1AnalyticMethodsinAccidentResearch:MethodologicalFrontierandFutureDirectionsFredL.ManneringCharlesPankowProfessorofCivilEngineeringSchoolofCivilEngineeringPurdueUniversity550StadiumMallDriveWestLafayette,IN47907-2051Email:flm@ecn.purdue.eduChandraR.BhatAdnanAbou-AyyashCentennialProfessorinTransportationEngineeringDepartmentofCivil,ArchitecturalandEnvironmentalEngineeringTheUniversityofTexasatAustin301E.DeanKeetonSt.StopC1761Austin,Texas78712Email:bhat@mail.utexas.eduSeptember4,20132AbstractTheanalysisofhighway-crashdatahaslongbeenusedasabasisforinfluencinghighwayandvehicledesigns,aswellasdirectingandimplementingawidevarietyofregulatorypoliciesaimedatimprovingsafety.And,overtimetherehasbeenasteadyimprovementinstatisticalmethodologiesthathaveenabledsafetyresearcherstoextractmoreinformationfromcrashdatabasestoguideawidearrayofsafetydesignandpolicyimprovements.Inspiteoftheprogressmadeovertheyears,importantmethodologicalbarriersremaininthestatisticalanalysisofcrashdataandthis,alongwiththeavailabilityofmanynewdatasources,presentsafetyresearcherswithformidablefuturechallenges,butalsoexcitingfutureopportunities.Thispaperprovidesguidanceindefiningthesechallengesandopportunitiesbyfirstreviewingtheevolutionofmethodologicalapplicationsandavailabledatainhighway-accidentresearch.Basedonthisreview,fruitfuldirectionsforfuturemethodologicaldevelopmentsareidentifiedandtherolethatnewdatasourceswillplayindefiningthesedirectionsisdiscussed.Itisshownthatnewmethodologiesthataddresscomplexissuesrelatingtounobservedheterogeneity,endogeneity,riskcompensation,spatialandtemporalcorrelations,andmore,havethepotentialtosignificantlyexpandourunderstandingofthemanyfactorsthataffectthelikelihoodandseverity(intermsofpersonalinjury)ofhighwaycrashes.Thisinturncanleadtomoreeffectivesafetycountermeasuresthatcansubstantiallyreducehighway-relatedinjuriesandfatalities.Keywords:Highwaysafety,crashfrequency,crashseverity,econometricmethods;statisticalmethods;accidentanalysis31.IntroductionWorldwide,morethan1.2millionpeopledieannuallyinhighway-relatedcrashesandasmanyas50millionmoreareinjuredand,by2030,highway-relatedcrashesareprojectedtobethe5thleadingcauseofdeathintheworld(WorldHealthOrganization,2009).Inadditiontothestatisticsondeathandinjuries,highway-relatedcrashesresultinimmeasurablepainandsufferingandmanybillionsofdollarsinmedicalexpensesandlostproductivity.Theenormityoftheimpactofhighwaysafetyonhumansocietieshasresultedinmassiveexpendituresonsafety-relatedcountermeasures,lawsgoverninghighwayuse,andnumerousregulationsconcerningthemanufacturingofhighwayvehicles.Whilethesuccessofmanyoftheseeffortsinreducingthelikelihoodofhighwaycrashesandmitigatingtheirimpactcannotbedenied,thetollthathighwaycrashescontinuetoextractonhumanityisclearlyunacceptable.Criticaltotheguidanceofongoingeffortstoimprovehighwaysafetyisresearchdealingwiththestatisticalanalysisofthecountlessmegabytesofhighway-crashdatathatarecollectedworldwideeveryyear.Thestatisticalanalysisofthesecrashdatahashistoricallybeenusedasabasisfordevelopingroad-safetypoliciesthathavesavedlivesandreducedtheseverityofinjuries.And,whilethequalityofdatahasnotalwaysprogressedasquicklyasmanysafetyresearcherswouldhaveliked,thecontinualadvanceinstatisticalmethodologieshasenabledresearcherstoextractmoreandmoreinformationfromexistingdatasources.Withthissaid,asinmostscientifi...