Texting Driving Study - Center For Transportation Safety

2y ago
9 Views
2 Downloads
1.19 MB
69 Pages
Last View : 2m ago
Last Download : 2m ago
Upload by : Rafael Ruffin
Transcription

Technical Report Documentation Page1. Report No.2. Government Accession No.3. Recipient’s Catalog No.SWUTC/11/476660-00024-14. Title and Subtitle5. Report DateAN INVESTIGATION OF THE EFFECTS OFREADING AND WRITING TEXT-BASEDMESSAGES WHILE DRIVINGAugust 20117. Author(s)8. Performing Organization Report No.Joel Cooper, Christine Yager, and Susan T. ChryslerReport 476660-00024-19. Performing Organization Name and Address10. Work Unit No. (TRAIS)Texas Transportation InstituteThe Texas A&M University SystemCollege Station, Texas 77843-313511. Contract or Grant No.12. Sponsoring Organization Name and Address13. Type of Report and Period CoveredSouthwest Region University Transportation CenterTexas Transportation InstituteTexas A&M University SystemCollege Station, Texas 77843-3135Final Project ReportSeptember 2010 – August 20116. Performing Organization CodeDTRT07-G-000614. Sponsoring Agency Code15. Supplementary NotesSupported by a grant from the U.S. Department of Transportation, University Transportation CentersProgram16. AbstractPrevious research, using driving simulation, crash data, and naturalistic methods, has begun to shed light onthe dangers of texting while driving. Perhaps because of the dangers, no published work has experimentallyinvestigated the dangers of texting while driving using an actual vehicle. Additionally, previous research doesnot clearly differentiate the dangers associated with reading and writing text messages. To address these issues,42 participants drove an instrumented research vehicle on a closed driving course. Participants drove under acontrol, text reading, and text writing condition. Baseline text reading and writing data were also collectedoutside of the research vehicle. Results indicated that impairment associated with texting while driving may begreater than previously thought. Principally, when reading or writing texts, drivers exhibited reductions inreaction time that were nearly twice as great as previously thought. Drivers also exhibited nearly identicalimpairment in the reading and writing conditions, suggesting that both reading and writing text messages maybe equally dangerous. These results have immediate implications for improving our understanding of thedangers of texting while driving and may be useful for future public policy discussions.17. Key Words18. Distribution StatementTexting, Distracted Driving, Distraction, Mobile DeviceUse, ImpairmentNo restrictions. This document isavailable to the public through NTIS:National Technical Information Service5285 Port Royal RoadSpringfield, Virginia 2216119. Security Classification (of thisreport)20. Security Classification (ofthis page)21. No. of PagesUnclassifiedUnclassified69Form DOT F 1700.7 (8-72)22. PriceReproduction of completed page authorized

An Investigation of the Effects of Reading and WritingText-Based Messages While DrivingbyJoel Cooper, Ph.D.Center for Transportation SafetyAssistant Research ScientistTexas Transportation InstituteChristine Yager, E.I.T.Center for Transportation SafetyAssociate Transportation ResearcherTexas Transportation InstituteandSusan T. Chrysler, Ph.D.Center for Transportation SafetySenior Research ScientistTexas Transportation InstituteReport SWUTC/11/476660-00024-1Project 476660-00024Project Title: An Investigation of the Effects of Reading and WritingText-Based Messages While DrivingSouthwest Region University Transportation CenterTexas Transportation InstituteThe Texas A&M University SystemCollege Station, Texas 77843-3135August 2011

DISCLAIMERThe contents of this report reflect the views of the authors, who are responsible for thefacts and the accuracy of the information presented herein. This document is disseminated underthe sponsorship of the U.S. Department of Transportation University Transportation CentersProgram in the interest of information exchange. The U.S. Government assumes no liability forthe contents or use thereof.ACKNOWLEDGMENTSThe authors recognize that support for this research was provided by a grant from theU.S. Department of Transportation University Transportation Centers Program to the SouthwestRegion University Transportation Center.The authors would also like to thank those at the Texas Transportation Institute whohelped with the project. Kathryn Bennett was a tremendous help with the data collection efforts.Kevin Spharler was a big help configuring the instrumented vehicle, and along with TaylorHaby, Katheryn Davila, and Brittany Badillo, helped reduce the collected data prior to analysis.Katie Connell managed the participant recruiting and scheduling, in addition to helping withreport formatting. Sue Chrysler offered invaluable guidance and direction from the initial projectidea to the final report.v

vi

ABSTRACTPrevious research, using driving simulation, crash data, and naturalistic methods, hasbegun to shed light on the dangers of texting while driving. Perhaps because of the dangers, nopublished work has experimentally investigated the dangers of texting while driving using anactual vehicle. Additionally, previous research does not clearly differentiate the dangersassociated with reading and writing text messages. To address these issues, 42 participants drovean instrumented research vehicle on a closed driving course. Participants drove under a control,text reading, and text writing condition. Baseline text reading and writing data were alsocollected outside of the research vehicle. Results indicated that impairment associated withtexting while driving may be greater than previously thought. Principally, when reading orwriting texts, drivers exhibited reductions in reaction time that were nearly twice as great aspreviously thought. Drivers also exhibited nearly identical impairment in the reading and writingconditions, suggesting that both reading and writing text messages may be equally dangerous.These results have immediate implications for improving our understanding of the dangers oftexting while driving and may be useful for future public policy discussions.vii

viii

TABLE OF CONTENTSAbstract . viiList of Figures . xList of Tables . xiExecutive Summary . xiiiIntroduction . 1Approach and Methodology. 5DEMOGRAPHICS . 11Order of Events and Instructions . 12Findings. 15Light Response Time . 17Missed Light Responses . 18Mean Speed . 19Standard Deviation of Speed . 20Standard Deviation of Lane Position . 21Text Messaging Characters Per Minute . 22Reading Rate Per Minute . 23Subjective Performance Rating. 24Summary and Conclusions . 25Furthur Discussion . 27References . 29Appendix A: Types of Mobile Devices . 31Appendix B: Cellular Information Form . 33Appendix C: Stories Sent to Participants . 35Appendix D: Verbal Instructions to Participants . 43Appendix E: List of Counterbalanced Segments . 45Appendix F: Post Experiment Interview Responses. 47ix

LIST OF FIGURESFigure 1. Closed course divided into a work zone and open sections . 6Figure 2. Diagram of course layout . 7Figure 3. Green LED mounted on vehicle hood . 8Figure 4. Participant responding to light by pressing joystick button . 9Figure 5. TTI’s instrumented vehicle. 9Figure 6. A forward facing camera was mounted on each side of the vehicle to tracklateral lane position . 10Figure 7. A camera was aimed at the driver to track head position and eye glances . 11Figure 8. A second camera was aimed at the driver’s profile to track hand and bodymovements . 11Figure 9. Definition of box plot components . 15Figure 10. Light response time statistical analysis . 17Figure 11. Missed light responses statistical analysis. 18Figure 12. Mean speed statistical analysis . 19Figure 13. Standard deviation of speed statistical analysis. 20Figure 14. Standard deviation of lane position statistical analysis . 21Figure 15. Text messaging characters per minute statistical analysis . 22Figure 16. Reading rate per minute statistical analysis. 23Figure 17. Subjective performance rating statistical analysis . 24x

LIST OF TABLESTable 1. Summary of demographic data for all participants . 12xi

xii

EXECUTIVE SUMMARYGovernment estimates in the United States suggest that around 25% of all fatal crashesare directly related to driver distraction. Research from the Virginia Tech 100-car study suggeststhat cellular device usage, either for conversation or other visual manual tasks, is the single mostcommon source of distraction. When used to send text messages, cellular phones are thought toincrease fatal crash risk by an astounding 6-23 times over baseline. Although the number ofannual fatalities caused by text messaging is not known, the potential scope of the problem canbe appreciated by the facts that: in 2009, an estimated 5 billion text messages were received perday in the United States, and that 20% of all U.S. drivers have admitted to texting while driving.To date, only a small handful of experimentally controlled research studies havespecifically looked at the dangers of texting while driving, and these have typically been carriedout using a driving simulator. The primary objective of this project is to assess the distractionpotential of sending and receiving text messages while driving under varying roadway andtexting response demands. In order to assess the individual contributions of reading and writing,these activities were blocked in separate experimental conditions. Driving demand wasmanipulated through changes in roadside objects. Drivers operated an instrumented vehicle on aclosed course.In order to better understand the problem of texting while driving in the real world, whilemaintaining a high degree of experimental control, this research used an instrumented researchvehicle on a closed driving course. Research was conducted at TTI’s Riverside campus, which isa 2000-acre complex of research and training facilities situated 10 miles northwest of TexasA&M University’s main campus. The site, formerly an Air Force Base, has large expanses ofconcrete runways and parking aprons that are ideally suited for experimental research and testingin the areas of vehicle performance and handling, visibility, distracted driving, and drivertraining. Participants drove a closed course, on a portion of a straight, mile-long runway at theTTI Riverside Campus. The course was set up toward the outer edge of the runway andmaintained widely paved boundaries for additional safety. This research setting and designallowed participants to drive an actual vehicle with its accompanying dynamics and interact withtheir personal cell phones, all in a highly controlled research setting that reduced risk exposure toxiii

tolerable levels. This approach leveraged the strengths associated with using an actual vehicle,with research control that is typically only found in the laboratory.A total of 42 subjects participated in this driving study. Participant ages ranged from 16to 54, and gender was approximately split. Attempts were also made during recruiting efforts toevenly split participants by the type of mobile device they had, touch screen or QWERTY raisedkeys, to see if there were any differences in driver performance or texting proficiency.Results were analyzed using 1- and n-way repeated measures ANalysis Of VAriance(ANOVA). Violations of sphericity in the ANOVA testing were corrected by adjusting thedegrees of freedom following the Greenhouse-Geisser procedure. Violations of the homogeneityof variance assumption did not lead to any adjustments or corrections. Though potentiallybiasing, the importance of variance homogeneity may be marginalized when larger sample sizesare used and when groups of equal sizes are compared; both of which are present in the currentset of analyses. For clarity and readability, the unadjusted degrees of freedom will be reported,regardless of whether an adjustment was made in interpreting the F statistic, thus preserving theability of the reader to identify the exact number of samples that were used to compute eachstatistic.Compared to the control condition, reading and writing text messages led to a significantdelay in response time, an increase in the number of missed response events, an overall reductionin speed, an increase in the standard deviation of speed on the open roadway sections, anincrease in the standard deviation of lane position on the open roadway sections, a reduction inwriting and reading rates, and a reduction in the number of glances to the forward roadway. Ofthe various dependent measures considered in this analysis, only response time to the light taskwas differentially affected by the writing and reading tasks, with the greatest response timeimpairment associated with writing a text message. Thus, overall performance on each of themeasures herein considered clearly indicated significant impairment from both writing andreading text messages. Results also suggest that any possible difference in driving impairmentassociated with writing and reading are likely very small.These findings suggest that previous research may have underestimated reaction timedelays associated with texting and driving, that reading and writing text messages are equallydifficult, and therefore, equally dangerous, and that when text messages are sent from thevehicle, the efficiency of both the texting and driving tasks are dramatically reduced. However,xiv

many questions remain unanswered. Thus, our recommendation is two-fold. First, that enough isknown about reading and writing text messages to suggest that drivers should never text behindthe wheel. Second, that additional research is needed to better understand when drivers are likelyto engage in these types of tasks in the real world.xv

INTRODUCTIONGovernment estimates in the United States suggest that around 25% of all crashes aredirectly related to driver distraction (1). Research from the Virginia Tech 100-car study suggeststhat cellular device usage, either for conversation or other visual manual tasks, is the single mostcommon source of distraction (2). When used to send text messages, cellular phones are thoughtto increase fatal crash risk by an astounding 6-23 times over baseline (3, 4). Although thenumber of annual fatalities caused by text messaging is not known, the potential scope of theproblem can be appreciated by the facts that: in 2009, an estimated 6 billion text messages weresent/received per day in the United States (5), and that 20% of all US drivers have admitted totexting while driving (6).In response to the perceived dangers of sending and reading text messages while driving,38 out of 50 U.S. states now have at least partial bans on texting, yet many important questionsremain unanswered. Similar to conversing via cellular phone, sending and reading text messagesis an activity that may allow drivers to selectively initiate the activity during moments of lowtask demand. However, unlike conversing on a cellular phone, the actual act of sending orreading a text message is one that can be interrupted at any moment, without social consequence(momentarily at least), allowing the driver to pay full attention to the road. An importantdistinction can be drawn between cell phone conversation and texting; texting is amenable toresumption after selective disengagement while conversation may be more difficult to resumeand, once initiated, is usually terminated slowly. However, the question of whether or not driversactually modulate text messaging engagement, based on task demand, is not well addressed inthe literature.To date, only a small handful of experimentally controlled research studies havespecifically looked at the dangers of texting while driving, and these have typically been carriedout using a driving simulator. The first published study came out in 2009 by Hosking, Young andRegan (7). They had 20 participants drive a computer simulated roadway that contained anumber of emerging threat events, a car following episode, and a lane-change task. Resultsindicated that drivers were particularly impaired when sending text messages and less so whenreceiving. In particular, they found that drivers’ ability to maintain their lateral position, theirability to detect and respond to traffic signs, the amount of time spent looking at the road, and1

their following distance, were all impaired when sending and receiving text messages. However,Hosking, Young, and Regan did not control for task exposure time during the text receiving andsending episodes and thus, it is not clear whether the results they obtained were due todifferences in task exposure time, or differences in the distracting nature of the tasks themselves.Additionally, participants did not use their own phones, the effect of which was unknown on theresults.A second study published in 2009 by Drews et al. was also completed using a simulateddriving environment (3). Here again, 20 participants drove a simulated roadway while sendingand receiving text messages. Unlike earlier work, in this research, participants were able to usetheir own phones. Additionally, the text messages sent and received in this study were sharedbetween actual friends, thus the actual communication was likely more representative of everyday text messages. The driving tasks consisted of following a periodically braking lead vehicledown a 65 mph two-three lane roadway. Results indicated that when texting, participantsexpressed greater following variability, greater lateral variability, reduced response time to thelead vehicle, and an increase in collision frequency. Brake response times associated withreading were reported to be higher than those associated with writing. However, because thereading and writing portions of this research were not balanced, the actual amount of drivingtime associated with reading was likely very low. Indeed, just 1.2 brake response observationsper subject were made while reading an incoming text, compared to an average of 14 brakeresponses while writing. Thus, the actual effects of reading and writing a text message on drivingperformance cannot be determined from this research.An additional factor that may play a crucial role in determining the extent of drivinginterference from texting is whether drivers are actively reading or writing the text message.While there are various theoretical reasons why differentiating the effects of reading and writingon driving could be interesting, from a practical standpoint the answer could have immediateimplications for the types of activities that are and are not allowed in vehicles. Currently, thereare a number of common secondary driving tasks that are heavily reliant on reading text basedinformation from a screen that may not necessarily require text entry. On the one hand, if textentry is significantly more impairing than reading, then future laws could be selectively writtento ban the writing but not reading of text based communications. On the other hand, if the2

writing and reading of text based messages is found to be equally impairing to drivers, thensimply banning one, but not the other, may not adequately address the problem.The primary objective of this project is to assess the distraction potential of sending andreceiving text messages while driving under varying roadway and texting response demands. Inorder to assess the individual contributions of reading and writing, these activities were blockedin separate experimental conditions. Driving demand was manipulated through changes inroadside objects. Drivers operated an instrumented vehicle on a closed course.3

4

APPROACH AND METHODOLOGYDriving simulators are often used to address research questions that are deemed to be toodangerous or impractical to study using other methods, which may account for the fact, to date,no published research has experimentally evaluated the effects of texting on driving using anactual vehicle. While there are several advantages to using driving simulation to addressquestions about an activity as potentially dangerous as texting while driving, there are alsoseveral limitations that make it difficult to confidently generalize the findings of a study usingdriving simulation to the real world.In order to better understand the problem of texting while driving in the real world, whilemaintaining a high degree of experimental control, this research used an instrumented researchvehicle on a closed driving course. This allowed participants to drive an actual vehicle with itsaccompanying dynamics and interact with their personal cell phones, all in a highly controlledresearch setting that reduced risk exposure to tolerable levels. This approach leveraged thestrengths associated with using an actual vehicle with research control that is typically onlyfound in the laboratory.This research was driven by three primary research questions: How well do texting while driving results obtained using an actual vehiclecompare to those using a driving simulator? Do drivers change the way they interact with non-driving tasks as the driving taskbecomes more demanding? When texting while driving, does driving impairment from reading differ fromwriting?In order to address these questions, a driving study was conducted at TTI’s RiversideCampus, which is a 2000-acre complex of research and training facilities situated 10 milesnorthwest of Texas A&M University’s main campus. The site, formerly an Air Force Base, haslarge expanses of concrete runways and parking aprons that are ideally suited for experimentalresearch and testing in the areas of vehicle performance and handling, visibility, distracteddriving, and driver training. Participants drove a closed course, on a portion of a straight, milelong runway at the TTI Riverside Campus. The course was set up toward the outer edge of therunway and maintained widely paved boundaries for additional safety.5

To address the question of whether drivers change the way they interact with non-drivingtasks as the driving task becomes more demanding, researchers had participants drive a coursethat was evenly divided between an open and unrestricted lane and one that was bordered bylane-restricting construction barrels (see Figure 1). In the barreled roadway section, a total of 17pairs of barrels, placed 140 feet apart, marked the edges of the lane boundary. In the open sectionof the roadway, lane boundaries were demarcated by existing pavement markings and/or seamsin the concrete paving blocks. Barrels were placed at the outside edges of the 12-foot wide lines.The barreled section was 2,380 feet, and the open section was 2,140 feet.Open.41 MilesStart/EndBarreled.45 MilesFigure 1. Closed course divided into a work zone and open sectionsEach drive began in the middle of the course, at the transition between the barreled andopen roadway sections (see Figure 2). Half of the participants began each drive in the work zone,and half began each drive in the open section, the order of which was counterbalanced to reduceany biasing effects on the data. Beginning in the middle, participants drove the full course twotimes for each of the driving conditions. At the end of the barreled and open sections were pairsof cones, which indicated to participants that they needed to make a U-turn and continue again6

down the length of the course. As an example, a participant may have begun in the open section,driven .41 miles to the end of the course, made a U-turn, driven back through the open section(.41 miles), driven through the barreled section (.45 miles), made a U-turn, returned through thebarreled section (.45 miles), driven the open section (.41 miles), made a U-turn, returnedthrough the open section (.41 miles), driven the barreled section (.45 miles), made a U-turn,returned through the barreled section (.45 miles), and then stopped, having completed two fulllaps of the course (3.5 total miles).Figure 2. Diagram of course layoutTo evaluate the effects of reading and writing texts on driving, three separate drivingsegments were completed (control, reading, writing), each approximately 10 minutes in length.The order with which participants completed the three segments was counterbalanced acrossparticipants.The three segments consisted of: Control: No text messages were sent or received; driver focused on maintainingspeed of 30 mph, maintaining lane position, and responding to light response task. Writing: Driver composed a story on their mobile device while driving and tryingto maintain speed of 30 mph, maintain lane position, and respond to lightresponse task. Reading: Driver read a story on their mobile device while driving and trying tomaintain speed of 30 mph, maintain lane position, and respond to light responsetask.7

The light response task consisted of a green LED that was attached to the hood of theinstrumented vehicle within eye line of the driver (see Figure 3 and Figure 4). The light wasprogrammed to initially turn on after 20 seconds and remain on for 15 seconds unless a responsewas received from the participant via the joystick button. Thereafter, the light was programmedto turn on at pseudo-random intervals defined by a normal distribution with a mean of 45seconds and a standard deviation of 5 seconds. Light state was recorded at 60 Hz and stored forlater analysis.Figure 3. Green LED mounted on vehicle hood8

Figure 4. Participant responding to light by pressing joystick buttonParticipants drove TTI’s instrumented 2006 Toyota Highlander on the closed course (seeFigure 5). The principle data collection system within the instrumented vehicle is the DewetronDEWE5000 data acquisition integration system. Essentially a large portable computer, theDEWE5000, serves as the data acquisition device for all the peripheral systems in the vehicle.Figure 5. TTI’s instrumented vehicle9

A Trimble DSM232 Global Positioning System (GPS) was used to track the speed andposition of the subject vehicle. It employed a high gain GPS antenna mounted on the roof of thevehicle, directly over the driver’s seat. The receiver was mounted in the rear cargo area near theDEWE5000.Two fo

the dangers of texting while driving. Perhaps because of the dangers, no published work has experimentally investigated the dangers of texting while driving using an actual vehicle. Additionally, previous research does not clearly differentiate the dangers associated with reading and writing text messages. To address these issues,

Related Documents:

texting is now ubiquitous, there remain gaps in the literature on texting and driving. A Deadly Wandering (2014), by Matt Richtel, examines texting and driving from both a behavioral and legal standpoint. Richtel's book introduces the topic of texting and driving, as well as studies relating to brain activity and attention, to a lay audience.

accidents from texting and driving has reached 1.3 million (Texting and Driving Statistics, 2002-2014). This amount will only keep rising if something is not done about this major distraction. All of the accidents occurring from texting and driving comes from the driver thinking he or she can multi-task.

Bruksanvisning för bilstereo . Bruksanvisning for bilstereo . Instrukcja obsługi samochodowego odtwarzacza stereo . Operating Instructions for Car Stereo . 610-104 . SV . Bruksanvisning i original

TEXTING & DRIVING PSA COMMUNICATION CHECK Driver Conversations Alexandria, VA & Dallas, TX May 4-6, 2015. BACKGROUND The stats on texting and driving are sobering: 18% IN 2012 of fatal crashes were caused by distracted driving (NHTSA) of teens feel

Only prohibits drivers from texting while driving Texting while driving is a primary offense in Michigan (and in all but 4 states), meaning that police can pull a vehicle over and issue a ticket just for texting Mich. Stat. §257.602b —Cell Phone Usage Distracted Driving Laws in Michigan Talking to Our Kids

– States have banned both texting while driving and hand-held cell phone use while driving in an attempt to decrease distracted driving – New York banned texting in 2001 – By 2010, 23 states (and DC) had banned texting and 6 states (and DC) had banned hand-held cell use HOWEVER: A 2011 analysis of the Governor’s

No Texting While Driving This chapter walks you through the develop-ment of No Texting While Driving, an app that autoresponds to text messages you receive while you're driving. The app, first created with App Inventor by a beginning computer science student, is similar to a now-mass-produced app developed by State Farm Insurance. It is a prime

interaction from texting while driving. This task involved a scenario in which participants receive a text message while driving, and they rated the likelihood of replying to a text message immediately versus waiting to reply until arriving at a destination when the amounts of a fine for texting while driving ranged from 1 to 300.