Assessing Significant Factors Affecting Risky Riding Behaviors of Vietnamese Motorcyclists Using a Contextual Mediated Model

Tis study explores the signifcant factors afecting risky riding behaviors of Vietnamese motorcyclists using a contextual mediated model (CMM) in Hanoi, the capital of Vietnam, where motorcycle crashes are prevalent. Te afecting factors include personality traits, riding self-confdence, riding attitude, and risk perception. Personality traits and riding self-confdence are distal factors of CMM that afect risky riding behaviors. On the other hand, riding attitude and risk perception are proximal factors in CMM. A survey was conducted to collect information onmotorcyclists’risky riding behaviors relatedto thefour factorsmentionedthrough a self-reportedquestionnaire.Statistical PackageSocialScience(SPSS)andstructuralequationmodeling(SEM)withanalysisofmomentstructures(AMOS)areusedtodetermine theefectsofthefactorsonriskyridingbehaviors.Teresultsdiscoveredthatridingattitudeandriskperceptionweretheintermediate variablesofpersonalitytraitandridingself-confdenceafectingtheriskyridingbehaviors,andpersonalitytraitandridingself-confdence alsoafectedtheriskyridingbehaviorsdirectly.Findingsinthemodelalsoshowthatridingattitudewasperceivedtoplayasignifcantrole inincreasingriskydrivingbehavior.Terecommendationistoincreasethesafetyeducationprogramsthatreduceriskydrivingbehavior.


Introduction
Many countries in Southeast Asia, including Vietnam, heavily rely on motorcycles for transportation.Tis is because motorcycles have several advantages over cars, such as being cheaper, easier to park, faster to navigate urban roads, and more fuel-efcient [1].However, crashes and fatality rates on motorcycles are higher than in other transportation modes, primarily passenger cars.In Hanoi (Vietnam), there are over 700 registered motorcycles per 1,000 people, which account for more than 80% of the trips taken [2].Te World Health Organization (WHO) [3] reported that motorcycles are related to 43% of all road crash deaths in Southeast Asia.In Vietnam, road trafc accidents (RTAs) involving motorcycles account for over 66.7%, and human factors account for 71.6% [4].Even in developed countries, motorcycle-related road crashes are still a recurring problem with no solution.
In contrast to passenger cars, motorcycles are less safe and vulnerable due to the absence of protective devices and the high number of motorcycles on the road in mixed trafc conditions.Motorcyclists are also considered to have more risk-taking and speeding behaviors [5], easily leading to road crashes.Te previous studies found that road crashes related to motorcycles have been associated with risky behaviors and afecting factors of the risky behaviors of motorcyclists [1,6].It is essential to identify and understand riding behavior that helps predict motorcyclists' actions on the road, identify crash mechanisms and causes, and then control crashes to improve their health.
Contextual mediated model (CMM) using a self-reported questionnaire survey is an efective tool to explore efecting factors on riding behavior found in many studies related to car drivers.So far, research on the risky riding behaviors of motorcycles is still far from universal compared with other road users such as cars and lacks general consideration of various contributing factors [7].Most studies on motorcycle behavior using psychological factors are also mainly conducted in developed countries, where a few motorcycles with considerable power are used.Tere are limited studies on the riding behaviors of motorcyclists in developing countries like Vietnam [1].
Numerous factors can impact the risky behavior of motorcyclists.Tese may include inadequate motorcycle maintenance, insufcient motorcycle facilities, age group, lack of attention, careless riding, failure to follow trafc rules, riding attitude, self-confdence while riding, perception of risk, personal preferences, and personality traits.Exploring all the factors infuencing driving behavior in a model would be more complicated and unnecessary than focusing on the key factors.Terefore, this study emphasizes Vietnamese motorcyclists' personality traits, riding selfconfdence, riding attitude, and risk perception of selfreported performed risky riding behavior using the CMM model.Terefore, a complete understanding of the hazardous behaviors of motorcyclists is necessary to establish a plan for improving trafc safety.
Te remainder of this study is structured as follows: Section 2 introduces the literature review, Section 3 presents the materials and methods utilized, Section 4 describes the study's results along with statistical testing and discussion, and Section 5 provides conclusions and future work.

Literature Review
Engaging in risky riding behavior, such as speeding, riding without a helmet, and other violations, is considered unacceptable in public due to the potential negative consequences of negligence [8,9].Since risky riding behavior is closely linked to motorcycle accidents, it is crucial to identify and comprehend the factors that contribute to such behavior.Studies have endeavored to defne and recognize risky riding behavior by combining various elements and assessing their interactions, such as personality traits, riding self-confdence, riding attitude, and risk perception, among others, using diferent models, including the CMM model.
Te contextual mediated model (CMM) of risky behavior was frst proposed by Elander et al. [10] and generalized by Sümer [11] based on the horizontal model of Lajunen [12].Te CMM points out that the combination of various distal and proximal contextual factors in certain conditions can lead to risky behaviors and crashes.Te proximal factors mainly emphasize the factors that directly contribute to risky behaviors or accidents.Te distal factors mainly emphasize social background, lifestyle, daily habits, and other factors.Elander et al. [10] described a relationship among personality traits, driving behaviors, and crash risk using CMM.Iversen and Rundmo [13] found that risky behaviors (e.g., speeding) partially mediated the impacts of personality on accident involvement in Norway.Sümer [11] proposed an indirect connection between distal factors (e.g., demographic traits and individual diferences) and outcomes (e.g., accidents) through the proximal environment (e.g., driver actions).Zhang et al. [14] investigated a relationship among driving anger, aberrant behaviors, and crash risk based on CMM.According to the study, abnormal behaviors were found to be the main factor that caused an increase in the risk of crashes due to driving anger.Just like the risk of crashing or getting into an accident, aggressive driving can also be considered an outcome in the absence of a crash [15].According to Ullerberg and Rundmo [6], distal factors afect the outcomes of a CMM involving personality, attitude, and behavior.Several studies have used or expanded the model of Ullerberg and Rundmo, and their results indicated that the framework of road trafc safety was helpful from a CMM model [16][17][18][19][20]. Te CMM can describe a relation between distal factors, proximal factors, and outcomes.We can easily increase variables and change the roles of variables in diferent CMM models.Analyzing shows that previous studies often use only a few latent variables in the CMM models in various areas with diferent study samples under trafc fow conditions predominated by cars.Tus, in the context of Vietnam's transportation, it is vital to construct a CMM model that suits motorcycles for better analysis of motorcycle accidents, training behavior of motorcyclists, and intervention.
Tere have been limited studies in Vietnam that focus on the behaviors of motorcyclists and the factors that afect them.For instance, Trinh and Vo [21] used a combination of the theory of planned behavior and the health belief model to explain the relationship between drunk driving, illegal changing of direction, and speeding, but they did not focus on motorcyclists.However, Trinh and Le [22] found a signifcant link between speeding and helmet-wearing among motorcyclists using the theory of planned behavior.Truong et al. [23] examined the relationship between phone use while riding and other risky behaviors among university students on motorcycles, with calling while riding being the most common risky behavior.Nguyen-Phuoc et al. [24] investigated the risky behaviors of app-based motorcycle taxi riders.Tey found that phone use while riding was the most frequent, followed by failing to use turn signals, encroaching on car lanes, exceeding speed limits, running red lights, and carrying more than one passenger.Meanwhile, Luot et al. [25] found no correlation between risk behaviors and risk perception among Vietnamese motorcyclists.Li et al. [26] conducted a cross-sectional study in Ho Chi Minh City and found an increase in the use of improper and poor-quality helmets, which is one of four main issues (i.e., drink-driving, motorcycle helmets, seat belts, and speed management) implemented by road safety projects in 10 low-and middleincome countries [27].Bui et al. [28] provided a Motorcycle Rider Behavior Questionnaire (MRBQ) with 36 items adapted from the original MRBQ of Elliott et al.However, they did not consider the factors infuencing risk behavior in a given model.Nguyen-Phuoc et al. [29] investigated the factors that afect the frequency of turn signals at intersections, such as environmental characteristics, perceived risk, and beliefs.In addition, Nguyen-Phuoc et al. [30] examined the factors related to the risk-taking attitude and road safety compliance of motorcycle delivery riders, such as job demands and resources.Vuong et al. [4] explored riding attitudes and attempted to explain the relationship between 2 Journal of Advanced Transportation the detected factors and the riding attitudes of motorcyclists.Lastly, Can et al. [31] considered the reliability and validity of a risk perception scale of motorcyclists in Vietnam.Generally, in recent years, some scholars have been interested in and tried to discover the risky behaviors of motorcyclists in Vietnam.However, they mainly focused on given violations and have not provided comprehensive models or a general picture of the risky riding behaviors of motorcyclists in the transport context of Vietnam.
In our study, we utilized a deductive approach by leveraging the results of prior research [15,32] and engaging with industry experts through qualitative interviews to construct a theoretical framework using the CMM model.Tis model signifcantly emphasizes fve key elements-personality traits, riding attitudes, risk perception, and riding self-confdence-to comprehensively analyze the efects of risky riding behaviors on road trafc safety in developing nations like Vietnam.Our ultimate objective is to propose efective safety measures and interventions that can meaningfully reduce the incidence of dangerous riding practices among motorcyclists.

Formulation of the Teoretical Model.
In this study, the proposed theoretical model using CMM considers personality traits, riding self-confdence, riding attitude, and risk perception as factors contributing to risky riding behaviors among motorcyclists.Te frst two factors (i.e., personality traits and riding self-confdence) are distal factors, and the remaining factors (i.e., riding attitude and risk perception) are proximity factors of the proposed CMM, as shown in Figure 1.We review the considered factors and propose the research hypotheses of the proposed model as follows: 3.1.1.Personality Trait.Te most studied personality traits are those that cause individuals to disregard or underestimate the dangers of the road, leading to reckless and unsafe behavior.Tese traits are characterized by an increased probability of engaging in risky riding practices and being involved in accidents.Several scales of personality traits are used in the context of risky behaviors, such as the Revised NEO Personality Inventory [33], the Big Five Questionnaire [34], the Zuckerman-Kuhlman Personality Questionnaires Cross Cultural [35], and normlessness [36].For instance, personality traits in the Big Five Questionnaire have comprised extraversion, neuroticism, agreeableness, openness to experience, and conscientiousness; the NEO Personality Inventory measures excitement-seeking, anxiety, and aggression.In addition, many scales of personality traits have combined and developed from the abovementioned scales [6,16,[37][38][39][40].
Previous studies have indicated that certain individual characteristics can signifcantly encourage unsafe motorcycling behaviors [41,42].Tese traits can exert both a direct and indirect infuence on risk perception [43] and riding attitudes.Our current study postulates that personality traits can impact hazardous riding conduct among motorcyclists via both direct and indirect routes, encompassing risk perception and riding attitudes.[37][38][39]44].It expresses the individuals' judgment of their riding ability.In [45,46], the scale is expanded and expressed in terms of motorcyclists' technical capacity and judgment ability.Compared with other scales, this scale is still less applicable when studying the riding behaviors of motorcyclists.

Riding Attitude.
A positive safety mindset is essential for riders to prioritize trafc safety while on the road.Tis mindset encompasses their thoughts, emotions, and cognitive processes [47], signifcantly impacting their riding behaviors.Riders who engage in risky behaviors often hold negative attitudes towards trafc safety.To assess a rider's attitude towards riding, various scales such as Ulleberg and Rundmo's scale (which evaluates trafc fow and rule obedience, speeding, and enjoyment of riding) [6,16] and Iversen and Rundmo's scale (focusing on attitude towards rule violations and speeding, careless driving by others, and drinking and driving) are utilized [47].

Risk Perception.
Risk perception is "the subjective experience of risk in potential trafc hazards" [48].Terefore, risk perception is considered a precursor to actual driving behavior and is also commonly examined concerning trafc safety.Risk perception can be assessed using several scales.For example, the scale of Rundmo and Iversen consists of three diferent dimensions (i.e., worry and insecurity, probability assessments, and concern) [49].Te scale of Machin and Sankey is compacted into three subfactors (i.e., worry and concern, the likelihood of a crash, and aversion to risk-taking) [42,48].Te scale of Wang et al. includes three diferent subfactors (i.e., level of danger, worry, and probability assessments) [45,46].

Risky Riding Behavior.
In 2004, Sexton et al. [50] employed a 24-item Motorcycle Rider Behavior Questionnaire (MRBQ) to evaluate the dangerous behaviors exhibited by motorcyclists, which were classifed into four distinct categories: trafc errors, speeding, stunts, and control errors.Later, Elliott et al. [51] expanded on the original MRBQ, drawing on Reason et al.'s [52] research to identify the various factors that infuence motorcyclist behavior in Great Britain.Tese factors were categorized into fve groups, including trafc errors, control errors, speed violations, performing stunts, and the use of safety equipment.Subsequent studies have also used the MRBQ to explore the risky behaviors of motorcyclists.For example, Motevalian et al. [53] developed the MRBQ for Persian motorcycle riders, which included six subscales (i.e., speed violations, trafc errors, safety violations, trafc violations, stunts, and control errors).In their model, some items related to protective clothing, as in Elliot's MRBQ, were deleted to match Persian motorcycle riders.Besides, some items were also added, such as helmet use, red light running, etc. Özkan et al. [54] deployed the MRBQ for Turkish people with factors similar to those of Elliott et al.According to researchers, the correlation between rider accidents and ofenses and annual mileage was signifcant, while rider age had a comparably more minor infuence.Stunts were identifed as the primary cause of accidents and ofenses, whereas trafc violations were primarily caused by speeding.In a study of Australian riders, Sakashita et al. [55] used the MBRQ to identify four important factors: control errors, speeding violations, stunts, and protective gear.Similarly, Uttra et al. [56] studied Tai riders and found that trafc errors, control errors, stunts, and safety equipment were critical factors.Stephens et al. [57] also analyzed the correlation between the MBRQ and trafc accidents in Australia.Tey discovered that while speeding violations and control errors were uncommon, they still increased the likelihood of an accident.Furthermore, stunts raise the chance of an accident.Bui et al. [28] investigated the factors of the MRBQ for Vietnamese motorcyclists.Teir scale included 36 items, which were arranged into four dimensions in terms of previous studies on the factors infuencing riding behaviors.Some studies considered either four factors [28,50,55,56], fve factors [51,54,57], or six factors [53].From there, the consideration of factors is not consistent, depending on the trafc characteristics of each region.Besides, most of the above studies have been conducted in a trafc environment with predominantly cars, which difers from the mixed trafc fow that depends on motorcycles, like in Vietnam.Besides, several items related to the use of safety equipment and safety devices, such as protective clothing, are unsuitable for Vietnamese trafc conditions due to the lack of regulations and trafc culture.Terefore, they are not considered in this study.Te mediated efect through riding attitude and risk perception of the theoretical model is signifcant when the product of paths is signifcantly diferent from zero [58,59].

Data Collection.
To match the theoretical model proposed in Sub-section 3.1, we designed a structured threesection questionnaire as a self-reported questionnaire.For the frst section, we gathered socio-demographic information from the participants, such as their age, gender, and riding experience.Te second section was dedicated to analyzing trip patterns, such as the purpose and frequency of rides.Lastly, in the third section, we sought to collect data on personality traits, riding attitude, self-confdence, risk perception, and any risky riding behavior.
Te research analyzed personality traits (PT) through fve subfactors that comprised a total of 24 items.Tese subfactors were anxiety (AN, four items), anger (AG, fve items), sensation-seeking (SS, fve items), altruism (AL, six items), and normlessness (NL, four items), which were sourced from materials such as [6,16,38,40].Participants were assigned a rating on a fve-point Likert-type scale, with responses ranging from "strongly disagree (1)" to "strongly agree (5)."A higher score indicated a greater level of the personality traits under examination.
Riding self-confdence (RC) was evaluated utilizing seven items adapted from [45,46] (except "I can judge the speed of a motor vehicle accurately").All items were evaluated based on a 5-point scale from low to high (strongly disagree (1) to strongly agree (5)).Te higher the score, the higher the self-confdence.
To assess one's riding attitude (RA), we analyzed three key components as a set of subfactors.Tese subfactors included attitudes towards trafc fow (AT), consisting of three distinct items; attitudes towards rule adherence and speeding (AR), consisting of nine items; and attitudes towards enjoying the ride (AF), consisting of three items.Tese were sourced from [4,16,60].Te quantitative items were evaluated on a 5-level scale, ranging from strongly disagree (1) to strongly agree (5).A higher score indicates a more negative attitude, indicating a greater inclination towards disregarding trafc laws and regulations [4].
Risk perception (RP) was considered using three aspects as three subfactors, including worry and concern (WC, four items), probability assessments (PA, four items), and cognition of danger level (WC, six items), as modifed from [45,46,49].Quantitative items were assessed on a 5-level scale from low to high (strongly disagree/no danger (1) to strongly agree/very dangerous (5)).Te higher the score, the more worried you are about trafc risks and feelings.Te higher the risk of behavior is subjective, the greater the perceived consequences [31].
Te questionnaire used for the ofcial survey was initially written in English and later translated into Vietnamese.Pilot tests were conducted to fnalize the questionnaire before data collection began in Hanoi, Vietnam's largest city by area and second-largest by population, from January to March 2019.Before participation, all participants were informed of the survey's purpose and gave their consent.
After collecting 750 responses, we carefully screened for incomplete and unreliable responses, resulting in a fnal sample of 716 participants to test the proposed model.Of those participants, 52.5% were male, and 50% were between the ages of 16 and 25.About 33.2% had a college or intermediate-level education, while 66.8% had a university degree or higher.In terms of income, 42.8% earned 5-10 million VND per month, 51.7% earned below 5 million VND per month, and 5.9% earned over 10 million VND per month.

Analytical Method.
Te model underwent rigorous testing and validation using structural equation modeling (SEM), a widely accepted analytical technique for discovering patterns and relationships among variables and parameters across various felds, such as trafc behavior modeling [65,66].To execute SEM, this study employed two key stages, namely, confrmatory factor analysis (CFA) for measurement models and SEM for a structural model, with the aid of SPSS software and the AMOS tool.
Several goodness-of-ft statistics are used to test the hypothesis, as well as to judge the ft and validity of the measurement model and structural model, as follows: (i) Factor loading is not less than 0.5 for CFA [67,68].
In addition, to ensure reliability and convergence, average variance extracted (AVE) is not less than 0.5 [67], and the composite reliability index (CRI) is not less than 0.6 [69,70].
(ii) Te goodness-of-ft index (GFI) and the root mean square error of approximation (RMSEA) are evaluated as absolute ft measures.Te comparative ft index (CFI) [71] and incremental ft index (IFI) are taken as incremental ft indexes.Besides, some other indexes are also used to test the hypothesis, such as the value of Chi-square/df and the Tucker-Lewis index (TLI).In general, 3 to 4 relevant indicators are present for each study, and the evidence is sufcient to provide the appropriateness of the model without considering all other indicators [67].Te CFI, TLI, IFI, and GFI values are between 0 and 1, and close to 1 is better [72].Te value of Chisquare/df should be smaller than 5 [73,74].RMSEA should be smaller than 0.08 [73].TLI should be equal to or larger than 0.80 [75].

Results and Discussion
4.1.Results of the CFA.We employed the CFA method to evaluate the reliability of measurement items and confrm the suitability of the factorial structure.Trough this process, we pinpointed key personality traits such as riding selfconfdence, risk perception, riding attitude, and risky riding behavior.We eliminated items that exhibited low loading values, as determined by the multiple goodness-of-ft statistics discussed earlier.
First, we obtained four dimensions for personality traits, including anger (AG, fve items), sensation-seeking (SS, four items), normlessness (NL, three items), and altruism (AL, two items).Secondly, riding self-confdence was extracted from one dimension of six items.Tere are three subscales for riding attitude: attitude towards trafc fow (3 items), attitude towards rule obedience and speeding (6 items), and attitude towards fun-riding (2 items).After that, risk perception consisted of three dimensions: worry and concern (WC, four items), probability assessments (PA, four items), and cognition of danger level (CD, six items).Finally, we achieved the scale of risk-riding behavior with four subscales: ordinary violations (OV, four items), speed and riding violations (SV, four items), lapses and errors (LE, fve items), and aggressive violations (AV, seven items).Te items of the factors are shown in Table 1.

Results of SEM.
Tis study tested the theoretical model using SEM to analyze the factors that afect risky riding behaviors among motorcyclists.In the frst analysis of the theoretical model, the goodness-of-ft statistics met the requirements.However, we did not see a signifcant relationship between personality traits (PT) and riding self-confdence (RC) (Sig.� 0.122), as well as between personality traits (PT) and risk perception (RP) (Sig.� 0.931).Tis means that hypotheses H1b and H1d have been rejected.After removing the insignifcant relationships from the theoretical model, we achieved a fnal model, as shown in Figure 2. Te model had the value of Chi-square/df � 2.454 < 5; RMSEA � 0.045 < 0.08.In addition, the GFI, CFI, IFI, and TLI values were all greater   Journal of Advanced Transportation than 0.8, close to 1.0, indicating that the model had a reasonable ft.In the model, only the relationship between riding self-confdence (RC) and risky riding behaviors (RB) was accepted with a signifcance of 0.05 (Sig.� 0.024), and other links were accepted with a signifcance of 0.001.Since then, we have obtained the results of the research hypotheses, as shown in Table 2.

Journal of Advanced Transportation
Te direct and indirect efects of various factors, such as riding self-confdence, personality traits, risk perception, and riding attitude, are listed in Table 3.
After considering both direct and indirect efects, it was found that the most infuential factors on risky riding behavior were one's riding attitude (β = 0.60) and one's personality traits (β = 0.44).Factors that had less impact included risk perception (β = −0.04)and riding selfconfdence (β = 0.06).

Discussion
, Findings, and Implications.Tis study frst described the relationship between personality traits, riding self-confdence, riding attitude, and risk perception related to risky riding behaviors in Vietnamese motorcyclists using the CMM.After that, we estimated the efects of these factors on risky riding behaviors using a path analysis of SEM.Te factors were considered in many aspects, so they were also appropriate to human behavior.
Our research reveals fascinating insights into the intricate relationship between a rider's personality, confdence level, attitude, and risk perception and how they impact the likelihood of engaging in risky riding behavior.We observed that a positive riding attitude, strong self-confdence, and certain personality traits were associated with an increased likelihood of engaging in risky behavior, both directly and indirectly.Conversely, a negative perception of risk was linked to a decreased likelihood of such behavior.Moreover, we found that riding attitude and risk perception partially mediated the relationship between behavior, personality traits, and riding self-confdence.Particularly, riding attitude partially mediated the association between risk perception and risky riding behavior.Our study uncovers the intricate interplay between these factors and their impact on risky riding behavior.
For personality traits, there was a direct positive causal relationship (c 13 � 0.32) between personality traits and risky riding behaviors.Similarly, a positive causal relationship (c 11 � 0.20) was also found between personality traits and riding attitudes.Tis was also found in [16,76]; therefore, the results of this study reinforce the statement related to personality traits.Hence, personality traits may be mitigated to address risky riding behaviors.
Multiple studies have indicated a favorable connection between risk perception and a reduction in risky behavior [8,43,77].Tis implies that an individual who perceives a particular behavior as hazardous is less likely to partake in it.Nonetheless, this association is sometimes not as strong as anticipated.Moreover, some research suggests that risk perception might be an efect rather than a causal factor.In other words, there was no cause of risky behavior by risk perception [6,25] or only an indirect efect on risky behavior via riding attitude [78] or a negative relationship on reduction of risky behavior [8,77,78].Hence, risk perception was found to infuence risky riding behaviors in both positive and negative, depending on the trafc characteristics and culture.Tis study found a positive direct relationship (β 23 � 0.25) between risk perception and risky riding behavior, suitable for the results found in [8,78,79].Tis result can be described when motorcyclists perceive and behave consistently with those perceptions while riding on the road.It stands to reason that one's likelihood of a potential crash infuences one's driving habits.Individuals who engage in riskier driving behaviors may see themselves as more susceptible to a collision, which may result in further risky driving behavior [42].Moreover, a few motorcyclists could accept a certain risk when riding to unafect their enjoyment and may take risky actions [80].Many motorcyclists believe that they are immune to accidents [46].However, our research revealed a negative correlation (estimated at −0.29) between risk perception and risky riding behaviors, primarily infuenced by riding attitude.Tis indirect impact is more signifcant than the direct impact.Terefore, our fndings indicate that risk perception has a negative impact on risky riding behavior.Simply put, the higher an individual's risk perception score, the lower their score in risky riding behavior.As a result, it is crucial to educate motorcyclists about the dangers of reckless behavior and the severity of trafc accidents through laws and awareness campaigns.
Tere was a positive correlation (0.60) between risky behavior and unsafe riding attitudes found in this study and sources [37,76,81,82].It found that high scores of riding attitudes indicated negative attitudes towards trafc safety, leading to a positive relation to risky behaviors.In other words, participants who reported unsafe attitudes towards riding also reported more risky behaviors.Besides, there was an association between riding attitudes and risk perception, as well as the afecting riding behaviors [6,81].We found a negative relationship (β 21 � −0.49) between risk perception and riding attitudes.It revealed that a high-risk perception score indicated a negative attitude towards risk-taking behaviors, i.e., a positive attitude towards trafc safety.Tis was consistent with the results in [37].Tese results reinforced the direct efect of risk perception on riding attitudes.Terefore, strategies intended to change riding behaviors will be efective only by changing riding attitudes towards unsafe behaviors.In other words, trafc laws and education programs are necessary to improve safety-riding attitudes.
Regarding riding self-confdence, there was a positive efect of riding self-confdence on riding attitude towards risktaking behavior found in [37,39].At the same time, there was a negative relationship between it and risk perception [45,46].Tis study also found these results, i.e., riding self-confdence increased attitudes toward unsafe behaviors and decreased perceived risk.However, we found a negative direct relationship (c 23 � −0.11) between riding self-confdence and risky riding behavior, the same as in [76].Experienced motorcyclists often exhibit greater self-assurance while on the Journal of Advanced Transportation road, which can reduce the likelihood of accidents for themselves and others [76].Moreover, there are indirect correlations between self-confdence and risky behavior, as evidenced by risk perception (with an estimated value of −0.12), riding attitudes (with an estimated value of 0.15), and the combination of both (with an estimated value of 0.14).In general, the advantages of riding self-confdence outweigh the disadvantages.To sum up, riding self-confdence positively impacts risky riding behavior.Nonetheless, it is worth noting that excessive confdence can result in increased risk-taking and a greater chance of accidents.Improving the trafc infrastructure for motorcycles is necessary to increase new riders' confdence on the road.In addition, stronger law enforcement is needed to address motorcyclists who exhibit excessive confdence.
For risky riding behavior, as presented in Table 1, 20 items of the riding behavior scale were found to result from complex trafc conditions and the limitations of safety legislation in Vietnam.Several behaviors (e.g., SV04 "Illegal overtaking," OV04 "Get into the wrong lane," and SV03 "Exceed the speed limit on a road," OV06 "Use a cell phone while riding," etc.) need to be prioritized for handling.Furthermore, aggressive violations (AV) were also identifed as important risky behaviors that impact the crash risk of motorcyclists.Aggressive violations have been associated with personality traits of motorcyclists, such as anger.Tey also refect the low culture of a part of motorcyclists.
Based on the abovementioned discussions, the controlling measures and changes in risky behaviors of Vietnamese motorcyclists are proposed as follows: (i) Motorcycles will still be the primary means of transportation people choose in the long term, but facilities (e.g., motorcycle lanes, signal lights, etc.) for motorcycles are still limited.Te facility is a key part of riding behaviors, and it impacts the behaviors of the motorcyclist.It is necessary to increase facilities for motorcycles.(ii) Strengthening and consistency in enforcing road trafc rules and laws can increase awareness and result in a change in riding behaviors.(iii) Installing surveillance cameras at intersections and road sections will make motorcyclists pay attention to their surveillance, thereby helping to control risky behaviors related to violations.(iv) To enhance road safety in crucial zones, it is imperative to adopt efective measures that cover various aspects such as speed control, enforcement, and communication.Strategies such as imposing higher penalties, introducing a license point system, and defning speed limits can be employed [27].It is also vital to regularly conduct awareness campaigns to educate motorcyclists on safe riding practices and risk perception.Furthermore, imparting safety education can positively infuence the personality traits of motorcyclists, particularly in managing anger and avoiding sensation-seeking behavior.
(v) Analyzing the causes of motorcycle crashes to increase public awareness is necessary.
Tis study has the following implications: (i) Tis study can help policymakers introduce a riding behavior test in training.(ii) We can predict future behavior tendencies of motorcyclists, as well as determine how to relate to motorcycle crashes through the afecting factors, including personality traits, riding self-confdence, riding attitude, risk perception, and so on.(iii) Te study's fndings can inform road safety measures and programs for motorcyclists.(iv) Tis study also provides future scholars with new insights into exploring the riding behaviors of motorcyclists in developing countries like Vietnam.

Conclusions
A CMM model study explored how personality traits, riding self-confdence, riding attitude, risk perception, and risky riding behavior were interrelated among motorcyclists in Hanoi, Vietnam.Te fndings indicated that these factors played a signifcant role in shaping the risky riding behavior of the participants.Both personality traits and riding selfconfdence directly and indirectly afect risky riding behavior, mediated by riding attitude and risk perception.Te study revealed that riding attitude and personality traits were the most infuential factors, whereas risk perception and riding self-confdence had a comparatively lower impact.Tese results align with earlier research and establish the validity of the measurement scales for varying trafc cultures.Te study provides crucial groundwork for developing strategies to mitigate risky riding behavior among motorcyclists.Te study's data collection scope was confned to urban areas of Hanoi, Vietnam, which could potentially limit its ability to represent riding behaviors in other areas accurately.Moreover, the study did not incorporate controlling factors for risky behaviors.Tese constraints will be thoroughly scrutinized in forthcoming studies.

Table 1 : 74 I
Te structures of factor scales measured.It is OK to get round laws and rules as long as you don't break them directly 0.665 0.51 0.75 If something works, it is less important whether it is right or wrong 0.813 Some things can be wrong to do even though it is legal to do it 0help me get out of trouble when in danger 0.688 I am a good rider so i can ride exceeding the speed limits 0.615 I am familiar with the roads around me and everything is under my control 0.816 I can often judge whether my riding behavior is dangerous or not based on my experience 0.791 I can accurately judge the movement of nearby vehicles 0.699 Riding attitude (RA) Attitude towards rule obedience and speeding (AR) Sometimes it is necessary to break the trafc rules in order to get ahead 0.759 0.52 0.87 Sometimes it is necessary to take chances in the trafc 0.687 Sometimes it is necessary to bend the trafc rules to arrive in time 0.765 A person who take chances and violate some trafc rules is not necessary a less safe driver 0.628 If you have good skills, speeding is OK 0.think it is OK to speed if the trafc conditions allow you to do so 0.745 Attitude towards trafc fow (AT) Tere are many trafc rules which cannot be obeyed in order to keep up the trafc necessary to bend the rules to keep trafc going 0.821 It is more important to keep up the trafc fow rather than always follow the trafc rules 0.646 Attitude towards fun-riding (AF) Riders may be increase fun and excitement in trafc 0 assessments (PA) How probable do you think it is for yourself to be involved in a trafc accident 0.906 0.73 0.91 How probable do you think it is for yourself to be injured in a trafc accident 0.959 How probable do you think that you are more likely to have a trafc accident than others?0.855 How probable do you think that riding a motorcycle in a trafc accident is higher than that of other vehicles 0.662 Worry and concern (WC) Feeling unsafe that you yourself could be injured in a trafc accident while riding 0.716 0.62 0.87 Worried for yourself being injured in a trafc accident 0.782 Feeling unsafe that persons could be injured in a trafc accident 0.853 Concerning about a trafc accidents and thinking that you yourself could be victimized 0.784 Risky riding behavior (RB) Aggressive violations (AV) Get angry at other rider and try to give chase with the intention of letting them know what you mean 0.694 0.60 0.91 Running through an intersection that causes riders of the same direction to stop and yield to you 0.777 Horn (verbally) to indicate your discomfort with other riders 0.817 Racing away from a trafc light with the intention of passing the vehicle ahead 0.841 Riding too close to rider in front ignoring safe gap 0.795 Riding very close to the rider in front as a signal to its rider to go faster or get out of the way 0.772 Getting involved in unofcial races with other riders 0.724 Lapses and errors (LEs) Not paying attention to observe should almost collide with pedestrians (trafc equipment and objects) or misjudging the speed of the vehicle in front, but running too close 0.766 Do not pay attention to the "give way yield " signs or signal lights 0violations (SVs)Accelerating into an intersection when the trafc light is changing from yellow to

Table 2 :
Results of the research hypotheses.

Table 3 :
Results of direct and indirect efects.