1
Creatingtheenvironmentfordriverdistraction:Athematicframeworkof
sociotechnicalfactors
KatieJ.Parnell*a,NevilleA.Stanton
a,andKatherineL.Plant
a
*Correspondingauthor
aTransportationResearchGroup,FacultyofEngineeringandtheEnvironment,BoldrewoodInnovation
Campus,UniversityofSouthampton,BurgessRoad,Southampton,SO167QF,UnitedKingdom.
Abstract
Asmodernsocietybecomesmorereliantontechnology,itsusewithinthevehicleisbecomingaconcernfor
roadsafetyduetobothportableandbuilt-indevicesofferingsourcesofdistraction.Whiletheeffectsof
distractingtechnologiesarewelldocumented,littleisknownaboutthecausalfactorsthatleadtothedrivers’
engagement with technological devices. The relevance of the sociotechnical system within which the
behaviouroccursrequiresfurtherresearch.Thispaperpresentstwoexperiments,thefirstaimstoassessthe
driversself-reporteddecisiontoengagewithtechnologicaltaskswhiledrivingandtheirreasoningfordoing
so with respect to the wider sociotechnical system. This utilised a semi-structured interview method,
conductedwith30driverstoinitiateadiscussionontheirlikelihoodofengagingwith22differenttasksacross
7differentroadtypes.Inductivethematicanalysisprovidedahierarchicalthematicframeworkthatdetailed
the self-reported causal factors that influence the drivers’ use of technologywhilst driving. The second
experiment assessed the relevance of the hierarchical framework to a model of distraction that was
established from within the literature on the drivers use of distracting technologies while driving. The
findingsprovidevalidationforsomerelationshipsstudiedintheliterature,aswellasprovidinginsightsinto
relationships that require further study. The role of the sociotechnical system in the engagement of
distractions while driving is highlighted, with the causal factors reported by drivers suggesting the
importanceof considering thewider systemwithinwhich thebehaviour is occurringandhow itmaybe
creating theconditions fordistraction tooccur.This supportspreviousclaimsmadewithin the literature
basedmodel.Recommendationsareproposedthatencourageamovementawayfromindividualfocused
countermeasurestowardssystemicactors.
Keywords:In-vehicletechnology,Driverdistraction,Qualitativemethods,Sociotechnicalsystems.
1. Introduction
Technologicaldevelopmentsarelargelydrivenbyindustrialorcommercialrequirementswhich,Dorf(2001)
claims,areharnessedbymankindtochangeormanipulatetheirenvironment.Thedrivingenvironmenthas
changedmarkedlythroughtheimplementationoftechnologyoverrecentdecades(Walkeretal,2001).This
hashadramificationsforthedesignanduseofvehicles(Wierwille,1993;Walkeretal,2001).Driversnow
2
expect thedesignof thevehicle to include technological facilities thatenableentertainment,navigation,
communication,connectivity,efficiencyandcomfortwhiledriving.Yet,thereisaneedtoensurethatthe
implementationof such technologiesdoesnotadverselyaffect roadsafety (Leeetal,2008;Youngetal,
2011).
The distractive effects of hand-held phones have been evidenced,with adverse consequences to driver
performancemetrics,suchasvehiclecontrol(Tsimhonietal,2004),attentiontunnelling(Reimer,2009),and
hazarddetection(Summalaetal,1998)amongothers.Yet,despitebeingmadeawareoftherisksposedby
mobile phoneswhile driving and legislation to ban their use acrossmany countries, drivers continue to
engagewiththem(Dingusetal,2006;Lerneretal,2008;Walshetal,2008;Zhouetal,2012;Young&Lenné,
2010;Metz et al, 2015; Tivesten&Dozza, 2015).While previous research has informed on the adverse
consequencesofmobilephones,thecontextualandmotivationalfactorsthatleadtoengagementinother
technological tasks isunder-researched (Young&Regan,2007;Youngetal,2008;Young&Lenné,2010;
Tivesten&Dozza,2015;Horreyetal,2017).
Some research has been conducted into the decisions that driversmake to engagewith distractions in
simulators(Metzetal,2011;Schömig&Metz,2013),ontesttracks(Horrey&Lesch,2009)andthroughthe
analysisofdataderivedfromnaturalisticstudies(Metzetal,2015;Tivesten&Dozza,2015).Achallengein
the assessment of driver distraction research is the dichotomy between high levels of control and the
naturalistic studyofbehaviour (Youngetal,2008), thus thebenefitsand limitationsof thesestudiesare
inherenttothevalidityofthefindings.Whilesimulatorsoffercontroloverexternalvariables,suchasroad
type and other road users, capturing realistic behaviour is compromised (Young et al, 2008). Yet, in
naturalistic studies the focus of data collection is on the driver and their triggered engagement with
secondary tasks as they allow very little control, and thusmeasurement of, the contextual factors that
influence drivers’ engagement with secondary tasks (Metz et al, 2015). TheWorld Health Organisation
(WHO)nowacknowledgesthesociotechnicalsystembasedapproachwhichidentifiesdriverbehaviour,not
asaproductoftheindividual,butasaproductofsystemicelementssuchastheroadlayout,roaddesign,
vehicledesign,andthecontextsurroundingthedrivingtask(WHO,2004).Despitethis,theapplicationof
systemsbasederrormanagementapproacheshavebeenlargely ignored(Salmonetal,2010).Thecausal
errortaxonomysuggestedbyStantonandSalmon(2009)statesfivekeyelementswithinthesociotechnical
systemwhichinfluencetheconditionsthatleadtoerror;thedriver,thevehicle,roadinfrastructure,other
roadusersandenvironmental conditions.Thus, it canbesuggested that thecauseofdistraction related
errorsisnotlimitedtothedriver,insteaditisinfluencedbyamultitudeofothersystemicactors.
Reviewingdistractionwiththesociotechnicalsystems‘riskmanagementframework’(RMF)developedby
Rasmussen (1997) revealed the impact that hierarchical levels of the systemhave on the emergence of
distraction.Actorswererevealedfromtheinternationalandnationalcommittees(Parnelletal,2017a)who
3
setthelawsthatareenforcedbylocalgovernmentsandregulatorsthatthenfeeddowntheframeworkto
themanufacturersofdevicesandtheinteractiontheyhavewiththeenduser(Young&Salmon,2012;Parnell
etal,2017a).Ratherthanfocusingonthedrivers’decisiontoengageastheinitiationoferror,thesystems
approach gives an insight into the conditions through which the driver was permitted to engage with
distractingtechnologiesandhowthisbehaviourinfluencestheemergenceofsafetywithinthesystemasa
whole.Yet,appropriatemethodsarerequiredtoassessthesociotechnicalsystem(Youngetal,2013;Salmon
etal,2017).
Inafirstattempttoassessandmodeldriverdistractionfromasociotechnicalsystemsapproach,Parnellet
al(2016)developedthePARRC(Priority,Adapt,Resource,Regulate,Conflict)modelofdistraction,thefirst
model of the behaviour to account for the contribution of systemic factors. This encompasses five key
mechanismsthroughwhichin-vehicletechnologymayleadtodistractionacrossthesociotechnicalsystem,
including‘goalpriority’,‘adapttodemand’,‘resourceconstraints’,‘behaviouralregulation’and‘goalconflict’
(Parnell et al, 2016). The PARRC model was developed through grounded theory methodology which
determined thekey factors involved in theemergenceofdistractionasevolved from the literature. The
interconnections made between these mechanisms were shown to influence how distraction related
behaviouremergedfromthesystem,aswellastherelevanceofothersystemicactorsonthemechanisms.
Readers are directed to Parnell et al (2016) for further information. Application of the PARRC model
mechanismstoanAccimapanalysissuggestedhowactorsinthesystemmaybepreventingtheemergence
ofdistractionorconverselyleavingthesystemopentodistraction(Parnelletal,2017a).Thishighlightedthe
roleoflegislation,developedthroughinternationalandnationalcommitteesthatisthenenforcedthrough
national laws, that targets hand-held mobile phone use but is more ambiguous on the use of other
technologies.Theambiguityinlegislationwasshowntohaveledtotheadvancementoftechnologiesand
theirimplementationwithinthevehicle,despitealackofevidencetosuggestthemtobesaferthanhand-
heldmobilephones(Parnelletal,2017a).Yet,themechanismsofthePARRCmodelweredrawnfromthe
literatureusinggroundedtheoryandthereforerequirevalidationthroughtheirapplicationtootherdata
sources,methodsand/or investigatorsthroughtheprocessoftriangulation(Hignett,2005;Raffertyetal,
2010).
Thispaperseekstogaindatafromdriversontheirselfreportedreasonsforengagingwithtechnologywhile
driving. Questionnaires and online surveys have strived to gather responses on drivers’ frequency of
engagingwithdistractionsandtheirviewsontherisksindoingso(e.g.McEvoyetal,2006;Young&Lenné,
2010;Walshetal,2008;Zhouetal,2009;Zhouetal,2012;).Yet,theyareoftenprescriptive,posingclosed
questionsthatmaylimitthedatatotheagendaoftheresearcher(O’Cathain&Thomas,2004).Instead,the
causalfactorsthatdriversdeemtoinfluencetheirdecisiontoengagewithdistractions,andhowthismay
resultindistractionrelatedevents,isofinterest(Youngetal,2008;Young&Lenné,2010;Lee,2014).The
first experiment within this paper sought to obtain the drivers self-reported reasons for engaging with
4
technology while driving using a semi-structured interview method to engage drivers in open-ended
discussionsonwhytheymaybemore,orless,likelytoengagewithvarioustypesoftechnologywhiledriving.
The inductive thematic analysis thatwas applied during the data analysis aimed to develop factors that
driversthemselvesdeemto influencetheirengagementwithtechnologicaltasks.Thesecondexperiment
aimedtoassesshowthecausalfactorsderivedfromthedriversintheinterviewstudyrelatedtothecausal
factorsthatweredevelopedfromtheliteratureinthedevelopmentofthePARRCmodel(Parnelletal,2016).
Thismodelisusedforitsabilitytoassessthesociotechnicalsystemsurroundingthebehaviour(Parnelletal,
2016;2017a).Thefindingsseektoassistintheprovisionofcountermeasuresthattargetthesourceofthe
issue,ratherthanobservingwithhindsighttheeffectsofdistraction.
2.Experiment12.1Aim
This experiment aimed to understand the drivers self-reported reasons for engaging with technological
deviceswhiledrivingandtheinvolvementofthesociotechnicalintheirdecision-makingprocess.Previous
researchhassoughttocapturethedrivers’useoftechnologiesusingquestionnairesandonlinesurveys,yet
thisstudyaimstocapturethedrivers’subjectiveperspectiveintheirownwords.Thiswillinvolvetheuseof
semi-structured interviews toelicitdiscussionswithdriverson their likelihoodofengagingwithdifferent
technologicaltasksacrossdifferentroadtypes.
2.2.Method
2.2.1Participants
DriverswithexperienceofUKroadswerespecifiedastheroadtypesincludedwithinthesemi-structured
interviewsrelatedtothosecomprisingtheUKroadwaysystem(Walkeretal,2013).Atotalof30participants
wererecruited(15females,15males),acrossthreeagecategories(18-30,31-49,50-65),withfivefemales
andfivemalesineachcategory.ParticipantswererecruitedundertherequirementthattheyheldafullUK
drivinglicenseandhadaminimumof1-yearsexperiencedrivingonUKroads(meanyearsexperience=19.5,
SD=13.08).Theywerealsorequiredtobefrequentdrivers,drivingonaregularweeklybasisinorderforthem
tobeexposedtosituationswheretheymaybeinclinedtoengagewithtechnology(meanhoursspentdriving
aweek=9hrs45min,SD=6hrs20mins).Participationwasvoluntary.
2.2.2Datacollection
To obtain the drivers own views on why they engage with technological devices while driving, semi-
structuredinterviewswereconducted.Semi-structuredinterviewshavebeenusedeffectivelytoinvestigate
otheraspectsofdrivingbehaviour(Simon&Corbett,2007;Gardner&Abraham,2007;Tonetto&Desmet,
2016),buttheyhavenotbeenappliedtostudyhowdriverdistractionisviewedbydrivers.Theirapplication
within this research allowed for open-ended questions that enabled drivers to generate concepts they
5
deemedimportanttotheiruseoftechnologicaldeviceswhilstfacilitatingastructureddatacollectionmethod
that could be reliably applied across all interviewees. They also allowed the researcher to probe into
interestingconceptsastheyarose(Cohen&Crabtree,2006).
The interviewswere structured around a table that encouraged the driver to discuss their likelihood of
engagingwitharangeofdifferenttechnologicaltaskswhiledrivingacrossdifferentUKroadtypesinorder
toprovideadiscussionsurroundingthesituationsandenvironmentswhichmayinfluencetheuseofdifferent
technologicaldevices.Table1presentsthelistoftechnologicaltasksposedtoparticipantsintheinterview.
Thetasksweredrawnfromthecurrentliteratureinvestigatingdistractionfromin-vehicletechnology(e.g.
Young&Lenné,2010;Nealeetal,2005;McEvoyetal,2006;Harveyetal,2011a),aswellasreportsfrom
roadsafetyorganisationsandpolicereports(RAC,2016;DepartmentforTransport,2015a).Theroadtypes
presentedhavebeenshowninpreviousresearchtoinfluencedrivers’situationalawareness(Walkeretal,
2013)andcrashrates(Bayliss,2009).Theyincludedmotorways,majorA/Broads,urbanroads,ruralroads,
residential roads and junctions. Participants were presented with a road type classification sheet with
definitions,imagesandcontextualinformationrelatingtoeachoftheroadtypesforclarity.
Table1.Listoftechnologies/specifictasksthatdriverswereaskedtoratetheirlikelihoodofengagingwith.
Technology Task
NavigationsystemMonitorroute
Enterdestination
Hands-freesystem
Findnumberfromaddressbook
Answeracall
Talktoother
In-vehiclesystem
Changeclimatecontrol
Changesong/radiostation
Adjustvolume
Listentomusic
Verballycommunicatewithin-builtsystem
Mobilephone/Portabledevice
Enterdestinationintonavigationapp
Monitornavigationapp
Write/sendatext
Readatext
Answerphonecall
Talkonthephone
Enter/Findanumber
Changesong/audiotrack
Usevoiceassistfeatures
Takeaphoto
Usesocialmediaapps
Checkyouremail
Participantswerefreetogeneratetheirownreasoningbehindwhytheymayormaynotengagewitharange
ofdifferenttechnologicaltaskswhiledrivingacrosstheroadtypes.Theresearcherprobedtheparticipantto
6
expandontheirdiscussionpoints forclarityandfurther informationwherenecessary.Thesameprimary
researcherconductedtheinterviewswithallparticipantsforconsistency.
Apilotstudywasconductedtodetermineifthetechnologicaltaskswererepresentativeofthoseusedby
driversandtoestablishagreementonthedescriptionsoftheUKroadtypes.Thisrevealedanoverlapinsome
of thetechnological tasks, suchassearching forapointof interestandadestination inasat-nav. Italso
revealedthatwhendriversdiscussedtheirbehaviouratajunctiontheyseemedtodifferintheirdiscussions
surrounding the use of technologywhen stopped at a junction e.g. at traffic lights compared to driving
throughanintersection.Therefore,thejunctionroadtypewassplitintwotorepresentbothdrivingthrough
ajunctionandstoppedatajunction.
Theinterviewslastedapproximately30minutesalthoughthisvarieddependingonthediscussionsengaged
bytheparticipantandtheresearcher(average=34.21mins,SD=14.07).Interviewswereaudiorecordedand
transcribed. Due to the sensitive content, i.e. if they were likely to engage in an activity that may be
consideredillegalunderUKlawssuchasusingamobilephonewhiledriving,confidentialitywasensuredto
allow theparticipant to talkopenly.The interviewstudywasapprovedby the research institutesEthical
ResearchandGovernanceOrganisation(ERGOreference:24937).
2.2.3Dataanalysis
Transcriptionsoftheinterviewsprovidedthedatasetfromwhichtoanalyseanddrawinferencesonthe
causalfactorsthatdriversreportedtoinfluencetheirdecisiontoengageinthetechnologicaltaskswhile
drivinganddeterminetherelationshipofthefactorstothewidersociotechnicalsystem.Thematicanalysis
wasusedtoorganise,analyseandinterpretkeythemeswithinthedata(Braun&Clarke,2006).Atheme
wasdefinedas“…somethingimportantaboutthedatainrelationtotheresearchquestion,andrepresents
somelevelofpatternedresponseormeaningwithinthedataset”(Braun&Clarke,2006;p10).The
flexibilityofthemethodisbothanadvantage,infacilitatingadaptabilityacrossdifferentapproaches,anda
limitationduetocommentsoflimitedrigourfromunclearmethodologies(Braun&Clarke,2006).Yet,
BraunandClarke(2006)commentthatwithclearlydefinedmethodsandcommentary,thematicanalysis
canbeahighlyusefultoolindrawingmeaningfromqualitativedata.Themethodologyappliedtothedata
setisthereforegivenindetailwithinthispaper.
ThethematiccodingprocesswasconductedinNvivo11softwaretoaddrigourtothequalitativeresearch
(Richards&Richards,1991;Welsh,2002).Italsoassistedinreviewingthesub-themes(seeExperiment2),
andallowedforqueriestoberunonthecodeddatatointerrogatethecodesaftertheyweredeveloped.
7
2.2.3.1Inductivethematiccoding
Thethematicanalysisutiliseda‘bottom-up’approach,wherebycontentwascodedwithoutapre-existing
framework, rather the frameworkdevelopedthroughtheanalysisof thedata (Boyatzis,1998).Thus, the
themesthatdevelopedwerestronglylinkedtothesourceofthedata(Patton,1990).Incontrasttodeductive
thematicanalysis,whichseekstolookataspectsofthedatathatrelatetotheresearchframeworkunder
investigation, inductive analysis provides a richer insight into the data set as a whole, using naturally
occurring themes (Braun & Clarke, 2006). It was not desirable to impose a framework on the drivers’
verbalisations,butratherdrawontheconceptsthattheparticipantsdeemedtobeimportant.Braunand
Clarke(2006)statethattheclarityofthemethodologyusedtodevelopthematiccodes isessential to its
validity, which led to the development of their own guidelines on conducting thematic coding. Their
guidelineswerefollowedandtheprocessofapplyingthemtothedatacollectedfromtheinterviewsisshown
inFigure1.Thesameprimaryresearcherthatconductedtheinterviewsalsoconductedthethematicanalysis,
astheywerefullyimmersedinthedataset.
Figure1.Stagesoftheinductivethematicanalysis.
Theiterativenatureofinductivethematicanalysismeantthattheinitial,descriptive,subthemeswerecoded
asmultiple individualconceptstodrawasmuchinformationfromthetranscriptsaspossible(Stages1-3,
Figure1).Thesewerecodedinthedriversownwords,in-vivo,tostaytruetothedata(Richie&Lewis,2003)
If framework does not fit data set return to further review and refine codes.
Stage 1Collect data - Interviews
Stage 2Transcribe data and familiarisation with data set
The primary researcher transcribed approximately 33% of interviews with the remainder sent to a transcription agency. The primary researcher spent time reading through all transcripts before coding
Stage 3Generate initial descriptive codes
The primary researcher conducted all interviews for consistency
Initial codes were generated based on the descriptions of the responses using the drivers own words where applicable. Braun and Clarke’s (2006) advice was followed: • Code as many themes as possible• Include context surrounding each excerpt• Allow exerts to be coded at more than one theme
Stage 4Review and collate themes into semantic themes
Stage 5Review and collate themesinto systemic themes
Re-read the the data extracts coded to each of the initial ‘descriptive’ themes generated at stage 3 and review to assess their coherence. Collate themes on their semantic meaning.
Once the descriptive and semantic themes are set, review for patterns across the themes that relate to systemic elements. This requires some interpretation.
Stage 6Contrast themes to data set as a whole
Go back to the initial data set with the developed thematic structure to observe its fit with the data set as a whole and code any further extracts that may have been missed.
DescriptionStageOutput
Audio recordings of the interviews with all participants
Complete transcriptions of all the interviews. Ready for input into Nvivo 11.
Extensive list of initial descriptive codes
Revised list of descriptive codes aggregated into semantic themes
Organised list of descriptive and semantic themes with systemic high-level themes that identify the patterns of the data
Refined theoretical framework encompassing the patterns observed in the data set
Stage 7Define and name themes appropriately
Review content coded to each theme to construct a meaningful definition of each theme and what it represents in the data. A codebook was generated encompassing the descriptive, semantic and systemic themes.
Clearly defined themes and subthemes that are easily identifiable and collated into a codebook.
8
and allow an emerging framework to reflect the real-life experiences and thought processes of drivers
(Tonetto&Desmet,2016).Themultipledescriptivethemeswerethenanalysed,organisedandrefinedinto
semanticthemes(Stage4).Thisrequiredsomeinterpretationandinductivetheorisingonthesignificanceof
conceptsandtheirbroadermeaning(Patton,1990).
2.2.3.2Inter-raterreliabilitytest
Allowingothers,independenttotheresearchproject,toapplythecodethathasbeenappliedtoadataset
isausefulwayofassessingthereliabilityofthecoding(Boyatzis,1998).Totestthecodingframework10%
oftheparticipants’transcriptswererandomlychosenforinter-raterreliabilitytesting.Twocolleagueswithin
the Human Factors research team were recruited to test inter-rater reliability. They were given the
codebook, which presented the full thematic framework (see the Online Appendix), during a 45minute
trainingbriefingwheretheinterviewstudywasexplainedandthethemesweredescribedbeforetheywere
askedtoindependentlycodethesame10%ofthetranscripts.Theinter-raterswereonlyrequiredtocodeat
thesemanticlevel,whilereferencingthelowerleveldescriptivethemestoaidtheircoding.Nvivo11software
wasusedfortheinter-ratercodingwiththeresearcherscodinghidden.Percentageagreementwasusedto
assessthereliabilityscoresoftheinter-raters incontrasttotheresearchers initialcoding.Thismethodis
widelyappliedininter-raterreliabilitystudies(Boyatzis,1998;Plant&Stanton,2013)andwhilethereare
stillnoestablishedstandardsontheacceptablelevelofagreementbetweenraters,Boyatzis(1998)deems
70%agreementasanecessary levelofagreement.Bothratersreachedagreementpercentagesoverthis
level(rater1=81.24%,rater2=74.87%).Thusindicatingthattheraterswereabletousetheframeworkto
codethedataata level thatwasmuchhigherthanchance,attestingtothereliabilityof theresearchers
codingandtheapplicationofthethematicframework.
2.3Results
Inductivethematicanalysisresultedinthedevelopmentofahierarchicalframeworkofthemesthatreflected
thedrivers’self-reportedlikelihoodtoengageineachofthetechnologicaltasksoneachoftheroadtypes.
AnoverviewofthehighlevelsystemicandsemanticthemesisshowninTable2.Thefullthematicframework,
includingthedescriptive,semanticandsystemiclevelsthatweredevelopedthoughtheprocess(shownin
Figure1)ispresentedintheOnlineAppendix.
9
Semanticsubthemes Description ExampleQuote
Driver Referencesmadebythedriverincludingtheirmental/physiologicalstate,experience,knowledge,skills,abilitiesandcontext-relatedbehaviour
(D1)Attitudeofdriver
Negative:Referencetonegativeattitudesofthedrivertowardsperformingthetaskwhiledriving “Because,Ijustthinkitistheworstthingintheworld,Ijustwouldn’tdoit,it’sterrible”Positive:Referencetopositiveattitudesofthedrivertowardsperformingthetaskwhiledriving “Idon’tseeanyproblemwithitpersonallywhatsoever”
Unnecessary:Referencetothedriverperceivingthetasktobeunnecessarytoperformwhiledriving “That’ssomething–it’sjustsomethingthatcanwaituntilwhenyougethomeIthink”
(D2)Tendency Referencetothedriversstatedtendencytoperformthetaskinthepastand/orthefutureasanindicatoroftheirlikelihoodtoengage “Ihavebeenknowntodothat”
(D3)Viewofself Referencetothedriversstatedviewofthemselvesandtheirownbehaviouraltendencieswhenstatingtheirlikelihoodtoperformthetask ‘ifIamstoppedIgenerallyamalittlebitmorenaughty”
(D4)Influenceofothers Referencetootherpeopleandtheirinfluenceonthedriverandtheirlikelihoodofperformingthetaskwhiledriving “theshameifyoudidsomethingbad,thateveryonewouldthinkyouaresostupid”Infrastructure Referencetothespecificroadtypewithintheroadtransportsystem,includingthelayout,contents,policy,andregulatedconditions
(I1)Perceptionsofsurroundingenvironment
Referencetothecontextsurroundingtheroadenvironmentofaspecificroadtypethatisinterpretedasaninfluencingfactorinthelikelihoodtoperformthetaskinthespecificroadenvironment
“fortheseroadsandjunctions,itwouldrequirealotmoreconcentration”
(I2)RoadLayout Referencetofeaturesofthespecificfixedroadenvironmentthatinfluencethedriver’slikelihoodtoperformthetask“becausetomeamotorway,onceyouareonit,itisallmovinginthesamedirectiongenerally”
(I3)Illegality Referencetothelegislationontheuseofthetaskwhiledriving “Iusuallyholditinalowposition,sothepolicecan’tsee”
(I4)Task-roadrelationshipReferencetotheinteractionbetweenfeaturesoftheroadandthetaskthatinfluencehowthetwomaybecompatiblesuchthatthelikelihoodofperformingthetaskisinfluenced.
“Yeahitwouldbestilted,Iwouldprobablymakethepersononthephoneaware,sayhangonaminutebutIwouldprobablysoundnotasengage”
(I5)RoadrelatedbehaviourReferencetotheactionsandresponsesthataretypicalorrequiredofthespecificroadtypewhichinfluencesthelikelihoodofperformingthetaskondifferentroads.
“urbanroadIthinkismorebusyaswellsoIthinkthemoresortofdecisionsyou’vegottomake”
Task Referencetothedetailssurroundingthespecifictaskandengagementwithit
(T1)Complexity Referencetothedifficultyoreaseofperformingthetaskwhiledriving “ifyouhavetounlockthephonescreenorwhateveretc.,itisnotassimple–wellitisquitedistracting”
(T2)Interaction Referencetophysicalfeaturesofthetaskthatrelatetotheinteractionrequiredtoperformthetaskwhiledriving.Thisrelatestotheinterfacedesign,devicelocationanddriverrequiredactionsinordertoengagewiththetask.“It’sonlyonebuttontopress,sothat’snotanissue”
(T3)Duration Referencetothetimeand/orlengthofthetask “ifit’salongtextyoumightnotreadit”
(T4)Desirability Referencetofeaturesofthetaskthatinfluencehowdesirableitmaybetoperformwhiledriving.Thismayincludeitsuse,performanceorqualityandoptionsforalternativemethodsofcompletingthetask.“Idon’treallyusemyphoneverymuchanywaysoit’sneverbeensomethingthatIhavefeltIhaveneeded”
(T5)EngagementregulationReferencetothefactorsthatinfluencetheconditionssurroundingtheonsetofthetask.Theymayrelatetothephysicality’softhetaskand/orthedrivers’motivationrelatingtothetask.
“IwillalwaysfigureoutwhatI’mgoingtolistentoandsetitgoingbeforeIleave”
(T6)Abilitytocomplete Referencetofeaturesofthetaskwhichinfluenceitsabilitytobecompletedinfullwhiledriving “becauseI’vehadthecarforages,Iknowwheretheswitchesare”
10
Table2.Thesystemsandsemanticsubthemesofthethematicframework.
Semanticsubthemes Description ExampleQuote
Context Referencetothecircumstancessurroundingthebehaviourdescribed
(C1)JourneyContext Referencetocircumstancesthatformthesettingforajourneythatmayinfluencethelikelihoodtoperformthetask.“ifIaminastrangecity,IwouldbelesslikelytomessaroundbecauseIdon’tknowwhereIamgoing”
(C2)TaskContext Referencetocircumstancesthatformthesettingfortheuseofthetaskthatinfluencethedrivers’likelihoodtoengagewithit.
“It’sstuffwhenIactuallyfeellikeIneedtosendamessagequickly,soifI’veagreedtocomehomeatacertaintimeandI’mrunninglateforinstance”
(C3)Roadcontext Referencetocircumstancesthatformthesettingsurroundingtheroadingeneral(notrelatedtospecificinfrastructure)thatinfluencethelikelihoodtoperformthetask“Ithinkitwouldbesituationaldependent,justhowbusyisit?Ithink”
11
ItisevidentfromthefulltablepresentedintheAppendixthattherewasanextensivelistofreasonsthat
drivers gave for engaging, or not engaging, with the technological tasks while driving. A total of 168
descriptive themes were iteratively generated and revised into 18 semantic thematic categories. The
generationofthesethemeswasalengthybutin-depthprocess.Clusteringthesesemanticthemesintohigher
level systemic actors that contribute to the occurrence of the causal factors gave another level to the
frameworkthatreadilydemonstratesthecontributionofthesystemwithinwhichdriverdistractionoccurs.
Inlinewithpreviousindividualfocusedapproaches,thedriveremergedasakeyactor.Thedrivercategory
suggests that the driver is influenced by their own attitudes (D1), perceptions of themselves (D3) and
tendencies(D2)intheirengagementwithdistractionswhiledriving,aswellashowtheyfeeltheymaybe
viewedbyothers(D4).Yet,thedevelopmentoftheothercategoriessuggeststhattheyarealsoinfluenced
byothersystemicactors.
Theroleofinfrastructurewasalsoreportedwhenrespondingtothedifferentroadtypesthatwereposedto
thedriverduringthesemi-structuredinterviews.Theirperceptionsoftheroadenvironment(I1)alteredhow
likelytheywouldbetoengageduetotherequirementsofthedrivingtaskintheseconditionse.g.increased
concentrationrequiredatjunctionsorreducedperceptionofriskonmotorways.Roadlayoutacrossroad
types(I2)wasalsowidelydiscussedwiththediscussionofcorners,roadturnings,androadvisibilitystated
aselementscontributingtothedecisiontoengage.Driversalsomadeconnectionsbetweentheroadand
theirbehaviouronit(I5),aswellasbetweenthetaskanditsuseinrelationtotheroad(I4).Forexample,the
speedofparticularroadsortheavailabilityofplacestostopwasdiscussedascontributingtotheirdecision
to engage while driving. Furthermore, the behaviour required in the driving task was also reported to
influence their ability to engage in the different secondary taskswhile driving, for example driving on a
motorwaywasdeemedtobeeasierbysomedriverswhichmotivatedtheirengagementinmorecomplex
secondarytasks,thaniftheywereinamorecomplexroadenvironment.Illegality(I3)wasathemethatis
alsomentionedbydriversandisincludedintheroadinfrastructuretheme,asitisinStantonandSalmon
(2009). Interestingly, the lawwasonlyoneof the168other factors thatdriversstated to influence their
decision to engage. This highlights the potential for the development of other techniques to tackle the
numerousothercontributingfactors,asafearofthelawwasonlyasmallcontributiontothedriversself-
reportedcausalfactors.
Thetaskitselfgenerateddiscussiononhowitinfluencedthelikelihoodofthedrivertoengagewithitwhile
driving.Notably,thisincludedhowlongthetaskwouldtaketocomplete(T3),themethodthroughwhich
theycouldinteractwithit(T2),thecomplexityofthisinteraction(T1),itsdesirability(T4),theabilityforthe
task tobecompleted (T6)aswell ashow itsonsetmaybe regulated (T5). The task themehas themost
semantic subthemes,highlighting thenumberof variables relating to the task that influence thedrivers’
engagement with it. It was evident that drivers were aware of the differing complexities and ways of
interactingwiththedifferenttechnologicaltasksandhowtheycouldmanagethesewhiledriving.Therole
12
of the manufacturer in facilitating engagement and the influence of developments in Human Machine
Interface(HMI)designconceptswasparticularlyevidentthroughoutthesediscussions.
Theothersystemiccategorywasthewidercontextwithinwhichdriversdiscussedtheirengagementwith
thetechnologicaltasks.Theydiscussedthetypeofjourney(C1)thattheymaybegoingonthatmayrequire
themtoengagewiththetaskmore.Forexample,theuseofasat-navwasmorelikelyonalongerjourney
whentheydidn’tknowwheretheyweregoingoraphonecallmaybemorelikelyiftheywerecommuting.
The context of the task itself (C2)was also discussedwith the importance of the task to their priorities
referenced frequently. Notably, when discussing phone based tasks drivers reporting who was
communicatingwiththemandtheirperceivedimportanceofthecommunicationtogreatlyinfluencetheir
engagementwiththetask.
2.4Discussion
ThedevelopmentofthethematicframeworkinTable2isthefirstattempttodevelopanextensivelistofthe
drivers self-reported reasons for engagingwith technological tasks. Previous efforts to assess the causal
factorsofdrivererrorhavesuggestedthatthekeyfactorsrelatetothefollowingsystemicelements:road
infrastructure,thevehicle,thedriver,otherroadusersandenvironmentalconditions(Stanton&Salmon,
2009; Salmon et al, 2010). Thus far, the development of error taxonomies has been heavily theoretical,
emergingfromtheaggregationofpreviousliteratureandaccidentreports(e.g.Stanton&Salmon,2009).
Thehierarchicallevelsoftheframeworkthatwereinductivelygeneratedgaveaninsightintothehigherlevel
factorsthatarecloselytiedtothedriversowncommentsanddiscussions.Thehigh-levelfactorssuggestthe
importanceofthedriver,theroadinfrastructure,thetaskandthewidercontextonthedrivers’decisionto
usethetechnologicaltaskswhiledriving.DifferencestothecausalfactorstaxonomypresentedbyStanton
andSalmon, (2009) includeamorespecific focusonthetask in thethematic frameworkrather thanthe
vehicleasawhole.Thisislikelyduetothedesignandaimsofthestudywhichrequireddriverstotalkthrough
theirlikelihoodofengagingwithavarietyofdifferenttechnologicaltasks.Hadtheparticipantsbeenasked
todriveortalkmoreabouttheinteractionbetweencompletingthetaskwhiledriving,morereferencesto
thevehiclemayhaveemerged(Pedic&Ezrakhovich,1999).‘Otherroadusers’isalsoabsentfromthehigh-
level themes of the thematic framework, but does appear within the infrastructure theme under the
‘perceptionsofsurroundingenvironment’semanticsubtheme(seetheOnlineAppendix).Thiscouldsuggest
thatthedrivers’viewofotherroadusersistightlylinkedtoinfrastructureandthesurroundingenvironment
whendecidingtoengagewithtechnologicaltasks.Otherresearchconductedinanaturalisticdrivingstudy
has suggested that other vehicle in front of thedriver donot influencedrivers’ decision to engagewith
technologicaltasks(Tivesten&Dozza,2015).
Whiletherewasanevidentinvolvementofsystemicactorsthatinfluencedengagement,thedriveremerged
asakeysystemicthemeduetothereferencesthatparticipantsmadetotheirattitudes,perceptionsand
13
viewsof engagingwith the technological tasks. This compliments other research that utilised surveys to
identifythatdrivers’intentiontoengageisstronglyinfluencedbytheirattitudetowardsthebehaviourand
theirperceived riskof the task (Welshetal,2008;Zhouetal, 2012).Thevoluntaryaspectofdistraction
(Beanland et al, 2013) and its self-regulatory association (Tivesten & Dozza, 2015) are inherent to the
behaviour, yet this should not be studied independently to the wider context and system within the
behaviouroccurs(Young&Salmon,2015;Parnelletal,2016).
Theroad infrastructurewasdiscussedextensively leading tomultiple themeswithin the framework.This
compliments theresearchconducted inanaturalisticstudybyTivestenandDozza (2015)whofoundthe
drivers’abilitytoanticipatetheroadinfrastructure,suchastightcornersorstraightroads, influencesthe
drivers’engagementwithdistractingtasks.Whiletheysuggestedthatotherroadconditionsdidnotinfluence
thedrivers’engagementastheycouldnotbeanticipated,thefindingsfromthisinterviewstudysuggestthat
road environment and the relationship between the task and the road is discussed as a causal factor in
engagementandthusdriverscananticipatetheeffectismayhaveontheirdrivingperformance,withinthe
interviewsetting.Otherthemesintheframeworkhavealsobeensuggestedintheliteraturesuchastask
context(Lerneretal,2008),taskcapabilities(Zhouetal,2009),journeycontextandtheinfluenceofothers
(Tivesten&Dozza,2015).Yet,theaggregationoffactorsthatwereinductivelygeneratedfromasampleof
driversisnovelandhasstrongtheoreticalapplications.
Theextensiverangeofcausalfactorswithintheframeworkincludesthecontributionof legislationtothe
drivers’decisiontoengage,yetthereareahostofothercontributingfactorsthatsuggestthepotentialfor
other measures through which to tackle the drivers’ engagement with technologies that can lead to
distractionrelatedevents.ThePARRCmodelofdistraction(Parnelletal,2016)highlightstherelevanceof
systemic actors to the causal factors that are attributed to driver distraction in the literature. The
developmentofthethematicframeworkinExperiment1offersthepossibilitytocontrastthecausalfactors
thatdriversreportintheinterviewstudytothosethatarereportedintheliterature.Furthermore,thisoffers
theopportunitytodeterminethefurtheravenuesfordistractionmitigation,asthePARRCmodelhassought
toachieveinthepast(Parnelletal,2016;17).ThisisexploredwithinExperiment2.
3.Experiment23.1.Aim
The inductive thematic analysis conducted in Experiment1allowed the causal factors that influence the
drivers’likelihoodofengagingwithtechnologiestobedirectlylinkedtothedrivers’discussions.ThePARRC
modelofdistractionsoughtcausalfactorsdirectlyfromtheliteratureusinggroundedtheorymethodology
(Parnell et al, 2016). Yet, like the thematic framework developed in Experiment 1, it highlights the
involvementofthewidersociotechnicalsysteminthedevelopmentofdistraction.TheaimofExperiment2
willbetodeterminetherelationofthedriversreportstotheclaimsmadeintheliteraturebyapplyingthe
14
thematicframeworkinTable2tothePARRCmodelofdistractiondevelopedinParnelletal(2016).Thiswill
seektoassessthevalidationofthePARRCmodelthroughtriangulationwith itsapplicationtoalternative
datasources(Hignett,2005).Itwillalsoassessifthereareconceptsthatarereportedbythedriversthat
havenotbeenstudiedintheliterature.Itthereforeseekstopromotefutureresearchaswellasproviding
sociotechnicalsystemsrecommendationstothemitigationofdistraction,ashasbeenachievedwithprevious
applicationsofthePARRCmodel(Parnelletal,2016;2017b).
3.2.Method
The PARRCmodel of distraction (Parnell et al, 2016) was reviewed to assess how the literature-driven
mechanisms relate to causal factors stated by drivers. The process through which this was achieved is
detailedinFigure2.
Figure2.StagesintheapplicationofthethematicframeworktothePARRCmodelofdistraction.
The relationship of the semantic causal factors stated by drivers to the key factors identified from the
literatureinthePARRCmodelwasreviewedthroughdiscussionswithsubjectmatterexpertswithover40
yearsofHumanFactorsexperience(Stage1,Figure2).Therelationshipofthesemanticfactorstosystemic
Stage 1Code the thematic framework to the PARRC model at the semantic level using Nvivo11.
Stage 2Asses the interconnections made between the PARRC mechanisms across the causal framework using a matrix query in Nvivo11.
The semantic themes developed from the thematic coding of the interview transcripts were coded to the key mechanisms of the PARRC model; ‘Goal priority’, ‘Adapt to demand’, ‘Resource constraints’, ‘Behavioural regulation’ and ‘Goal priority’. This was conducted with the discussion of subject matter experts.
Coding of the transcripts to the PARRC model mechanisms with the semantic themes in Nvivo11 allowed a matrix query to be run in the software to calculate the number of references to co-occurring PARRC mechanisms. The frequency of connections made between the mechanisms can be used to infer the strength of the connections which can then be compared to the original PARRC model interconnections that were derived from observations in the literature.
Stage 3Review the content of the transcripts coded to interconnecting themes.
From the matrix queries in Nvivo11 the context of the transcripts that were coded to co-occurring PARRC mechanisms can be reviewed to understand what the interconnections mean to the driver themselves. These can then be contrasted to the interconnections that were identified by researchers in the literature that the PARRC model was grounded within.
DescriptionStageOutput
Expansion of the PARRC mechanisms to include drivers self-reported causal factors
Frequency counts on the number of references in the interview transcripts that make connections between mechanisms of the PARRC model.
An understanding of what the interconnections of the PARRC model relate to in the drivers discussion.
Stage 4Contrast the original PARRC model developed from the literature on driver distraction from technology to the model developed from the drivers self-reported reasons for engaging with technological tasks.
The PARRC model was constructed from the interconnections that were referenced by the drivers, with the frequency of the interconnections visualised through the strength of the interconnections. This was then contrasted to the original PARRC model whose interconnecting mechanisms were constructed from the number references and significant relationships made between the mechanisms in the literature. Comparisons between the drivers stated behaviour and the behaviour studied in the literature.
A version of the PARRC model developed from the literature on driver distraction from technology and a version of a PARRC model developed from the drivers self reports in the interview.
15
actors(stage5oftheframeworkinFigure1)meantthatthecontributionofthesystemicactorstothecausal
mechanismsofthePARRCmodelcouldbeidentified.Furthermore,interconnectionsareimportantwithin
systemsmodelsassociotechnicalsystemsemphasiseemergenceofsafety fromthecomplex interactions
betweensystemicelements(Leveson,2004).TheinterconnectionsintheoriginalPARRCmodelwerederived
fromempiricallytestedconnectionsmadeintheliteratureaswellasassociationsmadebyauthorsinrelating
conceptstooneanother.Connectionsbetweenthecausalfactorsreportedbythedriversintheinterview
studyinExperiment1wereidentifiedusingamatrixqueryinNvivo11softwarethatwasusedtocodethe
data.Matrixqueriesallowthenumberofco-occurringcodedthemestobequantifiedandhighlighted,to
determinethenumberandtypeofdataexcerptsthatrelatetoco-occurringthemesofinterest.Stage3of
theinductivethematicprocess(Figure1),statestheprocessforgeneratingtheinitialdescriptivecodeswithin
thedata.Thisprocessrequiresexcerptstobecodedatmultiplethemes(Braun&Clarke,2006),allowingco-
occurringthemestobereviewedafterwards.ThelinkingofthesemanticsubthemestothePARRCmodel
mechanismsinFigure2,allowedthelinksbetweenthesubthemesrepresentingeachofthePARRCfactors
to be explored. The total number of interconnecting statements in the interviews between the PARRC
mechanism subthemes was calculated. The connections could then be reviewed through the Nvivo11
softwaretofurtheranalysetheconceptsthatwerecodedtotheinterconnectingPARRCmechanisms(Stage
3,Figure2).ComparisonscouldthenbemadebetweentheoriginalPARRCmodel,groundedintheliterature,
and the reconstructedPARRCmodelbuilt from thedrivers self-reporteddiscussionson their reasons for
engagingwithtechnologywhiledriving.
3.3Results
Application of the semantic themes detailed in Table 2 to the PARRC factors and assessment of the
interconnectionsreferencedbythedriversledtotheconstructionofthePARRCmodeldevelopedfromthe
drivers’selfreportedreasonsforengagingwithdistractivetechnologieswhiledriving(Figure2,Stage4).This
ispresentedinFigure3.InsightsthatweregainedfromtheapplicationofthePARRCmodelframeworkto
thethemesidentifiedandinductivelygeneratedfromtheinterviewsarediscussed.
16
Figure3.Applicationofthethematicframework,referencedinTable2,tothePARRCmodelofdistraction.
EachofthePARRCfactorswerefoundtoberepresented inthethematicframeworkdevelopedfromthe
driversself-reportedreasonsforengagingwithdifferenttechnologicaltaskswhiledriving.Therelevanceof
thesethemestothefactorsisdiscussedbelow.
Adapttodemands:Driverssupportedthenotionthattheyadaptboththeirbehaviourinthedrivingtask
andthesecondarytaskinlinewithincreasedmentalandphysicaldemandwhendiscussingtheirlikelihood
ofengagingwithatechnologicaltaskwhiledriving.Thesemanticsubtheme‘Roadcontext’(C3)highlights
theneedtoalterandadapttheirdrivingbehaviourinlinewiththechangingdemandsoftheroad
environment.Thesemanticsubtheme‘Road-taskrelationship’(I4)suggeststhatdriversalsoadapttheir
behaviouracrossdifferentroadtypes,suchthattheyareawareofthedifferentdemandsofdifferentroads
andaltertheirengagementwithtechnologyaccordingly.Intermsoftheadaptionofthetechnologicaltask,
thesubtheme‘abilitytocomplete’(T6)suggeststhatdriversadaptthefunctionalityofthetask,adjusting
thecompletenessofitasillustratedintheexamplequote,inordertomeetthedemandsofthedriving
task.
Behaviouralregulation:Theinterviewsprovidedinformationonthecognitivethoughtprocessesofthe
driverandtheirperceptionsoftheirownbehaviourrelativetothesurroundingenvironment.Ithasshown
thatthedrivers‘attitude’(D1),‘tendencies’(D2),‘viewofself’(D3),andthe‘influenceofothers’(D4)are
keyfactorsthatrelatetotheregulationoftheirbehaviourwithrespecttoengagingintechnologywhilst
Examples:Not worth it
Too dangerousHappily do it
Examples:Drawn in IgnoreTriggered responseExamples:
Eyes off the roadHands on the wheel
Examples:Task TimeLength of text
Examples:Required attentionOther road users
Examples:Poor visibilityCorners
Examples:SpeedDriving task difficulty
Examples:Journey lengthJourney TypeFamiliarity
Examples:UrgencyNecessityIn a hurry
Examples:Do it all the timeDon’t like doing itTry not o
Examples:GuiltyStupidToo clumsy
Examples:Others do it tooSocial peer pressureOthers seeing you
Examples:Know I shouldn’t
Examples:Concentration required from drivingBusyness of road environment
Examples:Driving SlowerWait for right conditions
Examples:Adapt for driving functionalityLimited completion
Examples:Ease of taskCognitive processing required Examples:
UtilityReliabilityUse outside of car
Attitude
Driver (D1)
Engagement regulation
Task (T5)
Task (T6)Ability to complete
Task (T2)Interaction
Road Infrastructure (I5)
Road Related behaviour
Road Infrastructure (I2)
Road layout
Road Infrastructure (I1)Perceived surrounding
environment
Context (C2)
Task Context
Context (C1)
Journey Context
Tendency
Driver (D2)
View of self
Driver (D3)Influence of others
Driver (D4)
Context (C3)
Road Context
Road Infrastructure (I4)
Road-Task relationship
Illegality
Road Infrastructure (I3)Task (T3)
Duration
Task (T1)
Complexity
Desirability
Task (T4)
Adapt to demands
Behavioural Regulation
Goal Priority
Resource constraints
Goal Conflict
17
driving.Furthermore,asisdemonstratedintheexamplequote(Table2,I4),thebehaviourofthedriveris
alsoshowntoberegulatedbytheroadinfrastructurewithinthetask-roadrelationshiptheme(I4),with
driversdiscussinghowtheyregulatetheirbehaviourinrelationtotheroadenvironmentandattainmentof
thedrivinggoal,whichisalteredacrossroadtypes.
Goalconflict:Driversdiscussedthelimitationsofrespondingtoco-occurringdrivingandtechnologicaltask
goalswithrespecttothefeaturesofthetechnologicaltasks(T4&T5)andtheirknowledgeonthelaws(I3)
whichstatetheconflictinggoalsshouldnotbeachievedinunison(i.e.drivingwhileusingahand-held
phoneunderUKlaw).Featuresofthetechnologicaltaskwhichrelatetoitspotentialtoconflictwiththe
drivingtaskwere‘desirability’(T4)and‘engagementregulation’(T5).Technologieswithinthevehiclehave
developedovertimetoprovidenovelinteractionsandfunctionalitiestothedriverthatwerenotpreviously
available.Thismakesthemdesirabletowould-beusers(Walkeretal,2001)andthereforeplacesthemin
conflictforattentionwiththemaindrivingtask.Driversdiscusstheutility(T4)ofthetechnologiesandhow
thisrelatestotheirusewhiledriving,asillustratedintheexamplequote(Table2,T4).Theyalsodiscussthe
featuresofthetaskthatdeterminehowabletheyaretoregulatetheonsetofthetask(T5)andthearising
conflictthismayhavewiththedrivingtask.Forexample,manydriverscommentedonthetriggered
responsethatoccurredwhentheyreceiveatextwhiledriving,e.g.“’Readingatext’,youseeIwouldreada
textjustbecauseofthenatureofthefactthatitflashesuponyourphone.”[Participant:7,Task:Readtext
onmobilephone].Thissuggeststhatdriversdidnotalwayswishtoengageinthetaskbutthedesignofthe
deviceallowedittocompeteforthedrivers’attention,divertingitawayfromthedrivingtask.
Goalpriority:Theprioritisationofthegoalswasfoundtobeinfluencedbythecontextandcircumstance
whichsurroundtheinteraction,notjusttheroadinfrastructurebutthespecificcircumstancesthatmay,or
maynot,leadtointeractionwithtechnology.Thejourneytypeand/orlength(C1)influencedthe
requirementtoengage.Forexample,longerjourneysmayleadtomoreinteractionswiththemusicsystem.
Thefamiliarity(C1)withtheroutewasalsosuggestedtoalterconfidenceinprioritisingthetechnological
task.Thecircumstancessurroundingthetechnologicaltaskwerealsoimportant(C2),suchashow
importantorurgentthetaskwas,ashighlightedintheexamplequote.Thecontextualfactorssuggestthere
aremanysituationsinwhichengagementwithtechnologyismoreorlesslikelytooccur.Thesedonot
relatetoroadtypeorenvironmentbutspecificmomentsandcircumstancesthatcannotbeforeseenand
directlyimpactondriver’swillingnesstoengage.
Resourceconstraints:Theattentionalresourcesofthedriverwerereportedtobeconstrainedbyfeatures
ofthetask(T1-T3)andfeaturesoftheroadenvironment(I1-I2).Theroadtypespresentedtothedrivers
ledthemintoadiscussiononthefeaturesoftheroadthatinfluencestheirdecisiontoengagewith
technology.Discussionsontheroadlayout(I2)andtheirinterpretationofthesurroundingenvironment(I1)
highlightedthedrivers’awarenessfortheelementsoftheroadenvironmentwhichmaylimittheir
18
attention.Likewise,therewasalsoanin-depthdiscussiononthecharacteristicsofthetechnologicaltasks
presentedtodriversandhowfeaturessuchasits‘complexity’(T1),methodof‘interaction’(T2),and
‘duration’(T3)influencedtheattentionthatitrequired.Insomecases,suchastheexamplequote,the
perceivedresourcesrequiredtointeractareminimalwhichincreasesthelikelihoodofengaging,whereas
forothertaskstheperceivedresourcesaretoogreattocompletethetaskwhiledriving.
Interconnections:JustasthemechanismsofthePARRCmodelwerefoundedingroundedtheory
methodology,theinterconnectionsbetweenthemwereidentifiedfromtheempiricallytestedconnections
madeintheliterature(Parnelletal,2016).Thetotalnumberofinterconnectingstatementsinthe
interviewsbetweenthePARRCmechanismsubthemesisshowninFigure4b.Inlinewithstage4inthe
applicationofthePARRCmodeltothethematicframeworkinFigure2,thisiscontrastedtothe
interconnectionsfoundduringthegroundedtheoryapproachintheoriginaldevelopmentofthePARRC
model(Figure4a).ThenumbersontheinterconnectionsinFigure4arelatetothenumberofstudies
empiricallytestingtherelationshipbetweenthefactors.Thesizeoftheconnectinglinesrepresentsthe
numberofconnectionsmadeinbothdiagrams.Thematrixcodingoftheinterviewdataonlystatesthe
frequencyofco-occurringsubthemes,notthedirectionofanyrelationshipthatmayoccur,sothe
connectionsareshownaslinesratherthandirectionalarrowsinFigure4b.
a) b)
Figure4.PARRCmodelscreatedusinga)empiricallytestedrelationshipsintheliterature(takenfromParnelletal,2016)b)interviewtranscripts.
Figure4contraststheinterconnectionsbetweenthePARRCfactorsintheoriginalmodel(Figure4a)and
theonebasedonthedriverscodedtranscripts(Figure4b).Asidefromthequantitiesoftotal
interconnectionsbeinghigherintheinterviewdrivenconnections(Figure4b),whichislikelytobedueto
richdatasourceoftheinterviewsincontrasttothe33studiesidentifiedintheliteraturereview(Parnellet
al,2016),therearesimilaritiesintheconfigurationsoftheinterconnections.Thissuggestssupportforthe
underlyingliteraturethatthePARRCmodelwasgroundedin,astheresearchistargetingconceptsthatare
alsogeneratedtobeimportantbydrivers.
19
Connection Description Quote
Resourceconstraints GoalConflictConstrainedresourcesmeansdriverscannotperformtwotasksatoncesotheyareinconflictwitheachother.
Well,againitisjustIwouldfindittoodistracting,Iwouldn’tbeabletodriveandoperatethesoftwareonmyphoneinordertoenterthedestination,becauseIwouldn’tbeabletoseeproperly!
Resourceconstraints Adapttodemands Limitedresourcesrequiresadaptationofbehaviour.
Sothepicturethatyou’vegothaslotsofcarsonthesideoftheroadforexampleandhousesandI’dbethinking,“Ahacarisgoingtopulloutinfrontofme”or,“I’mgoingtogetveryclosetoacar”sotherefore100%needstobeontheroadatthatpoint.
Resourceconstraints Behaviouralregulation Driversmustregulatebehaviourinlinewiththeirlimitedresources.
Yes,Idon’tthinkthatcanbedoneinasafeamountoftime.ItwouldhavetobelessthanasecondIthinkandeventhen,eventhatonamotorwayisrisky.ButIdon’tthinkIcoulddothatinlessthanabout20seconds,evenifIknewthepostcodeandeverything.SoIjustwouldn’t
GoalConflict GoalPriority Driversprioritisetoovercomeconflict.Soalsodependingonwhoiscalling,soifforexampleworkiscallingI’llprobablyturnthatofforsomenumberthatIdon’tknow,whereasifit’ssomeonethatI’mmorelikelytoknoworsomeonewho’smorelikelytotellmesomethingimportantthenIwillanswermorereadily
GoalConflict Adapttodemands Driverscanadapttothedemandsoftheconflictingtasks.
Iwouldbehighlylikelytodothisonamotorway,becauseIthinkitisquiteeasy,againlikeusingaphoneIwoulddoitstaggeredtouchsomethinglookup,touchsomethingelse,lookup.
GoalConflict BehaviouralregulationTheabilitytoregulatebehaviourtowardsnewtasksthatconflictsthedrivingtaskinfluencestheconflictforattentionbetweenthetasks.
Justbecauseit’smoreofa–forme,forsomereasonit’sanautomaticresponse,sothephoneflashesandyounaturallyjustlookoverandthenyoureadthetext.
Adapttodemands GoalConflict Driversadaptbehaviourwhichcanleadtofurtherconflictbetweengoals.
Yeah,intothecarbutIlistentoplaylistssoit’sneverreally…actually,onthemotorwayit’sfindingasong,ifI’mfindingasong,searchingonSpotifywhereasifIwasonaruralroadoraresidentialarea,I’djustbeskippingasong.
Adapttodemands GoalPriority Driverscanadapttodemandsbyprioritisingtheirgoals.
TheonlytimeImightdoitisatajunction,stopped;there’sachanceImightifIreallyneedtomakeacall,butotherwiseIwouldn’tmoving,no.
Adapttodemands Behaviouralregulation Driverscanregulatetheirbehaviourtoadapttothedemandsoftheenvironmentandthetask.yeah,oksochangingclimatecontrolsIdefinitelywouldonthemotorway,aslongasyouarejustcruisingalong
Goalpriority AdapttodemandsTheprioritisationofgoalsrequiresadaptionoftheprimaryandsecondarytasksinlinewithcurrentdemands.
Ruralroadprobably,itdependshowrural,ifitareallyreallytinyroadIprobablydon’tansweritbecauseitisquitenicetobeabletoheartheroad
Additionalconnections
BehaviouralRegulation GoalConflict
Thedriversattitudetowardsthetaskandstatedintentiontoengagewithitwhiledrivinginfluencethepotentialforthetechnologicaltasktocomeintoconflictwiththedrivingtask
No,never,never,never,never,never.Readatest?Honestly,Iwouldneverdothisstuff.
Resourceconstraints GoalPriority Driverslendresourcestothedrivingtasktodetermineitsprioritybeforedecidingtoengagewithit.Iwouldneveropenupawholemessage,butImightglancedownandlookatwhoitisfromatleastandwhatiswrittenonit.
20
Table3.InterconnectionsbetweenthePARRCfactorsasdrawnfromtheinterviewdata.
Theevidencefortheseconnectionsfoundwithinthedriversself-reportedlikelihoodofengagingwithtechnologies
providessupportforthePARRCmodel,andtheliteraturewithinwhichitwasdeveloped.Yet,somedifferencescanbe
seenbetweenthetwomodelsinFigure4.Thesewereassessedbylookingtowardsthetranscriptsthatwerecodedto
theconnectingthemesinthematrixqueryandunderstandinghowthedrivers’reportsmaydiffertothosestudiedin
theliterature.Theyincludethefollowing:
• Areductionintheprominenceoftheconnectionbetween‘goalconflict’and‘goalpriority’.
InthedevelopmentoftheoriginalPARRCmodelconnections,theliteraturewasfoundtofrequentlyconsidertheneed
toprioritise thegoalsof thedriving taskand thesecondary task inorder to resolveanygoalconflict thatmaybe
occurringbetweenthetwo(e.g.Doganetal,2011).Yet,thetranscriptsofthedriversverbalisedthoughtprocesson
theirlikelihoodtoengageintechnologicaltaskssuggestsprioritisationofgoalstobelessofaconcern.
• Anincreasedprominenceoftheconnectionbetween‘resourceconstraints’and‘adapttodemands’.
Adaption of behaviour tomanage resources was a well represented notion in the interview transcripts, with 72
statements connecting ‘resource constraints’ themes to adapt to demand themes. The connectionwas originally
deemedtoreflecttheideathatadaptionrelatestotheattentionalresourcesthatareavailable(Parnelletal,2016).
ThePARRCmodel alignswith the theory that the attentional resourcesof drivers are limited, such that both the
primaryandsecondarytasksmustcompeteforavailableresourcestofacilitatetheireffectiveperformance(Wickens,
2002).Onewayofensuringthatattentionalresourcesareefficientlyallocatedacrosstheprimaryandsecondarytask
isthroughadaption(Noyetal,1989;Cnossenetal,2000).Figure4bandtheexamplequotegiveninTable3suggest
thatdriversdoreporttheadaptionoftheirbehaviourinordertofacilitatetheirsecondarytaskgoal.Italsosuggests
thatthisadaptiontofacilitatemultiplegoalsismoreimportanttodriversthantheneedtoprioritiseonegoalover
another.
• Theadditionofaconnectionbetween‘resourceconstraints’and‘goalpriority’,whichwasnotfoundinthe
literaturethattheoriginalPARRCmodelwasdevelopedfrom.
AnotherkeydifferencebetweenFigure4aand4bistheadditionofanextraconnectionbetween‘resourceconstraints’
and‘goalpriority’,whichwasnotfoundintheliteraturethatthePARRCmodelwasdrawnfromandisthereforeabsent
intheoriginalmodel(Figure4a).Parnelletal(2016)reasonedthattheabsenceofthisconnectioncouldstemfroma
number of causes which relate to the other connections in themodel. These include the prominent connection
between‘goalpriority’,‘adapttodemands’and‘resourceconstraints’,whichpreventsresourcesfromconstraining
theprioritisedgoal,andfacilitatesadaptionofbehaviourinlinewithresourceavailabilityinordertoprioritiseone
goal over another (Parnell et al, 2016). Yet, the findings from the interview data suggest that drivers domake a
connection between ‘resource constraints’ and ‘goal priority’, although it is minimal. Table 3 gives an example
statement of the connection made by the drivers and suggests the connection relates to drivers lending some
21
attentionalresourcestowardsthetaskinordertodetermineifitrequiresprioritising.Thisisofinterestasitsuggests
a concept that was not previously been considered in the literature from which the original PARRC model was
conceived.
• Anincreasedprominenceofconnectionbetween‘goalconflict’and‘behaviouralregulation’.
Figure 4a suggested the connection between ‘behavioural regulation’ and ‘goal conflict’ reflected the bottom-up
processthroughwhichcertaingoalsmayresultintriggeredresponsesthatrealignstheconflictinggoals(Parnelletal,
2016). The interview data presents a stronger connection between ‘goal conflict’ and ‘behavioural regulation’, in
Figure4b.Areviewofthecodedtranscriptsconnectingthemechanismssuggestitmayreflectaconnectioninthe
oppositedirectiontotheoriginalPARRCmodel,goingfrom‘behaviouralregulation’to‘goalconflict’.Theexample
quotefromTable3highlightstheinfluencethatdrivers’attitudestowardsthetechnologicaltaskscanhaveontheir
potentialtoconflictwiththedrivingtask.Thedriverintheexampleisadamantthattheywillneverusethedevice
while driving, stating theywill turn it off so that itwill not pose a conflict to their driving goal. This illustrates a
reoccurring theme within the interview transcripts, many drivers held strong attitudes towards the use of
technologicaltaskswhiledrivingstatingthattheywouldnevereverattempttoengageinthetaskwhiledriving.
3.4.Discussion
Experiment2haspresentedtheapplicationofthethematicframeworkthatwasdevelopedinExperiment1tothe
PARRCmodelofdistractionwhichpresentsthecausalfactorsinvolvedindriverdistractionasstatedintheliterature.
ThishasvalidatedandextendedtheoriginalPARRCmodelandtheliteraturewithinwhichitiscomprised.Ithasshown
howthedrivers’verbalreportsrelatetothestudyofthebehaviourintheliterature.Driverdistractioncanbeadifficult
behaviour tostudy in itsnaturalenvironmentduetootherconfounding factorsandtheethical issuesofexposing
participantstodistractingactivities(Carstenetal,2013).Therefore,capturingthedriversreportedbehaviourthrough
open-endeddiscussionsandestablishingitsrelationtotheexistingliteratureprovidesmuchneededvalidationofthe
research.
ThecausalfactorsreportedwithintheinterviewsandtheirassociationwiththePARRCmodelfactorssupportsother
researchfromnaturalisticdrivingstudies.Thenotionofdriveradaptionisparticularlyevidentwithstudiesthathave
lookedatdrivers’ engagementwith secondary tasks,withdrivers slowingdown (Metzet al, 2015) and increasing
headway(Tivesten&Dozza,2015)whenengagingwithsecondarytasks.Thissupportstheadaptionofbehaviourat
thecontrollevelthathasalsobeenfoundinsimulatorstudies(Schömig&Metz,2013;Cnossenetal,2004).Yet,there
wasalsoasuggestionthatdriversstrategicallyplantheirengagementwiththetechnologicaltasksinadvance.The
notionthatdrivers’engagementwithtechnologydependsonjourneytypeandroadinfrastructurehavealsobeen
foundinnaturalisticdrivingstudies(Tivesten&Dozza,2015).
Therewere,however,commentsmadebydriversthatwerenotincludedinthedevelopmentoftheoriginalPARRC
model from the literature,which furtherhighlights the importanceofassessing thevalidationofmodelswith the
applicationtodifferentsources(Hignett,2005).Thisincludesanadjustmentwithinthestructureofthemodelformed
22
fromtheinterviewdatawhichsuggestedlessofafocusonprioritisinggoalsdirectlyandmorereportsofadapting
theirlimitedresourcepoolandattributingsomeresourcestodeterminingthepriorityofthesecondarytaskbefore
lendingitfurtherresourcestocarryoutthetask.Thishighlightstheeffectthattaskssuchastextmessageshavewhen
theydrawthedrivers’attentiontowardsthedeviceandleadthedrivertomakeadecisiontoprioritisethedrivingtask
orthetexttask.Ifthenotificationdidnotarisewhilethevehiclewasinmotiontheywouldnotbealertedtoitand
theywouldnotneedtore-establishtheirpriorities.
Conversely,theconnectionbetweenthe‘behaviouralregulation’and‘goalconflict’factorssuggeststhatsomedrivers
wereabletocontrolthegoalsthatconflictedwiththedrivingtaskduetotheirattitudes,perceptionsandbehaviour
that regulated their engagementwith the taskswhile driving. This reflects otherswho suggest the role of driver
attitudeintheirintentiontoengagewithdistractionswhiledriving(Walshetal,2008;Zhouetal,2009;Zhouetal,
2012).The suggestion thatdrivers’engagementwithdistractionsare largelyvoluntary (Lee,2014;Beanlandetal,
2013), suggest that they do have an element of control over their behaviour but that they choose to become
distracted.ThedevelopmentofthethematicframeworkanditsapplicationofthePARRCmodelaimstosuggestthat
thedrivers’decisiontoengageisnotentirelystraightforwardandthatbanningthebehaviourthroughlegislationis
nottheonlyoption.Insteadtherearenumerousfactorsandactorstoconsiderwithinthesociotechnicalsystem,which
arecomplexlyinterconnectedindeterminetheemergenceofdistraction.
4.GeneralDiscussion
It isunderstoodfrompreviousresearchthatthemisuseoftechnologicaldevices,suchassat-navs (Tsimhonietal,
2004),musicplayers(Youngetal,2012),hands-freephones(Horrey&Wickens,2006)mobilephones(McCarttetal,
2006)andwearabletechnologies(Sawyeretal,2014)poseathreattoroadsafety.Whilst,researchfocusesonthe
adverseeffectsoncethetechnologyisengagedwithbythedriver,itfailstoaccountforwhythedriverschooseto
engagewiththetechnology inthefirstplace.Anunderstandingofthekeyunderlyingcausalfactorsthatmotivate
driverstoengagewithtechnologiesiscriticalinprovidingrecommendationsandcountermeasurestolimittheadverse
effectsofdriverdistraction(Walshetal,2008;Atchleyetal,2011;Atchleyetal,2012;Zhouetal,2009;Young&Lenné,
2010).Furthermore,therelevanceofsystemsbasedmeasurestocounteringdriverdistractionarerequiredtoimprove
thesafetyoftheroadtransportsystemasawhole(Young&Salmon,2012;Salmonetal,2012;Parnelletal,2016).
Thesystemiclevelofthethematicframeworksuggestshowactorsoutsideofthedrivers’controlmaybecreatingthe
conditions for distractions to be engaged with. Application of the thematic framework to the PARRC model in
Experiment2suggestedhowtheunderlyingthemesrelatetothePARRCfactors,validatingthemwithconceptsderived
fromdriversreportsoftheirownbehaviour.Furthermore,explorationoftheinterconnectionsbetweenPARRCfactors
hassuggestedsomedifferentstructuralconnectionswithinthemodel.Astheoriginalmodelreflectedtheempirically
testedrelationshipsintheliterature,thedifferenceintheinterconnectionsfoundinthispapersuggestpotentialgaps
intheliterature.Theinductiveanalysishasprovidedfactorsthatdriversthemselvesreporttobeimportantintheir
decisiontoengagewithtechnologicaltaskswhiledriving.Thishassupportedtheagendaofpreviousresearchthathas
appliedbehaviouralintentionliteraturetodriverdistraction(Welshetal,2008;Zhouetal2009;Zhouetal,2012).Yet,
23
it has also highlighted the importance of the interacting elements in the road transport system in creating the
conditionsfordriverstomakethedecisiontoengagewithtechnologicalsecondarytaskswhiledriving.
4.1Recommendationstopractise
The adverse implications of using specific technological tasks while driving are known (Horrey &Wickens, 2006;
Tsimhonietal,2004;McCarttetal,2006;Youngetal,2012),yetthefacilitatingconditionsarelessacknowledged.By
targetingthecausalfactorsofdistraction,countermeasurescanbedevelopedthatfocusontheunderlyingcausesof
theissue,ratherthanlimitingitseffectsonceengagementhasbeeninitiated.Thethematicframeworkdevelopedin
Experiment 1 highlights the importance of the wider road transport sociotechnical system and its influence on
technologyuse,includingthedriver,task,contextandroadinfrastructure.Thissupportstheimportanceoflooking
beyondindividualfocusedmethodsoftargetingdriverdistraction(Young&Salmon,2012;Salmonetal,2012;Parnell
etal,2017).
Todeterminetheimportanceofsystemicactorstoissuesthatarefoundwithinsociotechnicalsystems,thehierarchy
ofthesystemcanbemappedusingtheRMF(Rasmussen,1997).Thisrepresentationofthesystemshierarchyisuseful
indetermininghowactorsinteractwitheachother(Rasmussen,1997),whichcanthenbeusedtoassessthepotential
forincidentaswellasidentifyingfuturesolutions(e.g.Young&Salmon,2012;Parnelletal,2017).Thelocationofkey
actors inthehierarchycanfacilitatetheprovisionofcountermeasuresthattargetelementshigherup,toproduce
widespreadchangeatlowerlevels(Branfordetal,2009).TheadaptedRMFhierarchy(Parnelletal,2017)thatincludes
anadditional twohigh level themes, thenational and international committees, alongside theoriginal levels (the
government,regulators,companyandmanagement;Rasmussen,1997)wasapplied.Theactorsrelevanttoeachof
thesystemiclevelsofthethematicframeworkwereassessedtodeterminetheactorsacrossthehierarchicallevels
thatcouldbetargetedforfuturecountermeasures.
24
Figure4. Systemsactorsacross thehierarchyof the sociotechnical systems thatare related to the fourhigh-level
systemicthemesidentifiedinthethematicanalysis.
4.1.1Driver
Thedrivers’attitudetowardstheuseofthetechnologywasfoundtoplayanimportantroleinlimitingitsconflictwith
thedrivinggoal.Elementsofthesystemthatimpactonthedriverandtheirattitudestowardstechnologicaldevices
appearacross thehierarchy fromthosedirectly interactingwiththedriver,suchaspassengersor thepresenceof
otherroaduserswhomaybewatching,tohigherlevelactorsuchaseducationalproviders,themediaandroadsafety
charitieswhocancontrolattitudesinamoretopdownmanner.ArecentroadsafetycampaignbyTHINK!intheUK
withthetag line‘maketheglovecompartmentthephonecompartment’,guidesthedriverawayfromplacingthe
phonegoal intoconflictwiththedrivinggoal (THINK!,2017).Thiswouldalsopreventthedriverfromdetermining
25
wherethetaskliesintheirgoalprioritiesandhowtheymayadapttheirbehaviouraccordingly.Furthermore,theviews
wehaveontheuseoftechnologywhiledrivingasasocietyareinfluencedbynationalcommitteeswhodeterminethe
importanceofbehaviourwithinnationalculture.Theissueofroadsafetyisasocialresponsibilitythatshouldbeshared
bythetoplevelofthesystem(Larssonetal,2010)andthereforetheuseoftechnologieswhiledrivingneedtobe
portrayedasanti-socialwhenitconflictswiththesafemonitoringofthedrivingtask.
4.1.2Infrastructure
Asan integralpartofthetransportationsystem,road infrastructure isregulatedatthe international,nationaland
governmental levels with the aim to develop an efficient road transport system (e.g. Department for Transport,
2015b).Theinteractionofroadtypewithtechnologicalengagementhasbeenexploredheretoidentifythatdrivers
doconsidertheroadenvironmentwhentheydecidetousetechnologywhiledriving,ashasalsobeenidentifiedin
naturalisticdrivingstudies (Tivesten&Dozza,2015).Onmotorways,descriptivethemessuchas ‘justcruising’and
‘consistent’roadlayoutssuggestthatdriversdeemthedrivingtasktobelessdemandingontheseroadscomparedto
ruralroadswhere‘poorvisibility’and‘corners’mayincreasedemandinthesecondarytask.Yet,attheindustrialand
resource providers level there are no actors directly influencing which tasks are compatible with different road
infrastructures (Salmonetal, 2012). Future research shoulddetermine if certain tasksand technologiesaremore
compatiblewithcertainroadenvironments.Forexample,interactingwithasat-navmaybeeasieronamotorwaybut
ithaslimitedusehereasthereisgenerallymoreroadsideinformationandthedistancebetweenjunctionsisgreater
thanonruralorurbanroads,whichmaybemoredemandingbutholdagreaterrequirementtoengagewiththesat-
nav to navigate through fast changing environments. Tivesten andDozza (2015) also came to similar conclusions
suggestingthepotentialforsomeroadareaswherephoneusemayberegulated,ratherthanbanned.
4.1.3Task
Thetechnological task isassociatedwithanumberofsystemselements fromtheverytopofthehierarchytothe
bottom.Thedesignoftechnologicaltasksisinfluencedbyanumberofguidelines,standardsandcriteriathatstem
frominternationalandnationalactors,whicharethenfeddowntotheindividualmanufacturersanddevelopers.Yet,
thereisalsoaneedtorepresenttheviewsoftheenduserandapplyiterativedesignproceduresthatallowforthe
evaluationofin-builtsystemsusabilitywithrespecttothedriverandthecontextofuse(Harveyetal,2011b).The
driversthatwereinterviewedmadenumerousreferencestothetaskfeatures,suchashowitmaytaketheireyesoff
theroadortheirhandsoffthewheelunderthe‘interaction’subtheme,aswellasreferencingthe‘complexity’and
‘duration’ofthetask.Thissuggestsdrivershadanunderstandingoftheattentionalrequirementsofthetechnological
tasks. Design standards and guidelines have aimed to inform what is achievable while driving, yet they do not
necessarilytakeintoconsiderationthedesirethatdrivershavetousethetechnologyattheenduserlevel.Indeed,
otherresearchpresentedinthisjournalsuggeststhatdifferentdriversneedanddesiredifferentinformationunder
different contexts (Davidsson & Alm, 2014). By facilitating the functioning of the technology in the vehicle, the
temptationforthedrivertoengagewillendure,thisisparticularlytrueofmobilephones(Nelsonetal,2009).The
multi-functionalityofphonesprovidesextratemptationforthedrivertoengagewithitwhiledrivingandthisshould
berespondedtobydevicedevelopersbylimitingfunctionalitywhiledriving.Therewerenumerouscommentsrelating
26
tothenotificationsreceivedonmobilephonewhendriversreceivedtextsorphonecallswhichtriggeraresponse
fromthedriver.Thepresenceoftechnologieswithcapabilitiestotriggeraresponsethattakesthedriversattention
awayfromthedrivingtask,evenmomentarily,shouldberevisedbydevicemanufacturersasitforcestheenduserto
assess theirprioritieswhichshouldpredominately focusonthedriving taskandroadsafety (Leeetal,2008).The
manufacturerApplehastakenstepstowardsthiswiththeirrecentphoneupdate(ios11,releasedSeptember2017)
thatincludesa‘donotdisturbwhiledriving’modethatcansensewhenthedeviceisinamovingvehicleand,once
promptedbytheuser,willturnoffnotifications(Apple,2017).Thereis,therefore,thepotentialfornewregulations
that target the desires and engagement regulation of technology use to stem from the very top levels of the
sociotechnicalsystemandfocusontheinfluencesacrossthelevelsofthesystem,notjustthedriver.
4.1.4Context
The framework highlighted the importance of circumstance in the drivers’ decision to engage. There are many
complexlyinteractingfactorsinfluencingtheuseoftechnologywhiledrivingthatrelatetotheroadenvironment,the
driverandthetaskitself.Theinformationthatdriversrequireanddesireunderdifferentcontextualdemandsislikely
todiffer(Davidsson&Alm,2014;Tivesten&Dozza,2015).Theeffectsofcontextoccurinapredominantlybottomup
fashionwithinthesystemhierarchyastheyaredeterminedbytheinteractionswiththesurroundingenvironmental
conditions.Figure5showsthelackofhighlevelactorsoncontextwithinthesociotechnicalsystemwhichisreflective
ofthelimitedcontroloverthecomplexlyintegratedfactorsthatcompriseindividualcircumstances.Roadconditions
arehardtocontrolastheyareinfluencedbyenvironmentalconditionssuchastimeofday,roadtypeandweather
conditions.Taskconditionsrelatingtotheurgencyornecessitytointeractsuggestthatdriversassesstheirpriorities
astasksandrequirementsarise.Yet,determiningandsettingprioritiesinadvance,orpredefiningsituationswhere
engagementwouldbemoreor lessnecessarycouldlooktocontroltechnologyengagementathigher levels inthe
system.Researchcentresofferingfacilitiestotestdifferentcontextualfactorsthroughtheuseofdrivingsimulators
andhighlycontrolledenvironmentscanofferpromisinginsightsintotheroleofcontextondriverengagement(e.g.
Konstantopoulosetal,2008)andfutureworkshouldassesthecomplexlyinteractingconditionsthatinfluencedrivers
desiretoengage.
4.2EvaluationandFuturework
Theframeworkwasdevelopedfromtheself-reportedbehavioursofasampleof30participants,which issmall in
contrast to the number of participants that can be recruited from online surveys that facilitate far-reaching
recruitmentdatabases.However, thedataobtainedwasmuchricher,withover17hoursofaudio recordingsand
transcribeddata.Furthermore,thesamplestrivedtoincludedriversofanequalrangeofageandgenderstogenerate
a framework based on a representative sample.While the research focused on UK drivers, the laws relating to
technologyuseinthevehiclearesimilaracrossEurope,Australia,NewZealand,JapanandIndiawhospecificallyban
theuseofmobilephones.However,futureworkshouldseektoexplorehowtheframeworkofcausalfactorsmay
alterwithindividualandculturaldifferences,asthismayinfluencetheuseoftechnologybothinsideandoutsideof
thevehicle(e.g.McEvoyetal,2007;Shineretal,2005;McEvoyetal,2006;Young&Lenné,2010;Horberryetal,2006).
27
Anadvantageofthethematicframework is itsgrounding intheself-reportedbehavioursofdrivers,suchthatthe
causalfactorsaredirectlyinformedbythosewhoexperiencethem.Theuseoftheinterviewsettingallowedthedrivers
toopenlydiscussallfactorsthattheyfeltinfluencedtheirlikelihoodofengaging.Whileotherstudieshaveexplored
thedrivers’willingness toengagewithdistractions insimulators (Metzetal,2011;Schömig&Metz,2013)and in
naturalistic driving settings (Metz et al, 2015; Tivesten & Dozza, 2015) they have looked at the drivers’ physical
engagementwiththetaskrelativetothecontextoftheroadenvironment,ratherthanenablingthedriverstodiscuss
theirmotivationstoengage.ThestudyconductedbyLerneretal(2008)facilitateddriverstodiscusstheirwillingness
toengagewithdistractionswithinfocusgroups,yetthismayhavebeeneffectedbythesocialdynamicsandbiases
whendiscussingbehavioursthatbeundesirable(Smithson,2000;Lajunen&Summala,2003).Thedrivers’discussions
inExperiment1wereconductedwithconfidentialityandanonymitywhichencourageddriverstorevealtheirtrue
views.Thisdiscussionalsoallowedforinsightsintothewidersociotechnicalsystemanditsinvolvementintheissue
ofdriverdistractiontoberevealed,aconceptthatwasnotobtainedbytheobjectivemeasurementsinsimulatoror
naturalisticstudies(e.g.Metzetal,2011;Schömig&Metz,2013;Metzetal,2015;Tivesten&Dozza,2015).Whileit
is important to review the drivers’ willingness to engagewithin the context that the behaviour arises, the open
discussionswithdrivers,promptedwithdifferentroadtypesandtechnologicaltasks,hasuncoveredavarietyoffactors
that influence theirwillingness to engage that has not been explored previously. Furthermore, this has provided
validationofpreviousresearch,aswellasproposingthepossibilityoffutureexplorationsofthebehaviourandways
tomitigateagainstit.
5.Conclusions
Thispaperhaspresentedthedevelopmentofathematic frameworkofthecausal factorsthatmotivatedriversto
engagewithtechnologywhiledriving.Theuseofsemi-structuredinterviewsenableddriverstodiscusstheirlikelihood
ofengagingwithavarietyoftechnologicaltasksacrosscommonUKroadtypes.Thestructureoftheinterviewsallowed
allparticipantstodiscussthesametasksandroadtypeswhilefreelygeneratingthekeyconceptsthatwereimportant
to their perceived likelihood of performing the task. Inductive thematic analysis facilitated the development of a
hierarchicalframeworkofcausalfactorsthatwasdrivenfromthedriverowninterpretations,ratherthanapplying
predefinedtheories.Thishasshowntheinfluenceofsystemicactorsonthecausalfactorsinfluencingtechnologyuse
whiledriving,highlightinghowtheroadtransportsystemmaybecreatingtheconditionsfordriverdistractiontooccur.
Thisshouldencourageamovementawayfromindividualfocusedcountermeasuresandtowardstheroleofsystemic
actors,whichhasbeenhighlightedthroughtheapplicationofthethematicframeworktotheinteractingmechanisms
ofdistractionfromthePARRCmodel(Parnelletal,2016).ThethematicanalysishasshowntherelevanceofthePARRC
factors to thesystemicandsemantic themesthatdrivers report to influencetheirengagementwith technological
devices. This has supportedprevious literature in the field,whilst also suggesting additional concepts of interest.
Assessmentoftheactorsimpactingonthesystemicthemesidentifiedfromtheinterviewsacrossthehierarchyofthe
sociotechnicalsystemshashighlightedfutureareasforresearchandcountermeasureimplementation.
28
6.Acknowledgements
This researchwas fundedby theEngineeringandPhysical ScienceResearchCouncil (EPSRC)grantEP/G036896/1,
undertheIndustryDoctoralTrainingCentreinTransportandtheEnvironment.
7.ReferencesAltman,D.G.(1990).PracticalStatisticsforMedicalResearch.CRCpress.Apple(2011).HowtousetheDoNotDisturbwhileDrivingfeature.https://support.apple.com/en-gb/HT208090Atchley,P.,Atwood,S.,&Boulton,A.(2011).Thechoicetotextanddriveinyoungerdrivers:Behaviourmayshapeattitude.AccidentAnalysis&Prevention,43(1),134-142.Atchley,P.,Hadlock,C.,&Lane,S.(2012).Stuckinthe70s:Theroleofsocialnormsindistracteddriving.AccidentAnalysis&Prevention,48,279-284Bayliss,D.(2009).AccidentTrendsbyRoadtype.MotoringTowards2050–RoadsandRealityBackgroundPaperNo.9.March. Royal Automobile Club Foundation. Bayliss, D. (2009).http://www.racfoundation.org/assets/rac_foundation/content/downloadables/roads%20and%20reality%20-%20bayliss%20-%20accident%20trends%20by%20road%20type%20-%20160309%20-%20background%20paper%209.pdfAccessed01/08/2017.Boyatzis,R.E.(1998).Transformingqualitativeinformation:Thematicanalysisandcodedevelopment.Sage.Branford, K., Hopkins, A., & Naikar, N. (2009). Guidelines for AcciMap analysis. In Learning from High ReliabilityOrganisations.CCHAustraliaLtd.Braun,V.,&Clarke,V.(2006).Usingthematicanalysisinpsychology.QualitativeResearchinPsychology,3(2),77-101Carsten,O.,Kircher,K.,&Jamson,S.(2013).Vehicle-basedstudiesofdrivingintherealworld:Thehardtruth?AccidentAnalysis&Prevention,58,162-174.Cnossen,F.,Meijman,T.,&Rothengatter,T.(2004).Adaptivestrategychangesasafunctionoftaskdemands:astudyofcardrivers.Ergonomics,47(2),218-236.Cohen D, Crabtree B. (2006) Qualitative Research Guidelines Project. July http://www.qualres.org/HomeSemi-3629.htmlAccessed22/04/2017.Davidsson,S.,&Alm,H.(2014).Contextadaptabledriverinformation–Or,whatdowhomneedandwantwhen?.AppliedErgonomics,45(4),994-1002.DepartmentforTransport(2015a).ReportedRoadCasualtiesGreatBritain:2015.AnnualReport.https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/568484/rrcgb-2015.pdfAccessed28/04/2017.DepartmentforTransport(2015b).2010to2015GovernmentPolicy:Roadnetworkandtraffic.May.TheStationaryOffice,London.Dingus,T.A.,Klauer,S.G.,Neale,V.L.,Petersen,A.,Lee,S.E.,Sudweeks,J.D.,...&Bucher,C.(2006).The100-carnaturalisticdrivingstudy,PhaseII-resultsofthe100-carfieldexperiment(No.HS-810593).WashingtonUSA.Dogan,E.,Steg,L.,&Delhomme,P.(2011).Theinfluenceofmultiplegoalsondrivingbehaviour:Thecaseofsafety,timesaving,andfuelsaving.AccidentAnalysis&Prevention,43(5),1635-1643.
29
Dorf,R.C.(2001).Technology,humans,andsociety:towardasustainableworld.AcademicPress.Gardner,B.,&Abraham,C.(2007).Whatdrivescaruse?Agroundedtheoryanalysisofcommuters’reasonsfordriving.TransportationResearchPartF:TrafficPsychologyandBehaviour,10(3),187-200.Harvey,C.,Stanton,N.A.,Pickering,C.A.,McDonald,M.,&Zheng,P.(2011a).Totwistorpoke?Amethodforidentifyingusabilityissueswiththerotarycontrollerandtouchscreenforcontrolofin-vehicleinformationsystems.Ergonomics,54(7),609-625.Harvey,C.,Stanton,N.A.,Pickering,C.A.,McDonald,M.,&Zheng,P.(2011b).Ausabilityevaluationtoolkitforin-vehicleinformationsystems(IVISs).AppliedErgonomics,42(4),563-574.Hignett,S.2005.“Qualitativemethodologyinergonomics”.InEvaluationofhumanwork,Editedby:Wilson,J.R.andMegaw,E.TaylorandFrancis.London,UK.Horberry,T.,Anderson,J.,Regan,M.A.,Triggs,T.J.,&Brown,J.(2006).Driverdistraction:Theeffectsofconcurrentin-vehicletasks,roadenvironmentcomplexityandageondrivingperformance.AccidentAnalysis&Prevention,38(1),185-191.Horrey,W.J.,&Wickens,C.D.(2006).Examiningtheimpactofcellphoneconversationsondrivingusingmeta-analytictechniques.Humanfactors:TheJournaloftheHumanFactorsandErgonomicsSociety,48(1),196-205.Horrey,W.J.,Lesch,M.F.,Garabet,A.,Simmons,L.,&Maikala,R.(2017).Distractionandtaskengagement:howinterestingandboringinformationimpactdrivingperformanceandsubjectiveandphysiologicalresponses.AppliedErgonomics,58,342-348.Jamson,A.H.,Westerman,S.J.,Hockey,G.R.J.,&Carsten,O.M.(2004).Speech-basede-mailanddriverbehaviour:Effectsofanin-vehiclemessagesysteminterface.HumanFactors:TheJournaloftheHumanFactorsandErgonomicsSociety,46(4),625-639.Konstantopoulos,P.,Chapman,P.,&Crundall,D.(2010).Driver'svisualattentionasafunctionofdrivingexperienceandvisibility.Usingadrivingsimulatortoexploredrivers’eyemovementsinday,nightandraindriving.AccidentAnalysis&Prevention,42(3),827-834.Lajunen,T.,&Summala,H.(2003).Canwetrustself-reportsofdriving?Effectsofimpressionmanagementondriverbehaviourquestionnaireresponses.TransportationResearchPartF:TrafficPsychologyandBehaviour,6(2),97-107.Larsson,P.,Dekker,S.W.,Tingvall,C.,(2010).Theneedforasystemstheoryapproachtoroadsafety?SafetyScience.48(9),1167–1174Lee,J.D.,Young,K.L.,Regan,M.A.,(2008).Definingdriverdistraction.DriverDistraction:Theory,EffectsandMitigation,pp.31–40.BocaRaton,FL:CRCPress.Lerner,N.,Singer,J.,&Huey,R.(2008).Driverstrategiesforengagingindistractingtasksusingin-vehicletechnologies(No.HS-810919).NationalHighwayTrafficSafetyAdministration.Leveson,N.G.,(2004).Anewaccidentmodelforengineeringsafersystems.SafetyScience.42(4),237–270.Leveson,N.G.,(2011).Applyingsystemsthinkingtoanalyseandlearnfromevents.SafetyScience49(1),55–64.McCartt,A. T.,Hellinga, L.A.,&Bratiman,K.A. (2006).Cell phonesanddriving: reviewof research.Traffic InjuryPrevention,7(2),89-106.McEvoy,S.P.,Stevenson,M.R.,&Woodward,M.(2006).Theimpactofdriverdistractiononroadsafety:resultsfromarepresentativesurveyintwoAustralianstates.InjuryPrevention,12(4),242-247.
30
McEvoy,S.P., Stevenson,M.R.,&Woodward,M. (2007).Theprevalenceof,and factorsassociatedwith, seriouscrashesinvolvingadistractingactivity.AccidentAnalysis&Prevention,39(3),475-482.Metz,B.,Schömig,N.,&Krüger,H.P. (2011).Attentionduringvisualsecondarytasks indriving:Adaptationtothedemandsofthedrivingtask.TransportationResearchPartF:TrafficPsychologyandBehaviour,14(5),369-380.Metz,B.,Landau,A.,&Hargutt,V. (2015).Frequencyandimpactofhands-freetelephoningwhiledriving-Resultsfromnaturalisticdrivingdata.TransportationResearchPartF:TrafficPsychologyandBehaviour,29(0),1-13.Neale,V.L.,Dingus,T.A.,Klauer,S.G.,Sudweeks,J.,&Goodman,M.(2005).Anoverviewofthe100-carnaturalisticstudyandfindings.NationalHighwayTrafficSafetyAdministration,(Paper05-0400).Nelson,E.,Atchley,P.,&Little,T.D.(2009).Theeffectsofperceptionofriskandimportanceofansweringandinitiatingacellularphonecallwhiledriving.AccidentAnalysis&Prevention,41(3),438-444.Noy,Y.I.(1989).Intelligentrouteguidance:willthenewhorsebeasgoodastheold?.InVehicleNavigationandInformationSystemsConference,September.Toronto,Ontario,Canada(pp.49-55).IEEE.O'Cathain,A.,&Thomas,K.J.(2004)."Anyothercomments?"Openquestionsonquestionnaires–abaneorabonustoresearch?BMCMedicalResearchMethodology,4(1),25.Parnell,K.J.,Stanton,N.A.,&Plant,K.L.(2016).Exploringthemechanismsofdistractionfromin-vehicletechnology:ThedevelopmentofthePARRCmodel.SafetyScience,87,25-37.Parnell,K.J.,Stanton,N.A.,&Plant,K.L.(2017).What’sthelawgottodowithit?Legislationregardingin-vehicletechnologyuseanditsimpactondriverdistraction.AccidentAnalysis&Prevention,100,1-14.Patton,M.Q.(1990).Qualitativeevaluationandresearchmethods.SAGEPublications,Inc.Pedic,F.,&Ezrakhovich,A.(1999).Aliteraturereview:thecontentcharacteristicsofeffectiveVMS.Road&TransportResearch,8(2),3.Plant,K.L.,&Stanton,N.A. (2013).What isonyourmind?Usingtheperceptualcyclemodelandcriticaldecisionmethodtounderstandthedecision-makingprocessinthecockpit.Ergonomics,56(8),1232-1250.RAC,(2016).RACReportonMotoring2016.TheRoadtotheFuture.https://www.rac.co.uk/pdfs/report-on-motoring/rac-report-on-motoring-2016-outline.pdfAccessed28/04/2016.Rafferty,L.A.,Stanton,N.A.,&Walker,G.H.(2010).Thefamousfivefactorsinteamwork:acasestudyoffratricide.Ergonomics,53(10),1187-1204.Reimer,B.(2009).Impactofcognitivetaskcomplexityondrivers'visualtunnelling.TransportationResearchRecord:JournaloftheTransportationResearchBoard,(2138),13-19.Richards,L.&Richards,T.(1991).TheTr