AN EVALUATION OF THE EFFECTIVENESS OF RED-LIGHT CAMERAS AT SIGNALISED INTERSECTIONS Alfred-Shalom Hakkert Transportation Research Institute, Technion, Israel 1. INTRODUCTION Crashes at intersections in general and at signalized intersections amongst these, form a significant part of the safety picture in many countries. In the US, in 1999, some 1,347,000 crashes occurred at traffic signal controlled intersections out of a total of 6,279,000 reported crashes (21 percent) (NHTSA, 2000). Of a total 2,054,000 reported injury crashes in the US in 1999, 476,000 occurred at signalized intersections (23 percent) indicating a slightly higher degree of severity than crashes at other locations. Mohamedshah et. al (2001) indicate an increasing trend in fatal crashes at signalized intersections between 1992 and 1996, increasing by some 18 percent from 1,888 in 1992. Retting et. al (1998) reported that annually some 260,000 red-light-running (RLR) crashes occurred in the US of which 750 crashes are fatal. The above numbers indicate that crashes at signalized intersections and, in particular, RLR crashes are worthy of attention and counter-measures ought to be sought to alleviate this problem. One of the counter-measures generally considered is police enforcement and in the case of red-light running, automatic enforcement with the aid of red-light cameras. The generally held view on the effectiveness of enforcement is that a combination of factors should be looked at. The deterrence effect is the combined result of enforcement and punishment as manifested in the subjective risk of apprehension. In risk of apprehension, it is certainty that produces the greatest effect. Certainty is a question of type of enforcement and intensity. In punishment, the most important factor seems to be swiftness (Shinar and McKnight, 1984). The following paper will bring together literature that has been published on the subject of RLR crashes. It will briefly deal with driver behaviour at signalized intersections as manifested in violation rates, i.e. the number of drivers passing against the red light and with the characteristics of the intersections associated with RLR, but most of the attention will be focused on the crash data at signalized intersections in the context of the installation of red-light cameras. 2. BRIEF OVERVIEW OF HISTORIC DEVELOPMENT The use of automatic cameras at signalized intersections started in Europe in the early l970s, in Australia on a wide scale in the 1980’s and in the US in the early © Association for European Transport 2002 1990’s. Levinson (1989) reports on the use of red-light cameras in Israel as early as 1969. The cameras are generally activated a fraction of a second up to one second after the onset of the red signal. A vehicle crossing on red is captured twice, with a one-second interval between photographs to document its progress. Video cameras, which have been introduced recently, work on a different principle and record a sequence of frames of the violating vehicle as it passes through the intersection. New York city was the major jurisdiction in the US that started applying RLCs on a large scale. Initially 18 intersections were equipped with cameras. Other jurisdictions followed-Howard County, MD and San Francisco, CA in 1996, Oxnard, CA in 1997 and others somewhat later. To date, about 70 communities in the United States have implemented RLC enforcement (IIHS, 2001). Automatic camera enforcement is attractive to many police forces. There is the potential for a police enforcement agent to be present at a spot for 24 hours every day, it could be unobtrusive (if that is the preferred tactic) and it produces a record of evidence that can not be easily disputed. Furthermore, it can generate a large number of citations thus influencing drivers in an efficient and costeffective manner. On the other hand, this type of enforcement created new issues and challenges- legal, technical and operational. The simple chain of evidence in a manual citation has the observing police officer identify an offending driver and hand her or him a citation, in person. With automatic cameras, there is no eyewitness and the identity of the offending driver cannot be ascertained at the time of the offence. The evidence has to be fool proof and the identity of the offender must be somehow extracted from the photographic record. Processing large number of film (or video) records, guarantying tamperproof handling, maintaining equipment, all present special challenges. In all cases, the legal system has to be adapted to accept this type of enforcement. Great Britain is one of the countries with widespread use of automatic enforcement of speeding offences and of red-light running. About 34 percent of all speeding tickets are issued by the use of speed cameras and 22 percent of all junction related offences are from RLCs. Out of 313,000 tickets issued in 1997 in England and Wales, 88 percent were for speeding and 12 percent for running red lights (Zaidel, 2000) 2.1 Automated systems The current RLC systems use wet film, digital film and video to record the violation. Although sometimes called automatic enforcement, these systems are by no means automatic. The loading and unloading of the film, the transfer of the films for processing and parts of the ensuing processing and ticketing systems are done by hand. A truly automatic system would enable the process to run from the stage of recording the violation till the issuing of the ticket without human intervention, keeping in mind, that safeguards would have to be built into such a © Association for European Transport 2002 system for a human operator to intervene in all cases where the slightest doubt arises. Some applications have been developed and tested that approach such an automated system. A Dutch company developed a digital system for speed measurement along a stretch of road that incorporates video cameras, a licenseplate reader enabling automatic vehicle identification of violators, transmission of the pictures to a central processing location and automatic issuing of tickets. After an initial trial period, the system has yet to be put into wider operational use. 3. DRIVER VIOLATIONS AT TRAFFIC SIGNALS AND BEHAVIOURAL ASPECTS A large number of studies have dealt with the aspect of drivers traveling through the red-light at signalized intersections (Retting et. al, 1999a; Retting et. al, 1999b; Chin, 1989; Baguley, 1988; Greene, 2000). Some studies were mostly concerned with the number and frequency of red-light violations while other studies have also dealt with the mechanism of red-light violations, i.e. the time interval since the onset of the red till the violation, the vehicle manoeuvre, characteristics of the drivers, etc. Some of these aspects will be summarized and described in the following sections. Baguley (1988) was one of the first to study the red light running phenomenon in detail. Drivers approaching a traffic signal are in some cases caught in what is termed a dilemma-zone. The driver is said to be in the dilemma-zone if the amber period (a standard three seconds in the U.K) is insufficient for him either to stop comfortably, or to clear the stop-line before the red-phase begins. In this situation the driver has the possibility to either brake hard with a possibility of coming to a halt after the stop-line, or to continue and cross the stop-line after the onset of the red. A partial, engineering solution is provided for these cases by providing an all-red phase to the signal, which delays conflicting traffic from entering the intersection. Studying the characteristics of drivers entering the intersection after the onset of the red light Baguley divided such drivers into three groups: drivers who were caught in the dilemma zone, drivers who should have been able to stop comfortably, but deliberately continued into the red and those who appeared completely unaware that they continued through a red signal. Various traffic engineering measures can be selected to improve safety in the case of drivers from the first group. Such measures include green phase extensions for drivers approaching at high speeds, all-red clearance intervals and measures to reduce speeds on traffic signal approaches. For drivers of the other two groups, red-light camera enforcement offers an effective solution. 3.1 Studies of violations at signalized intersections This section discusses the number and proportion of drivers that violate the red light according to the various studies that dealt with the subject. Table 1 summarises the results from ten various studies. © Association for European Transport 2002 Table 1 Summary table with results of studies with red-light violations at signalized intersections Reference Country Description of study Retting et. al (1999a) USA 5 camera sites 2 noncamera sites 2 comparison sites Retting et. al (1999b) Chin (1989) Woolley and Taylor (1998) USA Singapor e Australia Number of violations Before After 192 334 Great Britain Percent change Comments -44 Rateviolations 4 per 10 vehicles Violationsentering 0.4 seconds or more after red onset Rateviolations 4 per 10 vehicles Violationsentering 0.4 seconds or more after red onset Rateviolations per cycle 188 135 37.8 25.0 -34 51 53 7.6 7.7 +5 9 camera sites 3 noncamera sites 2 comparison sites 719 98 455 48 12.9 16.0 7.7 8.0 -40 -50 22 19 7.0 6.7 -4 23 camera approaches 20 noncamera approaches 14 comparison sites 12 sites N/a N/a 0.344 0.201 -42 0.326 0.239 -27 0.283 0.331 +17 Range 0.180.84 Range 0.390.83 at pedestrian crossing 0.33, 0.39 ranked 3 and 4 out of 12 Range5.7 – 33.4 percent of free flowing vehicles N/a Rateviolations per 100 passing vehicles N/a Free flowing vehicles is 2.3 – 3.2 % of total major road flow Total- 298,049 vehicles were counted 2 sites with cameras Baguley (1988) Rate of violations Before After 20.4 36.3 7 rural sites 3830 free flowing vehicles 670 red-light runners © Association for European Transport 2002 Kent et. al (1995) Australia 3 RLC in Melbourne and Geelong Metropolitan areas 38000 vehicle movements observed 123 encroachments 0.32 percent encroachments N/a Greene (2000) Australia 15 sites 133,238 vehicles observed 522 red-light incidents N/a Lawson (1991) Great Britain 14 sites in Birmingham N/a 2.4 incidents per hour 3.9 incidents per 1000 vehicles 1 incident per 25 cycles 1% of vehicles crossed against the red. At one site- 6% Oei et. al (1997) The Netherla nds 4 sites in Amsterdam 3000 1308 3.4% 1.5% -56 Thompson et. al (1989) Great Britain Nottingham Site 1 Site 2 60 149 44 168 1.4% 3.1% 1.1% 3.5% -22 +13 © Association for European Transport 2002 N/a 93% of encroachm ents occurred during the all-red phase Rateincidents as proportion of all passing vehicles Percent of vehicles in the decisionarea that had opportunity to violate Rate – percent of entering vehicles Violations range 1.6%-8.2% before to 0.5%-3.4% after Violations as percent of vehicles arriving and stopping on red The various authors have adopted different rates to describe the violation patterns. Retting et. al (1999a, 1999b) define the rate of violation as the number of violations per 10,000 entering vehicles. Chin (1989) uses the number of violations per signal cycle, whereas most other studies, including Woolley and Taylor (1998), Green (2000), Kent et. al (1995), Lawson (1991) and Oei et. al (1997) use violations per hundred passing vehicles. Baguley measures violations as a proportion of free-flowing vehicles at a distance up to 150 metres from the stop line, not crossing during amber. The proportions of violations vary largely between studies and countries, but a number of generalizations can be made. The studies from the US, Great Britain and the Netherlands report violation rates of between 0.5 to 7 percent of entering vehicles. The studies from Australia generally report lower violation rates of below one percent. This is presumably related to driving and enforcement practices in the various countries. Australia has a long tradition of red-light camera enforcement, which has been accompanied by high-profile publicity. Those studies that measured violation rates before and after the installation of RLC invariably found large reductions in violation rates at camera sites or on camera approaches, in the order of 40 to 50 percent (Except an early study at two sites in Nottingham - Thompson et. al, 1989). Rocchi and Hemsing (1997) reviewing a number of studies most of which are also covered in the present study, report on reductions in violations ranging between 0 and 75 percent. With publicity the reductions in violations range from 21 to 75 percent. The Nottingham study with a zero reduction in violations shows that it is important to advertise the introduction of a camera program. Large reductions were also found at sites near camera locations or near camera approaches, in the order of 30 to 50 percent. This occurred at a time when violation rates at sites not related to the camera sites, i.e. control sites, generally went up (Retting et al, 1999a and Chin) or decreased slightly (Retting et al, 1999b). Taking changes in violation rates as an indication of possible changes in crash rates, this seems to indicate that studies where crash rates at camera approaches are compared with those at other approaches of the same intersection, or with rates at nearby intersections are seriously underestimating the RLC effect on both violation rates and on crashes. The reason is that drivers are adjusting their behaviour in a more general way and not only on the camera approaches, a “halo” effect. One other aspect that should be considered is that a red light violation is defined differently from country to country. In many US jurisdictions, violations are defined as a passing of the stop line 0.4 seconds or more after the onset of the red light. An additional requirement is that the vehicle travels at a speed of 15 miles or more, so that slow moving right turning vehicles (the right turn on redafter stop manoeuvre, where permitted) are not captured. In Maryland a violation is recorded from 0.1 second after red onset. In South Australia (Earl and Coxon, 1996), a violation starts only 1 second after the onset of red. A shorter delay after red onset will invariably increase the number of violations. © Association for European Transport 2002 4. CRASH STUDIES 4.1 Aggregate studies of intersection crashes and crashes related to redlight running Intersection crashes form a large part of the crash population and a significant proportion of such crashes can be associated with red light running at signalized intersections. Table 2 shows that intersection and intersection-related crashes in the US make up 45 percent of all police reportable crashes in 1999, 49 percent of all injury producing crashes and 23 percent of the fatal crashes. Crashes at signalized intersections make up 20, 23 and 7 percent respectively. Table 2. Data on crashes in the US by type of location for the year 1999 (NHTSA, 2000) All levels of severity number percentage All crashes 6,279,000 100 Intersection 2,806,000 45 crashes Signalised 1,278,000 20 intersection crashes Injury crashes Fatal crashes number percentage number percentage 2,054,000 100 37,043 100 1,015,000 49 8,514 23 476,000 23 2,691 7 Table 3 summarises the results from the studies conducted in the US, Great Britain and Australia, which have looked at details of crashes at signalized intersections. The studies conducted in the US show a range of 5 to 9 percent of crashes at signalized intersections associated with RLR. Within the US the various results are very consistent. The Australian study indicates 15 to 19 percent of crashes can be RLR related. The study from Great Britain identifies 25 to 36 percent of crashes as RLR related. Some of these differences, which are very large, might be associated with behavioural differences between countries and differences in setting inter-green times at signalised intersections. Other differences in police coding procedures might also have contributed to these differences. 4.2 Before-after evaluations of red-light camera installation Considering the effect of camera installation on the number of red-light violations, the reasonable assumption being that, in many cases, it is the red-light violation that precedes a crash. However, it cannot be said that red-light violations in most cases lead to a crash, especially those that occur during the first two seconds of the red signal, part of which may be part of an all-red phase. It is therefore not enough to study the changes in rates and patterns of violations and one has to look also at changes in crashes that occur after the installation of cameras. © Association for European Transport 2002 Table 3. Studies with details of crashes at signalized intersections Source Year Country Retting et al 2000 USA Retting et al 1999 c USA Lawson Greene Mohamedshah, Chen, Council 1991 2000 2001 Great Britain Australia USA Number of intersections Total - 122 51experimental 56 comparison Large nationwide data bases, five years data 1992-1996 29 intersections In Birmingham with RLR problems 5 years 168 signalised intersections in Birmingham, one year’s data 1989 5 years data 1985-1990 in Birmingham Years 19941998 Victoria, Western Australia, Queensland 1756 signalised urban intersections © Association for European Transport 2002 Total number of crashes 1,044experimental Before 941experimental After 1323 comparison Before 1298 comparison After 6,833,669 policereported crashes Number of RLR crashes N/a Percentage 3753 RLR fatal crashes 257,849 RLR police-reported crashes (of which 208,355 are injury producing) 276 accidents N/a 7 fatal 68 serious injury 343 slight injury 418 total 2018 accidents N/a N/a 508 RLR accidents 25.2 percent Crashes at signalized intersections Relevant crashes to RLR - 2688 crashes 1479 756 accidents with injuries 53,247 crashes 8 percent reduction at experiment al sites after signal re-timing (not significant) 5 percent 36.5 percent 19 percent 15 percent 1589 21 percent 4709 crashes 8.8 percent 4.2.1 Description of studies with information on crashes and their results at intersections with cameras Overall, 12 studies and reviews were identified, from the USA (2 studies), Australia (6), Great Britain (2), Singapore (1) and Norway (1 - review), that had quantitative results on the number of crashes at signalized intersections before and after the installation of red-light cameras. The main results of these studies and reviews are summarized in table 4. The changes in the numbers of crashes shown in table 4 are the numbers quoted by the authors, which are the result of the methods they used in their studies. As will be shown in the joint analysis, in some cases different results are obtained by using a different statistical technique. Looking at the results it can be concluded that most of the studies reported crash reductions as a result of the installation of red-light cameras at signalized intersections. Some of the studies pointed to the fact that the reduction in crashes was not found to be statistically significant. Some of the studies noted a reduction in right-angle crashes and an increase in rear-end crashes (Andreassen, 1995; South et. al, 1988; Hillier et. al, 1993; Ng et. al, 1997). But other studies, notably the studies by Office of Auditor General (1996) and Mann et. al (1994) did not find such increases. Other studies did not disaggregate the crash results by type and provide results of overall changes. Some studies differentiated between crashes with injuries and all reportable crashes, including property-damage-only accidents. The range of reported crash reductions is wide and varies from a reduction of 7 percent in all crashes (Retting and Kyrychenko, 2001) to 18 percent (Hooke et.al, 1996) and 41 percent in property damage crashes in Queensland (1995); from a 6.7 percent reduction in crashes with injuries (South et.al, 1988), to 16-60 percent in Queensland (1995) and 29 percent in (Retting and Kyrychenko, 2001). Generally, reductions in right-angle crashes were larger and in some cases the number of rear-end crashes increased. The more severe crashes were reduced more than the non-injury crashes. Rocchi and Hemsing (1997) report on 12 studies from Australia, the United States and the United Kingdom. They found reductions ranging from 10 to 50 percent in the number of angle collisions. 4.2.2 Joint analysis To quantify the results of the various studies in a more meaningful way a joint analysis was conducted for all those studies that provided the required details. The joint analysis weighted the results obtained from the different studies using Odds ratio evaluations. Each study is weighted according to the relative number of crashes it contributes to the total. To calculate the Odds-ratio, the number of crashes in the before and after periods is needed, for both the © Association for European Transport 2002 Table 4. Summary and results from 12 studies with details of crashes at sites with cameras Number of intersection s with cameras 11 Average crash reduction Incl. in metaanalysis 7%-all crashes, 29%-injury crashes Yes 46 10.4%-all casualties, 6.7%-all crashes with injury 32%-right angle crashes 50%-in right angle crashes, 25-60% increase in rear-end crashes Changes in numbers of crashes were not found statistically significant Changes in number of crashes were not found statistically significant 40% in right-angle crashes, little change in rear-end crashes Yes Referen ce Year Town, state country Details of study Retting and Kyryche nko 2001 Oxnard, USA South et. al 1988 Victoria, Australia Number of crashes and injury crashes at camera sites and non-camera sites in treatment community and comparison communities Number of crashes with injuries, by type of crash at camera and control sites Hillier, et. al 1993 NSW, Australia Number of crashes and crashes with injuries by crash type at camera and control sites 20 Andreas sen 1995 Victoria, Australia Number of crashes, incl. with injury, by type of crash at camera approaches and non-camera approaches 41 Mann et. al 1994 South Australia Number of crashes and crashes with injury at camera and control sites 8 Office of Auditor general 1996 Western Australia Number of crashes per intersection, by type at camera and non-camera intersections 44 © Association for European Transport 2002 Yes Yes Yes No Review of studies in various US communities, incl. Oxnard Rates of crashes with injuries at sites with red-light cameras before and after camera installation Literature review and meta-analysis of three Small Scandinavian studies Varying numbers Varying results No 113 18% in crash rates per intersection No N/a No Queenslan d Australia Crashes and crashes with injuries at signalized intersections with and without cameras 79 1990 Great Britain Crashes with injuries before and after camera installation by type of crash 5 1997 Singapore Rates of crashes with injuries before and after camera installation, at camera sites and at comparison sites All camera sites-125 Beforeafter-42 sites 45%-fatal crashes 12%-injury crashes 9%property damage crashes Over 50% in fatal crashes 16-60% in crashes with injuries 41% in property damage crashes 10%-all crashes 68%-“fail to conform” crashes 16%-in crashes with casualties at 125 sites 19% at 42 sites Hasson 2000 USA Hooke, et. al 1996 Great Britain Elvik et. al 1997 Various countries Queensl and 1995 County Surveyo rs Ng et. al © Association for European Transport 2002 Yes No Yes treatment and valid comparison (control) sites. Of the thirteen studies described above, only seven studies provided detailed results, which enabled to include them in the joint analysis of the effect of camera installation on the crash frequencies at signalized intersections. These were the studies by Retting and Kyrychenko (2001), South et al (1988), Hillier et. al (1993), Andreassen (1995), Mann et. al (1994), Queensland (1995) and Ng et. al (1997). The first study relates to the USA, the last one is from Singapore and all the others relate to results from Australian studies. Results (Table 5) were attained first for studies that included both crashes with casualties and crashes with property damage above a certain monetary value, these groups combined are termed reportable crashes. A further calculation was performed on those studies that provided separate results for crashes with injuries. The injury crashes were further split up into right-angle crashes, which are those that might most likely be affected by the camera installation and rearend crashes which, according to some of the reports might be adversely affected. Only some of the seven studies provided further details about the types of crashes involved. These were the studies by South et. al (1988), Hillier et. al (1993) and Mann et.al (1994). Right-angle crashes include collisions between vehicles on adjacent approaches and other right-angle crashes. Rear-end collisions include crashes between vehicles on the same approach and other rear-end crashes. Table 5. Results from the joint analysis with crash reductions and statistical significance Type of analysis Total crashes Injury crashes Right-angle injury Rear-end injury Weighted odds- Change in ratio crashes 0.796 20% reduction 0.730 27% reduction 0.836 16% reduction Statistical significance yes yes yes(just) 1.096 not significant 10% increase No. of studies 5 6 3 3 The definition of sites included in each study differ somewhat. Retting and Kyrychenko (2001) include signalized intersections in the treatment group and use two types of comparison sites. One group is signalized intersections in comparison communities, whereas the other type uses non-signalized intersections in the treatment community. The study by South et. al (1988) uses slightly differing before and after periods, defined in terms of intersection-years, i.e. the number of intersections times the number of years in each period. Because the differences were less than 3 percent this effect was ignored. The study by Andreassen (1995) used intersection approaches on which a camera © Association for European Transport 2002 was installed as the treatment sites and other approaches at the same intersections as comparison sites. It has by now been well documented that when cameras are installed, the number of red-light violations decreases at both camera sites and at sites nearby, either other approaches of the same intersection or nearby intersections. The real effect of cameras is thus underestimated. The study by Mann et. al (1994) looked at eight sites at which cameras were installed and at another five sites (termed RLC modified) at which also various engineering improvements were made. The joint analysis included only the eight RLC sites without modifications. In the study from Queensland (Queensland, 1995) the cameras were installed over a period of five years 19911995. These years were excluded from the analysis making the years 1986-1990 the before period and the years 1996-2000 the after period. Table 5 shows that the studies reviewed demonstrate an overall reduction of 20 percent in crashes as the result of the installation of red-light cameras. The overall reduction in the number of injury crashes is estimated at 27 percent. A reduction of 16% is noted in the number of right-angle crashes and a notsignificant increase of 10% is observed in the number of rear-end-crashes. Looking at the contribution of each study to the weighted overall results, it was noted that the study from Queensland, being much larger than the other studies, contributed most to the overall weights and also had the largest reduction in crashes. The results were therefore calculated a second time without the Queensland data. Results are presented in Table 6. The Queensland study did not provide details of the types of crashes at camera and non-camera locations Table 6. Results from the joint analysis without the Queensland study Type of analysis Total crashes Injury crashes Weighted odds-ratio 0.926 0.832 Change in crashes 7% reduction 17% reduction Statistical significance yes yes Table 6 shows that the overall results remain the same, although the effect is somewhat reduced. The overall conclusion of the combined analysis of the effectiveness of the camera installation in the seven studies included is that the installation of the cameras proved effective in all studies reviewed. The overall effectiveness can be estimated as a 20 percent reduction in crashes of all severities, a 27 percent reduction in crashes with injuries and a 16 percent reduction in right-angle crashes. The increase in rear-end crashes of 10 percent, although not statistically significant was noted in four of the seven studies. © Association for European Transport 2002 5. DISCUSSION AND CONCLUSIONS The ultimate goal of police enforcement is to achieve normative changes in the public's behaviour to come into closer agreement with the laws and regulations. Only in this manner is it likely that enforcement can be effective, in the long run, within reasonable bounds of manpower and resources. Literature dealing with the effectiveness of conventional enforcement seems to indicate that such normative changes are unlikely to be achieved with the levels and methods of enforcement presently used in most countries (Makinen and Oei, 1992; ETSC, 1999). Whereas some changes, sometimes drastic, can be achieved locally in road user behaviour, these changes are generally limited in time and distance, and do not produce lasting impacts, neither on behaviour nor on the number of road accidents associated with these types of behaviour. The explanation seems to lie in the fact that normative changes are not achieved because the changes in behaviour are not internalized. It is believed that a considerable increase in the risk of apprehension is required in order to achieve the desired effect. Such an effect might be achieved with the aid of a variety of methods of automatic enforcement. This type of enforcement could be applied on a much wider scale. The technologies have been developed to make this kind of enforcement possible and feasible. In many countries, the laws and regulations have been adapted to make automatic enforcement possible, and public opinion, although not always very favorable, seems to agree with a gradual introduction of this type of enforcement. Various traffic engineering measures can be selected to improve safety in the case of drivers who enter the intersection during the first 1-2 seconds of red. Such measures include green phase extensions for drivers approaching at high speeds, all-red clearance intervals and measures to reduce speeds on traffic signal approaches. For drivers who enter the intersection deliberately, red-light camera enforcement offers an effective solution. As has been demonstrated in this study, red light cameras are highly effective in reducing the number of drivers entering an intersection on red. Reductions in the order of 40- 50 percent were found in those studies reviewed that observed violation rates before and after the installation of red light cameras. Not unexpectedly, cameras were also effective in reducing crashes. Overall reductions of 20 percent in all crashes (based on five studies) and of 27 percent in crashes with injuries (based on six studies) are the best estimates on the basis of those studies that provided detailed information on the number of crashes. Other studies that provided information only on changes in crash rates show a wider range of crash reductions but no study indicated an increase in crashes. © Association for European Transport 2002 One of the criticisms raised in the context of red-light enforcement programs concerns the outsourcing of the process to private companies. This criticism is mostly relevant to US practices and not to other countries where outsourcing is not common. Although the outsourcing is done for matters of efficiency and is always conducted under strict instructions and supervision of the responsible jurisdiction, some of the opposition could be alleviated by activating the camera system only after one second of red and not, as is now customary in most US jurisdictions, 0.3-0.4 seconds after red onset. It was shown in various studies that the safety problem created by drivers entering the intersection during the first 1-2 seconds of red can be treated with various traffic engineering solutions. 6. References Andreassen D.(1995) A long term study of red light camera and accidents, ARRB report ARR 261, Australian Road Research Board, Victoria, Australia. Baguley. C. (1988) Running the red at signals on high-speed roads, Traffic Engineering and Control, 29, 7/8. Chin H.C. (1989) Effect of automatic red-light cameras on red-running, Traffic Engineering and Control, April. Country Surveyors’ Society (1990) Traffic signal cameras, Report no. 1/12, County Surveyors’ Society, Special Activity Group on Accident Reduction, County Surveyors’ Society, England. Earl R., Coxon C. (1996) Red-light cameras in South-Australia - A review of the current situation and recommended future directions, Report Series 5/96, Office of road safety, South Australian Department of Transport, Australia Elvik, R., Borger-Mysen, A. and Vaa, T. (1997) Trafikksikkerhekshandbok (Traffic Safety Handbook), Institute of Transport Economics, Oslo, Norway. ETSC (1999) Police enforcement strategies to reduce traffic casualties in Europe, European Transport Safety Council, Brussels, Belgium. Greene F. (2000) Red light running, in Proceedings of the Conference on Road safety - Research, policing and Education, Brisbane, Queensland, Australia. Hasson P. (2001) Red-light cameras, FHWA Internet site http://safety.fhwa.dot.gov/fourthlevel/pro_res_srlr_camera.htm Hillier W., Ronczka J., Schnerring F. (1993) An evaluation of red light cameras in Sydney, Research Note RN 1/93, Road Safety Bureau, Roads and Traffic Authority, New South Wales, Australia. © Association for European Transport 2002 Hooke A., Knox J., Portas D. (1996) Cost benefit analysis of traffic light and speed cameras, Police research series paper 20, Police Research group, Home Office, London, England. IIHS (2001) Website www.highwaysafety.org/safety facts/red light cameras, Insurance Institute for Highway safety, Arlington, VA, USA. Kent S., Corben B., Fildes B., Dyte D. (1995) Red light running behaviour at red light camera and control intersections, Report no. 73, Monash University Accident Research Centre, Monash University, Australia. Lawson S.D. (1991) Red-light running: Accidents and surveillance cameras, AA Foundation for Road Safety Research and Birmingham City Council, England. Levinson J. (1989) Gotcha! How Israeli police catch red-light jumpers on film, Traffic Safety, Nov/Dec, USA. Makinen T., Oei H.L. (1992) Automatic enforcement of speed and red light violations, Research report, SWOV, Leidschendam, The Netherlands. Mann T., Brown S., Coxon C. (1994) Evaluation of the effects of installing red light cameras at selected Adelaide intersections, Report Series 7/94, Office of road safety, South Australian Department of Transport, Australia. Mohamedshah Y., Chen L. W. (2001) Association of selected intersection factors with red light running crashes, to appear in Journal of the Institute of Transportation Engineers, Washington DC, USA. Ng C.H., Wong Y.D., Lum K.M. (1997) The impact of red-light surveillance cameras on road safety in Singapore, Journal of Road and Transport Research, 6 (2), Australian Road Research Board, Australia. NHTSA (2000) Traffic safety facts, FARS/GES, National Highway Transportation Agency, Department of Transportation, Washington, USA. Oei H.L., Catshoek J.W.D., Bos J.M.J., Varkevisser G.A. (1997) Project Red-light and Speed(PROROS), SWOV report R-97-35, Dutch Road Safety Research Institute, Leidschendam, The Netherlands. Office of Auditor General (1996) Improving road safety - Speed and red-light cameras - The road trauma trust fund, Report 1, Office of the Auditor General, Western Australia, Australia. Queensland (1995) Queensland’s road toll - Queensland Transport Submission to the parliamentary Travelsafe Committee, Queensland Transport. © Association for European Transport 2002 Retting R.A., Williams A.F., Greene M.A. (1998) Red-light running and sensible countermeasures, TRB Record 1640, Transportation Research Board, Washington, USA. Retting R., Williams A.F., Farmer C.M., Feldman A.F. (1999a) Evaluation of red Light camera enforcement in Fairfax, VA, USA, ITE Journal, August. Retting R.A., Williams A.F., Farmer C.M., Feldman A.M. (1999b) Evaluation of red light camera enforcement in Oxnard, California, Accident Analysis and Prevention, 31, 169-174. Retting R.A., Ulmer R.G., Williams A.F. (1999c) Prevalence and characteristics of red light running crashes in the United States, Accident Analysis and Prevention, 31, 687-694. Retting R.A., Chapline J.F., Williams A.F. (2000) Changes in crash risk following re-timing of traffic signal change intervals, Report, Insurance Institute for Highway Safety, Arlington, VA, USA. Retting R.A., Kyrychenko S.Y. (2001) Crash reductions associated with red Light camera enforcement in Oxnard, California, Report, Insurance Institute for Highway Safety, Arlington, VA, USA. Rocchi S., Hemsing S. (1997) A review of the road safety benefits of red light cameras, ICBC report, Insurance Corporation of British Columbia, Vancouver, Canada. Shinar, D., McKnight, A.J. (1984) The Effects of Enforcement and Public Information on Compliance, in Human Behaviour and Traffic Safety, Evans, L. & Schwing, R.C. (Eds.), Plenum Books, N.Y. South D., Harrison W., Portans I., King M. (1988) Evaluation of the red light camera program and the owner onus legislation, Report SR/88/1, Road Traffic Authority, Victoria, Australia. Thompson S.J., Steel J.D., Gallear D. (1989) Putting red-light violators in the picture, Traffic Engineering and Control, March. Woolley J., Taylor M. (1998) Red light running in Adelaide, Proceedings of the 19th ARRB Transport Research Ltd. Conference, Sydney, Australia, pp 191-209. Zaidel D. (2000) Legal measures and enforcement, EU project Gadget, Working paper WP5, VTT, Finland. © Association for European Transport 2002
© Copyright 2026 Paperzz