Development of a Road Protection Value Model as an Indicator for National Road Safety Performance Assessment from the Perspective of Drivers of Four or More Wheeled Motorized Vehicles

ABSTRACT


IJSSR Page 2225
Basically, the iRAP star rating safety performance measure is developed through an assessment of road elements (Road Assessment) (iRAP, 2009;iRAP, 2010b;iRAP, 2012).An approach through direct assessment of road elements is considered more realistic compared to the approach developed so far which is more oriented towards accident data.The quality and condition of the road and the road environment through the study of road elements greatly determine the star rating given by the road.The better the technical standards implemented, the better the star rating obtained.This concept is used by iRAP, known as assessing road protection for road users.The amount of road protection value for road users (RPS: Road Protector Scores) is determined from a number of road parameters and attributes.This value is then ranked into a 5 star rating which describes the overall safety performance of a road section.A 5-star rating indicates the best performance, whereas a 1-star rating indicates poor safety performance.
To calculate the road protection value, a number of road attributes are required as elements that are expected to contribute to road user safety (iRAP, 2010b;iRAP, 2012).Each attribute has an indicator value called risk value, which is developed from the values of Crash Modification Factors (CMF) (iRAP, 2010b;iRAP, 2012;Elvik et al, 2009;AASHTO, 2010;PIARC, 2003).
The study and development of the RPS or SRS model is generally carried out in developed countries, so this model is likely to be more suitable for countries that have traffic characteristics and road technical standards that are in accordance with those of developed countries.For Indonesian conditions, the use of this model still requires modification because it is based on the traffic characteristics and road environment which are different from those of developed countries.Likewise with the fulfillment of standards and technical specifications for national road sections, not all of which have been fulfilled ideally.This certainly affects traffic movements which ultimately has an impact on traffic accidents on national roads.Therefore, in addition to fulfilling road technical standards and specifications, of course the differences in the characteristics of traffic accidents on these national road sections are seen as influencing the RPS value provided by these roads to road users.This assumption is the premise of this research with the assumption that the national road SRS model is different from the RPS or SRS iRAP model that has been practiced in many countries.
The RPS or SRS model developed by iRAP in road assessment is designed for 4 perspectives of road users, namely car occupants, motorcyclists, bicyclists and pedestrians.This model is also designed to utilize the Accident Modification Factor (AMF) value or the Crash Modification Factor (CMF) value.In the RPS calculation that has been developed by road safety researchers.The use of CMF in the RPS or SRS iRAP model makes the road protection calculation model very measurable, so that its use is seen as providing many effective technical recommendations from a number of existing treatment options.
Even though this model has been widely used in various countries, it is not yet known to what extent the accuracy and effectiveness of this model can be applied to improve road safety and/or reduce the number of accident fatalities on Indonesian national roads.The SRS model is seen as still needing to be adapted to conditions in Indonesia, bearing in mind the many traffic and road problems in Indonesia, such as compliance with road standards (geometric, road quality, signs and markings, facilities for accident-prone groups) which have not been maximized, mixed traffic (high proportion of motorbikes), traffic behavior that causes many traffic conflicts, and high side obstacles.
Various traffic and road problems in Indonesia are currently seen as factors that will influence the SRS model for conditions in Indonesia, especially for the SRS of Indonesian roads.In general, this study focuses on the study of the development of influencing factors on the SRS model which is in accordance with the national road conditions along 47,000 km which is referred to as the SRS model of the Indonesian National Road.This SRS model is not only based on an assessment of a number of road elements as part of the National Road SRS calculation model, it is also based on an analysis of traffic accident characteristics from 283,519 accident data on national roads obtained from 2012 to 2019.This accident data is sourced from IRSMS data base of Korlantas POLRI Headquarters as one of the main data used in this research.

METHODS
Based on the SRS model developed by iRAP, the types of accidents mainly result in deaths (fatal) and serious injuries (serious injuries) as well as road factors that influence accidents and the level of seriousness of accidents.The SRS model developed for national road sections in Indonesia also follows the same concept as that developed by iRAP.
Indonesia has traffic characteristics, the provision of road infrastructure, and the behavior of road users which causes Indonesia to have different accident characteristics.Therefore, the study of typical traffic accidents as a cause of fatal accidents and serious injuries is one of the foundations for developing this SRS model.As a first step, the development of the SRS model for national roads, especially from the perspective of users of 4-wheeled or more motorized vehicles in various provinces and islands, is important.The uniformity of traffic accident characteristics through statistical tests will show that the selected accident types based on statistical tests will be a basic consideration in developing the SRS model for Indonesia's national roads.
The statistical test used to assess the uniformity of accident characteristics in this paper is carried out using a two sign test known as the Wilcoxon Test.This test is used to determine consistency whether there is a difference between the proportion of accident types from each zone to the average proportion of zone accident types from all traffic accident class categories and fatal accident and serious accident categories on national roads in Indonesia.
The next stage is to carry out an inventory of the model tools developed by iRAP, especially with road attributes which are part of likelihood, severity, and other factors that are considered to influence traffic accidents.The road assessment attributes shown in Table 3 are the results of a study of various previously developed SRS models which were then combined with the concept proposed for Indonesian road sections (Idris et al, 2022).Attributes marked with an asterisk (*) are original attributes developed by iRAP.Meanwhile, the other attributes in the table are additional attributes that are deemed necessary to consider for road and environmental conditions as well as traffic conditions on Indonesian national roads.In total there are 53 road attributes that are considered for national road sections which include 23 attributes for front-rear collision accidents, 20 attributes for front-side collision accidents when turning around, 20 attributes for collisions on property access, 25 attributes for collisions leaving the body of the road, 22 front-end collision attributes, and accidents at intersections with 20 attributes.From a total of 53 attributes, all parameters are divided into element likelihood and road geometric attributes (12 attributes), road condition likelihood (2 attributes), traffic control likelihood (3 attributes), road equipment likelihood (5 attributes) turning facility likelihood (3 attribute), likelihood of intersection (5 attributes), severity factor (12 attributes), speed factor (1 attribute), and external factors of traffic flow (4 attributes).
Figure 2 is a design model for calculating road protection values from a car occupant perspective based on the results of benchmarking road attributes from the various models proposed in this study.This study uses a survey of expert perceptions of a number of proposed attributes which include likelihood, severity, external factors of traffic flow, and operational speed which are considered to contribute to a type of traffic accident.
There are two stages of the questionnaire used in this study.The first stage of the survey was aimed at capturing a number of attributes for each type of accident using snowball sampling with road safety expert respondents.Meanwhile, the second phase of the survey was directed at assessing the level of importance and ease of application of the assessed attributes.Several statistical analysis tools in the form of data adequacy tests, uniformity tests, validity tests and reliability tests (Walpole et al, 1995;Ott, 1991;Sprent, 1991;Siegel, 1997) have been used in this study.The analytical method for the level of importance and attribute application of each parameter uses the IPA (Importance and Performance Analysis) method approach (Zeithaml et al, 1990).The IPA method is used to map the importance and performance levels to identify the attributes of the proposed road assessment (Zeithaml et al, 1990).The IPA method maps the average attribute weights into 4 quadrants.

A. National Road Accident Characteristics
To test the consistency of typical fatal and serious injury accidents (FIs: Fatality and Serious Injury) between zones and the typicality of all accidents on all national roads in Indonesia, this paper uses the Wilcoxon Paired Sign Rank Test.The null hypothesis (H o) is that there is no difference between the zone average proportion of fatal accidents and serious injuries from all accidents and the proportion of all typical accidents on national roads.The alternative hypothesis (H i ) is that there is a difference between the proportion of fatal accidents and serious injuries in the zone average of all accidents.The critical value of W or W table in the Wilcoxon Paired Rating Sign Test for a significance level of α=0.005; α=0.001; α=0.025; and α=0.05 is given as in the Wilcoxon Test Table .While the assessment criteria are given if W count > W Table 2 below is a summary of typical data for fatal and serious accidents (FSIs) which juxtaposes the average observation zone (Sumatra, Java, Kalimantan, Sulawesi, Bali & Nusa Tenggara, Maluku, Papua) with typical accidents at all levels of road accidents national.The Wilcoxon test shows that the calculated W value =23 is greater than the W Table value =11.This test concludes that there is no significant difference from the typical fatal and serious traffic accidents in each zone with all types Inrernational Journal of Social Service and Research https://ijssr.ridwaninstitute.co.id/IJSSR Page 2229 of accidents for all accident classes.These results further indicate that the typical accidents in all zones for both Fatal and Serious Accidents (FSIs) and for all accidents on Indonesian national roads have relatively the same typical.

B. National Road SRS Model Attribute Analysis
Phase-1 questionnaire data (93 samples) and Phase-2 questionnaire (43 samples) have been tested for adequacy and uniformity of data for a 95% confidence level.Likewise, the validity test based on the Pearson correlation coefficient with an error rate of 5% is also fulfilled.The reliability test has met the Cronbach Alpha coefficient value > 0.60.The validity test on the Phase-1 questionnaire data succeeded in eliminating several attributes proposed in the research design.The results are shown in Table 5, which include 20 attributes for front and back crash accidents, 16 attributes for front and side crash accidents when turning around, 18 attributes for collisions on property access, 24 attributes for collisions leaving the road, 21 attributes for front-on collisions, and accidents at intersections with 16 attributes.

Table 3 Results of Phase-1 Survey Analysis
In the same way as in the Phase-1 questionnaire analysis, the results of the statistical analysis test in the Phase-2 questionnaire show data adequacy and data uniformity which meets statistical

SRS Head-on Attributes
tests with a confidence level varying between 90%-95%.The validity test with a confidence level of 95% shows that all data on the level of importance and level of application of attributes is categorized as valid.Meanwhile, the reliability for testing the reliability of the instrument has a Cronbach Alpha coefficient value above 0.70; which indicates that the research instrument has reliability that varies between high and very high.
Table 6 shows the results of the IPA analysis between the level of importance and application based on the perception survey of road safety experts from various professional groups for the SRS Front-Rear Collision Accident model.There are 7 (seven) attributes mapped into quadrant IV which fall into the category of having a low level of importance and a difficult level of application.The seven parameters include road shoulder width, land use, road side utilization, roadside parking, number of accesses, number of motorbikes and bicycles are seen as having no significant effect on the model being developed.Therefore, these seven attributes are not considered in the SRS model.
Selected attributes for likelihood include number of lanes, shoulder width, shoulder width, Rbend, corner quality.Longitudinal slope of the road (grade), superelevation, road pavement condition.The attribute for severity is the effective width of the road, while the attribute for the selected thermal effect is the number of slow vehicles, while the attribute 85%-tile speed is the attribute for operational speed.
This front-rear crash parameter is one of the new parameters that was not previously known in the SRS model developed by iRAP.The consideration factor for entering this parameter is because the majority of front-rear crash accidents occur on narrow roads and on roads with type 2/2-UD.The factor of availability of ideal road infrastructure is still very dominant on Indonesian national roads.

Table 4 Analysis of the Level of Importance and Application of Road Assessment Attributes for Front-Rear Crash Accident Parameters
Furthermore, Table 7 and Table 8 below show the results of the IPA analysis of the perceptions of road safety experts on a number of attributes of the SRS model for Front-Side Collision Accidents at two different locations.Table 6 presents the results of the IPA analysis on front-side collision accidents during U-turns at various locations on national roads.Table 7 presents the results of the IPA analysis of front-side collision accidents at property access locations and other road access.
Furthermore, the results of the IPA analysis in Table 7 show a number of attributes that are in quadrant-IV, namely the condition of the road pavement, the number of motorcycles and the number of bicycles.Attributes that are part of the likelihood of front-side collision accidents when turning around include the number of lanes, lane width, visibility, grade, road median, skid resistance, road surface conditions, facilities and turnaround signs, speed limit signs.The selected severity attributes are R-turnover and median width.The speed factor of vehicles around the turn-around location indicated by the 85%-speed tile is an important attribute to consider.Two attributes related to the external influence of traffic flow, both the number of vehicles and the number of motorbikes, are seen as having no significant effect on the SRS model of Front-Side Collision Accidents at U-turn locations.
The attributes in Quadrant IV in Table 8 are attributes that are categorized as attributes that are not considered in the SRS model of Front-Side Crash Accidents at road access or property access locations.The attributes in Quadrant IV are median type, road side utilization, volume of motorbikes and bicycles.Attributes that are likely in this model include number of lanes, lane width, visibility distance, grade, road perspective, pavement condition, skid resistance, surface condition, land use, area type.The severity attribute based on IPA analysis for the SRS model of Front-Side Collision Accidents on access roads and property access is the number or intensity of access, side obstacles, and on-road parking.Another attribute considered in this model is operational speed.
Parameters of front-side crash accidents both at road access locations or property access as well as at turning locations are the main problems encountered on national road segments.Uncontrolled road access and property access on national road sections that have arterial functions make these locations high points of conflict that have the potential for traffic accidents.
Likewise, front-side crash accidents at U-turn locations were also found to be very dominant, especially on roads that do not have ideal U-turn facilities.Many median roads have been opened by the community or local government to meet local needs without considering the required median opening standards.High intensity median openings and median widths that are less than ideal which are often found on national roads designated as arterial roads are listed as dangerous locations.

Table 5 Analysis of the Level of Importance and Application of Road Assessment Attributes for Front-Side Crash Accident Parameters When U-turning Table 6 Analysis of the Level of Importance and Application of Road Assessment Attributes for Front-Side Crash Accident Parameters in Property Access
Table 6 further shows the results of the IPA analysis of the attributes of the Single Accident SRS model leaving the road.Based on the IPA analysis of the Front-Rear crash accident attributes, there are 6 attributes mapped into Quadrant IV.These attributes are grade, emergency lane, median traversability, traffic volume, volume of motorbikes and bicycles.

Table 7 Analysis of Levels of Importance and Application of Road Assessment Attributes for Single Accident Parameters Exiting the Road Body
The likelihood attributes of this SRS model include road type, lane width, road shoulder, shoulder width, sight distance, R-bend, curve quality, superelevation, road pavement condition, skid resistance, road surface condition, safety fence, delineation, rumble strip.The severity attributes of this model are the dangerous object on the side of the road, the distance of the dangerous object to the traffic lane, and the condition of the safety fence.The SRS model for a single accident leaving the roadway also considers the operational speed factor as an influential factor.Table 8 below shows a number of SRS model attributes for front-end crash accident parameters based on IPA analysis.The SRS model of a front-to-front collision accident for this national road does not distinguish whether the cause is due to loss of control so that it enters the opposite lane or due to failure to anticipate when overtaking another vehicle.This condition is based on the fact that most of Indonesia's national road sections have type's 2/2-UD and 4/2-UD which have not been designed to have an ideal road width and median.
Based on Table 10, a number of attributes are mapped into Quadrant-IV, namely road shoulders, shoulder width, grade, super elevation, skid resistance, median traversability, volume of motorcycles and bicycles.In fact, it is still very possible to consider the attributes that have been defined as influential attributes, given the importance and difficulty values are very close to the average value.Thus, the likelihood attributes for front-to-front collision accidents based on IPA analysis are road type, lane width, R-bend, curve quality, pavement condition, road surface condition, speed limit signs, delineation, rumble strip.Meanwhile, the selected severity attributes are road median and effective road width.The attribute of the external factor in this SRS model is traffic volume.This front-to-front crash accident SRS model incorporates an operational speed factor given in the form of an 85%-tile speed as an influencing factor.

Table 8 Importance Level Analysis and Application of Road Assessment Attributes for Front-Front Crash Accident Parameters
Table 9 below shows the results of the IPA analysis of the SRS model attributes for accidents at intersections.Based on the IPA analysis, a number of attributes are mapped into Quadrant-IV.These attributes are categorized as having a low level of importance and a high level of difficulty in implementation based on the perceptions of road safety experts in Indonesia.Attributes that are not considered in this SRS model are lane width, pavement condition, skid resistance, median type, motorbike volume, bicycle volume.These six attributes are seen by a number of experts as attributes that have little influence on the SRS model for traffic accidents at intersections.Attributes that influence the SRS model of intersection accidents which are seen as the likelihood of an accident include right-turning lanes, visibility, grade, canalization, traffic control lights (APIL), speed limit signs, and speed reduction devices.The median factor based on IPA analysis is seen as not an influential factor considering that traffic accidents on national road sections predominantly occur at small intersections with 2/2-UD road types.
The results of this IPA analysis also show that traffic volume factors including motorcycle and bicycle traffic volume are seen as having no effect on the SRS model of accidents at intersections.It is possible that traffic accidents are dominant at small unregulated intersections.Generally, accidents occur at Y-junctions or T-junctions which have poor visibility.These intersections generally have not been designed ideally because many of them are found on corners that have less than ideal visibility.

DISCUSSION
Overall the results of the IPA Quadrant analysis of a number of attributes for each National Road SRS calculation produce a number of attributes from each parameter which are divided into SRS attributes for front-rear crash parameters (13 attributes), SRS parameters for front-side crash accidents when turning around (13 attribute); The road attributes are the same for several SRS types of accidents, so that in total 43 attributes were selected for National Road SRS as shown in Table 12.Based on the analysis and sharing of statistical tests as well as analysis of the level of importance and level of application as well as the design of the National Road SRS model, in general it shows the final model of the special National Road SRS from the perspective of passenger vehicle users.The resulting model as shown in Figure 4 has significant differences compared to the SRS iRAP model.The difference between the National Road SRS model and the iRAP SRS model is determined by the type of accident as the main parameter and the attributes of each of these parameters.The National Road SRS model provides two completely new parameters, namely front-front crash accident parameters and the type of front-side crash accident which especially occurs when one of the vehicles makes a U-turn.

Figure 2 SRS Car Occupant Model for National Road
In addition to these two parameters, there is one SRS parameter for National Roads which actually has a different type of accident but has the same parameter name, namely the front-side crash accident parameter in property access.The SRS iRAP model property access accident parameters are not specifically stated as to the typical accidents that are dominant at the property access.Meanwhile, the accident parameters at property access on the National Road SRS model are very much based on the typical dominant accidents on national roads, namely in the form of front-side crash accidents.Therefore, the accident parameter on property access in the National Road SRS model emphasizes the dominance of typical front-side crash accidents on the property access.The inclusion of these three accident parameters in the National Road SRS model brings a number of consequences with new attributes according to the type of accident.For other accident parameters, it was also found that there were several new attributes that were not yet available in the RPS model or the SRS iRAP model.In general, the calculation of special National Road SRS from the perspective of passenger vehicle users/drivers is based on statistical analysis.

CONCLUSION
Basically, the Star Rating Scores (SRS) calculation model or Road Protector Scores (RPS) calculation model was developed from typical dominant accidents, especially from road elements and the road environment.With the same concept, this research develops an SRS calculation model for Indonesian national roads, especially from the perspective of motor vehicle users with 4 or more wheels, also based on the typical dominant accidents on national roads.There are 6 typical accidents that are dominant on national roads which are closely related to the condition of road infrastructure based on an analysis of 283,158 traffic accident data from 2012 to 2019.The six types of accidents are rear-end collisions, rear-end collisions.front-to-front (head-on collisions), front-to-side collisions (head to side collisions) both at property access and at U-turn locations, single accidents leaving the road (run-off accidents), accidents at intersection.This SRS model takes into account 2 different main parameters and 4 main parameters that are the same as the main parameters of the SRS International Road Assessment Program (iRAP) and a total of 51 road attributes.The two main parameters that differ from the SRS iRAP model are the parameters for front-rear crashes and fore-side crashes when turning.While the 4 parameters are the same as frontside collision accidents at property access, single accidents leaving the road body, front-to-front collision accidents, and accidents at intersections.
Based on the Importance and Performance Analysis (IPA) method, it was possible to formulate 43 road attributes for the National Road SRS model, consisting of 30 likelihood factor attributes, 10 severity factor attributes, 2 traffic external influence factor attributes, and 1 operational speed factor attribute.Of the 43 attributes there are several attributes used in several SRS models, so that out of the 43 attributes each is divided into 13 attributes of front-rear crash accident SRS, 13 attributes front-side SRS when turning, 13 front-side SRS property access as many as 14 attributes, single accident SRS from road bodies as many as 18 attributes, SRS front-to-front collision as many as 13 attributes, and SRS at intersections as many as 10 attributes.

Figure 1
Figure 1 Design of the SRS Calculation Model for National Road Sections from the Perspective of Motorized Vehicle Drivers with 4 or More Wheels American Association of State Highway and Transportation Officials (AASHTO), (1997), Highway Safety Design and Operation Guide, American Association of State Highway and Transportation Officials, Washington, DC 20001 American Association of State Highway and Transportation Officials (AASHTO), (2010), Highway Safety Manual, 1 th Edition, American Association of State Highway and Transportation Officials, Washington, DC 20001

Table ,
then the hypothesis accepts H o. Conversely, if W count < W Table, the hypothesis rejects Ho or accepts Hi.