Examining Risky Riding Behaviors: Insights from a Questionnaire Survey with Middle-Aged and Older Motorcyclists in Thailand

Abstract

This research endeavors to achieve three primary objectives. Firstly, it seeks to develop a measurement model capable of assessing the motorcycle riding behavior of middle-aged individuals in Thailand. The construction of this model integrates the motorcycle rider behavior questionnaire (MRBQ) with statistical and descriptive analyses. Secondly, the research examines the accuracy of the measurement model using a factor analysis approach, comprising both exploratory and confirmatory factor analyses. Lastly, the study aims to furnish the people of Thailand with a set of guidelines for self-assessment of their motorcycle riding habits. The examination involves 399 middle-aged motorcycle riders aged 35 years or older, a significant majority of whom (81%) possess the requisite licenses for motorcycle operation, with the majority (83%) accumulating over a decade of riding experience. Through analysis, a set of 39 items capturing nuanced behaviors exhibited by middle-aged motorcyclists in Thailand is identified. These items are categorized into four distinct clusters: speed violations, control errors, traffic errors, and adherence to safety equipment protocols. The utilization of the MRBQ in this study holds significant importance, as it provides valuable insights into the riding practices of the Thai population. The resultant findings hold the potential to inform safety initiatives and strategies aimed at enhancing the overall motorcycle riding experience.

1. Introduction

Worldwide, road traffic accidents are the cause of death of about 1.2 million people every year, making it one of the leading causes of death. Moreover, the World Health Organization also showed that motorcycle accidents in South-East Asia account for 43% of those road traffic accident deaths, and 74% of motorized accidents occur in Thailand [1]. Otherwise, people across the globe are enjoying longer lifespans, with the majority of individuals now living into their sixties and beyond. The older population is growing both in numbers and as a proportion of the total population in every country. By 2030, roughly one in six people worldwide will be aged 60 or older. The number of people in this age group is set to grow from a billion in 2020 to 2.1 billion by 2050. The population of those aged 80 and older is also predicted to triple, reaching 426 million between 2020 and 2050. While the trend of population aging initially started in countries with high incomes, such as Japan where over 30% of the population is already over 60, now it is the low- and middle-income countries that are experiencing the most significant changes. By 2050, around two-thirds of the world’s population aged 60 and above will be living in these lower- and middle-income countries [1].

From the records of the Police Information System Center, the Royal Thai Police indicates there was an increasing trend of accidents from 2011 to 2022 of approximately 36.25%. For the economic losses due to road accidents, persons injured by road accidents increased approximately by 152.18%. Moreover, motorcycles have been the most popular vehicles among Thai people over a long period; the statistics from 2020 show that approximately 19.40% of accidents involved motorcycles, compared with passenger cars being involved in 15.67%, and light trucks being involved (pick-ups) in 9.20%. Evans et al. [2] found that humans influence 95 percent of accidents, while the road environment and vehicle-related factors influence 28 percent and 8 percent of accidents. It might be suggested that traffic accidents resemble a severe and chronic disease that occurs every day and whose threat to health, quality of life, and people’s property is significant. Therefore, this is an urgent problem that everyone and every department must take responsibility for and pay considered attention to working together to solve these problems and prevent accidents. However, in order to prevent accidents, we must first be aware of the cause of accidents.

Risky behaviors while riding a motorcycle can lead to accidents on the road [3]. To reduce these accidents, it is crucial to understand the behaviors that put motorcyclists at risk. That is where the motorcycle rider behavior questionnaire becomes involved, developed by Elliot and colleagues [4]. They took inspiration from the driver behavior questionnaire (DBQ), a well-known tool used to study how truck and car drivers behave on the road, including their involvement in accidents [5]. Research has shown that people who score higher on the DBQ are more likely to have been in accidents in the past and are also at a higher risk of future accidents [6]. Thus, in addition to car drivers and motorcycle users, there is the WBQ version for pedestrians, the CBQ for cyclists, among others (e.g., Useche et al. [7]). This demonstrates the validity of the questionnaire and its effective applicability for different users and cultural contexts. Building on the success of the driver behavior questionnaire, Elliot and his team created the MRBQ specifically for motorcycle behaviors. The original MRBQ examines five aspects: mistakes in traffic unintentional errors by the rider; problems with controlling the motorcycle; speeding; performing stunts intentionally for excitement; and the use of safety gear. Over time, different ways of looking at the MRBQ were suggested, and extra questions were added to the questionnaire. The varying results from MRBQ studies show that road safety differs significantly between countries.

All of this discussion points to the need for studying how motorcycles are used in Thailand. Previous research has used MRBQ in various locations and with different groups of people. So, collecting and analyzing data from surveys to understand risky riding behaviors, especially on motorcycles, is currently of real importance. This type of data could help with forward planning, prevention of accidents, and lowering the chances and seriousness of severe injuries if accidents do happen—like fractures or dislocations, where organs can be injured, causing a lot of pain and serious harm. From the world-wide reports that the rates of accidents for people in low- and middle-income countries are increasing, likewise, The 5th Road Safety Master Plan of Thailand (2022–2027) places a strong emphasis on addressing significant risks and threats to the country’s security in a serious and timely manner [8]. It addresses concerns related to all road users, with a particular focus on high-risk groups such as middle-aged and older individuals. The MRBQ was used as a research tool to collect data from middle-aged motorcyclists in Thailand. Statisticians and factor analysis techniques use variables to identify unsafe driving behavior in Thailand. The findings from this study can contribute to enhancing, rectifying, or designing strategies aimed at effectively preventing accidents and minimizing losses from accidents in the future.

2. Literature Review

2.1. Risk and Older Motorcyclists

The growing number of older people in the world requires a thorough investigation of the practices of senior riders, including how they engage with different forms of transportation. The issue of safety holds significant importance, as scholarly studies have demonstrated that older individuals are more susceptible to accidents and injuries due to their physical fragility and reduced sensory capabilities. This vulnerability is particularly evident in the context of driving, where older adults face heightened risks at intersections and during lane changes. These risks can be primarily attributed to cognitive decline and a decrease in reaction times [9]. In addition to safety considerations, it is imperative to comprehend the travel preferences of older adults, as they demonstrate unique tendencies towards various modes of transportation. Convenience, cost, and accessibility are some of the factors that influence these inclinations, which ultimately determine whether people use public transportation, continue to rely on their personal vehicles, or ask family members for help [10]. Furthermore, the various obstacles faced by older adults, including diminished physical mobility resulting from health conditions, financial limitations, and hurdles in utilizing public transportation services, exert a substantial impact on their decision-making process regarding transportation options [11]. It is imperative to acknowledge and tackle these obstacles in order to encourage increased engagement in riding activities among the older population. This necessitates the implementation of customized interventions, including driver training programs that are suitable for their age group, enhancements in public transportation services, and the development of a pedestrian and cycling infrastructure that accommodates seniors [12].

The study of older riding behaviors is of paramount importance in light of the aging global population. Safety remains a central concern, with older adults facing increased vulnerability to accidents, particularly during driving, attributed to cognitive decline and reduced reaction times [13]. Understanding their travel preferences, which are influenced by convenience, cost, and accessibility, is crucial for designing inclusive and age-friendly transportation systems [14]. Addressing the challenges faced by the older population, including reduced mobility due to health issues and financial constraints, is essential to promote active riding behaviors [15]. Tailored interventions, such as age-appropriate driver training programs, improved public transit services, and a senior-friendly infrastructure, hold the potential to enhance the mobility and overall quality of life for older adults [16]. The body of research in this field continues to evolve, and ongoing efforts are required to adapt transportation systems to meet the specific needs and preferences of the older population, ultimately ensuring their safe and efficient mobility in an aging society.

2.2. Motorcycle Rider Behavior Questionnaire

According to earlier studies conducted by Vlahogianni et al. [17], there are a few different techniques to evaluate rider behavior and how it relates to accidents. In situations where there is a limited availability of resources for direct observation and official traffic records, the utilization of the motorcycle rider behavior questionnaire assumes a critical role. The present questionnaire provides significant insights into the behavioral patterns exhibited by individuals during motorcycle riding in various traffic scenarios, rendering it an invaluable instrument for both motorcycle safety research and practical application. The incorporation of stunt performance in the MRBQ has demonstrated a notable correlation with accidents among motorcyclists in Australia [18]. According to [19], this particular feature has also played a substantial role in actual accidents and instances of traffic violations among riders in Turkey. Although safety equipment was frequently evaluated, its impact on the occurrence of accidents was not significant. The paramount significance lies in guaranteeing the reliability and validity of the MRBQ when evaluating motorcycle safety initiatives with the aim of comprehending and mitigating accidents. Numerous scholarly investigations have extended beyond the conventional MRBQ methodology. For example, the authors of [20] conducted a study in which they examined individuals who had a minimum of three years of experience in motorcycle riding.

Researchers [21] have utilized the MRBQ to evaluate socio-demographic factors and the frequency of accidents among the participants. A factor structure was constructed that exhibits alignment with the structures of the MRBQ in other nations. While earlier studies primarily concentrated on industrialized nations such as Australia, Slovenia, and the UK, subsequent studies have extended the utilization of the MRBQ to emerging countries like Vietnam. Nevertheless, the outcomes of these inquiries exhibit variability, underscoring the imperative to reassess and validate the efficacy of the MRBQ in various foreign settings [22]. It is clear that the MRBQ is an essential instrument for comprehending riding behavior and how it relates to accidents, especially in situations where resources are limited. Recent studies have shown the correlation between stunt performances and accident engagement across different nations, while also shedding light on the relatively minimal influence of safety equipment. In order to uphold the effectiveness of motorcycle safety initiatives, it is crucial to sustain the reliability and validity of the MRBQ. The research related work shown in

Table 1. Related works and factors in previous motorcycle riding behavior research.

			Factor 1
Sources	Location	Sample Size	SV	TE	CE	ST	SE	AL	Analysis Method
Elliott et al. [4]	UK	8666	+	+	+	+	+		Principal component analysis, linear modeling
Stephens et al. [18]	Australia	470	+	+	+	+			Descriptive statistics, exploratory factor analysis
Özkan et al. [19]	Turkey	451	+	+	+	+	+		Descriptive statistics regression, principal component analysis, theory of planned behavior, health belief model, locus of control
Sumit et al. [20]	India	300	+	+	+	+	+		Principal axis factoring, and direct oblimin method
Sakashita et al. [21]	Australia	2375	+	+		+	+		Exploratory factor analysis, confirmatory factor analysis, Poisson regression
Trung Bui et al. [22]	Vietnam	2254	+	+	+		+	+	Confirmatory factor analysis, principal axis factoring, and direct oblimin method
Sunday and Akintola [23]	Nigeria	500	+		+	+			Principal component analysis, generalized, linear modeling
Topolšek and Dragan [24]	Slovenia	205	+	+		+	+	+	Exploratory factor analysis, confirmatory factor analysis, structural equation modeling
This research	Thailand	399	+	+	+		+		Exploratory factor analysis, confirmatory factor analysis, structural equation modeling

¹ (+) = Considered factor that included in the research; SV = speed violations; TE = traffic errors; CE = control errors; ST = stunts; SE = safety equipment; AL = alcohol.

.

2.3. Analysis Methods in Behavior Assessment

A popular statistical technique for testing research hypotheses by incorporating pertinent theories or concepts is structural equation modeling (SEM) [25,26,27]. A recent study utilized SEM to investigate the personality traits, socio-cognitive features, and risky riding behaviors in a sample of 1028 inexperienced Italian drivers. The findings of the study indicate that there is an indirect influence of personality on riding behavior, mediated by risk perception, in addition to a direct influence. The relationship between riding behavior, social norms, and personality was found to be substantially influenced by risk perception. Both genders can provide evidence for this claim [28]. Furthermore, the study conducted by Mokarami et al. [29] employed the Mokarami method to examine the associations among latent factors pertaining to safety culture, unsafe behavior (acting as a mediator), and accidents. This approach elucidated the causes of accidents by utilizing observable variables. Typically, researchers employ descriptive measurements and goodness-of-fit statistical indices to evaluate the degree of alignment between the model and the observed data.

In order to better understand rider behavior, researchers in this discipline have also employed a variety of statistical techniques. Significantly, researchers have utilized exploratory factor analysis (EFA) as a methodology to discern latent components that contribute to the manifestation of hazardous riding behaviors. Stanojevic et al. [30] employed EFA as a methodological approach to investigate the psychological determinants that underlie risk-taking behaviors among motorcyclists. Their study identified specific components such as sensation-seeking and impulsivity as significant contributors to this phenomenon. Furthermore, within the realm of cycling helmet usage, a study conducted by [31] employed this analysis method to uncover factors that influence helmet-wearing behavior. The study identified variables such as safety attitudes and peer influences, shedding light on the intricate dynamics involved in this behavior. Moreover, confirmatory factor analysis (CFA) has been important in the validation of models and measurement devices employed in the investigation of rider behavior. Moreover, a previous study utilized confirmatory factor analysis as a methodological approach to establish the reliability and validity of a motorcycle rider behavior scale [32]. This scale serves as a crucial instrument for evaluating rider behaviors and attitudes. Furthermore, another research study used it to validate the suggested factor framework in the area of risk perception among car drivers, showing how important factors like risk acceptance and tolerance are in shaping driver behavior [33]. Finally, the utilization of analysis of variance has been crucial in examining the influence of external factors on rider behavior. The researcher carried out an investigation to look into the influence of age on motorcycle riders’ behavior. The results indicated noteworthy disparities between younger and older riders, with younger riders demonstrating a propensity for engaging in more hazardous activities [34]. In a similar vein, a recent study utilized this method to examine gender discrepancies in the behavior of individuals who engage in bicycle riding [35]. The researchers emphasized noteworthy disparities in terms of compliance with rules and the occurrence of accidents among male and female riders.

3. Methodology

3.1. Sample Size and Participants

This study focuses on the risky riding habits of Thai motorcycle riders who are middle-aged and older, considered aged 35 to over 64 years old [36,37,38,39,40]. The participants, who were randomly selected from Thailand’s motorcycle riding population, were all above 35 years old, capable of riding motorcycles, and either in possession of or without a riding license. The required sample size for analysis comprised a viable sample size greater than five to fifteen times the observed variables [41]. Given that there were 39 items in the survey, between 195 and 585 samples were required. Nonetheless, the replies were gathered from 399 individuals in order to enhance the current research quality. Ultimately, after deleting 21 incomplete survey forms, 399 of the 420 responses were included in the final analysis.

3.2. Data Collection and Survey Instrument

The surveys were conducted utilizing the pre-designed questions from 1 December 2022 to 20 February 2023. The survey was distributed using an online Google Form through various social media platforms like Facebook, Line, and X. This was conducted through random sampling, reaching people from all across Thailand. The form of the questionnaire was created in accordance with motorcycle riding behaviors by reviewing the relevant literature and studying conceptual ideas and associated work that is shown in Appendix A,

Table A1. Variables of MRBQ and model modification [4,18,19,22].

Variable	Description	Modification of Original MRBQ 1	Additional 1
X1	Taking a corner so fast that it makes you nervous.	+	−
X2	Going over the speed limit on a residential street.	+	−
X3	Exceeding the speed limit on a country or rural road.	+	−
X4	Accelerating too quickly and lifting the front wheel off the ground.	+	−
X5	Failing to notice pedestrians crossing when turning onto a side street.	+	−
X6	Approaching a corner so fast that you feel like you might lose control.	+	−
X7	Trying, or actually managing, to do a wheelie.	+	−
X8	Participating in informal races with other riders or drivers.	+	−
X9	Riding when you suspect you might be above the legal alcohol limit.	+	−
X10	Slipping on a wet road or a manhole cover.	+	−
X11	Struggling to stop in time when traffic lights change while you’re riding at the same speed as other vehicles.	+	−
X12	Racing to outpace the vehicle next to you when traffic lights turn green.	+	−
X13	Drifting wide while cornering.	+	−
X14	Attempting to overtake without noticing a right-turn signal.	+	−
X15	Following so closely that you would not have enough space to stop in an emergency.	+	−
X16	Having your visor or goggles fog up.	+	−
X17	Not using headlights properly.	−	+
X18	Texting while riding.	−	+
X19	Eating while riding.	−	+
X20	Failing to keep to the left on four-lane highways.	−	+
X21	Riding in lanes meant for pedestrians.	−	+
X22	Neglecting to signal when parking.	−	+
X23	Feeling drowsy.	−	+
X24	Using a cell phone while riding.	−	+
X25	Riding with one hand.	−	+
X26	Using a GPS to navigate while riding.	−	+
X27	Going against the flow of traffic.	−	+
X28	Running a red light.	−	+
X29	Not seeing a person waiting to cross the road at a marked crossing, like a zebra or pelican crossing, when the signal has just turned red.	+	−
X30	Pulling out onto the main road in front of a vehicle that you didn’t notice, or misjudging its speed.	+	−
X31	Failing to anticipate that another vehicle might pull out in front of you, and struggling to stop in time.	+	−
X32	Not noticing someone stepping out from behind a parked vehicle until it’s almost too late.	+	−
X33	Not realizing that pedestrians are crossing the street while you’re making a turn from the main road into a side street.	+	−
X34	When waiting to turn left on the main road, focusing so much on the main traffic that you almost hit the vehicle in front of you.	+	−
X35	Missing the “Give Way” signs and narrowly avoiding a collision with traffic that has the right of way.	+	−
X36	Being distracted or lost in thought, and suddenly realizing that the vehicle in front of you has slowed down, forcing you to brake suddenly to avoid a crash.	+	−
X37	Carrying passengers while riding.	−	+
X38	Riding in lanes intended for motor vehicles.	−	+
X39	Not wearing a helmet while riding.	−	+

¹ (+) = included variables; (−) = not included variables.

; meanwhile, the validity was then confirmed by experts. The survey instrument employed in this study was partitioned into halves. The initial segment comprised sociodemographic information of the participants, such as their gender, age, educational attainment, occupation, driving license type, frequency of motorcycle usage, riding experience, and traffic knowledge. Respondents were then requested to assess their own safety conduct. The second section of the questionnaire comprised the adapted MRBQ, which includes 24 items from the original questionnaire [4] and 15 extra items that pertain to risky riding behavior seen in Thai motorcyclists. On a 6-point scale, respondents were requested to indicate the frequency with which they engaged in risky driving practices over the last year. A rating of 1 indicates never, 2 indicates rarely, 3 indicates often, 4 indicates rather often, 5 indicates frequently, and 6 indicates always.

3.3. Data Analysis

First, we examined the participants’ basic information using percentages to understand their demographics. Then, we used IBM SPSS Statistics version 26 to analyze the data relating to risky riding behaviors. Our analysis involved various approaches, including calculating average scores and assessing the reliability of the results. To ensure reliability, we applied a statistical measure called Cronbach’s alpha coefficient, considering values above 0.60 as a reliable ratio [42,43,44]. Next, we used exploratory factor analysis (EFA) to discover any patterns in participants’ riding behaviors. For EFA, we set certain criteria: Kaiser–Meyer–Olkin (KMO) values needed to be above 0.70, and factor loadings had to surpass 0.50 to be considered significant. To validate the quality of our model, CFA, using AMOS version 21.0 applications, was employed. This technique utilized a refined method known as maximum likelihood estimation.

To measure the good fit of our model aligned with the data, we employed several metrics: chi-square (χ²), degrees of freedom (df) ratio, comparative fit index (CFI), Tucker–Lewis index (TLI), residual mean squared error of approximation (RMSEA), and standardized root mean square residual (SRMR). A model was considered satisfactory if the CFI and TLI values exceeded 0.90 and if the RMSEA and SRMR values were below 0.07 or very close to it. We applied these criteria to assess model fit, drawing inspiration from the work of Browne and Cudeck [45].

Moreover, composite reliability (CR) and average variance extracted (AVE) are commonly used in the analysis of survey data for the reliability and convergent validity of a measurement instrument [46]. This can be achieved by calculating Cronbach’s alpha coefficient, which represents the extent to which the items in a construct consistently measure the same underlying concept [47]. This measurement considers the threshold greater than 0.50 to be significant. The following equation was used to compute the CR values of each latent variable [46]:

$$CR=\frac{{\left({\sum }_{i=1}^n{FL}_i\right)}^2}{{\left({\sum }_{i=1}^n{FL}_i\right)}^2+{\sum }_{i=1}^n{ME}_i}$$

(1)

where ${FL}_i$ denotes the standardized factor loadings of measurement item i, n denotes the number of items within a factor, and ${ME}_i$ refers to error in measurement.

The average variance extracted is a measure used to assess how much of the variability in a construct is accounted for by the construct itself compared to the variability caused by measurement errors. The following equation could be used to compute the AVE value [46]:

$$AVE=\frac{{{\sum }_{i=1}^n{FL}_i}^2}{n}$$

(2)

where ${FL}_i$ denotes the standardized factor loadings of the measurement item i, and n represents the number of items within a factor.

4. Results

4.1. Demographics of Participants

In this study, background demographic information was obtained for the 399 participants who took part in the questionnaire survey. The results are shown in

Table 2. Participant characteristics (n = 399).

Category	Description	Frequency	Percent
Gender	Female	151	37.8
	Male	248	62.2
Age	35–44 years old	110	27.6
	45–54 years old	178	44.6
	55–64 years old	63	15.8
	Over 64 years old	48	12.0
Education level	Under bachelor’s degree	64	16.0
	Bachelor’s degree	261	65.4
	Over bachelor’s degree	74	18.6
Occupation	Student	20	5.0
	Business owner	94	23.6
	Government employee	105	26.3
	General employee	111	27.8
	Freelance	24	6.0
	Worker	33	8.3
	Farmer	12	3.0
Monthly income,	Under 9000	33	8.3
Thai baht (THB)	9001–15,000	44	11.0
	15,001–25,000	96	24.1
	25,001–35,000	71	17.8
	Over 35,000	155	38.8
Rider’s license	Yes	319	81.0
	No	80	19.0
Motorcycle Driver’s license type	Temporary driving license (2 year)	13	3.3
	Driving license (5 years)	178	44.6
	Permanent driving license	132	33.1
	None	76	19.0
Riding experience	Less than 1 year	4	1.0
	1–3 years	12	3.0
	4–7 years	25	6.2
	8–10 years	27	6.8
	More than 10 years	331	83.0
Level of traffic knowledge	High	185	46.4
	Medium	211	52.8
	Low	3	0.8

Note: USD 1 = THB 36.52 [48].

and

Table A2. Description of the questionnaire data.

Variable	Value
Gender	1: Female, 2: Male
Age	1: 35–44 years old, 2: 45–54 years old, 3: 55–64 years old, 4: Over 64 years old
Education	1: Under degree, 2: Bachelor’s degree, 3: Over bachelor’s degree
Occupation	1: Student, 2: Business owner, 3: Government employee, 4: General employee, 5: Freelance, 6: Worker, 7: Farmer
Monthly income (Thai baht)	1: Under 9000 baht, 2: 9001–15,000 baht, 3: 15,001–25,000 baht, 4: 25,001–35,000 baht, 5: Over 35,000 baht
Rider’s license	1: Yes, 2: No
Motorcycle Driver’s license type	1: Temporary driving license (2 year), 2: Riding license (5 years), 3: Permanent riding license, 4: None
Riding experience (years)	1: Less than 1 year, 2: 1–3 years, 3: 4–7 years, 4: 8–10 years, 5: More than 10 years
Level of traffic knowledge	1: High, 2: Medium, 3: Low
X1–X39	1: Never, 2: Rarely, 3: Often, 4: Rather often, 5: Frequently, 6: Always

. Most of the participants were male, outnumbering females. The majority fell within the 45–54 age bracket. Among the respondents, 44.6% held a motorcycle riding license. Within the group with licenses, almost all of the respondents had more than 10 years of riding experience. About 52.9% of respondents had a moderate level of traffic knowledge. In terms of education, around 65.4% of participants held a bachelor’s degree.

4.2. Empirical Analysis

Table 3. Analysis of riding behavior versus age in years of motorcyclist (n = 399).

		Age (Years Old) 1
Category	Description	35–44	45–54	55–64	Over 64
Area of using motorcycles	Urban	61	90	19	25
	Municipality	6	14	4	3
	Municipal district	26	56	27	11
	Suburban	17	18	13	9
Motorcycle engine capacity	100–150 cc.	84	136	48	37
	151–200 cc.	16	27	8	6
	201–500 cc.	6	13	6	3
	Over 500 cc.	4	2	1	2
Riding times	Day	104	176	62	48
	Night	6	2	1	0
Riding hours per day	Under 6 h	106	166	57	44
	6–8 h	4	10	5	2
	8–10 h	0	2	0	2
	Over 10 h	0	0	1	0

¹ Counts.

shows the crosstab analysis between age groups and riding behavior from all of the 399 respondents, composed of four parts including the area in which motorcycles were used, the motorcycles’ engine size, riding times, and riding hours. As can be seen from the table, almost everyone used their motorcycles in the urban area, used a motorcycle size of about 100–150 cc., rode during the daytime, and usually used their motorcycle for under 6 h per day. Also, people aged around 45–54 years old had the highest frequency of usage among all of the groups, followed by the group aged 35–44 years old.

In addition, the Pearson chi-square of people’s age association between variable contents was ages and area of using motorcycles (χ² = 16.920, df = 9, p > 0.05); ages and engine size of motorcycles (χ² = 4.175, df = 9, p > 0.05); ages and riding times (χ² = 7.375, df = 3, p > 0.05); and ages and riding hours (χ² = 13.625, df = 9, p > 0.05). Following the results obtained, there are no significant differences under the five percent error condition between the observed and the expected frequencies, by ages [49]. Thus, there is no correlation between the respondents’ age and riding behavior.

4.3. Motorcyclist Characteristics on Risk Contexts

The group of ages was analyzed by riding behavior that comprised traffic violations, accidents, and level of injuries using an ANOVA test, as shown in

Table 4. ANOVA testing of risk and age groups (n = 399).

Risk	Age	n	Mean	SD	F	Sig.
Traffic violations	35–44 years old	110	1.35	0.882	0.312	0.817
	45–54 years old	178	1.26	0.754
	55–64 years old	63	1.25	0.822
	Over 64 years old	48	1.25	0.729
	Total	399	1.28	0.797
Accidents	35–44 years old	110	1.15	0.522	1.522	0.208
	45–54 years old	178	1.08	0.344
	55–64 years old	63	1.02	0.126
	Over 64 years old	48	1.10	0.472
	Total	399	1.09	0.397
Level of injury	35–44 years old	110	1.72	1.042	4.837	0.003 1
	45–54 years old	178	1.43	0.801
	55–64 years old	63	1.27	0.574
	Over 64 years old	48	1.71	1.091
	Total	399	1.52	0.896

¹ Confidence level 99%.

. The results show that statistically significant differences (p < 0.05) were found between the age group and level of injury, with people who were 35–44 years old showing a mean and standard deviation of 1.72 ± 1.042; thus, they showed a higher chance of having more injuries than the other groups, followed by the group aged over 64 years old (1.71 ± 1.091). However, the statistics illustrated that traffic violations and accidents had no significant effect on the people’s age group (p > 0.05). Meanwhile,

Table 5. ANOVA testing of risk and riding experience (n = 399).

Risk	Riding Experience	n	Mean	SD	F	Sig.
Traffic violations	Less than 1 year	4	1.00	0.000	0.764	0.549
	1–3 years	12	1.58	1.165
	4–7 years	25	1.40	1.000
	8–10 years	27	1.33	0.832
	More than 10 years	331	1.26	0.767
	Total	399	1.28	0.797
Accidents	Less than 1 year	4	1.00	0.000	3.492	0.008 1
	1–3 years	12	1.50	1.000
	4–7 years	25	1.12	0.332
	8–10 years	27	1.07	0.267
	More than 10 years	331	1.08	0.370
	Total	399	1.09	0.397
Level of injury	Less than 1 year	4	1.25	0.500	0.455	0.769
	1–3 years	12	1.75	1.215
	4–7 years	25	1.64	1.036
	8–10 years	27	1.44	0.892
	More than 10 years	331	1.51	0.879
	Total	399	1.52	0.896

¹ Confidence level 99%.

shows the statistically significant difference between riding experience and accidents, where the people with 1–3 years (1.75 ± 1.215) of riding experience have a greater chance of causing or being in an accident than the other age groups, followed by the people who have 4–7 years of riding experience (1.64 ± 1.036). The other factors (traffic violations and level of injury) show that the results are not statistically significant.

4.4. Exploratory Factor Analysis

Exploratory factor analysis was performed with the data from 399 participants. Principal component analysis with rotation, or more precisely, varimax rotation with Kaiser normalization, was the technique we employed. The findings are summarized in

Table 6. Factor analyses of MRBQ.

		Exploratory Factor Analysis			Confirmatory Factor Analysis
Construct	Variable	Loading	Mean	SD	UE 1	SE 2	t	p-Value
Speed violations	X1	0.785	1.38	0.806	0.951	0.907	25.022	<0.001
	X2	0.781	1.48	0.844	0.949	0.862	22.873	<0.001
	X3	0.771	1.50	0.896	0.989	0.847	22.233	<0.001
	X4	0.765	1.24	0.711	0.791	0.859	22.686	<0.001
	X5	0.761	1.51	0.885	0.961	0.835	21.663	<0.001
	X6	0.750	1.41	0.790	0.919	0.893	21.998	<0.001
	X7	0.748	1.23	0.722	0.762	0.816	20.732	<0.001
	X8	0.715	1.42	0.867	0.898	0.799	20.133	<0.001
	X9	0.715	1.31	0.802	0.836	0.801	20.164	<0.001
	X10	0.699	1.40	0.820	0.858	0.804	20.235	<0.001
	X11	0.664	1.49	0.905	1.000	0.850	-	<0.001
	X12	0.655	1.67	0.920	0.921	0.780	19.624	<0.001
	X13	0.628	1.56	0.869	0.935	0.828	21.454	<0.001
	X14	0.624	1.46	0.884	0.956	0.832	21.581	<0.001
	X15	0.616	1.49	0.893	0.961	0.838	27.705	<0.001
	X16	0.559	1.60	0.934	0.811	0.671	15.493	<0.001
Control errors	X17	0.798	1.30	0.733	0.823	0.773	15.514	<0.001
	X18	0.790	1.36	0.769	0.881	0.790	16.083	<0.001
	X19	0.774	1.23	0.611	0.689	0.776	15.614	<0.001
	X20	0.747	1.40	0.792	0.862	0.754	15.149	<0.001
	X21	0.737	1.31	0.718	0.822	0.789	14.044	<0.001
	X22	0.725	1.43	0.766	0.878	0.790	15.921	<0.001
	X23	0.693	1.38	0.766	0.836	0.745	15.072	<0.001
	X24	0.656	1.52	0.823	0.924	0.774	15.777	<0.001
	X25	0.600	1.65	0.931	1.000	0.740	-	<0.001
	X26	0.598	1.59	0.980	0.871	0.608	12.201	<0.001
	X27	0.555	1.54	0.785	0.806	0.710	14.305	<0.001
	X28	0.539	1.51	0.811	0.825	0.711	14.514	<0.001
Traffic errors	X29	0.797	1.60	1.079	1.000	0.771	-	<0.001
	X30	0.783	1.63	0.970	0.974	0.836	18.324	<0.001
	X31	0.769	1.69	0.971	0.991	0.851	18.693	<0.001
	X32	0.740	1.69	1.076	0.940	0.728	19.489	<0.001
	X33	0.689	1.59	1.062	0.918	0.720	20.038	<0.001
	X34	0.673	1.55	0.944	0.969	0.855	18.725	<0.001
	X35	0.641	1.60	0.946	0.94	0.830	18.012	<0.001
	X36	0.595	1.51	0.879	0.867	0.827	17.736	<0.001
Safety	X37	0.740	2.44	1.286	0.922	0.540	9.022	<0.001
	X38	0.641	1.84	0.990	0.942	0.721	10.42	<0.001
	X39	0.590	2.01	1.127	1.000	0.667	-	<0.001

¹ Unstandardized estimate; ² Standardized estimate.

, which explains about 51.50% of the total patterns we observed. Only factor loading that is stronger than 0.40 is discussed. We identified four main categories: speed violations, control errors, traffic errors, and safety. The items are listed from the most important to the least important based on their contribution. The Kaiser–Meyer–Olkin measure came out as 0.967, which means that our sample was good for this type of analysis. Another test we performed, called Bartlett’s test of sphericity, showed that the data were appropriate for this type of analysis (χ² = 15294.799; df = 741, p < 0.000).

• Speed Violations: Sixteen items in this group had factor loading thresholds ranging from 0.559 to 0.785. This group included items like riding after drinking, exceeding the speed limit, racing on the road, and having vision problems.
• Control Errors: Factor loading thresholds of the riding behaviors ranged from 0.539 to 0.798 for twelve items in this group. Among these were “not using headlights appropriately”, “riding with one hand”, and “not staying in the proper lanes when driving”.
• Traffic Errors: The factor loading levels for the eight items in this group ranged from 0.595 to 0.797. The main reason for these acts was that the participants were oblivious to pedestrians and “Give Way” signs.
• Safety: Three items with factor loading criteria ranging from 0.590 to 0.740 were included in this group. It covered people not wearing helmets and carrying passengers while cycling. Additional information is provided in

  Table 7. Model fit indices of the CFA model [50,51,52].

  Model Fit Index 1	Model	Recommended Values	Resultant
  ${\chi }^2$/df	2.377	<5.000	Accepted
  CFI	0.940	>0.900	Accepted
  TLI	0.932	>0.800	Accepted
  RMSEA	0.059	<0.080	Accepted
  SRMR	0.035	<0.080	Accepted

  ¹ ${\chi }^2$/df = ratio of chi-square to degrees of freedom; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

  .

4.5. Confirmatory Factor Analysis

To validate the model, the maximum likelihood estimating approach was used for the confirmatory factor analysis (CFA). There were 39 measurement items in all across the four components that made up the original model. However, a modified model was created by modifying the free parameters until the model fit indices attained levels that were acceptable in order to obtain a satisfactory fit. Table 6 shows the CFA assessment methodology. The fitting indices for the modified model are shown in Table 7.

Overall, the results of the CFA demonstrate a well-fitted model, and the standardized estimates, ranging from 0.540 to 0.907, indicate robust relationships between the variables under consideration. To further evaluate the internal consistency of the constructs, we utilized the composite reliability index and average variance extracted, with recommended thresholds of CR ≥ 0.7 and AVE ≥ 0.5 [46]. The relationships between variables were examined through correlation testing at a significance level of p < 0.01. The results of the correlations are presented in

Table 8. Correlation matrix.

Construct	Speed Violations	Control Error	Traffic Error	Safety
Speed violations	1
Control error	0.654 1	1
Traffic error	0.827 1	0.511 1	1
Safety	0.441 1	0.619 1	0.347 1	1

¹ p < 0.01; correlation is significant at the 0.01 level (2-tailed).

.

4.6. Validity and Reliability

We examined the precision and dependability of the measurement model and evaluated the correctness of the indicators by utilizing Cronbach’s alpha. It has been suggested that a Cronbach’s alpha value exceeding 0.6 is the minimum threshold for establishing the reliability of a self-report questionnaire [42,43,44]. The reliability analysis yielded Cronbach’s alpha values between 0.972 and 0.699 for the factors of speed violations, control errors, traffic errors, and safety; meanwhile, the cumulative of the variance explained greater than 50 percent is also shown in

Table 9. Construct validity and reliability result.

Construct	Alpha	Eigenvalue	Explanation of Variance (%)	AVE 1	CR 2
Speed violations	0.972	20.085	51.50	0.686	0.972
Control error	0.935	3.955	10.14	0.476	0.915
Traffic error	0.940	1.485	3.81	0.510	0.892
Safety	0.699	1.353	3.47	0.436	0.697

¹ AVE = average variance extracted; ² CR = composite reliability.

.

Additionally, the AVE values obtained for these factors ranged between 0.436 and 0.686 while CR values obtained ranged between 0.697 and 0972 as shown in Table 9. As a result, the AVE values for control error and safety fell below the recommended threshold. According to the criteria outlined by Fornell and Larcker, the convergent validity of the construct is supposed to be satisfactory when the AVE is less than 0.5 and the CR surpasses 0.6 [53]. Thus, the measurement model we used in our study is consistent and reliable and supports the validity of our research.

5. Discussion and Conclusions

In this research, our objective was to develop a model for assessing motorcycle riding behaviors specific to the Thai population. We achieved this by adapting the motorcycle rider behavior questionnaire (MRBQ) through exploratory factor analysis. Additionally, we aimed to validate this measurement model using confirmatory factor analysis. This validation approach had been previously employed in studies by [4,17,18,19,21].

Furthermore, our aim extended to establishing guidelines for a self-assessment report concerning motorcycle riding behaviors among Thai individuals. To achieve this, we utilized MRBQ, a tool designed to study motorcycle riding behaviors. We applied this tool to Thai riders, incorporating the original 24 items from Elliott et al.’s [4] work in 2006, as well as 15 additional items related to unsafe riding behaviors. Utilizing factor analysis, we separated these behaviors into four different factors: speed violations, control errors, traffic errors, and safety.

It is important to note that our study specifically focused on collecting data from middle-aged individuals in Thailand. Our findings in terms of factors closely resembled those of Sakashita et al. [21]. When analyzing these factors, we observed that speed violations exhibited the highest loading factor value. This category contained actions such as excessive speed, breaking speed limits, riding under the influence, and engaging in racing on roads. Regarding control errors, our study identified frequently reported behaviors of Thai respondents, such as using electronic devices while riding, failing to stay in the correct lane, and riding with just one hand. These behaviors are often driven by time constraints and a reliance on convenience devices. In terms of traffic errors, a combination of behaviors occurred, including disregarding “GIVE WAY” signs on narrow roads and failing to yield to pedestrians at zebra crossings. The agility of motorcycles often leads riders to weave through traffic, sometimes resulting in sudden encounters with obstacles, particularly at intersections. A significant portion of accidents at traffic light junctions involved motorcycles crashing with other vehicles. Lastly, our observations regarding safety indicated a concerning trend: a notable portion of Thai riders do not wear helmets while operating motorcycles. This behavior can significantly increase the quantity of injuries in the event of an accident.

6. Limitation

The study faces several limitations that should be addressed in future research. Firstly, despite the inclusion of a substantial sample size of 399 middle-aged motorcycle riders in Thailand, the study’s sample may not fully capture the diversity of motorcycle riders in the country, thus potentially limiting the generalizability of findings to the broader population of riders. Secondly, reliance on self-reported data introduces the possibility of self-reporting bias where the participants may underreport certain unsafe behaviors or overemphasize adherence to safety protocols, thereby impacting the accuracy of the findings. Lastly, while survey-based methods were primarily utilized for data collection, the incorporation of complementary methods such as observational studies or interviews could offer deeper insights into motorcycle riding behaviors, providing a more comprehensive understanding of the factors influencing rider behavior.

7. Future Work

Several avenues for future research could enhance our understanding of motorcycle riding behaviors and contribute to the development of effective safety initiatives. Firstly, conducting longitudinal studies to track changes in motorcycle riding behaviors among different age groups over time would provide insights into the development of riding habits and the effectiveness of safety interventions. Secondly, comparative analyses across different regions or countries could be employed to identify cultural and environmental factors influencing riding habits, thereby informing the development of tailored safety initiatives. Moreover, enhancing quantitative data with qualitative research techniques like interviews or focus groups would provide more profound understandings of motorcycle riding habits, adding context and subtlety to numerical results. Lastly, investigating the incorporation of technology, such as smartphone applications or wearable devices, for real-time monitoring and enhancement of motorcycle riding behaviors is warranted. These research directions hold potential for advancing our understanding of motorcycle safety and contributing to the reduction in accidents and injuries among riders.