An Influencing Factors Analysis of Road Traffic Accidents Based on the Analytic Hierarchy Process and the Minimum Discrimination Information Principle

Abstract

Safe traffic is an important part of sustainable transportation. Road traffic accidents lead to a large number of casualties and property losses every year. Current research mainly studies some types of traffic accidents and ignores other types of traffic accidents; therefore, taking various types of road traffic accidents as a whole, an overall study of their influencing factors is urgently needed. To improve road traffic safety, taking various types of road traffic accidents as a whole, this paper analyzes the influencing factors and finds out the causative factors of road traffic accidents. A new index system of road traffic accident influencing factors is constructed based on the existing literature and real traffic data, and their subjective weights and objective weights are obtained by the analytic hierarchy process based on the subjective data and the normalization of the actual traffic data for Yizheng City, Yangzhou, China from January 2020 to December 2020, where the subjective weights are the main weights, and comprehensive weights are obtained by the minimum discrimination information principle correcting the subjective weights with the objective weights. Finally, the global weights, their ranks, and their weight differences are obtained. The main findings are as follows: (1) compared with the real traffic data, experts generally overestimate the impact of road factors on traffic accidents and underestimate the impact of human factors on traffic accidents; (2) in the first-level, human factors and road factors are the causative factors; (3) in the second-level, “motor vehicle drivers’ misconduct”, “road condition”, and “road section” are the causative factors; and (4) in the third-level, “slippery road”, “rain and snow weather”, “intersection”, and “untimely braking” are the causative factors. The research results can provide some scientific basis for improving road traffic safety.

1. Introduction

At present, the world’s road traffic safety situation is not optimistic. More than 33.5 million people have died in road traffic accidents worldwide since road traffic accident data became available. Every year, about 1.35 million people die in road traffic accidents, while up to 50 million people are injured, and the economic loss is about 518 billion US dollars, accounting for about 1% to 3% of the world GDP [1]; the impact on emerging countries is even more severe, with economic losses of around 5% of GDP in these countries [2].

The number of traffic accidents shows an increasing trend year by year, which has led to a large number of casualties and property losses every year [3], and therefore many scholars have studied the factors influencing various types of road traffic accidents to improve road traffic safety. However, road traffic accidents mainly comprise traffic accidents between motor vehicles, non-motor vehicles, pedestrians, and passengers, excluding crashes between people, and the current research mainly studies some types of traffic accidents and ignores other types of traffic accidents; therefore, taking various types of road traffic accidents as a whole, an overall study of their influencing factors is urgently needed.

In order to identify the new safety influencing factors, confirm the known influencing factors, obtain the weighting and ranking of the influencing factors, find the causative factors, and improve road traffic safety, based on the existing literature, real traffic accident data in Yizheng City, Yangzhou, China in 2020 and the effect evaluation of traffic accident influencing factors by traffic safety experts, this paper analyzes the factors influencing various types of road traffic accidents as a whole and finds out the causative factors for road traffic accidents.

The contributions of this paper are as follows: (1) Based on the actual data and the existing literature, from the perspective of human–vehicle–road–environment, we take various types of road traffic accident influencing factors as a whole and add the relevant indicators of non-motor vehicle drivers and pedestrians and passengers, and build a new hierarchical model of the factors influencing traffic accidents. (2) Using the analytic hierarchy process and the normalization of the actual traffic data, the subjective and objective weights are obtained and compared. (3) Using the principle of minimum discriminant information, we correct the subjective weight obtained by AHP based on the subjective data with the objective weight based on the objective data and obtain the comprehensive weight. (4) According to the global weight, the rankings of the influencing factors and the causative factors at all levels are obtained.

The other parts of this paper are as follows. The second section reviews the relevant literature. The third section simply describes and analyzes the subjective data and the real traffic data. The fourth section puts forward the research method. The fifth section constructs the new hierarchical model of road traffic accident influencing factors and calculates various weights. The sixth section is the discussion of the results, including the new hierarchical model of influencing factors, the subjective and objective weight difference and the causative factors at all levels. The seventh section is the conclusion and research prospect of this paper.

2. Literature Review

Road traffic safety has become the focus of all countries in the world, and traffic accidents are one of the most typical concerns in the field of road traffic safety. In order to improve the safety of the road traffic system, many scholars have studied the factors influencing various types of road traffic accidents based on some research methods.

2.1. Study on Road Traffic Accident Influencing Factors

A traffic system is a complex system composed of human–vehicle–road–environment, and various traffic factors have an important impact on the safety of a traffic system. We review the relevant research from the perspective of human–vehicle–road–environment, providing a basis for the construction of a new index system.

2.1.1. Research on Human-Related Factors

Drivers’ bad driving states have an important influence on traffic accidents. Driving experience and driving age have an important influence on traffic accidents [4,5]. Inexperienced drivers are more likely to be involved in traffic accidents, and more training and education for inexperienced drivers is necessary [5]. The proportion of elderly drivers in traffic accidents is higher, the mortality rate is higher, and older drivers are three to twenty times more likely to be involved in a fatal collision than non-older drivers for the same driving distance [6,7]. Driver emotion has a significant impact on driver behavior, which in turn has a significant impact on traffic accidents [8,9]. Drivers with higher anger characteristics are more likely to drive angrily on the road, make more dangerous driving behaviors, and cause traffic accidents [8]. Drunk driving affects drivers’ perception and response, leading to an increase in traffic accidents [10,11]. Both the driver’s reaction time increases and proper steering control ability decreases [12,13] and drivers are more likely to be involved in fatal traffic accidents [14] under fatigue driving.

Drivers’ dangerous driving behavior is one of the main causes of traffic accidents [10,15]. Shao et al. [15] analyzed the response mechanism of drivers to warnings considering driver characteristics and real-time driving risk level and compared with normal and conservative drivers, impulsive drivers follow the car in front of them closer and the probability of slowing down is lower. Palk et al. [16] found that young drivers often engage in anti-social and dangerous driving behaviors such as illegal street racing and speed testing on public roads, and these behaviors show a positive correlation with crash events. Urban bus drivers’ behaviors have an important impact on traffic accidents. Zhu et al. [17] studied the relationship between urban bus drivers’ rule violation behaviors such as running red lights, not yielding to pedestrians, changing lanes suddenly, not renewing driver’s licenses in time, and bus crashes.

Motorcyclists, cyclists, pedestrians, and passengers account for approximately 50% of the deaths of vulnerable road users worldwide, and some scholars have studied the impact of unsafe driving behaviors of non-motor vehicle drivers on traffic accidents [18]. Research on the behavior of non-motor vehicle drivers of express delivery vehicles mainly focuses on some illegal driving behaviors such as red-light violations [19,20] and not wearing a safety helmet [21]. Qian et al. [22] compared the accident injury severity of e-bikes with that of other types of two-wheelers based on accident data and analyzed the influencing factors such as “not riding in the prescribed lane” and “turning in”. Oviedo-Trespalacios et al. [20] compared the riding behavior of food delivery and private bicycle riders.

The factors influencing pedestrian and passenger traffic accidents have also been well studied, based on which traffic accident countermeasures have been proposed [22,23,24,25]. The number of passengers carried by non-motor vehicles has an impact on the severity of passenger and non-motor vehicle traffic accidents [22]. Pedestrian and passenger distraction is an important cause of pedestrian and passenger traffic accidents [23,25]. Haleem et al. [24] used the actual traffic accident data in Florida, United States, and analyzed the important factors affecting the severity of pedestrian injury at both signalized and unsignalized intersections. Based on actual traffic data, Zeng et al. [25] put forward a model to study the severity of vehicle–pedestrian collision injuries at urban intersections and found the main factors affecting the severity of pedestrian injuries, such as illegal crossing lanes and illegally crossing the traffic barrier.

2.1.2. Research on Vehicle-Related Factors

Some scholars have studied the impact of vehicle safety conditions such as tire blowout, steering failure and brake failure on traffic accidents [26]. In many crashes, the impact of vehicle defects is obvious [26], among which tire blowout and vehicle brake failure are more important vehicle accident influencing factors [27,28]. Haq et al. [29] studied the severity of crashes and injuries associated with brake failure on mountain roads in Wyoming. Some scholars have studied the impact of vehicle safety hazards on traffic accidents. Excessive speed is an important factor leading to traffic accidents [30,31,32]. Speeding is a complex problem involving various traffic factors, such as drivers’ attribute behavior, road design and characteristics, and traffic management [30]. Vehicle weight is closely related to traffic accidents. In China, a small number of heavy vehicles often lead to traffic accidents. Wang et al. [33] proposed a traffic conflict risk measurement method considering the influence of vehicle weight. Bunn et al. [34] studied driver injuries in traffic accidents involving heavy trucks versus light and medium trucks in the United States from 2010 to 2019.

2.1.3. Research on Road-Related Factors

Some scholars have studied the influence of road sections such as uphill and downhill, sharp curves, and intersections on traffic accidents. Wang et al. [33] studied the heterogeneity of traffic accident collision mechanisms among different road types. Road geometric alignment has an important impact on traffic accidents [10,35]. In order to reduce traffic accidents, the geometric alignment design of roads needs to comprehensively consider various factors. For example, when the radius of the horizontal curve is less than 200 m, the risk of traffic accidents increases [36]. Wen et al. [37] studied the comprehensive impact of slope and curve on the severity of truck traffic accidents on mountain expressways. It is crucial to understand which geometric design elements will affect driver expectations and lead to unpredictability of the road. Sharp turns lead to more traffic accidents [10,38], and the injury degree of traffic accidents is often more serious. Pedestrian traffic fatalities often occur at intersections, and many scholars have studied the causes of pedestrian traffic accidents at intersections [39,40]. Some scholars have studied the influence of road conditions such as a slippery road surface, road construction, and traffic signs on traffic accidents. Poor road conditions or rainy days will increase road traffic accidents [35], and slippery roads in winter significantly increase road accidents [41]. Good road conditions are conducive to reducing traffic accidents [42,43]. Zhang et al. [43] studied the impact of pavement construction on traffic accidents, and the research results showed that with a high traffic volume (e.g., >20,000 vehicles per day) and medium length (e.g., 2000 to 5000 m), the causal effect of the working area on the occurrence of collision was significantly positive. Traffic signs such as warning signs and speed limit signs have an important impact on traffic safety on rural curved roads [44]. Uddin et al. [45] showed that on rainy days, variable speed limit signs should be used to reduce the speed of trucks.

2.1.4. Research on Environment-Related Factors

Weather conditions have an important impact on traffic accidents, such as rain, snow, fog, and other bad weather conditions are prone to traffic accidents [45,46]. Uddin et al. [45] established the logit model under normal weather, rainy days, and snowy days to study the severity of truck collision injuries under different weather conditions. Abdel-Aty et al. [47] studied crashes when fog and smoke caused visibility barriers. In order to ensure traffic safety, Wang et al. [21] studied the reasonable speed limit in the dynamic low-visibility foggy environment by using simulation driving research, and drew some conclusions, such as the speed limit of 30 km/h when the visibility was between 35 m and 60 m. Some scholars have studied the effect of sight conditions on traffic accidents. Lighting conditions are important factors affecting traffic accidents [37,48]. Low road visibility causes a large number of traffic accidents, and visibility is one of the most important environmental factors affecting traffic accidents [21,49]. Due to the low visibility in fog, drivers’ behavior and performance will be negatively affected by foggy weather conditions, which can easily lead to traffic accidents [50,51].

As mentioned above, in order to improve the safety of a road traffic system, many scholars have studied the factors influencing various types of road traffic accidents from the perspective of human–vehicle–road–environment; however, road traffic accidents are mainly composed of traffic accidents between motor vehicles, non-motor vehicles, pedestrians, and passengers except for crashes between people, and the current research mainly studies some types of road traffic accidents and ignores other types of road traffic accidents. Therefore taking various types of road traffic accidents as a whole, an overall study of their influencing factors is urgently needed.

2.2. Influencing Factors Analysis Based on the Analytic Hierarchy Process

Proposed by Saaty [52], the analytic hierarchy process (AHP) is a flexible multi-criteria decision-making method, which transforms complex problems into a hierarchy of influencing factors. Many scholars have used AHP to analyze the weight of various decision-making or result-influencing factors; for example, scholars have used the AHP to study the factors influencing tourism decision-making [53,54]; the factors influencing transportation logistics decision-making [55,56]; and the factors influencing various safety accidents [57,58,59,60]. Some scholars have used AHP to study the factors influencing traffic accidents in various travel modes. These include the factors influencing railway accidents [61,62,63]; for example, in recent years, the increasing trips on the European railway network endanger the safety of operators and passengers, and Sangiorgio et al. [63] used AHP and put forward a new index system to evaluate the safety performance level of the railway transport system. The demand for the transport of dangerous goods is gradually increasing worldwide [64], and the loss caused by transport accidents of dangerous goods is $70 billion per year [65]. One of the most important problems in the transport of dangerous goods by highway is the truck and its safety conditions. Ghaleh et al. [66] used the fuzzy analytic hierarchy process (FAHP) to study the safety risk-contributing factors (SRCFS) and sub-safety risk-contributing factors (Sub-SRCFS) of road tanker trucks and their weights. Karahalios [67] used fuzzy sets and AHP to design a scorecard to identify the key points of ship accident prevention, and studied ship collision accidents. Yoo et al. [68] used AHP to determine the relative importance of various influencing factors in airport passenger security inspection, so as to improve airport passenger security inspection. The results of AHP showed that human resources were the most important factor to improve passenger security inspection performance. Manca et al. [69] proposed a quantitative evaluation method of highway tunnel emergency preparedness and response based on AHP, and determined the hierarchical structure of influencing factors necessary to measure the performance of road tunnel accident emergency response systems and evaluated the relative importance of the influencing factors.

As mentioned above, the analytic hierarchy process (AHP) is a flexible multi-criteria decision-making method, which transforms the complex problems into a hierarchy of influencing factors, and therefore many scholars have used AHP to analyze the weight of various decision-making or result-influencing factors. However, the weight of influencing factors obtained by AHP is a subjective weight, which needs to be revised by the objective weight based on the actual data; at the same time, the actual data cannot reflect the importance of the decision-makers to different influencing factors, and since it is impossible to obtain all objective data, the objective weight obtained based on the objective data is somewhat different from the actual weight. To this end, this paper takes the subjective weight obtained by AHP based on the subjective data as the main weight, and modifies the traditional hierarchical model of road traffic accident influencing factors and corrects the subjective weights of the factors affecting the traffic accidents with the real data using the principle of minimum discriminant information, which can correct the subjective weight obtained by AHP based on the subjective data with the objective weight based on the objective data and obtain the comprehensive weight, and these are also the innovation in the application of AHP and some innovations of this paper.

The occurrence of road traffic accidents is the result of bad human conditions and unsafe behavior, unsafe vehicle states, dangerous road factors, and unfavorable environmental factors. Therefore, to improve traffic safety, it is necessary to analyze the factors influencing road traffic accidents and find the causative factors of traffic accidents. Current research mainly studies some types of traffic accidents and ignores other types of traffic accidents; therefore, taking various types of road traffic accidents as a whole, an overall study of their influencing factors is urgently needed.

To this end, taking various types of road traffic accidents as a whole and with the traffic accident data for Yizheng City, Yangzhou, China from January 2020 to December 2020, which are good quality detailed data including the main types of traffic accidents such as road traffic accidents between motor vehicles, non-motor vehicles, pedestrians, and passengers but excluding crashes between people and road traffic accident human–vehicle–road–environment information, this paper takes the subjective weight obtained by AHP based on the subjective data as the main weight, and uses the analytic hierarchy process to obtain the subjective weight, normalizes the real traffic data to obtain the objective weight, uses the minimum discrimination information principle to correct the subjective weight with the objective weight to obtain the comprehensive weight, and finally obtains the global weight and its ranking of road traffic accident influencing factors, obtains the causative factors of traffic accidents, and provides a scientific basis for improving traffic system safety.

3. Data

3.1. The Subjective Data

The subjective weight obtained by AHP is the main weight and based on the subjective data. According to the calculation steps of the analytic hierarchy process, the subjective data are obtained by selecting and inviting ten traffic safety experts from universities and traffic safety departments to judge the relative importance of influencing factors at all levels according to their own at least 10 years of expertise and experience in traffic safety. More specifically, we designed an “urban road traffic accident influencing factors relative importance questionnaire”, sent the questionnaire to the experts who judged the relative importance of influencing factors at all levels with the relative importance scale of the AHP, and then collected the questionnaire, and finally the relative importance judging results of the influencing factors at all levels of the experts were averaged, and then the subjective data were obtained.

The “urban road traffic accident influencing factors relative importance questionnaire” contains three main parts. The first part of the questionnaire introduces the background and purpose of the questionnaire, and expert selection criteria; the second part of the questionnaire briefly introduces the contents of the questionnaire, namely, urban road traffic accident influencing factors at all levels of this paper and the relative importance scale of AHP; and the third part of the questionnaire is the judgment matrixes of urban road traffic accident influencing factors at all levels waiting for the experts to judge according to their own at least 10 years of expertise and experience in traffic safety and the relative importance scale of the AHP. The judgment matrixes waiting for the experts to judge and fill include one first-level indicator judgment matrix, four second-level indicator judgment matrixes and ten third-level indicator judgment matrixes, and after the experts’ judgment and filling, 150 judgment matrixes were collected in total.

3.2. The Objective Data

The purpose of the objective data: The objective data in this paper are mainly used to determine the factors influencing traffic accidents by combining the existing literatures [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51], and calculate the objective weight to correct and compare with the main weight, namely, the subjective weight obtained by the AHP based on the subjective data.

The representative dataset criteria: The samples with good quality detailed data include the main types of traffic accidents such as traffic accidents between motor vehicles, non-motor vehicles, pedestrians, and passengers but excludes crashes between people and road traffic accident human–vehicle–road–environment information. These samples are a representative dataset and suitable for the study of this paper and can realize the research goal of this paper.

According to the purpose of the objective data and the representative dataset criteria, the objective data on road traffic accidents used by this paper were collected from Yizheng City, Yangzhou, China from January 2020 to December 2020, and the total number of accidents was 101. The sample size was a compromise between the purpose of the objective data and obtaining a representative dataset, and combining with the sample size of the existing studies such as [70,71,72], we can obtain a feasible objective data sample size for this paper.

Traffic accident information is classified according to human–vehicle–road–environment aspects in

Table 1. Summary of human information.

Attribute	Category	Quantity	Attribute	Category	Quantity
Motor vehicle drivers in bad condition	Fatigue driving	7	Driving experience of motor vehicle drivers	≤5 years	16
	Drunk driving	9		6~8 years	20
	Emotional driving	10		9~14 years	23
Motor vehicle drivers’ misconduct	Driving without a license	6		15~19 years	13
	Illegal U-turn	5		>20 years	15
	Illegal overtaking	8	Age of motor vehicle drivers	≤25 years old	12
	Illegal lane change	16		26~30 years old	14
	Traffic signal violation	10		31~40 years old	29
	Failure to maintain a safe distance	10		41~50 years old	23
	Not yielding to pedestrians at zebra crossings	9		50~60 years old	6
	Untimely braking	37		>60 years old	5
Non-motor vehicle driver factors	Swerve	10	Pedestrian and passenger factors	Illegal crossing lanes	5
	Crossing the road	12		Illegally crossing the traffic barrier	7
	No safety helmet	22		Traffic signal violation	4
	Occupying motor vehicle lanes	4		Not observing traffic environment	11

,

Table 2. Summary of vehicle information.

Attribute	Category	Quantity	Attribute	Category	Quantity
Vehicle safety condition	Tire burst	4	Vehicle safety hazard	Overloaded	5
	Steering failure	6		Over speed	18
	Brake failure	13		Large truck	22

,

Table 3. Summary of road information.

Attribute	Category	Quantity	Attribute	Category	Quantity
Pavement condition	Dry	61	Road section	Flat straight section	55
	Wet and slippery	38		Uphill and downhill section	10
Construction situation	Road construction	14		Sharp turn section	8
	No road construction	87		Intersection	28
Traffic sign	There are traffic signals or lines	79
	Lack of traffic signals	22

and

Table 4. Summary of environment information.

Attribute	Category	Quantity	Attribute	Category	Quantity
Weather condition	Clear Weather	26	Sight condition	Day	75
	Overcast sky	33		Lighting at night	11
	Rainy and snowy day	34		No lighting at night	15
	Foggy weather	8		Visibility less than 100 m	8

, where “Quantity” means the corresponding number of traffic accidents.

Humans in the traffic system can mainly be divided into four categories: motor vehicle drivers, non-motor vehicle drivers, pedestrians, and passengers. According to the road traffic accident classification of the National Bureau of Statistics of China [73], traffic accidents caused by pedestrian and passenger factors are considered together. The information on motor vehicle drivers includes the driver’s age and driving experience, and the driver’s bad conditions and improper driving behavior. The summary information is shown in Table 1.

Vehicle information includes vehicle safety conditions and vehicle safety hazards, as shown in Table 2.

Road information includes the road condition and the type of road section. The type of road section can be subdivided into flat straight section, uphill and downhill road section, sharp turn road section, and intersection road section. For details, see Table 3.

Environment information includes weather condition and sight condition, as shown in Table 4.

4. Method

4.1. Analytic Hierarchy Process Calculation Steps

(1) Construct the judgment matrix $A$

The matrix $A$ is constructed as follows: indicators $a_i$ and $a_j$ are compared pairwise; $a_{ij}$ indicates the importance of indicator $a_i$ relative to indicator $a_j$ and $1/a_{ij}$ indicates the importance of indicator $a_j$ relative to indicator $a_i$, which means $a_{ji}=$ $1/a_{ij}$ and $a_{ii}=1.$ If there are $n$ indicators, a judgment matrix of order $n$ will be formed, denoted as $A={\left(a_{ij}\right)}_{n\mathrm{*}n}$. The analytic hierarchy process usually uses a digital scale of 1–9 to represent the relative importance, as shown in

Table 5. Relative importance scale of AHP [74].

Scale	Degree of Importance
1	Equally important
3	Moderately important
5	Strongly important
7	Very strongly important
9	Extremely important
2, 4, 6, 8	Intermediate values

.

For example, if $a_i$ is more important than $a_j$, $a_{ij}$ is assigned 9; then $a_j$ is less important than $a_i$, and $a_{ji}$ is 1/9, and the judgment matrix $A$ is as follows:

$$A=\left[\begin{array}{ccc}\begin{array}{cc}{a}_{11}& a_{12}\\ a_{21}& a_{22}\end{array}& \cdots & \begin{array}{c}{a}_{1n}\\ a_{2n}\end{array}\\ \begin{array}{cc}⋮& ⋮\end{array}& \ddots & ⋮\\ \begin{array}{cc}{a}_{n1}& a_{n2}\end{array}& \cdots & a_{nn}\end{array}\right]$$

(1)

(2) The eigenvector ${W^s=\left[W_{1,}^s{W}_{2,\dots ,}^s{W}_n^s\right]}^T$ of the judgment matrix $A$ is solved by the root method:

$$W_i^s={\displaystyle \frac{\sqrt[n]{\prod _{j=1}^n{a}_{ij}}}{\sum _{i=1}^n\left(\sqrt[n]{\prod _{j=1}^n{a}_{ij}}\right)}}(i=1,2,\dots ,n)$$

(2)

where $W_i^s$ is the subjective weight of the $i$th indicator and $a_{ij}$ is the element of the judgment matrix $A$.

(3) Calculate the maximum eigenvalue ${\lambda }_{max}$ of the judgment matrix $A$:

$${\lambda }_{max}={\sum }_{i=1}^n{\displaystyle \frac{B_i{W}^s}{n{W}_i^s}}$$

(3)

where $B_i$ is the ith row vector of the judgment matrix $A$.

(4) The consistency index $CI$ of judgment matrix $A$:

The judgment matrix $A$ is the result of the indicators’ relative importance as judged by experts, whose judgements are to be somewhat imperfect or inconsistent. Therefore in order to check for the extent of the consistency of the judgment matrix $A$, the consistency index CI should be calculated as follows:

$$CI={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}$$

(4)

where $n$ is the dimension of the judgment matrix $A$.

(5) The consistency ratio $CR$ of judgment matrix $A$ is calculated as follows:

$$CR={\displaystyle \frac{CI}{RI}}$$

(5)

where $RI$ is a random index, which is related to the dimension $n$ of the judgment matrix.

Table 6. Standard values of random index $RI$ [75].

$\mathit{n}$	1	2	3	4	5	6	7	8	9	10
$RI$	0	0	0.58	0.90	1.12	1.24	1.32	1.41	1.45	1.49

shows the corresponding relationship.

Under normal circumstances, when $CR<0.1$, it is considered that the judgment matrix $A$ passes the consistency test, otherwise it is necessary to reconstruct the judgment matrix until it passes the consistency test.

4.2. Principle of Minimum Discrimination Information

As mentioned above, the weight of influencing factors obtained by the AHP based on the subjective data is a subjective weight, which needs to be revised based on the objective data. Therefore, this paper uses the principle of minimum discrimination information to correct the subjective weight with the objective weight based on the objective data, and obtains the comprehensive weight. Based on the subjective weight $W_i^s$ determined by the analytic hierarchy process and the objective weight $W_i^o$ obtained by the normalization of the accident data, the comprehensive weight $W_i^c$ closest to the subjective weight and the objective weight is determined by the principle of minimum discrimination information, and the corresponding objective function is established as follows:

$$minF={\sum }_{i=1}^n{W}_i^c\left[\ln {\displaystyle \frac{W_i^c}{W_i^s}}\right]+{\sum }_{i=1}^n{W}_i^c\left[\ln {\displaystyle \frac{W_i^c}{W_i^o}}\right]$$

(6)

Among them, the comprehensive weight $W_i^c>0$ and satisfies $\sum _{i=1}^n{W}_i^c=1$.

The Lagrange function is generally used to solve the value of the minimum discrimination information:

$$L\left(W_i^c,\lambda \right)={\sum }_{i=1}^n{W}_i^c\left[\ln {\displaystyle \frac{W_i^c}{W_i^s}}\right]+{\sum }_{i=1}^n{W}_i^c\left[\ln {\displaystyle \frac{W_i^c}{W_i^o}}\right]-\lambda \left({\sum }_{i=1}^n{W}_i^c-1\right)$$

(7)

When extreme values exist, the following applies:

$$\left\{\begin{array}{c}{\displaystyle \frac{\partial L}{\partial W_i^c}}=\left(\ln {\displaystyle \frac{W_i^c}{W_i^s}}+1\right)+\left(\ln {\displaystyle \frac{W_i^c}{W_i^o}}+1\right)-\lambda =0\\ {\displaystyle \frac{\partial L}{\partial \lambda }}=\sum _{i=1}^n{W}_i^c-1=0\end{array}\right.$$

(8)

We have the following:

$$W_i^c={\displaystyle \frac{\sqrt{W_i^s{W}_i^o}}{\sum _{i=1}^n\sqrt{W_i^s{W}_i^o}}}$$

(9)

5. Hierarchical Model and Weight Calculation of Influencing Factors

5.1. Hierarchical Model of Road Traffic Accident Influencing Factors

According to the existing literature [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51] and the actual traffic data and its summary information in Table 1, Table 2, Table 3 and Table 4, the factors influencing road traffic accidents can be selected, and the hierarchical model of factors influencing road traffic accidents can be built. Specifically, this paper takes urban road traffic accidents as the target layer and based on the existing literature and the real traffic accident data and the analysis of influencing factors, the human factors, vehicle factors, road factors, and environmental factors are taken as the first-level indicators, and then the human factor is subdivided into four second-level indicators: motor vehicle drivers’ bad conditions, motor vehicle drivers’ misconduct, non-motor vehicle drivers’ unsafe behavior, and unsafe behavior by pedestrians and passengers. Among the human information, “driving experience ≤ 5 years” is defined as a lack of experience, and “age > 60 years” is defined as old and infirm. Vehicle factors are subdivided into two second-level indicators: vehicle safety condition and vehicle safety hazard. Road factors are subdivided into two second-level indicators: road section and road condition. Environmental factors are subdivided into two second-level indicators: weather condition and sight condition. Then, the third-level indicators are divided, and finally a new hierarchical model of factors influencing urban road traffic accidents is built, as shown in

Table 7. The hierarchical model of factors influencing urban road traffic accidents.

Result	First-Level Influencing Factors	Second-Level Influencing Factors	Third-Level Influencing Factors	Selected Studies
Urban road traffic accidents	Human factor U1	Motor vehicle drivers’ bad condition U11	Inexperience U111	[4,5]
			Old and infirm U112	[6,7]
			Emotional driving U113	[8,9]
			Drunk driving U114	[10,11]
			Fatigue driving U115	[12,13,14]
		Motor vehicle drivers’ misconduct U12	Driving without a license U121	[17]
			Illegal U-turn U122	[16]
			Illegal overtaking U123	[16]
			Illegal lane change U124	[17]
			Traffic signal violation U125	[17]
			Failure to maintain a safe distance U126	[15]
			Not yielding to pedestrians at zebra crossings U127	[17]
			Untimely braking U128	[15]
		Non-motor vehicle drivers’ unsafe behavior U13	Swerve U131	[22]
			Crossing the road U132	[19,20]
			No safety helmet U133	[20,21]
			Occupy motor vehicle lanes U134	[22]
		Unsafe behavior by pedestrians and passengers U14	Illegal crossing lanes U141	[25]
			Illegally crossing the traffic barrier U142	[25]
			Traffic signal violation U143	[24]
			Not observing traffic environment U144	[23,25]
	Vehicle factor U2	Safety condition U21	Tire burst U211	[27,28]
			Steering failure U212	[26]
			Brake failure U213	[29]
		Safety hazard U22	Over speed U221	[30,31,32]
			Overloaded U222	[33]
			Large truck U223	[34]
	Road factor U3	Road section U31	Uphill and downhill section U311	[37]
			Sharp turn section U312	[10,38]
			Intersection U313	[37,38]
		Road condition U32	Slippery road U321	[41]
			Pavement construction U322	[42,43]
			Traffic sign problem U323	[44,45]
	Environment factor U4	Weather condition U41	Rain and snow U411	[45,46]
			Foggy U412	[21,47]
		Sight condition U42	No lighting at night U421	[37,48]
			Visibility below 100 m U422	[21,49,50,51]

.

5.2. Weight Calculation of Road Traffic Accident Influencing Factors

5.2.1. Analytic Hierarchy Process Determines the Subjective Weight

According to the calculation steps of the analytic hierarchy process and the subjective data, the judgment matrixes of indicators at all levels such as the one of first-level of indicators under the target layer shown in

Table 8. Judgment matrix for first-level influencing factors.

	U1	U2	U3	U4
U1	1	3	2	2
U2	1/3	1	1/3	1
U3	1/2	3	1	4
U4	1/2	1	1/4	1

were obtained. The following is a calculation example of first-level indicators under the target layer using Equations (1)–(5). The calculation of other indicator weights can refer to this process.

(1) The judgment matrix of the first-level indicators under the target layer is constructed, as shown in Table 8.

(2) By computing the fourth root of the product of each row’s elements of the judgment matrix $A$ and the normalization process

$$\left[\begin{array}{ccc}1& 3& \begin{array}{cc}2& 2\end{array}\\ \begin{array}{c}1/3\\ \begin{array}{c}1/2\\ 1/2\end{array}\end{array}& \begin{array}{c}1\\ \begin{array}{c}3\\ 1\end{array}\end{array}& \begin{array}{c}\begin{array}{cc}1/3& 1\end{array}\\ \begin{array}{c}\begin{array}{cc}1& 4\end{array}\\ \begin{array}{cc}1/4& 1\end{array}\end{array}\end{array}\end{array}\right]\stackrel{\mathrm{the}\mathrm{fourth}\mathrm{root}}{\to }\left[\begin{array}{c}\begin{array}{c}1.8612\\ 0.5774\end{array}\\ 1.5651\\ 0.5946\end{array}\right]\stackrel{\mathrm{the}\mathrm{normalization}\mathrm{process}}{\to }\left[\begin{array}{c}\begin{array}{c}0.405\\ 0.126\end{array}\\ 0.340\\ 0.129\end{array}\right]$$

We obtain the subjective weights $W_i^s(i=1,2,3,4)$ of the first-level indicators as 0.405, 0.126, 0.340, and 0.129, respectively.

(3) Calculate the maximum eigenvalue ${\lambda }_{max}$

$${\mathsf{\lambda }}_{\max }={\displaystyle \frac{1}{4}}\left({\displaystyle \frac{1.7208}{0.405}}+{\displaystyle \frac{0.5032}{0.126}}+{\displaystyle \frac{1.4367}{0.340}}+{\displaystyle \frac{0.5423}{0.129}}\right)=4.169$$

(4) Calculate the consistency index $CI$

$$CI={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}={\displaystyle \frac{4.169-4}{4-1}}=0.056$$

(5) Calculate the consistency ratio $CR$

$$CR={\displaystyle \frac{CI}{RI}}={\displaystyle \frac{0.056}{0.9}}=0.062$$

Because $CR=0.062<0.1$, the judgment matrix $A$ of the first-level influencing factors has a reasonable level of consistency. According to the above calculation steps, the subjective weights of the second-level and third-level influencing factors are solved, whose consistency tests of judgment matrixes are shown in

Table 9. The consistency test of judgment matrixes for the second-level and third-level influencing factors.

Judgment Matrixes	${\mathit{\lambda }}_{\mathit{max}}$	CI	RI	CR
U11–U14	4.233	0.078	0.9	0.086
U21–U22	2	0	0	0
U31–U32	2	0	0	0
U41–U42	2	0	0	0
U111–U115	5.306	0.076	1.12	0.068
U121–U128	8.949	0.136	1.41	0.096
U131–U134	4.184	0.061	0.9	0.068
U141–U144	4.121	0.04	0.9	0.045
U211–U213	3.054	0.027	0.58	0.046
U221–U223	3.054	0.027	0.58	0.046
U311–U313	3.094	0.047	0.58	0.081
U321–U323	3.104	0.052	0.58	0.089
U411–U412	2	0	0	0
U421–U422	2	0	0	0

.

Table 9 shows all of the CR values are less than 0.1, which means the second-level and third-level influencing factors judgment matrixes all have a reasonable level of consistency and the subjective weights obtained by AHP are shown in

Table 10. Weight, its rank, and weight difference for the first-level influencing factors.

First-Level Factors	Subjective (Global) Weight	Objective (Global) Weight	Comprehensive Weight	First-Level Global Weight	Rank	Weight Difference
U1	0.405	0.468	0.437	0.437	1	−0.063
U2	0.126	0.143	0.135	0.135	4	−0.017
U3	0.340	0.252	0.294	0.294	2	0.088
U4	0.129	0.137	0.134	0.134	3	−0.008

,

Table 11. Weight, its rank, and weight difference for the second-level influencing factors.

Second-Level Factors	Subjective Weight	Subjective Global Weight	Objective Weight	Objective Global Weight	Comprehensive Weight	Second-Level Global Weight	Rank	Weight Difference
U11	0.194	0.078	0.211	0.099	0.202	0.089	6	−0.021
U12	0.429	0.174	0.453	0.212	0.441	0.193	1	−0.038
U13	0.230	0.093	0.215	0.101	0.223	0.098	4	−0.008
U14	0.147	0.060	0.121	0.057	0.134	0.058	8	0.003
U21	0.333	0.042	0.338	0.048	0.336	0.045	10	−0.006
U22	0.667	0.084	0.662	0.095	0.664	0.089	5	−0.011
U31	0.4	0.136	0.383	0.097	0.392	0.115	3	0.039
U32	0.6	0.204	0.617	0.155	0.608	0.179	2	0.049
U41	0.667	0.086	0.646	0.088	0.656	0.088	7	−0.002
U42	0.333	0.043	0.354	0.048	0.344	0.046	9	−0.005

and

Table 12. Weight, its rank, and weight difference for the third-level influencing factors.

Third-Level Factors	Subjective Weight	Subjective Global Weight	Objective Weight	Objective Global Weight	Comprehensive Weight	Third-Level Global Weight	Rank	Weight Difference
U111	0.372	0.029	0.340	0.034	0.356	0.032	11	−0.005
U112	0.110	0.009	0.106	0.011	0.109	0.010	34	−0.002
U113	0.226	0.018	0.213	0.021	0.220	0.019	21	−0.003
U114	0.146	0.011	0.191	0.019	0.167	0.015	26	−0.008
U115	0.146	0.011	0.149	0.015	0.148	0.013	28	−0.004
U121	0.072	0.013	0.059	0.013	0.066	0.013	30	0
U122	0.087	0.015	0.050	0.011	0.066	0.013	29	0.004
U123	0.085	0.015	0.079	0.017	0.083	0.016	24	−0.002
U124	0.173	0.030	0.158	0.034	0.167	0.032	10	−0.004
U125	0.128	0.022	0.099	0.021	0.114	0.022	17	0.001
U126	0.116	0.020	0.099	0.021	0.108	0.021	18	−0.001
U127	0.099	0.017	0.089	0.019	0.095	0.018	22	−0.002
U128	0.241	0.042	0.366	0.078	0.300	0.058	4	−0.036
U131	0.217	0.020	0.208	0.021	0.213	0.021	19	−0.001
U132	0.258	0.024	0.25	0.025	0.254	0.025	16	−0.001
U133	0.434	0.040	0.458	0.046	0.446	0.044	7	−0.006
U134	0.091	0.009	0.083	0.008	0.087	0.009	36	0.001
U141	0.160	0.010	0.185	0.011	0.173	0.010	32	−0.001
U142	0.227	0.013	0.259	0.015	0.243	0.014	27	−0.002
U143	0.160	0.010	0.148	0.008	0.154	0.009	35	0.002
U144	0.453	0.027	0.407	0.023	0.430	0.025	15	0.004
U211	0.184	0.008	0.174	0.008	0.179	0.008	37	0
U212	0.232	0.010	0.261	0.013	0.246	0.011	31	−0.003
U213	0.584	0.025	0.565	0.027	0.575	0.026	13	−0.002
U221	0.396	0.033	0.4	0.038	0.398	0.036	8	−0.005
U222	0.105	0.009	0.111	0.011	0.108	0.010	33	−0.002
U223	0.499	0.042	0.489	0.046	0.494	0.044	6	−0.004
U311	0.225	0.031	0.217	0.021	0.221	0.025	14	0.010
U312	0.165	0.023	0.174	0.017	0.170	0.020	20	0.006
U313	0.610	0.083	0.609	0.059	0.609	0.070	3	0.024
U321	0.550	0.112	0.514	0.080	0.532	0.095	1	0.032
U322	0.189	0.039	0.189	0.029	0.189	0.034	9	0.01
U323	0.261	0.053	0.297	0.046	0.279	0.050	5	0.007
U411	0.8	0.069	0.810	0.071	0.805	0.071	2	−0.002
U412	0.2	0.017	0.191	0.017	0.195	0.017	23	0
U421	0.667	0.029	0.652	0.032	0.659	0.030	12	−0.003
U422	0.333	0.014	0.348	0.017	0.341	0.016	25	−0.003

. The subjective weight obtained by the AHP based on the subjective data is the main weight, which needs to be corrected by the objective weight based on the actual data.

5.2.2. Data Normalization Determines the Objective Weight

Based on the accident summary information in Table 1, Table 2, Table 3 and Table 4 and the traffic accident hierarchy model in Table 7, the objective weight is calculated in a normalized way and the specific step is to divide the number of accidents caused by the indicator at this level by the number of accidents caused by the corresponding indicator at the upper level, and the objective weights obtained are shown in Table 10, Table 11 and Table 12.

5.2.3. The Principle of Minimum Discrimination Information Determines the Comprehensive Weight

In order to correct the subjective weights of the factors affecting the traffic accidents with the real data, we used the principle of minimum discriminant information to correct the subjective weight obtained by the AHP based on the subjective data with the objective weight based on the objective data to obtain the comprehensive weight. According to the subjective weights obtained by the analytic hierarchy process and the objective weights obtained by the normalization of the actual traffic data and Equation (9), the comprehensive weight of all levels of indicators can be obtained, as shown in Table 10, Table 11 and Table 12.

5.2.4. Weight, Its Rank, and Weight Difference of Road Traffic Accident Influencing Factors

In Table 10, Table 11 and Table 12, the global weight calculation ideas of the corresponding column are as follows: The subjective global weight of the indicator at this level $=$ the subjective global weight of the corresponding indicator at the upper level $\times$ the subjective weight of the indicator at this level. The objective global weight of the indicator at this level $=$ the objective global weight of the corresponding indicator at the upper level $\times$ the objective weight of the indicator at this level. The global weight of the indicator at this level $=$ the global weight of the corresponding indicator at the upper level $\times$ the comprehensive weight of the indicator at this level. In addition, the ranking is based on the global weight, and weight difference $=$ the subjective global weight at this level $-$ the objective global weight at this level.

Table 10 shows the weight, its rank, and the weight difference for the first-level influencing factors.

According to the above the global weight calculation of the corresponding column, Table 10 can be obtained. More specifically, take the weights of U1 for example. Its subjective global weight in the second column is 0.405 $=$ 1 $\times$ 0.405; its objective global weight in the third column is 0.468 $=$ 1 $\times$ 0.468; its global weight in the fifth column is 0.437 $=$ 1 $\times$ 0.437, and according to the global weight, it ranks first in the sixth column; and in the last column, its weight difference is −0.063 $=$ 0.405 − 0.468. It is noted that for the first-level indicators, their upper level is the target layer urban road traffic accident, and for the target layer urban road traffic accident, its various weights are 1, and therefore, in the second column, the subjective global weight $=$ 1 $\times$ the subjective weight of the indicator at this level; in the third column, the objective global weight $=$ 1 $\times$ the objective weight of the indicator at this level; in the fifth column, the first-level global weight $=$ 1 $\times$ the comprehensive weight of the indicator at this level. Table 10 shows that among the first-level influencing factors, human factor (U1) and road factor (U3) are the two most important factors affecting traffic accidents, and the global weights of traffic accidents are 43.7% and 29.4%, respectively. For the weight difference, compared with the objective weights, experts underestimated the impact of human factors on traffic accidents the most, reaching 6.3%; and experts overestimated the impact of road factors on traffic accidents the most, reaching 8.8%.

Table 11 shows the weight, its rank, and the weight difference for the second-level influencing factors.

According to the above the global weight calculation of the corresponding column, Table 11 can be obtained. More specifically, take the weights of U11 for example. Its subjective global weight in the third column is 0.078 $=$ 0.405 $\times$ 0.194; its objective global weight in the fifth column is 0.099 $=$ 0.468 $\times$ 0.211; its global weight in the seventh column is 0.089 $=$ 0.437 $\times$ 0.202, and according to the global weight, it ranks sixth in the eighth column; and in the last column, its weight difference is −0.021 $=$ 0.078 $-$ 0.099. Table 11 shows that among the second-level influencing factors, “motor vehicle drivers’ misconduct (U12)”, “road condition (U32)”, and “road section (U31)” are the causative factors, and their global weights of traffic accidents are 19.3%, 17.9%, and 11.5%, respectively. For the weight difference, compared with the objective weights, the experts underestimated the impact of motor vehicle drivers’ misconduct (U12) on traffic accidents the most, reaching 3.8%, and experts overestimated the impact of road conditions (U32) and road section (U31) on traffic accidents the most, reaching 4.9% and 3.9%, respectively.

Table 12 shows the weight, its rank, and the weight difference for the third-level influencing factors.

According to the above the global weight calculation of the corresponding column, Table 12 can be obtained. More specifically, take the weights of U111 for example. Its subjective global weight in the third column is 0.029 $=$ 0.078 $\times$ 0.372; its objective global weight in the fifth column is 0.034 $=$ 0.099 $\times$ 0.34; its global weight in the seventh column is 0.032 $=$ 0.089 $\times$ 0.356, and according to the global weight, it ranks eleventh in the eighth column; and in the last column, its weight difference is −0.005 $=$ 0.029 − 0.034. Table 12 shows among the third-level influencing factors, “slippery road (U321)”, “rain and snow (U411)”, “intersection (U313)”, and “untimely braking (U128)” are the causative factors, and their global weights of traffic accidents are 9.5%, 7.1%, 7%, and 5.8%, respectively. For the subjective and objective weight difference, compared with the objective weights, experts underestimated the impact of untimely braking (U128) on traffic accidents the most, reaching 3.6%. Experts overestimated the impact of slippery road (U321) on traffic accidents the most, reaching 3.2%.

6. Discussion

6.1. Hierarchical Model of Influencing Factors

As described above, road traffic accidents are mainly composed of traffic accidents between motor vehicles, non-motor vehicles, pedestrians, and passengers but not crashes between people, and the current research mainly studies some types of traffic accidents and ignores other types of traffic accidents. Therefore, taking various types of road traffic accidents as a whole, an overall study of their influencing factors is urgently needed. To this end, based on the existing literature [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51] and the actual data, from the perspective of human–vehicle–road–environment, we take various types of road traffic accident influencing factors as a whole and add the relevant indicators of non-motor vehicle drivers and pedestrians and passengers, and build a new hierarchical model of factors influencing traffic accidents as shown in Table 7, which is one of our contributions to the current studies.

The classification method of the National Bureau of Statistics of China [73] and the data for Yizheng City, Yangzhou, China from January 2020 to December 2020 consider traffic accidents caused by pedestrians and passengers together. The reasons for this might be that some accidents mainly occur when passengers are not yet in the vehicle and that pedestrians and passengers often mix together, and therefore the impact of pedestrians and passengers on traffic safety is often difficult to distinguish; therefore, for simplicity and without a loss of generality, traffic accidents caused by pedestrians and passengers are considered together, which is also the reason why the same classification method is used in this paper.

6.2. Subjective and Objective Weight Difference

In Table 10, Table 11 and Table 12, using the analytic hierarchy process and the normalization of the actual traffic data, the subjective and objective weights are obtained. In determining the weight according to the decision-maker’s awareness, a subjective weighting method such as the AHP has more advantages than an objective weighting method, but its objectivity is relatively poor and subjectivity is relatively strong. Analyzing objective data to obtain the objective weight can overcome this defect and has objective advantages, but it cannot reflect the importance that decision-makers attach to different indicators. The subjective and objective weight difference can reflect some characteristics of traffic accident influencing factors in real traffic. Therefore, based on Table 10, Table 11 and Table 12, this paper analyzes and discusses the subjective and objective weight difference.

Among the global weight of the first-level influencing factors in Table 10, the weight difference for the environmental factor is −0.8% $=$ 12.9% − 13.7%, which is almost no difference, and it indicates that the experts’ estimation of the impact of environmental factors on traffic accidents is very consistent with the actual traffic data. The weight difference for vehicle factors is also relatively small, at −1.7% $=$ 12.6% − 14.3%. The most significant difference between experts’ emphasis on an influencing factor and objective data is the weight difference for road factors, followed by the one for human factors, reaching 8.8% $=$ 34% − 25.2% and −6.3% $=$ 40.5% − 46.8%, respectively, which indicate that the experts emphasize the impact of road factors and the experts’ estimation of human safety awareness is a little high. That may be because in real traffic, the road factors of traffic accidents are significant and often present to experts, and although the number of traffic accidents caused by road factors is less than that caused by human factors, traffic accidents caused by road factors are often more serious and impress the experts, such as the serious traffic accident caused by the road collapse on the Meizhou Expressway in Guangdong Province, China on 1 May 2024, while the human factors of traffic accidents are often not significant, accidental, and unpredictable, and do not often present to and impress experts. In this paper, the subjective and objective weights of environmental factors are 12.9% and 13.7%, respectively, indicating that although the weight difference is very small, environmental factors have an important impact on traffic safety, which is consistent with the existing research [46,76,77]; the subjective and objective weights of vehicle factors are 12.6% and 14.3%, respectively, indicating that although the weight difference is very small, vehicle factors have an important impact on traffic safety, which is consistent with the existing research [27,28,29].

Among the second-level influencing factors in Table 11, the subjective and objective weights of most indicators are not much different, and there are some differences for the subjective and objective weights of motor vehicle drivers’ bad condition (U11) and motor vehicle drivers’ misconduct (U12), whose weight differences are −2.1% $=$ 7.8% − 9.9% and −3.8% $=$ 17.4% − 21.2%, respectively, and it may be because although these factors cause many accidents, they are often not significant, accidental, and unpredictable and do not often present to experts, and therefore experts underestimate them. For road section (U31) and road condition (U32), their weight differences are 3.9% $=$ 13.6% − 9.7% and 4.9% $=$ 20.4% − 15.5%, respectively, and it may be because these factors contain some of the more significant factors that often cause traffic accidents and impress the experts, such as uphill and downhill section [37], sharp turn section [10,38], intersection [37,38], slippery road [41], pavement construction [42,43], and traffic sign problem [44,45].

Among the third-level influencing factors in Table 12, for untimely braking (U128), compared with the real traffic data, the weight of experts underestimating is 3.6%, and that may be because untimely braking is determined by many hidden factors such as driver fatigue, drunk driving, and distracted driving, and its specific performance is a long perception–reaction time, and for a driver who is driving, it is difficult to judge whether she/he brakes in time or not and for experts, they rarely experience or encounter traffic accidents caused by not braking in time. Compared with the real traffic data, for intersection (U313), the weight of experts overestimating is 2.4%, and for the slippery road (U321), the weight of experts overestimating is 3.2%. This indicates that compared with real traffic data, the experts pay more attention to the impact of the intersection and the slippery road on traffic accidents and that may be because drivers, including experts, drive through intersections almost every day and have seen or personally experienced traffic accidents that occur at intersections, so it is not surprising that experts highlight the impact of intersections on traffic accidents; roads can become slippery for a variety of common reasons, such as rain and snow, and slippery roads have led to a significant increase in traffic accidents [41,78], which also impressed experts.

To sum up, we can find an interesting phenomenon that, in general, experts’ opinions on the impact of vehicle factors and environmental factors on traffic accidents are consistent with the real traffic data, but compared with real traffic data, they generally overestimate the impact of road factors on traffic accidents and underestimate the impact of human factors on traffic accidents. Possible explanations for the above phenomenon are as follows: in real traffic, before traveling, drivers will check the environmental factors, for example, weather conditions such as rain, snow, or fog and sight conditions such as visibility, through the Internet, and will also regularly check vehicle factors, for example, vehicle safety conditions such as tires, steering, and braking conditions, and pay attention to vehicle safety hazards on the road such as speeding, overloading, and large trucks when driving, which makes the impact of vehicle factors and environmental factors on traffic accidents more certain and measurable. However, the influence of human factors and road factors on traffic accidents is random and unpredictable, which are some important reasons for the small difference between the subjective weights and objective weights of vehicle factors and environmental factors at all levels, and the larger difference between the subjective weights and objective weights of human factors and road factors at all levels. In addition, the subjective weights of road factors are generally greater than their objective weights, while the subjective weights of human factors are generally less than their objective weights and it may be because in real traffic, the road factors of traffic accidents are significant and often present to experts, and although the number of traffic accidents caused by road factors is less than that caused by human factors, traffic accidents caused by road factors are often more serious and impress the experts, such as the serious traffic accident in which 48 people died due to road collapse on the Meizhou Expressway in Guangdong Province, China on 1 May 2024. On the other hand, the human factors of traffic accidents are often not significant, accidental, and unpredictable and do not often present to and impress the experts.

As described above, using the analytic hierarchy process and the normalization of the actual traffic data, the subjective and objective weights are obtained and compared, which is one of our contributions to the current studies.

6.3. Causative Factors

6.3.1. First-Level Causative Factors

Human and road factors are the main factors in the first-level influencing factors, and the global weights are 43.7% and 29.4%, respectively. A traffic system is a complex system composed of human–vehicle–road–environment. The human is the main body of this system, who is both the perpetrator and the victim of traffic accidents. Human factors are also the most active factors in the system, and the most direct factors affecting traffic accidents, which is consistent with previous studies [2,79] and is also consistent with real traffic such as the serious traffic accident caused by human factors in which eight people died and one person was injured on State Route 510, Weichang Manchu Mongolian autonomous county, Hebei Province, China on 19 April 2024. One difference between this paper and most previous studies is that taking various types of road traffic accidents as a whole, this paper carries out the overall study of their influencing factors including motor vehicle drivers, non-motor vehicle drivers, pedestrians, and passengers. Many aspects of human factors can affect traffic accidents, such as motor vehicle drivers’ bad conditions [4,7,9], motor vehicle drivers’ misconduct [10,15], non-motor vehicle drivers’ unsafe behavior [80], and unsafe behavior by pedestrians and passengers [22,23,24,25].

Road factors are the second most important causative factors of traffic accidents in the first-level influencing factors, which is consistent with the existing research [33,36,37]. Road factors also include many contributing factors such as road section [37,39] and road condition [41,44]. In the complex traffic system of human–vehicle–road–environment, “human” is the center and “road” is the foundation. In the process of driving on the road, the information used by the driver mainly comes from the road section and the road condition, and the driving behavior choices are made through the drivers’ perception and judgment of the information. Any mistake in driving behavior choices may cause traffic accidents, and the number of traffic accidents often increases with the increase in the number of the driving behavior choices mainly related to the road section and the road condition information. As the road traffic infrastructure and the basic condition of vehicle driving, road factors play an important role in many accidents such as the serious traffic accident caused by road factors in which 13 people died that occurred at the Xijiata tunnel of G59 Hubei Expressway (Beihu direction), Shanxi Province, China on 19 March 2024, which is well reflected in the research of this paper.

6.3.2. Second-Level Causative Factors

The causative factors in the second-level influencing factors are “motor vehicle drivers’ misconduct (U12)”, “road condition (U32)”, and “road section (U31)”. The global weight of “motor vehicle drivers’ misconduct (U12)” is 19.3%, which is the most important causative factor in the second-level influencing factors. The driver’s behavior is different due to the heterogeneity of driver attributes such as driving state and obtaining and processing information ability. When the driver’s psychology and behavior are consistent with the actual traffic situation, the driver’s behavior is safe and reliable, and when they are not consistent, the driver’s behavior is unsafe. This result is consistent with the actual traffic. In the actual traffic, the driver directly controls the vehicle, the driver’s driving behavior is directly reflected by the vehicle, and the vehicle is very sensitive to the driver’s driving behavior, and therefore a minor behavioral error by the driver can cause a major traffic accident, which is also consistent with the existing research [10,15,81].

The second-level influencing factor “road condition (U32)” is the second most important causative factor, with a global weight of 17.9%. Poor road conditions will increase road traffic accidents [35]. Good road conditions are one of the preconditions for safe traffic, which is conducive to reducing traffic accidents [42,43]. This is consistent with the actual traffic phenomenon, in the actual traffic, bad road conditions will induce the improper driving behavior of drivers; for example, when the road is slippery, many drivers will make wrong driving behaviors, and sometimes when road conditions are very bad, even if the driver’s driving behavior is correct, traffic accidents will occur, for example, when there is pavement construction or a traffic sign problem, traffic accidents very easily happen. This is also consistent with previous studies [35,42,43,82].

The second-level influencing factor “road section (U31)” is the third most important causative factor, and the global weight is 11.5%. In actual traffic, road sections are the infrastructure of road traffic and the fundamental conditions for vehicle running, and the “non-mobility of accident-prone road sections” characteristic of road traffic accidents reflects the important impact of road sections on traffic accidents. Intensive traffic accidents are usually associated with road sections, which can be called accident-prone points or black spots. The accident-prone points are closely related to road design. Due to insufficient information provided by road design to drivers, or road sections inconsistent with drivers’ visual and psychological reactions or against drivers’ expectations, the driver’s reaction time increases, the driver is too late to deal with unexpected information or makes mistakes in judgment, and finally traffic accidents occur. This is also consistent with previous studies [10,33,35].

6.3.3. Third-Level Causative Factors

The most important causative factor in the third-level influencing factors is “slippery road (U321)”, and its global weight reaches 9.5%. A slippery road can be caused by rain and fog [35] or snow in winter [41,78], which are also the third-level indicators of environmental factors. The wet road reduces the friction on the ground, and the maximum speed limit of the vehicle will drop rapidly, but some drivers with low safety driving awareness will not consciously reduce the driving speed, resulting in traffic accidents, and when the road is extremely wet, even if the driver reacts correctly, there will be a big traffic accident, such as the one on 23 February 2024, in Suzhou, China, in which the icy wet road caused a collision of hundreds of vehicles. This finding is different from the one of Clarke et al. [81]. Some useful measures can reduce the negative impact of slippery roads on traffic accidents; for example, DiLorenzo et al. [41] showed that in Finland, the VMS (variable message signs) relayed road conditions to the drivers, such as when the road was slippery.

“Rain and snow (U411)” is the second most important causative factor in the third-level influencing factors, and its global weight reached 7.1%. Rain and snow will make the road surface slippery, and the visibility of the road environment such as road signs will be greatly reduced [82]. As a result, the driver’s perception accuracy of the surrounding environment will be greatly reduced, and the driver’s reaction ability and accuracy will also be reduced, which will easily lead to traffic accidents. This is consistent with existing studies [45,46,83]. Zhai et al. [83] found the hot and rainy day increased the effects of jaywalking and risky driving behavior on crash severity, and the possibility of injury or death in traffic accidents. Yasanthi et al. [46] found slight snow substantially reduced the reliability of the current speed limit.

Traffic accidents often occur at intersections in real traffic. On the one hand, intersections are the places where motor vehicles, non-motor vehicles and pedestrians cross and the traffic situation is very complicated [84]; on the other hand, they are also places where traffic subjects compete for the right of way. Some traffic subjects, such as pedestrians, may rush to pass through [39,40,85] and often do not comply with the right-of-way assignment of traffic lights, leading to traffic accidents, which is in line with the result of this paper that “intersection (U313)” is the third most important causative factor among the third-level influencing factors, and the global weight reaches 7.0%, which is also consistent with previous studies [39,40].

Among the third-level influencing factors, the last one with a global weight of more than 5% is “untimely braking (U128)”, which should be the most direct third-level influencing factor leading to the rear-end traffic collision. Many traffic factors can cause untimely braking, such as drunk driving, fatigue driving, distracted driving, and so on, which lead to an increase in the driver’s perception–reaction time, slow braking, and eventually traffic accidents [86]. If braking in time is performed under the existing braking technology conditions and a safe distance, traffic collisions will not occur naturally, and improving the existing braking technology and reducing the driver’s perception–reaction time, such as through the development of a higher level of autonomous vehicles, will also reduce traffic accidents, but under the current braking technology conditions, untimely braking is a major cause of traffic accidents, which is in line with the existing study [15].

As described above, using the principle of minimum discriminant information, we correct the subjective weight obtained by the AHP based on the subjective data with the objective weight based on the objective data and obtain the comprehensive weight; according to the comprehensive weight, we obtain the global weight, and finally according to the global weight, the rankings of the influence factors and the causative factors at all levels are obtained, which are some our contributions to the current studies.

The results of the study are generalizable. This year, 2020, was affected by COVID-19, but it does not mean the results of the studies using traffic data during the COVID-19 pandemic are not generalizable, such as the results of [15,20,22]. In this paper, the subjective weight obtained by the AHP is the main weight and based on the subjective data, and the objective data in this paper are mainly used to determine the factors influencing traffic accidents by combining the existing literatures [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51], and calculate the objective weights to correct and compare with the subjective weights obtained by the AHP based on the subjective data. Although in 2020 the real traffic data may be affected by the COVID-19 pandemic, the existing literature [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51] and the subjective data are not much affected by COVID-19, because the literature years of the existing literature [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51] from 1995 to 2023 and the subjective data are obtained by inviting ten traffic safety experts from universities and traffic safety departments to judge the relative importance of influencing factors at all levels according to their own at least 10 years of expertise and experience in traffic safety, not only their expertise and experience in traffic safety in 2020 or during the COVID-19 pandemic, which means the results of this study are generalizable.

The factors influencing traffic accidents determined by the objective data are consistent with the factors influencing traffic accidents determined by the existing literature [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51], and the differences between the subjective weights obtained by the AHP based on the subjective data and the objective weights obtained by the real traffic data are not very large, which also verify that the objective data sample size is feasible and the results of the study are generalizable.

7. Conclusions

Road traffic accidents are mainly composed of traffic accidents between motor vehicles, non-motor vehicles, pedestrians, and passengers, but not crashes between people, and current research mainly studies some types of traffic accidents and ignores other types of traffic accidents. Therefore, taking various types of road traffic accidents as a whole, the overall study of their influencing factors is urgently needed. To this end, this paper analyzes the factors influencing various types of road traffic accidents as a whole and finds out the causative factors of road traffic accidents. We obtained some insights and conclusions, and it is noted that the data in this paper do not include highway traffic accident data, so the insights and conclusions obtained in this paper are not necessarily suitable for highway traffic conditions. The results indicate experts’ opinions on the impact of vehicle factors and environmental factors on traffic accidents are generally consistent with the actual data, respectively, but compared with the real traffic data, they generally overestimate the impact of road factors on traffic accidents and underestimate the impact of human factors on traffic accidents. The global weights and their ranking of traffic accident influencing factors are obtained, which show that in the first-level influencing factors, human factors and road factors are the two most important factors affecting traffic accidents, and their global weights of traffic accidents are 43.7% and 29.4%, respectively. In the second-level influencing factors, “motor vehicle drivers’ misconduct”, “road condition”, and “road section” are the causative factors, and their global weights of traffic accidents are 19.3%, 17.9%, and 11.5%, respectively. In the third-level influencing factors, “slippery road”, “rain and snow”, “intersection”, and “untimely braking” are the causative factors, and their global weights of traffic accidents are 9.5%, 7.1%, 7.0%, and 5.8%, respectively. The research results can provide some scientific basis for traffic managers to formulate traffic management policies to reduce traffic accidents and improve traffic system safety.

Although this paper has carried out a detailed influencing factors analysis of urban road traffic accidents, there are still some aspects for further research:

(1) There are numerous factors affecting urban road traffic accidents, and it is necessary to further study the literature and collect more road traffic accident data to further improve and supplement the new index system and build a more comprehensive index system of influencing factors.

(2) Although it is feasible to use the analytic hierarchy process to determine the subjective weight, the normalization of accident data to obtain the objective weight and the minimum discrimination information principle to obtain the comprehensive weight, a further research direction could be to introduce more comprehensive methods to determine various weights.

(3) Furthermore, a robust analysis is also one of the future research directions.