Safety Performance Indicator for Signalized Intersections Review
Introduction
Road safety performance indicators have been the subject of research over the years. This follows continuous endeavors to make roads safer. The research is however conducted in an environment where there exists little to no data on safety performance indicators. Where existent, the quality of the data is adversely affected by factors of randomness whereby both accidents, near-miss incidents, and the lack of both accidents and near-miss incidents are characterized by randomness and the inability to predict and assign probabilities (Yuan et al., 2008). Starting with the general understanding of the environment within which research on safety performance indicators for signalized intersections is conducted, this research seeks to provide a more informative perspective into the road safety concerns of researchers especially for signalized intersections.
Accidents and traffic incidents often result from the unsafe operational conditions of road traffic systems. Consequently, understanding the potential areas of probable unsafe operational conditions of the road traffic system becomes the beginning point for researchers’ intent on making the road traffic system safer. One of the areas where researchers can reduce the impact of randomness in accidents and traffic incidents is in the signalized intersections considering the possibility of improving the road systems in order to reduce and eliminate road accidents and incidents (Essa & Sayed, 2018).
Understanding safety performance indicators begins with the determination of what safety on the road entails. This research presents the definition of safety to include all factors, events, events, situations, and incidents that present a potential threat to road users who include pedestrians, riders, drivers, and every other road users. This definition delimits road safety from the perspective of actual crashes to include any situations considered as near-miss situations as well as conflicts on the roads. The conflicts include conflicts among motorized and non-motorized vehicles, motorcycles, bicycles, and pedestrians. The consideration appreciates that with the many different road users, traffic accidents and near miss incidents among vehicles may be well recorded. However, there are many other traffic incidents that may go unnoticed but which pose a major threat to the users of road intersections and the road transport systems in general (Ma, Shao, Ma & Ye, 2011).
In recent years a significant potential has been recognized in improving roads safety and different approaches have been developed to improve road safety (Hauer et al., 2002; Elvik and Vaa, 2004; Sørensen and Elvik, 2007; AASHTO, 2010). As a result of continued research, developed nations such as the United States and European countries have adopted highway management systems that identify locations with potential for safety improvements based on a variety of road safety prediction models. Authorities have continued to modify locations and road sections that are identified for safety improvements and have continued to implement countermeasures for road safety issues. Besides the implementation of data-informed safety measures and countermeasures, research has continued to focus on the benefits and costs of the measures and countermeasures.
It is difficult to find comprehensive data on road safety issues for a particular road or even a particular intersection especially considering the factors of randomness in road safety accidents and incidents (Lu et al., 2008; Bergel-Hayat, 2012). Consequently, models used in the management of highway safety systems borrow heavily from data on others roads as well form different road systems across the world. It is as a result of such widely varied data that safety-conscious road designs are increasingly vital. The goal is to continually incorporate safety data into the designs of roads across the world by observing comparative road designs and models. In so doing, all areas of possible safety incidents are identified and corrected long before road designs are implemented. Often, this affects all or majority of road segments. However, many intersections remain increasingly difficult to alter thereby making it necessary to continue research in this area (Gashaw, Goatin & Härri, 2018).
Research confirms that road intersections of weaving sections of the road are the weakest points in a rod network (Hakim et al., 1991; Buckholz, 1993). It is at this point that many road incidents are likely to occur as a result of both road design factors and human factors. The high levels of traffic at intersections also increases the potential for road accidents and incidents at intersections. Additionally, the very fact that the signalized road management system at the intersections may fail makes the intersections highly susceptible to road accidents and incidents.
A common issue of consideration is the case of drivers misunderstanding the signals on the signalized sections of the roads. This is particularly caused by the presence of multiple signals direct people and traffic to different sections on the roads. The failure of a single signal in the signalized intersections is also a possible cause of potential misunderstanding by the drivers and this increases the risk of accidents (Cafiso, Cava & Montella, 2007). Such cases have the potential of causing severe road crashes especially at the signalized intersections.
When designing the signal systems at intersections, traffic engineers consider many factors including the expectations of the drivers. For instance, the lead and lag times are an important factor of consideration for the design of road intersection signal systems. Modern systems use traffic data along the intersection to influence the lead and lag timing of the systems. However, this does not always function in accordance with the expectations of the drivers and it presents high potential for conflicts at the intersections. These are some of the complexities experienced by engineers in the design of traffic management systems. The factors also influence the concept of road safety in such areas (Sawalha & Sayed, 2001).
A properly functioning traffic light signal system should be able to provide smooth and orderly flow of traffic in the intersections. It should also help in reducing conflicts as well as delays when vehicles are merging to a particular side of the road. This links to the enhancement of lane integrity with is often affected by the traffic lights systems and also by human factors such as the behavior of the drivers (Gross, Lyon, Persaud & Srinivasan, 2013).
About 15% to 20% of movements in signalized intersections involve left turning (Wang & Abdel-Aty, 2008). The left-turning traffic is considered a major cause of conflicts in signalized intersections. When there is heavy traffic it becomes difficult for vehicles to cross from right to left. The implications are that this also results in growing traffic along the intersections and though the left-turning exits may be clear, it becomes difficult for the vehicles to move resulting in growing traffic levels as well as further delays. These factors directly result in traffic accidents and incidents at the signalized intersections.
The conflicts caused by left-turning vehicles, linked with delayed traffic lights signaling, result in in growing disobedience among the drivers. Some left-turning drivers may keep the inner lanes when approaching the intersections, especially when the inner lanes appear to be moving faster (Harwood, Bauer, Potts, Torbic, Richard, Rabbani, & Griffith, 2003). At the same time, other drivers disobediently keep the outer lane even when they do not intend to exit or take the left-turns. Such disobedience of the road traffic signals and rules heavily affects the capacity of traffic on the intersections. Drivers may then grow impatient resulting in further worsening of traffic conflicts at such intersections. Some drivers will be forced to directions they did not intend to move to while others are bullied along such intersections due to the impatience. In the midst of such issues the potential of accidents and traffic indents arises. This indicates some of the many reasons why traffic incidents and issues may be high in the signalized intersection sections.
The high number of possible causes of conflicts, accidents, and incidents on signalized intersections are an indicator of the variety of factors that affect safety (Persaud & Nguyen, 1998). Consequently, safety performance indicators for signalized intersections must put into consideration all the possible causes of conflicts, incidents, and accidents involving people and traffic in such intersections. Human factors such as the behavior of drivers, behavior of pedestrians, and behavior of other people such as law enforcement officers close to such intersections are an important consideration in the determination of safety performance indicators (Vitetta, Musolino & Marcianò, 2007). At the same time, more objective considerations concerning the design of the road are critical in influencing safety performance indicators in the signalized intersections. Road design factors would include geometrics, road makings and traffics signs, the conditions of pavements, the lighting conditions, and traffic characteristics. This perspective appreciates the fact that alterations to the road designs at intersections, as an enhancement to the signalized traffic lights systems is one of the key considerations in safety performance of signalized intersections (Asalor, 1984).
Empirical review of literature as herein presented is important in demonstrating various perspectives on safety in signalized intersections. This is because the review of literature helps in highlighting some of the reasons commonly linked to accidents and incidents on the roads and most importantly in demonstrating how best to deal with such challenges in signalized intersections. The following section focuses on theoretical review, focusing on theoretical frameworks that are used in explaining the common causes of accidents and incidents on different sections of a road.
Theoretical Review
Various theoretical models have been developed to help explain safety issues in road systems and to help in the identification of safety performance indicators. These models have guided traffic infrastructure planning and in many have ways helped in reduction of accidents and incidents on the roads. Since development, the models have been employed in the improvement of traffic management systems by providing data for analysis of the potential for traffic incidents and accidents.
This section will focus on three theoretical perspectives of the safety performance indicators in traffics systems. The three theoretical perspectives include crash based models, conflict models, and non-crash non-conflict models. The models are briefly discussed in the following sections.
Crash Model
Crash based models draw inspiration from crash data. In the models, the likelihood of an accidents occurring in a particular area is considered from the perspective of the number of observed accidents in the same or similar area. The causes of such accidents are considered and the data collected used in improvement and redesigning of the sections as well as in the development of new transport systems and road designs that seek to eliminate the particular causes of accidents.
There is a wide variety of factors that may lead to accidents or incidents at particular section such as intersections. From the review of literature, the possible cause of accidents include human factors involving drivers, riders, pedestrians, and even the police officers directing traffic at intersections from time to time. The cause of accidents also involve the road design characteristics such as the presence of sharp turning angles and other factors of geometrics. Road design factors also include the presence or lack of traffic lights and where there are traffic lights, factors such as lead and lag time, the colors, numbers, and locations of the traffic lights relative to the various users of the traffic light systems. Additionally, there may be other factors such as motor vehicle characteristics including what types of vehicles are likely to cause accidents. For instance, long trucks may comprise the highest number of accidents and reported traffic incidents reported in a particular area under analysis. The consideration of all such factor sin the crash models helps in continuous improvements of the road models with the goal of eliminating the failures in the design of the roads and to reduce the accidents.
Crash frequencies naturally fluctuate over time. The implications are that short-term averages may be significantly different from the long-term data trends. Similarly, the fluctuations also encompass aspects of randomness as well as the myriad of human error elements that give the model a weak predictive power. Using the crash models it is not possible o confidently state when and how an accident may occur.
Despite the influence of randomness and human factors elements in crash models, crash models are considered simple and informative in helping the identification of common causes of accidents and how they can be mitigated, mostly by focusing on the design of road systems and traffic management systems.
One of the reasons why crash models have remained significant despite the concerns of randomness, is the finding that crash observations are not always independent of each other. Researchers have continued to model different road design factors such as Annual Average Daily Traffic (AADT), degree of horizontal curvature, lane, shoulder and median width, urban/rural, and the section’s length, on the frequency of accident occurrence (Abdel-Aty & Radwan, 2000). Being highly subjective factors, the determination of how such factors influence the probability of accidents in different sections of the roads has been instrumental in influencing road safety. Particularly, these factors have been important in elimination of road design features that are characterized with highest linkages to road accidents. This is one of the areas where crash models have been instrumental in definition of safety performance indicators on the roads.
Crash models are used not just in real road environments. Rather, there are also used in simulated environments for the development of high performance road systems and road sections. Particularly, the considerations of the road usage conditions with the goal of minimizing road accidents had seen simulations using crash models becoming an essential part. This is especially important at points where the traffic volumes are high, including road intersections, and where other factors such as terrain and topography have an impact on the road design processes. These factors have been consistently critical in ensuring the development of safe road systems.
In summary, Poisson models (Jones et al., 1991; Miaou and Lum, 1993) and negative binomial (NB) models (Miaou, 1994; Poch and Mannering, 1996; Abdel-Aty and Radwan, 2000) have been widely used to capture the relationship between traffic crashes and contributing factors. The models rely heavily on the availability of data and correct definition of the contributing factors. When correctly developed the models have been instrumental in establishing probabilities of accidents and incidents occurring at specified spatial points, during specified days of the week or even months, and have ben key in identifying areas of improvement in road systems and in the case of signalized intersections, how leads and lags can be set to minimize the capacity of traffic and also to eliminate traffic accidents and incidents. What is important about these models is that they are considered ahead of the design of road designs as well as signalized systems the results of the models are well included in the design of the roads and signalized intersections. These models have been key in making road sections safe.
There has been a major concern with the use of crash models where data on crashes is nearly inexistent or where crashes are nearly inexistent. The zeros, when used in prediction models, become a challenge. Researchers have bene considering the use of zero-inflated negative binomial models thereby helping in the overcoming of the weaknesses associated with the zero incidents in the predictive models (Kumara & Chin, 2003). While the predictive power of zero-inflated negative binomial models may be low, it helps in the determination of the conditions linked to accidents. In a study conducted in Singapore it was determined that uncontrolled left-turn slip road, permissive right-turn phase, existence of a horizontal curve, short sight distances, large number of signal phases, total approach volume, and left-turn volume may increase accident occurrence (Kumara & Chin, 2003).
As it appears in this analysis, the use of crash models is not as important in predicting accidents as it is in helping the identification of the factors linked to accidents. Essentially, the goal is to help in the identification of the safety performance indicators which is why this theoretical framework is considered essential in this research which focuses on safety performance indicators for signalized intersections. In application, crash models would be considered from the aspect of how they have been employed to improve safety at signalized intersections. Using Generalized Estimating Equations (GEE), Wang and Abdel-Aty (2008) showed that there are obvious differences in the factors that cause the occurrence of different left-turn collision patterns.
Conflict Models
Traffic conflicts are measures of accident potential and operational problems at a highway location. By definition, a traffic conflict is an observable situation in which two or more road users approach each other in space and time for such an extent that there is a risk of collision if their movements remain unchanged (Amundson and Hyden, 1977). Perkins and Harris define a traffic conflict as the interactions between vehicles, and such interactions can result in actions such as braking and changing direction of movement.
Early attempts by practicing highway engineers to diagnose operational or safety deficiencies included the simple technique of observing erratic driving, unsafe maneuvers and “near misses” at problem locations (Baker, 1977) This method was first formalized by McFarland and Moseley (1954) who observed “near misses”, judged as “emergency situations or critical incidents which could easily have led to an accident” experienced by intercity bus and truck drivers.
Besides “near misses” there are other methods used in identifying traffic conflicts. The traffic conflicts technique (TCT) was developed in an attempt to objectively measure the accident potential of a highway location without having to wait for a suitable accident history to evolve. Using the virtual trajectories with the relative speeds and angular direction for each trajectory to predict the occurrence and the severity of accidents traffic conflict is evaluated by the term time to collide (TTC).
The first formalized procedure for identifying and recording traffic conflicts at intersections was developed by Perkins and Harris of General Motors Corporation in 1967. Major types of conflicts at intersections include rear-end, left-turn, cross-traffic, red-light violation, and weave conflicts. While crashes may be avoided during many such conflicts, the analysis of the conflicts may help in identifying possible areas of improvement in intersections (Guido, Astarita, Giofré & Vitale, 2011).
The use of conflict models advances the discourses on crashes by recognizing not just the crashes but also the incidents that could have easily resulted in accidents. It is a recognition that there are many incidences that go unreported because there was no actual crash or traffic accident. By virtue of including more data, this aspect ensures that a lot more information is captured in the models thereby providing more accuracy on reasons for accidents and traffic incidents (Flannery, Elefteriadou, Koza, & McFadden, 1998). By studying these aspects, it becomes possible to identify the safety performance indicators more accurately.
Traffic conflict analysis is applicable where data on crashes may not be available. A good example is in the case of new highways. Traffic conflict analysis helps in the identification of near miss instances which can be observed in relatively short period of time. A researcher does not need to wait for long hours in order to collect data on crashes (Lu, Pan & Xiang, 2008). The key areas of focus would include the cover conflicting points, number of conflicts, conflict rate, conflict distribution, and conflict forecasting models. The weaknesses of TCA include the fact that TCA uses judgment and determination of traffic conflicts are more subjective. Different observers may provide different traffic conflict judgments. In addition, TCA is a time-consuming task.
There are different methods of collecting data on traffic conflicts. These include observation, simulations, and video-recording. Field observations considered to be the first and the basic method of data collection, where the data collector follow the general role of the conflict which is recording all the “near miss” events. The decision to record the near miss situation may vary from person to person. Most of the recordings shows that the results were inconsistent, techniques for analyzing the relevant data either are lacking altogether or differ markedly from study to study and successive reports omit a critical assessment of earlier work. One of the most important aspects to consider when utilizing conflict data is the reliability of data collected by observers (Eisele & Frawley, 2005). There are many factors which will account for variation in conflict counts including alertness, experience, and different driving attitudes of the observers, location of the observer at the site, and traffic volumes. Field observers are not only expensive, but inter- and intra-observer variability is a common challenge for the repeatability and consistency of results (William, 1972).
The second method is simulations. In recent years, traffic conflict simulation models have attempted to analyze traffic safety performance at roadway intersections. Pirdavani, Brijs, Bellemans & Wets, 2010) considered what they considered micro-simulation in understanding safety indicators in intersections. Model development for traffic conflict simulation needs long-term data accumulation, and simulation model parameters could change as environmental conditions change, limiting its applicability in real situations (Yuan et al., 2008). Focusing on simulation of conflicts for both T and 4-leg unsignalized intersections, Sayed et. al (1994) considered traffic conflicts as critical-event traffic situations and the effect of driver and traffic parameters on the occurrence of conflicts. The simulation was found useful for assessing safety performance and feasible solutions for other unsignalized intersections.
Huang and Pant (1994) simulated measures of effectiveness of traffic control at a high-speed intersections. The model considered three critical elements which included probability of being caught in a dilemma zone, speed of a vehicle in different segments of the intersection approach, and vehicle conflict rate. The inclusion of the three components of the model showed the wide variety of issues that should be considered in safety performance analysis of intersections. Similarly, Rao and Regaraju (1998) focused on intersections and showcased the variety of issues that cause conflicts at intersections.
Persaud and Mucsi (1995) simulating accidents in rural areas under different conditions used regression package that allowed the assumption of a negative binomial error structure. Regression models were calibrated for the different combinations of time periods (24 hr, day hours, and night hours) and geometric (roadway and shoulder width) characteristics. The research showed that the effect of day/night conditions is different for single-vehicle and multivehicle accidents. For single-vehicle accidents, the accident potential was higher during the night, whereas for multivehicle accidents the opposite is true. The simulation indicated the importance of differentiating between single-vehicle and multivehicle accidents and day/night conditions (Persaud & Mucsi, 1995). Importance of mixed conditions was also highlighted in Li, Yue & Wong (2004).
The last method is video-recording. Video sensors are selected as the primary source of data due to advantages of richness in details, inexpensiveness, and ubiquitous usage for monitoring purposes. The automated extraction of road users’ positions from video data using techniques in the discipline of computer vision has been advocated as a resource-efficient and potentially more accurate alternative (Ismail, Sayed, Saunier, 2009b). Qu, Kuang, Oh, & Jin, 2014; and Cunto 2008 noted the importance of video recording in observing microscopic indicators.
Non-crash non-conflict model
Non-crash non-conflict model is the last theoretical framework for consideration. As the title suggests, the model does not rely on traffic crashes or traffic conflicts data. Instead, the method relies more on professional expertise and experience of road and transport systems’ designers. An important aspect of this framework is that it emphasizes the professional skill and expertise of the analysts and also looks into aspects of design that may be invisible to the inexperienced eye (Oh, Park & Ritchie, 2006).
This type of safety Assessment is conducted by the personal judgments of traffic safety experts. In most cases those traffic safety experts are road designers because the focus is on the dimensions that are related directly to the road safety such as, the geometrics of the design, characteristics of road and the type of traffic control.
Usually, non-crash- and non-conflict-based Assessment methods can provide a safety Assessment for a roadway facility in a relatively short time; these methods are relatively easy to implement in real applications. This type of approach has advantages; it is low cost, has high efficiency, and is less time-consuming. Furthermore, this type of approach can be better used by field safety engineers to find potential safety problems and corresponding countermeasures can be implemented in a short time so that possible traffic safety problems can be prevented (Oh, Park & Ritchie, 2006).
In Assessment of signalized intersections, the installation of traffic lights systems, lane integrity, width and length, the shoulders, and locations of turning points as well as crossing points become the important indicators that the expert focuses on. The information from the non-crash non-conflict analysis may be analyzed together with the data from the other two models discussed above.
Summary
Complex analyses involve the use of Accident prediction models or Safety Performance Functions, or Bayes` empirical analysis. Prediction models include the use of correlation model that was derived from the analysis of actual crash data. This means that many models are applicable to local areas and often they cannot be applied to the analysis of traffic safety in other regions or countries. The application of these models to other areas requires modification factors that are related to local areas. It is difficult to define clear procedures by which the modification of factors can be carried out in a simple way and allow application of the model in other regions. In ecological analysis, traffic accidents are viewed as aggregated data, and they bind to a geographical level (section, crossroads, district, province, etc.). Statistical models that use frequency in order to assess the correlation between several routes, road and environment factors, as well as transport risks are applied on these data. Identification of black spots on roads aims to find road parts on which the traffic risk is higher, while the environmental analysis seeks to determine the factors that influence the increased risk of accidents. The spatial unit size, which is the subject of analysis depends on the objective of the analysis and the available data. Because of the problems that analysts are faced by, they often aggregate data on road accidents compared to several spatial levels starting from the intersection or section of road which is a micro-level, to regions, provinces or countries that belong to the macro level.
References
Abdel-Aty, M. A., & Radwan, A. E. (2000). Modeling traffic accident occurrence and involvement. Accident Analysis & Prevention, 32(5), 633-642.
Al-Ghafli, A., Garib, A., Sarhan, M., & Al-Harthi, H. (2013). Effect of Signal Phase Sequences on Signalized Intersection Safety Performance Case Study from Abu Dhabi, U.A.E. Published in 17th IRF World Meeting & Exhibition, Riyadh, Saudi Arabia.
Amundson, F., & Hyden, C., (1977). Proceedings of first workshop on traffic conflicts. Institute of Economics, Oslo.
Archer, J., (2001). Traffic Conflict Technique Historical to current State-of-the-Art. Effektmodeller för vägtrafikanläggningar (EMV), TRITA-INFRA 02-010, Stockholm, September, 2001. ISSN 1651-0216, ISRN KTH/INFRA-02/010-SE.
Asalor, J. (1984). A general model of road traffic accidents. Applied Mathematical Modelling, 8(2), 133-138. doi:10.1016/0307-904x(84)90066-0
Bergel-Hayat, R., (2012). Time-series models of aggregate road risk and their applications to European countries. Transport Reviews: A Transnational Transdisciplinary Journal 32, 211–246.
Buckholz, J. (1993). The 10 major pitfalls of coordinated signal timing. ITE Journal, 63(8), 26-29.
Cafiso, S., Cava, G., & Montella, A. (2007). Safety index for Assessment of two-lane rural highways. Transportation Research Record: Journal of the Transportation Research Board, (2019), 136-145.
Candappa, N., Logan, D., Van Nes, N., & Corben, B. (2015). An exploration of alternative intersection designs in the context of Safe System. Accident Analysis & Prevention, 74, 314-323.
Cunto, F. (2008). Assessing safety performance of transportation systems using microscopic simulation.
Eisele, W., & Frawley, W. (2005). Estimating the safety and operational impact of raised medians and driveway density: experiences from Texas and Oklahoma case studies. Transportation Research Record: Journal of the Transportation Research Board, (1931), 108-116.
Essa, M., & Sayed, T. (2018). Traffic conflict models to evaluate the safety of signalized intersections at the cycle level. Transportation research part C: emerging technologies, 89, 289-302.
Flannery, A., Elefteriadou, L., Koza, P., & McFadden, J. (1998). Safety, delay, and capacity of single-lane roundabouts in the United States. Transportation Research Record: Journal of the Transportation Research Board, (1646), 63-70.
Gashaw, S., Goatin, P., & Härri, J. (2018). Modeling and analysis of mixed flow of cars and powered two wheelers. Transportation research part C: emerging technologies, 89, 148-167.
Gross, F., Lyon, C., Persaud, B., & Srinivasan, R. (2013). Safety effectiveness of converting signalized intersections to roundabouts. Accident Analysis & Prevention, 50, 234-241.
Guido, G., Astarita, V., Giofré, V., & Vitale, A. (2011). Safety performance measures: a comparison between microsimulation and observational data. Procedia-Social and Behavioral Sciences, 20, 217-225.
Hakim, S., Hakkert, S., Hochermann, I., & Shefert, D. (1991). A critical review of macro models for road accidents. Accident Analysis & Prevention, 23(5), 379–400.
Harwood, D., Bauer, K., Potts, I., Torbic, D., Richard, K., Rabbani, E., … & Griffith, M. (2003). Safety effectiveness of intersection left-and right-turn lanes. Transportation Research Record: Journal of the Transportation Research Board, (1840), 131-139.
Highway safety manual 2010 (HSM 2010)
Huang, P., & Pant, P., (1994). Simulation neural-network model for evaluating dilemma zone problems at high-speed signalized intersections. Transportation Research Record 1456, 34–42.
Ismail, K., et al., (2009). Automated Analysis of Pedestrian-Vehicle Conflicts Using Video Data. Transportation Research Record: Journal of the Transportation Research Board, Washington, D.C: s.n., 2009, Vol. 2140, pp. 44-54.
Ismail, K., Sayed, T., Saunier, N., (2009b). Automated pedestrian safety analysis using video data in the context of scramble phase intersections. In: Annual Conference of the Transportation Association of Canada, Vancouver, BC.
K. Ismail, T. Sayed, N. Saunier. (2010). Automated Analysis of Pedestrian-vehicle Conflicts: A Context for Before-and-after Studies. Washington, DC. : TRB, 2010. Transportation Research Board Annual Meeting.
Kumara, S. S. P., & Chin, H. C. (2003). Modeling accident occurrence at signalized tee intersections with special emphasis on excess zeros. Traffic Injury Prevention, 4(1), 53-57.
Li, J., Yue, Z. Q., & Wong, S. C. (2004). Performance Assessment of signalized urban intersections under mixed traffic conditions by gray system theory. Journal of Transportation Engineering, 130(1), 113-121.
Lu, J., Pan, F., & Xiang, Q. (2008). Level-of-safety service for safety performance Assessment of highway intersections. Transportation Research Record: Journal of the Transportation Research Board, (2075), 24-33.
Lyon, C., Haq, A., Persaud, B., & Kodama, S. (2005). Safety performance functions for signalized intersections in large urban areas: Development and application to Assessment of left-turn priority treatment. Transportation Research Record: Journal of the Transportation Research Board, (1908), 165-171.
Ma, Z., Shao, C., Ma, S., & Ye, Z. (2011). Constructing road safety performance indicators using fuzzy delphi method and grey delphi method. Expert Systems with Applications, 38(3), 1509-1514.
Mehmood, A., Saccomanno, F., Hellinga, B. (2001). Simulation of road accidents by use of systems dynamics. Transportation Research Record 1746, 37–46.
MNDOT (2005). Minnesota Manual on Uniform Traffic Control Devices. Mn/DOT, U.S. Department of Transportation, FHWA, Part 4.
Oh, C., Park, S., & Ritchie, S. G. (2006). A method for identifying rear-end collision risks using inductive loop detectors. Accident Analysis & Prevention, 38(2), 295-301.
Persaud, B., & Mucsi, K. (1995). Microscopic accident potential models for two-lane rural roads. Transportation Research Record 1485, 134–139.
Persaud, B., & Nguyen, T. (1998). Disaggregate safety performance models for signalized intersections on Ontario provincial roads. Transportation Research Record: Journal of the Transportation Research Board, (1635), 113-120.
Persaud, B., Lord, D., & Palmisano, J. (2002). Calibration and transferability of accident prediction models for urban intersections. Transportation Research Record: Journal of the Transportation Research Board, (1784), 57-64.
Pirdavani, A., Brijs, T., Bellemans, T., & Wets, G. (2010). Assessment of traffic safety at un-signalized intersections using microsimulation: a utilization of proximal safety indicators.
Qu, X., Kuang, Y., Oh, E., & Jin, S. (2014). Safety Assessment for expressways: a comparative study for macroscopic and microscopic indicators. Traffic injury prevention, 15(1), 89-93.
Rao, V., Regaraju, V. (1998). Modeling conflicts of heterogeneous traffic at urban uncontrolled intersections. Journal of Transportation Engineering 1 (23).
Sawalha, Z., & Sayed, T. (2001). Evaluating safety of urban arterial roadways. Journal of Transportation Engineering, 127(2), 151-158.
Sayed, T., Brown, G., Navin, F. (1994). Simulation of traffic conflicts at unsignalized intersections with TSC-Sim. Accident Analysis and Prevention 26 (5), 593–607.
Vitetta, A., Musolino, G., & Marcianò, F. A. (2007). Safety of users in road evacuation: Supply and demand-supply interaction models for users. WIT Transactions on the Built Environment, 96.
Wang, X., & Abdel-Aty, M. (2008). Modeling left-turn crash occurrence at signalized intersections by conflicting patterns. Accident Analysis & Prevention, 40(1), 76-88. doi:10.1016/j.aap.2007.04.006
William, T.M. (1972). An Assessment of traffic conflict technique. Highway Research Record 384, 1–8.
Xie, K., Wang, X., Ozbay, K., & Yang, H. (2014). Crash frequency modeling for signalized intersections in a high-density urban road network. Analytic methods in accident research, 2, 39-51.
Zegeer, C.V., Deen, R.C., 1978. Traffic conflict as a diagnostic tool in highway safety. Transportation Research Record 667, 48–55.
