310 likes | 433 Views
Scientific evidence on road safety effects of section control and red light cameras. Charles Goldenbeld, Institute for Road Safety Research SWOV. This presentation. Institute for Road Safety Research SWOV Scientific evidence section control Separate studies
E N D
Scientificevidence on roadsafetyeffects of section control and red light cameras Charles Goldenbeld, Institutefor Road Safety Research SWOV Dublin, ITS Ireland, 17 September 2013
Thispresentation • Institutefor Road Safety Research SWOV • Scientificevidencesectioncontrol • Separate studies • Soole et al. review 2013 • Dutch experience • Acceptance • Conclusions • Scientificevidence red light cameras • International reviews (Cochrane 2005, Erke 2009, Høye 2013) • Dutch experience • Acceptance • Conclusions Dublin, ITS Ireland, 17 September 2013
What SWOV stands for • Mission: to improve road safety with knowledge from scientific research • Public knowledgefor professionals • Network organization, top institutenationalandinternational Dublin, ITS Ireland, 17 September 2013
What we do • In-depth studies into the causes of accidents • Experimental research • Evaluation studies • Data analysis • Calculating the effects of policy plans • Second opinions Independent, multidisciplinary, high quality Dublin, ITS Ireland, 17 September 2013
Who we workfor • Road safety professionals: • National, regionalandlocalauthorities • Policeandjudicialauthorities • Consultants • Trade andindustry Dublin, ITS Ireland, 17 September 2013
Section control: Terminology • Section speed control • Average speed control • Point-to-point speed control Dublin, ITS Ireland, 17 September 2013
Section control Figurefrom: Soole, Watson, & Fleiter, J.F. (2013). Effects of average speed enforcement on speed compliance and crashes: A review of the literature. Accident Analysis and Prevention, 45, 46-56. Dublin, ITS Ireland, 17 September 2013
Scientific evidence section control Dublin, ITS Ireland, 17 September 2013
Section control separate studies Dublin, ITS Ireland, 17 September 2013
Separate studies Dublin, ITS Ireland, 17 September 2013
Section control: Effects speeds/crashes/cost-benefit estimates. Soole et al. 2013 review • Uitvoering Dublin, 17 September 2013
Section control: comparative performance & cost benefit. Sooleet al. 2013 review Outcomes Dublin, ITS Ireland, 17 September 2013
Dutch Experience:Section control in the Netherlands • Pilot testing 1997 • Introduced 2002 • Financed by funds enabled by the Dutch ‘Climate Bill’ (1998) • Systems distinguishes motorcycles, cars, buses, trucks, lorries • Fines starting from 7 km/hr. above limit (80, 100, 120, 130) • In 2013: 13 section control systems in operation in the Netherlands • 3 section control systems include several sections (only one fine is given!) • A 2002 evaluation of section control (including limit change 100 km/hr. 80km/hr.) on motorway near Rotterdam (140.000 vehicle per day 10% heavy trucks) showed positive effects: • Speed: < 1% offenders of 80 km/hr. limit • Crashes: 47% reduction all crashes, 46% reduction persons injured • Climate: Absolute NO2-concentrations loweredby 4% to6% • Noise: Reduction of daily noise levels by 0,4 dB • In general: 99,5% drivers comply with speed regime under section control Dublin, ITS Ireland, 17 September 2013
Section control acceptance Dublin, ITS Ireland, 17 September 2013
ConclusionsSection control • Preliminary findings are very encouraging (positive & large effects: average speeds, extreme speeds, homogenised speeds, crashes, vehicle emissions, reliability journey time, public acceptance, favourable cost-benefit estimates) • Broad applicability - Motorways - UK also on: Urban arterials - Road works - Tunnels - Mobile systems • Word of Caution: 1.although the available evidence is consistently positive, much of the evidence is lacking in strict scientific rigour (e.g. presence adequate control group, adjustment regression to the mean) 2. section control is complementary to other speed management measures, not a replacement for road design or maintenance deficiencies Dublin, ITS Ireland, 17 September 2013
Red light cameras = RLCs Dublin, ITS Ireland, 17 September 2013
Scientific evidence: major quantitative review studies RLCs Dublin, ITS Ireland, 17 September 2013
Aeron-Thomas & Hess 2005, (Cochrane review) • Two reviewers independently extracted data on study type, characteristics of camera and control areas, and data collection period • 10 controlled before-after studies from Australia, Singapore and the USA met inclusion criteria concerning method quality • Conclusion 1: RLCs are effective in reducing total casualty crashes (based 4/5 studies; only 1 very good). • Conclusion 2: The evidence is not conclusive as to whether RLCs reduce right-angle or rear-end casualty crashes or total crashes (including property damage only crashes) and traffic violations. Dublin, ITS Ireland, 17 September 2013
Meta-analysis Erke (2009) • Meta-analysis = statisticalmethod of combining effect results of several studies toassess overall effect andinfluence of moderating variables • 21 studies (10 USA, 4 Australia, 3 Singapore, 3 UK, 1 Norway) • 5 well-controlled studies Dublin, ITS Ireland, 17 September 2013
Meta-analyses Erke (2009): Takinginto account studyquality Dublin, ITS Ireland, 17 September 2013
Main results Erke 2009 ( blue boxes) Dublin, ITS Ireland, 17 September 2013
Lund et al. 2009 criticismErke 2009 meta-analysis • Basic criticism: Erkedidnot look criticallyenoughto the methodquality of studies she has included in her meta-analysis • Two of the five studies listed as controlling forregressionto the meanand spillover didnot in fact control for these factors • Three ‘well-controlled’ studies included in meta-analysis were non-peer reviewedandreceivedconsiderablestatisticalweights • Among the five studies Erke labels “strong,” the two weaker studies received more statistical weight than the three stronger studies. Doel • Uitvoering Dublin, ITS Ireland, 17 September 2013
Recent studyHøye 2013: new meta-analysis & answertoearliercriticism • The aim of this study was to replicate the results from the study by Erke (2009) based on a larger sample of RLC-studies, and to investigate more thoroughly the effects of study methodology. • A closer look is especially taken at those studies that have been critizedby Lund et al. (2009) and several analyses are performed to test if these or other studies can be regarded as outliers. Dublin, ITS Ireland, 17 September 2013
Meta-analysis Høye 2013 • 19 studies included in Erke 2009 plus 9 more recent studies • Most studies USA (17), Australie (7), UK (3), Norway (1), Canada (1), Singapore (1) Dublin, ITS Ireland, 17 September 2013
Høye3 step approach meta-analysis • Initial meta-regression = tostudyeffectsregression-to-the-mean (RTM) and spillover effects • Overall effects meta-analysis: effect estimatesfor studies withand without control RTM • Exploratory analyses toexplainremainingheterogeneityresults • Moderator variables (.e.g. warningsigns at allintersections vs. generalwarningentrancecities) • Publication bias, outlier bias, bias bystatisticalweighting Dublin, ITS Ireland, 17 September 2013
Høye2013 Mainresults( blue boxes) Doel • Uitvoering • Table 3: Results from meta-analysis of the effects of RLC on numbers of intersection crashes, summary effects and confidence intervals from before-after studies with and without control for regression-to-the mean. Dublin, ITS Ireland, 17 September 2013
Høye 2013 Furtherresults • General warningsignstendtobe more effectivethan separate signs at eachintersection (however: notallresults in support!) • Someevidenceforpublication bias, but direction of the resultsnotchangedbyit • In answertoLund et al. 2009 criticism: no evidencethat meta-analysis resultswereexcessivelyinfluencedbyoutliers or bystatisticalweighting Dublin, ITS Ireland, 17 September 2013
Dutch experience:Red light cameras in the Netherlands • Estimate: > 600 speed/red light cameras • In 2013 fine for red light running: € 220 • Few thoroughevaluations • 2005 study Amersfoort: Positive crash reduction found (15 red light camera intersectionscomparedwith 30 control intersections) • 2011 nationalroad users survey: • 2% drivers report to have been finedfor red light running; • 71% in favour of installing more red light cameras Dublin, ITS Ireland, 17 September 2013
Acceptance red light cameras • Dutch survey 2011: 71% in favour of more red light cameras • European drivers survey 2010: 72% support (SARTRE-4) • Insurance Institutefor Highway Safety Status Report April 2013: 87% support RLC amongthosewho live in areaswith long-standing automatedenforcement • Acceptance even higherwhenpedestrian, cyclist viewpoints taken into account Dublin, ITS Ireland, 17 September 2013
Conclusions red light cameras • Høye2013 meta-analysis: Evidenceforroadsafetyeffects red light cameras is positive but stillless strong than we wouldlikeittobe Best estimate, best studies: -12% injury crashes (not significant) • Best estimate, best studies: -33% right-angleinjurycollisions (significant) Best estimate, best studies: +19% rear-end injurycollisions (significant) • Complexity of intersections: Large inter- and intra-study variation: variables that may be of importance are: the phasing of the traffic lights, the offence rate before the cameras were placed, early warning signs about the cameras, and the cameras' visibility. • Red light cameras are onlyonecandidatemeasureforproblemswith red light running; start with proper, complete problem-analysis • Red light cameras are likelynot the best solution when crashes are relatedtocongestion, inattentivedriving, or long signaldelays Dublin, ITS Ireland, 17 September 2013
Thanksforyou attention! Dublin, ITS Ireland, 17 September 2013