fashion nova models
so good morning to you all. i want to thank youfor joining us here this morning in washington. and i also want to say agreeting to those joining us online for the webcast. this will be an incredible day--very informative, and a thought provoking experience, so thankyou again for joining us. i'm so pleased tobe here with you on the eve of a veryimportant month.
april is national distracteddriving awareness month, and i'm thrilledto be participating by kicking off this symposium. my name is chris mullen. i'm director of technologyresearch for the strategic resources departmentat state farm. as you may or maynot know, state farm has a long commitment toconducting and supporting research related toauto safety, and i'm
honored to be hereto work with you all on this important issue. it only makes sense thatstate farm has a desire to understand therisks faced by those who travel throughour neighborhoods and on our nation's roadways. if you truly want to be agood neighbor-- and state farm does-- you want to help peopleavoid risks wherever possible. through our "research to action"philosophy at state farm,
we focus on research toidentify specific risks and evidence-basedinterventions that can be evaluatedfor effectiveness, and optimized for the mostsignificant positive impact. we followed that modelin the 1940s and '50s when we focused oncreating awareness by publishing information aboutthe most dangerous times of day to drive, and by becominga founding member of the insurance institutefor highway safety.
in the '60s and'70s, we continued education and awareness efforts,conducting anti-drunk driving and roadside death trapcampaigns, which included radio and tv commercials, andaccompanying magazine and newspaper pieces aboutways to make highways safer. during the '80s, '90s, and thefirst part of the 21st century, we became a founding memberof the advocates for highway and auto safety, andactively supported seat belt and airbagsafety initiatives,
including researchand policy efforts. we work with stateand local governments to address dangerousintersections, and develop safety programsaimed at youthful drivers, such as project ignition andour steer clear driver discount program. in 1996, state farm joinedwith the children's hospital of philadelphia--or chop-- to embark on one of the mostambitious ever auto safety
research initiativesrelated to our most vulnerable and valuablepassengers, our children. to begin that effort, statefarm identified crashes involving children,obtained permission from our policyholdersto use data from these crashes for research,and provided a plethora of unique data foranalysis by chop. through the partners for childpassenger safety project-- or pcps-- chop researcherswere able to closely examine
more than 600,000 crashesinvolving 875,000 children. it was the largest sourceof data ever collected on children involved inmotor vehicle crashes. the groundbreakingresearch provided the first clear picture of whathappens to children in a crash, and helped us to understandhow to prevent injuries in children notin booster seats. we used the data tosupport enactment of legislative measures thatsaved the lives of hundreds
of children, and reducedthe severity of injury for tens of thousandsof children each year. in 2006, as many of the childrenwho were the focus of our pcps research werebecoming teen drivers, state farm began an effort toenhance and augment knowledge around teen driver safety. the disturbing fact--that motor vehicle crashes are the number onekiller of teenagers-- provided the impetus forthis young driver research
initiative. continuing theresearch partnership with children's hospitalof philadelphia, and broadening our workwith other institutions, this research helpedidentify and expand awareness of the most commoncauses of teen driver crashes. we pursued effective,science-based interventions, including thedevelopment of tools that help young drivers bemore aware of driving hazards,
and a program thatprovides practical advice to assist parents in teachingtheir new drivers how to be safer on our roads. state farm also advocates forgraduated driver's licensing laws, and conducts teendriver awareness campaigns, such as our successful annualcelebrate my drive effort. we continuously monitor thetraffic safety landscape to identify theissues most in need of evaluation and intervention.
so in 2012, the autotechnology research staff that i lead at state farmconducted an environmental scan to identifysignificant auto safety issues needing to be addressed. distracted driving shotto the top of the list, and has been scoped now asa state farm engaged driving initiative, or edi. engaged driving occurswhen drivers actively provide sufficientattention to actions
critical to safe driving. today, drivers are challengedwith demands on their attention by multiple devicesand activities that may distract them fromthe task of safe driving. every day, each of usmakes conscious decisions to mitigate ourrisk while we drive. under what conditions dowe manage this effectively? what behaviors are most criticalto a safe driving experience? and how can we encourage driversto take personal responsibility
in adopting more of thesesafety-conscious behaviors? there are so manyrelevant question surrounding this effortthat it is very difficult to know where to begin. based on our successin past partnerships, we knew our engageddriving research had the best chance forsuccess if we reached out to others with a broadrange of perspectives and areas of expertise.
but how best toidentify and reach the people ingroups with the most to contribute to this effort? for this, we turnedto the association for the advancement ofautomotive medicine, or aaam. as an internationalmultidisciplinary organization dedicated to motor vehicle crashinjury prevention and control, aaam was the perfect partnerto help us bring together a panel of learnedexperts to quantify
the current state ofscience, and identify gaps needing furtherinvestigation. the panelists youwill hear from today approached this topic from awide range of perspectives, and will share some of thelatest scientific knowledge to help us better understandthis multifaceted issue. we hope what you will hear todayfrom this expert panel provides the spark the drivesresearch further, and breakthroughs makingthe task of driving safer
for everyone. before i turn the program overto frank pintar, president of aaam, i want totake this opportunity to thank a few ofthe people that made today's event possible. first, i would like to recognizedr. pintar, and the staff of aaam and its members. they have proven tobe the perfect fit, as a facilitating organizationwhose mission and purpose
closely aligned with thegoals of this effort. in particular, i wantto say huge thank you to aaam's administrativedirector, irene herzau, for her contributions. and of course, i want tothank our panelists, who represent the topminds in their field, for providing keyresearch in this issue. in addition to theirother responsibilities, they have givengenerously of their time,
and produced a tremendouslyrobust body of work in very short order, toincrease our understanding of relevant topics, andhelp us drive next steps. i also want to thank our keynotespeaker, david strickland, former administrator of thenational highway traffic safety administration, for takingtime from his busy schedule to share his perspectiveon this important issue. so thank you, david. and finally, we owea large thank you
to our panel chairperson,dr. sue ferguson. sue's background andexpertise position her fabulously for this role. in an effort like this,with an aggressive timeline and an ambitious scope, requiresdedicated and professional leadership. and we would not be here todaysharing this work with you, if it were not forsue's commitment. so thank you, sue.
i would also like to nowintroduce dr. frank pintar, president of the association forthe advancement of automotive medicine, aaam. aaam is a professionalmulti-disciplinary organization, dedicated entirelyto motor vehicle crash injury prevention and control. founded in 1957, itsmembers represent engineers, physicians, and researchers frommore than 20 countries working together to reduceroad traffic injuries.
its membership's combinationof clinical, research, and administrative backgroundsforms a unique blend of leaders in trafficinjury and control. aaam's currentpresident, dr. pintar, is a professor in the departmentof neurosurgery at the medical college of wisconsin,adjunct professor of biomedical engineeringat marquette university, and director of the neuroscienceresearch laboratories at the va medical center inmilwaukee, wisconsin.
these laboratories occupyover 35,000 square feet, employ 9 phd scientistsand 30 staff, and include expanded functionsof an acceleration service led and vehicle crash test facility. over the years, dr. pintar hasserved as a research mentor to more than 30 graduatestudents, and post-docs, residents, fellows,and junior faculty in the medical schooland engineering school. dr. pintar has authored morethan 600 research publications,
including two books onhead and neck trauma, and over 220 peer-reviewedjournal papers. his current research includesbrain injury mechanics under blunt and blast loading,vehicular and military trauma biomechanics, and optimizingspinal surgery techniques. he is the engineeringprinciple investigator of the wisconsin ciren center,one of six in the nation. as a member of aaam, hehas been actively involved in the board of directors,has served as its treasurer,
and currently its president. he has also servedas an instructor in biomechanics courses, andhas given invited lectures at speciality conference. he contributed papers to theannual conference since 1995, and received the aaamaward of merit in 2012. please join me in welcomingdr. frank pintar to the stage. [applause] well, thank you.
as the president ofaaam, i also want to welcome you to thisengaged driving symposium. thank you, chris, very much,for a nice introduction. as aaam, we are sopleased and proud to be able to partner with statefarm to provide some clarity on this most important topic. i do want to thank chrismullen for her tireless efforts in spearheading this endeavourwith state farm and aaam-- just getting us together.
i know that shewould humbly tell you that there were a lot ofother people involved-- and there where-- i do want toacknowledge her critical role in assembling the right peopletogether, at the right time, to make thissymposium a reality. thank you, chris. a little about our organization. the aaam-- associationfor the advancement of automotive medicine--was founded in 1957
by a small group of physiciansinterested in automotive racing safety. but they soon realizedthat-- i'm sorry, they soon developed a broaderinterest in vehicle safety, as they saw how safetyimprovements in racing could be implemented in thegeneral public's vehicles. aaam leaders establishedan annual conference to bring together individualswith similar interests, and in 1964, opened up themembership to professionals
that were not physicians. today we enjoy the benefitsof a multi-disciplinary international groupof professionals that are dedicated to roadtraffic safety and vehicle crash injury controland prevention. aaam membership includesdiverse medical practitioners, engineers, epidemiologists,and other scientists who share the same interests andgoals of road traffic safety. besides our annualmeeting, people often
don't realize that aaam is theoriginator of the abbreviated injury scale, or ais. the ais is ananatomically-based coding system to classify anddescribe the severity of specific individual injuriesrepresented as a threat to life assessment. the ais is one of the mostcommon anatomical scales for traumaticinjuries, and is used throughout the world bymedical trauma centers,
as well as governmentsand industries to quantify injuries. through the voluntaryefforts of our membership, the ais dictionaries aresustained and updated with the latest information. we also maintainedprofessional staff that provide ongoing trainingto individuals and organizations on the use of the aisto score injuries. also a product ofaaam's annual meetings
and our scientificpresentations today, is our journal, the annalsof advances of automotive medicine, which is indexedon medline and in pubmed. the scholarly productsof today's symposium, the individual papers thatthe panelists have written, have all been peer reviewedthrough our usual process. and this scientificevaluation has been made possible becauseof the diverse membership available for peer review andaffiliations with the aaam.
i'm happy that we wereable to provide you all a copy on your way intoday when you registered. i'm really looking forwardto today's presentations on the state of the art sciencein our understanding of what engaged drivingreally looks like. we have all hadour bad experiences of driving down the road andnoticing a distracted driver weaving, or talkingon a cellphone, or eating, or somepersonal grooming,
and having that sickeningfeeling that something bad might be about to happen. i believe there's beena lot of confusion in the general publictoday on the dangers of distracted driving,and i look forward to the results oftoday's symposium, with the promotionof state farm, to try to bring some clarityto this subject matter. i do want to thankeach of the panelists
today for participatingin this effort. and a special thank you,again, to sue ferguson for agreeing to chair the panel. you will see that thepanelists assembled are truly the nation's expertson a variety of aspects of this problem, and i urgeyou all, as the audience, to participate withdiscussions and questions that will stimulate the meeting. so finally, i wantto thank state farm
for making thissymposium possible. i know that aaam looks forwardto continuing to work together with state farm on similartopics in the future. thank you very much. thank you, frank. it is now my distinct honor tointroduce our keynote speaker for the morning. david strickland beganhis career as an associate at a washingtondc based law firm
before moving to a positionon the staff of the us senate committee on commerce,science and transportation, as democratic senior counsel. for eight years, heserved as lead counsel for the subcommittees overseeingthe federal trade commission, the consumer product safetycommission, the national highway trafficsafety administration, and the department of commerce. during his servicein the senate,
his accomplishmentsincluded serving as the lead democraticstaff drafter for multiple legislativeactions, including the 2003 do not call registry, the2005 highway reauthorization, the 2007 fuel economy mandates,and the 2008 re-authorization of the consumer productsafety commission. mr. strickland then servedas the 14th administrator of the national highwaytraffic safety administration, or nhtsa as the top automotivesafety official in the united
states, he was responsiblefor executing the agency's mission-- to reducecrash-related fatalities and injuries, while insuringthe highest standards of safety on our nation's roads. mr. strickland oversaw abroad range of vehicle safety programs, includingvehicle safety standards, defect investigations, safetyrecalls, and distributing over $600 million in highway safetygrants to the states each year. mr. strickland'saccomplishments while at nhtsa
include, among otherthings, development of the first national fueleconomy programs for passenger vehicles and heavytrucks in conjunction with the environmentalprotection agency, and implementing the vehiclesafety and highway safety grant mandates includedin the 2012 highway authorization, or map 21. he also included the first everejection mitigation standards to help keeppassengers from being
partially or fully ejectedfrom vehicles during a rollover crash, and mandatedlap and shoulder belts be installed on allnew motor coaches. launched the nation's largestconnected vehicle safety pilot program, and issued the firstever automated vehicle policy. if that's not enough,mr. strickland brought national attention tochild passenger safety issues, and was a true leaderin the campaigns to fight impaired anddistracted driving.
now a partner in venable'sregulatory group, mr. strickland'scurrent practice focuses on transportationpolicy, consumer protection, internet privacy, data security,and legislative and government affairs. please help me inwelcoming david strickland. well, good morning, everyone. it's truly a pleasureand an honor to be here. i see a lot ofvery friendly faces
and long friends in the room. it's really greatto be back, talking to a bunch of very,very smart people that are committed to thepurposes of reducing fatalities and injuries onthe nation's roads. it's an interestingperspective, after 12 years of public service and trying tofind the right ways to approach safety, how todemocratize safety, how to lift safety up inthe minds of individuals
every single day, howto be able figure out how to empower the agencies,and frankly, the congress, in making the right decisions. and all those decisionssurround data. and i find it interestingthroughout the scope of my career, not only atthe commerce committee, but at nhtsa, thedifficulties in trying to get the answersto the questions. or even more importantly,trying to figure out what
are the right questions toask even before you begin to think about how yourun down those answers. and i know there's a lotof people in this room that have committed their livesto this very endeavor. and it's just soexciting for me to be able to have a fewmoments with you today to talk about engaged driving. it's interesting, because whensecretary lahood took this up at his keynote issue whenhe became secretary in 2009,
it was interesting inhow we came about it. because you're thinkingabout the constant fight of, i would say, the more maturerisks that we all appreciate. impaired drivingstill kills one third of all people intraffic crashes. unbelted passengersand drivers account for half of thefatalities in the fleet. distracted driving at the timethat the secretary sort of took it up, there really wasn'tany real strong data
as to how many people werebeing lost to inattentive or distracted driving. so that was sort of the firstquestion that had to be run down and asked, and some reallyasked for the agency to begin to endeavour upon. and then the nextquestion is, how do you deal with what seemsto be a threat that was already incumbentinto the fatality matrix, but clearly wasevolving, because
of the utility ofdevices that people began to become more and moredependent upon over the years. and you really had this sortof anecdotal social notion that there's a lot ofpeople using devices in very bad ways onthe roads, but it does beg the broader question. there's always been inattentivedriving in the fleet, ever since the first timesomebody decided eat a sandwich behind the wheel of his model t.
it just has always been there. it's just neverbeen focused upon because of theallure of devices. now that we have the power,and people recognizing that texting and drivingis not a good thing to do, now the question is how do youembrace the broader notion, which is the reason why i'm soexcited for aaam and state farm because i think insome ways, the campaign led by the secretaryand i over the years
has been very successfulin some aspects, but has actually sortof given short shrift to the broader and broaderproblems in others. and i think this came tolight to me specifically in secretary fox's firstconvening of a group of experts regarding talking aboutdistracted driving. the one big takeaway forthe secretary-- and fox understands it-- heknows this thing. he knows how passionatelahood was about it.
but also the fact thatit is a driving emerging threat that you don't wantto end up making gains in other areas, as we havein drunk driving and in belted driving, to possiblyhave a countervailing wind with inattentive driving. and the campaigns to gettexting bands across the country have been intensely successful. the question is now tryingto get resources and ability to enforce.
but then there was somesort of a strange subtext that happened. people think that textingwhile driving is bad, but looking at a web page on anandroid device isn't texting. so how do you capturethose activities? surfing the web, playing,doing other things that aren't necessarilytexting, but clearly drive to inattentive ordistracted driving. and so it was interesting.
the secretary said, nowthe next behavioral goal is, you have to make itbroader than texting. so a bit of the problemfrom the policy side is, sort of chasing the dragon. which is, how do youtruly define the problem, and thinking about engageddriving holistically. i personally think it's theright approach, because you can't simply define inattentionby if you put your device down, you're better off.
it's also making sure that youare not doing other things. trying to eat a wendy's triplebaconator with two hands, while driving with your knees onthe steering column-- also bad. or, as frank mentioned,deciding to shave while driving. or putting that book inthe middle of the steering column while you're ona straight long road in the middle of nebraska. i've seen that happen, too. all inattentive drivingnot related to a device.
how do you put together ascheme which really conveys culturally to theamerican people how to stay engagedbehind the wheel? and that's, once again, sortof the thing i circle back to. perhaps in 2009 and2010, the department in how it phrasedits first question, we got a very effective answerfor one sliver of the problem. and it's activity such asthese-- thinking about how to unpack thething, so we can now
ask the right holisticquestions to then empower social researchersand the engineers to think about theholistic answer to deal with inattentivedriving, writ large. and reason why this symposiumis so incredibly important. the second aspect i wanttalk about is in the data, and is something that the agencyis working very hard upon, i'm sure, at this very second. i see two of my favoriteformer colleagues
here in the room who workedvery hard on this issue. nhtsa issued the firstdistraction guidelines for visual manualdistraction, as the phase 1 part of the effort. phase 2 talking aboutpneumatic or handheld devices. and phase 3 is going to bevoice or cognitive distraction. i'm sure the nhtsa team willbe able to talk about this in a much broader andbetter detail than i because i, once again,was a dumb political
in the back of the office,and the smart engineers are the ones that doall the hard work. but once again, it'sabout data, and being able to come up withthe answers that we see in this particular space,and how you can actually change this into effectivesafety and regulatory policy. my fear, now that i am nowthe former administrator, and i can have a bitof an unplugged moment because i'm no longerat the agency anymore.
my fear is that everyone isworking on actionable data that they have intheir hands right now. the manufacturers are. the software developers,the apple developers are. everybody wants to try to do theright thing and still innovate. but what happens if theagency gets to phase 3 and we find that thereare cognitive load issues in the waythings are particularly designed right now whichcut against the basic notion
of whether this particularapplication, or device, or display is truly safe. that is going to bethe open question. the agency will answer it. and then we're all goingto have to wrap ourselves around the fact we may haveto have some very, very difficult decisions to make interms of how vehicle design is managed, how we deal withit, and how frankly, we build the cultural programs.
and i think the one thingthat everybody in this room should-- i wouldencourage you do-- don't shy away fromthe hard answer. because you never shyaway from the truth. and if that's wherethe chips fall, that's where the chips fall. but the important thing isto try to wrangle the answer, because you shouldnever compromise safety against any other value interms of individual mobility.
safety is number one priority. and with that beingsaid, the one thing that the agency wasvery focused upon is thinking about howto attack the problem of inattentive driving. more specifically, withtechnological countermeasures. the behavioral program isone that i am very proud of. the reasons that we have hadthe recent historical lows of fatalities is becauseof not only great vehicular
design, better crashworthiness, but also the power of thebehavioral program and working with the stateswith better education, stronger enforcement, and overall,a change in culture. you don't see a child todaynot strapped in a baby seat. the problem we havenow is they may not be properly strappedin that baby seat. but yeah, i'll see mostbabies in baby seats. and you'll also see people thatdo see an unrestrained child,
they will call 911and report them. that is a beautifulculture change. and that's what you want toachieve in all these things. that's what the power of thebehavior program gives you. we're in the very beginnings ofthose stages with distraction, and we have to fightthat behavioral fight. we're going to haveto use those grant resources to empowerpolice departments to be able to figure out waysto enforce those texting bans,
or those cellphone bans. that's very, veryimportant stuff. and that is a long tail onhow you get to a solution on driving down thosefatality numbers regarding to distraction, the over3,300 people that are lost. but the other piece thati feel is very important is the technological piece. is figuring out, for the casesof distraction by a device, being able to have thevehicle identify the driver
position from thepassenger positions, and then inter-lock the driver'sphone unless it's properly tethered in and ina safe interface that limits particularapplications that are not appropriate to thedriving environment. it's a doable goal. lots of manufacturersare looking at it. lots of cellphone manufacturersare looking at it. the question is whether you canget all those folks together
in a room to come together in astandardization process that's not going to takeus 15,000 years, or a long, bruisingrule-making process, or a legislative attemptwhich may necessarily be too broad or toolimiting, depending upon the data that's received,and the political decisions that are made. but that's a keycomponent of it. the second attempt ispart of the agency's
significant and seamlessinitiative, which focuses on threetechnological interventions to drive on thethree biggest risks. one is for seat belts. they're looking at aseat belt interlock standard that couldbe voluntarily applied by the manufacturers,which i think is very important toaddress those numbers. also, the driver alcoholdetection system,
or dads-- which a numberof folks in this room are also workingon-- to making sure that we have aauto-grade alcohol interlock, that if the driver'sis seen as detected as being above 0.08, thevehicle interlocks. very point in addressingthose one third of crashes. and the last, which alsoapplies to inattentive and engaged driving isforward collision avoidance and mitigation, or fcam.
so when you're thinkingabout drivers and crashes, 90% of crash causation has anelement of human error in it. embedded in that humanerror is inattention-- looking away from theroad, not paying attention to what you're doing. so if you have avehicle that can, when a person failsto pay attention or is not engaged indriving as they should be, having the vehicle be able toapply a full braking to keep
you from running into thecar that's in front of you, or helping you maintaina lane, or noticing maybe they're even sleepy, andgiving you an indication to get you off the road. technology isn't thesolution for all things, but it has to be clearpart of the solution matrix in the long term,and how we deal with these behavioral problems. we've done a greatjob as an industry,
as researchers, asregulators, as states to deal with crash worthinessover the past four years. we're now in the windowof crash avoidance, and i think it's a veryimportant window to be in. but you cannot take your eyesoff of the fact that we have opportunities to getthe data right now, to have the right culturalinterventions now, as we build towards a futurewhere you have vehicles that can support the driving task,and make it a safer endeavour
every single day. this is a long-tailedthing, because even if you do have atechnological innovation, it will take severalyears-- over a decade-- to get the fleet turned overto have it fully penetrated. so you can't takeyour eye off the ball in any of these notions. so having everyonehere today to talk about the research and data thatthey have right now regarding
engaged driving isclearly a foundation to fill in a number of gapsthat frankly, everybody's having around the issue ofengaged or distracted driving. i know the folks atnhtsa will definitely appreciate these effortson a going-forward basis. and i know for a fact that theamerican people will derive the benefits of the fantasticwork that's not only going on at this symposium today,but as you all return back to your educationalinstitutions,
your hospitals-- hello, dr.dennis, always great seeing you-- and your manufacturingfacilities across the country, know that thiswork today is going to save lives in the future. so thank you so much for thisopportunity to join all of you today. as you can tell, eventhough i am now back to being a simplecountry lawyer, my passion for highwaysafety has not diminished,
and it never will. because it's something, frankly,as everybody in this room knows, it gets in your marrow. and i will always bevery close to this topic, and several others,as the years go on. and i will becheering [inaudible] for all the smart people inthis room to get the thing done. thank you so much, everybody. have a wonderful symposium.
take care now. thank you, david, again,for your passionate comments on this issue. we are amazinglyearly, and i think we will appreciatehaving that extra time, so i will say, on withthe show, with the program for our panelistsand their findings. and again, it is my honorto introduce dr. susan ferguson, our panel chairperson.
sue is a highwaysafety consultant with 22 years experienceconducting research in many differenthighway safety areas. her main emphasis has beenon vehicle safety issues, including alcohol and driving, occupant restraints, young and older drivers, andchild occupant protection. during that time, she haspublished more than 100 scientific research papers. most recently, she hasserved as the program manager
for the driver alcoholdetection system for safety, or dads,that david mentioned programmed multimilliondollar, multi-year research program for the development ofan advanced alcohol detection system that is funded by thenational highway traffic safety administrationand the automotive coalition for traffic safety. doctor ferguson serves on manycommittees and advisory boards, both within the usand internationally,
including chairing thetransportation research board committee on alcohol, otherdrugs and transportation. and as i saidearlier, this effort would not have come togetherwithout sue's dedication. so sue , thank you so much. good morning, everybody. a lot of familiar faces,and some not so familiar, but i'm glad you couldall be here today. thanks again to statefarm and the aaam
for providing this reallywonderful opportunity for me to be part of this effort. it's given me theopportunity, basically, to work with the who's who indistracted driving research, and the flip side,engaged driving research. we do have a full program today. my role is basicallyto keep it on the road, to keep things moving,to get people introduced. we have a littleextra time already,
which is great becausei know that you'll have lots of questions,and our speakers will have a lot of goodto things talk about. so with that, i'm going tointroduce our first speaker. our first speakeris dr. mike regan. he's a professor in transportand road safety research group in the school of aviation atthe university of new south wales in sydney, australia. we do have aninternational team.
he has a bsc and phd degreesin experimental psychology and ergonomics from theaustralian national university, and more than 20years experience in transportationsafety research. he is the author or co-authorof more than 220 published documents, includingthree books, and has career researchearnings totaling around 18 millionaustralian dollars. he sits on the editorial boardof five peer review journals,
including human factors, andseveral international expert committees. in 2012, he became the 25thpresident of the human factors and ergonomicssociety of australia. i also want tointroduce his co-author on the first paper, whichis dr. david strayer. he is a professor in thedepartment of psychology at the university ofutah, and the director of the applied cognition lab.
he received his phd fromthe university of illinois at urbana-champaignin 1989, and has been studyingattention, multitasking, and driver distractionfor the last two decades. doctor strayer is afellow of the association for psychological science, hasauthored over 100 peer reviewed articles, and in 2010, was therecipient of the university of utah distinguished, scholarlyand creative research award. with that, i'd liketo welcome mike regan
to the podium for thefirst presentation. thank you very much forthose kind words, sue, which i must confessi wrote myself. i've just got to work out howto work out the ergonomics here. thank you very much, sue. i just wanted to saythanks to state farm for funding my involvementin this activity, for bringing an aussiehere to america, especially to this great city.
and to the aaam, to sue andirene, and others for all the hard work in supporting me. thank you. and of course, to dave, whowrote part of this paper, and he's going to bepresenting on his contribution. so i'm going to navigatethrough it like this-- to try to talk aboutwhat destruction means, what inattention means. i'll talk about ataxonomy of inattention
that was developeda couple of years back, with a couple ofcolleagues, one on whom is my phd student,charlene hallett-- or was. i'll talk about some work that'sbeen done in australia that actually used thattaxonomy for the first time to classify crash data. i'll talk aboutanother taxonomy that was developed that i was alsoinvolved in developing when i was working in europe, withchris monk and a few others.
so it quite interestingfor me to say to work on anothertaxonomy, even though i'd developeda previous one. dave's going to talk about thetheoretical characterization of the originaltaxonomy, and then i'll finish with conclusionsand research needs. people talk about a distractionas if they know what it means, but if you ask a dozen peoplewhat they think it means, you'll get a dozendifferent answers.
and for people like me and someothers who work in this field, it can be quite frustratingbecause it can actually cause some issues. and the main issue is reallythat if you have inconsistent definitions, it meansthat you can't really be comparing research findingsacross studies, because you can't be sure that you'remeasuring the same thing. and differentdefinitions can lead to different classificationschemes for coding data.
and so for example, do we codeas a distraction someone who attends to on anunexpected warning light in an unfamiliar carwhen they're driving at night. or do include asdistraction daydreaming? some people would, andsome people wouldn't. it just depends onwhere you come from, and how you thinkabout the problem. and so, the fact that that's theway the world sort of operates, means that these differentclassification schemes
can lead to differentestimates of the role of drive distraction in crashesand critical incidents. and so when we saythat we think we know what the scaleof the problem is, i'd probably say that wedon't really yet still know exactly what thescale of the problem is, although i thinkwith converging evidence, i think it's apretty big problem. here are a couple ofdefinitions of distraction
but two groups of experts,totally independently, about five years apart. i was on the secondgroup, but not the first. an interesting thing here isthat the first definition's essentially saying thatdistraction's not distraction unless you're payingattention to something that's not related to driving. if it was driving-related-- likethat unexpected fuel warning light-- then itwouldn't be distraction.
the second the definition'sa little more easy on that aspect. and the first one's assumingthat distraction has to involve some sort of adverse outcome,whereas the second definition doesn't. again it's a bitopen on that issue. and whatever thedefinition, it seems that there are certain elementsthat have been thought about in thinking about distraction.
it's a diversion of attentionaway from driving or activities critical for safe driving. it's diverted towards somesort of competing activity. it could be inside oroutside the vehicle, and some definitionsdiffer on that. the competingactivity, as i said, may or may not bedriving-related. it could be voluntaryor involuntary. i saw a definitionrecently in which
the person from the netherlandsthought that distraction was only distractionif it was involuntary. if you reach for a cellphone and distract yourself, as far as thatperson was concerned, it's not distraction. and as i said, there can be animplicit or explicit assumption that safe driving isadversely affected. so inattention is aninteresting kettle of fish, too. and there's a bit of confusionabout the relationship
between distractionand attention. they are relatedconstructs, but there aren't that manydefinitions of inattention, and the ones that do existtend to vary quite widely-- even more so thandefinitions of distraction. and here are just three that ichose fairly randomly, in fact. the first one isbasically saying that inattention isattending to something else in the absenceof distraction.
the second one'sbasically saying that inattention isattending to the wrong thing. and the third oneis basically saying that inattention is diminishedattention to the right thing. and so, like distraction, thereare lots of different elements that have been thought about inthinking about what inattention actually means. it could mean anyone of those things. and in fact, some peoplethink that inattention
is synonymous with daydreamingand internalized thought. that rather than itbeing an outcome, they'll say distraction orsome other form of inattention that is, in fact,itself a process. and that's what iwas really trying to get at here in this slide. so a couple of colleaguesand i put together a taxonomy a couple of years agowe published in aap to try to piece togetherthe whole area a little bit,
and to try to makesense of it all. and so the model basicallysays that inattention is the outcome of a numberof different processes which we characterize asdriver-restricted attention at the top, all the down todriver diverted attention. and driver diverted attentionis what other people would call driver distraction. and what these forms ofinattention are telling us is a bit aboutthe mechanism that's
involved to leadto this intention. and for us at the bottomthere, distraction or driver diverted attention, can beboth non-driving related, or driving related. so there are things that cancompete for your attention that are non-driving related ordriving related that we would regard as being distraction. and for both of those, fornon-driving related and driving related sources ofdistraction, they
can also includeinternal thought. so we're sayingthat when people are attending to internal thoughts--whether they be daydreaming or thinking about what they'vegot to get at the supermarket-- that that's distraction. and some people might notcall that distraction, but that's the waywe thought about it. and so we developedthis taxonomy from-- it was data-driven.
so we looked at crash data,at the way other people had classified data, rather thangoing back to first principles and looking at theory. and that's what we didfor the second taxonomy that i'll talk aboutbriefly in a moment. and so in thistaxonomy, distraction is conceptualized as oneof a number of mechanisms of driver distraction. and different driverconditions and driver states,
we see in the contextof this taxonomy, as being thingsthat can give rise to the differentforms of inattention, or moderate their impact. so a young driver could be moreaffected by distraction simply because they have lessspare capacity than a more experienced driver todeal with the distraction. and so in terms of definition,we defined in this paper and for the purposesof the taxonomy,
driver inattentionas insufficient or no attention to activitiescritical to safe driving. and used essentiallythe same definition that we used in abook that i co-wrote with john lee and christieyoung a few years ago. but then on the end ofit, in the red writing, added this bitabout inattention. and so for us, i think it wasprobably-- and might still be-- the first definitionthat's directly
linking distractionand inattention. so it's saying that distractionis a diversion of attention away from activitiescritical for safe driving, toward a competingactivity which may result in inattention. that's how i would probablyabbreviate that definition further. and so i talked aboutthese different mechanisms of inattention inthat model, as you'll
recall, on the right hand side. and here are a few examples. driver restrictedattention, for example, is inattention brought aboutby biological restrictions that we just have todeal with as humans. so when you sneeze,you close your eyes. and that means that youcould be inattentive. and so sneezingwould be an example of what we call driverrestricted inattention.
drive misprioritizedattention is when you've got multiple sectorcritical things going on, and the timesharing breaks down. and you end up not attendingto the thing that's safety critical,for whatever reason. driver neglected attention ismore that faulty expectations. you cross a little crossing. you never see a train, so youstop looking for the train. and so you're inattentiveto the train, which
is the activity criticalfor safe driving. driver cursoryattention has more to do with being in a hurry andnot, in this case, for example, doing a full head check. so you're not activelyinterrogating the environment as actively as you could. and then finallywith distraction, it could be that someone'slooking at a cell phone while dialing a friend,and that causes problems.
and these are just-- iwon't to go into these, but these are otherexamples of distraction that relate to attending tointernal thought, for example. daydreaming, andthat sort of thing. both driving andnon-driving related. a colleague, vanessabeanland and some of her othercolleagues in australia used this taxonomy recentlyto conduct a study, and applied it to 340 crashesin which drivers were at fault.
and what they foundwas that inattention was an actual possiblecontributing factor in 64% of crashes. and that in itselfis interesting because, it's quite consistentwith other data that has been published. interestingly, ifyou look at the pie chart in the bottomleft hand corner, you can see that driverdiverted attention--
because they used the taxonomyto carve up the data-- that counted for about a quarterof the inattention related crashes. so the big one was driverrestricted inattention. but again, i'm not surethat they interpreted the taxonomy in quite the waywe would, because they included people being unconscious asdriver restricted inattention and i'm not sure i a sortof tightly agree with that, because i think ifyou're unconscious,
it's not physicallypossible to attend. so how can you be inattentive. so i've got to think aboutthat one a little more. but they suggestedthat the taxonomy could be refined ina number of ways. firstly, they suggestedthat this driver restricted attention category could besubdivided into two categories. but as i've said, i'm notsure i agree with that. secondly, theythought there should
be a new category ofinattention, which they called undifferentiated driverneglected attention, or driver cursory attention. in the case where peoplefailed to detect something, but they can't workout whether it's because they failedto look, or looked but didn't properly attend. and i think that'sa fair comment, and that's one ofthe things that
happens when people startusing these taxonomies. finally, they saidthat there was a bit of a problem in being ableto differentiate between what we call driver misprioritizedattention and driving related distraction. the latter beinggiving attention to less safety criticaldriving alleged activities, like that unexpectedwarning light. and with misprioritizedattention, having your time
sharing breakdown betweenmultiple safety critical tasks. so finally, beforei hand over to dave, i just wanted tomention that there has been this other taxonomythat's been developed. quite a number of us were ona committee that developed it. the interesting thingis that this one was developed from theoryrather than from crash data. and the interesting thing isthat when it all boils down, they're actually quitesimilar taxonomies.
there's not muchdifference at all. here you've got two generalforms of inattention, called insufficient attention,and misdirected attention. and if you look atinsufficient attention, it's broken down into sleeprelated attention impairment. that's pretty much whatwe were calling driver restricted attention. except they'resaying that it's only when you're sleepy that youcan be less likely to attend
to activities criticalto safe driving. but it excludes things likething drunk or drugged, and that sort of thing. and in that taxonomy, wewanted to include those things, because they can actuallyresult in inattention. the other thingis that-- i don't know whether this thinghas a pointer on it. i think it has. this one here.
insufficient attentional effort. that but basicallyencompasses what we called in theoriginal taxonomy, driver misprioritized attentionand driver neglected attention. and in this bottom one downhere, driver distraction is what we called drivercursory attention, and driver diverted attention. so the only differencebetween these two is that they're talking hereabout driver distraction
external to the vehicle,and internal to the vehicle. and in our taxonomy--the original taxonomy-- we talked aboutdriver distraction as being driving relatedor non-driving related. so they're pretty muchsimilar taxonomies, and you could takeone or the other and use it, andsee what you get. and as i'll say when ifinish off this talk, i think that's exactlywhat we need to be doing.
so i'll hand over to dave. he's going to talk brieflyabout the theoretical characterization ofthe original taxonomy. and then i'll have just twominutes at the end to finish. and i hope that'llbe at 20 minutes. thanks, mike. thanks for describingthe taxonomy. you did a really nice job ofarticulating how that worked. it was initially designedas a data driven effort,
looking at crash data. and the goal thatmike and i tried to do is say, let's try and takenot only the data driven crash data. next one. how do we basically take that--the crash data and the taxonomy that was developedby mike, and try to formalize it usingattentional theory. and the idea behindthis was that you
can think about just a resourcebased metaphor, where you have the activities associatedwith driving, that are kind of category x. and there'sa set of non-driving related activities category y. and then you can thinkabout the kind of activities within the driving scope thatare hierarchically organized in terms of traffic safetyand safety criticality. so that, for example,category x-a, or activity a withinthe context of driving,
would be more important atany particular point in time than b, then c and so forth. and so you'd have some kind ofprioritized list of activities that will besupportive of driving. and that you couldbasically then develop they an optimal frameworkfor describing how a driver that is fullyattentive would look. and then start to look and map,from a theoretical perspective, the different kinds ofcategorical taxonomies
that mike developed. and it might look somethinglike this using a simple model. this is an old model thatdanny kahneman developed. but you have anumber of activities. those activities can bereceiving different levels of attention based onthe momentary demands and intentions, and howwell your performing. and the more attentionyou pay to something, the better you'regoing to perform.
and so you could havedifferent kinds of activities that are either from the set ofdriving related activities, x, they're getting moreattention or less. and it might looksomething like this. so this is kind of abusy slide, but think about the blackas-- first of all, you have a numberof activities that are related to driving-- thecategory of x. that might be things like maintenanceof lane positions,
steering, heading, speed,scanning for hazards, and so forth. and don fisher is going totalk a little bit about some of the categories that areassociated for safe driving and attention, and that domain. and you can have the relativeprocessing priorities. so the optimalmight be something like this, where the mostimportant task associated with driving atthat moment in time
would have the highestlevel of attention. and it would go down in aprioritized list associated with the most relevantto the least relevant, and the highest prioritywith respect to driving. you can have the differentkind of categorical taxonomy that mike developed,for example. where you might havemisprioritized attention. where you have actuallyallocating more attention to some subordinate,secondary activity
that's related todriving, and not as much as you should to theprimary, most critical thing at any point in time. and without goinginto too much detail, you have situationsover here where you have driver divertedattention that's either-- in the case of thisorange bar, situations where people areallocating attention to some activity in thevehicle that's not really
that important for safety. like maybe aflashing light that's saying that you'rerunning low of fuel. or you can have driver divertedattention where you're actually taking attentionand allocating it to some secondary task liketalking on a cell phone, and not allocatingproper attention to the maintenance ofthe overall vehicle. so the idea behindthis is, frankly,
that you've got a taxonomythat was really wonderfully developed by mikeand colleagues. the question is,is it sufficient. do we need to basically addmore categories to that taxonomy to be able to be fullyrepresenting the data. are there some categoriesthat can be pulled together. you can basically taketwo approaches to that. one of them--beanland, for example, came up with some ideasabout how driver cursory
attention and driverneglected attention. they suggest thatin some situations, you can't reallydifferentiate between that. either that means weneed another category, or maybe those twocategories need to be pulled together into one. same kind of thing with drivermisprioritized attention, and this activity overhere, where you're basically diverting attention tosome activity that's
not that safety relevant. we think that tomake this useful, we need to basically tryand use both a data driven and a theoretical approachto decide whether or not you're going to want tohave more categories that are actuallymeaningful, or maybe consolidate some ofthese categories. and that's kind ofthe research effort. at this point i'll let you kindof move on to the next thing,
if you want, aboutthe conclusions. thank you dave. there's just two slides left. so in conclusion, thecategories that were proposed seemed to be sufficientlydifferentiated for classifying index crash dataexcept, as dave said, for a couple of thosecategories where they had a little bit oftrouble distinguishing, in particular,driver misprioritized
attention from driverdiverted attention that's driving related. i talked about that before. we also mentioned thatthe inattention categories in the originaltaxonomy don't appear to be sufficient forclassifying crash data because, at least for the firstgroup of people to use it, they felt there needed to beanother category included, this so-called undifferentiateddna, or dca category.
and finally, the taxonomy. when we had a closelook at the taxonomy by johan engstromand his colleagues, it seems to be reallyquite similar to ours. and i think it's justpackaged differently. and so you could takeone or the other. so i think, finally, in termsof further research needs, i think it's very important todo further research to validate certainly the originaltaxonomy, and and,
the more recent taxonomy byjohan engstrom and colleagues, using not onlyin-depth data, but data from naturalistic drivingstudies, which will give us extra insights into theutility of the categories. we need to createnew research programs to understand the impact of eachof these different categories of inattention, ifwe call them that, on driving performance,crash start and crash risk. there haven't beenany concerted programs
that have looked at each ofthose different mechanisms in isolation. and finally, andvery importantly, if we're going to makethese taxonomies useful, we've got to revise existingin-depth crash investigation protocols so that they, infact, align with the taxonomy, so that they're capable ofactually collecting data to confirm theexistence or otherwise of these differentmechanisms in crash data,
and so that we can providemore comprehensive, reliable, and consistentinformation regarding the contribution ofinattention generally to crashes of all types. so they key messagein this talk is we need to know whatwe're measuring, and it's only by knowing whatwe're measuring that we can make some sense out of what isactually happening out there. so i thank you for your time.
yes. ok, great. we do have sometime for questions. there are two microphoneson either side of the room. if you'd like to go toone of those for questions you might have. and please identifyyour yourself, your name and your affiliation. this question from somebodywatching the webcast.
this is for mike. you claimed that you derivedthe taxonomy without theory. but how did you score thedata without a theory? how do we score-- iknow richard young. i know his questionsare ones that have to be thoughtabout carefully. i'm not sure exactlywhat he means by how do we score thedata without theory. but all i can say isthat vanessa beanland
and her colleagues were ableto score the data in the sense that they had crashdata, they could go through theindividual narratives, and they could attemptto make some attempt to determine into whatcategory of inattention the data fitted into. so i can't say anymoreabout the question because i don't totallyunderstand what he means. unless you do, dave.
basically the ideais, you started off with a data driven approach,letting the data speak for itself in terms of whatthe sources of inattention distraction were. and from that youbuilt the taxonomy. it was tested in kind ofa empirical assessment with beanland et al. and now you'retrying to add to it a theoreticalinteraction to look
at the components ofdifferent levels of attention, and how they map ontothat data driven approach. but you definitelywere using a framework for initially studyingdistraction to begin with. i think the next stepnow is-- the next step in the validation, too, isbeing able to try and make predictions based on thetheoretical characterization of each of these differentforms of inattention, of how that might affectcrash performance,
or might affectbehavior, and hence, crash types and crash outcomes. hey, mike. don fisher. umass amherst. i think you mentionedin one of the studies-- it was the 2014 study--that 64% of the crashes were categorized asone's due to distraction. is that roughlycorrect as a memory.
it was 64%. 64%. ok. can you who recall whether theywere able to partition those into crashes which largelyoccurred while the driver's eyes were on theroad, and largely occurred while driver's eyeswere off the road, or at least their eyes being off the roadwas a part of that cause. because if we thinkabout what it is we
need to do in order toreduce distraction, where your eyes are whenthe crash occurs is of some considerableconsequence. it may be orthogonalto what you did. i don't know, but i am curious. that's a good question. i didn't do the study, so i-- that won't get youout of answering it. but i'd say it was probablyalmost impossible for vanessa
and her colleagues to digthat deeply into the data. because remember, thiswas an in depth study, but it wasn't anaturalistic driving study. and even with mostnaturalistic driving studies these days, you've onlygot fairly crude measures of where peopleare looking anyway, unless you've gotan eye tracker. and even with eye trackers, aswe know, there are problems. so i think the short answeris, i'd say probably 99% sure
that they weren't ableto collect that data, or have access to that data. i wanted to askmore about, i think it was the mis-prioritizedthere were the two categories that we'rekind of similar, and you had anexample where someone is looking too long over theirshoulder when they're merging. just thinking commonsensically about that, it seems like you would onlyknow that after the fact.
and that it would beextremely situation specific. that might be a mistake mergingone time, but not another. so just in terms ofusing this to try to develop countermeasures, orto apply it to help drivers, that seems very hard. those ones. where the driverseems to be doing what you want the driver to do,including looking at the road, but because of thespecific circumstances,
it's not the right thing to do. that's a very good question. and that's one ofthe problems when you have thesesorts of taxonomies that are data driven. that they're off in data. that it's with thebenefit of hindsight that you can make ajudgment that they've mis-prioritized their attention.
i think the interesting thingfrom a countermeasure point of view is that we knowthat they are suspect. and you could predict. it sort of gets back to theissue of making predictions from the forms ofinattention that you've got. i would predictthat young drivers, probably, if you lookedat them as a group, would suffer morefrom what we call driver mis-prioritized attentionthan more experienced drivers,
simply because-- andi've done research in this area, whichi need to publish. but basically theissue there is the that because they'reinexperienced, they don't often knowwhat is the right priority to give to the differentactivities involved in driving. and so they may not focuson the activities that are critical for safe driving,and may focus instead on what's happening just immediatelyin front of them,
rather than what'shappening way ahead of them. so i think the answer is, iagree with you that hindsight bias is a real issue here. and it's one of the thingsthat we talked about in trying to developthe second taxonomy. and if you read the report thatcontains the second taxonomy, it has quite a discussion aboutthat issue of hindsight bias. and as a result, the definitionsof inattention and distraction are quite differentfrom the ones
that i've talkedabout here today. deliberately so, so there'sno application there that the driver's to blame. and that's all abouthindsight bias. jim foley. toyota collaborativesafety research center. when you talk aboutinsufficient classification, is that really a weaknessof your taxonomy or model, or is it simply thatthe coders can't
discriminate betweenthe two cases. so is it an applied problemor a theoretical problem. that's a very goodquestion, jim, because when wedeveloped the taxonomy, we didn't want tosort of restrict it to only those categories ofan inattention for which data could be practicallyextracted from in depth crash studies or naturalisticdriving studies. we tried to keep itas open as possible.
and so when we're sayingthat the taxonomy is a little insufficient,what we're saying is exactly whatyou said, i think, which was the latter point. that it was insufficientin the sense that from a practical pointof view, the people weren't able to clearlydistinguish easily between the datathat fell into one or the other ofthose two categories.
that's what wemean by sufficient. so in that sense, we neededanother category in there that could-- so that'sthe case for that. and then the otherone-- the other category that they discovered,which was more to do with insufficiencyrather than practical issues in differentiatingbetween categories. which was not being able totell whether the paper looked, or looked but were notproperly attending.
so that one was moreabout the sufficiency, and whether we haveenough categories. the other one was moreabout the practical problems in differentiating betweensome of the categories. they're two different issues. we've got time forone more question. this is john lee from theuniversity of wisconsin. and i'd like to followup on anne's question. it's a really interesting point.
there's this problemof hindsight bias, and causal attribution post hoc. looking at the crash, andthen attributing distraction-- rather causes--resulting in that crash. so the way out of that seemsto be this allocation policy, the optimal allocation policy. that you can sortof say this is how people should belooking at the road, and the difference fromthat is, in a sense,
the cause of the crash. so related to that, i'vegot a couple questions. one is, if somebodycrashes and they're following the optimal allocationpolicy, are they distracted. or how do you labelthat situation. and then second, where doesconversation, texting, surfing the internet fall into theoptimal allocation policy. does the optimalallocation policy allow for us to tune theradio, turn up the volume,
turn down the volume. and if so, have you thoughtabout the cost function that balances safety withentertainment, enjoyment in driving. that's a complicated questionsfor a jet-lagged aussie on a monday morning, john. that's three into thatlast one question. i think the first one-- alli can say about the first one is that you can have anoptimal allocation policy,
i'm sure, andstill have a crash. and it's not necessarilydue to distraction per se. there could be lots of factorsthat could lead to a crash, even if you've got theoptimal allocation policy. now what was the second one. can you just remind me. is there ways in which youcould be allocating attention to a cell phone or somethinglike that would still be optimal.
i think the answerto that basically would require us togo back to the earlier work on time sharing. and one of the things i'venoticed about this field is that we often don't go backto the very early work that was done by wickens,and posner, and mccloud, and people like thatback in the '70s. and i think one of theconclusions i'm coming to-- and i had a chat with a guycalled marco dozza in sweden
recently, whose donesome quite interesting work on self regulation. we call it how youmanage a distraction. he came up with a littleequation, which is something like distractionequals interference minus self regulation. so what he's sayingis, someone can drive, and you can have someinterference going on between activitiesconducive to safe driving
and a second activity thatwon't diminish your performance. but ultimately, the neteffect is what's important, and that is how well youmanage the interference, and overcome the interference. i suppose the answerto the question is that there are somesituations, i think, in which you can effectivelytime share between activities critical for safe drivingand secondary activities and get away with it.
but i think it's verycontext dependent. i think that's the issue. in the middle of the desertwith nothing else going on, you can time share quiteeffectively sometimes, but as we know, demandfluctuates over time. and that and otherfactors can lead to a breakdown intime-sharing performance. thanks to mike and dave. our next speakeris dr. dan mcgehee.
and dan, i know you'vegot to change out the equipment, right? i thought you wereusing your computer. all right. i don't have your--that's what we discussed. if you'll just getme a minute, i'll try to find your presentation. do you have it separately? that requires the computer.
just one moment. sorry. i don't know if i downloadedit, is the problem. let's do that. that's easiest. because we had that discussion. anyway, while we're doingthat i will introduce dan. so dr. dan mcgeheeis the director of human factors invehicle safety research
division at the university ofiowa public policy center, ppc. in addition to hisposition at the ppc, he also holdsadjunct appointments in the colleges of engineeringpublic health, medicine, injury prevention researchcenter, and the national [inaudible] driving simulator. his interests are in driverbehavior and performance assessment, crash avoidance,driver distraction, crash decomposition, in-vehiclesystem design and evaluation,
naturalistic driving,teen driving, traumatology and safety policy. dr. mcgehee earnedhis phd in england at the university of leeds. his phd dissertation examineddriver pre-crash response, and was funded by the uscenters for disease control and prevention, and bynissan north america. general motorsfunded his ms thesis at the universityof idaho, where
he developed the first prototypeforward collision warning system for gm. and with that, dan, doyou need audio on this? thank you very much,sue, and thanks to state farm and the aaam. i really appreciate beinginvolved in this process. it's really a greatopportunity to work with so manydifferent individuals across the spectrum.
as frank mentionedearlier, we really have a greatinterdisciplinary team of physicians, policymakers,engineers, psychologists, which is really neededto come to the table here to look at these really complexissues in driver distraction. so what i'm goingto talk about today is some differentmethods and metrics that are out there, thatare evolving over time. there's really still nodefinitive metrics and methods.
over the last 20 years--over 20 years now-- we've been studyinglots of different ways to measure driver distraction. but they many times don'tagree with one another. so we're constantly insearch of the right way to measure distraction. so when you take a look atwhat we're really examining, it's really kind of takinga look at how the driver is neglecting the roadway, andthe performance is declining.
and so the kindsof measures that we use our speed variation. so when you're involved in aheavily demanding task, what we see-- and the firsttime that i really saw this was 20 years ago. we did a study in orlandoon a navigation system called travtech. and that system was reallyone of the first devices that was put into cars.
the first generationgps navigation system, where it was really visuallydemanding in the sense that you're trying tonavigate to a route. and i sat in the backseat ofan oldsmobile toronado trofeo for 14,000 miles aroundorlando in three months. it was quite interestingto firsthand see the results of these kindsof attention demands. and what was most salient-- whenpeople are really working hard trying to figure out whatthat display was saying,
or what that paper mapwas trying to adjust, even turning that paper map,trying to stand on their head to make sure that they'returning left instead of right-- was that their speedwould vary greatly. and we would be drivingdown the freeway, sometimes at 40 miles anhour, while they're trying to figure out if they'regoing to make that exit. other times that speed wouldgo up to 80 miles an hour. so we see lots ofvariation in that speed.
then we get to lateral control. so when you're lookingaway from the road, we tend to drift in ourlane, because we're not maintaining that lane position. and certainly eye glance is acritical aspect of distraction that's actually veryeasy for us to measure, not only in theexperimental context where we use eye trackers--computer eye trackers that enable us to look preciselywhere a driver is looking.
but even fromnaturalistic data, we can essentially plot outdifferent areas of interest. for a number ofdifferent projects, we can get aboutnine different areas of interest outside ofthe forward roadway. looking at mirrors, lookingat the center stack. lots of different places,and go frame by frame, and look at thoseeye movements, those transitions to thedifferent areas of interest.
then headway. when we are following avehicle, that gap between us can also vary quitea bit as we're modulating that acceleratorpedal while we drive. and of course, there's responseto different roadway objects, whether they're vehiclesor sudden emergencies-- pedestrians, and so forth. how do we respondto those while we're having a dividedattention task going on.
in terms of the metrics,we use a number of tools, from simulators, toinstrumented vehicles that may be on the road, where we dofield experiments where we set up preplanned tests, where welook at a specific scenario. we're able to capture driverattention demand in context. and then of course,naturalistic driving, where we put systemsinto your own car, where we actuallysee how you perform in your ownenvironment over time.
now we're really seeingin the real world the consequence of distraction. most crashes, asmany of you know are called single vehicleroadway departures. and when you livein a rural state, like i do, many crashes--many crashes in the us, north america-- are singlevehicle roadway departures, where are we drift in the lane. we drop a tire and there'snot a lot of room to recover.
and so unmaintainedlips of roadways-- you have a foot before you'reoff into the ditch, you over correct, so we see thosekinds of rollover crashes. you hit a culvert. you hit a tree. there's really not a lot offorgiving roadway departures in the rural environment,whether they're off to the right into aditch, or off to the left into a head on crash.
so the mostpractical consequence is really the lane departure. now, we have a greatnumber of tools that have advantagesand disadvantages, and how they play togetheris very interesting. and we really needall of these metrics. so we have in the upperright here, the fmri scanner, or some kind of scanner that'slooking at brain activity. now you can argue whetherthe kinds of research that's
been done out there, where youlay-- i don't know how many of you've had an mribefore, but you're put into this tube where there'sreally no information in there, but you can sort of puttogether a series of mirrors and everything, and kind ofbroadcast a scene of a road. and then you'll sort of tap yourfinger to control the drive. and so through these differentfinger tapping tasks, there are some studiesthat have been published that purport thatthis kind of thing
can look at changes in brainbehavior through imaging. as we move across thespectrum of tools, we have simulators like thenational advanced driving simulator-- which is reallythe big daddy of simulators-- but all the way down tosmaller fixed-based simulators. test tracks out there. field experiments like thefield operational tests that have been done. the university ofmichigan has done
a number of reallygreat studies. then we can go intonaturalistic driving, whether it's drive cam or thevirginia tech sharp 2 devices. and then finally, we havecrash epidemiology data, so [inaudible] and a number ofother really terrific databases are out there. but really the issue that welook at across this spectrum is that experimental controlvaries greatly along this. so in the tightly controlledlaboratory experiment,
we can put every single personinto the same exact scenario every time. as we move acrossthe spectrum, we start to lose thatexperimental control, and the further you move tothe right in the spectrum, the more we move into sortof the case study domain. so we can look atnaturalistic driving, but every case that we lookat in terms of distraction is really to that exact moment.
so the roadwayinformation is different, the traffic informationis different, the environmentalinformation is different. is it raining. is it dark. is it light. the sleep state of thedriver is different. other issues on board--whether they have passengers, whether they havethe radio on-- all
contribute to a differentexperience at that moment. so all of those things needto be carefully considered in doing this kind of work. so we're always lookingfor how and what the gold standard is in this. and this is really, of allthe traditional scientific methodology,experimental psychology, we really kind of take a lookat first asking a question. we construct a hypothesis.
we manipulate andcontrol variables. then we do the analysis. and then we refineour hypotheses, run the experiment again,then report those results. in naturalistic driving,things are a bit different. we can still, obviously,construct hypotheses, but now we're using fieldobservations and aggregated data to try to understand howto answer those questions, and data mining to reallyget at the key issues of how
distraction and otherbehaviors work out there. so we say on thebottom here, really both approaches arevery complimentary. we need to use themtogether to really try to understand thesecomplex phenomenon. in terms of the methods, iwant to just go through briefly here, and describe someof the measures that are being used right now. there are many, manydifferent methods.
so i concentrated on somethat the sae and the iso are using and adapting. these are very muchliving documents. were learning a lot about thesedifferent methods over time. but really, they'reused to sort of predict whether a secondary taskis distracting, especially in the context of the designof new infotainment systems, and other devices thatare in the vehicle. and really, these aremeant to sort of predict
whether a product is going tobe distracting in the future. and so standardizingbodies like the sae and iso are really importantto organizations to sort of follow,shepherd, understand, iterate these differentkinds of standards. now a quick history in termsof where this all got started. and some of you have seen this. [video playback] --as shown by this experimenton the attention span
needed for safe driving. the experimenter is [inaudible]director, psychologist john senders. driving along asection of route 128, and i'm trying to get anestimate of the demand-- [end video playback] what you're seeing here-- let'svisualize-- is professor john senders with a helmeton with an occlusion screen that essentiallycomes over his face.
and this was donein the 1960s, where he was able to actuallyput people onto the road with this visorthat would come down over the top oftheir face, to try to understand how muchattention it took to drive. so what he had was alittle pedal on the floor, and you would essentiallystart driving down the highway. and he would closethis mask, and then you would press a buttonto open it again,
to get a gulp of attention,and then it would close. and so he determined howmuch attention-- and this is oversimplifyingthings a bit-- it would take to drivedown the road. and this videosort of illustrated this really interesting,quite primitive methodology to really sort of occludethe driver's attention, and see how much youreally need to drive. it's a very interesting andgroundbreaking experience.
in fact, john lee and i'vetalked to professor senders about getting the old film--this black and white film-- of these experiments, andputting together and sort of recoding this information,in a light of what kind of tools we have today. so later on-- and you can'tread this from the back, but what were really in the1980s and early '90s sort of developed a visualsampling model where essentially,you're looking
at how much time it takesto extract information, or to chunk thatinformation before you need to return your eyesto the roadway. and so he found that aboutone and a half seconds, up to two seconds, you needto look back to the road before you had toget back to driving. and in the paper, it describesthis a little bit more. so this is a very importantmodel in terms of the history of how we sample informationoff of the roadway.
so we'll talk aboutseveral metrics that you are beingused out there today. so starting with differentperipherals stimuli, and how we can geta subject driver to respond to andacknowledge that they're seeing some kind ofperipheral stimuli. whether that's using a tactor--sort of little vibrating buzz that we put ontoessentially your clavicle. whether it's aperipheral stimuli
that's down andout of your view. and then the modernway on the lower left here is something calledocclusion goggles, and i'll zoom in there. and then we also have somethingcalled the lane change task, which essentiallyasks you to change lanes, sort of predictablywhile you drive, while you're doinga secondary task. so we can look atthe different lags.
and then ourcolleague, joel cooper, has also anotherperipheral detection test there that i'll talk about. so this is on oneof our simulators at the university of iowa,looking at a secondary task. so on the front we have an eyetracker-- the seeing machine's eye tracker on thedashboard, there. and then we havea participant here that's doing someperipheral detection task.
so we have a little ledthat's pointing just on the outside of her view. and then she haslittle finger switches, so that she can put her handsanywhere on that steering wheel while she's driving,and then sort of tap them to give a response. so while we're doing adistracting task-- and this is part of the connectedvehicles project experiment here, where we're lookingfor different features
of a roadway. so say you're looking fora starbucks that's ahead. you're getting acoupon flashed to you. it's time to get your$2.00 off your latte. how hard do youhave to study that, and what kind ofperipheral information do you miss while doing that. so we can look at notonly the head mounted peripheral detection, whichis really good because you can
move your headaround and still have a peripheral object out there. we can keep a fixedperipheral target, and then we can have thistactile information, as well. so they're differentareas of examination of attention usedwith that method. and then these areour occlusion glasses. this kind of looks like a prisonphoto or something like that, but this is one of our students.
and joel cooper alsosort of developed a more nuanced, amore advanced version of the occlusion glasses. but this really gets atwhat john senders was doing almost 50 years ago,with this really big visor. so we can use lcdcrystals, essentially, to turn off what you see ahead,and adjust the timing of what we block out while you drive,and look at your performance. sort of simulating how much timeit takes while things are not
being seen out there. then the lane change taskis a really simple task that was developed ingermany, where you essentially drive down a simulatedroad, and then you get these differentsigns that tell you to change laneswhile you drive. and so you're drivingdown the center, and this particularone says ok, now you've got to get into the left lane.
and so, while you're driving,you're doing a secondary task. you're punching a navigationdisplay, a new infotainment system, some sortof sustained task. and then you arecommanded to change lanes throughout thisstraight road drive. so what we do then is thatwe measure the difference in the lag in yourresponse, relative to what would be the perfectpath out there. and then finally, we lookat other situation awareness
queues, and don fisher'sgoing to be talking about this in much more depth, but i justwanted to sort of mention this, because this is one areathat we're looking at. and this could bejust while you're under some sort ofa load, but there are different roadwayobjects and features that are out there. and then we'll come back laterand say, did you see object a, that trunk in the road.
or did you see thatbroken tire out there. so a couple of issuesfor consideration. we have a reallygood grasp, i think, on the visual manualdistraction elements. pretty easy to measurethat kind of work. the auditory vocal distraction,cognitive distraction is much moredifficult to examine. and i can show manyvideos, and many of you have seen a lot of thekinds of drive cam videos
that we have ofcrashes, near crashes, and so forth, that addadditional context, and show the differentkind of complexity, provide information on what ispresent during these things. and i think, as wasmentioned in the last talk, the hindsight issueis significant. and the way we sortof examine that is that we only codewhat's present. we don't make any judgmentsof yes, that was the cause.
we just say, the radio wason, they're on the phone, they're not on the phone. we just say here are allthe things that are present. because those kinds ofcognitive distractions are just reallydifficult to interpret. because i can show avideo, for instance, of a phenomenon that'sbeen around for a long time in the literature called,look but don't see. so far before cellphonescame on the scene,
we had-- rear-end crashesrepresent a substantial part of the kinds of crashes outthere-- but we have crashes where people arelooking straight ahead, and they drive right intothe vehicle ahead of them. and that's due to somecomplex perceptual factors that we publishedover 20 years ago, and sort of decomposing allthose different perceptual factors. now when you put a person onthe phone, talking on the phone,
they're still susceptible tothe same perceptual limitations that they had before the phone. so it's reallynot clear how much the phone has involvementin that kind of a crash. so naturalistic drivingcan be a challenge for us to interpretbecause we certainly have lots of mind wandering,blank stares out there while they'redriving, but we really don't know what's going on.
and there's been some reallygood recent literature on mind wandering that will bevery instructive in the future. and this is really whereexperimental studies come in, and try tofill in the gaps. where we can nuance some of ourquestions a little bit better, put them into controlledcircumstances, and really try to getfurther at those behaviors. so we really needboth of those things to answer some ofthese questions.
so, what is thefuture of all of this in terms of measuresand metrics. first is that we reallyneed to understand that distraction-- thisexposure crash paradox. and really, what we'reseeing is this vast increase in the amount ofinfotainment systems, in cellphone use, smartphoneuse, the kinds of social media that are intenselydemanding in terms of visual mentalmanual interaction.
changes fromtalking on the phone to much more visual manualinteractions with twitter, with facebook, withinstagram, with snapchat. all of these things arechanging almost daily. and at the sametime, we're still seeing a fairly dramatic falloverall in the crash rates. and so that'sreally, i think, one of the most puzzlingand vexing questions, is to try to figure out how wecan have all of this seemingly
really dangerous behavior thatwe look at in the laboratory, but it doesn't really translateinto big crash rate increases overall. so the future is reallytrying to understand the role of selfregulation in this. and there's been acouple of good studies. i think i really likemarco dozza's work at chalmers universityin gothenburg. some of the other work that'sbeing done in self regulation,
because this is really going tobe the essence of how we study how people are choosingto be distracted. what kind ofstrategies they use, if they're thinkingabout it all. but it can be assimple as, i'm going to wait to send a textmessage, or read a message, or type a message wheni'm at a stop sign, or at a traffic light. that's a strategy.
it may not be an overtstrategy, but that's something that we're seeing a lot moreof in our naturalistic data. so we also need tobe able to figure out how we can predict the differentpersonality variables of who chooses to be distracted,and who self regulates this kind of behavior overall. and we really need totry to get some real time metrics in the car to alertthose that are driving. we have some reallyfabulous prototypes,
and even someproduction systems now, that look at head pose and eyesoff road to kind of wake people up to get them back on the road. and then of course, we havedifferent safety systems now that will kind ofkeep you in your lane, warn you when you drift out,when you start to drift out. because all of us sort of have aunique signature, whether we're impaired, alcoholimpaired, or we're texting, our steering profile changes.
the way our steering variabilitychanges is very unique to us under different circumstances. and then finally, wereally need to ensure that all types ofdistraction are covered. in a study that we're just goingto be publishing here shortly, funded by the aaafoundation, we're finding that among teens,that passenger distraction is by far the most dangerouskind of distraction for younger drivers.
far, far moreimportant to examine than even texting,which seems like that would be the reallyimportant element. so we need to really look at alldifferent kinds of distraction. external distraction--there's just so many different categoriesof that out there. so i will be happy toanswer your questions. could you go to the mic, please. dan, i'm not going toask anything too tough.
oh, good. i've just got a comment. it's something that'sbugged me for a while, because you've got allof these different tests that we've developedto try to understand the impact that distractionhas on performance and safety. and one of thethings that's bugged me a lot about thisexperimental work-- as i said, it's acomment-- is that i
think there's beena general failure to control for a lot ofconfounding factors that could influence the degree of impactthat that distractor has. and we've beenfocusing on the test, and i think maybenot focusing enough on the methodologies that areused to utilize those tools. so for example, we know thatthere are characteristics of the driver thatmight moderate the impact of the distraction.
if they're moreexperienced, they might find that thesource of distraction is less severe than forsomeone who's is inexperienced, for example. and it could be that we furthercontrol for the driving, for the context, for the demandsin which the person's driving so in a simulator. because the demands of drivingwill have an affect on-- will influence the degree towhich that particular source
of distraction has an impacton behavior and performance. and then finally, as youalluded to, self regulation. the ability of thedriver to self regulate will have a very importantinfluence on the overall impact of that source ofdistraction on driving. so i think we probably need tothink a little more about how we control ourexperiments, so that when we're comparing the impact ofa given source of experiment between different layers, we'veutilized a common methodology
that has taken into accountthese different moderating the second thing thati just wanted to say is that i was reallyinterested in that comment that you made, becausei've heard this repeatedly. that people are beingexposed more and more to sources of distraction,and yet the crash rates are going down. and i just wanted to clarifyexactly what you mean by that. because if it's just crash ratesgenerally that are going down,
they could be going downfor all sorts of reasons that have nothing todo with distraction. but if you're talking aboutdistraction related crashes-- and i'm not sure whether areor if that's another issue-- but it could also be thatdistraction related crashes are going down because thepolicies that you have here in the us formanaging distraction are effective, even though youmight not have evaluated them. so i was just interestedmore in that latter issue.
are you talking aboutdistraction related crashes, or crashes generally. in general, we're seeinga pretty marked decline in crash fatalitiesoverall, and it's a really complicated question. it can involve some economics,and certainly our vehicles are getting safer. there's many different issues. and i think that topicalone is something
that we've been talking abouthaving a panel at the driving assessment conference to tryto get different perspectives. certainly with this kindof focus on engaged driving that's being talkedabout here today, the purpose is really totry to even further reduce those crashes and fatalities. that's all of ourinterest out there. and so this is reallythat opportunity to be able to havethis dialogue,
to try to reducethose even more. but i think just asa general question i ask, we see prettydramatic declines, and really sharp increases in the overalldistractability of things. i think john lee and otherswill be talking about some of those issues inmore detail later on. for me, i can't see theparadox for that reason. there are lots of other factorsthat could bring it down yes, absolutely.
thank you, dan. i have another question fromsomeone watching the webcast. you pointed out that differentmeasures of driver performance can give different answers,which is certainly true. do you agree thatone reason may be that these different measuresare each giving a valid answer, but they are just measuringdifferent dimensions of driver performance. just like your height andwidth are different dimensions,
and an object can be tallbut narrow, short or wide. yeah, absolutely. i think there aredifferent times that you're going to employ thesedifferent kind of metrics for differentkinds of questions. the purpose of this sortof high level review is to really show whatthe spectrum of issues are with regard to themeasurement of driver distraction.
so i've got a question followingup on the second bullet there. developing a test to predictthose that can self regulate. would the consequence offailing that test be euthanasia? [laughter] it's difficult to know exactlyhow you would take that. whether you get remedialtraining, or what. so i wonder ifyou thought about, instead of developinga test of the driver to see how wellthey self regulate,
whether it would befeasible to develop a test of thetechnology to see how well the technologysupports self regulation. no, i think that'sa great point. i think in thinkingabout this and talking to many of thepeople on the panel, the originalthought was if there were different personalitycharacteristics that could help predict driving.
so from an insuranceperspective, could you identify ahead oftime somebody that may not be as prone to self regulatethan another person. but certainly i thinktechnology mitigation is a critical partof sort of pacing the driver with informationso that you can have a better context to present informationthan at the wrong time. and i think one issue--you can argue on one hand, for instance, thatchecking your text messages
or even typing out amessage at a stoplight is better thanwhen you're moving. at the same time, intersectionsare inherently complex. we have pedestrians. we have other vehicles turning. there's a lot ofother stuff going on. so it really becomes areally delicate balance of choosing what is rightin those situations. yeah.
with respect to, i guess,the self regulation, it seems like thatinformation would be very key to feed intothe driver training aspects. i know a lot of people in theus go oh, driver training. that's kind of athrowaway thing. but i think the debateis usually, right now, like this is bad, don't do it. and the kids are kindof like ok, well, what do you mean by it's bad?
yeah, texting is bad, but i seelots of gradations in driving. and what i try to teachmy kids is ok, well, if you're going to have tolook away, keep in mind, you've got to getback pretty quick. and that 1 and 1/2 secondsis kind of your duration. so think about whenyour radio tuning, to try to keep your eyesgoing back to the road instead of just trying to findthat station for four seconds straight.
so i think there may besome benefit to trying to provide some of thatinsight to help drivers learn how to self regulate. because you don't startoff driving knowing that. you have a lot ofoh, shit, moments. and hopefully you survive them. and you learn thatbased on that. and so the betterway we could feed some of that collectivelearning of how
do you self regulateinto the driver training, i think might have some benefit. right, i think that's a greatpoint. and one of discussions i have with some of my otherpublic health colleagues is really trying to look atthe history of abstinence only training in other domains. whether it's sex, oralcohol, or drugs. what's worked and whathasn't in those domains. because clearly people arestill going to do it, right?
and so what kindof information can we provide to helpmake better decisions when they're out on the road. hi, dan. a couple comments. bryan reimer from mit. first of all, i think yourpoint on self regulation is really key. and self regulation does presentitself quite differently.
highway, we see slower driving. we can all design anexperiment in a simulator where we don'tallow that behavior, and look at fardifferent findings than we observe on the road. but i think somethingthat we really need to think aboutin this area is, when we think about urbancontext, self regulation congestion, selfregulation-- i.e.
dial that cellphone whenyou're parked at a light-- is becoming a congestionproblem in many urban cities. so while it's not necessarilya safety related behavior-- we are safe sitting there at atraffic light-- it's promoting congestion, promoting road rage,and all those other things. i want to talk about a biggerpoint in some of our work that's arisen. we did some datacollection in 2008, and then recast ita few years back--
i think it's inaap-- where we looked at drivers under amulti-dimensional driving task on a highway, where theyhad no phone engagement at all. and they were engaged insome of our cognitive manned activities. but we took about 20minutes of data there. we looked at drivers whoself-reported engaging in conversations througha phone while driving, and those who don't.
and the folks that arewilling to at least admit that they talk on thephone while driving are fundamentally adifferent risk category. they are the folks whoare driving faster. the folks who are changinglanes more frequently. and they're folkswho are probably making all their poordecisions while driving. so we may be able to take thetechnology out of their hands. but there are the folkswho are going to eat,
the folks who are going topotentially drive with alcohol. so i really like themessage in engagement here in self regulation,because it really talks to a much higherpoint in the equation. how do we do begin to manageall the different conflicting elements that arecoming together, beyond the technologyof focus today. and as we talk aboutself regulation, we talk aboutdesigning interfaces
that allow for an eruption,or pacing more effectively, we're really talkingabout a topic that we can reallymake inroads and strive to make change around. yeah, i agree completely. and i think this is really--fora such as these are really going to be important inmoving that kind of discussion forward. with all the greattechnologies that we have,
clearly there is an importanteducation component to this, too. thanks, dan. i'd like to introduceour next speaker. doctor don fisherjoined the university of massachusetts amherstin the fall of 1982. he became the head of thedepartment of mechanical and industrialengineering in 2009, and has served inthat capacity since.
he is also a faculty fellowat the volpe national transportation systems centerin cambridge, massachusetts, and director of the arbellainsurance human performance laboratory. professor fisher receivedhis bachelor's degree from bowdoin college,master's degree from harvard, and phd from theuniversity of michigan. doctor fisher is the principalinvestigator or co-principal investigator on over $23million in research funding,
including grants fromseveral federal, state, and regional governments,private and public foundations, and private industry. his current research interestsare human factors in safety, particularly in transportationand health care. thank you, sue. i am-- are we live? can you bring up something? sue's responsible, iguess, for the display.
i'm presenting thiswork with david strayer at the university of utah. and i want to thank,as has everyone, the association for theadvancement of automotive medicine, along withstate farm, for making this opportunity possible. i'm going to begin bytalking about the difference between a novice andexperienced driver crashes. and i'm going todo that because i
want to create a linkbetween novice drivers with very little experience,and experienced drivers who are cognitively distracted. so in order to dothat, i need first to talk about novice drivers. as many of you know, the crashrates among novice drivers are some six tonine times higher than they are among moreexperienced drivers. and this is a tragedy thathas been long in the making.
the question iswhy novice drivers are more of a crash risk thanmore experienced drivers. and an answer was proposedmcknight & mcknight make in back in 2003. their answer wasthat novice drivers were largely clueless,not careless. and this was a radicalsuggestion at the time, because most people thoughtthat, indeed, novice drivers were careless.
so we set out in our labs todetermine whether, indeed, novice drivers werelargely clueless. and the reasoningwent as follows. if novice driversare largely clueless, then they ought notto anticipate hazards. and this had a nicelogic to it, since we knew from research by horswilland others that indeed, if you don't anticipatehazards, you're much more likely to crash.
the research we were going todo on novice drivers involved us putting them insituations where dangerous, and as dan mcgehee said, wewanted to be in a situation where indeed, we didn'texpose them to any risks, we put them in adriving simulator. this is our driving simulator. it's typical of many drivingsimulators around the world. there's a car. saturn sedan, 1995,surrounded by three screens.
in order to determinewhether novice drivers don't look for hazards, whatwe have the novice driver do is put on an eye tracker. and we use that eyetracker to determine where in the visualworld they are looking, while they are navigatingthrough that visual world using the steering wheel brake andaccelerator of the saturn sedan. this is a typicalexample of a scenario
we used to determine whetheror not novice drivers are likely to anticipatehazards less frequently than ourexperienced drivers. the videos i won't show. but i'll pretendi'm an eye tracker. if i am-- this representsthe cross-hairs of an experienced driver. the experienced driverwould move along here, then shift to the right to seewhether a pedestrian was
emerging from behind that truck. the novice driver,if i had the video, would indeed just move rightalong up and down the road, without ever lookingto the right. this is called amulti-threat scenario, because the pedestrian who couldemerge from behind this truck could be hit bythis truck, or could be hit by the car that'sin the right travel lane passing by the vehicle.
so the videos wouldhave suggested how we measurehazard anticipation, but hopefully the littlepointer did the same thing. what we want to know iswhether the driverr-- who is right there, andcan't see because of an obstruction, a pedestrianwho might be crossing behind the vehicle-- actuallylooks in that direction when, in fact, theypass the vehicle. so if they lookin that direction,
we assume theyanticipate the hazard. if they don't lookin that direction, we assume they don'tanticipate the hazard. simple enough. here are the resultsfrom that scenario. it was eye-opening at the time. this was back in the early1990s, for us anyways. 10% percent ofthe novice drivers failed to anticipate a hazard.
only 10% anticipateda hazard, i'm sorry. 30% of the experienced drivers,25 to 35 anticipated a hazard. and then some 60%of the 60-year-olds anticipated a hazard. so is it real. is the scenario i just showedyou on the driving simulator representative of whathappens in the real world? as many of you areall to aware, it is. there is a car right there.
you can see it'sbrake lights on. it's stopped in front of amarked mid-block crosswalk. you see a car in theleft travel lane getting ready to pass the car in theright travel lane-- in fact it is passing theright travel lane. this is an intersection on theumass campus, and as i said, unfortunately, thisscenario is all too real. allison lustig losther life here in 1998 at that scenario becauseof just that reason.
and unfortunately forme, it's particularly haunting because inone of the many offices that i've occupied,my window overlooked this exact intersection. so i want to segue nowfrom novice drivers to experienced drivers whoare cognitively distracted. and i want to do such, andask the following questions. imagine that insteadof the novice driver approaching the intersectionwhere allison lustig lost
her life, it's anexperienced driver who's cognitivelydistracted, who's approaching that intersection. and the questioni want to ask is whether the decrementedand the hazard anticipation skills of the experienced driverwho's cognitively distracted, compared to the experienceddriver who's not cognitively distracted, is the same asthe detriment in the hazard anticipation skillsof the novice driver,
when compared with theexperienced driver. so john lee and i werefascinated by this question. we decided to run an experiment. again, what you need to doin a hazard anticipation scenario like this,is track the eyes of the driver who's lookingat the virtual world. you need to know whether they'reanticipating those hazards as they're manipulatingthe controls of the vehicle through that virtual world.
we used many ofthe same scenarios in the study of the effectsof cognitive distraction on experienceddrivers that we used to differentiate betweennovice and experienced drivers. this, again, is themultiple threat scenario. the only thing i want tomention about the method here is what we used formock cellphone tasks. the participantswho were driving were read sentencesover headphones.
that's the bottom bullet. they were asked to repeatthe subject and object of the sentence, and to indicatewhether the sentence made sense. that is, we wanted to knowwhether they were really paying attention tothe conversation. here are the results ofthe study for the five multi-threatscenarios we measured. remember, there were a totalof 18 different scenarios
in the study, but these are thefive multi-threat scenarios. the dark blue barsrepresent the percentage of experienced drivers who werenot cognitively distracted, who anticipated a hazard. the light blue barsrepresent the percentage of experienced drivers whowere cognitively distracted, and anticipated a hazard. as is clear here, the blue barsare much higher on average. dark blue bars aremuch higher on average
than the large bluebars, indicating that, in fact, experienceddrivers who are cognitively distracted are much, muchless likely to detect a hazard thanexperienced drivers who are not cognitively distracted. scenario fivethere, is this one. it's a little different than thetypical multi-threat scenario. instead, the hidden threat,latent threat, is on the left. and you'll see that, indeed, 90%of the experienced drivers who
aren't cognitively distractedactually do look to the left, whereas people who havebeen driving all their life, and are doing thesecondary task, look to the left for a childthat might be crossing in front of the cars only20% of the time. david strayer, in the field, haslooked at much the same thing, only he's lookedat intersections. here you have drivers onand off the cell phone. on the y-axis is representedthe glance probability.
you'll notice that for driverswith handheld and hands free cellphones, theirchances of scanning for hazard at theintersection are significantly less than for thedrivers who are not engaged in a secondarycellphone task. and in general, weknow that in fact, it's the case that if you'reengaged in the secondary task, you're much likely to detecta hazard than in fact, if you're not engagedin that secondary task.
so the bottom line. in summary, justas novice drivers are clueless largelybecause of inexperience, it would be my hypothesisthat experienced drivers who are on the cellphoneare clueless, largely because ofcognitive distraction. to put it slightly differently,either way, novice drivers, because of inexperience,or experienced drivers because of cognitivedistraction,
are much less likely todetect an unexpected threat, to be aware of the situation,than are those that so david strayerand i put together a model with theacronym of spider which can be used topredict the likelihood that a hazard's anticipated. and the spider, if it'sattentive and constructs the web carefully, will catchthe latent threat in its web. but the spider, if it's notattentive and doesn't construct
the web carefully, willlet the latent threat slip through the web. now all of you canimagine at this point that i enjoyed somethingan awful lot as a parent, and that was reading to my kids. so i apologize forthe drama, but i don't get to do it very often. so the spider canpredict the likelihood that a hazard is anticipated,but we're really interested in
is the likelihood of a crash. and if we're lucky, we can movefrom predicting the likelihood a hazard is anticipated tothe likelihood of a crash. you can also seethe relation here between spider andsituation awareness. they're basicallythree processes in situation awareness--perception, identification, and prediction. so spider not onlypredicts the likelihood
that a hazard isanticipated, but for each of those processes, it willalso predict the likelihood that you perceive, youidentify, and you predict. the backbone of spideris five processes-- scanning, predicting,identifying, deciding, and executing and responding. what i want to do is go through,one at a time, the spider processes, and talk about howcognitive load can impact each of those processes inways we think are harmful.
so let's begin with scanning. it goes without saying that youneed to scan broadly from side to side. numerous studies have shownthat if you're cognitively distracted, in fact, youscan less far side to side, and less far vertically. given you've scanned and you'veseen that a crosswalk may, indeed, be located rightin front of the truck, you have to predict it.
but as navon and gopherand many others have shown, you have a much hardertime doing two tasks it once than you doone task at once. so the likelihood you'llsucceed at predicting even if you see theinformation is less. if we move on to the nextprocess identification, remember there is a lag betweenwhen you predict that, indeed, you should look here, and whenyou actually do look here. and as the peterson andpeterson paradigm show,
if the interval between whenyou're given a list of things to recall, and youactually recall something, is one which is filled witha cognitive distraction, you're much less likely toremember that, indeed, you were supposed to lookat a particular area, remember a particular word. so indeed, ifyou're multitasking, you're much less likelyto identify this. right here we have the decision.
again we've got multitasking. the driver is both focusedon the primary task, driving, but also the secondarytask, deciding, and so it's less likelythat that task will succeed. and finally we haveexecution and response. and i refer here to theinattention blindness studies, where the gorillawalks across the stage. you aren't likely to detect it. it's unexpected when youhave only a primary task,
as you saw on thatslide which i showed. if you have asecondary task, you're still much lesslikely to detect it. so what i hope i'veshowed you here is a concrete articulationof the spider processes, and how we couldexpect each of those processes to be impacted bythe cognitive distraction. so now i want to show how wecan transform spider, which is a model withqualitative processes,
into a model withquantitative predictions, predicting, indeed,how likely it is you are toanticipate a hazard. and for that, weuse a hidden mark off process witha state identified with each of thespider processes. before i do that,i want you to look at what we call the stopsign ahead sign curve. there is a curve herewith a stop sign ahead.
that's the yellow arrowpointing to the stop sign ahead. and around the curveis an intersection, and you can't seethe stop sign there, which is why you havethe stop sign ahead. i'm reminded of a trip itook with my older daughter. i told her aboutthe stop sign ahead. and she yelled at me. she said, daddy, i seethe stop sign ahead. what's your problem?
nothing, jennifer. nothing. no problem whatsoever. so as we came aroundthe curve, of course she didn't see the stop sign and sheblew through the intersection. anyways, it was a goodlesson to me, and to her, i think, in how not toteach, and how not to drive. anyways, here is thehidden mark off process. each of the boxes is a statein the mark off process.
transitions betweenstates are indicated by arrows, the directed arrows. so you can go fromstate s-1 to state r-1. you can go fromstatus 1 to status 2. in the top of each box--let's focus on the left-- is the process. remember, spiderhad five processes-- scanning, predicting-- so inthe top of status 1 is scanning. you should be scanningfor the stop sign ahead.
in the bottom, ihave the letters s-a. that's situation awareness. so baseline situationawareness is coded as yellow. at the end of a longsequence of processes, either you'vemissed the stop sign or you haven'tmissed the stop sign. if you've missed thestop sign, you've probably missed thestop sign ahead, and you've executed none of thepassing processes correctly.
and so i have a big redhighlight in the bottom, meaning you're notsituationally aware at all. on the other hand,if you're doing well, you have indeed seenthe stop sign ahead. you've seen thestop sign, you've executed each of theprocesses correctly, and everything's hunky dory. finally, there'severything in between, indicated in the yellow.
the terminal states are t-6and t-5, you saw at the end. what we want to dois be able to predict not only the probabilitythat you missed the stop sign, that you stoppedor that you didn't stop, but also the probabilitythat you crashed, given you were distractedor you were not distracted. and if we're lucky,we can do that. and if we can dothat, then we can predict the relativeincrease in risk.
that is, the ratio,the probability you crashed given that you weredistracted, the probability you crashed given youweren't distracted. ok, there have been twothreads to the story. i'm done after i talkabout these two threads. the first thread is thatexperienced drivers who are cognitively distractedare some four to six times more likely to miss a latentthreat, such as a child crossing in front of a truck,then experienced drivers who
aren't cognitively distracted. and in fact, if welook at novice drivers and compare those ratios,experienced drivers who are cognitively distractedare about as clueless as novice drivers withvery little experience. and what about spider. how can it help us here? it can help ustheoretically in two ways, and practically in two ways.
theoretically, we coulduse it to test hypotheses about how cognitive distractionaffects each of the latent processes in the spidermodel-- scanning, prediction, identification, decision,execution and response. we can also useit to predict how different elements ofhazardous scenarios contribute to the likelihoodthat drivers distracted or not distracted will detectan actual threat. practically, we canuse spider to predict
the improvements that wouldcome from advances in signs, signals, pavement markings,in-vehicle distraction warnings, predictimprovements to the likelihood that a driver would actuallyanticipate a hazard. and practically, if we can moveinside the vehicle in real time an eye-movement monitor,we can predict-- and we've heard otherssay about this-- but we can predict with a spidermodel, when the driver has fallen below a givenlevel of distraction.
and hopefullythereby, we can strike at the achilles heelof collision warning systems, which have waytoo many false alarms, but picking off the driver onlywhen he or she is cognitively distracted. i've enjoyed talking with you. thank you for letting me indulgein my childhood fantasies. and please don't drivewhile distracted. our children needour full attention.
ok, we have timefor some questions. we do have one. thanks, charlie. you saved me. it's embarrassing when youdon't have any questions. either the audience is asleep,convinced that what you said is full of it, or anyways. so don, you and i have talkeda little bit about this, as i have with others.
and i'm curious if you couldanswer this question now, if you haven't been in the past. so the question i have is thatyou show this really compelling evidence that cognitivedistraction turns experienced driversinto novice drivers. and we all know howbadly, and how often, novice drivers crash. so the question is, ifcognitive distraction is so potent in thenaturalistic data sets,
we don't see a similardecline in safety. the odds ratio of a crashwhile somebody's on the phone is around 1, ormaybe even below 1. so i"d just like foryou to expand on that. i think that's awonderful question. of course, i think about it. i worry about it. is the research i'mdoing full of it. the answer is yes.
but i don't drivearound town anymore. i guess i'll answerit personally. i don't drive around townanymore with my cell phone on, just because i believein my very heart of hearts that i'm less likely to lookfor little kids stepping out from behind a bus, orsomething like that. on the other hand, wherethe risk of that happening is much smaller,where the statistics, indeed, suggest i'mno more at risk.
as a matter of fact,i may be less at risk because i'm on the cellphone. john has some wonderful datawhich he hasn't discussed, which he may discuss shortly,so i won't give it away. anyways, there are all sortsof reasons to think maybe at times you are advantagedby being on the cellphone. i don't know. but i certainly don't think ina school zone or in a work zone, there's any reason whatsoeverto be on a cellphone.
second, i thinkthe opportunities for crashes in situationslike that are just very small. so it would be very hardto pick up these examples that i'm showing you,where your son or daughter might lose their life. it's just hard to pick theseup in the overall scheme of things. third, i think that there may besomething going on here, which are experimentsthat capture it all,
and that is thewillingness of drivers to engage in acellphone conversation. we didn't let ourdrivers-- adults-- not engage in thecellphone conversation when they were inthe school zone. it may well bethat in fact, they decide not to engagein those conversations during that period of time. so i think all ofthose are factors
which could explainwhy overall, we're not seeing in thenaturalistic data the increase in risk due tocognitive distraction, but why, in oursimulator studies, we do see that increase in risk. it's still the case that indeed,i put down the cell phone or don't use it when i'm ina school zone, work zone, or in a situation where thereare lots of latent threats which are hard forme to anticipate.
does that get at it? good. hi, vicky. hi, bill. do you want to go up here,vicky, and just solve it quickly? for those of you onthe webcast, we're having a little problemwith the microphone. again, bill windsor,watching the webcast.
do you think that the wayyou simulated conversations required more attention thana typical phone conversation, resulting in the conversationbecoming the primary task. would drivers in realworld drop conversation if they detected hazard? i definitely thinkwe pushed the limits, and we intended topush the limits. and bill's exactly right. there are conversationswith my mother
or the dean that, indeed,i can ignore entirely. hopefully the dean'snot listening. but he has a goodsense of humor. so, bill, in answerto your question. absolutely we wanted topush it as hard as we could. and i think if wepushed it less hard and you weren'tpaying attention, we'd get identical results. so there's no doubtthat our results are
a reflection of how hardwe pushed the driver, yes. and this is from richard young. if you're conversingon a cellphone, do believe it's safer to keepyour eyes on the road ahead, or to look left and rightat billboards and scenery? well, i definitely believeit's safer to scan. in scanning, it'simpossible not to-- indeed, as you scan left to right,it's impossible for your eyes not to intersect in thedistance with billboards
and other events. so i definitely thinkscanning is important. but that doesn't mean ithink you should necessarily read the information onbillboard, because that, in turn, leads to problems. so, richard, yes, i'd scan. i wouldn't lookjust straight ahead. that would be a mistake. but i wouldn't necessarily takemy eyes and mind off the road,
and look at a cute cowthat was doing whatever. ok, thank you, don. you're welcome. our next speakeris dr. john lee. doctor lee is the emersonelectric professor in the department of industrialand systems engineering at the university ofwisconsin in madison, and director of thecognitive systems laboratory. he is co-author of the textbook,an introduction to human
factors, and is the author andco-author of over 170 papers. his research focuses onthe safety and acceptance of complex human machinesystems by considering how technologymediates attention. specific researchinterests include trust in technology, advanceddriver assistance systems, and driver distraction. he recently co-edited editeda book, driver distraction theory, effects and mitigation.
thank you very muchfor the introduction, and thank you, statefarm, and aaam, and sue for puttingall this together. i think this is areally important issue. driving safety killstoo many people, disrupts too many lives. and i think the attentionis really well deserved, and i'm pleased to behere to talk about this. in a sense, my title could bechanged slightly with a colon.
and something like the limitsof individual responsibility. and i'd like to startwith a brief story that sort of illustrates this. this happened to me as i wasdriving back from iowa city. i was collaboratingwith the folks at the national eventsdriving simulator there. driving back from iowacity to madison, wisconsin, where i live. and as i was enteringmadison, i was
listening to the infotainmentsystem in our car. sort of like theradio, but not quite. and it was adele. and nothing against adele, but ididn't want to listen to adele. i wanted to listento bruce springsteen. so i reached down and startedscrolling through the songs. in this car, thisinfotainment system, every time you put a cdin, it conveniently copies all of the tracks onto alarge, one terabyte hard drive.
and so my wife had beenlistening to adele. a lot of adele songs mixedin with bruce springsteen. and so i was scrolling throughthis long list of songs, trying to findbruce springsteen. so adele, springsteen,adele, springsteen. and then i realized i hadn't beenlooking at the road for like three,four, five seconds. i don't know exactly howlong, but it was a long time.
and when i looked back,i thought oh, my god. i could have crashed. nothing untoward happened. nothing bad happened,only because nothing happened in front of me. i was lucky. but this, i think illustrates,some important things about driving and distraction. and one of them is that becausei was coming into madison
on this rural highway,it wasn't very demanding, and i was somewhatdisengaged from driving. and then i had thistransition where, all of a sudden i was now inthe somewhat urban environment where i should have beenengaged, but wasn't. and i was frightened by thatwhen i looked back to the road. the second thing is that theseemingly familiar stereo system isn't. this isn't likechanging the radio.
it's not likechanging the radio. it's much more complicated. and the third thing thatis, i think, insidious about this problemof distraction is that if i had not beensomebody studying distraction and realizing the risk of along glance way from the road, this wouldn't havebothered me at all. nothing bad happened. i reached down,searched for some songs,
looked back to theroad, and continued on. nothing happened. the feedback is really poor. so these threeelements, i think, make distraction a really toughproblem, and somewhat question, can individualresponsibility mitigate the risk of distraction. so i learned somelessons, and i think they are related tosome of these topics
that i'll be talking about here. in title, choosing distraction,but what i want to emphasize is that drivers don'talways choose distraction. sometimes distractionchooses us. i'm going to addressthree general topics. the first one is distractionand disengagement. i think they're reallyimportant and separate concepts. the second issue isattractions of distraction. what is it about the technology,about the social situation,
about the psychologicalsituation that draws people in. and attraction, in thesense of seducing the driver into these distractingsituations, like the one i found myself in. and then the lastone, the last point there is looking at attentionattending, and choosing what is it that leads us intodistracting activities. so distractionand disengagement. i think this is an importantdistinction, and one
that was somewhat alludedto in the definitions that mike reganpresented earlier. these are from the bookthat i helped co-edit with mike and christie young. the first one,driver distraction, is a diversion of attentionaway from activities critical for safe driving,towards a competing activity. driver inattention,as defined here, is diminishedattention to activities
that are criticalfor safe driving, in the absence of acompeting activity. and there's a number ofsources for this inattention. you can think potentially offatigue, alcohol, contributing to this. but i would add tothat, disengagement. and i'll go into alittle bit more detail what exactly i meanby disengagement. disengagement, i think,is a prevalent problem.
so the basic messagei want to make here is just becauseyou're not distracted, doesn't mean you're engaged. so simply not beingon the phone doesn't mean you're engagedin driving, doesn't mean you're attending to theworld in a systematic fashion. in fact, 72% ofdrivers reported lack of concentration, lack of beingengaged, on a recent trip. disengagement leads to thesame sort of glance patterns,
gaze concentration, that'stypical of missing hazards along the lines that donfisher just described. another way ofdescribing disengagement is stimulus independent thought. this is also describedin terms of the default network of the brain. this is the statethat we gravitate to when we're notseeking or responding to the informationin the environment.
not a situation that you wantto be in when you're driving, but does occur. and maybe some ofus, maybe all of us are familiar with thestimulus independent thought that occurs when you traversewell-traveled, familiar routes. you can't remember how and whatyou did for the last 10 minutes on a familiar route. you just did it automatically. you're sort of on auto pilot.
i believe that thisissue of disengagement is going to become moreprevalent in the near future, as cars literally havetheir own auto pilot, and drivers will nothave to be as engaged. so this routefamiliarity effect that occurs when driverstraverse a familiar route and then engage in this mindwandering and disengagement driving, i thinkthe same thing can occur with vehicle automation.
vehicle automation may inducethis route familiarity effect, because you don'thave to do anything with these increasinglyautonomous vehicles for certain parts of theroute, and then you do. so this effect is nicelydocumented by young and stance a number of yearsago, where they showed attentionalresources actually diminish when youare on auto pilot, when the car's driving itself.
and then when youhave to re-engage, your ability to doso is diminished because you have fewerattentional resources to devote to the driving task at thatpoint of re-engagement. so this issue of distractionand engagement i think are important issues, and definetwo dimensions of a space. so you have highengagement, low engagement, high distraction,low distraction. and i think they are, tosome degree, independent.
for example, in theupper right quadrant, you can be highly engaged. you can be looking at theroad, scanning frantically, and you can be distracted,reading text messages in between the frantic scans. so that's overload situation. that's a prevalentway of talking about driverdistraction-- exceeding the cognitive resourcesof the driver.
that's one type ofdistraction associated with high engagement. you can also havelow engagement. and in the left, mind wandering. the car's driving itself. you're not thinkingabout the road. your mind is justsort of zoning out. you're not distracted,but you're not engaged. the right-- lowerright quadrant is
the situation i found myself in. it was a lowworkload environment. i was disengaged from driving. i was just scrollingthrough, in a relaxed sort of fashion, the playlist. so that's low load distraction. so i think importantdistinctions that have implicationsfor how we deal with the attractionsof distraction.
this shows a variety oftypes of distraction. we focus on the technologybecause that's maybe one of the most quicklychanging types of distraction. but as we heard this morningfrom david strickland, distractions have been withus since the beginning, when somebody tried to eat asandwich while they drive. that's certainly the case now. but increasingly teen drivers--the picture in the lower right-- are being drawn intodistraction because many
of their social interactionsare mediated by technology. they spend more time connectingto people through the phone than they do face to face. the 24-hour society,indicated by the upper left, is pushing society. pushing all of us to beresponsive, in the moment, to demands. we get a textmessage, people expect us to respond right away,even if we're driving.
and in the upper right, there'sthat sort of physiological need to sustainourselves-- with thirst for beverages, hunger for food. i think the same sortof thirst and hunger exists for information. we need to occupy our minds. if we're not getting enoughstimulation from the roadway, we seek it from other sources. and those other sources arebecoming more prevalent.
apps for the car, eitherin the smartphone, or in built in systems,and increasingly, touch screens that allowcomplex menu systems, allowing interaction with thingslike pictures and movies while you drive. touch screens inthemselves, i think, are demanding enough,but then to be able to access all this richcontent, it seems problematic. this particular touch screeninduced distraction in me.
i was driving arental car in seattle. turned on the radio, whichhas a nice button here. very clear. it actually has buttons. hard buttons forthe volume up, down. but then i wanted tochange the station from the default, serious,advertisement station, to a local radiostation, which i think is under the audio menu.
but as i tried to punchat the touch screen, i quickly realized i might die. so i stopped. the future, i thinkwill bring more. on the left issergey brin talking about his vision of the glass,freeing the hands and eyes, so you can streaminformation to yourself all the time, in all situations. you can imagine that in the car.
this is an imagefrom wired magazine showing the varietyof information that could come into the car. and i think there's ahuge technological push for this information. money to be made. and an email froma friend of mine said it really seems to bea race between those trying to help the driver--detroit, sort of simply put.
and those trying to distractthe driver-- silicon valley. and i think this tensionbetween those distracting and those helping thedriver is an important one in terms of distraction. so this issue of engagement anddistraction-- the availability of technology does two things. and this is nicelydocumented in a paper by actually-- with surveydata from teen drivers, as to why they engagein distractions.
and they did a factor analysis,and identified two factors. one was task andsocial pressure. your phone buzzeswith a text message, you feel compelledto look at it. so that's one. and that can lead tothis overload situation. another force fordistraction is boredom. sitting there withnothing to do, your mind thirstingfor information.
that leads you toengage in distraction when there's nothingdemanding it. so this issue of attention,attending and choosing distractions, i think,is a tricky one. this shows a useful wayof looking at distraction in terms of cognitive resources. developed by chris wickens,multiple resource theory is a good guide to design. it identifiesplaces where, if you
have this competitionfor resources, you will have declinesin driving performance. reading whiledriving is dangerous. and that comes out ofthis sort of model. a more simplistic descriptionis in terms of visual, manual, and cognitive. the components ofthose adding up to-- in the case, a moderatedistraction with a hands free phone, a moresevere distraction
in terms of reaching, where youhave a much larger component of the visual and manual, thanyou do with a hands free phone. so these are ways ofthinking about distraction in terms of cognitiveresources sort of assume a situation like this. where you have competingdemands, distracting activities, passingthrough to the driver, leading to decrementsin driving perform. so if we could only describethe resource demands
that a task imposes, wecould redesign it or identify its risk, and all would be good. so this is a way of thinkingabout this, graphically shown, with the image of a brain,which charles spence told me that when you putan image of a brain, it increases the credibilityof your presentation substantially. so i've got toput that in there. but the basic idea here isyou have a distraction task
index in this verysimple visual manual. cognitive components, demandingcertain mental resources, leading to crash risk. if that was the stateof the situation, things would be easy. one of thecomplicating components is that the drivingdemands vary. it goes from the sort of thing,where there's a whole lot to do for a long time, tothis sort of situation, where
the driving demands are intense. and so it's really-- theconsequence of distraction really depends onthe roadway demands, and how they lineup with the demands from the in-vehicle device. so here you see a situationwhere the roadway demands are peaking just as the competingdemands are competing. so if this-- in thecontext of my drive back into madison, if the leadvehicle had braked suddenly
during my eyes offthe road period, i would have had avery different story to share with you. if you shift thecompeting demands away from that peak inthe roadway demands, you can see on theright the decline in driving performance is lower. so depending on howthat shifts, you get a greater or lesserdriving performance decrement.
the same thing withthe roadway demands. drivers can activelymodulate their speed, their positionrelative to other cars, and reduce the roadway demands. so drivers play anactive role in defining how and when theybecome distracted. and that has importantimplications for safety. and behind this idea of driversmodulating when and where they engage in distraction, is thisidea of perceived crash risk.
so drivers perceivethe situation, they see the hazards, andthen they schedule tasks, or they manage the interruptionsof the driving activity so that they minimize thatrisk they're being exposed to. so that perceivedcrash risk influences when they do thedistracting task. so the direction of causalityis a little bit different. and this is a niceillustration of how drivers shift theirdistracting activities.
this shows six secondsof time, of 100-car data. naturalistic data, withthe blue-purple line there being drivers whocrashed, or were involved in anear-crash situation. and on the vertical,you have the proportion of glances to the roadway. so what you see is thatfor those drivers who had a near-crashexperience or a crash, their proportion ofglances to the roadway
dipped precipitously duringthis precipitating effect. so they lookedaway from the road when the lead vehicle braked. the red line isthe baseline case. so what you see in thebaseline condition. in the baseline, people aren'tall looking at the roadway. they're notcompletely attentive. and what's sort offascinating is that top line-- those are people involved ina conversation on the phone.
so they're looking at theroad-- forward roadway-- more. what they're absorbing,how they're scanning, is a question that's still open. maybe they're not lookingaround attentively. maybe they're not processing. but i think anotherinterpretation of this is that when they're lookingat the forward roadway because they're on the phone,the phone is, in a sense, keeping them from doing otherstupid stuff as they drive.
if i had been on thephone coming into madison, i wouldn't have been scrollingthrough that playlist. so there may be aninteresting protective effect, in some sense, of providing amild distraction keeping you from a more severe distraction. so one of the other elements ofthis modulation of the roadway demands, i think, isreally importantly changing is how automationis going to influence that. so in other literature, inthe aviation literature,
for instance, they'vedescribed clumsy automation as automation thatis shown here, increasing the easypart of the trip, and making the difficultpart even more difficult. so it makes the boringstuff more boring, and the challengingstuff, more challenging. that's clumsy automation. coordinated automationwould do the opposite. it would sort ofsmooth out the trip,
making the high demandsituations easier to deal with, and makingthe rest of the drive somewhat more engaging,and less boring. how automationplays out, i think, is quite difficult to know. and not only doesautomation influence the workload over time, interms of clumsy or coordinated automation, but itchanges the feedback the driver gets qualitatively.
so that feedback influencesthat perceived risk, which influences when you'regoing to be engaging or not in a distraction task. so one of the thingsthat we assume is that when people engagein the distracting task, there's some crash riskthat they experience, and that influencestheir perceived risk. cognitive dissonance theorysuggests sort of the opposite. you engage in the task, andby engaging in the task, that
changes your perception of risk. so for example, with smokers,those people who smoke believe smoking to be lessrisky than it actually is. so engaging in the activitychanges your perception of risk. not because it's lessrisky, but because you have to resolve thatincompatibility of perception and reality. you're engaging in somethingrisky, but if it's risky,
you can't be engaging in it. to resolve thatcognitive dissonance, people reduce theirperception of risk. so in terms of driving,i think this interplay of feedback-- riskperception, risk adjustment, behavior adjustment-- playsout in a very complicated way. and fowler identifiessome safety traps that i believe are applicable toautomation and distraction that are really useful.
the first one iscontingency traps. and this is the sortof thing the don fisher talked about interms of novices. novice drivers don'tsee risks, so they don't adjust forthe consequences. they're not awarethat the risks exist, so they behave accordingly. so that's a contingency trap. novice drivers areparticularly prone to that,
and all of us will be novicedrivers when we drive with the next generation ofautomated vehicles , and new distracting technology. so this isn't justabout 17-year-olds. consequence traps. driver's see the risks, and yetthey choose to engage anyway. teen drivers seethe risk of texting, but they choose to text anyway. the amount of variance accountedfor by risk perception,
or the perception ofrisk in behavior to text is very small to nonexistent. so consequence trap, i thinkis an important element. people are seeing the risk,but choosing to become distracted anyway. and then the last one ismaybe the most insidious one. these are conditioning traps. situations where people dothe same behavior over time, even though it's risky.
and they get into the habitof repeating that behavior. so it's a conditioningit occurs over time. and because they don'texperience negative outcomes, they continue to do it,and it becomes a habit. so conditioning traps andhabit versus the intention to become distracted,i think are really important as we tryto intervene and create safer driving. habitual texting can occurwithout the driver's awareness,
and without their attention. we don't choose totext, we just do it. habit is much moreeffective in explaining the decision or thebehavior associated with texting andother distractions than other factors, includingsocial norms and attitudes-- the perception of risk. so this, i think, has somereally important implications for what we do as acommunity and a society
in terms of thisreal health problem. so in other areas,health improvement plans, the theory ofplanned behavior has been used to guidean intervention. the idea there is that youchange people's attitude. that changes their intentions,and that manifests itself in terms of theiractions or behaviors. i think what we need issomething to complement that. maybe a theory ofunplanned behavior,
that instead looks itchanging people's actions. and through changingtheir actions, we change their attitude, whichthen changes their intention. which is sort ofthe backwards logic, but i think it's thatbackwards logic that is really important tothink about in terms it's not aboutpeople knowing risk, and then using that riskto shift their intention. habits play a huge role.
so going back to thisissue of distraction being guided by habit. to some degree, perceivedcrash risk, and then also social pressures. with perceived crash risk shouldbe written in small letters, because i think its influenceis relatively small. this is a great picture. i was over in swedentalking to a similar group about a similar issueearlier this year.
and this was in thelobby of the hotel. it was sort of anabba groupie hotel. it had an abbatheme to the hotel. they played abba musicin the elevators. and this was abovethe registration desk. and as a visual attention taskhere, anybody notice something a little bit odd? this is sort of yourtypical drunken rock stars in the back of a plane,celebrating after a concert.
i found this picture surprising. anybody notice something odd? they've got their seat belts on. only in sweden woulddrunken rock stars put on their seat belts. but i think this illustratesa powerful influence of safety culture,either because of habit or social pressures,they have their seat belts on. i just focused onengagement of drivers
at the individual level. and that's primarilywhat we talk about. i think it's alsoimportant to think about engagement ata societal level. so how are we thinking aboutdriving at a societal level. this is a fascinating graphfrom the sartre survey of european countries. on the vertical axisis the fatalities per million inhabitants.
on the horizontal axisis how concerned are you about road safety. so this is a surveyto 600 people from all of thesecountries, asking them how concerned are youabout road safety. and you see kind of aninteresting distribution. in the upperquadrant there, those are high concern, low safety. those are people like myself.
i hardly drive. i'm a scary, dangerous driver. i'm really concerned about this. i'm in the upper quadrant. the left quadrant there, withpoland, the czech republic. low safety, low concern. maybe a teen driver. overconfident in their ability. the lower left there.
you have highsafety, low concern. complacent countries. i think thesecountries should be more concerned becausetechnology is changing quickly. safety five years ago, withthe new technology coming in, may not be the same assafety five years from now. so we need to be vigilant, ascountries like ireland, sweden, and israel are. and interestingly, i just cameback from ireland, israel,
and i visitedsweden earlier, all to talk to people inthis sort of conference, about distractionand driving safety. so these are vigilant countries. so the question is, where doesthe us stand in this space. anybody want to hazard a guess? i don't have data for thehorizontal, so who knows there. but on the vertical,this is where the us is. we're competing to be the bestin terms of higher fatality
rates, maybe. if that's the case,we're definitely winning. if we're trying to compete interms of safety, were losing. only poland, serbia, andgreece outperform us. or under perform us interms of fatality rates. are we concerned as a society? i'm kind of thinking we're not. i think we're overin the left quadrant, among the sort ofsocieties of teen drivers
who are low concernand low safety. so what can we do to moveto the lower right quadrant. i'm not quite sure. so just to conclude, i thinkthe absence of distraction is not engagement. i think we need to think ofthese two concepts separately. just because you're notengaging in distraction, doesn't mean you'reengaged in driving. and this is relevant atthe individual level,
and the societal level. the attractionsof distraction, i think seduce driversinto situations that they shouldn'tfind themselves, but do. and the process of attention,attending, and choosing is really complicated. and much of it is outsidethe conscious control of the driver. and so the issue ofindividual responsibility,
i think is really somewhatquestionable in this domain. interesting, i had atalk with, arguably, the guy who changed israel frombeing less safe than the us to twice as safe as the us. and he said that oneof the things that shifted in how they werethinking about driving safety, is that it was ashift from an individual responsibility,to the government needs to design a systemto protect drivers.
it was a health policy issue. so with that, i'd just liketo conclude with a quote from bruce springsteen,who's written eloquently about driving. and i think it sortof points to the need to be engaged in thinkingabout driving safety. "i just lay there inthe middle of the night, thinking about thewreck on the highway." and i think if more of us didthat, thought about safety
as a society andindividually we might be off. so thank you for your attention. thanks, john. do we have some questions? jim fowley, toyota collaborativesafety research center. more of a commentthan a question. but the thing that was missingin the earlier presentations and the laboratoryexperiments is basically, in the laboratoryexperiments, you always
assume that thedemand of the roadway is coinciding withthe distracting task. in the natural environment,that doesn't happen. and that's why you sometimeshave conflicting results between the twotypes of studies. yes, i agree, and i thinksome of the laboratory-based surrogate tasks for evaluatingthe safety consequences of some vehicle system designmight benefit from thinking about how to incorporate thiselement of allowing drivers
to modulate their attention tothe device and, by consequence, allow us to evaluate howwell a device supports that modulation of attention. so going back todan's talk earlier, where he showed sendersand the occlusion goggles that come down. interestingly, withsenders, much of his work used a situation where driverscould request information. they pushed a button,the goggles come up.
but with the occlusiontechnique that's used to evaluatedriver distraction, the goggles flash on and offevery one and a half seconds. there's no ability of thedriver to choose how much and how long to lookaway from the road effectively, as the occlusiontechnique simulates. so i think that's aninteresting challenge. how do we create an evaluationtechnique, surrogate driving tasks that build in someof that ability of drivers
to modulate the demands. it's something i'mthinking about, and we're actually working onsome of these issues for nhtsa. hey, john. dan fisher, umass amherst. a lot of the scenariosthat we have designed were ones that iactually designed. as a driver untilabout age 40, 45, was largely not payingattention to the road.
and i thought it was a taskwhich, like my stepson, required absolutelyno attention. and perhaps i'm as immaturenow as i was back then, but i've since learned thatactually driving can be fun. and it can be funsimply because, indeed, in the boston environmentin which i drive, there are a huge numberof latent threats. so you were talking aboutthe thirst and hunger for stimulation that goes on.
could it be the casethat if we develop programs which actuallytaught people-- not on the open highway, sincethere's relatively little going on-- but taught people aboutall the different things to which they should be payingattention that could cause actual crashes in theurban environment, or in congested environmentswherever that occurs-- it could become a source ofstimulation, rather than the alternative devicesthat indeed decrease safety.
i wondered as i waslistening to you whether that mightbe a possibility. i sort of alluded tothis in a very vague way in terms of thedesign of automation. vague because i have littleidea of exactly what to do, but i think it's aninteresting direction. how can we design automationto enhance rather than diminish engagement. and one illustrationmaybe along the lines
of what you're thinking there,don, if we reverse back here, is the image frothe wired article. there we go. so this has a bunchof advertisement that's distracting andcompeting for attention. but there's onehere that i thought was really fascinating interms of identifying hazards on the road, which might bekind of fun for people to have. this highlights this vehicleahead to the left as somebody
who has a dui citation. and so you could have,in a sense, this game of where should i be attendingto in the environment. and monitoring the person whowas drunk a couple weeks ago, and might be drunknow, might not be a bad thing to bedoing as you're driving, and maybe moreengaging then just having a blank windshieldwith none of this clutter. i had not thought about that asan enhancement to google glass.
yeah, i'm sure there's someprivacy issues, and all that. and maybe even highlight someof dan's poor task management people. on the road, youcould get highlights of who's a good daydual-tasker, and who's not. or more seriously, whois a novice driver. learned drivers, if youcould highlight them in the environment,that could be really useful in termsof giving them a bit more
leeway, a bit moreaccommodation. thank you, john. in england, they callthat learner plates. they put them onthe back of the car. good afternoon. i hope you had a good lunch. it was certainly delicious,and good to have a break. we have a greatprogram this afternoon. we have, again, just agreat slate of speakers.
and we're going to beginthe program with dr. linda angel, who's president oftouchstone evaluations, inc., and principal scientist there. she started the company uponretiring from general motors-- about 27 years in their humanfactors and safety operations. at gm, she was atechnical fellow in the safety center ofgm's engineering division, where she worked withinthe crash avoidance system development group.
she holds a phd in experimentalcognitive psychology from the university ofcolorado, and her work has addressed the effects ofadvanced technologies on driver behavior, the developmentof user interfaces, as well as crash analysis,engineering countermeasure development, andcountermeasure effectiveness. doctor angel has been recognizedwith a number of awards, including the 2008 a.r. lauersafety award from the human factors and ergonomics society.
doctor angel alsoholds an appointment as a research scientist for thevirginia tech transportation institute, where she's part ofthe center for vulnerable road users safety. and thanks also to aaam andstate farm for this very unique opportunity to be part of thispanel, and part of this event. can you hear me ok? i know my voice isa little bit soft. i'll let you settleback and get used
to having lunch just finished. but i want to move alittle bit rapidly, because i have a lot to cover. in the scientific effort tounderstand driver distraction over the last 15 or 20years, a large variety of research methodshave been applied to exploring some keyquestions, many of which have been already raised bythe speakers this morning. among those questionsare what activities
are drivers doing onthe road, and which of those lead to distraction. and what are thereprevalence during driving, and which of them elevate risk. specifically, crash risk. the focal point for thistalk is on a category of methods and datasources that i'll describe asnon-naturalistic, and what they've contributed tostudying distracted driving.
this term, non-naturalisticmethods, is a broad one, and encompasses a large numberof methods, some of which are listed on that slide inthe black box in the lower left corner of the screen. methods ranging fromgovernment crash databases, to fixed video observation,to insurance claim databases, and even experimentalmethods and survey methods. the term non-naturalisticis used in this talk also to strike a contrastwith naturalistic driving
methods, which is the topic ofthe next talk following mine, which will be givenby dr. tom dingus. whereas naturalisticmethods draw data from the driver in their ownvehicle, in their own car, being used in everyday driving,the methods that i'll focus on are not taken from thatsetting, but recorded from sources external tothe driver vehicle unit, such as from policeaccident reports of crashes, from cellphone records ofcalls, from video cameras posted
above intersections,or from claims filed with insurance companies. these are the types ofmethods that i'll focus on. and because there are so manymethods in this category, i can't do justice to eachof them in the 20 minutes that i have. so what i will doinstead is give you a sense for alarger picture that emerged during thechallenge that i was given,
which was to review the workin this category of methods, and identify gaps andopportunities for future work. let me back up here. in this area, there aremany methods, many results, but a picture thatstill lacks clarity. and this is true in spiteof the very deep interest that all of us who arestudying this topic have in resolving someof these questions. the field has been characterizedby a great deal of controversy
and debate over findings. many puzzles, some of whichhave already been talked about. but one of the things thatis most apparent to me as i've beenreviewing this area, is that there havebeen few efforts to integrate findingsacross methods. not only in this domain,but across methods in other categories, as well. but this is not terribly unusualin the history of science.
it tends to happen whena science is young, and there are newand emerging methods. it tends to happen whenissues are highly complex, and it's really hard to thinkof anything more complex than the humanbehaviors involved in driving within thecontext of a dynamic driving environment, and a changingtransportation system. and it tends to happen whenthe methods that are in use bear both similarities to oneanother, and also differences.
in the search that we sharefor answers in this domain, this nugget of wisdom fromchesterton from long ago has some relevance. he wrote, the smallman said, where does a wise man hide a leaf. and the other answered,in the forest. and it's easy for smallnuggets of new findings to become lost or hidden ina forest of different methods that may each be generatingslightly different answers
or information. indeed, when a variety ofdifferent methods are applied, it can be quite easy tomisunderstand methodological details that differbetween the approaches, but which are fundamentallyimportant for reaching appropriate conclusions. and when this happens,often key answers get lost. and not only get lost,but erroneous conclusions can be reached.
in complex fieldssuch as ours, it may take a special effortto ensure that information is extracted, rather thanhidden in the richness of the techniques thatwe bring to the table. it may take aspecial effort to see both the leaves in theforest, and the forest itself. the overall understandingof distraction risk that is being rendered bythe individual findings that are emerging from thestudies that were doing.
so what i'd like to dois explore this question, as i describe thestate of findings from naturalnon-naturalistic methods. namely, can abetter understanding of prevalence of riskof distracted driving be built by harnessingfindings across different types of methods anddata sources, both within the non-naturalisticdomain, and across other domains within the fieldof distracted driving, too.
an idea for us toconsider as we try to harness the powerof different methods and knowledge basesis to apply something called convergence science. this is a new idea thathas emerged recently. it's been identified as animportant scientific trend for fields that inherentlydraw upon multiple disciplines. for fields that are socomplex that advances lie beyond a single field,let alone an individual.
in 2011, mit conveneda blue ribbon panel on convergence science. it was led by phillipsharp, the nobel laureate. and he characterizedconvergence science as a broad rethinking ofhow scientific research can be conducted to capitalize ona range of knowledge bases. he pointed out thatthis involves more than simply bringing togetherconverging operations. it involves more than simplybringing together experts
from different disciplines. it involves an exchange ata deep level of mindsets, and a much more fundamentalintegration of approaches that were originally viewedas separate and distinct. an example of aconvergence field is the field ofcomputational biology, which emerged from the convergence ofcomputer science, engineering, physics, and molecularand cellular biologies. it was only when those werefused that really new concepts
and innovations began to emerge,and some new outcomes began to be available to us. to apply such an approachlike this fruitfully requires understandingstrengths and limitations of each method thatmight be brought together in a convergence approach. and i'd just liketo consider this in the context of three topics,as i very briefly give you a snapshot of workthat's been done
in the non-naturalisticdata and methods domain of drivingdistraction work. they areepidemiological analyses using non-naturalistic data,experimental studies, and then what's been done to pulltogether a larger picture. as i go throughthese, i will first present a very highlevel summary of findings from the area, and then giveyou a little bit of detail under that summary slide.
i'd like to turn firstto studies of prevalence. prevalence, of course,is a formal measure for quantifying driverswithin a population who are exposed to an activity ata particular point in time. so we might say, the prevalenceof us drivers eating a sandwich while driving at anygiven moment is 15%. that's just illustrative. we don't actually know what theprevalence of eating sandwiches while driving is.
but as we review studiesof prevalence of attention diverting activitiesin crashes based on non-naturalistic data,what we see at a summary level is that these have tendedto focus selectively on cellphone use. they've tended to incorporateself-reported data, and they've become outdated. this has left asignificant knowledge gap. there are bigsections of the forest
that we can't see becausethey have not been studied. and this has createdan opportunity for future research,and also for convergence with other types of methods thatmay yield data on these topics. very briefly, let me give yousome supporting details here. when i say that mostnon-naturalistic studies of activity prevalence havefocused on cell phone use, one good example isthe [? nopa ?] study that usdot does.
this is a study in which theydo observations at intersections when drivers are stopped. and they've estimated thatbetween 6 to 11% of drivers are using cellphones at anygiven moment of driving. flanigan and sayerat [? umtree ?] have been careful to pointout that this tells us about that moment of timewhen the drivers are observed at the intersection, andwhether or not that generalizes to during driving,we don't really know.
and this is oneinstance of where it would be useful to couplethis piece of information with converginginformation to help us have a sense of theprevalence of cellphone use across all ofdriving situations. when we look at what weknow about other types of secondary activitiesbesides cellphone activity, most of those studieshave been based on surveys usingself-reported data.
and this can bevery important data. data that provides informationabout driver beliefs and perceptions. and in particular, itcan give a lot of insight into willingness to engage, anddriver beliefs about conditions of engagement thatare important. but self-reported data hasvery important limitations, and if it's used to estimatefrequency of engagement or durations ofengagement, it is
subject to gross inaccuracies,and the literature tells us that. and i'll come back to thisissue, because it has sometimes been used inappropriately. so there would be benefitin updating the information in this area, butthere would also be benefit in usingwhat has been done. sorry, let me back up. there are prevalentstudies that have
been done on distractionrelated crashes. so i talked about prevalenceof activities during driving. now let's talk aboutprevalence of crashes that are distraction related. these studies have been doneusing non-naturalistic data-- that is, crash datafrom the databases. various kinds of databases,but they have not been updated for many years. what's interesting, however,is that in spite of their age,
they are quite consistentwith each other. and beyond that, theones that have been done are also quite consistentwith the naturalistic studies and i'd just like to verybriefly show you this. this table compares the rankorders of secondary activities that were coded prior to crashin two different crash database studies. they used very differentsources of data. they used very differentcomputational methods.
and i'm talkingabout the columns in white that are to the left. one study was done by wang,nibbling, and goodman in 1996, and the other by stutz andher colleagues in 2001. and if you only looked atthe numbers that came out, you'd think they founddifferent things. but if you look atthe rank orders that are shown in redcircles from one to six, there's a strikingconsistency between these
that shows that the mostprevalent source of distraction in 1996 and in 2001was diverting attention outside the vehicle. and all six of them are quitesimilar between both studies. furthermore, the technologyrelated sources of distraction added all together donot sum up to the extent of the top source. so they play a muchlower role in prevalence then the outside thevehicle distraction.
in the us, there havebeen no studies since 2006 to give us an updatedpicture of what's happening withinfotainment systems, or how prevalent is there use. what's happening with allof the other things that might be carriedinto the vehicle. however, in 2007, mcavoyin australia did a study, and her results were veryconsistent with these results. we can also look in pink atwhat naturalistic studies are
showing us. and again, we see astriking correspondence. and i think it'sfascinating to look at what is and isn'tprevalent during driving. but we have to rememberthat prevalence is not the same as risk. these activities maybe equally prevalent during baseline driving, wherecrashes aren't occurring. to look at crashrisk, we have to have
to look at relative risk. and to give you asnapshot of what has been found fromnon-naturalistic data sources, these studies alsohave focused very selectively on cellphoneuse, rather than on the full range of attentiondiverting activities. the results here have beenwidely varied and inconsistent. at the low end, therelative risk ratios have been as low asone or below one.
meaning not higherthan just driving. and at the high end, havebeen as high as four, or a little above four timesas risky as just driving. and i should add that not onlyhave they been inconsistent with each other,but they have been inconsistent with thenumbers of crashes recorded in the national crash databases. the comment that dan broughtup this morning about exposure versus crash risk, that wecan come back to in questions,
if we have time. the reasons forthe inconsistencies between the studies have beenidentified, and when corrected, these studies now appear toconverge on a similar finding-- that the risk of crashwhile talking on a phone during driving is not elevatedabove the risk of crash when not talking on aphone while driving. but these findingsare quite new, and would benefit fromfurther scientific review.
thus, we have a knowledgegap in this area, as well. we need study onactivities and crash risk outside of cellphoneuse, and it would be useful to havefurther clarity on the topic of cellphone use. i'm running a littlebit long here, so i'm going to skip over someof this lower level detail that i had expectedto share with you, and just touch briefly on theuse of experimental studies,
which dan mcgehee did areally nice job of covering this morning. they certainly are one of themethods in this domain that can contribute, but theirfindings are not well suited for estimatingprevalence or crash risk. they nonetheless can be usedfor some very important things in studying distraction. and one of the mostimportant, as dan pointed out, is that they allow usto study distraction
under repeatablecontrolled conditions. this allows us to get adeeper understanding of some of the effects that activitieshave on driving performance, and to develop an understandingof underlying mechanisms. for example, of cognitiveworkload, or visual workload. but because experimentaltesting involves controlling the conditions underwhich activities are performed, it removes many of thecritical elements that are important forassessing risk.
in particular,usually the driver does not have theability to choose when the activity is done. they do not have the experienceof actually experiencing risk, because they aresafe at all times. and these are criticalelements of determining, on the road, whetherto engage or not engage that play into what levels ofcrash we see on the roadway. so experiments playa critical role,
but they are not appropriatefor assessing prevalence in risk questions. when i reviewed theliterature in this domain, for any efforts to pull togethera holistic picture across risk studies, or prevalencestudies, i did not find any. there have been a few macroanalyses of cellphone studies in the literature,but no studies have extended afocus beyond this. it's as if each study is tossedinto the literature like bits
of knowledge, or bricksonto a pile of bricks. and it is this that makes itvery difficult to knit together the findings that arethere, which may well paint a consistent picture,but from which it's difficult to tell. so to close-- a couple ofresearch recommendations. first, that new studiesof prevalence in risk would be useful to performto fill the gaps that exist, and they should harness lessonslearned from prior work.
i haven't had thetime to explain what some of thoselessons are, but i think that new studies wouldbe very useful in this domain. and secondly, a big picture. formulating a big pictureof prevalence in risk needs to be constructed,and findings placed into a larger, moreintegrative framework that harnesses knowledgefrom multiple methods. as i mentioned, we may wellhave more understanding
than we think we haveat this point in time. and convergence science mightbe one option to consider here. so in closing, that possibilityis ours to consider. and maybe brick by brick, alarger conceptual structure can be built, making moreeffective use of the tools that we do have inour toolbox, not only in the non-naturalistic domain,but across the other domains, as well. time.
we have time for questions. i have just a quick question. in thinking aboutconvergence science and the differentapproaches you could use. presumably you coulduse a top down method, where experimental studies wereused to derive hypotheses that could be testedin the real world. or you could do it inthe reverse order, where you have naturalistic studiesthat then could be studied
in more depth in morecontrolled settings. could you comment on that. i think that's right. you could use both approaches. i think the key thing isto co-locate scientists from different traditions, andto have them working together towards a common goal. i think that's really the key. i think that whether it'stop down or bottom up
may make less difference thanhaving people approach it together fromdifferent perspectives. i think that wassort of my question. i just wanted to know, linda,how scientific convergence science is, and what sortof scientific techniques are used to come to aconclusion when you're bringing these fields together. is it just sort oflike meta analysis or is it an established science.
i don't think thatit's any different kind of scientific techniques. i think that like withcomputational biology, the molecular biologistsare bringing their tools in, along with thecellular biologists. the computer scientists arecoming in with their tools. but what's different isthey are working together. they're not in their silocompeting against each other. they are co-located in a center.
their incentives areto work toward a goal. the same goal, which is tosolve the same problems. and so they are nowlinking their knowledge. the computerscientist is working with the molecular biologist tounderstand the same question. and so what comes out ofthat are creative solutions. and it's notanymore, for example, in our field, somebody in onesilo arguing against somebody in another silo about a relativerisk ratio, and who's right.
it's suddenly that they'retogether-- pardon me-- working to meldtheir techniques, and to come up with somenew estimates, for example. are there any other questions? ok, thank you, linda. our next speaker todayis dr. tom dingus. doctor dingus is director ofthe virginia tech transportation institute, and is the newportnews shipbuilding professor of civil and environmentalengineering at virginia tech.
he is center director of thetier one connected vehicle infrastructureuniversity transportation center, which comprisesa consortium of virginia tech, vtti, theuniversity of virginia, and morgan state university. since 1996, dr.dingus has managed the operations andresearch at vtti. he has conducted transportationsafety and human factors research since 1984, includingthe safety and usability
of advanced in-vehicledevices, crash avoidance countermeasures, truck driverfatigue, and driver distraction and attention. he has pioneered studies ofnaturalistic driving, which involved instrumenting cars,trucks, and motorcycles with video cameras andsophisticated instrumentation designed to assess crashand near-crash causation, and to test a variety ofcrash countermeasures. doctor dingus was recentlynamed a white house champion
of change, and was selectedfor his exemplary leadership in developing or implementingtransportation technology solutions. he is also a fellow of the humanfactors and ergonomics society, for which he has receivedseveral awards, including the a. r. lauer award foroutstanding contributions to the field of safety. thanks, tom. so i'm going to talk aboutthe naturalistic piece,
and i, too, would like tothank aaam and state farm. this has been a great experienceworking with these panel members. we don't alwaysagree on everything, but we agree on a lot. and if we were to focuson what we agree on, i think we could makea lot of progress. i'd also like to thankall my co-authors. essentially i'm going to reporton four separate studies,
and each of these studieshad sometimes six or seven co-authors. i borrowed shamelessly fromthem, so i would recommend, if you read thepaper, to go back to the original pieces of work. they have a lot moredetail about somethings. in particular, i'd liketo recognize dr. charlie klauer, who's the lead authoron one of those studies. she's in the audience, so if youhave any really hard questions,
i'll call on charlie. so we've talked about alot of traditional methods. and back in theday when we really started thinking aboutnaturalistic driving, we really wanted tocomplete the puzzle. and that is, naturalisticdriving doesn't necessarily replace any of thesethings, but it really is a missing puzzlepiece in a lot of ways that i'll talk aboutin a few minutes.
so if you're trying to estimatecrash risk from police reports, for example, it's very hardfor an investigating officer, or even the ntsb to determineexactly what went on pre-crash. you have to make alot of assumptions, and things like that. surveys are dependenton subjective reporting. if you look at pleaseaccident reports, eyewitness testimony isreally relatively unreliable. you have driverimpairment often.
focused attention,so they may not have been looking atthe right direction. witness bias, and other things. and then simulation andtest tracks are great, but there's a lot of thingsthat go on in the real world that they don't account for. and that's whereyou get the control, as several of the authorshave talked about. so anyway, i thinkeverybody here
is now pretty familiar withwhat naturalistic driving is, so i'll move onrelatively quickly, but essentially we'reputting pretty highly capable instrumentationon people's own cars, trucks, motorcycles, now, andcollecting data for years. the studies i'm going totalk about collect data continuously, and so whatthat allows you to do is estimate exposureand estimate prevalence, which i'll talkabout in this talk.
as well as begin toput together crash risk estimates by using bothcrashes and near-crashes. so we try to make it asnaturalistic as possible. so these people have thesystems in their cars for up to three years. we just finishedthe data collection for the shrp 2naturalistic driving study. we had over 3,000drivers, and we collected over five milliontrips in that data collection.
and that's all continuous, soi'm tired of collecting data. very tired. but there's minimalintervention. we try really hard notto contact the subjects or do anything else sothat it's unobtrusive. we try to make it as unobtrusiveas possible in the vehicle and outside the vehicle,so that the drivers driving around the vehicle don'ttreat it any different than any other vehicle.
so i borrowed fromfour data sets. one was 100-car, which manyof you are familiar with. one was naturalisticteen driving study, which dr. klauer just hadpublished in the new england journal of medicine. one was a heavy vehicle drowsydriver field operational test, and the fourth was thenhtsa cellphone report. the lead authorwas dr. greg fitch, that just came out last year.
and so i took all thesedata from these-- and these are all published reports--across the three populations that are there, which areadult light vehicle drivers, adult heavy vehicledrivers, and then teens. the methodologywas very similar. there were a fewsubtle differences that you can get byreading the paper, but for the purposes of thissort of broad brush overview, i don't think they wereparticularly important.
so i took those and i estimatedboth odds ratios-- relative risk, if you will, using againcrash and near-crash events, and several of their studies. but also, they were alittle different in the fact that the smaller studies usedsafety critical events that were not true near-crashes, aswell-- and created odds ratios, because we couldhave the baseline. so each study hadits own baseline, which was at least5,000 samples.
so once a safety criticalevent was identified via some kind of trigger orjust by reviewing baseline data, they were validatedvia video review. in other words, somethinghappens, and then you can go in and you can watchit unfold, which is a real powerful aspectof naturalistic driving. i mean, you cango frame by frame and figure out exactly wherethe driver was looking when the brake lights cameon for a rear-end crash,
how long their eyeswere off the road, when they looked back, whatthe break reaction time was, what their evasive maneuver was,and whether it was successful or not. so it's a very powerfultool in that regard. it's also a powerfultool for prevalence, because you pull up asample from the baselines and you know exactlywhat they're doing. they're talking ontheir cell phone,
they're dialing ontheir cellphone, they're working on the centerstack, they're reading a map, or whatever the case may be. so prevalence waspretty easy to estimate based on what i just said. and you just essentiallylook at over 5,000 samples across thesestudies, essentially what percentage of timepeople are engaging in these various secondarytasks, or other distraction
or inattentive factors,depending on your definition. so this is sort of alittle bit of an eye chart, but this takes asample of tasks-- this is not every task thatwe observed in this study, but i picked the ones thati picked for a reason. so if you think about thisfrom a classic system safety example, you have thecriticality of a safety related event is typically the frequencywith which it occurs times the severity of the event.
and so we're fortunateto have both, because we have theodds ratio, which is the severity, if you will. and then the prevalencewhich, in fact, is the frequency,loosely coined. this one's a littleeasier to see, but essentiallysays the same thing. ok, so you see-- i got it.
so you see up herecellphone related tasks. and then for any oneof these studies, you see the percentage oftime for that particular study that we saw that behavior. and then the odds ratioor risk estimate for that. and then the ones in bold arestatistically significant, and the ones not in bold arenot statistically significant. so you take-- one thingyou've got to understand is some of these studieswere relatively small,
so you might havean odds ratio that's relatively high, like 2.6,but it's not significant because-- it could be becauseit's just not significant, it could be because ofthe size of the sample. so the ones i'vehighlighted here are the ones that we reallyneed to be concerned about. if you look at all the visualmanual associated things with a cellphone,they're all pretty bad, with a very few exceptions.
and again, theywere categorized. the 2013 cellphone study, i hadto combine categories, that's why these sort of categoriessit out by themselves because, again, it was arelatively small study. but you see-- and i don'tthink anybody in the room would agree that handheldelectronic devices are dangerous. and that's one of the importantthings that we all agree about. and if we couldagree, if you will,
to pass a law-- there wasa federal law, primary law banning handheldelectronic devices, we'd save a lot of lives. a whole lot of lives. one thing you'll noticeis the relative frequency of these things. some of them are veryhigh, relative to things that you don't see. so for example, one thingyou don't see on here
is interaction witha navigation system. people ask me, navigationsystems are distracting , right. well, there's twoaspects of that that i'll talk about in moredetail in a few minutes. but one of the aspectsof a navigation system in thisparticular case is it happened so rarely icouldn't even put it on a chart. so it's really not done thatoften, as opposed to using hand
electronic devices whichare, again, quite serious. one thing you'll noticeis across all studies, there's talking andlistening to handheld, and then where thedata were available, hands-fee portable werenot nearly as risky, and in some caseswere not risky at all, relative to the handheld pieces. and you'll noticethat-- and this has been a prettywidely reported result--
is that for heavy truck drivers,talking on a hands-free phone was actually aprotective effect. as was talking to a passenger,or talking on a cb radio. now why would that be? well, heavy trucks are oftenon the open road, open highway, often with not a lot of traffic. and i believe that that's howthey keep themselves alert. and in fact, if you do focusgroups with truck drivers, they'll tell you that.
and so we need to really bekind of careful about this. one thing that doesn't getnearly enough attention in my experience overthe last few years are heavy vehicledispatching devices. so this is a relativelypopular thing, and if you're a truck driverand you're out on a load, and you've got lessthan a truckload, or your dispatcher wants you topick up another load after you drop off the one you have,they communicate with you
via a dispatching device. and then you interact to sayyes, i want that load or not. and many of them are fairlycompelling in terms of time. and that is, ifyou want that load, you have to interactwith the system, and you have to make a response. well, that's an80,000 pound vehicle. that's got an odds ratio of 10. and it's a fairlyfrequent occurrence,
but you never seeanything about it. there's a sort ofinherent assumption, truck drivers areprofessionals, and so they ought to be able tohandle these devices. some of them arevery poorly designed, and they were killing people. other things you see here. reading a map. this is a paper map.
truck drivers do that. that's not a great idea, either. it's a little bit less--about half the frequency. truck drivers do alot in-vehicle office kinds of things. they do writing, forexample, while they're driving-- which you don't see inlight vehicles quite so much-- and other office related tasks. and those are bothpretty bad ideas.
for essentiallyall of the drivers, although it wasn'tsignificant in 100-car, reaching for an object,particularly a moving object, was really quite risky. if you think aboutit, you often-- and you see this particularlywith the truck drivers or teens, you'reoften leaning over to get something kindof in the back seat, or between the seats, oron the passenger seat,
and that's highly distracting. i thought thiswas a fascinating. so looking out the windowat some kind object. and the object couldbe lots of things. man, teens do that all time. they are easily distractedby external objects. it has a high odds ratio, butlook at the percentage of time that they're justsort of looking around at nothing in particular.
on the other hand, truckdrivers do it a whole lot, too, but for them, it'sa protective effect. this could be due to the roadwayenvironment, could be due to experience, allthose kind of things. and then even in 100-car, yousaw a significant odds ratio, although the frequencywas quite a bit less. another thing ithought was interesting is teens tend toreally get overloaded, where adults don't all the time.
and one of thosecases was eating. so you can see adults, theydo they do pretty well. at least truck drivers. truck drivers eat a lot, butthey do pretty well with it. in the 100-car study theydidn't do it nearly as often, and it was in thewrong direction, but it wasn't significant. but man, look at the teens. they have real troubleeating and driving.
and this is probably due tothe relative inexperience. they are just moreeasily overloaded. i threw a couple in here justbecause i hear it all time. we all talk about applying--just the other day i saw somebody applying makeup. and i've always beenimpressed by that, actually. that's the first thing. if you're putting on mascara andyou're not poking your eye out while you're driving,that's pretty impressive.
but the reason i wanted toput it up here is that we've seen it, but it really, relativeto these things in particular up here, it doesn'thappen very often at all. and the same is really truewith grooming, as well. although truck drivers,they don't do it very often, as you might expect, butthey're not very good at it. so that's somethingto keep in mind. but anyway, if you go backto the original reports, there's bigger lists, and theodds ratios are all presented.
and you can calculateyour own prevalences with the data that are there. but these are justthe ones that i picked it to makeseveral points. so, you know, thoseof you that know me, i'm a big eye off road guy--eye off road time kind of guy. we see people in allof these studies. they're looking away,like john lee apparently does when he's trying toget adele off of this car,
for four five seconds. that's the length of afootball field or longer. and we have lots of crasheswhere they just completely missed-- these are the kindof crashes where you really see somebody run intothe back of somebody else with no brake lights coming,and no the brake activation. it's serious stuff. and these kind of tasks includetyping, texting, reading, dialing a cellphone orreaching for a cellphone.
and also, if you're a heavytruck driver, also writing, as you see down here. that's a big deal. so if we can figure out how tokeep driver's eyes on the road by banning handheld cellphones,by requiring design standards for truck dispatching devices,we'll save a lot of lives. i found, as i said, the externaldistractions, particularly by teens, to bequite disturbing. the magnitudes of the oddsratios are quite different.
i think we are beginning to geta handle on why that might be. so for example, the oddsratios for the heavy trucks tend to be higher. but if you think about it,the crash rate per mile traveled for a heavy truckis about 15 times lower than a passenger car. and if you think about the waythe odds ratio, the numerator and denominator arecalculated, that would make up for alot of that difference.
so secondary tasks thatcompete for visual resourcess-- and again, i think weall agree on this-- are the most serious,and something that needs to be dealt with. and if we can deal with it,we'd make some real impact. the risk of cognitivedistraction based on what we've seen is somewhatless, and i believe it's there. i believe we see sort ofinklings of it being there. i think john lee and donfisher do great studies,
and i believe it's there. i think it's probablysignificantly lower risk than visual distraction. and if you think about itjust in terms of one result, and john lee and danmcgehee and others did a paper oncognitive distraction, and they found reactiontime was delayed by about 300 milliseconds. and that's relatively serious.
that's a 300 millisecond delay. but if your eyes are off theroad for four or five seconds, that's an orderof magnitude worse in terms of reaction time. because it's hard to reactto something if you're not looking at it. and so i think cognitivedistraction exists. i think it's something tostudy and be concerned about, but i don't thinkwe'll ever find
that it's anywhereclose to the risks associated withvisual distraction. if you talk abouthandheld cellphone use-- i talked about thisalready, but there's a good reason why statesare banning these things, and other countries arebanning these things. and they need to be banned. it's a very serious problem. and it's a serious problem notonly because the odds ratio
are relatively high, butalso because people use them all time. and if you add up theprevalence numbers-- the most recent cellphonestudy is a good example. so if you take thesenumbers and you add them up, and you take into accountthe different categories, you get the nhtsa number. it's about 10-- ibelieve it's about 10% of the time people are doingsomething with their cellphone.
one thing we don'tagree on, that i believe we spend a whole lot oftime not agreeing on, are hands freecellphone conversations versus passenger interactions. and i believe thereis some difference, but i don't think there'sa lot of difference. like i said, i believecognitive distraction exists. i believe it can happen whenyou're talking hands free on a cell phone.
i believe it can happenwith a passenger in a car. i believe it canhappen when you're listening to a book on tape. and i believe that it willhave a detrimental effect. but it's not that different. i mentioned navigationsystems and again, we almost didn't see this at all, whichi thought was interesting. and of course, when you'rein familiar territory like many of our subjectswere, and your commuting,
and you're doingother things, you don't need a navigation system. not to say peoplenever use them, and they shouldbe well designed, but it's not somethingthat we should probably focus a lot of our attention on. the other reason we shouldn'tfocus a lot of our attention on it is because,unlike cellphones or other in-vehicletasks like infotainment,
you've got tonavigate by some means to get to yourdestination if you don't know where you're going. and so what thatmeans is you're either going to use apaper map, or you're going to use a directionlist, or you're going to use anelectronic system. and many studiesgoing back to poor dan in the back of thattoronado for 14,000 miles,
would prove to you that,in fact, a well designed electronic system isthe best way to go. so i already talked aboutcognitive distraction. here's another littletheory that i like. this goes back toancient history. these two guys operationallydefined three levels of driving-- navigation,guidance, and control. so i think whencognition is overloaded-- and this will be agood debate over a beer
this evening-- whencognition is overloaded, i think you really losethe navigational layer. and that's thephenomenon that john and others we're talking about. i was on auto pilot, i don'treally remember how i got here, i don't remember anypart of my commute. i missed my turn, even. but then, the real questionis, does losing the navigation layer reallyincrease crash risk.
and i think we'rebeginning to see that, not as muchas you might think. so i talked about this. a four second glance isnot unusual, unfortunately. and it goes by fast. sometimes too fast whenyou're about to hit the car front of you. but we see people adapt innaturalistic driving studies. so adults, relativelyprudent drivers,
will increase their headway. they'll slow theirspeed, and then they'll engage in a secondary task,and do everything right. but they often underestimatehow fast time and space changes. how fast their buffer goesaway, particularly when their eyes are off the road. so a naturalisticdriving studies and work are coming onlineinternationally. there are studies going onin asia and europe, and soon
to be australia and canada. so we're going toget a lot more data. we're going to get a lotmore powerful results in the very near future. you should come to ournaturalistic driving symposium, because you'll actually be ableto interact with the shrp 2 data during the workshop. and shrp 2 data is going to bean extremely powerful database. it's good not onlyfor a prevalence.
it's good for risk assessment. and again, we now havealmost 1,000 crashes. so we're beginning to get reallya significant number of crashes that we can analyze. and we should continueto ask these questions. we should continue to studythese things as these data come online, and more and moredata becomes available. i see you might have afew questions here, tom. i'm not surprised.
didn't you get thememo about no panel members were allowedto ask questions? is this on? go ahead. this is not so much a commenton your particular paper, but on thisconference in general. you're about the onlyone who has actually talked about solutionsin ways of putting technological changes in eitherthe devices or the vehicles
to deal with thedistracted driving. and i'm quitedisappointed that nobody has talked about how doyou actually put something in a vehicle, or in itsinfotainment system, or whatever, sothat people will not be unnecessarily distracted. i have a new ford that hasabsolutely the most utterly distracting system. it's impossible to use.
it's extremely distractingwhen you try to use it. there's also thingslike, 60 years ago, general motors did a studythat showed the small driving changes in your steering wheelcan be measured, and used to determine whether a person'spaying attention to the driving task. and there's certainlylots of new work on things likehead position, eye position, that sort of stuff.
and i don'tunderstand why nobody has said a word aboutany of this stuff. i think that's a good comment. in defense of my carcompany colleagues, i think they worryabout this a lot. there are differencesbetween car companies. some worry about it alot more than others. for example, at vtti, wework with them all time trying to developbetter interfaces.
and actually, we work withtruck dispatching companies to come up withbetter interfaces. so it's possible. you've got to pay someattention and really work at it. so i think there'sa lot of work that goes on that's not necessarilypublished right away, it just shows up in designs. but, yeah, you're right. i could give you ahall of shame like i
do to my class aboutinterfaces, and i think you saw onein an earlier paper. i think it was john lee's. your car, man. what's up with that? why would you buya car like that? but i think that'sa good comment. i believe-- john lee and i weretalking about this at lunch. i believe a really goodnext step would be for us--
and we owe linda and others, andsort of our own shamelessness-- but i think that maybe the nextstep would be for the panel members to get together, and puttogether a paper of what we all believe in, and what thecountermeasures could be. and i believe we could makea big dent in this problem. like an 80% kind ofdent in this problem. not 100%, so we should focuson the 80% we agree on, not the 20% we don't. but i think that wouldbe a really good idea.
what do you think, sue? i'll defer to the members. don fisher, umass amherst. tom, i really enjoyedthe talk, and i want to follow up on theother individual's question. the dispatcher devices. what is it about the interfacethat's creating the problem? i assume there are a numberof especially long glances. is it obvious whereit's going wrong?
you said you're doingsome research on that. i'm just curious. we work with interfaces, too. it's a more detailedquestion than maybe we should answer herein this forum. but still, i'm interested inwhat it is about the dispatcher device in particularthat's created a problem. well, i thinkthere's two aspects. they're trying to convey afair amount of information.
and so many of them choose todo it via paragraphs of text, that you actuallyscroll between screens. and nobody in thisroom would believe that that's a good idea. but the other is sort oflike reaching for a moving object, which we found in100-car was very dangerous. it's sort of thecompelling-- you're compelled to dealwith it right now, or i'm going to losethat load, unless i
respond that i want it. so regardless of thetraffic circumstances, i have to read thatparagraph of text, and say yes or no,i want to load. otherwise one of mytruck driving colleagues might get it, and i might notmake as much money on this run. so those there's two things. i want to be the first tosecond your suggestion to sue. i'll leave it toothers to add to it.
we have a questionfrom the webcast. phil cordoro. the point about truckdrivers being less impacted by distraction isreally interesting. and i was wondering aboutwhether we have any reason to believe that such aneffect could generalize towards a population ofmore experienced drivers. do we have any reason tobelieve that some slight amount of distraction couldactually be helpful for very
experienced driversbecause it prevents them from going on auto-pilot? yeah, i think in long haultrucking, i think that's true. i think there arecircumstances where they can engage in something--like talking on a cb radio, for example-- andthey're fine doing it. and it's probably somerelatively small level of distraction thatyou're sacrificing, but there's a muchbigger level of alertness
benefit that sortof counteracts it. the other thing that ihear all the time that drives me a little crazy is thattruck drivers are professional. cab drivers are professional. they can operate. police officers, nooffense, are professional. they can operate that laptopwhile driving at high speeds, and do these things. they can't.
i mean, vision isnot omnipresent. if you're notlooking at the road, you're not processing anything,and i don't care who you. and i have the greatestrespect for truck drivers. 90% of them arehighly professional. they're very safe drivers. but there are somepercentage-- in fact, this was a study that richhanowski, rebecca olson, and others from vtti did.
they showed that 10% of thedrivers really create about 50% of the crash risk inthe trucking industry. and that's one plug fordriver monitoring systems, because if you canidentify what that 10% is, you can make a huge differencein the safety of your fleet by remediatingthem, or encouraging them to get another job,or doing things like that. but the other aspect is,they'll tell you, well, you've got to go throughtraining to get a cdl.
but these guys aren'tfighter pilots. i mean, they're notselected because they're the top 1% in terms ofreaction time and vision, and all these things. and they don't gothrough tens of thousands of hours of training. they go through a six weekcourse in how to drive a truck. and so they're not-- i thinkthat while we should respect what they do for a living, andmany of them are safe at it,
we need to be quite cautiousabout statements like that, because you canwreak a lot of havoc with an 80,000 pound vehiclewhen you're not looking at the road for the lengthof a football field. that was a long answer. we have time forone more question. sue chrysler,university of iowa. this is more of a comment. and can i add to your answer?
sure. i think we're overlookingthe effect of roadway type. and you mentioned it, butdespite the higher speeds, limited access freeways havea fraction of the crashes. there just aren't theopportunities for conflict. and one of thespeakers this morning touched on that interms of training. we always have beenjoking for 10 years, are going to start trainingpeople how to text and drive.
or at least when. but i think that when you justlook at crash data, distracted or not, freeways aresafer than local roads. but the perception of driversis that you're driving fast, so they're more dangerous. and so you'll see people textingin a shopping center parking lot, which is probably themost dangerous place to text. but they have this idea thatit's slow here some, so i'm ok. that's a good point.
i agree with you 100%. thank you, tom. we've heard a lot about teens. a lot of people have alluded tothem and their special risks. our next speaker,dr. dennis durbin, will talk aboutspecial considerations in distracteddriving with teens. and he did a combination paperwith other panel members. doctor durbin-- well, isee i have an old bio.
are you still this? anyway he was, as of today, apediatric emergency physician and clinicalepidemiologist, as well as a professor of pediatricsand epidemiology at the children'shospital of philadelphia, which you've heardof today as chop, and the universityof pennsylvania school of medicine. an internationallyrecognized leader
in pediatric traumaresearch, dr. durbin is the scientific director ofthe center for injury research and prevention at chop. his research interest inthe area pediatric injury epidemiology, withparticular emphasis on the prevention of motorvehicle occupant injuries to children, and the preventionof teen driver crashes. he served as the co-principalinvestigator of partners for child passenger safety.
chris mullin discussed thatthis morning-- a joint effort between chop and statefarm insurance companies-- that resulted in theworld's largest child focused motor vehiclecrash surveillance system. his research has beenpublished in a number of scientific journals andconference proceedings, including a long listwhich i won't go into. he's also been elected to thesociety for pediatric research. it's great to be here.
i first want tothank my co-authors. one of the reasons i agreedto participate in this panel was because i wanted to workwith some of the people who were on the panel withwhom i don't often get a chance to work. and maybe reflecting thatnotion of convergence science that linda spoke about. i've had a great timeworking with dan and don, who you've both met,and ann mccarte, who
will be speaking momentarily. learned a tremendousamount from them. so thank you very much. and my presentation reallyreflects the combined work of all of us in this area. and i'll add my thanks tostate farm and to aaam. and maybe just to add to whatchris had mentioned earlier, i am one of the folks atthe children's hospital of philadelphia who's hadthe very distinct honor
and pleasure of collaboratingwith state farm for over 17 years in auto safety. i know the company very well. i appreciate its commitment. i'm not at allsurprised that they decided to committo this endeavor, and to actually adopt aresearch oriented focus to it. to really understand whatthe evidence tells us before deciding whatwe should do about it.
and so i really appreciateit a great deal. so we have heard alittle bit about teens. and the panel did think it wasimportant to summarize what we know about teens anddistracted driving-- focus a little bit of aheadlight on them, so to speak. and i think manyof you in the room may know that automobilecrashes by far and away are the leading causeof death to teens. so it's a very importantpublic health, as well as
traffic safety, issue whenit comes to teenagers. crashes kill 1 in 4 14 to19-year-olds in this country. so it far exceeds manyother causes combined. so it's worth ourtime and effort to focus on teens and crashes. and while there have been anumber of great reviews written globally on the topic ofteens and driving safety over the last several years,some of which by my co-authors, none of those shineda particular light
on the issue of distractionwhen it came to teens. and so that was ourgoal, was to summarize the currentliterature of what do we know about teensand distraction. and we wanted to focuson five issues that are sort of unique to teens. maybe the first one'snot entirely unique, but i'll just gothrough one by one and summarise what does thecurrent literature tell us
about these five issues. and at the end,our recommendations for where future researchshould be focused. so the wide prevalenceof mobile device use. some of us on the panelhave been talking about one of the challenges, ithink, in traffic safety when it comes to mobile devicesbeing a source of distraction is they are so widely prevalent. they're really omnipresent.
and i was thinking whenit comes to seat belts, we think about the policyand individually oriented interventions thathave been designed to get people towear seat belts. well, that's a behavior youonly do when you're in the car, separate from ridingon an airplane. so in theory, youopen the car door, you should be thinking aboutputting on you seat belt, but it's the only timeyou think about that.
and so certain approaches to getpeople to change their behavior can really be focusedon environment. take drinking anddriving, for example. well, we drink bothin the car, sometimes, but we also drink a lotoutside the vehicle. so drinking is not abehavior unique to the car, but from a societal standpoint,we've sort of defined the boundaries of whereit's appropriate to drink and where it's notappropriate drink.
and in the car iscertainly one place where it's notappropriate to drink. and people don't walk aroundwith open bottles of liquor in their jacket pocket,like i have this 24/7. so i think one of the challengesin dealing with mobile devices as sources of distractionis they're always with us. and as a society,we haven't even defined yet the sociallyacceptable times when it's ok to usethem and when it's not.
we use them all the time,whether it's really polite and that is certainlytrue for teens, who have grown up with thesedevices sort of permanently affixed to them. so this really does providea very unique challenge. and then perhaps as an exampleof convergence science, getting an estimateof how commonly teens might text or talk on acellphone while driving, you'll get differentanswers depending
upon the methodologies used. so in largepopulation-based surveys, if you ask for sort ofa cumulative prevalence of over the past30 days, how often have you done thisat least once. you'll get a conservativeestimate of a quarter of 16 to 18-year-olds saying thatthey've texted while driving, and over half talking onthe phone while driving. whereas using thenaturalistic driving methods
that tom just mentioned, andthe naturalistic teen driving study just recently publishedthis in the new england journal, about 5% of therandomly sampled video clips involved a teen interacting witha mobile communications device. they're both correct. they tell us two complementarypieces of information. the naturalistic gives usthat sort of instantaneous, at any given moment intime, any given trip, what's the likelihood thatthey'll be interacting.
you could erroneously concludethat it's not that common. but if you ask someone overa longer period of time, how often do you engagein this behavior, you begin to get some insight. this is not a one timething that people do. they repeatedly do thisover and over again. there's some mixed evidenceon how commonly-- how the frequency of teen mobiledevice use compares to adults. early studies andself-reported surveys
would suggest thatin general, teens report texting more commonlywhile driving than older adults, but talking onthe phone less commonly than older adults. that recent naturalisticdriving study showed actually fairlyequivalent frequencies between teens and adults atboth overall secondary task engagement, as well asuse of the cellphone. so i think that'sa rapidly evolving,
kind of a moving target. that naturalistic teen drivingstudy that dr. klauer published in the new englandjournal of medicine just a few monthsago also provided this interesting finding,which is over the first 18 months of independentlicensure, how did the engagement and secondarytasks by newly licensed teens change? and what they found wasthat it generally increased.
the more experience teensgot driving independently, the more likely they were toengage in secondary tasks. and that's adisturbing finding that deserves a littlebit more attention. and then this is--again, my thanks to dr. klauer forallowing me to use the slides from apresentation she gave to us recently-- asummary of the data, much of what youjust saw from tom.
the numbers areslightly different. i'm assuming just somedifferent analytic techniques were used here. but the message isas tom described it, which is certainly thosetasks that involve both visual and manual distraction areassociated with very high risks of crash/ near-crash events. so you can see texting,dialing, reaching for the phone, reaching for other objects.
and you'll also noticethat there are activities that seem to beproblems for teens then there are forexperienced adult drivers. so again, this is someinsight into the fact that as relativelynovice drivers, teens haven't reallyfigured out how to engage in someof these things without it imparting risk. so the second issue ofteens are just novices.
don spoke earlier aboutthis concept of teens are clueless, not careless. i actually don't likethat characterization, as the father of teenagers. but this is what inexperiencelooks like quantitatively. these are graphs again,over a decade old. we've known thisfor a long time. that teens who are newlylicensed to drive independently without an adult supervision areat their highest lifetime risk
of being in a fatal crash. and that risk declines overabout the first six months or 1,000 miles ofaccumulated practice. these types of studies havebeen recently replicated. this has not changed over time. i put up the order datajust to make the point that we've known thisfor a very long time. and as don pointedout earlier, the teens are both less skilledstrategically,
so they tend to engagein more risky behaviors than experienced adults. in part, becausethey're less likely to assess various activities asrisky than experienced adult drivers. and that's just a byproductof their inexperience. teens are also lessskilled tactically, and don went into tothis in some detail. a lot of this has come fromhis work, where eye glance
behaviors are our worsefor inexperienced novice drivers compared toexperienced adults. they'll take longer looksaway from the roadway, take more frequent looks awayfrom the forward roadway. the good news is, asdon's group has shown, is that you can improvesome of this behavior with pc-based training. and the evidence for thatis fairly robust to date, but i'll talk in amoment about where
we think it shouldgo moving forward. now the issue of peers. this, of course, is one ofthe defining characteristics of adolescence. adolescents spend moretime with their peers than any other age group does. they spend more time with theirfriends and peers than we do, or than younger children do. peers are a verypowerful influencer
of adolescent behavior. i put up these data to suggestthat we shouldn't start thinking about therole of peers and teens in the vehicle wheneverybody turns 16, because it's aproblem that begins to rear its ugly headat much younger ages. these are data from a studyflaura winston and our group did several years ago. so along the x-axis are agesfrom 8 to 17 years of children
killed as passengersin automobile crashes. these are data from fars. and what's depicted isthe age distribution of the driver whowas behind the wheel when that child was killed. and so what you cansee is certainly for the vast majority ofkids 8 to 11 years old, the driver wasover 25 years old. it was most commonly a parent.
but beginning atage 12, and then increasing exponentiallythrough age 17, the proportion of drivers that are youngincreases pretty substantially. not only do the total numbers ofdeaths increase exponentially, but the characteristicsof the driver also change. so i think we need to bethinking about who our 11, 12, and 13-year-oldsare driving with, because there's a substantialamount of risk involved there. so it's not just about16 to 18-year-olds.
and again, we've knownfor over a decade that adding peer passengerswhen there's a 16 or 17-year-old behind the wheel isexponentially dangerous, increasing the riskof fatal crashes. and again, thesedata have recently been replicated, both with somenaturalistic methods, as well as more population-basedsources of crash data. so the data on peers andthe risk with fatal crashes is pretty robust andcompelling, and has
been known for quite some time. the story's a littlebit more complicated when it comes tonon-fatal crashes. these are data from allisoncurry in our group, who utilized the national motorvehicle crash causation survey, a data sourcei think dan mentioned in passing this morning. nhtsa did a study that detailedinvestigations of over 5,000 crashes, looking at the causalchain that led up to the crash.
this was a data source createdto help with the development and evaluation of crashavoidance technologies. we looked at all thecrashes in [? nimbix ?] that involved a teendriver, and looked at the association of whatwere the things that happened immediately proximalto the crash when teens were driving eitheralone or with their peers. and the first twobars on the left show in-car distractionswhen there was a male driver.
and what you see isabout one quarter of the crashes involved anin-car distraction, according to the investigationthat was done for boys. and that didn't change whetherthey had friends in the car so it's fairlycommon, but it didn't seem to be influenced by whethera peer was in the car or not. contrast that to the girls, whoare next to them, where only 5% of crashes involvingfemale teen drivers involved an in-vehicledistraction immediately
preceding the crash. but if that girl hadpeers in the car, the risk went up four times. so girls seem tobe quite influenced by the presence ofpeers when it came to in-vehicle distractions. the third cluster ofbars is aggressive acts-- speeding, tailgating,these kinds of things, breaking the law, runningstop signs-- counted
for 14% of crashes whenmales were driving alone, and over 36% when with peers. so here's evidence that oneof the ways in which peers influence crasheswith male drivers is perhaps encouraging the sortof aggressive driving behavior. and girls are notparticularly prone to this. they don't engage inaggressive driving behaviors as frequently as boys, interms of it being a preceding cause of their crashes,and it doesn't particularly
change when theirfriends are in the car. so again, it's complicated. there are a varietyof circumstances that one must look atin order to understand the influence of peers. and i think one of the thingsthat i found particularly interesting about doing thisliterature review was dustin albert and larry steinbergat temple university in philadelphia have put fortha model for decision making
and the influence of peerson adolescent decision making that i think isreally thought provoking. and it is the following. whenever any of usmake a decision, our decisions are theresult of a balance between cognitive processesand emotional or affective processes. so the cognitive is likewhat's the right thing for me to do here.
and the emotionalis how am i'm going to feel once imake this decision. so some of us haddessert with lunch today, and some of us didn't. so those who haddessert with lunch might have thoughtwell, i probably shouldn't have this dessert. i'm trying to lose weight. but i'm going to feelreally good if i eat this.
it's going to tastereally good, so i'm going to go ahead and do it. so every little decisionwe make is often influenced by these two things. and adolescentsare no different. but what it is differentabout adolescents is that duringadolescence, the pathways involved in that affectivecomponent of regulating decision making get sortof a pulse of development,
while the cognitivepathways, which are generally in the prefrontalcortex, just go along a sort of lineardevelopment over a much longer period of time. people now on commonlyknow that brains don't stop developinguntil 25, or so. it's that prefrontalcortex, that cognitive piece that continues to developthroughout young adulthood. so teens generally overvaluethe emotional component
of decision making. and what is interesting is thatthere's some early evidence to suggest that in thepresence of peers, both whether the peer isphysically present as well as maybe justan influence remotely, teens will overvalue shortterm rewards of their decision making, instead oflonger term consequences. so again, i don'tdoubt that there's a teen in the unitedstates of america
that doesn't know thatwearing a seat belt is the right thing to do. it's not a knowledge gap problemthat explains why 30% of teens don't wear their belts. it's because at the momentthey get in their car, there's a differentthought process that's going on that'sovervaluing short term rewards of deciding not towear the belt in that moment, than the longer termconsequences of doing so.
and this is important. we need to understand thisbetter to know well, then how do you change teenbehavior despite that. the other obviouscharacteristic of most teens when they begin the drivingprocesses is they're living in families. they live at home. they have parents that caninfluence their behavior, and that serve as very powerfulrole models of their driving
behavior. and these are data thatthere in the paper that come from thenielsen company, that show that the use ofsmart-phones is increasing. that's not just a teen thing. most of us in the room probablyhave them, and our use of them has increased exponentiallyover the last few years. so there's a lotof modeling that's being done by parents whenthey're behind the wheel,
perhaps with theirteens in the car. we know that modeling isa very powerful influencer of teen behavior. so if we want tochange teen behavior, the parent is a powerful ally. but they've got to be doingthe right things themselves, or they'll undermine their role. and lots of groups have triedto categorize parenting, and the role of parenting,in adolescent behavior.
one common way ofclassifying parent behavior is along these two domains ofincreasing rules and control, and increasing so-calledwarmth or support. and so in the lowerleft hand corner you have parentswithout a lot of rules, and without a lot ofwarmth and support. they're commonly referredto as uninvolved parents. so these are folks you mighthear say oh, kids will be kids. just go do it.
you'll learn from your mistakes. they're kind of checked out. permissive parents, in the upperleft hand corner, still not a lot of rules, but theyreally love their kids. and so the message that alot of these kids get is, i trust you'll dothe right thing. you decide. i know you'll be fine. in the lower righthand corner, any of us
who were kids inthe 1950s and '60s were probably raisedby this type of family. the authoritarian parentwith lots of rules, but not a lot ofwarmth and support, or explanation for those rules. so this is the,you'll do as i say, and you'll do itbecause i said so. and then the upper right handis the authoritative parent who has rules, butcan impart those rules
in the context of warmth,support, explanation. so they can identifycertain things that are non-negotiable becausethey're safety relevant, but also do understand theconcept of giving teens some responsibilityand growing into that. and some work that ourgroup did several years ago in a population-basedrepresentative survey of high schoolstudents in the us found that teens who describetheir parents as authoritative
were less likely to reportcrashes in the past year, drinking and driving, or usinga cellphone while driving, as compared to teens who describedtheir parents as uninvolved. and in more directobservation studies, several folks havementioned drive cam and other in-vehiclevideo-based monitoring systems. this was a study that brucesimons-morton published. the figure at the top shows thenumber of high g-force events as a function ofwhether the teen had
a device in the car thatwas giving them just above a light indicatorthat an event had happened. that's the top line. versus the light plusfeedback to the parents, where the teen knewthat their parent would get the video clip, have achance to review it with them. again, dan has donea lot of this work, indicating that to theextent that parents remain engaged and actuallytake advantage of this,
they can really influencetheir teen's behavior. the number of those eventsdropped significantly over time. whereas if the team justgets immediate feedback but no real consequenceto their behavior, the device doesn't reallyhave an effect at all. there have to beconsequences, is the message. and there's someemerging literature that i find interesting as botha parent and as a pediatrician.
we often think about theparent-adolescent relationship as a one-way street,where parents just tell teens what to do. they make rules and teensjust sort of passively receive that orfight against it. and there's some reallygood information coming out that that, of course,is very bi-directional. that a parent'sparenting style is often influenced by thekind that teen they
have, and the way theteen interacts with them. so this is very mucha two-way street. and we don't, ithink, really know yet how to best influence that sothat the parents and teens can be having a productivediscussion that results in safer behavior onthe part of teens. the last issue thispaper focused on was teen-specific laws. and again, ann is going to gointo some detail about laws
in general, so i'll justtouch on this briefly here. 47 states have laws that forbidtexting while driving by teens, but 37 states have what arecalled all cellphone use bans for young drivers. and of course, evenamong those states that forbid the use of allcellphones by young drivers, the provisions are different. the secondary versusprimary enforcement is different fromstate to state.
and there's little informationon the effectiveness of these studies. a nice natural experimentoccurred several years ago in north carolina, whichpassed a cellphone ban in 2007. and researchers at theuniversity of north carolina did direct observationsat entry and exit points at high schools in both northcarolina and south carolina, which did not have a ban. and before the ban, you seethat about 11 to 15% of drivers
were noted to beusing a cellphone. and post-ban, thenumber didn't really change in north carolina. it dropped a littlebit in south carolina. i'm not sure that's actuallystatistically significant. so this pretty muchsummarizes most of what we know aboutempiric evidence on cellphone bansfor teen drivers. and john did mention earlierwith the google glass
potentially beingable to point out who the intermediatedriver is on the road. and sue correctlypointed out something i was going to say, too, whichis we have a low tech solution to that that exists now. many of you knowthat in australia and some europeancountries, there are indicators placed in thevehicle or on the license plate to indicate that the driver isa provisional or learner driver.
until new jerseypassed this provision in its graduated driverlicensing law in 2010, no such law existed in the us. and the concepthere, of course, is that we have all thesegraduated driver licensing laws, but their effectiveness islimited by a police officer's ability to actually enforcethe law on the street. and so it's commonin many states that a 17-year-olddriver is held
to certain restrictionsin their driving, but an 18-year-old may not be. and the average policeofficer can't distinguish a 17 from an 18-year-old whiledriving down the road. so new jersey passedthis decal provision. the decal is a little redsquare in the upper left hand corner of that license plate. in theory, it's designedto increase police ability to enforce the law.
because now we knowthat the person behind the wheel in this vehicleis subject to different driving laws than a fullylicensed driver. and so this is sort ofthe theoretical model for how these decals would work. they would work through apathway of increasing law enforcement. that would influenceteen behavior. if you know you're morelikely to get pulled over
for a violation, you'reless likely to commit that violation. and all of that wouldreduce teen driver crashes. so what do we knowof the evidence? well, the insuranceinstitute for highway safety did the firststudies of this law, and demonstratedthat while there was an increase in policecitations for graduated driver licensing violations, whileself-reported behaviors
of teens didn't really changebefore and after the decal. and then dr. curryin our group looked at state statewide sources ofboth crash and licensing data. and this is a graph of the rateof gdl citations per 10,000 intermediate driversin new jersey. and the two verticallines-- the law was passed a year before it wenton the books, which is common. and so when the lawwas passed, there were some publicity aboutthe passage of the law.
and then a year later, the lawactually went on the books. and for what yousee here is-- these are smoothed forseasonal variation. and you can see there isseasonal variation to gdl citations, at least inthe state of new jersey, and it tends to peak in thesummertime months, likely when driving exposurepeaks by adolescents. and in the first yearafter passage of the law, accounting forseasonal variation,
we found there was a 14%increase in gdl citations, and a commitment 9%decrease in young driver crashes involvingintermediate drivers. so the initial evidenceof new jersey's law would suggest that thismay be an effective way to enhance enforcementof gdl provisions. and there are a number ofother states considering this. so based on thatreview the literature, these are our-- my co-authorsand i-- our recommendations
for future researchfocus on teens. first of all, weneed better studies of the association ofdistraction and crash risk for teensrelative to adults. i think the early naturalisticdriving studies give us a good indication of whatthat probably looks like. and as tom mentioned,the shrp 2 data should give us a much betterlook, more precise estimates, better characterizationof that relationship.
the last commenter mentionedthat we don't really know much specific toteens how technology-- either the infotainmentsystems or technologies designed to control thefunctionality of the phone-- whether they'rebeneficial for teens. or could they, themselves,be a source of distraction. we don't reallyunderstand, again, for novice teens as muchabout the importance of visual versus cognitivesources of distraction.
as i mentioned, don'swork has indicated that the pc-based trainingfor eye glance behaviors can be effective when evaluatedin simulated or limited on-road environments. we really need toscale that, because to the extent thatthat's really beneficial, that would be agood example of how to increase the sort ofpotential successive driver training programs to focus onthe very specific skills we
know that novicedrivers don't have when their first drivingindependently on the road. an area of interestof mine is really, so what about these friends. obviously, we know thathaving peers in the car is detrimental to teen drivers. the most effective strategywould be to eliminate them. that's why the provisions ingraduated driver licensing laws are there, and they've beenrepeatedly proven effective.
but we also know thatthat's not something for which there's100% compliance. so better understanding of thespecific situations under which peers really impartrisk, and how to modify both thebehavior of the driver as well as the behaviorof the passenger, i think is going to be really critical. i mentioned parents are an allyin this, as long as they're given the tools that we knowcan put them in a position
to be most effective in servingtheir role as coach, monitor, supervisor, delivererof consequences. and i just don't think we haveenough evidence yet-- some of us have been involved, again,with state farm's support, and have identified oneparticular intervention designed to helpparents be better supervisors of theirteens during the learning to drive process, and havefound that to be effective. so most parents aremotivated to try
to keep their teenssafe behind the wheel, they just don't know whatexactly they're supposed to do. and then when it comesto laws, of course, we have verylimited data on what the effects of cellphoneand texting ban laws are, both on teenbehavior in terms of how they're using thedevices, as well as on crashes. so we need to characterizethat a little bit better. with that, i'll stop, and wehave time for some questions.
we have time for one ortwo questions for dan. dr. durbin. i happen to have one or two. first of all-- it's a good thingyou're not driving. here we go. have you studiednovice adult drivers? without that control, can yousay whether novice teenager risk is worse becausethey are teenagers,
or because they are novices. i have not personally, thoughi'm involved in a study with dr. currey, againusing new jersey data, where young drivers are underthe graduated driver licensing law to age 21. so all novices up to 21 aresubject to all the provisions of the graduateddriver licensing law. and the early resultsfrom analyzing that as a functionof age-- as well as,
i think, there'ssome european data, because licensingtends to happen later-- is that the shape ofthat curve, that peak, in the early stages ofnovice driving, and then it going down with experience,the general shape of the curve is likely preserved, but thepeak is moderated by age. so novice drivers, regardlessof how old they are, need to gain experiencebehind the wheel. there is a so-calledmaturity effect,
that the younger you arewhen you start that process, the higher your initial risk is. but there's definitely an effectof being a novice at any age. and then there'sanother question. you mentioned increasein non-driving tasks under some conditions. might the same occur with somedriving aid safety systems, like lane-keeping assiststhat nearly completely assumes that particular task.
in the perfect world,such partial automation would allow a driver toreallocate the newly available resources to theother driving tasks. however, a driver mayreallocate those resources to texting, et cetera. would teens be more prone toengaging in non-driving tasks under such conditions. the short answeris, i don't know. i think that's a veryinteresting question.
as we move moretowards semi-autonomous and fully autonomous vehicles,it's one thing for us as experienced driversto think about how we will interactwith those vehicles. it's another thingentirely if you have zero driving experience. and i think that's a very goodtestable question about is it beneficial orharmful as teens learn to drive to partiallyautomate some
of the functions of driving. that's a good empiricquestion to my knowledge. anybody in the roomknow if anybody's got an answer that question? i don't believe we knowanything about that, but that's ananswerable question. thank you, dennis. thanks. cathy chase isthe vice president
of governmental affairs ofadvocates for highway and auto safety, commonlyknown as advocates, a coalition of nationalconsumer, health and safety organizations, and the majorproperty and casualty insurance companies, andtrade associations. she joined the staffof advocates in 1996 as manager oflegislative affairs, and was promoted to directorof state affairs in 1999. her professional career has beenin the areas of transportation
advocacy and policy,teenage violence prevention, and support for at-riskyouth, political fund-raising and campaigning, andgovernment relations. she has worked and volunteeredfor numerous nonprofit organizations in the washington,dc area related to education, at-risk populations, andlegislative advocacy. she has givennumerous presentations on highway andauto safety issues, and has testified beforestate legislatures.
with that, cathy is goingto talk to us about us state and federal laws thattarget distracted driving. thank you, sue, forthat nice introduction. i appreciate it. i know we're heading toward theend of the day, so i appreciate you still being here, andawake, and moving around. so i thank you forlistening today. as sue said, iwork for advocates, so i come from this from avery different perspective
than the other speakers thatyou heard from today, because we are not academics, andwe are not researchers. we are a nonprofit highwaysafety lobbying group. so what i'm goingto talk about is what happened to get usto where we are today, and what we hope toaccomplish in the future to address the problemof distracted driving. for advocates, we base allof our policy decisions on research that we garnerfrom other organizations,
including the insuranceinstitute for highway safety, and a number oforganizations that are represented in thisesteemed audience before me. we are proud to be celebratingour 25th anniversary as an organization this year. so as i said, we lobbyon the state, federal, and regulatory levels. and for the last 11years, one of the things that we have issued is astate highway safety report.
it's kind of like a reportcard for the states on 15 different basic highwayand auto safety laws. it usually garnersa significant amount of national and local attention,because states usually like to see how they do. and we code them red,yellow, and green. and we work veryclosely with the media as part of that effort, and alsoas all of our advocacy efforts. this is the cover of the reportthat we issued this past year,
just to give you a littleflavor of what it looks like. and this is just a sampleof one of the articles that covered our report card. and you'll see that,obviously, the picture is why i picked thisone-- because he's texting whiledriving-- but south dakota was rated the lowest. and you would think thatsometimes the states would contact us and be irritated.
that doesn't happen that often. we hope that this tool inspiresthem to improve their highway and auto safety laws. and we have foundthat to be the case. not as quickly aswe'd always like, but we do find thatit is a helpful tool. so with the proliferationof in-vehicles devices, we have found, ofcourse, that distracted driving has becomea bigger problem.
while distracted drivingincludes any activity that diverts the attention from thedriver from the primary task that they arecharged with, we have found that the use ofelectronic devices which require visual, auditory,cognitive, and sometimes manual attention, isparticularly troublesome. i don't know that thegeneral public really understands the graspof this problem, including that in 2012-- thelast year for which we have
information-- more than3,300 people were killed, and more than 421,000have been injured in distracted drivingcrashes nationally. and teens are, unfortunately,the largest proportion of distracted drivers. and even though these numbersare large and startling, there is a significantproblem of under reporting, for which advocatesis very concerned. there are different databases inpolice crash reporting schema,
and the limitationof crash databases are a significant problem. the reason why this is a problemis because we really need an accurate portrayal ofthe magnitude of the problem to address it effectively. research has shown thatdistracted driving is the equivalent toimpaired driving, and i think that the generalpublic is kind of surprised by this, becauseit's a newer issue.
and the generalpublic understands, don't drink and drive. i think they arestarting to understand don't text and drive,especially the younger people. but whether thatis translated yet to understanding the fullproblem that is before us, i don't know if that'shappened quite yet. while there are technologiesto disable devices while cars are in motion,they are not in common use
by the general public right now. additionally, there are, ofcourse, engineering approaches to reduce distracteddriving, but there are no federal regulationsright now that require them in cars-- which i thinkgoes hand-in-hand. consequently, in theshort term, advocates works to effectuate somechanges via counter measures that have been proven toaddress other highway safety challenges, for theway to change behavior.
so a brief background. as i mentioned, i'dstart with the background to say how we've gottenwhere we are today. in 1997, nhtsa issueda comprehensive report which assessed thestate of knowledge regarding the safetyimplications of using wireless communicationwhile driving. and it's important to notethat at that time, only 1 in 10 americans wereusing cellphones--
a big change fromwhere we are now. a few years later, at theurging of safety organizations, nhtsa held a public meetingon the safety implications of telematics in communications. at that time, industryvoluntary efforts resulted in lax guidelines,including a total task time of up to 20 seconds. subsequent research underscoredthe need for concern about the then-emergingpublic safety risk,
including a 2003 study whichshowed that merely talking on a cellphone whileoperating a vehicle disrupts the driver'sattention, which can lead to what is calledinattentional blindness, or the inability to recognizethe driver's field of vision. the virginia techtransportation institute, which you've heardfrom earlier, i'm sure, released several large scalenaturalistic driving studies that highlighted the seriousdangers of distracted driving.
and additionally, in 2011, texastransportation institute study found that drivers respondedmore slowly to either writing or reading text. so in 2009, theobama administration issued an executiveorder prohibiting federal employeess-- which is aworkforce of approximately four million people-- fromtexting while driving on official government business,or while using government supplied the equipment whiledriving it any other time.
under the leadership of thentransportation secretary ray lahood, the us departmentof transportation convened national summitson distracted driving. i'm sure some ofyou in the audience here were part of that. also, dot issued somefederal regulations to address distracted drivingby commercial motor vehicle operators, and driverstransporting hazmats. dot also established anational education campaign,
and nhtsa issued somevoluntary guidelines. yet clearly, there areadditional counter measures to address distracteddriving that can be taken by the administration, includingregulating the use of current and emerging in-vehicleand portable technologies. in 2009, the ntsbchairman debbie hersman announced a ban ontexting and talking on cell phonesfor stafff-- which was approximately 400 staffpeople-- who were driving
during work hours, orwhen on personal time if the driver is using afederally-owned wireless device. in response to a 2009 highlyvisible fatal multi-vehicle collision in whichthe ntsb determined that the probable causewas distraction likely due to texting, thentsb issued a number of safety recommendations,one of which was that all statesshould enact bans
on portable electronicdevices while driving. so what have the states done. they started out in 2001, withnew york enacting the first ban on using handheldphones when driving. in 2007, washington statebecame the first state to enact an all-driver textingban, although at that time it was a secondaryenforcement law. it was subsequently improvedto a primary enforcement law in 2010.
and yet prior to thewidespread efforts to ban handheld cellphonesand texting while driving, the initial focus was onthe dangers of distracted driving on novice teendrivers through the passage of state graduateddriver licensing laws. in 2002, maine and newjersey were the first states to enact cellphoneplans for teens. on the federal level,we know that congress can enact laws that requirestates to pass laws or be
penalized, or they usuallydo incentive grants. and what happenson the state level is the states candecide whether they want to participate in this,and be granted additional funds or penalties being withheld. and so on the statelevels, what advocates does is we work with anumber of victims. and we try to mobilize themand work with them so that we can effectuate changeon the state level.
and the way that wehave found that is to be effective to do thisis what we call the three es. i know that in the engineeringcommunity, one of the es is engineering. but in terms oflegislation, we say that there's theenactment of the law, the education of the publicabout a safety hazard, and the enforcement of the law. a prime example of theneed for these three es
was low safety values. in 1981, only 14% ofamericans used a safety belt, despite 15 years ofeducational campaigns aimed at convincingdrivers to buckle up. however, after ashift in strategy that included not onlyeducation, but enactment of mandatory safety beltlaws and high visibility enforcement, the national beltuse rate rose to 86% by 2012. similarly with impaireddriving, a shift
occurred when the nationalminimum driving age act of 1984 was enacted, compelling everystate to pass a 21 drinking age law or face a reductionin federal aid highway construction funding. and then again, afterthe national 0.08% bac law was passed in 2002. these laws have resultedin notable changes to drunk driving deaths. the insurance institutefor highway safety
issued a specialreport in 2001, which provided convincing historicalevidence that the most effective highway traffic safetyprograms combine education with the enforcementof traffic laws, and that's what advocatesworks to accomplish. just to say the oldsaying, the devil's in the details, when we'reworking to advocate for safety laws, we pay specificattention to the difference between primary andsecondary enforcement.
primary, of course, meaningyou can be pulled over for violating thatspecific violation. secondary, you need to breakanother law before a police officer can pull youover for that violation. advocates for highwayand auto safety does not endorse secondaryenforcement laws, as they have beenproven not to be as effective as primaryenforcement laws. additionally, thereare the problems
with the exceptions tothe rule, such as texting in stopped traffic. i think we've all probablybeen behind someone when the light has turned, andyou see someone looking down at their device. and it is a problembecause other cars may advance while thisperson is texting. so we work to try to preventthese special exemptions to these safety laws.
we have found thatweak highway safety laws, with these exceptionsand special circumstances, have been shown to createenforcement challenges, and thereby weaken thefull impact of the laws. i think that it'sinteresting to note that for handheldcellphone bans, from 2001 till now-- a total of 13years-- only 12 states and the district of columbiahave passed primary enforcement handheld cellphone bans.
compared to from 2007until now, 43 states and the district of columbiahave passed texting bans. yet only 38 of those andthe district of columbia are primary enforcement. still that's aninteresting comparison to make with the speed of whichtexting bans have burgeoned in our nation. and also comparatively,teen cellphone bans-- a total ban on any useof cell phones-- from 2002
until now, 37 states andthe district of columbia have passed those bans. 31 and the district of columbiaare primary enforcement. the only total ban thathas been introduced has been in chapelhill, north carolina, and currently it's pending. it's been underreview of the courts. and on march 17th, the supremecourt heard oral arguments, and we're waiting to see howthe court decides on that ban.
so this just provide some chartsthat you can see in comparison. there are no states thathave secondary enforcement handheld cellphone bans,which is interesting to note. and then these are thestates-- how it pans out with the differentprimary enforcement, secondary enforcement, partialban-- meaning that there are significant exceptions,like i just mentioned-- and then threestates without a law. and then states that banthe use of cellphones
while driving for teens. you'll see that asignificant amount have primary enforcement, and yet westill have a lot of work to do. there are a significantamount of states that have no law on this issue. so then i wantedto bring us to what is happening on thecongressional level, just to provide a brief summary. on the federallevel, congress began
to address distracted drivingin approximately 2009, when both the senate and the housecommittees held hearings. the senate commerce, scienceand transportation committee held a hearing, during whichsecretary lahood and fcc chairman genachowski testified. senator schumerfrom new york also spoke during thathearing about a bill that he'd introduced-- thealert drivers act-- which would impose financialpenalties on states that did not
pass adequate texting bans. but this bill did notmove out of committee, either in the houseor the senate. similarly the housetransportation and infrastructure committee,and the house energy and commerce committee heldhearings in the fall of 2009, during which, again,secretary lahood testified, as well as members ofcongress and representatives from industry and thehighway safety community.
yet with all of thesehearings, it brought attention to the issue ofdistracted driving, but no legislation was passed. which brings us to thereauthorization bill called map-21. map-21 is the resultof significant lobbying from safety groupsand the leadership of a number ofmembers of congress, including commerce chairman,senator rockefeller.
in 2012, congressenacted a program to incentivize the statesto pass primary enforcement all-driver textingbans, and cellphone bans for young drivers. this law, map-21, expiresseptember 30th of this year. so one of the things thatadvocates is working on is our next reauthorizationpriorities-- what to include in continuingsome of these grant programs. so public opinion is alwaysimportant to us as a lobbying
group, because it kind ofbuttresses our positions on where we want to go. and it certainlyis helpful when we can point to polls that show theamerican public is behind this and, congressman, youshould be as well. the public has consistentlysupported efforts to combat distracted driving,including a very high 90% showing supportfor texting bans. which may be evidence of thehand-in-hand relationship,
since 43 states have this law. a strong majority--this is a kind of a summary ofa number of polls that we did-- a strongmajority of approximately 70% support handheld cellphone bans. and an average of about halfsupport total cellphone bans. so in conclusion, we stillhave some work to do. steps have been taken bythe obama administration, the national transportationsafety board, congress,
safety advocatesand, as i mentioned, victims and survivors ofdistracted driving crashes. federal and statelaws have ensued, but we still have alot of work to do. with more than 3,300 deathsand 400,000 injuries annually in distracted driving crashes--and cellphone use known to be a leading distraction--it will take continued and more expansive state and federalgovernment efforts to address the dangers posed by currentand future emerging technologies
while driving behind the wheel. the implementation of the threees approach to highway safety issues-- enactment, education,and enforcement of laws-- should be continuedand enhanced. this method has beensuccessful in addressing the issues of low safety beltuse and impaired driving. and moreover, the publicis primed for support. lastly, we urge theunited states department of transportation to take anaggressive leadership role
in regulating the useof current and emerging in-vehicle andportable technologies. use of technology to temporarilydisable electronic distractions and prevent theiruse by drivers needs to be evaluated and implemented. driver use of any technologythat presents an unreasonable risk to safetydue to distraction should be temporarily disabled. regulations, and notvoluntary guidelines,
should apply to portable as wellas built-in electronic devices. i wanted to thank sue andeveryone here for having me i'd be glad to answer anyquestions that anybody had. any questions for cathy on thelegislative or regulatory side. and it looks like,vickie, you might have one from our viewers on the webcast. this is from someonewho wants to know if you're working on speciallaws regarding rural roadways, because she had a tragedy whereshe lost two grandchildren
in a car crash ona rural roadway, and doesn't feel that that'sgetting enough attention. well, first of all, i want toextend my deepest condolences to that family. and we do. we work on all roads,including rural roads. there are different problems fordifferent regions, certainly. and if this family isinterested in working with us, our website is saferoads.org.
and we really relishworking with these families to advocate change. jim foley, toyota collaborativesafety research center. thank you for an excellentsummary of the laws. i do have to take issue withone of your initial statements, where you said there isresearch that equates distracted driving to drunk driving. true, that researchexists, but that analogy is a horrible analogy.
a distracted driver can bedistracted in one second, and alert in the next. a drunk driver is drunkfor a very long time, and much more dangerous. and to repeat thatanalogy, i think minimizes the importanceof reducing drunk driving, and is not a useful one to use. thank you for that response. that wasn't my research.
yeah, yeah. and as you know, advocatesworks on all of these issues, so i appreciate youpointing that out to me. i have another question froma webcast viewer about map-21. she says only onestate, connecticut, qualified for thedistracted driving grant program under map-21. what is advocates' positionon the current requirements for states?
should they be modified inthe next reauthorization? i don't know that that's quiteaccurate-- that connecticut was the only state forthe distracted driving. i know that there werevery strict provisions added to the graduated driverlicensing program which prohibited, i think, any statefrom qualifying for the grant programs. and we are workingto kind of fine tune that in the nextreauthorization, because it
is our hope thatthese grants inspire the states to effectuate change. we don't hope themto be onerous. the goal is to getthese laws on the books. ok, if there aren't any furtherquestions, thank you, cathy. so our last speaker--i should say last, but definitely notleast-- dr. anne mccartt is going to talk to us. having heard about thelaws, obviously the question
is, how effective are they. and anne will betalking about that. dr. anne mccartt issenior vice president for research at the insuranceinstitute for highway safety in arlington, virginia. the institute is an independentnonprofit scientific and educationalorganization dedicated to reducing the losses--that is deaths, injuries, and property damage--from highway crashes.
at the institute, she overseesa multi-disciplinary research staff whose workcenters on finding ways to change driver behavior,improve roadway design, and make vehicles safer. she's authored more than160 technical reports and scientificpapers on such topics as alcohol impaired driving,automated enforcement, distracted driving, youngdrivers, older drivers, and occupant protection.
she was the president ofthe association of aaam, of automotivemedicine, and serves on expert committeesand advisory boards of the transportationresearch board, and other safety organizations. i'm really pleasedto be here today. i want to add my thanksto aaam, especially irene, and to state farm,for gathering together this interesting andvery thoughtful group.
i'd also like tothank my co-authors. david kidd is aresearch scientist. he's here in the audience today. and eric teoh, a seniorstatistician at the institute. so our task was to take alook at the effectiveness of cellphones and texting bans. so we did a review of theliterature in this area. and i want to start using seatbelt laws as an example, which follows closely with cathy.
i want to remindus of what it takes to do a strong evaluation. and i think you'll see by theend of the presentation that is very difficult to do strongevaluations of cellphone laws. so ideally, whena law is passed, there's solid evidencethat the behavior that the law istargeting is associated with injuries or crashes. and that was certainly thecase with seat belt laws.
it was really clear, based onlots of research and experience with laws in other countries,that wearing a seat belt reduces your risk of injuryor fatality substantially. there are four typesof evaluation measures, and you heard theterm causal chain used in a different context. but in [? evaluative ?]research, the term causalchain is often used to refer to taking measuresthat lead from the law
to the ultimate outcome. so they are measures that getat the law implementation. how many citations are written? was the law publicized? then you want to knowwhether the message was received by drivers. do they understand the law? are they aware ofthe enforcement? and then the two mostimportant measures
would be driver behavior-- howdoes the law affect behavior-- and how does the law ultimatelyaffect crashes or injury measures. these are the most importantmeasures, as i said, and these are the measuresthat studies looking at cellphone lawshave focused on. if you think aboutseat belt laws, it was very interesting thatthe first studies of seat belt laws in this country foundthat while belt use went up,
over time it began to go down. and that led researchersto circle back and look at the lawimplementation issues, and led to what we nowaccept-- that it's not enough to pass a law. the law has to beenforced and publicized. then in terms of theevaluation components, we measure our behavior incrash outcomes before the law. we measure themagain after the law.
and then critically important,to be a strong evaluation, there has to be somesort of a control. so it might be,if you're looking at a law in a particularstate, the control might be a comparisonstate without a law. the importance ofa control is it lets us know whatwould've happened. what we would expectwithout a law. and it helps us eliminateother explanations
for whatever we find. so as you heard from cathy,and i think as we all know, strong, publicizedenforcement of laws has been a reallycritical countermeasure in highway safety. and at this point,only three states don't have some kind of lawtargeting cellphone use. there is some researchlooking at the effects of these laws on behavior.
so our institute did a series ofstudies-- observational survey studies-- that found thatall driver laws targeting handheld phone conversationsdid have a significant effect on handheld phone conversations. there weresignificant reductions even several years afterthe laws were implemented. based on a nationaltelephone survey, we found that driversin banned states reported higher ratesof hands-free phone use,
but overall lower ratesthan non-banned states. so in other words,we found evidence that some drivers didswitch to hands-free, but not all drivers did so. overall use rates were lower. you heard from dennis, basedon a study in north carolina, there is some evidencethat laws that ban any kind of phoneuse among teenage drivers don't appear to haveaffected their phone use.
there's basically no evidenceon the effects of texting bans on texting behavior by drivers. and lastly, i'llmention, as you might know, nhtsa took the"click it or ticket" model, and applied it to ademonstration project in hartford, connecticutand syracuse, new york, where there was very heavyticketing and publicity about that. and it did result in lowerrates of handheld phone
conversations, andphone manipulations based on observations. so now looking atthe studies that have examined the effects ofcellphone bans on crashes. one set of studies haslooked at the states that ban all drivers fromhaving a conversation on a handheld phone,and you see them here. there are 12 ofthem today, and dc. the first one was passedin 2001 in new york.
a few years later, new jerseyand connecticut passed laws, and then most of them havebeen passed more recently. so when we took alook at the studies-- i should say wefocus in our review on studies that were publishedin peer reviewed journals or technical reports. we found nine of these studieslooking at cellphone laws. and as i say onthe slide, they use a variety of measures--insurance collision claims.
most use some kind of measuresof fatal crash involvements. and basically, what yousee is mixed findings. so there were four studieslooking at state specific laws, using either fatal ornon-fatal crash measures. three of those were conductedlooking at new york's law. the fourth study looked at newyork and three other states. so some of the analysesfound increases in crashes, some found reductionsin crashes, some found nosignificant changes.
there were also mixedfindings from five states that looked either at--excuse me-- five studies that looked either ata group of states or cross-state national studies. all these studies usedfatal crash measures. i might point out-- andyou'll hear this theme for me throughout my presentation--that some of the studies, actually most of them,had important limitations. often it had to dowith mis-coding laws.
not taking intoaccount-- these studies looked at handheld cellphonelaws-- not taking into account that there might be texting bansor teen-specific bans in place. not accounting for aconfounding factor. so doing a state study where youdon't have a state comparison. and then sometimes, briefafter-ban study periods. or with the nationalstudies, in a lot of cases they were basically lookingat new york in many cases, because new york hadthe first law in place.
so it really turned out to bemostly a study of new york. the second set ofstudies looked at laws that banned alldrivers from texting. and actually, since iprepared this slide, south dakota passed a law. so they're now 43 states theyban all drivers from texting. and as you can see, someof the laws are primary, some are secondary. and the first lawspecifically targeting texting
was implemented inwashington in 2008. so these are really recent,making it a challenge to evaluate-- thatso many of them have been implementedvery close to one another. so there are fewerstudies of texting bans. we found threethat are criteria. two peer reviewed,one technical report. one of the studieswas an analysis of insurance collision claims.
it looked at four stateswith texting bans, and control states without bans. and the authors foundsignificant small increases in three states. no change in the four state. there were two peerreviewed national studies. and as you can see here,they had mixed findings. actually, withinthe studies, there were mixed findings, and thencomparing the two studies.
one study lookedat single vehicle, single occupant fatalcrashes, on the theory that these were the crashesthat would be most affected by texting. the other studylooked at fatalities. and again, it's a challengeto do these studies because there's so manystates that are passing law. there are already handheldcellphone bans in place. so i want to talk brieflyabout one major challenge.
and i think thisdoesn't just relate to doing studies of bans. it relates to doingother studies. and this has come up acouple of times today. what this slide showsis the difficulty in relying onpolice crash reports to identify crashesinvolving distraction. these are the percents ofdeaths in the fatality analysis reporting system, going backto 1999, and then through 2012.
these are the percentof deaths coded as involving driver distraction. you would see a similarpattern if you focused just on crashes involvingcellphone use specifically. the red line isthe national rate. and then these otherlines represent states. and the two points i wantto make with this slide are that if youlook within states-- in maryland, forexample-- the purple line.
you see these odd anomaliesoccurring from year to year without apparent explanation. and you can alsosee the differences across these states. and again, there reallywould be no logical reason that one state might have 60%of deaths involving distraction, while one might have 10%. and i think when you think abouthow police crash reports are filled out, when apolice officer arrives
at the scene ofthe crash, there's basically usually noobjective evidence. especially in afatal crash, where the driver might have died. there's no objective evidenceto indicate whether or not distraction was a factor. if we knew the crash isinvolving distraction, it would make it a lot easierto study the effects of laws, but we don't.
so we concluded from our reviewthat despite an increasing number of laws limitingphone use in the us, it's unclear whetherthey're having the intended effects on behavior in crashes. now the exceptionto that might again be the handheldbans, where we have found that thehandheld bans appear to be effectivein reducing people talking on handheld phones.
and then the secondbullet, again, gets back to this issue we'veheard today so much about. which is, there really isunsettled science at this point about the crash riskassociated with phone use, especially seriouscrashes, and even more so, fatal crashes, whichmost studies use. and the lack of ouragreement on this makes it difficult toformulate reasonable hypotheses about what you'd expect tosee when you look at crashes.
or to choose appropriatecrash measures. and then there are otherchallenges identifying controls with texting bans, for example,with so many states having texting bans. it's difficult to identifyappropriate controls. in most of the studies,there's very little evidence-- or no evidence-- about howthe law was implemented, whether it wasenforced, publicized, or whether it affecteddriver behavior.
and then you have thischanging picture over time, where you have laws beingenacted-- lots of variety in laws, and that presentsa challenge to evaluators. so we came up-- moving ahead--with three research priorities. the first is--you've heard again, this from otherspeakers-- trying to do better studies lookingat the crash risk associated with phone use that addresssome of the limitations of prior studies.
the second pointis, we believe there needs to be astronger validation from the naturalisticstudies of the association of non-crash surrogatesfrom naturalistic studies with crashes ofdifferent severities. and again, especially injurycrashes or fatal crashes. if you think about texting, forexample, there's not a study-- a sound study-- that hasrelated texting to crash risk. and then finally, we think weneed more, better evaluations
of cellphone and texting laws. not just looking atcrashes, but trying to, again, build that causal chainfrom the enactment of the law, to the enforcement, to thedriver behavior change, and then to crashes. so a couple more points. you can probablygather from what i said that webelieve it's the case that cross-statenational studies have
special challenges. they use fatal crashesbecause fatal crashes are available atthe state level, whether or not that makes sense. but when you get down to monthlydata, which many studies use, and state data,you're still getting small samples of crashes. it's difficult to identifycontrol variables. the studies tend to be largestatistical models that
incorporate economic factorsand other variables that might be affecting crashes. but if you think aboutthe last few years where we've had severe economicrecession, unstable gas prices, again, the strengthof the study rests on whether that study canidentify the factors that might have made a difference,other than the law. i've mentioned it'sdifficult to account for all the variations in laws.
and again, to thesort of reiterate, there's just generally alack of data on compliance. state specific studies, ithink offer advantages-- partly just simplicity. but there can be avery strong design if an appropriate controljurisdiction can be identified. it's much easier to build thatcausal chain within a state, where you can lookat driver behavior in the implementationof the law.
and you don't have torely on fatal crashes when you're lookingat just one state. i want to conclude by-- toillustrate these points. this is a study that was done byour sister agency, the highway loss data institute. these are collisionclaim frequencies. and frequencies, inhldi speak, means rates. these are collision claimsper insured vehicle. and this is a look at oneof the states, connecticut,
that passed an all-driverhandheld ban, which is the vertical line there. it shows when it took effect. and claims inconnecticut are graphed in the blue lineon a monthly basis. and then there were twocomparison states, graphed here in the red line. and i want to make a coupleof points with this slide. the first one is, if you lookbefore the line in what we call
the baseline period, therewas a very strong similarity in the trends incrashes on this slide. so that would suggest this is avery good control to be using. the other couple ofthings i want to point out is that it might seem unusualto look at collision claims as a crash measure,but they tend to be low severity, mostlyproperty damage only crashes. and one of the strongest studiesof crashes, also conducted by our institute-- oneconducted by our institute
looked at personalinjury crashes. but there was asimilar study done in canada looking atproperty damage only crashes. these are thestudies that you've heard referred to that founda fourfold increase in crash risk. so when this studywas conducted, there was a strong, sortof theoretical basis for looking at collisionclaim frequencies.
and finally, in connecticut. connecticut was oneof the studies where we had looked at whether thelaw affected driver behavior. and it did. it led to reductionsin handheld phone use. so there was a logicand a coherence to this study in terms of tryingto bring the pieces of evidence together, to take agood hard look at this. so i'd be happy toanswer any questions.
and i want to thankyou for persisting to the end of the day. thank you, anne. ok, i see we have questions. oh, oh. tom dingus from virginia tech. so the original new yorklaw-- and part of it as an artifact of the wayphones were back in the day, was really a headset law.
and that didn't really getto a lot of the problems with manual phones. in other words, now,when you're driving, you fumble withyour manual phone, and you fumble with you headset,and you put your headset on, but you still may dial,and things like that. so really, the onlything it was doing was getting the phoneout your hand, which was a minor part of the problem.
do you know if any of thenewer laws address that issue, or talk about-- you sayit's a total handheld ban. do you know if they allallow headsets, still? my understanding-- and cathymay know the laws better-- but my understandingis that the laws today are much like thelaw in new york. and actually, theselaws do allow dialing. they are worded differently,but generally, what's against the law is holdingthe phone to your ear
and conversing on it. and there's differences. some states don't allow talkingwhen your stopped, some do. like at a stoplight,for example. i would hypothesizethat's really going to limit theeffectiveness of those laws. i mean there's probablybetter, and may we could collectively givesome advice to the legislators. but there's probablya better ways
to put those lawsinto effect where you would see a bigger benefit. something to think about. carl nash. i have no particularaffiliation. iihs spends a lotof money rating the crash safety of cars. have you thoughtabout doing a rating of the degree to whichthe various navigation
and entertainmentsystems distract drivers to give a safety ratingof those systems in cars? what did you mentionbesides navigation? i couldn't hear. entertainment. in other words, all the stuffthat's in the middle there. well, i don't thinkwe're at a point where we could do that yet. we are doing someresearch with researchers
from mit, who arehere today, to try to look at some of theinfotainment systems, especially focusingon voice activation. sort of leadingup to that study, we did a study-- david kiddand ian reagan, who's also here today, took a look at some ofthe voice activated systems. and counted the steps involvedcompared to the manual versions of the systems,and actually found that some of the voice activatedsystems required more steps.
and our study is lookingat several questions, but one of thequestions is looking at whether a voiceactivated system has less effect ondriving performance than other measures,than manual systems. i don't think-- we're just atthe beginning of that research, i would say, andknowing more about whether, in fact, these systemsare less distracting then the manual systems.
and if so, to what extent. after my last comment,somebody in the audience stated that there's a realdifference between the systems in different kinds of cars. so if anecdotallythat's correct, it seems to me you should beable to do a more formal study of which cars are betteror worse in that respect. well in our study, we'relooking at a system we judged to beeasier to use, and one
we judged to be harder to use. but we're looking justat two at this point. i might mention,too, since i can, that the institute thinks thata solution-- we haven't talked really about it much-- iscrash avoidance technologies. we would say, of the availablealternative countermeasures, we believe that in the longrun, may be the most successful at reducing distraction andother driver-- drowsiness, for example, thatlead to crashes.
so we are putting a lotof effort into that. dan mcgehee from theuniversity of iowa. more of a comment. one thing that's really great. law enforcementis evolving in how they investigatetheir own crashes. and we're seeing now it'spretty standard protocol for law enforcement to ask forcellphone records for almost all crashes to tryto understand that.
at least we've beenpromoting that greatly. the other issue i think weshould all be looking at, too, is really promotingmore advanced event data recorders in vehicles, as well. from an engineering perspective,from a data recording perspective, we can tap into theelectronic systems of vehicles. so we can actuallyreconstruct what kind of button presses,what kind of systems were in operation atthe time of an impact.
so that would be onearea to perhaps look at. there's a lot ofsensitivity and politics to data in vehicles,especially crash recorder data. but it's something toconsider in the broader conversations of this. well, you hit on oneof the big challenges in doing goodstudies of crash risk which is, it's verydifficult to know exactly when the crash occurred.
and to relate that. and in the study thatwe did in australia and the canadian study, wegot cellphone billing records. and we looked atdifferent windows leading up to thetime of the crash. but we acknowledged,as a limitation, that there was certainly someuncertainty about exactly when the crash occurred relativeto talking on the phone. what we use in some ofthe recording packages,
because time,universal time, is so common with in-vehiclesystems because they're linked to the internet somehow. in the currentnon-sophisticated world, we use the 911 callsthat are generally made fairly soon after acrash was reported-- usually within a minute, dependingon where you are. it can be several minutes,but at least a 911 call is sort of theanchor of when we can
start to look at those records. but it is imprecise. this viewer wantsto know if you've studied the impact ofpenalties on compliance. no, i don't know. the question waswhether there are studies looking at thefines, for example. no, i don't know of any studiesthat have looked at that. john lee from theuniversity of wisconsin.
i have a follow-up questionin response to tom dingus, in a way. and that is, a lot of yourstudies-- like the last one you showed-- was looking atbans before smart-phones. and just for myself, the talkingon the phone, at least for me, is the functionthat i use least. lots of other things can be usedwhile driving, beyond the phone conversation. and just to clarify-- iwant to make this clear.
i do get confused and distractedwith adele and springsteen, but i never use my phone inthe car, either for talking, texting, or otherwise. but i can imagine doingso, and it frightens me. so i'm just curiousif you speculate about the effectivenessof some of those bans today, when we havesmart-phones, versus the less sophisticated stuff in the past. well, i think if you lookat the-- if you think back
to the first slide i showed,and you want this sound basis for laws, i think it'snot entirely logical that the laws have remainedfocused on handheld laws. and then texting, which is quitespecific, the texting bans, as i understand it, are quitespecific to texting and not other manual manipulationsof the phone. and among the things thatwe don't know for sure-- in our case, the crossoverstudy in the canadian one, we found handheld and handsfree phone conversations
were equally-- had equallyelevated crash risk. but then, if you look at thenaturalistic studies based on near-crash measuresmostly, you don't see that. so in the lack ofsome of this evidence, i honestly think it'sdifficult to craft laws that are reallysound in terms of based on what weknow about the risk. and also, in termsof enforcement, i think we all can agreethat if someone is texting,
you don't know if they'rechecking their social media, or sending a text message. so law enforcement,as i mentioned before, is further complicated becauseof these different exceptions. there are some statelaws, as i understand it, that get specificin terms of there's no surfing on the internet,or there's no this or that, but the enforcement ofthat is quite challenging. ok if there are nofurther questions, that
finalizes our listof presenters. i'd like to ask chrismullen of state farm to just sum up for the day. and thank you, once again,all of you, for attending. and thank you,again, all of you, for staying thebalance of the day. and i want take this opportunityto thank our incredible panel and sue ferguson, once again,for the content that you heard today.
i want to thank theaudience, in particular, for your engagement, as wellas our audience on the webcast for participatingwith your questions. that was fantastic. i want to remind everyonethat the findings of this incrediblepanel will be available as a peer reviewed publication,the annals of advances in automotive medicine,available for download from aaam.org.
for the next month,free of charge. as well, the webcast willbe available on demand shortly followingthis symposium. so again, on behalfof state farm, i want to thank susan ferguson. i want to thank ourpanelists, and all of you, really, for your interestin this important issue, and for your continuedcommitment to auto safety. so safe travels, everyone.
and thank you very much.
0 komentar :
Posting Komentar