Extending the DBQ Framework: A Second-Order CFA of Risky Driving Behaviors Among Truck Drivers in Thailand

Abstract

Background: Truck drivers are a vital workforce sustaining Thailand’s freight transport, particularly in Northeastern Thailand (Isan), a major logistics hub connecting with Laos, Vietnam, and Cambodia via Highway No. 2 and the AEC network. However, these drivers face disproportionately high risks of severe road accidents due to occupational factors such as fatigue, time pressure, and long-distance driving. Methods: This study developed and validated a second-order confirmatory factor analysis (CFA) model to examine the multidimensional structure of risky driving behavior among Thai truck drivers. Grounded in the Driver Behavior Questionnaire (DBQ), the framework was extended to include seven dimensions: traffic violations, errors, lapses, aggressive behavior, substance use, technology-related distractions, and pedestrian-related risks. Results: Data were collected from 400 truck drivers in Isan using a structured questionnaire. CFA results confirmed the model’s structural validity, with satisfactory fit indices (X²/df = 2.122, CFI = 0.913, TLI = 0.897, RMSEA = 0.053, SRMR = 0.079). Conclusions: The findings reveal that risky driving behavior in this group extends beyond traditional DBQ categories, incorporating emerging risks specific to the commercial transport environment. This framework can be effectively utilized for risk assessment, behavioral screening, and the development of targeted safety interventions for this high-risk occupational group.

1. Introduction

Truck drivers serve as the driving force of the national logistics and supply chain system, playing a crucial role in the transportation of goods, raw materials, and economic products both domestically and internationally. The continuity and efficiency of freight transport depend heavily on the skills, endurance, and decision-making of truck drivers working under demanding and highly competitive conditions. In many countries, particularly in Southeast Asia, including Thailand, this occupation has been steadily increasing in number to meet the growing demand for freight transport driven by economic expansion and regional connectivity within the ASEAN Economic Community (AEC). Despite their critical role in the national economy, truck drivers face significantly higher safety risks compared to the general driving population. According to the World Health Organization [1], Thailand recorded a road traffic fatality rate of 25.4 deaths per 100,000 population, one of the highest in Southeast Asia. Data from the Ministry of Transport’s Traffic Accident Management System (TRAMS) indicated that in 2024, there were 377 fatalities involving truck crashes, representing a 20.96% increase compared to 2020 [2]. Similarly, the Department of Land Transport reported that in the same year, Thailand had over 1.256 million registered trucks, an 8.48% increase from 2020 [3]. These figures highlight a direct correlation between the rising number of trucks on the road and the increasing risk and severity of crashes, underscoring mounting pressures on road safety management.

The Northeastern region of Thailand, commonly known as Isan, serves as a major hub for agricultural and industrial freight transportation. This region forms a strategic corridor linking Thailand with neighboring countries such as Laos, Vietnam, and Cambodia, primarily through Highway No. 2 and the AEC cross-border road network. The rapid expansion of the logistics sector in this area has led to a high density of freight trucks operating on both major highways and secondary rural roads, significantly escalating road safety risks. As a result, Isan has become not only a primary transit route for heavy vehicles but also a critical hotspot for traffic safety concerns. Truck drivers operating in this region routinely encounter extreme work conditions, including time pressures, long-distance routes, congested traffic, and extended working hours. These factors collectively contribute to the likelihood of engaging in risky driving behaviors such as speeding, driving while fatigued, using mobile phones while driving, or even resorting to stimulant substances to maintain alertness over long hauls [4,5]. Such conditions are reflected not only in the frequency and severity of traffic accidents but also in risk behaviors that are closely tied to the occupational demands of truck driving, such as continuous long-distance driving, fatigue, and time-induced stress. They are also associated with psycho-physiological impairments that further reduce driving performance. On the other hand, certain protective factors such as accumulated driving experience and prior participation in safety training may mitigate some risks. Nevertheless, in the absence of adequate regulatory or organizational measures, these risk factors continue to foster hazardous driving behaviors, including excessive speeding, fatigue driving, or stimulant use to prolong driving [6]. These phenomena underscore that truck drivers’ behaviors constitute a critical factor that warrants systematic analysis in order to better understand the mechanisms leading to elevated risks and severe accidents.

Previous studies have categorized risky driving behaviors among truck drivers into several dimensions. These include traffic violations such as speeding, running red lights, and failure to wear seat belts [5,7]; aggressive behaviors such as honking excessively, tailgating, or abrupt braking in front of other vehicles [8]; and errors and lapses resulting from fatigue or momentary inattention. These behaviors are commonly regarded as unsafe acts that are strongly associated with the likelihood of severe traffic accidents. Importantly, such behaviors cannot be fully explained by structural or environmental factors alone.

One of the most widely accepted conceptual frameworks for studying risky driving behavior is the Driver Behavior Questionnaire (DBQ). This instrument classifies risky behaviors into three main categories: violations, errors, and lapses [9]. Later, Lawton et al. [10] extended the DBQ to include aggressive violations, capturing the emotional and affective dimensions of unsafe driving. This expanded framework has been widely applied and empirically validated in various countries, including France, Finland, Iceland, Austria, and Serbia, as well as Canada, a North American country whose driving culture closely resembles that of Western Europe [11,12,13,14,15,16]. In Asian contexts, the DBQ framework has also been confirmed in studies conducted in Qatar, the United Arab Emirates, China, and Malaysia [17,18,19]. Over time, it has become a foundational tool in road safety research for examining driver behavior across diverse cultural and regulatory environments. These studies highlight the DBQ’s adaptability to both developed and developing countries, particularly in contexts with differing traffic laws, road user norms, and enforcement mechanisms. Nevertheless, the original DBQ presents certain limitations when applied to specific subgroups, such as commercial truck drivers, whose behavior is shaped by unique occupational demands. These include chronic fatigue, high time pressure, and operational constraints imposed by the nature of heavy vehicles. As such, applying the DBQ to this group requires an expanded scope to better capture risk behaviors specific to the commercial driving context.

Emerging evidence from studies on truck drivers’ risky behaviors indicates that their behavioral patterns often extend beyond those defined by the original DBQ framework. These include behaviors such as substance and alcohol use as coping mechanisms for fatigue [20], engaging with social media or mobile phones while driving [21], and disregard for the safety of vulnerable road users, particularly pedestrians [22]. While these behaviors have been identified as critical contributors to serious traffic accidents, they are yet to be systematically incorporated into a comprehensive framework that accurately reflects the commercial driving context.

In summary, although the Driver Behavior Questionnaire (DBQ) has been widely applied and validated across different cultural and occupational contexts, it has not been systematically adapted to capture the unique risk behaviors of commercial truck drivers. This study addresses this gap by extending the DBQ framework to encompass seven dimensions of risky driving, including both traditional and context-specific factors. In doing so, the research makes two original contributions. Theoretically, it advances the DBQ taxonomy by integrating intentional, unintentional, and occupation-related risks into a more comprehensive framework that reflects the realities of truck driving. Practically, it provides an evidence-based foundation for developing specialized assessment tools, designing targeted training interventions, and informing transport safety policies tailored to the trucking sector in Thailand. Together, these contributions aim to reduce crash severity and promote safer logistics operations at both the regional and national levels.

2. Literature Review

2.1. Conceptual Framework of Risky Driving Behavior

The Driver Behavior Questionnaire (DBQ) was originally developed in the United Kingdom by Reason et al. [9] in 1990 to investigate aberrant driving behaviors. The initial version comprised 50 items designed to capture three core dimensions: violations, lapses, and errors. Subsequently, Parker et al. [23] revised the DBQ by reducing the number of items to 24, selecting those with the highest factor loadings from each of the three dimensions. Later developments further refined the DBQ. Researchers introduced an additional construct aggressive violations into the violation category. This included emotionally driven behaviors such as honking out of annoyance or tailgating to express frustration [10], thereby capturing the emotional component of risky driving more explicitly. Building on these revisions, Lajunen et al. [24] extended the questionnaire to include 27 items, distributed across four dimensions: errors, lapses, violations, and aggressive behaviors.

Despite its widespread application, the original DBQ still has limitations in addressing group-specific behaviors, particularly among commercial truck drivers, whose work routines, responsibilities, and stressors differ substantially from those of general drivers. These differences suggest the need to develop context-sensitive models that reflect the unique realities of the target population, thereby enabling more accurate identification and effective mitigation of driving risks.

2.2. Extension of the DBQ to Reflect the Context of Truck Drivers

Although the original Driver Behavior Questionnaire (DBQ) has been widely applied to assess multiple dimensions of risky driving, it demonstrates limitations when administered to commercial truck drivers, whose occupational demands and working conditions differ substantially from those of ordinary motorists. Truck drivers are typically exposed to intense time pressure, prolonged working hours, and chronic fatigue, which represent distinctive occupational stressors. Empirical evidence indicates that sleep deprivation is significantly associated with crash involvement among truck and bus drivers in Thailand [25]. Moreover, many drivers rely on energy drinks to counteract fatigue [26]. These findings are consistent with theories in human factors and occupational safety, which emphasize that fatigue and work overload increase the likelihood of impaired decision-making and traffic law violations [27,28]. Collectively, such factors suggest that the traditional DBQ may not be sufficient to capture the risky driving behaviors of truck drivers who face profession-specific constraints and pressures.

Several studies have highlighted specific risky behaviors that are not well represented in the original DBQ, one of the most prominent being mobile phone and social media use while driving. Such behaviors not only reduce situational awareness but also prolong reaction time and divert visual attention away from the road [29]. This issue is particularly relevant among long-haul truck drivers, who are more likely to use their devices for navigation, messaging, or social media engagement during trips. These activities considerably elevate the likelihood of high-risk incidents, especially in the case of social media use, which involves simultaneous visual, manual, and cognitive distraction, thereby greatly diminishing vehicle control [30]. Given these characteristics, mobile phone and social media use cannot be fully subsumed under the traditional DBQ categories. Unlike lapses, which typically reflect unintentional failures of memory or attention, or violations, which represent deliberate rule-breaking, this behavior constitutes a distinct form of distraction. It involves unique cognitive mechanisms and reflects modern technological challenges that the original DBQ framework did not anticipate.

In addition, the consumption of substances that impair driving performance remains a critical factor contributing to road accidents, particularly among truck drivers who often face long working hours and time pressure. This category includes alcohol and sedative drugs, which continue to be major causes of crashes, especially during festive seasons such as New Year and Songkran, when residual alcohol may still be present in drivers’ systems [31,32,33]. Similarly, the use of drowsiness-inducing medications, such as antihistamines or certain analgesics, can compromise vehicle control without the driver’s full awareness [34]. At the same time, although energy drinks are often perceived as a means of increasing alertness, empirical evidence suggests that excessive caffeine intake may lead to poor decision-making and rebound fatigue once the stimulating effects wear off [33,35]. Theoretically, these behaviors can be subsumed under the dimension of “Substance Use,” since alcohol, sedatives, and stimulants all exert negative effects on driving capacity in comparable ways. Substance or alcohol use, while resembling intentional violations in terms of impaired judgment, operates through psychopharmacological effects that distinguish it from traditional error or violation categories. Both depressants and stimulants alter psychomotor control in ways that are not captured by the original DBQ dimensions, thereby warranting its classification as a distinct factor within the extended taxonomy. Recognizing this as a distinct factor is therefore essential to accurately capture the risk profile of professional truck drivers.

Another critical yet often overlooked dimension of risky driving behavior is the disregard for the safety of vulnerable road users, particularly pedestrians. Among truck drivers, behaviors such as veering toward the roadside without checking for pedestrians, failing to notice individuals emerging from the roadside, or not stopping or slowing down at pedestrian crossings reflect a lack of awareness and concern for vulnerable road users [36], who are frequently the most severely affected in the event of a crash. Driving through crosswalks or pedestrian signals without stopping may indicate low safety consciousness or insufficient training in risk perception and situational awareness. These behaviors are especially concerning for large commercial vehicles, which are subject to greater blind spots and require longer braking distances than passenger cars. Therefore, it is essential to incorporate specific behavioral constructs into the DBQ that reflect these high-risk interactions, in order to better capture the safety challenges posed by truck drivers toward vulnerable road users. Neglect of pedestrian safety may superficially resemble a traffic violation, such as failing to stop at crosswalks, but it extends beyond mere rule breaking. It reflects a lack of situational awareness and social responsibility toward vulnerable road users, which represents a qualitatively distinct form of risky behavior that is especially critical in the context of heavy vehicles. This theoretical distinction supports its inclusion as a separate dimension in the extended DBQ framework.

Building upon the theoretical foundation of the Driver Behavior Questionnaire (DBQ) and empirical findings from previous studies, this research proposes a novel conceptual framework tailored specifically to the behavioral risk profiles of commercial truck drivers. As summarized in

Table 1. Summary of Conceptual Dimensions of Risky Driving Behavior among Truck Drivers.

Author (Year)	Driver Behavior Questionnaire (DBQ)				Other Factors
	Violation	Error	Lapses	Aggressive Behavior	Social Media	Substance and Alcohol	Pedestrian Risk
Rashmi and Marisamynathan [37]	√	√	√	√	-	-	-
Shams et al. [38]	√	√	√	√	-	-	-
Mehdizadeh et al. [39]	√	√	√	√	-	-	-
Han et al. [40]	√	√	√	√	-	-	-
Mehdizadeh et al. [41]	√	√	√	√	-	-	-
Naderi et al. [42]	√	√	√	√	-	-	-
Dotse and Rowe [43]	√	√	√	-	-	-	-
Taiwo et al. [44]	√	√	√	√	-	-	-
Maslać et al. [15]	√	√	√	√	-	-	-
Claveria et al. [29]	-	-	-	-	√	-	-
Valenzuela and Burke [30]	-	-	-	-	√	-	-
Zaharaddeen et al. [31]	-	-	-	-	-	√	-
Mabbott and Hartley [32]	-	-	-	-	-	√	-
Gates et al. [33]	-	-	-	-	-	√	-
Dassanayake et al. [34]	-	-	-	-	-	√	-
Poulsen et al. [35]	-	-	-	-	-	√	-
Tyndall [36]	-	-	-	-	-	-	√

, prior studies have primarily focused on only four behavioral components based on the original DBQ: traffic violations, errors, lapses, and aggressive driving. In contrast, other emerging risk factors such as substance or alcohol use, distraction from mobile phone use, and neglect of pedestrian safety have been examined separately in more recent studies. However, to date, no study has comprehensively integrated all seven dimensions into a single framework.

To address this gap, the present study introduces an extended conceptual model that combines the original four DBQ dimensions with the three additional context-specific factors. The resulting framework comprises the following seven behavioral components: (1) traffic violations, (2) errors, (3) lapses, (4) aggressive behaviors, (5) substance or alcohol use, (6) distraction from mobile phone use, and (7) neglect of pedestrian safety. These dimensions represent a comprehensive spectrum of both intentional and unintentional risky driving behaviors, particularly those relevant to the safety challenges of operating large commercial vehicles. To validate the structure of this extended framework, a second-order Confirmatory Factor Analysis (CFA) is conducted. Unlike a first-order CFA, which only tests the independence and loadings of each dimension, a second-order CFA provides additional analytical value by examining whether these seven correlated first-order constructs can be explained by a single higher-order latent factor—overall risky driving behavior. This approach is particularly important in this context because it allows us to test whether risky driving among truck drivers is best represented not merely as a set of separate behaviors but as manifestations of an overarching construct. Demonstrating such a higher-order structure supports the development of more holistic behavioral screening tools and integrated safety interventions specifically designed for the commercial trucking context.

3. Method

3.1. Data Collection and Variables

The questionnaire used in this study consisted of two parts. The first part collected data on the socioeconomic characteristics of truck drivers, while the second part focused on risk-related behavioral indicators, divided into seven categories based on the conceptual framework. These included: Violations (V1–V7, 7 items), Errors (E1–E5, 5 items), Pedestrian-related risk behaviors (P1–P4, 4 items), Lapses (L1–L5, 5 items), Aggressive behaviors (AB1–AB5, 5 items), Distraction from social media use (S1–S3, 3 items), and Substance and alcohol use (Al1–Al3, 3 items). Each item was measured using a six-point Likert rating scale, ranging from 1 (“Never”) to 6 (“Very often”), following the format used in previous studies [16,41].

The study employed Confirmatory Factor Analysis (CFA), focusing solely on the second part of the questionnaire, which contained a total of 32 observed variables. According to the Maximum Likelihood Estimation principle, Pett et al. [45] recommend a minimum sample size of at least 10 respondents per item, suggesting a threshold of no fewer than 320 participants. Accordingly, data collection was carried out with 400 truck drivers in the Northeastern region of Thailand. A self-administered and anonymous questionnaire was used, and participants were recruited on a voluntary basis from eligible drivers who held a valid truck-driving license and were actively employed in the transport sector. From this pool of willing respondents, simple random sampling was applied. The data were collected across four provinces Ubon Ratchathani, Sisaket, Nakhon Ratchasima, and Khon Kaen which serve as major logistics and transportation hubs in the region and included drivers from 18 transport companies to ensure diversity and representativeness. This sample size was deemed adequate and appropriate for the purposes of CFA. Furthermore, the study received ethical approval from the Human Research Ethics Committee of Suranaree University of Technology (COE No.178/2567, 1 December 2024).

3.2. Sociodemographic Characteristics of Truck Drivers

As shown in

Table 2. Socio-economic data of respondents.

Variables	Descriptive	Frequency	Percentage
Gender	Male	395	98.750
	Female	5	1.250
Marital status	Single	142	35.500
	Married	247	61.750
	Divorce	11	2.750
Age (years)	18–29	80	20.000
	30–44	213	53.250
	45–59	103	25.750
	more than 60	4	1.000
Education Level	Primary School	57	14.250
	Junior High School	100	25.000
	High School/Vocational Certificate	161	40.250
	Associate Degree/Higher Vocational Certificate	53	13.250
	Bachelor’s Degree	18	4.500
	Other	11	2.750
Average income (THB/month)	<10,000	2	0.500
	10,000–20,000	276	69.000
	20,001–40,000	113	28.250
	>40,000	9	2.250
Truck driving license	Yes	399	99.750
	No	1	0.250
Driving experience (years)	0–5	238	59.500
	6–10	117	29.250
	11–15	35	8.750
	16–20	8	2.000
	>20 years	2	0.500
Safe Driving Training (times)	0	20	5.000
	1	239	59.750
	2	104	26.000
	3	24	6.000
	>4	13	3.250
Average daily driving time (excluding rest breaks):		10 h/day
Average daily break time during work:		46 min/day
Average daily sleep duration:		7 h/day

, the demographic profile of truck drivers in the sample reveals that the vast majority were male (98.750%), with only 1.250% being female, reflecting the limited participation of women in this occupation. Regarding marital status, 61.750% of drivers were married, followed by 35.500% who were single and 2.750% who were divorced or separated, suggesting potential family-related responsibilities that may contribute to occupational stress and risk. In terms of age, the largest group fell within the 30–44 age range (53.250%), followed by 45–59 years (25.800%) and 18–29 years (20.000%), indicating a concentration of labor in the middle-aged workforce. The most common education level was upper secondary or vocational certificate (40.300%), followed by lower secondary (25.000%) and primary education (14.300%), while only 4.500% held a bachelor’s degree, highlighting generally limited educational attainment. In terms of average monthly income, the majority (69.000%) earned between 10,000–20,000 THB, followed by 28.250% earning 20,001–40,000 THB, and only 2.250% earning over 40,000 THB, reflecting a predominantly low- to middle-income status among truck drivers.

In terms of licensing and experience, nearly all participants (99.750%) possessed a valid truck driving license, confirming their legal eligibility for commercial driving, while only a negligible proportion (0.250%) did not. Regarding professional experience, the majority of drivers (59.500%) had between 0–5 years of experience, followed by 29.250% with 6–10 years, and a much smaller proportion with over 10 years of experience (combined 11.250%). This distribution indicates that the truck driver workforce is relatively young in terms of occupational tenure, which may have implications for both safety awareness and skill maturity.

Concerning safety training participation, 59.750% of drivers reported having received safe driving training once, while 26.000% had attended such training twice. Only 9.250% reported attending three or more times, and 5.000% had never received any training, highlighting a potential area for improvement in continuous professional development and reinforcement of safety standards. Work pattern indicators further illustrate the demanding nature of truck driving. On average, drivers reported 10 h of driving per day, excluding rest periods, and 46 min of break time, reflecting a work environment with long hours and potentially insufficient rest. The average daily sleep duration was 7 h, which meets the lower threshold of the recommended range but may still be inadequate for individuals under physically and mentally demanding conditions like long-distance trucking.

3.3. Data Analysis

The analysis began with preliminary data examination using descriptive statistics, including mean and standard deviation, to understand the general characteristics of the data prior to conducting the Exploratory Factor Analysis (EFA). To ensure the suitability of the data for factor analysis, normality of distribution was assessed as a prerequisite for improving statistical accuracy. This was evaluated using skewness and kurtosis values, with absolute skewness not exceeding 3 [46], and absolute kurtosis not exceeding 7 [47,48]. In addition, multivariate outliers were examined using Mahalanobis D² at the significance level of p < 0.001 [49], which identified 13 cases as potential outliers. However, these cases were not due to data entry errors but rather reflected genuine variability among truck drivers and given their relatively small proportion compared with the total sample, they were retained for further analysis. Finally, construct validity of the questionnaire was assessed through Bartlett’s Test of Sphericity, which evaluates whether the correlation matrix significantly differs from an identity matrix. A p-value less than 0.05 indicates the presence of significant correlations among at least some variables, supporting the suitability of the dataset for factor analysis [50]. Additionally, the Kaiser-Meyer-Olkin (KMO) measure was used to assess the adequacy of sample size for factor analysis. A KMO value greater than 0.80 is considered acceptable and indicates that the data are appropriate for further EFA procedures [51]. To maintain model stability, both EFA and CFA were conducted on the same dataset. The sample consisted of 400 valid cases with 32 observed indicators, which was insufficient to allow reliable splitting into subsamples. Dividing the dataset would likely have produced unstable parameter estimates and reduced model reliability.

3.3.1. Exploratory Factor Analysis (EFA)

The exploratory factor analysis (EFA) in this study was conducted using SPSS Statistics (Version 29) software. Principal Component Analysis (PCA) was employed as the extraction method to reduce the large number of observed variables into a smaller set of interpretable components. PCA has been widely adopted in exploratory research because it is frequently set as the default extraction method in statistical software and has been recommended as a useful preliminary framework for identifying the underlying structure of data [45,52]. Given the aim of this study to reduce 32 observed variables into 7 meaningful components PCA was considered both appropriate and efficient. To enhance interpretability and simplify the factor structure, Varimax rotation with Kaiser normalization was employed. This orthogonal rotation method maximizes the variance of squared loadings across factors, resulting in clearer separation between components and facilitating the extraction of distinct and easily interpretable factors [53]. In accordance with guidelines by Velicer and Fava [54], each retained factor was required to have at least three observed indicators. In this study, sub-factors with factor loadings greater than 0.40 [55], were retained. Additionally, consistent with the recommendation of Hair et al. [56], only factors with an eigenvalue greater than 1 were retained for further analysis. To strengthen this decision, the scree plot was also examined to confirm the appropriate number of factors.

3.3.2. Reliability of the Research Instrument

The reliability of the measurement model was evaluated using two key indicators: (1) Construct Reliability (CR) and (2) Average Variance Extracted (AVE). These indicators assess the internal consistency and convergent validity of the observed variables within each latent construct. The values were calculated using Equations (1) and (2), respectively.

$$CR={\displaystyle \frac{{\left({\displaystyle \sum _{i=1}^n{\mathrm{\beta }}_i}\right)}^2}{{\left({\displaystyle \sum _{i=1}^n{\mathrm{\beta }}_i}\right)}^2+{\left({\displaystyle \sum _{i=1}^n{\mathrm{\delta }}_i}\right)}^2}}$$

(1)

$$AVE={\displaystyle \frac{{\left({\displaystyle \sum _{i=1}^n{\mathrm{\beta }}_i}\right)}^2}{n}}$$

(2)

where:

$\mathrm{\beta }$ is the standardized factor loading of each observed variable.

$\mathrm{\delta }$ is the error variance of each observed variable.

$n$ represents the number of observed indicators.

According to Hair et al. [56], a CR value above 0.70 indicates an acceptable level of construct reliability, reflecting that the set of items consistently represents the underlying latent construct. Furthermore, an Average Variance Extracted (AVE) value greater than 0.50 demonstrates adequate convergent validity, meaning that the indicators share sufficient variance in representing the construct. Additionally, when the AVE of a construct surpasses the squared correlations between that construct and others within the model, it provides evidence of discriminant validity, thereby confirming the conceptual distinctiveness of each latent factor in the measurement model [57].

3.3.3. Confirmatory Factor Analysis (CFA)

Confirmatory Factor Analysis (CFA) was conducted using the maximum likelihood (ML) estimation method to assess the degree of alignment between the hypothesized measurement model and the empirical data. This analysis aimed to evaluate the structural validity of the proposed model in the context of road safety behavior. Several goodness-of-fit indices were used to determine the model’s adequacy, based on widely accepted criteria in CFA literature. Specifically, the chi-square to degrees of freedom ratio (χ²/df) should be less than 5.0 [58], the Comparative Fit Index (CFI) should exceed 0.90 [59,60], and the Tucker–Lewis Index (TLI) should be greater than 0.80 [61]. Additionally, the Root Mean Square Error of Approximation (RMSEA) should not exceed 0.07 [62], while the Standardized Root Mean Square Residual (SRMR) should remain below 0.08 [63]. These thresholds were used as benchmarks to determine whether the second-order CFA model provided a satisfactory representation of the latent constructs related to risky driving behavior among truck drivers.

4. Results

4.1. Descriptive Statistics

As shown in

Table 3. Descriptive statistics.

Code	Variables	Mean	SD	SK	KU
v1	You accelerate when the traffic light turns green in order to overtake other vehicles.	2.020	0.944	0.453	−0.630
v2	You drive through intersections at high speed while the traffic light is switching from green to yellow.	1.920	0.949	0.779	0.027
v3	You overtake in no-passing zones marked with solid lines.	1.820	0.853	0.802	0.006
v4	You exceed the speed limit in school zones, residential areas, or villages.	1.770	0.795	0.641	−0.516
v5	You drive over the speed limit, particularly during late night or early morning hours.	1.920	1.024	0.986	0.464
v6	You overtake slower vehicles on the left-hand side.	1.810	0.906	0.831	−0.119
v7	You remain in a closing lane until the last moment before merging into another traffic lane.	1.740	0.909	1.167	1.172
E1	You misjudge the duration of green lights, making it difficult to stop in time at traffic signals.	2.040	0.869	0.590	0.182
E2	While waiting to turn onto a main road, you focus only on oncoming traffic and nearly collide with the vehicle in front of you.	1.810	0.757	0.501	−0.557
E3	When the road is clear, you proceed through a red light without paying attention to the signal.	1.840	0.864	0.807	0.039
E4	You overtake another vehicle without signaling beforehand.	1.730	0.753	0.627	−0.513
E5	You often ignore “Yield” signs on narrow roads and fail to give way to vehicles with the right of way.	1.740	0.803	0.943	1.143
L1	You often forget to check the rearview or side mirrors before moving off or changing lanes.	1.890	0.723	0.297	−0.662
L2	You frequently react too late and have to brake suddenly to avoid hitting the vehicle in front.	1.770	0.736	0.515	−0.609
L3	You accidentally hit unseen obstacles while reversing the truck.	1.800	0.775	0.520	−0.661
L4	You find yourself driving on a different road than your intended destination.	1.720	0.778	0.746	−0.266
L5	You often drive into restricted zones due to failing to notice “No trucks allowed” signs.	1.690	0.758	0.683	−0.597
AB1	You honk the horn to express annoyance toward other drivers or pedestrians.	1.710	0.765	0.737	−0.090
AB2	You often drive too closely to the vehicle in front and flash your headlights to pressure them to move aside.	1.660	0.759	0.969	0.598
AB3	When angered by another driver’s behavior, you tend to follow their vehicle with the intention of confronting them.	1.530	0.721	1.497	3.075
AB4	You feel irritated or angry when encountering slow drivers and overtake them by switching to the inner lane.	1.570	0.725	1.092	0.824
AB5	You use high beams without concern for the discomfort it may cause to other drivers.	1.390	0.651	1.755	3.351
S1	You often use your mobile phone to make calls or send messages while driving.	2.300	1.110	0.692	0.167
S2	You frequently use your phone for navigation while driving.	2.680	1.299	0.588	−0.158
S3	You often browse social media platforms (e.g., Facebook, Twitter, Instagram, Line) while driving.	1.860	0.855	0.694	−0.094
Al1	You drive after attending a party or festival (e.g., New Year, Songkran) where alcohol was consumed.	1.590	0.915	1.835	4.039
Al2	You drive after taking medication that may cause drowsiness.	1.530	0.870	2.110	5.682
Al3	You consume energy drinks while driving a truck.	2.800	1.740	0.543	−0.960
P1	You veer toward the roadside without checking for pedestrians.	1.830	0.804	0.778	0.908
P2	You often fail to notice pedestrians about to step onto the road, nearly causing an accident.	1.800	0.792	0.524	−0.768
P3	You do not slow down or stop near pedestrian crossings to allow people to cross safely.	1.660	0.804	1.134	1.470
P4	You drive through pedestrian traffic signals without stopping or slowing down.	2.020	0.944	0.453	−0.630

Note: Skewness: SK, Kurtosis: KU, Standard Deviation: SD.

, the observed indicators of risky driving behavior among truck drivers exhibited skewness values ranging from 0.297 to 2.110 and kurtosis values ranging from −0.960 to 5.682, all of which fall within acceptable ranges, indicating that the data approximate a normal distribution. Results of Bartlett’s Test of Sphericity yielded a statistically significant value of χ² = 5754.169 (df = 496, p < 0.001), confirming that the correlation matrix is not an identity matrix and that the variables are sufficiently intercorrelated for factor analysis. Additionally, the Kaiser-Meyer-Olkin (KMO) measure was 0.868, suggesting an adequate sampling adequacy and supporting the suitability of the dataset for further exploratory factor analysis.

4.2. Results of Exploratory Factor Analysis

As presented in

Table 4. Result of EFA.

Code	Component
	1	2	3	4	5	6	7
v2	0.788	0.234	0.271	0.056	0.078	0.023	0.027
v1	0.776	0.164	0.232	−0.020	0.065	−0.031	−0.062
v3	0.665	0.240	0.126	0.102	0.053	0.162	0.104
v6	0.618	0.157	−0.097	0.083	0.169	0.143	0.014
v7	0.563	0.231	−0.094	0.267	0.159	0.151	0.020
v5	0.543	0.339	0.028	0.137	−0.051	0.361	0.196
v4	0.483	0.419	0.036	0.118	−0.071	0.205	0.050
E3	0.270	0.686	0.116	0.072	0.077	0.129	0.082
E1	0.305	0.665	0.187	0.159	0.136	0.040	0.042
E2	0.210	0.639	0.147	0.287	0.170	−0.021	−0.046
E5	0.149	0.592	0.055	0.165	0.096	0.178	0.092
E4	0.251	0.581	0.182	0.205	0.055	0.062	0.016
P4	0.016	0.147	0.723	0.081	0.067	0.081	−0.048
P3	0.004	0.102	0.722	0.098	0.107	0.207	0.110
P1	0.201	0.116	0.718	0.127	0.147	0.119	−0.021
P2	0.120	0.113	0.711	0.156	0.127	0.133	−0.028
L5	−0.038	0.227	0.088	0.711	0.131	0.040	0.060
L4	0.261	0.122	0.047	0.660	0.063	0.257	0.209
L3	0.088	0.185	0.144	0.614	−0.076	0.238	0.104
L1	0.227	0.342	0.325	0.537	0.115	0.086	−0.108
L2	0.401	0.265	0.252	0.458	0.212	0.023	0.033
AB5	0.050	0.001	0.103	0.166	0.699	−0.072	−0.085
AB3	0.008	0.208	0.056	0.042	0.684	0.296	0.033
AB2	0.178	0.272	0.242	−0.096	0.676	0.236	0.005
AB4	0.339	−0.079	0.161	0.356	0.532	−0.052	0.122
AB1	0.136	0.358	0.398	−0.202	0.479	0.275	0.043
S2	0.191	0.115	0.116	0.190	0.027	0.787	0.014
S1	0.165	0.139	0.308	0.103	0.122	0.716	−0.030
S3	0.082	0.079	0.218	0.151	0.235	0.570	0.011
Al2	−0.027	0.071	−0.107	0.116	−0.021	−0.073	0.906
Al1	−0.029	0.144	−0.053	0.121	−0.015	−0.025	0.889
Al3	0.263	−0.072	0.199	−0.009	0.039	0.156	0.560
Eigenvalue	9.155	2.526	1.983	1.529	1.443	1.326	1.166
% of variance	28.609	7.895	6.198	4.779	4.509	4.143	3.645
Cumulative %	28.609	36.504	42.702	47.481	51.99	56.133	59.778

, the exploratory factor analysis (EFA) identified seven principal components from a total of 32 observed indicators measuring risky driving behaviors among truck drivers, collectively explaining 59.778% of the total variance. The first component, Traffic Violations (V1–V7), accounted for 28.609% of the variance, followed by Driving Errors (E1–E5) at 7.895%, Pedestrian-Related Risk Behaviors (P1–P4) at 6.198%, Lapses (L1–L5) at 4.779%, Aggressive Behaviors (AB1–AB5) at 4.509%, Social Media Distraction (S1–S3) at 4.143%, and Substance and Alcohol Use (Al1–Al3) at 3.645%. These findings indicate that risky driving behaviors among truck drivers can be clearly categorized into distinct behavioral constructs, confirming the multidimensionality of the framework and supporting its appropriateness for further confirmatory factor analysis in the subsequent phase of the study. In addition, the scree plot (Figure 1) demonstrated a clear inflection after the seventh component, further supporting the retention of seven factors as the most appropriate solution.

4.3. Result of Confirmatory Factor Analysis

The second-order CFA was conducted to examine whether the seven first-order latent constructs could be integrated into a higher-order factor representing overall risky driving behavior among truck drivers, thereby validating the theoretical structure identified through EFA. Construct validity was evaluated using CR and AVE with the results summarized in

Table 5. Result of 2nd Order CFA.

Factor	Code	Beta	p-Value	Cronbach’s Alpha	CR	AVE
Violations	V1	0.768	0.000	0.850	0.889	0.534
	V2	0.805	0.000
	V3	0.764	0.000
	V4	0.748	0.000
	V5	0.730	0.000
	V6	0.635	0.000
	V7	0.650	0.000
Errors	E1	0.778	0.000	0.800	0.842	0.517
	E2	0.758	0.000
	E3	0.688	0.000
	E4	0.711	0.000
	E5	0.654	0.000
Lapses	L1	0.677	0.000	0.780	0.835	0.507
	L2	0.836	0.000
	L3	0.598	0.000
	L4	0.760	0.000
	L5	0.664	0.000
Aggressive Behavior	AB1	0.770	0.000	0.740	0.841	0.514
	AB2	0.753	0.000
	AB3	0.676	0.000
	AB4	0.674	0.000
	AB5	0.706	0.000
Social Distraction	S1	0.764	0.000	0.740	0.758	0.512
	S2	0.748	0.000
	S3	0.627	0.000
Substance and Alcohol Use	AL1	0.948	0.000	0.730	0.791	0.594
	AL2	0.889	0.000
	AL3	0.306	0.000
Pedestrian Risk	P1	0.769	0.000	0.790	0.876	0.638
	P2	0.793	0.000
	P3	0.828	0.000
	P4	0.803	0.000

Note: CR = Composite Reliability, AVE = Average Variance Extracted.

, with CR values ranged from 0.758 to 0.889, exceeding the minimum recommended threshold of 0.70 suggested by Hair [59], indicating good internal consistency and reliability of the observed variables in capturing the respective constructs. The AVE values ranged from 0.507 to 0.638, surpassing the 0.50 benchmark proposed by Fornell and Larcker [57], confirming convergent validity. In addition, the model fit indices, as presented in

Table 6. Goodness-of-Fit Summary for Second-order CFA Model.

Fit Index	Acceptable Threshold	Reference	Second-Order CFA Value	Fit Decision
X2/df	<5.000	Wheaton et al. [58]	2.122	Pass
CFI	>0.900	Hair [59], Bollen [60]	0.913	Pass
TLI	>0.800	Hooper et al. [61]	0.897	Pass
RMSEA	<0.070	Browne and Cudeck [62]	0.053	Pass
SRMR	<0.080	Steiger [63]	0.079	Pass
CR	>0.700	Hair [59]	0.874	Pass
AVE	>0.500	Hair [59]	0.508	Pass

, (X²/df = 2.122, CFI = 0.913, TLI = 0.897, RMSEA = 0.053 and SRMR = 0.079), demonstrated a good fit with the empirical data, indicating that the measurement model adequately represented the multidimensional structure of risky driving behaviors among truck drivers.

When examining the standardized factor loadings (γ) from the second-order CFA in Figure 2 and Table 5, traffic violations emerged as the most influential latent factor associated with overall risky driving behavior among truck drivers (Violation: γ = 0.949). The highest-weighted item was running through intersections as the lights changed from green to yellow (0.805), followed by aggressively accelerating at green lights to overtake other vehicles (0.768), while overtaking slower vehicles by shifting to the left lane had the lowest loading (0.635). These findings indicate that violations driven by impatience and time pressure are most critical, although less overt maneuvers also contribute to risk.

The next influential factor was driving errors (Error: γ = 0.835), which encompassed behaviors resulting from poor judgment or a lack of attentiveness. These include misjudging the duration of green lights (0.778), followed by failing to observe the vehicle in front when merging (0.758), while ignoring “yield” signs on narrow roads had the lowest loading (0.654). This suggests that perceptual errors and inattentiveness at intersections or during merging carry greater risk than sign neglect.

Another important dimension was lapses (Lapses; γ = 0.770), with delayed reactions requiring sudden braking showing the highest loading (0.836), followed by wrong-route driving (0.760), while colliding with unseen obstacles when reversing was the lowest (0.598). These findings emphasize the critical role of reaction time in truck driving, whereas navigation mistakes and reversing errors, though less weighted, still present safety concerns.

In addition, the substance and alcohol use factor (substance and alcohol; γ = 0.648) was associated with drinking alcoholic beverages during festive periods (0.948), followed by taking drowsiness-inducing medication before driving (0.889), while frequent energy drink consumption had the lowest loading (0.306). This highlights the direct impairment effects of alcohol and sedatives, while energy drinks may mask fatigue but contribute less strongly.

The analysis also revealed that social distraction (Social Distraction; γ = 0.578) was associated with mobile device use. Talking or texting while driving had the highest loading (0.764), followed by using a phone for navigation (0.748), while social media use had the lowest loading (0.627). These results suggest that active phone engagement poses the greatest distraction, while social media reflects a lower but still notable risk.

Furthermore, the aggressive driving factor (Aggressive Behavior; γ = 0.566) showed its strongest loading for honking to express annoyance (0.770), followed by tailgating and flashing headlights (0.753), with irritation at slow vehicles merging into the inner lane being the lowest (0.674). This indicates that overt, externally directed aggression has greater weight than internal irritation.

Lastly, the pedestrian-related risk (Pedestrian Risk; γ = 0.526) had the lowest factor loading among all constructs. The highest loading was for failing to slow down at pedestrian crossings (0.828), followed by disregarding pedestrian signals (0.803), while veering toward the roadside without checking for pedestrians was the lowest loading (0.769). These results highlight that neglect of pedestrian right-of-way represents the most severe risks, though lateral inattentiveness also compromises vulnerable road user safety.

In addition, the examination of standardized residuals and modification indices revealed that certain pairs of items showed significant residual correlations, particularly those with similar content and contexts, such as speeding and overtaking behaviors, observation and decision-making errors, aggressive violations, and pedestrian-related risks. Allowing residual covariances for these pairs was therefore theoretically and substantively justified. Full statistical details and explanations of the adjusted item pairs are provided in the Appendix A.

5. Discussion

The results of the confirmatory factor analysis clearly revealed the multidimensional structure of risky driving behaviors among truck drivers. These behaviors can be categorized into seven distinct dimensions: intentional violations, perceptual errors, lapses, aggressive behaviors, substance use, distraction from social media, and behaviors that endanger pedestrians. The discussion in this section focuses on an in-depth analysis of the most influential factors, linking them to relevant theories and empirical findings from previous studies. This approach aims to reflect the complexity of occupationally specific risk behaviors within the context of commercial truck driving.

The analysis revealed that traffic violations emerged as the most influential dimension of risky driving behavior among truck drivers. This factor reflects intentional breaches of traffic laws and safety regulations, which had the strongest overall impact on driving risk. The findings indicate that deliberate rule-breaking constitutes a primary safety concern in commercial trucking, consistent with prior research suggesting that violations remain a dominant risk factor, even when newly emerging behavioral dimensions are introduced especially in the context of large vehicles [64].

The findings of this study reaffirm that traffic violations remain a significantly influential risk factor, even when newly emerging behavioral dimensions are incorporated into the analysis particularly in the context of large commercial vehicles. This result aligns with prior studies [37,38,40,41,42,43] that have consistently identified intentional violations as a dominant predictor of crash risk among professional drivers. Among the most concerning behaviors were running intersections during yellow lights, accelerating to overtake as the light turns green, and overtaking in no-passing zones, all of which represent blatant breaches of fundamental road safety principles. These actions become even more alarming when considering the mass and momentum of heavy trucks, which can greatly amplify crash severity when such violations occur [65].

Additional violations such as speeding in residential or school zones, exceeding speed limits during nighttime hours, and frequent lane-related violations like driving in closing lanes until the last moment, abrupt lane changes, and improper overtaking maneuvers reflect a systematic disregard for traffic regulations. These behaviors pose particularly high risks for trucks, which have limited maneuverability, wider turned radii, and longer stopped distances compared to passenger vehicles. As such, these findings underscore the need for stricter enforcement, professional training, and context-specific interventions to address the persistent issue of traffic violations in the commercial transport sector [66,67,68].

The driving errors dimension emerged as the second most influential factor in explaining risky driving behavior among truck drivers. This category encompasses behaviors arising from faulty decision-making, inadequate hazard perception, or misjudgment of traffic situations, all of which reflect complex cognitive processes essential for safely operating large vehicles in dynamic traffic environments. Unlike traffic violations, which are intentional, these errors represent unintentional cognitive failures that may occur despite the absence of deliberate risk-taking [9]. Among the most significant errors identified were misjudging the duration of green lights and failing to observe other vehicles while merging into traffic, both of which indicate deficiencies in risk assessment and situational awareness. These skills are particularly critical for truck drivers, who require more time and distance to decelerate or change direction. Misjudgments in such contexts can lead to disproportionately severe consequences due to the size and weight of commercial trucks [66]. Additional error-related behaviors, such as overtaking without signaling and driving through red lights on seemingly empty roads, suggest inappropriate risk calculation possibly caused by incomplete information processing or a lack of awareness regarding vehicle-specific limitations including braking distance, blind spots, and turning radius [67,69]. Furthermore, ignoring “Yield” signs on narrow roads reflects a fundamental misunderstanding of road priority and right-of-way, a particularly problematic issue for drivers of large vehicles that require greater space and time for maneuvering compared to standard passenger cars.

Another key factor identified was lapses, referring to risky driving behaviors stemming from momentary cognitive failures or inattentiveness. These lapses reflect temporary breakdowns in the mental processes required for safe truck operation. Among the most critical behaviors in this category was delayed reaction requiring sudden braking, which directly compromises safety by reducing a driver’s ability to respond to rapidly changing road conditions. Given that trucks require significantly longer braking distances than passenger vehicles, even slight delays in reaction time can lead to severe collisions [70,71]. Other notable behaviors include getting lost or confused about the travel route, which may indicate deficiencies in spatial memory or navigation, potentially prompting abrupt corrective maneuvers particularly hazardous in large trucks with limited turning capability and maneuverability [38,40]. Additionally, failing to check side mirrors before changing lanes represents a serious error, especially considering the extensive blind spots typical of large trucks and the critical role of mirrors in ensuring safe lane changes [67].

Another common lapse involved entering restricted zones due to failing to notice truck-prohibited signs, which highlights a lack of attention to traffic signage specifically targeted at commercial drivers. Lastly, colliding with unseen obstacles while reversing is an example of unintentional oversight with potentially serious consequences, given the visual and spatial limitations associated with operating heavy vehicles [37].

Substance and alcohol use emerged as a moderately influential factor in shaping risky driving behavior among truck drivers. As one of the three newly introduced dimensions beyond the original DBQ framework, this construct addresses an important gap in evaluating substance-related risks, drawing upon prior studies that examined stimulant and alcohol use in the trucking population [31,32,33]. Among the indicators in this dimension, alcohol consumption during festivals and holidays was the most concerning, suggesting that some drivers may struggle to separate personal celebration from professional responsibility a disconnect that can lead to critical safety consequences [72]. Another high-risk behavior involved driving after taking drowsiness-inducing medication, which is particularly problematic in the trucking industry where irregular work schedules often disrupt normal sleep patterns. Drivers may take such medications to alleviate illness or sleep disturbances, but when combined with physical and mental fatigue from extended driving hours, the risk of accidents increases significantly [34]. Although the indicator for energy drink consumption (AL3) exhibited a relatively low loading, it was retained in the model to preserve content validity. Stimulant use reflects a theoretically distinct but practically important risk mechanism in long-haul trucking, functioning through fatigue compensation and arousal regulation rather than psychomotor impairment. Retaining this item ensures that the Substance Use construct captures both impairment-related and stimulant-related risky behaviors, thereby reflecting the full spectrum of substance-related risks faced by truck drivers.

Additionally, the frequent use of energy drinks while on duty highlights a context-specific coping mechanism in commercial driving. Drivers may rely on these beverages to combat fatigue caused by long hours, unpredictable shifts, and delivery deadlines. However, when categorized under stimulant use, it becomes evident that excessive caffeine intake can impair driving performance and elevate crash risk, particularly when followed by rebound fatigue once the stimulant effects wear off [33,35].

Distraction from social media emerged as a moderately influential factor affecting risky driving behavior and represents a newly introduced dimension that extends the traditional DBQ framework. This construct reflects the growing influence of technology and digital connectivity among commercial drivers. Particularly concerning behaviors include talking on the phone, texting, and using mobile devices for navigation while driving all of which divert attention from vehicle control. Such distractions are especially hazardous in the context of truck driving, which requires sustained concentration and situational awareness [29]. The use of social media while driving suggests that some drivers may engage in technologically mediated activities that are incompatible with the demands of professional driving. These behaviors compromise traffic monitoring and reaction times, representing clear violations of safety standards and professional responsibilities in the transportation sector [30]. Notably, the Social Media Distraction dimension showed higher mean values, such as frequent phone use for navigation (M = 2.680). This likely reflects drivers’ perception of mobile phone use as a functional necessity rather than intentional rule breaking. However, even when work-related, such behaviors still create serious distraction and safety risks. Although this dimension showed a relatively low factor loading in the second-order CFA, this may reflect drivers’ perception of mobile phone use as a functional necessity rather than an intentional violation. Nevertheless, prior studies have confirmed that DBQ combined with mobile phone or technology use is a strong predictor of risky driving and crash involvement [73,74].

Aggressive driving behavior emerged as a moderately influential factor in overall risky driving and remains one of the core dimensions of the original DBQ framework. Despite being a traditional construct, it continues to be highly relevant in the context of commercial driving, which involves unique challenges such as large vehicle sizes, strict professional standards, and occupational pressure. The most concerning behaviors within this dimension include honking to express frustration, tailgating with flashing headlights to force vehicles ahead to move, and inappropriate use of high beams. These actions are particularly alarming when performed by truck drivers, as the large size and visibility of commercial vehicles can intimidate other road users, potentially provoking road rage incidents and contributing to broader traffic safety risks.

Driving aggressively behind slow-moving vehicles as a means of expressing dissatisfaction may reflect work-related stress, such as tight delivery schedules, congested traffic conditions, or time constraints. These pressures can lead to emotional fatigue, which may manifest as aggression on the road. In commercial transport settings, drivers frequently face multiple stressors, including rigid time demands, unsuitable traffic infrastructure for large trucks, customer expectations, strict compliance regulations, and the psychological toll of long-distance and often isolated driving. When these pressures accumulate, they may heighten the likelihood of aggressive driving, posing a significant threat not only to the drivers themselves but also to the safety of surrounding road users [75,76].

Although pedestrian-related risk emerged as the least influential dimension among all constructs, its inclusion in the extended DBQ framework represents a significant academic contribution, particularly in addressing the interaction between trucks and vulnerable road users. This new dimension highlights safety concerns that are frequently overlooked in traditional driving behavior research but are of critical importance in the trucking context, where collisions with pedestrians often result in severe injuries or fatalities [36]. The most concerning behaviors in this category include failing to slow down or stop at pedestrian crossings, disregarding pedestrian traffic signals, and failing to notice pedestrians emerging from the roadside. These actions reflect a lack of awareness and concern for pedestrian safety, which can lead to catastrophic outcomes. Moreover, veering toward the edge of the road without checking for pedestrians further reinforces the elevated risk in this domain, particularly when considering the physical characteristics of trucks such as high driver seating positions and extensive blind spots that limit pedestrian visibility and complicate spatial judgment.

These findings underscore the need for targeted training to improve hazard perception in areas with limited visibility and to raise awareness of pedestrian safety risks. Such training is especially critical in the commercial trucking sector, where professional drivers bear a heightened responsibility for minimizing harm to vulnerable road users in complex traffic environments.

The results of the modification indices further highlighted the complexity of risky driving behaviors among truck drivers, showing that some behaviors, although belonging to different dimensions, were significantly correlated. For example, honking out of annoyance (AB1) was associated with speeding through yellow lights (V2), reflecting impatience and time pressure in truck driving. Mobile phone use while driving (S1, S2) was correlated with failures in traffic observation (E2) and wrong-route driving (L6), indicating the direct impact of distraction on driver attention. Similarly, alcohol-impaired driving (AL1) was linked with neglecting yield signs (E5), pointing to impaired decision-making under intoxication. These associations confirm that risky driving behaviors among truck drivers do not occur in isolation but overlap and interact under real working conditions. Allowing residual covariances only for theoretically justified pairs not only improved the model fit but also provided a more realistic representation of on-road behaviors in this occupational group.

6. Conclusions

6.1. Overall Conclusion of the Study

This study aimed to explore and validate the structural dimensions of risky driving behavior among truck drivers in Thailand. The research began with the development of a questionnaire comprising 32 observed indicators, covering both traditional DBQ dimensions violations, errors, lapses, and aggressive behaviors as well as additional factors reflecting the context-specific nature of commercial truck driving. These extended dimensions included substance and alcohol use, social media distraction while driving, and pedestrian-related risk behaviors.

In the preliminary analysis, Exploratory Factor Analysis (EFA) was employed to examine the underlying structure of the dataset. The results supported the hypothesized classification into seven distinct factors. Subsequently, Confirmatory Factor Analysis (CFA) was conducted to test the model’s goodness of fit with the empirical data. As presented in Table 6, The CFA results confirmed a clear and robust factor structure of risky driving behavior among truck drivers, identifying seven latent constructs derived from a total of 32 observed indicators, as illustrated in Figure 2. Each construct represents a unique dimension of risky driving behavior, categorized as follows:

• Traffic Violations: This dimension comprises seven indicators that reflect intentional disregard for traffic rules, such as speeding, running red lights, or overtaking in no-passing zones. These behaviors are often driven by time pressure, urgency to meet delivery deadlines, or familiarity with the route that leads to an underestimation of risk. In the context of trucking, such violations are particularly hazardous due to the large size, heavy weight, and longer braking distances required for commercial vehicles, which significantly amplify the consequences of unsafe maneuvers.
• Driving Errors: Consisting of five indicators, this dimension captures unintentional misjudgments, such as merging without checking blind spots or overtaking without signaling. Unlike violations, these errors arise not from deliberate rule-breaking but from fatigue, inattention, or vehicle-specific limitations including reduced visibility, high driver seating positions, and the challenges of navigating large trucks in confined or narrow spaces.
• Lapses: This factor includes five indicators related to momentary lapses in concentration or memory, such as forgetting to check mirrors, overlooking road signs, or delayed braking. These behaviors are common among long-haul drivers working extended hours without adequate rest. Both physical and mental fatigue contribute significantly to the emergence of such inattentive behaviors.
• Aggressive Behavior: This dimension consists of five indicators, such as tailgating, honking to express impatience, or using high beams to pressure other drivers. These actions reflect negative emotional expression while driving, often stemming from stress, workload pressure, or dissatisfaction encountered during professional driving tasks.
• Substance and Alcohol Use: Comprising three indicators, this factor reflects the use of alcohol or medications such as painkillers or tranquilizers particularly in response to fatigue or stress. For example, drinking during holidays or using sedatives due to irregular sleep patterns is still reported among some truck drivers. Although these behaviors may be intended to alleviate exhaustion, they can severely impair judgment, reaction time, and vehicle control.
• Social Media Distraction: This factor includes three indicators that capture inappropriate technology use while driving, such as texting, calling, or using navigation apps. While such behaviors may be normalized in the digital age, they pose greater risks in the trucking context, where large vehicle size, slower maneuverability, and longer response times make any visual distraction even for a few seconds potentially fatal in emergency situations.
• Pedestrian-related Risk: This dimension comprises four indicators involving failure to slow down for pedestrians at crossings or driving too close to the roadside in community areas. Although this factor carried the lowest weight in the overall structure, it is nonetheless critical from a safety perspective, as collisions between trucks and pedestrians often result in severe injury or fatality. Truck-specific characteristics, such as extensive blind spots and elevated driver seating positions, further limit visibility and increase the likelihood of pedestrian-related incidents, highlighting the need for enhanced attention to vulnerable road users in commercial driving contexts.

The findings of this study underscore the need for targeted training programs to improve hazard perception in areas with limited visibility and to enhance awareness of pedestrian safety risks. Such training is particularly critical in the commercial trucking sector, where professional drivers carry a heightened responsibility for minimizing harm to vulnerable road users in complex traffic environments. At the same time, the examination of standardized residuals and modification indices indicated that certain risky behaviors do not occur in isolation but are interrelated in real-world truck driving contexts for example, aggressive behavior linked with speeding violations, or mobile phone use associated with observation errors. These findings confirm that risky behaviors in this occupational group are complex and interconnected across multiple dimensions. The model developed in this study therefore not only provides a more comprehensive understanding of risky driving behaviors among truck drivers but also serves as a practical tool for behavioral risk assessment, the design of targeted training programs, and the development of safety policies tailored to the working conditions of truck drivers.

6.2. Practical Implications for Training, Regulation, and Safety Interventions

Based on the results, all proposed hypotheses were empirically supported, confirming the multidimensional structure of risky driving behavior and the significant influence of both traditional and newly extended DBQ dimensions. These validated findings allow for the development of concrete, context-specific implications to inform safety management in the Thai trucking sector:

• Scenario-based training modules. Since intentional traffic violations emerged as the most influential factor, training programs for truck drivers should focus on decision-making in high-risk scenarios such as intersections, overtaking, and lane-changing under time pressure. Scenario-based training that vividly illustrates the severe consequences of such violations in heavy vehicles can strengthen hazard awareness and foster safer behavioral norms.
• Enhancing hazard perception and attention management. The prominence of perceptual errors and lapses highlights the need to integrate hazard perception and attentional control into professional driver curricula. Exercises such as blind-spot monitoring, systematic mirror checks, and reaction-time training under fatigue should be incorporated into both licensing renewal and company training schemes.
• Managing technology-related distractions. The identification of mobile phone and navigation-device use as a new risky driving dimension underscores the necessity of embedding digital-distraction management strategies into both training and regulatory frameworks. Raising awareness of distraction-related impairments and implementing restrictions on mobile phone use while driving can help mitigate these risks.
• Strengthening regulatory enforcement. The findings support stricter enforcement of existing standards regarding speeding, alcohol and substance use, and working-hour limitations. Logistics companies and enforcement agencies could adopt systematic monitoring measures, such as random alcohol testing, GPS-based telematics to track speed and lane violations, and fatigue management systems to ensure compliance.
• Infrastructure measures for pedestrian protection. The inclusion of pedestrian-related risk in this study highlights the importance of interventions beyond the driver level. Infrastructure measures, such as clearer pedestrian signage along trucking routes and the establishment of designated crossing zones, can reduce potential conflicts between heavy trucks and vulnerable road users.

Taken together, these implications not only provide actionable strategies for improving training, regulatory compliance, and infrastructure design, but also highlight the unique contribution of this study in extending the DBQ framework to the occupational context of commercial trucking. By capturing emerging risks such as technology-related distraction and pedestrian interactions, the findings move beyond traditional dimensions of risky driving and offer evidence-based guidance for policymakers, enforcement agencies, and logistics companies. Implementing these measures in Thailand would not only strengthen driver competence and organizational safety culture but also contribute to reducing the broader societal burden of truck-related crashes.

7. Limitation and Future Work

This study has several limitations that should be taken into account when interpreting the findings. Most notably, both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were conducted on the same dataset, which may increase the risk of capitalizing on chance. This decision, however, was driven by practical constraints: with 32 observed variables and only 400 participants, splitting the sample into two subsamples would likely have produced unstable parameter estimates and reduced model reliability. To address this limitation, future studies should validate the measurement structure using independent samples or employ cross-validation techniques to strengthen the stability and generalizability of the model. In addition, the reliance on self-reported questionnaire data alone reflects drivers’ perceptions at the time of data collection and is subject to potential reporting bias, which may not fully capture actual on-road behavior.

Another limitation concerns the gender distribution of the sample. Only 1.250% of respondents were women, which mirrors the demographic imbalance of the trucking industry in Thailand, where female drivers remain extremely underrepresented. As a result, the findings primarily reflect the behavioral patterns of male truck drivers, and caution should be exercised when generalizing the results to female drivers.

These limitations also highlight directions for future research. Subsequent studies should seek to assess the stability of the identified multidimensional structure over time and integrate alternative data sources, such as GPS-based driving behavior monitoring, to detect deviant behaviors under real-world work pressures more effectively. In addition, the application of mediation or multilevel modeling could provide deeper insights into cross-level mechanisms operating at both the individual and organizational levels. For instance, safety climate, enforcement rigor, and scheduling systems may jointly shape risky driving behavior. Although this study highlighted occupational pressures such as time constraints and delivery schedules as important contextual factors, mediating variables such as fatigue, scheduling practices, and job stress were not directly tested. Future research should therefore incorporate these psychological and organizational variables into the analytical framework to clarify how occupational demands are translated into unsafe driving behaviors. Furthermore, it is recommended that future research incorporate mediating and moderating variables such as fatigue, scheduling practices, and income using multi-group CFA or regression-based approaches to strengthen explanatory power and enhance the practical value of the findings for designing more targeted safety interventions.

At the same time, future research should expand the scope to other contexts of commercial driving, including long-haul transport, cross-border logistics, and last-mile delivery, to evaluate whether the identified behavioral structure can be generalized across different groups. Moreover, the inclusion of psychological factors such as fatigue, job satisfaction, and risk perception may provide deeper insight into the cognitive and emotional mechanisms that contribute to unsafe driving.

Finally, although this study introduced Pedestrian-Related Risk items to capture unsafe interactions with vulnerable road users, these were measured using generalized statements rather than scenario-specific distinctions. Future studies should consider incorporating scenario-based items to provide greater precision in capturing situational variations in pedestrian safety risks. Taken together, these approaches will support the development of more robust predictive models of risky driving behavior and lay the groundwork for safety interventions that are more responsive to the realities of the transport sector.