liuimpactconstructionarealayout2025.md 104 KB
category: literaturenote citekey: liuimpactconstructionarealayout2025 title: Impact of the construction area layout on road safety in urban work zones authors: "Liu, Peng; Zhang, Chengyi; Arditi, David; Olorunnishola, Ayoola" year: 2025 date: 2025-00-00 2025 doi: 10.1016/j.aap.2025.108092 publication: "Accident Analysis & Prevention" url: "https://doi.org/10.1016/j.aap.2025.108092" zotero_key: QB8ZGNJI zotero_storage: CTN2SNHM collections: transpordi_planeerimine folder: 001_artiklid/Road Safety Indicators firstAuthor: "Liu, Peng"
status: converted
Contents lists available at ScienceDirect
Accident Analysis and Prevention
journal homepage: www.elsevier.com/locate/aap
Impact of the construction area layout on road safety in urban work zones☆
Peng Liu a , Chengyi Zhang a,* , David Arditi b , Ayoola Olorunnishola a
- a Department of Civil and Architectural Engineering and Construction Management, University of Wyoming, Laramie, WY 82071, USA
- b Department of Civil, Architectural, and Environmental Engineering, Illinois Institute of Technology, Chicago, IL 60616, USA
ARTICLE INFO
Keywords: Urban work zones Construction area layout Driver gender Ambient light condition Driver behaviors
ABSTRACT
Despite the crucial role that work zone configurations play in traffic safety, there is a limited understanding of work zone configuration from the perspective of construction area layout within urban work zones impacts overall safety. This study addresses this gap by examining the critical influence of construction area layout on lane-changing, driving behavior passing by the construction area, and driving stability. Construction area layout was reflected by the position of heavy equipment in this study. A driving simulator experiment was conducted with 26 participants (14 males and 12 females) to simulate real-world urban work zone scenarios and assess the impact of construction area layout on safety. The experiment also considered two key safety-related independent variables: driver gender and ambient light condition. To evaluate driver behaviors and identify safety–critical patterns, parametric survival modeling, Multivariate analysis of variance (MANOVA), and regression analysis were employed. The findings highlight the significant impact of the construction area: (1) the construction area layout whereby heavy equipment was positioned closer to the center of the work zone (Position 2) prompted drivers' merging distance to be 14.6% longer, underscoring the importance of heavy equipment at the start of the work zone (Position 1), and (2) Position 2 enables drivers to pass the work zone with a higher speed to pass by the construction area. The driver gender and ambient light condition can also have a significant effect. For example, the increased longitudinal velocity was observed during nighttime, suggesting a need for enhanced visibility and speed control. Male drivers tend to pass by the construction area with a more stable longitudinal velocity than female drivers. These findings are significant for improving work zone safety through careful consideration of construction area layout design, enhanced ambient lighting condition, and strategic safety interventions for specific groups. These insights offer valuable guidance for improving safety and operational efficiency in urban work zones, reducing the risk of accidents, and safeguarding both drivers and construction personnel.
1. Introduction
In recent years, the increased demand for transportation capacity has spurred frequent road maintenance, improvement, and construction activities, which enable a great number of work zones to appear in the traffic network (Duan et al., 2020). Work zones in traffic network not only lead to a reduction in transportation capacity and an increase in traffic congestion but also impact traffic safety (Fei et al., 2016; Meng & Weng, 2011; von der Heiden & Geistefeldt, 2016). According to the National Institute for Occupational Safety and Health, 2,222 were construction workers who lost their lives at work zones between 2003 and 2020 and 63 % of transportation-related work zone accidents involved workers being struck by vehicles either intruding from external traffic lanes or operating within the work zone itself, resulting in 577 fatalities from 2011 to 2020 (Demeke et al., 2024). A standard work zone consists of construction areas, buffer zones, and transition areas, with proper layout being crucial for ensuring safe vehicle passage (Herrero-Perez & Martinez-Barbera, 2010). Within the framework, the construction area is the section of the work zone where actual construction or maintenance activities take place, often involving active use of heavy machinery, construction materials, and personnel working on or near the roadway (Shen et al., 2016). The design of work zones is primarily governed by guidelines from the Manual on Uniform Traffic Control Devices (MUTCD), which provides standards for traffic signs and control devices to guide vehicles safely through construction zones ensure. MUTCD aims to establish safe work zones by implementing specific
E-mail addresses: pliu3@uwyo.edu (P. Liu), chengyi.zhang@uwyo.edu (C. Zhang), arditi@iit.edu (D. Arditi), aolorunn@uwyo.edu (A. Olorunnishola).
☆ This article is part of a special issue entitled: 'Transportation Hub Safety' published in Accident Analysis and Prevention.
* Corresponding author.
signs, barriers, and markings that alert drivers to lane closures, reduced speeds, and detours, thereby enhancing safety for workers and drivers. However, the current design of construction areas often lacks a comprehensive focus on safety considerations, which presents a critical area for improvement.
Navigating through work zones presents several challenges for drivers. The presence of traffic signs, construction workers, and equipment increases the complexity of the driving environment (Theofilatos et al., 2017). Additionally, drivers often adjust their driving habits to accommodate the altered conditions in these zones (Raju et al., 2020). For example, a lane needs to be closed to make room for a work zone and drivers need to change lanes to pass by the area. The configuration of the work zone itself plays a significant role as it directly influences the safety and efficiency of traffic flow in these areas. A large number of researchers have conducted many relevant studies in this field. For example, Duan et al. (2020) investigated lane-merging behaviors in work zone areas. According to their study, the position of the lane-end sign influenced the initiation point for lane changes and the merging distance. In the process of lane-merging, driver gender also presented a significant influence on the merging distance. Shakouri et al. (2014) studied the effect of conventional lane merge (CLM) and joint lane merge (JLM) and traffic density on driver behaviors and their subjective workload when passing by the area. They found that JLM can lower the maximum brake force and lead to a reduction in drivers' subjective workload. Based on their findings, JLM is an effective measure to enhance safety in the work zone areas. Ma et al. (2023) presented how drivers responded to work zones, including drivers' workload and behaviors when passing by the work zone with different transition areas. They found that expanding the number of lanes in the transition area can lead to an increase in driving workloads and vehicle speed, which can offer valuable insights for enhancing traffic management and ensuring safety in work zone areas on freeways. Petzoldt et al. (2016) investigated the influence of left lane width on driver behavior in highway work zones. Their study revealed that the width of the left lane did not significantly affect drivers' willingness to overtake. However, the findings indicated that drivers do perceive differences in lane width, which in turn increases their perceived stress levels. This perception leads to some behavioral adaptations among drivers.
While these studies provide valuable insights into drivers in different work zone configurations, there is a lack of research on the work zone configuration from the perspective of construction area layout for safety. Construction area layout refers to the organized arrangement of physical and operational elements on a construction site. They involve planning the positioning of temporary structures, access routes, work zones, material storage areas, machinery, and worker facilities to optimize efficiency, safety, and project management (Al Hawarneh et al., 2021). This study focuses on the influence of construction area layouts, which not only can offer useful recommendations to optimize the work zone configuration, but also highlight the need to develop new intervention measures to enhance the safety in these areas. In addition to the construction area layouts, the driver gender and the ambient light condition are also considered in this study because the significant influence of both driver gender and ambient light condition on driving behavior are justified by both theoretical considerations and empirical evidence (Cordellieri et al., 2016; van Huysduynen et al., 2017; Wei et al., 2024), particularly in challenging environments, such as work zone areas. The construction area layout, the driver gender, and the ambient light are considered in this study as three independent variables and are used to investigate their effect on three safety-related criteria: (1) lane-changing actions, including decision and merging process; (2) driving behaviors passing by the construction area, including longitudinal velocity and lateral offset, and relative longitudinal velocity and relative lateral offset; and (3) driving stability, including longitudinal dimension and lateral dimension.
This research is structured in six sections. Section 2 provides a comprehensive review of the literature related to the focus of this study. Section 3 outlines the methodology used in the study, particularly the apparatus and test scenarios, and the procedure of the experiment. Section 4 presents the analysis and findings, which are discussed in Section 5. Section 6 provides a summary of the study, draws conclusions, highlights the limitations, and suggests directions for future research.
2. Literature review
2.1. Work zone configuration
Work zone configuration is a significant physical factor to impact driver behavior and it encompasses a comprehensive design to ensure safety and efficiency for both workers and drivers. It begins with an advance warning area, where well-spaced signs such as "Road Work Ahead," "Lane Closed Ahead," and "Prepare to Stop" are placed to provide ample warning to drivers about the upcoming work zone (Baratian-Ghorghi & Zhou, 2017). Next, the transition area guides traffic smoothly into the new lane configuration with clear lane closure and merging signs, supported by cones or barrels and arrows indicating lane shifts (Raddaoui et al., 2020). This is followed by a buffer space, a critical safety zone devoid of any equipment or personnel, allowing for driver error and recovery without endangering the work area (Pesti & McCoy, 2001). The core of the setup is the construction site area, distinctly marked with barriers and delineators to separate construction activities from traffic, and equipped with appropriate lighting for nighttime operations to enhance visibility and safety. Finally, the termination area signals the end of the work zone, using "End Road Work" signs and clear pavement markings to guide drivers back to their normal traffic patterns, ensuring a smooth and safe transition. In line with these considerations, Duan et al. (2020) investigated the effect of the position of the lane-end sign (near, middle, and far) on lanechanging behaviors based on parametric survival models. Their findings indicated that the position of the lane-end sign significantly influenced the point at which drivers decided to change lanes, subsequently impacting the merging distance. However, this effect was reduced when the sign was situated at a considerable distance from the work zone. Similar to Duan et al. (2020), Hang et al. (2018) focused on the effect of the position of the lane-end sign and traffic volume on drivers' lanechanging behaviors in work zone areas. They conceptualized the merging process in three successive stages, including the "perception", "preparation", and "action" stages. According to the result, the positioning of the lane-end sign significantly impacts drivers' perception of the approaching work-zone situation, but this influence progressively diminishes through the subsequent "preparation" and "action" stages. This finding is consistent with the finding in the study conducted by Duan et al. (2020). In addition to the research that focused on advance warning area and transition area, the construction site area itself has a significant effect on driver behaviors. For example, the study of Banerjee et al. (2018) focused on how the movable work zone barriers influenced driving maneuvers on arterial roads. The findings of the study revealed a tendency among drivers to accelerate when driving adjacent to concrete Jersey barriers and drivers consistently deviated from the lane's center, moving away from the barrier, during their travel alongside these concrete structures. The impact of the length of the median opening in a work zone area was investigated by Difei et al. (2021). An intriguing pattern was observed in that vehicles' speeds exceeded the posted speed limits across all median opening lengths. In addition, a narrower median opening could potentially encourage hazardous driving behaviors, including abrupt deceleration and deceleration during lane changes. The taper length of the work zone also plays a significant effect on driver responses. The study conducted by Morgan et al. (2010) pointed out that a shortened taper length increases the risk for both drivers and work zone personnel, particularly when limited sight distance reduces drivers' ability to anticipate the lane shift. Even though a large number of studies were carried out to investigate driver behaviors in work zones, few researchers focused on the effect of the construction area layout on
driver behavior from the perspective of safety.
2.2. Driving behaviors at work zone
Since closed lanes are occupied by work zones, drivers are forced to drive their vehicles away from the closed lane into the open lanes, thereby passing by the work zone area. This lane-changing activity within work zone areas causes a significant rise in disputes between drivers, potentially increasing traffic congestion and the risk of collision. Hence, a large number of relevant studies have been conducted to investigate drivers' lane-changing behaviors in work zone areas with the objective of enhancing traffic safety in work zone areas. The first step is to explore the processes involved in lane-changing behaviors. A lane change is characterized as an action in which a vehicle transitions from one lane to another, with both lanes in the same direction of travel (Fei et al., 2016). The process of changing lanes begins with the driver adjusting their speed or position within the lane and concludes upon successfully merging into the desired lane (Chauhan et al., 2022). Prior studies have explored this process from various angles. Some studies emphasized the decision-making phase of the process. For example, in the studies conducted by Yang and Koutsopoulos (1996) and Pan et al. (2016), lane-changing behaviors were split into two categories, namely mandatory lane changing and discretionary lane changing. Lanechanging in work zone areas is a mandatory maneuver rather than an optional one. The lane-changing process involves intention, generation, planning, and completing the lane-change (Li et al., 2022). Some researchers focused on maneuvers used by drivers to change lanes. According to the study conducted by van Winsum et al. (1999), these maneuvers can be performed in three phases: Phase 1: rotating the steering wheel to its maximal angle to achieve lateral velocity; Phase 2: realigning the steering wheel to a neutral position; and Phase 3: adjusting the steering wheel to its extreme angle in the contrary direction to lower lateral velocity and ensure vehicle stability within the desired lane. Liu et al. (2021) proposed a dynamic trajectory planning method of lane-changing based on vehicle-to-vehicle communication for connected and automated vehicles and showed that their model was not only effective to make the necessary maneuvers but also could enhance safety. Wang et al. (2018) introduced a method based on "Reinforcement Learning" to execute lane-changing maneuvers that can be adapted to a wide array of driving scenarios, including those that are not predictable.
Apart from the lane-changing behavior, driver behaviors about how to pass by the work zone area have also been studied relative to longitudinal motion, lateral motion, and the gap to the head vehicle. Traffic capacity and safety are highly related to these behaviors. For example, Liu et al. (2016) demonstrated a significant correlation between driver behavior and the likelihood of injury in work zone crashes. Specifically, the data indicated that the probability of driver injury went up by 9.9 to 10.3 % above the baseline behavior (defined as the absence of improper actions by the driver) when the driver intentionally engages in improper actions or violations within work zone areas. In contrast, for crashes occurring outside of work zone areas, the increase in injury risk associated with such behaviors is comparatively lower, ranging from 1.7 to 5.7 % above the baseline. Racha et al. (2008) developed a non-linear hyperbolic model to demonstrate the relationship between the speed and the traffic capacity. Highway organizations can utilize the estimated capacity in conjunction with projected traffic demand to strategically plan work zone activities, thereby preventing extended durations of over-saturation.
2.3. Influencing factors on driver behaviors
The factors that affect driver behaviors when passing by the work zones can be divided into three groups: environmental elements, driver characteristics, and vehicle features (Meng & Weng, 2012; Silveira et al., 2023; Yang et al., 2015). The environmental elements that influence driver behaviors in work zone areas include a range of physical and visual factors present in or around work zones that can affect how drivers perceive, react to, and navigate these spaces. Weather and illumination conditions are significant components of environmental factors that impact driver behaviors in work zone areas. Hou and Chen (2020) introduced an integrated framework designed to assess traffic safety in work zone areas, which considered the prevailing weather and road surface conditions. The finding revealed that adverse weather can have a negative effect on crash risk in work zones. Similarly, Ahmed and Ghasemzadeh (2018) demonstrated that the probability of drivers reducing their speed by more than 5 km per hour below the speed limit was 23 % higher in conditions of light rain, and this probability increased to 29 % under heavy rain conditions. Driver characteristics, including gender, driving experience, and the driver's profession, can significantly impact how drivers pass by work zones. A study carried out by Kummetha et al. (2020) has uncovered significant evidence of differences in lateral control and gaze behavior between genders. Vehicle features can impact driver behaviors to an extent. Weng and Meng (2012) demonstrated that the tendency for risky driving behavior in work zones exhibited a difference dependent on the vehicle's age. Specifically, individuals operating newer vehicles, characterized by a service period of less than five years, are inclined towards adopting safer driving practices.
2.4. Research gap
Although a substantial body of research has investigated driver behavior in work zones, most studies have focused on specific elements such as advance warning signs, merging strategies, or environmental factors. However, there is limited research that systematically evaluates how the construction area layout itself, specifically the spatial arrangement and position of equipment, affects driver behaviors from a safety perspective. Moreover, while prior studies have typically examined the influence of individual factors in isolation, the combined effects of physical layout and contextual factors such as driver gender and ambient light conditions remain underexplored. Understanding these interaction effects is crucial, as driver responses to construction area layouts may vary significantly depending on such contextual conditions.
This study aims to address these gaps by analyzing the individual and combined effects of the construction area layout, the driver gender, and the ambient light condition on lane-changing behavior, driving behavior passing by the construction area, and driving stability using a controlled driving simulator experiment.
3. Methodology
This section presents the methodology of this study. It includes two parts, namely experiment and data analysis. The experiment was designed to collect participants' behaviors using a driving simulator and the collected data were analyzed to investigate how independent variables impact driver behaviors. Fig. 1 shows the flowchart of the study.
3.1. Apparatus and test scenarios
The driving simulator located in the Transportation Center of the University of Wyoming was used in this study. The driving simulator was developed by Realtime Technologies Inc., Royal Oak, MI. It comprises a ¼ size passenger vehicle cabin, an operation monitoring system, a digital video display system, and a control console. The passenger vehicle cabin has a steering wheel, an acceleration pedal, a brake pedal, an automatic transmission, and electronic accessories. Participants can interact with the scenario through this system. The operation monitoring system can monitor and record users' driving behaviors using cameras installed on the driving simulator. The digital video system projects the scenarios on three screens with a 1920 × 1080 resolution in front of the cabin, which enables users to have a front/peripheral field. The control console can
Fig. 1. Methodology of the study.
control the start and stop of the driving simulator.
Some parts of the driving simulator, including ¼ passenger cabin, the digital video display system, and the control console, are presented in Fig. 2. Internet Scene Assembler was used to build the scenario. Sim-Creator and SimVista were used to develop the vehicle model and to manage the driving simulator. The data update frequency was 60 Hz.
An experimental situation was crafted in this study based on an urban road section within a city context. A conventional bi-directional stretch of road, comprising four lanes and a speed limit of 30 miles per hour (mph), was emulated to observe driver actions. This particular road piece spanned approximately 1.5 miles, with each lane measuring 12 feet in width. A typical work zone was established on the rightmost lane near the terminal part of this road section. The unimpeded left lane was referred to as "the open lane", while the right lane housing the work zone was dubbed "the closed lane". Traffic signs, including speed limit,
road work ahead, and lane-end signs, were set based on the Manual on Traffic Control for Roadway Work Operations (Wyoming, US). In this study, the location of heavy equipment reflects work zone configuration from the perspective of the construction area layout. As shown in Fig. 3, Position 1 indicates that the heavy equipment is near the beginning of the construction area, while Position 2 places it at the center of the construction area.
3.2. Participant and experiment process
Some students at the university were invited to participate in the experiment. In total, 26 people, including 14 males and 12 females, completed the experiment. Their ages ranged from 19 to 65, with an average of 31.2 years. They all had valid driving licenses and at least one year of driving experience. The average duration of the driving
Fig. 2. Driving simulator.
Fig. 3. Layout of the test scenarios (not to scale).
experience was 6.7 years. The study received approval from the university's Institutional Review Board (IRB) prior to its commencement.
Before the experiment, a research team member provided participants with a comprehensive briefing on the experimental procedures. Participants then completed a short practice session consisting of fundamental driving tasks, such as accelerating, braking, turning, maintaining speed, and changing lanes. This exercise, lasting approximately five minutes, allowed participants to become acquainted with the driving simulator. Following the practice, participants were instructed to drive the simulator as they would under normal driving conditions.
The first experimental phase featured three daytime work zone scenarios: (1) no heavy equipment (No HE), (2) heavy equipment positioned near the start of the construction area (Position 1), and (3) heavy equipment positioned at the center of the construction area (Position 2). The second phase replicated the same three work zone scenarios under nighttime conditions. Lighting adjustments simulated nighttime driving, including reduced visibility and headlight use. This setup ensured that participants experienced identical scenarios to the daytime experiment, with the primary distinction being the altered ambient lighting. Participants were required to pass by the work zone safely. All their driving behaviors were monitored and recorded by the driving simulator.
3.3. Parametric survival model, MANOVA, and linear regression model
A parametric survival model was used in this study to measure the effect of the independent variables on lane-merging behavior based on decision distance and merging distance. Let T denote the temporal parameter within the context of the survival model. In this framework, the cumulative function F(t), frequently referred to as the failure function, characterizes the likelihood that a given individual experiences failure prior to a predetermined time t, with an associated probability density function represented by f(t). In contrast, the function known as the survivor function, denoted by S(t), encapsulates the probability that the individual will endure beyond time t. The mathematical relationship between these functions can be articulated as in the Equation (1).
$$F(t) = Pr(T < t) = 1 - S(t)$$ (1)
The parametric survival model is traditionally used with time as the dependent variable, but it can also be effectively applied to other factors, such as distance. This versatility arises from the model's fundamental capability to handle censored data and varying risk over different conditions.
Driver behaviors can be affected by various factors. In this study, several independent variables were considered, which can be separated into environmental variables and human variables. Environmental variables originate from the environment around the area, including the work zone layout and the ambient light. Human variables stem from drivers' inner characteristics, such as gender. In this study, changes in the position of heavy equipment reflect different construction area layouts. Ambient light condition includes the light in daytime and nighttime. The influencing factors are explained in detail in Table 1.
MANOVA is an extension of analysis of variance (ANOVA) that evaluates multiple dependent variables simultaneously (Boyce & Geller, 2002). While ANOVA assesses differences in means between groups, MANOVA examines the differences in vectors of means. The primary objective of using MANOVA is to investigate whether the response variables are influenced by the manipulation of independent variables (Worsfold et al., 2007). In this study, MANOVA was selected due to its capability to simultaneously evaluate the impact of multiple independent variables and their interactions on several correlated dependent variables. Specifically, this statistical method enables a comprehensive examination of complex research questions by identifying the overall main effects of independent variables, analyzing interactions among these variables, and assessing their collective influence on multiple dependent measures. MANOVA statistical method was employed to investigate how different construction area layouts, along with their interactions with driver gender and ambient lighting conditions, collectively influence driver behaviors when navigating through road work zones. Compared to alternative methods that consider dependent variables individually, MANOVA offers enhanced statistical rigor by controlling for the inflated Type I error rate and accounting for the interdependencies among independent variables (Warne, 2014), thus
Table 1 The explanation of influencing factors.
| Independent variables |
Description of variables | Notation |
|---|---|---|
| Position of heavy equipment in construction area | ||
| No HE | There is no heavy equipment in the work | Location (1) = |
| zone. | 0Location | |
| (2) = 0 | ||
| Position 1 | The heavy equipment is located close to | Location (1) = |
| the start of the work zone. | 1Location | |
| (2) = 0 | ||
| Position 2 | The heavy equipment is located close to | Location (1) = |
| the middle of the work zone. | 0Location | |
| (2) = 1 | ||
| Driver gender | ||
| Female | Gender = 0 | |
| Male | Gender = 1 | |
| Ambient light | ||
| Daytime | Light = 0 | |
| Nighttime | Light = 1 |
providing more reliable and detailed insights into driver behavioral patterns.
To examine the relationship between independent variables and driving behaviors, linear regression analysis was employed in this study. Linear regression enables the estimation of the unique contribution of each independent variable to variations in the dependent variable while controlling for the effects of the other elements (Roustaei, 2024). In addition, this approach provides interpretable coefficients that reflect the direction and magnitude of influence, allowing for a detailed assessment of how specific factors contribute to changes in the dependent variable. For example, Razmara et al. (2018) used linear regression to predict taxi drivers' safe-driving behaviors based on constructs from the theory of planned behavior and habit, demonstrating the method's effectiveness in understanding behavioral influences. Similarly, Katayama et al. (2024) utilized a multiple linear regression model to explore the relationship between driving ability and physical fitness factors among older adult drivers, highlighting the applicability of linear regression in analyzing complex interactions in driving studies. These examples underscore the suitability of linear regression for modeling the influence of multiple independent variables on driving behaviors, facilitating a comprehensive understanding of the factors affecting driving performance.
3.4. Metrics for assessing driver behaviors in the work zone
Lane-merging actions consist of decision-making and lane-changing process. Decision-making was measured by the decision distance and the lane-changing process by measuring the merging distance, as seen in Fig. 4. In this figure, decision distance is the longitudinal distance from the start point to the decision point associated with lane change. According to the American Association of State Highway and Transportation Officials (AASHTO), drivers should be able to see the lane-end sign at least 460 feet upstream when the speed limit is 30 mph. Merging distance is defined as the longitudinal distance traveled by the vehicle from the moment its lateral velocity becomes greater than zero, indicating the beginning of a lane change, until the point at which the lateral velocity returns to zero, signifying the completion of the lane change (Duan et al., 2020).
In this study, the longitudinal velocity and lateral offset are used to measure driver performance passing by the construction area. Fig. 5 shows the longitudinal velocity and the lateral offset.
The lateral stability indicator represents a critical metric for assessing a vehicle's lateral stability during motion. This indicator quantifies the extent of lateral deviation a vehicle experiences, utilizing the lateral displacement data extracted from trajectory analysis. In this study, the lateral stability indicator involves analyzing the variations in lateral offset when passing the work zone area. To evaluate lateral driving
stability, this study adopted the coefficient of variation of lateral offset. The lateral stability indicator was calculated using Equations (2) and (3) based on a systematic approach to quantifying lateral stability in vehicular dynamics.
$$D_x(t) = X(t) - X_0 \tag{2}$$
$$R_1(t) = \frac{std(D_x(0), \cdots, D_x(end))}{mean(D_x(0), \cdots, D_x(end))}$$ (3)
where R1(t) represents the lateral stability indicator, *D*x(t) represents the lateral offset of a vehicle at instant t, X(t) denotes the lateral position of the vehicle at instant t, and *X*0 is the central position of the open lane. If *D*x(t) is greater than 0, it means that the vehicle is driving on the left side of the center line, but when *D*x(t) is less than 0, the vehicle is leaning to the right side. *D*x(0) and *D*x(end) represent the lateral offset when the vehicle enters and leaves the work zone area, respectively. The standard deviation is denoted as std, while the mean is denoted as mean. The ratio of std to mean is referred to as the coefficient of variation, which serves as a measure of the dispersion of the lateral offset. An increased absolute coefficient of variation signifies a higher dispersion in lateral offset, indicating a tendency for serpentine driving, which suggests a decrease in driving stability in the lateral dimension.
The longitudinal stability indicator represents a critical metric for assessing a vehicle's longitudinal stability during motion. Fluctuations in longitudinal velocity can induce trajectory perturbations, with acceleration being employed as a parameter to quantify these variations. Longitudinal stability is predominantly represented by the disparity between instantaneous acceleration and the mean acceleration of a vehicle over duration T. Longitudinal stability is calculated by using Equation (4).
$$R2(t) = \sqrt{\frac{\sum\limits{i=t}^{T} |a_i - \overline{a}|}{T}}$$ (4)
where, R2(t) represents the longitudinal stability indicator at time t, ai denotes the acceleration at instant t, and a represents the average acceleration in the period T, which is the total time taken to pass by the work zone area. As *R*2 increases due to increased variability in longitudinal velocity, it leads to diminished longitudinal stability.
4. Result and analysis
The drivers' responses to the independent variables (i.e., the construction area layout, the driver gender, and the ambient light condition) when driving through the work zone were studied in this section. The effect focused on three aspects, namely lane-changing, driving
Fig. 4. Decision distance and merging distance.
Fig. 5. Longitudinal velocity and lateral offset.
behaviors passing by the construction area, and driving stability.
4.1. Effect of independent variables on lane-changing
In this section, the effect of the independent variables (i.e., the construction area layout, the driver gender, and the ambient light condition) on lane-changing actions was investigated. The descriptive results of drivers' decision distance and merging distance are shown in Table 2.
Figs. 6 and 7 show the results of the parametric survival model for decision distance and merging distance, respectively. In both figures, the Y-axis represents lane keeping probability, which indicates the likelihood of completing the decision about the merging process. A value of one signifies that no participants completed the process, while a value of zero signifies that all participants completed it. According to Fig. 6, the construction area layout does not have significant influence on decision distance. Male drivers tend to have longer decision distance and nighttime seems to require a longer decision distance too. Fig. 7 demonstrates that the three independent variables (i.e., the construction area layout, the driver gender, and the ambient light condition) do not have any effect on merging distance.
The MANOVA results for decision and merging distances are presented in Table 3. According to this table, a driver gender shows a significant effect on decision distance (Sig. = 0.095, slightly less than 0.10), but no effect on merging distance (Sig. = 0.293). However, the other two variables do not have a statistically significant effect on decision or
Table 2 The results about decision distance and merging distance for each independent variable.
| Independent | Number of | Decision distance | Merging distance | ||
|---|---|---|---|---|---|
| variables | samples | Mean | SD | Mean | SD |
| Position of heavy equipment |
|||||
| No HE | 52 | 235.797 | 87.737 | 115.038 | 40.613 |
| Position 1 | 52 | 237.953 | 92.475 | 112.628 | 33.537 |
| Position 2 | 52 | 239.664 | 100.111 | 126.632 | 51.345 |
| Driver gender | |||||
| Male | 84 | 249.586 | 97.799 | 114.754 | 40.659 |
| Female | 72 | 224.060 | 85.672 | 122.002 | 44.803 |
| Ambient light | |||||
| Daytime | 78 | 249.917 | 93.605 | 119.986 | 41.861 |
| Nighttime | 78 | 225.693 | 91.360 | 116.213 | 43.586 |
merging distances. In addition, interactions between these variables generally show no significant effects on decision or merging distances either, with significance levels well above the 0.10 threshold, indicating that these interactions do not substantially affect drivers' decision and merging distances.
The results of the survival regression model for decision and merging distances are presented in Table 4. As shown in Table 4(a), the only variable that has a statistically significant effect on decision distance is the driver gender. It is clear that male drivers' decision distance is 12.1 % longer than the decision distance of female drivers (p-value = 0.042 < 0.10). According to Table 4(b), the only variable that has a statistically significant effect on merging distance is the position of the heavy equipment in the construction area. It is clear that the merging distance is 14.6 % longer when the heavy equipment is located closer to the center of the work zone (p-value = 0.040 < 0.10).
4.2. Effect of independent variables on passing by the work zone area
The drivers' responses to the construction area layout, the driver gender, and the ambient light condition when passing by the work zone area were investigated in this section. Driving behavior passing by the work zone area is assessed by measuring (1) the longitudinal velocity and lateral offset, and (2) the relative velocity and the lateral offset before and when passing by the construction area. The relative longitudinal velocity and relative lateral offset are the differences of longitudinal velocity and lateral offset between passing by the construction area and the area of 300 feet of open lane area behind the speed limit sign. Table 5 shows the descriptive results of driver behaviors.
4.2.1. Effect of independent variables on longitudinal velocity and lateral offset
Table 6 shows the MANOVA results about driver behaviors passing by the construction area, which indicate that none of the independent variables and none of their interactions affect either velocity or lateral offset when passing by the construction area, as all p-values exceed the threshold of 0.10 for statistical significance. In this study, the abbreviation "Sig." stands specifically for "significance," and in practical terms, it directly refers to the p-value of the statistical test. "Sig." in statistical tables represents the exact probability (p-value) of obtaining the observed results.
The results of the linear regression for driver behaviors passing by the construction area are presented in Table 7. This table suggests that
Fig. 6. The results of the survival model about decision distance.
Fig. 7. The results of the survival model about merging distance.
Table 3 MANOVA results for decision and merging distances.
| Independent variables | df | Decision distance F ratio |
Sig. | Merging distance F ratio |
Sig. |
|---|---|---|---|---|---|
| Position of heavy equipment in the construction area |
2 | 0.019 | 0.981 | 1.645 | 0.197 |
| Driver gender | 1 | 2.833 | 0.095 | 1.115 | 0.293 |
| Ambient light | 1 | 2.452 | 0.120 | 0.435 | 0.511 |
| Position × Gender | 2 | 0.022 | 0.978 | 0.514 | 0.599 |
| Position × Light | 2 | 0.097 | 0.907 | 0.018 | 0.982 |
| Gender × Light | 1 | 0.169 | 0.682 | 2.043 | 0.155 |
| Position × Gender × Light | 2 | 0.245 | 0.783 | 1.284 | 0.280 |
how the construction area layout, the driver gender, and the ambient light condition affect the longitudinal velocity and lateral offset. According to Table 7(a), drivers pass by the work zone area with a speed of 33 mph (t-statistic = 30.860, p-value < 0.10), which is higher than the speed limit of this area. Table 7(b) presents a significant negative intercept of lateral offset (t-statistic = -9.071, p-value < 0.10), which reveals that drivers tend to drive left of the center line.
4.2.2. Effect of independent variables on relative longitudinal velocity and relative lateral offset
The MANOVA results that reflect the effect of the independent variables on the relative longitudinal velocity and the relative lateral offset are presented in Table 8. Relative longitudinal velocity represents the difference between the longitudinal velocity before and while passing by
Table 4 The results of survival regression for decision and merging distances.
| (a) Decision distance Independent variables |
Estimate | Standard error | Z-Statistic | p-value | Exp(β) | 90 %CI | |
|---|---|---|---|---|---|---|---|
| Lower | Upper | ||||||
| Intercept | 5.456 | 0.065 | 83.66 | <0.001 | 5.349 | 5.583 | |
| Position of heavy equipment in construction area | |||||||
| Position 1 | 0.021 | 0.069 | 0.030 | 0.765 | 1.021 | − 0.092 | 0.133 |
| Position 2 | 0.044 | 0.069 | 0.64 | 0.523 | 1.045 | − 0.069 | 0.157 |
| Driver gender | |||||||
| Male | 0.114 | 0.056 | 2.03 | 0.042 | 1.121 | 0.022 | 0.206 |
| Ambient light | |||||||
| Nighttime | 0.083 | 0.056 | 1.48 | 0.140 | 1.087 | − 0.010 | 0.206 |
| (b) Merging distance | |||||||
| Independent variables | Estimate | Standard error | Z-Statistic | p-value | Exp(β) | 90%CI | |
| Lower | Upper | ||||||
| Intercept | 4.881 | 0.066 | 73.56 | <0.001 | 4.772 | 4.990 | |
| Position of heavy equipment in construction area | |||||||
| Position 1 | − 0.055 | 0.066 | − 0.82 | 0.412 | 0.947 | − 0.164 | 0.055 |
| Position 2 | 0.136 | 0.066 | 2.04 | 0.040 | 1.146 | 0.019 | 0.238 |
| Driver gender | |||||||
| Male | − 0.062 | 0.055 | − 1.12 | 0.262 | 0.940 | − 0.152 | 0.029 |
| Ambient light | |||||||
| Nighttime | 0.014 | 0.055 | 0.26 | 0.791 | 1.014 | − 0.075 | 0.104 |
Table 5 The results about velocity and lateral offset for each independent variable.
| Independent variables | Number of samples | (mph) | Longitudinal velocity | Lateral offset (feet) | Relative longitudinal velocity (mph) | Relative lateral offset (feet) | |||
|---|---|---|---|---|---|---|---|---|---|
| Mean | SD | Mean | SD | Mean | SD | Mean | SD | ||
| Position of heavy equipment in construction area | |||||||||
| No HE | 52 | 33.742 | 6.407 | − 1.379 | 0.908 | − 3.396 | 8.695 | − 1.176 | 0.802 |
| Position 1 | 52 | 33.458 | 5.524 | − 1.470 | 0.840 | 0.116 | 5.340 | − 1.216 | 0.777 |
| Position 2 | 52 | 34.121 | 5.916 | − 1.433 | 0.892 | 0.409 | 4.827 | − 1.172 | 0.784 |
| Driver gender | |||||||||
| Male | 84 | 34.005 | 5.685 | − 1.386 | 0.905 | − 1.086 | 5.516 | − 1.223 | 0.777 |
| Female | 72 | 33.503 | 6.228 | − 1.476 | 0.844 | − 1.105 | 7.990 | − 1.226 | 0.723 |
| Ambient light | |||||||||
| Daytime | 78 | 33.336 | 6.088 | − 1.462 | 0.892 | − 0.08 | − 0.083 | 5.433 | − 1.266 |
| Nighttime | 78 | 34.211 | 5.769 | − 1.393 | 0.864 | − 2.11 | − 2.106 | 7.750 | − 1.183 |
Table 6 MANOVA results for driver behaviors passing by the construction area.
| Independent variables | df | Longitudinal velocity F-ratio |
Sig. | Lateral offset F-ratio |
Sig. |
|---|---|---|---|---|---|
| Position of heavy equipment in construction area |
2 | 0.177 | 0.838 | 0.115 | 0.892 |
| Driver gender | 1 | 0.265 | 0.608 | 0.394 | 0.531 |
| Ambient light | 1 | 0.802 | 0.372 | 0.248 | 0.619 |
| Position × Gender | 2 | 0.291 | 0.748 | 0.188 | 0.829 |
| Position × Light | 2 | 0.410 | 0.664 | 1.404 | 0.249 |
| Gender × Light | 1 | 0.000 | 0.985 | 0.037 | 0.848 |
| Position × Gender × Light |
2 | 0.442 | 0.644 | 0.235 | 0.791 |
the construction area. Relative lateral offset is the difference between lateral offset before and while passing by the construction area. According to Table 8, the position of heavy equipment can significantly impact the relative longitudinal velocity as the p-value associated with this F-ratio is 0.003, which is less than 0.10. Ambient light condition also demonstrates a similar result. However, other independent variables do not have a statistically significant influence on relative longitudinal velocity or on relative lateral offset as their p-values are all greater than 0.10. In addition, the interactions also present a similar result.
The linear regression results about relative longitudinal velocity are presented in Table 9(a). According to Table 9(a), the effect of ambient light on relative longitudinal velocity is significant, with a coefficient of 3.093 (p-value = 0.004 < 0.10), which means that drivers tend to have bigger relative longitudinal velocity at nighttime than daytime. Drivers may perceive nighttime conditions as less hazardous, leading to risk compensation and more aggressive driving. This finding initially appears counterintuitive, given the common expectation that reduced visibility increases perceived driving risks. However, this unexpected result may stem from the specific experimental scenario employed in our study, which excluded critical real-world factors such as traffic volume, moving construction machinery, and active workers. Additionally, inherent differences between simulated and actual driving experiences may have further influenced driver perceptions, potentially limiting the generalizability of this finding beyond the conditions tested. A study by Garber and Zhao (2002) supports these observations, indicating higher vehicle speeds at night due to perceived lower enforcement. In addition, position 2 enables drivers to pass the work zone with a higher speed as the corresponding to p-value is 0.058, less than 0.10.
The linear regression results about relative lateral offset are presented in Table 9(b). According to Table 9(b), drivers prefer to drive on the left of the center line as the intercept is negative and the p-value corresponding to the intercept is 0.001, less than 0.10. The other variables do not have statistically significant effects on the relative lateral offset.
4.3. Effect of independent variables on driving stability
Lateral stability indicator and longitudinal stability indicator are two
Table 7 Linear regression results about driver behaviors passing by the construction area.
| (a) Linear regression results about longitudinal velocity | ||||||
|---|---|---|---|---|---|---|
| Independent variables | Estimate | Standard error | t-Statistic | p-value | 90 %CI Lower |
Upper |
| Intercept | 33.658 | 1.091 | 30.860 | 0.000 | 31.853 | 35.464 |
| Position of heavy equipment in construction area | ||||||
| Position 1 | − 0.563 | 1.175 | − 0.479 | 0.633 | − 2.508 | 1.383 |
| Position 2 | 0.086 | 1.175 | 0.073 | 0.942 | − 1.859 | 2.031 |
| Driver gender | ||||||
| Male | 0.502 | 0.963 | 0.521 | 0.603 | − 1.091 | 2.095 |
| Ambient light | ||||||
| Nighttime | 0.007 | 0.960 | 0.008 | 0.994 | − 1.581 | 1.596 |
| (b) Linear regression results about lateral offset | ||||||
| Independent variables | Estimate | Standard error | t-Statistic | p-value | 90%CI | |
| Lower | Upper | |||||
| Intercept | − 1.448 | 0.160 | − 9.071 | 0.000 | − 1.713 | − 1.184 |
| Position of heavy equipment in construction area | ||||||
| Position 1 | 0.091 | 0.172 | 0.526 | 0.599 | − 0.194 | 0.375 |
| Position 2 | − 0.201 | 0.172 | − 1.169 | 0.244 | − 0.486 | 0.084 |
| Driver gender | ||||||
| Male | 0.090 | 0.141 | 0.640 | 0.523 | − 0.143 | 0.323 |
| Ambient light | ||||||
| Nighttime | 0.019 | 0.140 | 0.133 | 0.894 | − 0.214 | 0.251 |
Table 8 MANOVA results about the relative longitudinal velocity and about relative lateral offset.
| Independent variables | df | Relative longitudinal velocity |
Relative lateral offset |
||
|---|---|---|---|---|---|
| F ratio |
Sig. | F ratio |
Sig. | ||
| Position of heavy equipment in construction area |
2 | 5.936 | 0.003 | 0.049 | 0.952 |
| Driver gender | 1 | 0.000 | 0.985 | 0.001 | 0.979 |
| Ambient light | 1 | 3.882 | 0.051 | 0.445 | 0.506 |
| Position × Gender | 2 | 0.086 | 0.917 | 0.152 | 0.859 |
| Position × Light | 2 | 1.515 | 0.223 | 1.855 | 0.160 |
| Gender × Light | 1 | 0.333 | 0.565 | 0.011 | 0.915 |
| Position × Gender × Light | 2 | 1.155 | 0.318 | 0.156 | 0.856 |
Table 10 The results of driving stability expressed by R1 and R2 relative to the independent variables.
| Independent
variables | Number of
samples | Lateral stability
(R1) | | Longitudinal
stability (R2) | | |
|--------------------------|--------------------------------------------------|---------------------------|-------|--------------------------------|-------|--|
| | | Mean
SD | | Mean | SD | |
| | Position of heavy equipment in construction area | | | | | |
| No HE | 52 | − 0.023 | 0.472 | 1.082 | 0.870 | |
| Position 1 | 52 | − 0.081 | 0.235 | 1.175 | 1.006 | |
| Position 2 | 52 | − 0.029 | 0.147 | 0.982 | 0.790 | |
| Driver gender | | | | | | |
| Male | 84 | − 0.050 | 0.175 | 0.920 | 0.698 | |
| Female | 72 | − 0.038 | 0.426 | 1.266 | 1.047 | |
| Ambient light | | | | | | |
| Daytime | 78 | − 0.057 | 0.431 | 1.142 | 0.923 | |
| Nighttime | 78 | − 0.033 | 0.119 | 1.019 | 0.859 | |
Table 9 Linear regression results about relative longitudinal velocity and relative lateral offset.
| Independent variables | (a) Linear regression results about relative longitudinal velocity Estimate |
Standard error | t-Statistic | p-value | 90 %CI Lower |
Upper |
|---|---|---|---|---|---|---|
| Intercept | − 4.116 | 1.195 | − 3.444 | 0.001 | − 6.094 | − 2.138 |
| Position of heavy equipment in construction area | ||||||
| Position 1 | 1.937 | 1.288 | 1.505 | 0.135 | − 0.194 | 4.069 |
| Position 2 | 2.456 | 1.288 | 1.907 | 0.058 | − 0.324 | 4.587 |
| Driver gender | ||||||
| Male | 0.019 | 1.055 | 0.018 | 0.986 | − 1.726 | 1.764 |
| Ambient light | ||||||
| Nighttime | 3.093 | 1.051 | 2.942 | 0.004 | 1.353 | 4.833 |
| (b) Linear regression results about relative lateral offset | ||||||
| Independent variables | Estimate | Standard error | t-Statistic | p-value | 90%CI Lower |
Upper |
| Intercept | − 1.211 | 0.137 | − 8.853 | 0.000 | − 1.437 | − 0.984 |
| Position of heavy equipment in construction area | ||||||
| Position 1 | 0.053 | 0.147 | 0.362 | 0.718 | − 0.191 | 0.297 |
| Position 2 | − 0.198 | 0.147 | − 1.343 | 0.181 | − 0.442 | 0.046 |
| Driver gender | ||||||
| Male | 0.003 | 0.121 | 0.027 | 0.978 | − 0.196 | 0.203 |
| Ambient light | ||||||
| Nighttime | 0.065 | 0.120 | 0.537 | 0.592 | − 0.135 | 0.046 |
common metrics to measure driving stability passing by the construction area. Table 10 shows the descriptive results about driving stability relative to each dependent variable.
The results of MANOVA examining the influence of the independent variables (i.e., construction area layout, driver gender, and ambient light) and their interactions on driving stability are presented in Table 11. According to this table, none of the independent variables has a statistically significant effect on lateral or longitudinal stability as indicated by the high p-values (Sig. > 0.10). Notably, the only exception is the effect of the driver gender on longitudinal stability *R*2, where a significant impact is observed (F-ratio = 5.907, Sig. = 0.016), suggesting that the driver gender can significantly influence longitudinal stability. The interactions, including the three-way interaction, have no statistically significant effects on driving stability. This analysis underscores the nuanced role that only the driver gender affects driving stability.
A linear regression model was adopted to measure the exact effect of the independent variables on driving stability, and Table 12 shows the regression results. The only statistically significant effect is observed in Table 12 (b), according to which a driver gender has a significant negative effect (p-value = 0.015 < 0.10) on longitudinal stability, suggesting that male drivers tend to pass by the work zone area with a more stable longitudinal velocity. This result agrees with to the finding presented in Table 11. Male drivers' consistent speeds may be due to higher confidence and calculated risk-taking (Møller & Gregersen, 2008; Yagil, 1998). This stability can enhance safety by reducing abrupt speed changes, which are common causes of accidents (Harb et al., 2008). However, it also necessitates proper regulation to prevent excessive speeding. Understanding these gender differences is crucial for designing targeted interventions to improve work zone safety and efficiency (Ozkan ¨ & Lajunen, 2006).
5. Discussion
In this section, the findings of the study are discussed in terms of the impacts of three independent variables (i.e., the construction area layout, the driver gender, and the ambient light condition) on driving behaviors (i.e., lane-changing, driving behavior passing by the construction area, and driving stability).
5.1. Effects of the independent variables
The construction area layout has a significant effect on drivers' lanechanging behaviors in urban environments. More specifically, drivers tend to use a longer distance to complete the merging process when the heavy equipment is located closer to the center of the construction area. The finding indicates the heavy equipment located closer to the center of the work zone is likely to enhance drivers' perception of risk and to improve the driver's assessing of the environmental complexity. Longer merging distances represent more cautious merging behaviors. Drivers may not be able to predict the size of the maneuvering space when heavy equipment are positioned far away from the start of the construction
Table 11 MANOVA result for lateral stability and longitudinal stability.
| Independent variables | df | Lateral stability (R1) |
Longitudinal stability (R2) |
||
|---|---|---|---|---|---|
| F ratio |
Sig. | F ratio |
Sig. | ||
| Position of heavy equipment in construction area |
2 | 0.703 | 0.497 | 0.706 | 0.495 |
| Driver gender | 1 | 0.057 | 0.812 | 5.907 | 0.016 |
| Ambient light | 1 | 0.166 | 0.684 | 0.602 | 0.439 |
| Position × Gender | 2 | 2.261 | 0.108 | 0.530 | 0.590 |
| Position × Light | 2 | 0.923 | 0.400 | 0.802 | 0.450 |
| Gender × Light | 1 | 0.786 | 0.377 | 1.336 | 0.250 |
| Position × Gender × Light | 2 | 2.249 | 0.109 | 0.427 | 0.653 |
area, hence using a longer distance to safely assess and execute the merging actions. This cautious approach aligns with theories of risk compensation, whereby drivers adjust their behavior in response to the perceived level of risk in their environment.
In addition to the construction area, the driver gender is also a critical factor that affects lane-changing behaviors. Male drivers' decision distances are 12.1 % longer than female drivers' decision distances. Research has consistently shown that male drivers often exhibit higher risk-taking behaviors compared to female drivers (Duan et al., 2020; Horswill & Coster, 2002; Turner & McClure, 2003). The increased decision distance for male drivers could be indicative of a more aggressive approach to lane changing, where they may initiate the maneuver from a farther distance. In contrast, female drivers might adopt a more cautious approach, waiting until they are closer to the lane change point to make their decision, reflecting a higher sensitivity to potential hazards. However, this finding is contradicted by the findings of Duan et al. (2020) as their study revealed that female drivers tended to change lanes late compared with male drivers. In their study, traffic volumes and the gap were considered in the scenarios, which may be the factors for the contradictions.
Concerning driving behaviors through the construction area, drivers exhibit a tendency to drive left of the center line. This finding may be due to the work zone in the experiment conducted with the participants was located in the right lane because drivers may have perceived the space on the left as safer. The construction area layout has a significant effect on the relative longitudinal velocity, i.e., the longitudinal velocity before and during the passing by the construction area. The construction area layout may significantly affect drivers' perception of the environment, leading to adjustments in driving behavior, particularly longitudinal velocity. Therefore, drivers may reduce and/or increase the speed of their vehicle as necessary to navigate the area safely. In addition, drivers tend to pass by the construction area with a lower velocity at nighttime compared with daytime, which was supported by the study conducted by Bassani et al. (2016). They found that as illuminance levels rise, there is a corresponding increase in average longitudinal velocity.
Driving stability can also be significantly affected by the independent variables. For example, compared with female drivers, male drivers tend to pass by the work zone area with a higher longitudinal stability. This finding is supported by previous studies, which reported that (1) when distracted, female drivers are often associated with more erratic driving behavior compared to their male counterparts (Arca et al., 2022); and (2) females exhibited lower heart rate variability compared to males across all conditions, indicating heightened stress levels (Miller and Boyle (2015). Thus, to enhance female drivers' driving stability, it is essential to provide them with specific information such as real-time driving stability alerts, detailed updates on road conditions, and tips on managing stress and distractions while passing by work zone areas to ensure they can drive safely.
5.2. Implications of the study
A parametric survival model was used in this study to analyze the effects of environmental factors (such as construction area layout and ambient light) and human factors (such as driver gender) on various aspects of the lane-changing process, including decision distance and merging distance. The use of this model is supported by Duan et al. (2020) and Washington et al. (2020), who demonstrated its superior capability in characterizing lane-changing dynamics in work zone areas. This superiority stems from the model's ability to account for the variation in the probability of successful merging maneuvers over time or distance, making it particularly effective for handling duration data and merging distances.
The implications of adopting the parametric survival model in this study are significant. By quantifying the exact effects of decision distance and merging distance through residual analysis, this research provides a deeper understanding of how environmental and human
Table 12 Linear regression results about the driving stability.
| Independent variable | (a) Linear regression results for lateral stability Estimate |
Standard error | t-Statistic | p-value | 90 %CI Lower |
Upper |
|---|---|---|---|---|---|---|
| Intercept | − 0.066 | 0.058 | − 1.141 | 0.256 | − 0.161 | 0.030 |
| Position of heavy equipment in construction area | ||||||
| Position 1 | 0.089 | 0.062 | 1.434 | 0.154 | − 0.0014 | 0.192 |
| Position 2 | 0.040 | 0.062 | 0.640 | 0.523 | − 0.063 | 0.143 |
| Driver gender | ||||||
| Male | − 0.012 | 0.051 | − 0.236 | 0.813 | − 0.096 | 0.072 |
| Ambient light | ||||||
| Nighttime | − 0.031 | 0.051 | − 0.603 | 0.547 | − 0.114 | 0.053 |
| (b) Linear regression result for longitudinal stability | ||||||
| Independent variable | Estimate | SE | t-Statistic | p-value | 90%CI | |
| Lower | Upper | |||||
| Intercept | 1.401 | 0.160 | 8.768 | 0.000 | 1.136 | 1.665 |
| Position of heavy equipment in construction area | ||||||
| Position 1 | − 0.265 | 0.172 | − 1.538 | 0.126 | − 0.550 | 0.020 |
| Position 2 | − 0.196 | 0.172 | − 1.136 | 0.258 | − 0.480 | 0.089 |
| Driver gender | ||||||
| Male | − 0.346 | 0.141 | − 2.454 | 0.015 | − 0.579 | − 0.113 |
| Ambient light | ||||||
| Nighttime | 0.038 | 0.141 | 0.272 | 0.786 | − 0.194 | 0.089 |
factors influence lane-changing behavior in work zones. This understanding can inform the design and implementation of safer work zone configurations and targeted driver interventions. Additionally, the study contributes to the existing literature by demonstrating the practical applications of survival analysis in traffic safety research, offering a robust methodological approach for future studies in this field.
This study's findings can be of value to traffic management officials, safety professionals, and driver education, especially in the context of urban work zones. For example, the observation that drivers modify their lane-changing behavior in response to the position of heavy equipment in the construction area presents a valuable opportunity for enhancing traffic flow and safety through the strategic work zone configuration. By encouraging earlier lane changes, the risk associated with last-minute maneuvers can be significantly reduced.
Recognizing the gender-specific differences in lane-changing behaviors, targeted driver awareness campaigns can be developed. For male drivers, these campaigns could emphasize the benefits of cautious lane changing and could highlight the dangers of aggressive driving behaviors. Conversely, for female drivers, the focus could be on confidence-building strategies to improve driving stability and alleviate stress in work zone areas. This approach, combining the work zone layout design with gender-specific educational campaigns, offers a comprehensive strategy for improving safety and efficiency in work zone traffic management.
In addition, the findings of this study can be used as policy recommendations, such as gender-specific safety measures, speed limit enforcement, and illumination improvements. Concerning genderspecific safety measures, additional signage or alerts could be tailored to address the specific needs and behaviors of male and female drivers, potentially improving overall safety in work zones. The findings of the study support the enforcement of speed limits in work zone areas, especially considering the impact of the work zone layout and the time of day on driving velocity. Policies that focus on strict speed limit enforcement during nighttime and in areas with heavy equipment could reduce accidents. Since driving velocity and stability are affected by illuminance levels, improving lighting in work zones during nighttime could enhance safety by allowing drivers to navigate the area more confidently and at safer speeds. Hence, achieving a balance between traffic efficiency and safety is critical. Implementation of enhanced lighting systems within work zones to improve driver visibility and overall safety is necessary. Specifically, targeted illumination at critical zones, such as areas adjacent to construction equipment, can aid drivers in perceiving and responding to work zone more effectively. Additionally, adaptive lighting technologies that adjust brightness based on ambient conditions may offer a practical solution to balance safety and energy efficiency. These interventions can help maintain safe driving speeds and improve vehicle control, ultimately enhancing work zone safety. The advent of technologies within the realm of intelligent transportation systems, notably vehicle-to-vehicle and vehicle-toinfrastructure communications, holds significant potential for enhancing traffic management in areas affected by work zones.
6. Conclusion
This study explored the effects of construction area in the urban traffic network on lane-changing, driving behavior passing by the construction area, and driving stability. The driver gender, and the ambient light condition were also considered in this study. For this research, an experiment was designed to collect information about driver behaviors when passing by the work zone. A parametric survival model was used to assess how drivers respond to the independent variables. In addition, MANOVA and linear regression were integrated to measure which independent variables have statistically significant effects on driver behaviors when passing by the work zone. The statistically significant findings include the following aspects:
- (1) Lane-changing behaviors: Merging distance is 14.6 % longer when the heavy equipment is located closer to the center of the work zone. Male drivers' decision distance is 12.1 % longer than female drivers' decision distance.
- (2) Driving behaviors passing by the construction area: Drivers tend to drive left of the center line when passing by the construction area. Position 2 enables drivers to pass by the construction area with a higher speed. Drivers tend to have bigger relative longitudinal velocity at nighttime than daytime.
- (3) Driving stability: Male drivers tend to pass by the work zone area with a more stable longitudinal velocity than female drivers.
The findings of the study have implications for traffic management, road safety, and urban planning. The observed differences in lanechanging behaviors between male and female drivers, the impact of the work zone configuration from the perspective of construction area layout on merging distance, and the variations in driving behaviors and driving stability passing by different conditions highlight the complexity
of driving dynamics in work zone areas. These insights are crucial for developing targeted strategies to enhance safety and efficiency in such areas.
Although this study contributes much to understanding driver behaviors when passing by a work zone area, it still has some limitations that need to be addressed in the future. Firstly, this study covered only urban work zones and did not consider the traffic volume, which may be a critical factor that may impact driver behaviors. Secondly, this study recruited 26 participants to perform the experiment. The participants are not professional drivers, such as taxi drivers or truck drivers. Professional drivers may perform differently when they drive through a work zone. Lastly, this study utilized the position of heavy equipment to represent different construction area layouts. It should be acknowledged, however, that construction areas consist of numerous additional elements beyond heavy equipment, including various types of machinery and numerous construction workers. Moreover, the scenario developed for this experimental study was static, indicating that all components within the construction area remained stationary. Consequently, the simulated work zone environment may not fully capture the dynamic and uncertain nature inherent in actual work zones, which are characterized by continuous movement and variability. This limitation may influence the generalizability of findings in this study.
CRediT authorship contribution statement
Peng Liu: Writing – review & editing, Writing – original draft, Software, Resources, Methodology, Conceptualization. Chengyi Zhang: Supervision, Methodology, Funding acquisition. David Arditi: Writing – review & editing. Ayoola Olorunnishola: Writing – review & editing.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgement
This work was supported by USDOT, Office of the Assistant Secretary for Research and Technology (Project No.69A3551747108).
Data availability
The data that has been used is confidential.
References
- Ahmed, M.M., Ghasemzadeh, A., 2018. The impacts of heavy rain on speed and headway Behaviors: An investigation using the SHRP2 naturalistic driving study data. Transp. Res. Part C Emerging Technol. 91, 371–384. https://doi.org/10.1016/j. trc.2018.04.012.
- Al Hawarneh, A., Bendak, S., Ghanim, F., 2021. Construction site layout planning problem: Past, present and future. Expert Syst. Appl. 168, 114247. https://doi.org/ 10.1016/j.eswa.2020.114247.
- Arca, A.A., Chouljian, E., Mouloua, M., 2022. The role of gender differences in distracted driving behavior: a psychophysiological approach. In: Proceedings of the Human Factors and Ergonomics Society Annual Meeting. https://doi.org/10.1177/ 1071181322661453.
- Banerjee, S., Jeihani, M., Moghaddam, R., 2018. Impact of mobile work zone barriers on driving behavior on arterial roads. J. Traffic Logist. Eng. 6 (2). https://doi.org/ 10.18178/jtle.6.2.37-42.
- Baratian-Ghorghi, F., Zhou, H., 2017. Traffic control devices for deterring wrong-way driving: Historical evolution and current practice. J. Traff. Transport. Eng. (Engl. Ed.) 4 (3), 280–289. https://doi.org/10.1016/j.jtte.2016.07.004.
- Bassani, M., Catani, L., Cirillo, C., Mutani, G., 2016. Night-time and daytime operating speed distribution in urban arterials. Transport. Res. F: Traffic Psychol. Behav. 42, 56–69. https://doi.org/10.1016/j.trf.2016.06.020.
Boyce, T.E., Geller, E.S., 2002. An instrumented vehicle assessment of problem behavior and driving style:: Do younger males really take more risks? Accid. Anal. Prev. 34 (1), 51–64. https://doi.org/10.1016/S0001-4575(00)00102-0.
Chauhan, P., Kanagaraj, V., Asaithambi, G., 2022. Understanding the mechanism of lane changing process and dynamics using microscopic traffic data. Physica A 593, 126981. https://doi.org/10.1016/j.physa.2022.126981.
Cordellieri, P., Baralla, F., Ferlazzo, F., Sgalla, R., Piccardi, L., Giannini, A.M., 2016. Gender effects in young road users on road safety attitudes, behaviors and risk perception. Front. Psychol. 7, 1412.https://doi.org/10.3389/fpsyg.2016.01412.
Demeke, A.Y., Younesi Heravi, M., Dola, I.S., Jang, Y., Le, C., Jeong, I., Wang, D., 2024. Advancing highway work zone safety: a comprehensive review of sensor technologies for intrusion and proximity hazards. Transp. Res. Rec. 03611981241265845.https://doi.org/10.1177/03611981241265845.
Difei, J., Cancan, S., Zhongyin, G., Ran, L., 2021. Influence of the median opening length on driving behaviors in the crossover work zone—a driving simulation study. Transport. Res. F: Traffic Psychol. Behav. 82, 333–347. https://doi.org/10.1016/j. trf.2021.09.001.
Duan, K., Yan, X., Li, X., Hang, J., Yang, J., 2020. Analysis of driver's decision distance and merging distance in work zone area based on parametric survival models: with the aid of a driving simulator experiment. Transport. Res. F: Traffic Psychol. Behav.
71, 31–48. https://doi.org/10.1016/j.trf.2020.03.017. Fei, L., Zhu, H., Han, X., 2016. Analysis of traffic congestion induced by the work zone. Physica A 450, 497–505. https://doi.org/10.1016/j.physa.2016.01.036.
Garber, N.J., Zhao, M., 2002. Distribution and characteristics of crashes at different work zone locations in Virginia. Transp. Res. Rec. 1794 (1), 19–25. https://doi.org/ 10.3141/1794-03.
Hang, J., Yan, X., Ma, L., Duan, K., Zhang, Y., 2018. Exploring the effects of the location of the lane-end sign and traffic volume on multistage lane-changing behaviors in work zone areas: a driving simulator-based study. Transport. Res. F: Traffic Psychol. Behav. 58, 980–993.https://doi.org/10.1016/j.trf.2018.07.024.
Harb, R., Radwan, E., Yan, X., Pande, A., Abdel-Aty, M., 2008. Freeway work-zone crash analysis and risk identification using multiple and conditional logistic regression. J. Transp. Eng. 134 (5), 203–214. https://doi.org/10.1061/(ASCE)0733-947X(2008) 134:5(203).
Herrero-Perez, D., Martinez-Barbera, H., 2010. Modeling distributed transportation systems composed of flexible automated guided vehicles in flexible manufacturing systems. IEEE Trans. Ind. Inf. 6 (2), 166–180. https://doi.org/10.1109/ TII.2009.2038691.
Horswill, M.S., Coster, M.E., 2002. The effect of vehicle characteristics on drivers' risktaking behaviour. Ergonomics 45 (2), 85–104. https://doi.org/10.1080/ 00140130110115345.
Hou, G., Chen, S., 2020. Study of work zone traffic safety under adverse driving conditions with a microscopic traffic simulation approach. Accid. Anal. Prev. 145, 105698. https://doi.org/10.1016/j.aap.2020.105698.
Katayama, A., Abe, T., Hase, A., Miyatake, N., 2024. Relationship between driving ability and physical fitness factors in older adults: a multiple linear regression analysis. Int. J. Environ. Res. Public Health 21 (6), 660.https://doi.org/10.3390/ijerph2106066.
Kummetha, V.C., Kondyli, A., Chrysikou, E.G., Schrock, S.D., 2020. Safety analysis of work zone complexity with respect to driver characteristics—a simulator study employing performance and gaze measures. Accid. Anal. Prev. 142, 105566. https:// doi.org/10.1016/j.aap.2020.105566.
Li, B., Zhang, L., Meng, Q., Pan, W., 2022. A Novel Decision-making Framework for Lane Change Considering Intentions of Traffic Participants. In: 2022 6th CAA International Conference on Vehicular Control and Intelligence (CVCI). https://doi. org/10.1109/CVCI56766.2022.9964808.
Liu, J., Khattak, A., Zhang, M., 2016. What role do precrash driver actions play in work zone crashes?: Application of hierarchical models to crash data. Transp. Res. Rec. 2555 (1), 1–11. https://doi.org/10.3141/2555-01.
Liu, X., Liang, J., Fu, J., 2021. A dynamic trajectory planning method for lane-changing maneuver of connected and automated vehicles. Proceed. Institut. Mechan. Eng., Part D: J. Automob. Eng. 235 (7), 1808–1824. https://doi.org/10.1177/ 0954407020982712.
Ma, S., Hu, J., Wang, R., 2023. Impact of Transition Areas on Driving Workload and Driving Behavior in Work Zones: A Naturalistic Driving Study. Appl. Sci. 13 (21), 11669. https://doi.org/10.3390/app132111669.
Meng, Q., Weng, J., 2011. An improved cellular automata model for heterogeneous work zone traffic. Transp. Res. Part C Emerging Technol. 19 (6), 1263–1275. https://doi. org/10.1016/j.trc.2011.02.011.
Meng, Q., Weng, J., 2012. Classification and regression tree approach for predicting drivers' merging behavior in short-term work zone merging areas. J. Transp. Eng. 138 (8), 1062–1070. https://doi.org/10.1061/(ASCE)TE.1943-5436.0000412.
Miller, E.E., Boyle, L.N., 2015. Driver behavior in road tunnels: Association with driver stress and performance. Transp. Res. Rec. 2518 (1), 60–67. https://doi.org/10.3141/ 2518-08.
Møller, M., Gregersen, N.P., 2008. Psychosocial function of driving as predictor of risktaking behaviour. Accid. Anal. Prev. 40 (1), 209–215. https://doi.org/10.1016/j.
aap.2007.05.007. Morgan, J., Duley, A., Hancock, P., 2010. Driver responses to differing urban work zone configurations. Accid. Anal. Prev. 42 (3), 978–985. https://doi.org/10.1016/j. aap.2009.12.021.
Ozkan, ¨ T., Lajunen, T., 2006. What causes the differences in driving between young men and women? The effects of gender roles and sex on young drivers' driving behaviour and self-assessment of skills. Transport. Res. F: Traffic Psychol. Behav. 9 (4), 269–277. https://doi.org/10.1016/j.trf.2006.01.005.
Pan, T., Lam, W.H., Sumalee, A., Zhong, R., 2016. Modeling the impacts of mandatory and discretionary lane-changing maneuvers. Transp. Res. Part C Emerging Technol. 68, 403–424. https://doi.org/10.1016/j.trc.2016.05.002.
Pesti, G., McCoy, P.T., 2001. Long-term effectiveness of speed monitoring displays in work zones on rural interstate highways. Transp. Res. Rec. 1754 (1), 21–30. https:// doi.org/10.3141/1754-03.
Petzoldt, T., Rossner, P., Mair, C., Bullinger, A.C., Krems, J.F., 2016. How does left lane width in motorway work zones affect driver behaviour and perception? Adv. Transp. Stud. 1, 51–58.
Racha, S., Chowdhury, M., Sarasua, W., Ma, Y., 2008. Analysis of work zone traffic behavior for planning applications. Transp. Plan. Technol. 31 (2), 183–199. https:// doi.org/10.1080/03081060801948175.
Raddaoui, O., Ahmed, M.M., Gaweesh, S.M., 2020. Assessment of the effectiveness of connected vehicle weather and work zone warnings in improving truck driver safety. IATSS Res. 44 (3), 230–237. https://doi.org/10.1016/j.iatssr.2020.01.001.
Raju, N., Arkatkar, S., Joshi, G., 2020. Effect of construction work zone on traffic stream parameters using vehicular trajectory data under mixed traffic conditions. J. Transport. Eng., Part A: Syst. 146 (6), 05020002. https://doi.org/10.1061/ JTEPBS.0000353.
Razmara, A., Aghamolaei, T., Madani, A., Hosseini, Z., Zare, S., 2018. Prediction of taxi drivers' safe-driving behaviors based on the theory of planned behavior: The role of habit. J. Educat. Health Promot. 7 (1), 139. https://doi.org/10.4103/jehp.jehp\_61\_18.
Roustaei, N., 2024. Application and interpretation of linear-regression analysis. Med. Hypothes., Discov. Innovat. Ophthalmol. 13 (3), 151. https://doi.org/10.51329/ mehdiophthal1506.
Shakouri, M., Ikuma, L.H., Aghazadeh, F., Punniaraj, K., Ishak, S., 2014. Effects of work zone configurations and traffic density on performance variables and subjective workload. Accid. Anal. Prev. 71, 166–176. https://doi.org/10.1016/j. aap.2014.05.016.
Shen, X., Marks, E., Pradhananga, N., Cheng, T., 2016. Hazardous proximity zone design for heavy construction excavation equipment. J. Constr. Eng. Manag. 142 (6), 05016001. https://doi.org/10.1061/(ASCE)CO.1943-7862.0001108.
Silveira, R., Ferreira, S., Cunha, L., Rebelo, M., 2023. Driver behavior on road work zones: a systematic review. Transp. Res. Procedia 72, 1318–1325. https://doi.org/ 10.1016/j.trpro.2023.11.593.
Theofilatos, A., Ziakopoulos, A., Papadimitriou, E., Yannis, G., Diamandouros, K., 2017. Meta-analysis of the effect of road work zones on crash occurrence. Accid. Anal. Prev. 108, 1–8. https://doi.org/10.1016/j.aap.2017.07.024.
Turner, C., McClure, R., 2003. Age and gender differences in risk-taking behaviour as an explanation for high incidence of motor vehicle crashes as a driver in young males.
Inj. Control Saf. Promot. 10 (3), 123–130. https://doi.org/10.1076/ icsp.10.3.123.14560.
van Huysduynen, H.H., Terken, J., Meschtscherjakov, A., Eggen, B., Tscheligi, M., 2017. Ambient light and its influence on driving experience. In: Proceedings of the 9th International Conference on Automotive User Interfaces and Interactive Vehicular Applications. https://doi.org/10.1145/3122986.3122992.
van Winsum, W., De Waard, D., Brookhuis, K.A., 1999. Lane change manoeuvres and safety margins. Transport. Res. F: Traffic Psychol. Behav. 2 (3), 139–149. https://doi. org/10.1016/S1369-8478(99)00011-X.
von der Heiden, N., Geistefeldt, J., 2016. Capacity of freeway work zones in Germany. Transp. Res. Procedia 15, 233–244. https://doi.org/10.1016/j.trpro.2016.06.020.
Wang, P., Chan, C.-Y., de La Fortelle, A., 2018. A reinforcement learning based approach for automated lane change maneuvers. In: 2018 IEEE Intelligent Vehicles Symposium (IV). https://doi.org/10.1109/IVS.2018.8500556.
Warne, R., 2014. A primer on multivariate analysis of variance (MANOVA) for behavioral scientists. Pract. Assess. Res. Eval. 19 (1). https://doi.org/10.7275/sm63- 7h70.
Washington, S., Karlaftis, M.G., Mannering, F., Anastasopoulos, P., 2020. Statistical and econometric methods for transportation data analysis. CRC Press. https://doi.org/ 10.1201/9780429244018.
Wei, T., Zhu, T., Bai, H., Zhao, L., Wang, X., 2024. Effects of driver gender, driving experience, and visibility on car-following behavior. Transp. Res. Rec. 03611981241258988.https://doi.org/10.1177/03611981241258988.
Weng, J., Meng, Q., 2012. Effects of environment, vehicle and driver characteristics on risky driving behavior at work zones. Saf. Sci. 50 (4), 1034–1042. https://doi.org/ 10.1016/j.ssci.2011.12.005.
Worsfold, K., Worsfold, J., Bradley, G., 2007. Interactive effects of proactive and reactive service recovery strategies: the case of rapport and compensation 1. J. Appl. Soc. Psychol. 37 (11), 2496–2517. https://doi.org/10.1111/j.1559-1816.2007.00267.x.
Yagil, D., 1998. Charismatic leadership and organizational hierarchy: attribution of charisma to close and distant leaders. Leadersh. Q. 9 (2), 161–176. https://doi.org/ 10.1016/S1048-9843(98)90003-0.
Yang, H., Ozbay, K., Ozturk, O., Xie, K., 2015. Work zone safety analysis and modeling: a state-of-the-art review. Traffic Inj. Prev. 16 (4), 387–396. https://doi.org/10.1080/ 15389588.2014.948615.
Yang, Q.I., Koutsopoulos, H.N., 1996. A microscopic traffic simulator for evaluation of dynamic traffic management systems. Transp. Res. Part C Emerging Technol. 4 (3), 113–129. https://doi.org/10.1016/S0968-090X(96)00006-X.