

3. Dosage, Tolerance, Task Framework

16 of 49

Crash Types are Alike

  • Each have different triggering events (Onsets)

  • Each have different response times

  • Each have different factors that influence the response (Muttart, 2003, 2004, 2005)

  • Each have different Crash / No-Crash ratios (SHRP-2, 100-Car, Lee et al., 2003, Ma et al., 2020)

  • Expecting all drivers to response in an “average” way is unsound

16

17 of 49

Dosage, Tolerance, and Task Framework

  • How Different Secondary Tasks Affect Driver Performance (Dosage)

  • Impact of Driver Characteristics on Distraction (Tolerance)

  • Role of Driving Task Complexity (Task)

17

18 of 49

Vulnerable Road Users are Not Alike

  • Multitaskers
  • Novice drivers
  • Intoxicated drivers

  • Expecting all drivers to response in an “average” way is unsound

18

  1. Source: Strayer et al., 2010; Summala, Lamble, & Laakso, 1998; Victor, et al., 2018; Braitman, et al, 2008; Mayhew, et al, 2003; McKnight and McKnight, 2003

19 of 49

Secondary Tasks are Not Alike

  • Secondary tasks were categorized as follows:
  • None
  • Visual or auditory only (i.e., listening to music) [A/V]
  • Hands-free conversation [HF]
  • Hand-held conversation [HH]
  • Cognitive task [COG]
  • Visual and Manual tasks (i.e., dialing or texting) [VM]
  • Level 2 driving

19

20 of 49

x

4. Methodology for Evaluating Distraction

21 of 49

Dosage, Tolerance, and Task Framework

  • Task – the crash type
  • Dosage – The secondary task
  • Tolerance – compare the drivers’ responses to baseline responses for that crash type
    • Determine if the driver was “distracted” rather than assume distraction

21

22 of 49

Tolerance:

Classical

Case-Control

Scientific Approach

How drivers responded to similar baseline task in research

Driver’s Performance

23 of 49

We need a method

  • Standardizing the research results
  • Focused only on response to lead vehicle events (rear end alignments)

23

24 of 49

Our solution: Standardizing the research results

  •  

24

Source: Horrey, & Wickens, 2006

25 of 49

25

First Author Name

A/V

HF

HH

COG

VM

Alm

2.71

Brookhuis

0.00

Benedetto

0.03

0.02

Chisholm

0.00

Cole

0.55

0.64

Cooper

-0.48

Drews

0.56

Drews

5.21

He

0.25

0.43

Ishida

0.06

0.32

0.19

Lamble

0.61

0.61

Lin

0.42

0.85

Nilsson

0.00

Mohoebbi

0.88

0.94

Muttart a(cue)

0.46

Muttart a

0.03

Muttart b

0.96

Ranney

-0.56

Sawyer

1.00

Smiley

-1.37

Strayer

0.41

0.62

1.28

1.90

Terry

0.82

Xu

2.17

Shinar (Days 1-3)

-0.50

-0.50

Weaver

0.00

Metz

-0.34

 

 

 

 

Lee

 

 

 

 

1.86

26 of 49

26

Standardized Brake Response Times

Not significant

27 of 49

27

Standardized Brake Response Times

Slower than 58% of baseline responders

Quartile

Not significant

Z

28 of 49

28

Standardized Brake Response Times

Slower than 47% of baseline responders

Quartile

Not significant

Z

29 of 49

29

Standardized Brake Response Times

Slower than 54% of baseline responders

Quartile

Not significant

Z

30 of 49

30

Standardized Brake Response Times

*Slower than 75% of baseline responders

Quartile

significant

Z

31 of 49

31

Standardized Brake Response Times

*Slower than 93% of baseline responders

Quartile

significant

Z

32 of 49

32

A/V

HF

HH

COG

VM

Median

0.41

0.01

0.11

0.61

0.96

75th percentile

0.42

0.24

0.20

0.85

1.67

25th percentile

0.14

0.00

-0.10

0.03

0.63

Table 6. Z-score increase or decrease in PRT for median, 25th percentile, and 75th percentile responder. For responses to a LV while using a cell phone and engaged in one of five secondary tasks

33 of 49

33

Drivers’ BRT

1. Significantly greater when engaged in Cognitive or Visual/manual secondary tasks

34 of 49

34

Non-Standardized Response Times

L2 Vehicle studies

Sec.

35 of 49

35

Non-Standardized Response Times

L2 Vehicle studies

Not an emergency response

Sec.

36 of 49

36

Non-Standardized Response Times

Slower than 50% of baseline responders

L2 Vehicle studies

Sec.

37 of 49

37

Non-Standardized Response Times

*Similarly slow response as cognitive with no L2

L2 Vehicle studies

Sec.

38 of 49

38

Non-Standardized Response Times

*Slower than 93% of baseline responders

Quartile

L2 Vehicle studies

Sec.

39 of 49

39

Not standardized – general finding

  1. Slight delay due to switching
  2. Greater delay due to the 2nd task

40 of 49

Other findings: Handheld or Hands-Free on PRT

  • Edewaard, D.E., Muttart, J.W., Dinakar, S., Appow, S. (2021). Driver Response Time to Pedestrians and Bicyclists During Turning Events from the SHRP-2 Naturalistic Database, Transportation Research Board of the National Academies.
  • Muttart, J., Dinakar, S., & Edewaard, D. (2021). Drivers’ Responses to Lead Vehicles: Thresholds for triggering an emergency response, Age Differences, Crash Risks, and Influence of Secondary Task Engagement (No. 2021-01-0898). SAE Internal Paper.
  • Dinakar, S. and Muttart, J., “Driver Behavior in Left Turn across Path from Opposite Direction Crash and near Crash Events from SHRP2 Naturalistic Driving,” SAE Technical Paper 2019-01-0414, 2019, doi:10.4271/2019-01-0414.
  • Dinakar, S., Muttart, J. W., Edewaard, D. E., Giannone, M., & Dickson, C. (2021). Driver Response Time in Cut-Off Scenarios from the Second Strategic Highway Research Program Naturalistic Database. Transportation Research Record. https://doi.org/10.1177/03611981211045368

40

+

NS

NS

NS

Conversation

Significant?

Even texting did not reach significance – 85% < 1.8 s TTC

41 of 49

Definition of distraction for Crash Reconstruction

41

Driver Distraction occurs when a driver is involved in a non-driving-related task that results in a response that is inferior to the baseline responses observed for similar tasks.

A crash related to driver distraction takes place when this decline in response quality leads to a crash that would not have happened under typical response conditions.

42 of 49

1. Take caution in “distraction” data

2. L2 is generally associated with delayed response

3. 2nd tasks are cues or distractions only when associated with behavior changes associated with a change in the outcome

Can all response types be crash surrogates?

43 of 49

Other resources

43

Contrast gradient

For night photos

Training: Live and Remote

Book

Software

44 of 49

The information contained in this presentation is, to the best of our knowledge, authentic and reliable; the author gives no guarantee that he has exactly paraphrased prior research. The author acknowledges that other factors and procedures not discussed in this class may affect your conclusions and interpretation of these results.

We are not responsible for how these materials are used or interpreted. Many limitations and application of that data is explained verbally.

DISCLAIMER

We cannot teach all 160 years of research in one week – this presentation is a condensed sample and is meant to guide the attendees toward the available research.

These course materials must be taken with the lecture. The course materials discuss concepts and may not address all the details that must be considered.

45 of 49

Do not make a copy – for discovery, refer counsel and Courts to CSS, LLC.

No materials herein may be shared without written permission. These materials are the intellectual property of CSS, LLC. If someone wishes to purchase these materials, please contact info@crashsafetysolutions.com

The cost for these materials is $1000 per user unless provided as part of an authorized course of study provided by, or sponsored by CSS, LLC.

These materials may be used by only the person who paid for attendance, or purchased the materials.

Use of any part of this course material for any purpose other than reconstructing and investigating motor vehicle crashes by the single user is prohibited. No derivative works are allowed.

Therefore, do not use this material to…

    • teach a class,
    • write a study or article,
    • for any other purpose without the written permission of the author

Copyright

46 of 49

46

Source: Muttart and Dinakar, 2022

Drivers exhibit “Wait & See”:

Head-on

LTAP-OD > 3s. TTC

Turn into Path (RTOR)

Red light runners

47 of 49

Uncertainty = inverse of probability

Tap table when a circle appears

When Probability = 100%

Reaction time approaches 0 s.

Reaction time is longer with more uncertainty

  • Poor Contrast
  • Larger Eccentricity
  • Longer time between targets
  • Smaller size

48 of 49

GOOD!

49 of 49

KEY TAKEAWAYS

& IMPLICATIONS:

  • Nighttime
    • Consider CLAPS first
      • For contrast – was there elements of camo?
        • Poor contrast
        • Correlated
        • Multi-color against multi-color

    • Consider the rules for use of the nighttime recognition chart
      • Glare – See Chapter 4, Section 4.3 – oncoming headlights
      • Weather – See Chapter 4, Section 4.2 – weather / tint
      • Lighting? – See presentation re: adding light