Evaluating road work site safety management: A case study of the Amman bus rapid transit project construction

Abstract

This study explores safety perceptions in Jordan’s road construction sites, where work-zone hazards have been insufficiently addressed despite extensive research elsewhere. The study focuses on the impact of safety measures on workers and their environment, providing in-depth insights into roadway construction projects’ safety climate. The study analyzed responses from the Project-Based Group (PB) and the General-Based Group (GB). The PB group included 75 subjects interviewed face-to-face, while the GB group had 43 subjects who filled out an online form. The ratings indicate that staff training, traffic operations, loading/unloading, and site administration safety measures are perceived positively. However, there is a need to improve general site safety and maintenance/management protocols. Safety perceptions differ significantly between the PB and GB groups across multiple safety factors. While the response rates are similar to some extent, differences in safety measure evaluations emphasize the need for targeted interventions. The study highlights the importance of standardized safety protocols, particularly in areas like general traffic safety and maintenance. Using Exploratory Factor Analysis (EFA), the research identifies nine factors that shed light on safety perceptions. Sociodemographic elements such as age, job roles, education, and representation significantly impact perceptions. The correlation analysis reinforces the association of factors. The Confirmatory Factor Analysis model (1^st and 2^nd order) confirms a strong association between the latent and observed variables. However, some model fit criteria were still unmet, signaling the need for further refinement. Despite challenges, the research provides valuable insights into construction safety perceptions, guiding future research.

Keywords:

• work zones
• safety factors
• applications rating
• perception
• safety appropriateness rating
• exploratory factor analysisi
• confirmatory factor analysis

1. Introduction

Total quality management is a technique that assumes that all employees and personnel in the business or on the site are expected to improve operations, from development to production to execution and completion. Quality management in the construction industry differs from other sectors, making its implementation complicated as it involves cost and time overruns, not to overlook human errors and faults. To face these inherent challenges, on-site officials must change their mindset from monitoring to continuously seeking improvement opportunities to ensure appropriate quality and productivity. A diagnostics study investigated the causes of poor quality management in Iraqi construction projects that were divided into eight main groups (Equipment causes, Labours causes, Systems causes, Materials causes, Design and Execution causes, Subcontractors causes, Site staff causes, Contract causes) (Jasim, 2021). A similar study was conducted in Brazil to define the critical success factors affecting the implementation of total quality management in the construction industry (Reinaldo et al., 2021).

According to Manual Traffic on Uniform Control Device MUTCD, the work zone is “an area of a highway with construction, maintenance, or utility work activities (FHWA, 2009). A work zone is typically marked by signs, channelizing devices, barriers, pavement markings, and/or work vehicles. It extends from the first warning sign or vehicle equipped with high intensity rotating, flashing, oscillating, or strobe lights to the END ROAD WORK sign or the last temporary traffic control device” (See Figure 1).

Figure 1. A typical work zone layout (Sharma et al., 2017).

The construction industry in Jordan suffered human and financial losses due to poor safety performance. According to statistics from the Social Security Corporation, construction work accidents accounted for (1876) 12.2% of all accidents in 2016, down from (2022) in 2015. The latter constitute 13.9% of the total incidents registered, corresponding to (45.1) injuries per thousand insured, the highest in 2015. The 2021 work-related injury report from the Social Security Corporation highlighted the construction sector’s alarming fatality rate at 33.4 deaths per 100,000 insured workers. This sector reported 11 fatalities, making up 13.9% of recorded deaths. Within the construction sector alone, 635 injuries were reported, accounting for 4.9% of the total workplace injuries in 2021. The injury rate in this sector was 19.3 per 1,000 insured workers. Fatal road accidents were a significant contributor, comprising 46.8% of all fatalities reported to the corporation, with 37 deaths attributed to them (Al-Sharaa, 2022).

Construction labor in Jordan accounts for about 6.4% of the labor force (UNDP et al., 2013). The Ministry of Labor develops relevant legislation and enforces safety regulations (Articles 78–85 of the 1996 Labor Law). The prevailing legislation is continuously criticized for Articles 78–85 of the 1996 Labor Law not being detailed enough regarding significant safety issues (equipment safety, training, environment set-up, etc.). For highway projects, the Ministry of Public Works and Housing sets standards for roadway work zones. Police annual reports show that few causalities in road traffic are related to road defects. Crashes in a work zone are reported among these few causalities, accounting for 5.6% of all impacts associated with road defects (Public Security Directorate [PSD], 2019). In 2020, road defects caused around 1% of road accidents and produced casualties, of which 15.1% were related to work zones (PSD, 2019). Although official statistics cannot reflect the severity and danger of the situation, there are many accidents in work zones.

Consequently, injuries occur, time is lost, and money is lost. Contracts for road construction are required to include a traffic management plan and detour plan. Driving through the site is stressful, inconvenient, and hazardous because of the lack of follow-up and monitoring. Although significant, few studies were conducted on work zone safety in Jordan. The study examines safety culture and perceptions of road work zone construction in Jordan. It conducts a literature review and proposes research methods to reach conclusions.

2. Literature review

Safety metrics used in construction, such as accident rates and injury frequency, are reactive, as they only record events that have already occurred. This approach neglects near-misses and often results in underreporting, hindering proactive prevention measures. Behavioral observation measures are a better alternative but lack severity consideration and ignore organizational and management safety aspects. A comprehensive approach that combines quantitative and qualitative metrics is needed, distinguishes between operational and managerial levels, and includes leading indicators such as safety climate or culture perception. Understanding top-down organizational attributes and bottom-up perceptual approaches is crucial for practical safety performance assessment and benchmarking.

2.1. Safety at work zone

Many researchers have examined safety in work zones over the last five decades. All previous work emphasizes the impact of roadway work zones on safety in terms of crash frequency. Many researchers concluded that crash rates in highway work zones significantly increase as compared to the prework zone conditions (Garber & Zhao, 2002; Khattak & Council, 2002; La Torre et al., 2014; Pal & Sinha, 1996; Saleh et al., 2013; Srinivasan et al., 2011; Wang et al., 1996). Other studies showed that there is an increasing effect, not significant though, on crash frequencies when compared to “prework zone” conditions (Hall & Lorenz, 1989; Khattak & Council, 2002; Lisle, 1978; Nemeth & Migletz, 1978; Ozturk et al., 2014; Pal & Sinha, 1996; Srinivasan et al., 2011). Ozturk et al. (2014), based on data from 60 work zone sites in New Jersey between (2001 and 2011) found that the average number of crash rates increased by 18.8% and 24.4%, respectively, during work activities. Rear-end crash frequency was 8.6% higher compared to non-work zone conditions. A study by the University of Florence in Italy showed that the overall expected crash frequency on a motorway segment during work time in the work zone is about 32% greater than in the “prework zone” period (Saleh et al., 2013). The increase in crash rate could reach up to 65% when work activity occurs and temporarily involves lane closures, as indicated by an investigation of 64 freeway construction projects in four states in the USA (Srinivasan et al., 2011). The total crash rate during the during-work zone period was 21.5% higher than in the prework zone period, while the increases in PDO and injury crash rates were equal to 23.8% and 17.3% (Khattak & Council, 2002).

A state-of-the-art review study found that 48% of previous studies on work zone crashes did not provide conclusive evidence of increased crash severity under these conditions (Yang et al., 2015). In Ohio, data from nine work zones showed that crashes were slightly less severe than in “non-work zones” (Ha & Nemeth, 1995) and that crash severity in work zones increased significantly. Night crashes are usually more severe than during the day (Garber & Zhao, 2002). The prevailing types of crashes in the work zone occur at various locations and times. However, some studies agree that rear-end collisions are the most frequent crash types within work zones (Antonucci et al., 2005; Bai & Li, 2007; Garber & Zhao, 2002; Mohan & Gautam, 2002; Saleh et al., 2013; Ullman et al., 2008).

Analyzing 17,228 work zone crashes in four states (California, North Carolina, Ohio, and Washington) over six years (2000 to 2005) showed a significantly different distribution of work zone crash types between daytime and night-time conditions. However, rear-end collisions remain the most common type (Ullman et al., 2008). The primary causal factors in crashes in work zones are human errors, such as inattention, driving, and misjudging (Bai & Li, 2007; Chambless et al., 2002; Daniel et al., 2000; Mohan & Gautam, 2002). Human error could differ significantly by time of day (Hill, 2003). Speeding (Garber & Zhao, 2002) and inefficient traffic control (Ha & Nemeth, 1995) are significant factors causing work zone crashes. Fatalities on 720 work sites in the U.S. in 2008 showed that speed contributed to 31% of cases. The critical factor is seatbelt use-lack, contributing to 53% of all fatalities (Dissanayake & Akepati, 2009). The prevailing types of work zone crash types vary with different locations and times, but most previous studies agree that rear-end collisions are the most frequent crash types within work zones (Antonucci et al., 2005; Bai & Li, 2007; Garber & Zhao, 2002; Mohan & Gautam, 2002; Saleh et al., 2013; Ullman et al., 2008).

2.2. Safety perception and practice at construction worksite

Construction safety research has delved into hazard recognition and risk perception inadequacies, which are crucial for accident prevention. In Saudi Arabia, Suresh et al. (2017) identified widespread dissatisfaction with existing health and safety practices among workers and managers, underscoring the urgent need for improvements. Shah and Alqarni (2018) studied safety risks in highway construction, focusing on ergonomic injuries from postures and equipment collisions. Despite OSHA regulations, issues endure. Their study in Saudi Arabia highlighted vehicle collision injuries. To prevent these, they advocate for personal protective equipment use, strict safety compliance, improved traffic control, and fostering a safety culture, especially in developing nations. Erogul and Alyami (2017) highlighted safety protocol neglect in small construction projects, leading to worker unawareness and lenient inspections. Al-Shayea et al. (2019) delved into the motivational factors behind shortcuts, a common unsafe act, revealing nuanced insights from quantitative and qualitative data.

Pandit et al. (2019) discovered a positive safety climate significantly improved hazard recognition and risk perception, emphasizing its direct and indirect impact. Han et al. (2019) introduced a safety cognition framework, addressing biases in hazard perception and advocating targeted safety training programs. Elmoujaddidi and Bachir (2019) found a risk perception bias counterbalanced by a positive safety climate in Morocco. Tailored safety approaches, such as clarifying responsibilities, regular training, and open communication, are recommended, considering factors that could affect labourer perceptions (Chen et al., 2019).

The Queensland study on roadwork incidents with 66 participants highlighted issues like vehicles entering work areas and collisions with traffic controllers, often due to driver errors like speeding and distraction. The research emphasized the crucial need for understanding roadwork hazards and advocated effective site management. Worker insights revealed hazards such as driver aggression and adverse weather, suggesting safety measures like police presence and enhanced driver awareness. The study stressed integrating diverse worker perspectives into safety measures, addressing literature gaps, and aligning safety strategies with workers’ experiences (Debnath et al., 2013, 2015).

Xia et al. (2020) found from a survey involving 311 workers from 35 different workgroups that workers’ risk perception can affect their safety behaviours in construction. A positive safety climate among coworkers can mitigate the adverse effects of perceived risk, while a supervisor’s improved safety climate can lead to decreased motivation to be safe. Workplace climate is essential in influencing safety behaviour. Construction workers and managers in Iran have different perspectives on safety climate. Workers prioritize attitudes and relationships, while managers focus on rules and practices. Customized training and positive relationships between workers and managers are crucial to enhance safety. This should be considered when creating safety guidelines, particularly in developing countries’ construction sectors (Chan et al., 2021). Kim et al. (2022) provide valuable directions for enhancing safety management practices and developing context-specific guidelines to safeguard construction workers in challenging environments. A study in Taiwan, including 74 managers and 261 laborers, analyzed safety climate perceptions among construction managers and laborers and found varied perceptions, with managers having higher levels.

2.3. Measures to improve safety in road work

Investigating roadwork signs’ effectiveness, drivers glanced similarly at temporary and permanent signs within roadwork areas at a 40% frequency rate. Single roadwork signs attracted more attention and longer glances than multiple signs (Vignali et al., 2018). Investigating drivers’ acceptance of graphic-aided Portable Changeable Message Signs in work zones, drivers’ gender and age did not significantly impact their preferences (Huang & Bai, 2014). Various temporary traffic control strategies on freeways were compared to assess safety impacts, considering crash rates during construction and non-construction periods. Safety performance was related to segment length, duration, traffic volume, and closure type (Rista et al., 2017).

Examining the impacts of temporarily installed construction signage, a U.S. survey, based on 500 survey questionnaires through Amazon Mechanical Turk revealed drivers prefer dynamic characters over static signs. Drivers desire more work zone information but not the posted speed limit. Temporary signage was perceived as the least distracting feature compared to construction equipment, workers, and lighting (Jin & Gambatese, 2018). A study looked into improving multi-lane closure sign comprehension in work zones, addressing problems with existing signs like MUTCD W4–2, especially for multi-lane closures. Effective alternatives, such as the Upward Drop Arrow and one-arrow-per-lane designs, were identified through surveys and simulations. Field evaluations were proposed to enhance work zone safety and driver understanding, resolving road signage challenges effectively (Shaw et al., 2017, 2018).

A project funded by the European Union focused on improving work zone safety in eight European countries. The study examined incidents in work zones, psychological factors, and traffic management systems. The findings indicated that organizational structures, speed control, and warning techniques influence work zone safety. The recommendations included conducting regular safety assessments, prioritizing safety in contracts, enhancing worker skills, and educating drivers on work zone behaviour. Varhelyi et al. (2020) suggested implementing various measures, such as clear signage, universal symbols, adaptable illumination, vehicle restraint systems, and regular policy evaluations to improve safety in work zones. Evaluation of commercially available light towers (metal halide, LED, and balloon) showed that the type and orientation affect drivers’ visual performance, visibility, and glare perceptions (Bhagavathula & Gibbons, 2017).

2.4. Technology in construction safety

The use of technology and intelligent techniques in enhancing safety in construction sites has gained interest and become an attractive topic for researchers. A three-phase review of 500 articles on highway construction work zone safety was conducted using the number and location of publications and the types of technologies. Implementing smart work zone systems is a trend that is on the rise. The paper identified six additional research areas to understand better technology’s role in highway safety management (Chukwuma Nnaji et al., 2020). Innovative approaches and automation in highway construction can enhance safety and project quality. A study using a fuzzy index model in the U.S. identified key readiness indicators for construction automation adoption, emphasizing external factors’ substantial impact on quality management. Focusing on non-technological aspects is crucial for improving highway project outcomes (Ogunrinde et al., 2021). Recent studies highlight technology’s transformative impact on construction safety, mainly through virtual reality (VR) applications. Kim et al. (2021) used VR simulations to address risk habituation in road construction, effectively reducing inattentiveness to hazards and maintaining worker vigilance. Grégoire et al. (2022) explored VR interventions, showing sustained impacts on workers’ attentiveness through eye tracking and EEG data. Shohet et al. (2019) integrated information technology, leading to a 30% enhancement in construction quality and a remarkable 90% reduction in unsafe activities. These studies emphasize technology’s pivotal role in reshaping construction safety paradigms and offer innovative solutions for longstanding challenges.

2.5. Safety climate and culture

2.5.1. Concept and development

Zohar’s (1980) initial concept of safety climate (SC) as shared perceptions in the workplace has evolved through subsequent studies. Brown and Holmes (1986) expanded it to beliefs about safety, while Dedobbeleer and Béland (1991) emphasized management commitment and worker engagement. Coyle et al. (1995) defined safety climate as an objective measurement of attitudes, and Williamson et al. (1997) described it as an organization’s safety ethic. Mohamed (2002) focused on employees’ perceptions, while Cooper and Phillips (2004) emphasized shared employee perceptions of safety management. Hahn and Murphy (2008) highlighted shared perceptions, and Neale and Waters (2012) integrated psychological elements. Zohar’s later reflections (Zohar, 2010) reinforced the safety climate’s global importance and its links to leadership. In the construction industry, safety climate centres on management control failures rather than worker negligence, emphasizing the need for prioritizing safety and health. Mosly (2019) identifies 18 crucial safety factors in construction: work competence, pressure, hazard appraisal, physical environment, risk perception, worker involvement, supervision, adherence to rules, communication, and management commitment. In highway construction, inconsistencies in policy enforcement led to hazards and individual risk perceptions, impacting the overall safety climate. Mosly and Makki (2020) pinpoint 13 key factors, emphasizing communication and supportive environments providing insights to guide construction stakeholders in enhancing safety behavior and workplace culture. Understanding the distinction between safety culture and safety climate is pivotal in advancing safety outcomes in construction. Safety culture reflects organizational values, while safety climate captures immediate safety perceptions. Extensive research spanning sixty-three articles (2000–2021) underscores the vital role of safety attitudes, primarily focusing on organizational factors in developing nations. However, studies at the project level, especially in countries like Tanzania, are scarce. A tailored safety climate maturity framework for construction is proposed to address these gaps, emphasizing the need to consider diverse safety factors at different decision-making levels within organizations and projects (Kajumulo et al., 2023).

The balanced scorecard is a great way to evaluate safety culture in construction. It involves setting safety objectives, engaging stakeholders, and having management play a crucial role. Kaplan and Norton (1992) introduced this approach in 1992 and considered financial, customer, internal business, innovation, and learning perspectives. Mohamed (2003) recommends using the balanced scorecard as a benchmark for safety culture to ensure strategic safety goals, stakeholder engagement, management involvement, evaluation of performance, and improved safety.

2.5.2. Training in construction safety

Addressing global construction safety, Loosemore and Malouf’s (2019) study highlighted limited safety training’s impact on worker attitudes. Han et al. (2019) advocated tailored training through a safety cognition framework based on Chinese construction data. Alruqi et al. (2018) provided standardized safety dimensions for assessments. Abukhashabah et al. (2020) emphasized the need for improved safety training, especially for falling incidents. Chan et al. (2021) stressed personalized training and enhanced worker-manager relationships. Kornevs et al. (2018) offered diverse stakeholder perspectives valuable for policymaking in construction projects. These studies underscore the importance of targeted training initiatives in enhancing construction safety worldwide. In this context, the Malawi study (Chaswa et al., 2020) underscored the pivotal role of education and training in shaping construction workers’ risk perceptions. It recommended prioritizing educational initiatives and training programs for enhanced safety, aligning with the broader need for targeted training in global construction safety efforts.

2.5.3. Communication and organization for construction safety

Sanni-Anibire et al. (2018) emphasized poor communication and untested emergency response procedures among 196 construction workers in Saudi Arabia, stressing the importance of enhanced communication and standardized safety protocols. Effective supervisory safety communication in construction significantly impacts the safety climate and workers’ behaviours. In Melbourne, Australia, a study involving 20 workgroups across 11 sites found that traits like regular engagement, consistent messaging, active listening, and respectful delivery fostered a positive safety climate, influencing compliance and participation in safety practices (Zhang et al., 2020). Using interval-valued intuitionistic fuzzy logic and weighted divergence measures, another study assessed key stakeholders and refined analysis through linear programming. It identified crucial safety climate and perceptual factors, emphasizing the roles of clients, designers, construction managers, and contractors in managing workers’ unsafe behaviours (Khoshnava et al., 2020). Survey-based research employing a numerical rating scale highlighted the effectiveness of a novel construction safety culture and climate framework. It underscored the role of organizational culture, defined safety standards, and emphasized the influence of upper management and safety personnel on workers’ safety behaviour and motivation, offering valuable insights for construction firms to enhance site safety (Al-Bayati, 2021).

2.6. Structural Equation models in construction safety

In recent years, the construction industry has been a focal point of research exploring the complexities of safety climates. He et al. (2020) studied Chinese construction projects, employing Structural Equation Modelling (SEM) to dissect safety dynamics between supervisors and workers. The findings highlighted unique safety behavior patterns, necessitating tailored safety interventions for each group. Safeera and Bhavya (2020) delved into safety performance, unveiling intricate relationships between stress, unsafe events, injuries, and safety climate. Their research emphasized the multifaceted nature of safety in construction contexts. Seo et al. (2015) explored safety behaviors among temporary construction workers, illuminating the pivotal role of safety culture in shaping practices and attitudes. In Abu Dhabi’s healthcare sector, Al Faqeeh et al. (2019) unravelled the complexities of safety climates, emphasizing the role of safety attitudes as regulators, shaping perceptions and behaviours. Newaz et al. (2019) introduced the innovative concept of the “Psychological Contract of Safety,” shedding light on the impact of mutual safety obligations between supervisors and workers on construction safety behaviors. Finally, Nadhim et al. (2018) underscored the substantial positive correlation between safety climate and safety performance in retrofitting projects, emphasizing the indispensable role of safety climate awareness in construction safety. These studies collectively provide nuanced insights into safety climates, offering valuable directions for targeted safety interventions within the multifaceted construction industry landscape.

2.7. Safety in construction projects in Jordan

A study in Jordan based on a survey where a questionnaire was forwarded to contractors, consultants, and engineers involved in construction projects, and only 46 questionnaires were turned back. Research shows that safety violations, inadequate training, unfavorable conditions, and substandard equipment are the main reasons for accidents. Excavation is the most hazardous task, with 68% of accidents resulting in collapse. Companies with more experience have better safety procedures, and Safety is crucial to avoid expenses, delays, and injuries (Sarireh & Tarawneh, 2013). A study analyzed the safety conditions on the Amman-Salt highway, which saw 63 accidents during its two-year-long construction period. The researchers discovered that the road signs were poorly maintained and installed, failing to meet the standards set by MUTCD. Moreover, the traffic management plans lacked drawings indicating the placement of traffic control devices. These devices were later relocated following complaints from residents living in the area.

Additionally, some signs were not removed after the completion of work, creating confusion among drivers. The road lacked pavement markings, and the posted speed limit was reduced by 40 km/h from the actual speed limit of 80 km/h, which exceeds the MUTCD standard of 16 km/h (Abudayyeh et al., 2015). Suleiman et al. (2020) examined the impact of traffic management strategies on a work zone’s performance along Queen Rania Al-Abdullah Arterial Street in Amman, Jordan. The strategy, including Temporary Access Control, Limitation of Heavy Vehicles, and lane management, were analyzed using a microsimulation model created with VISSIM software. The heavy vehicle percentage, access road flow, parking, and driver behavior significantly influenced the work zone’s average delay (113.5 seconds/vehicle) and speed (10.1 km/hr). Implementing traffic management strategies reduced average delay to 89 seconds/vehicle (−22%) and improved speed to 29 km/hr (+30%).

2.8. Research problem

Extensive research has thoroughly studied the safety impact of road works and the effectiveness of measures to ensure safe traffic flow. Safety culture and climate have also been considered in construction projects, including roadway works. However, the factors addressed in the literature focus on the procedures and policies that impact worker safety on site, not their impact on the site surroundings. On the other hand, traditional safety studies address the Safety within the construction site or its premises and surroundings in terms of crash and fatality numbers. The perception of roadway officials as responsible for providing a safe workplace for workers and traffic has also been examined to address Safety within the site. The safety climate concepts, when it comes to road construction, manage the organization, perception of risk, and communication; not much work is devoted to the procedures and site management. Apart from a few studies completed in the transition and developing countries, road construction projects’ safety climate and culture are limited, including in Jordan. This study could contribute to bridging the gap of what would the impact of the measures within the site on its surroundings. It would also consider applying the safety climate concept in roadway construction projects.

Additionally, it will examine the factors that may influence road work zone safety. An exploration of the reflection of the professionals and laborers on their safety measures application and practice is reviewed. Based on the literature, the safety climate factors to be considered in this study would be supervision, guidance, inspection, education, and training; safety rules; adequacy of safety procedures; and management commitment to Safety (Mosly & Makki, 2020). This study examines the perceived effectiveness of safety measures related to safety climate factors among workers in roadway construction.

3. Research methods and data collection

3.1. Data collection

After identifying the aspects to be explored in this study regarding the safety climate, a series of questions related to factors affecting safety in each area were developed to create the questionnaire. The tool’s design also considers the safety audit checklist commonly utilized by road agencies globally. Data was gathered through on-site visits, in-person interviews, and Google Forms to ensure that each respondent provided only one response, thereby capturing the data accurately.

The survey involved two distinct groups. The first group, the “Project-Based” PB group, comprised employees working across four packages within the Amman Bus Rapid Transit (BRT) system, sharing their practical experiences. The BRT project was chosen due to its significant scale, involving multiple contracts, and spanning various parts of the city. The sample was selected to reflect different subjects’ roles on the site and professions. The second group, the General-Based “GB” Group, comprised general employees engaged in highway construction projects, reflecting their broader practical experiences. In the first group, approximately one hundred employees from the Amman BRT were randomly selected, of which 75 participated in face-to-face interviews to complete the questionnaire. Although not all questions were consistently answered, missing information was minimal, accounting for less than 10%, and varied across different items explored. The participants in this group included workers, supervisors, engineers, and managers. The second group’s questionnaire was transformed into a Google Form and distributed across various social media platforms. Despite being open for responses over a month, only forty-three completed forms were received, all of which contained complete and non-missing data, as respondents were required to answer all questions before proceeding. The population size of the sample would not be easy to define in numbers. Still, it would comprise all groups representing different jobs, genders, and qualifications, described as study factors.

3.2. Questionnaire design

A questionnaire assesses safety conditions within the road work zone and its surroundings to effectively meet the study objectives through the designed questionnaire. It is a modified version of various checklists for inspecting work zones globally. While not covering every on-site item examined in this study, the questionnaire sufficiently addresses the principal aspects for comprehensive evaluation. In the initial section, participants provide socio-demographic information, such as gender, age, qualifications, job roles, and experience, aiding in analyzing the diverse assessments based on these characteristics. The questionnaire’s second part delves into 49 safety measures related to each safety climate factor, asking respondents if these measures are implemented (with binary responses—yes or no) and to rate their assessment on a 5-point Likert scale. The scale ranges from 1 (Not appropriate at all) to 5 (Very appropriately applied), allowing for nuanced evaluations. Table 1 breaks down the assessed work zone safety measures in detail. It evaluates the implementation and appropriateness of these measures, categorized into “General Traffic” and “Within the Site.” These categories comprehensively cover safety climate factors, including supervision, guidance, inspection, education, training, safety rules, safety procedures, and management commitment to safety.

Table 1. Items describing assessed measures on work -zone by category and location and Safety Climate (SC) factors

Category:	Item No.	General Traffic	Category	Item No.	Within the Site
		Measure Description			Measure Description
SC1: Supervision, guidance, inspection:General Site	Q1	Suitable routes are provided for traffic and pedestrian	SC2 Education &Training: Staff Training	Q1	Drivers/employees gone through safety induction course
	Q2	Skid resistance surface are provided		Q2	Drivers/employee going through refresher training
	Q3	Vehicle traffic is separated from Pedestrian traffic		Q3	Drivers have a valid license and experience to operate the construction vehicle.
	Q4	Traffic Control devices are installed according to TMP		Q4	Flagmen been trained to guide the transport vehicles?
	Q5	Adequate Street lighting for all road users	SC3 Safety Rules: Traffic Operation	Q1	Appropriate transport vehicles are provided for the job?
	Q6	Road surface is even and stable to withstand heavy vehicles		Q2	Drivers’ compliance with regulatory laws is monitored
	Q7	Routes for transport vehicles and pedestrian walkways kept free from hazard objects		Q3	Site staff and workers wear safe outfit (e.g., visibility vests)
	Q8	Traffic routes properly marked		Q4	Drivers perform pre-operating tests before the actual operation
	Q9	Signs installed where required		Q5	Staff/workers evacuated from the areas before allowing powered vehicle to
	Q10	Other safety measures are provided (mirrors, rumble strips, etc)		Q6	Flagmen provide to guide the maneuvering of transport vehicles
	Q11	Designated traffic routes are wide enough		Q7	Flagmen position themselves at a safe position?
	Q12	Pedestrian pathways are wide enough		Q8	Vehicle properly parked talking all necessary precautions
SC1: Supervision, guidance, inspection: Maintenance & Management	Q1	Roads and walkways suitable and maintained for vehicular and pedestrian types and volumes	SC4 Safety Procedures: Loading & Unloading Activities	Q1	Loading or unloading operations conducted a way from traffic and pedestrians?
	Q2	Adequate Separation maintained between vehicles and pedestrians		Q2	Appropriate vehicle(s) used for loading or unloading
	Q3	Adequate parking places are provided and maintained to meet the demand		Q3	Loading or unloading activities conducted on flat, stable, and firm ground free from potholes
	Q4	Sufficient lighting level for vehicular and pedestrian traffic is provided		Q4	Parking brakes are applied to prevent hazardous movement (when coupling vehicles)
	Q5	Motorized vehicle traffic routes have firm and even surfaces that are well maintained.		Q5	Vehicles’ drivers used brakes and/or stabilizer to prevent unsafe movements during loading and unloading operations?
	Q6	Vehicular and pedestrian routes are kept free of hazards and obstructions		Q6	Lifting equipment, equipment or gears used for unloading vehicles possess valid legal inspection certificate (s)?
	Q7	Traffic route designed and maintained to avoid sharp or blind turns		Q7	Loading or unloading conducted by spreading load evenly to avoid vehicle instability
	Q8	Traffic route marking is correctly maintained according to traffic management plan (TMP)		Q8	Checks are made to ensure that loading is secured and no overloading before the permitting the vehicle to leaves the site
	Q9	Standard Traffic Control devices are installed and kept well where required		Q9	There is noticeable unsafe practice during loading and loading activities
	Q10	Convex mirror, road humps, or safety barriers are provided and supported where necessary		Q10	Employees follow safe work procedures when (un)coupling, (un)loading and securing loads?
	Q11	Pedestrian pathways are maintained wide enough	SC5 Management Commitment to safety Site Administration	Q1	Workplace traffic rules documented and distributed?
				Q2	Staff, workers, and visitors are aware of the workplace traffic rules
				Q3	Risk assessment conducted for work activities; communicated to all concerned personnel prior to operation
				Q4	Relevant actions are taken when the workplace’s traffic rules (if any) are violated

The questionnaire was streamlined to focus on critical aspects, reducing the time needed for online and face-to-face interviews and capturing essential site safety climate details. In the “General Traffic” category, there are subdivisions that focus on site measures (12 items) and their maintenance and management (11 items). Similarly, four categories fall under within-site safety measures. Beginning with the education and training aspect (4 items), staff training is assessed, while the safety rules factor is examined through site traffic operations (8 items). Safety procedures related to loading and unloading activities (10 items) are also scrutinized to assess safety practices. Lastly, management commitment to safety (4 items, categorized under site administration) encompasses administration on-site actions. Each category is detailed with specific items and corresponding descriptions, ensuring a systematic evaluation of safety protocols and a comprehensive understanding of the safety climate within work zones.

3.3. Study area amman BRT project description

Amman’s Bus Rapid Transit (BRT) system consists of two primary lines spanning 32 km. Line No. 2, covering 17 km from Sweileh to the Jordan Museum, commenced its trial operation in mid-2021. Line No. 1, a 15 km route from the station complex to Sweileh Station, started operating in the first quarter of 2022 (Figure 2). During the trial period, 96 high-specification buses run on the road at a frequency of 5 minutes, potentially doubling when the full service begins. The network includes 34 sub-stations serving key city attractions and is fully accessible to individuals with special needs. Operating hours during the trial phase are from 6 a.m. to 10 p.m., Saturday to Thursday, and 8 a.m. to 9 p.m. on Fridays. The infrastructure contracts, including stops and wearing courses, amounted to approximately 160 million US dollars.

Figure 2. Amman BRT and Amman-Zarqa BRT link layout.

Completing the project’s four mega packages took a decade, transforming the city of Amman into a bustling workshop with numerous construction sites, causing disruptions for its residents. Additionally, the Amman BRT project is linked at Al Mahatta Terminal with the Amman-Zarqa BRT, connecting the two cities over a 20 km distance with six sub-stations, including terminals at Zarqa and Al Mahatta. The estimated cost for this link is 282 million US dollars.

The interview evaluations were conducted at four distinct sites, each aligning with specific sections of BRT lines one and two. Site 1 covers the area parallel to Jordan University, aligning with line one. Site number 2 connects Sport City Station to Wadi Saqra along line 2. Site number 3 pertains to construction activities linking Sports City and the North Terminal on line one, while site four is affiliated with Tareq Station. Upon conducting site visits at the four construction sites included in the survey, there were notable deficiencies in traffic control measures.

Specifically, speed limits remained unchanged at 50 km/h (Figure 3(a)), even in areas where pedestrians were mixed with vehicle traffic with narrow sidewalks (Figure 3(b)). The existing traffic control devices were poorly maintained, and several electronic displays were non-functional (Figure 3(c)). Furthermore, there was insufficient separation between general traffic and ongoing construction activities (Figure 3(d)). The condition of traffic lanes deteriorated, with inadequate markings, and the pavements were slippery, posing potential hazards (Figure 3(e)). Additionally, construction materials were on-site but stored improperly and lacked proper markings for identification and organization (Figure 3(f)). These findings emphasize significant concerns about these construction sites’ safety, organization, and overall management practices. These issues and others form the basis of designing the study questionnaire.

Figure 3. Safety measures assessment in work zones: sites 1–4 for investigated safety factors.

3.4. Data analysis

3.4.1. Description and inferential analysis

The data were coded and analyzed using SPSS, employing both descriptive and inferential statistics, including analysis of variance (Figure 4). Descriptive statistics summarized responses using the mean and standard deviation to measure dispersion. The frequency of reactions was also presented, indicating the number of answers in each score category. Moreover, the average score for each response was also calculated by converting the five narrative scale into number from 1 (Not appropriate at all) to 5 (Very appropriately applied) The average score was compared against the following scale: 1–1.80 (exceptionally low application), 1.81–2.60 (low application), 2.61–3.40 (moderate application), 3.41–4.20 (high application), and 4.21–5 (extremely high application). Cronbach’s Alpha coefficients are statistical measures used to assess the internal consistency or reliability of a set of items in a questionnaire or test. A high Cronbach’s Alpha value, typically ranging from 0 to 1, indicates that the items in the test are highly correlated and measure a single construct consistently. Researchers often use a threshold value, such as 0.7 or 0.8, to determine whether the scale is reliable. A higher Cronbach’s Alpha α suggests excellent reliability, implying that the test produces consistent and dependable results. The rules of thumb scale per George and Mallery (2003) is α > 0.9– Excellent; α > 0.8– Good; α > 0.7– Acceptable; α > 0.6– Questionable; α > 0.5– Poor; and α < 0.5– Unacceptable.

Figure 4. A flow chart of the Research methods.

For inferential analysis, background information was used to explore its impact on subject responses. ANOVA tests assessed differences in responses based on subjects’ age group, job on-site, experience, and the representative side. ANOVA is a statistical test comparing the means of multiple groups to identify significant differences. It assesses variations within and between groups; if inter-group variance is significantly higher than intra-group variance, ANOVA detects mean differences. To draw statistical conclusions, it’s crucial to compare diverse factors concurrently, commonly used in experimental studies. Additionally, a t-test was employed to investigate potential response differences due to the subjects’ gender, data group (PB or GB), and project location (Urban or rural). P-value measures the likelihood of chance being responsible for any observed difference between groups. If the p-value is less than 0.05, it’s statistically significant, and the null hypothesis should be rejected. If it’s greater than 0.05, the deviation from the null hypothesis is not statistically significant, and the null hypothesis is not rejected. Correlation analyses examined the relationship between ratings and work zone conditions. The correlation scale ranged from 0 to 1, indicating the degree of association (0.8–1.0 very strong, 0.6–0.8 strong, 0.4–0.6 moderate, 0.2–0.4 weak, 0.0–0.2 very weak or no relationship).

3.4.2. Explanatory factor analysis

Explanatory Factor Analysis (EFA) is used in determining the number of factors using tools like the scree plot and selecting extraction methods like principal axis factoring. Completing item factor loadings involves techniques like principal axis factoring and maximum likelihood. Post-extraction, loadings are rotated to simplify the factor arrangement. EFA’s help in comprehending intricate relationships between variables. Varimax is used in this study as an orthogonal rotation method to simplify complex datasets by reducing variables into a smaller, interpretable set of factors. Varimax simplifies the interpretation of data by highlighting strong loadings and reducing inter-factor correlations

Principal Component Analysis (PCA) was employed to condense the dataset of each group into a single variable, aggregating the data to explore the original variation effectively. The data were standardized using IBM SPSS Statistics software, and covariance and correlation matrices were computed. Eigenvalues represented the variance explained by each principal component, and eigenvectors indicated the directions or loadings of the original variables in the principal component space. Eigenvalues exceeding one signified that the corresponding principal components captured more variance than a single variable. Eigenvalues and cumulative percentages of variation explained were used to determine the number of principal components, employing a minimum threshold of 60% for this project. Transformed variables were further analysed akin to the original or scaled variables. In addition to eigenvalues and eigenvectors, component loadings were examined to understand each variable’s contribution to the identified factors. Component loadings indicate the strength and direction of the relationship between the original and principal components, elucidating how much each variable contributes to the identified factors. This detailed analysis provided a comprehensive understanding of the data, shedding light on the intricate relationships between variables and factors within the work zone safety context.

3.4.3. Structural Equation modeming

Structural Equation Modelling (SEM) is a sophisticated statistical method that analyzes intricate relationships among variables, exploring connections between observed and unobservable factors (latent constructs) and accounting for measurement errors. Unlike data reduction techniques, SEM integrates various statistical methods like path analysis, factor analysis, and regression analysis within a unified framework, enabling researchers to test and model complex relationships. Latent constructs are deduced indirectly from observable variables as they cannot be directly measured. SEM models, characterized by structured equations, primarily serve to confirm hypotheses rather than exploration, providing insight into concepts influencing latent phenomena. Latent constructs are deduced indirectly from observable variables as they cannot be directly measured. SEM models, characterized by structured equations, primarily serve to confirm hypotheses rather than exploration, providing insight into concepts influencing latent phenomena. Coefficients in SEM are derived from hypothesized relationships and are limited to specified associations, allowing researchers to test multiple hypotheses and refine models by analyzing differences.

Confirmatory Factor Analysis (CFA), a part of SEM, tests hypotheses regarding relationships between observed and underlying latent constructs. CFA validates the factor structure of observed variables, confirming hypothetical data structures. This study utilized two levels of analysis. The first-order CFA explored relationships between observed variables and their underlying factors. The introduction of higher-order factors in second-order CFA expanded the concept, considering first-order factors as indicators for a superior factor. It explored the relationships between observed variables, immediate factors, and a higher-level latent factor, offering a more profound comprehension of inter-factor connections in the data structure.

In SEM, path coefficients signify the strength and direction of relationships between variables and are assessed for significance through associated p-values, usually focusing on p < 0.05 to indicate meaningful relationships. Path coefficients close to or exceeding 0.5 are considered large effect sizes, around 0.3 as medium, and near or below 0.1 as small, aiding researchers in interpreting relationships among variables. Significant correlations prompt the exploration of relationships among hypothesized latent variables (Mohamed et al., 2018). SEM’s goodness-of-fit indices evaluate the model’s fit to observed data. Absolute Fit Indices such as Chi-square statistic (CMIN) and Chi-square divided by Degrees of Freedom (CMIN/Df) ≤ 3 indicate acceptable fit, and ≤ 5 is reasonable. Root Mean Square Error of Approximation (RMSEA) ≤ 0.05 suggests acceptable fit, and up to 0.08 indicates mediocre fit. Incremental Fit Indices like Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Normed Fit Index (NFI), and Incremental Fit Index (IFI) compare models, with values ≥ 0.9 indicating good fit based on literature (Mulaik, 2007). These indices collectively assess the model’s suitability for representing observed data, which is crucial for valid interpretations in SEM analyses.

3.4.4. Tested hypotheses

Several hypotheses were formulated and tested to address the research problem and achieve its objectives. The following summarized the proposed hypotheses:

Hypothesis 1:

There is no association between socio-demographic factors (such as gender, age, education, job category, experience, and project geographic location) and variations in subjects’ perceptions of safety measures application and the appropriateness of the application. Additionally, it is believed that differences in perceptions are not due to reflecting on general experiences versus specific project practices.

Hypothesis 2:

Extracted factors describing investigated safety measures are not influenced by socio-demographic factors, project location, contract, or experience.

Hypothesis 3:

There is no association between the measures describing the within-site measures and those on the site’s premises.

Hypothesis 4:

There is no association between the latent variable describing measures the within-site measures and the latent variable relating to the measures serving the general traffic.

4. Results and findings

4.1. Sample characteristics and analysis

Table 2 presents a comprehensive overview of the characteristics of the PB Group and the GB Group. These groups are differentiated based on various factors such as gender, age, representation side, education level, professional category, and years of experience.

Table 2. Sample characterises

Group	Factor	Categories and Valid Response Percentage						Valid Percentage	Response rate
Project-Based	Gender	Female	Male					54 Cases	72.0%
		31.5%	68.5%
	Age	<30	30–39	40–49	50–70			74 Cases	98.7%
		39.2%	28.4%	17.6%	14.9%
	Representation Side	Consultant	Contractor	Project Owner				42 cases	56.0%
		28.6%	40.5%	31.0%
	Education Level	PhD	MSc	BSc	Associate degree	High School	Basic School	68 cases	90.7%
		1.5%	29.4%	61.8%	1.5%	2.9%	2.9%
	Professional Category	Project Director	Resident Engineer	Site Engineer	Supervisor	Labour		68 cases	90.7%
		10.3%	41.2%	14.7%	13.2%	20.6%
	Years of experience	1–3	4–6	7–10	10–15	<15		64 cases	85.3%
		26.6%	14.1%	20.3%	14.1%	25.0%
General-Based	Gender	Female	Male					43 Cases	100%
		27.9%	72.1%
	Age	<30	30–39	40–49	50–70			43 Cases	100%
		20.9%	37.2%	16.3%	25.6%
	Representation Side	Consultant	Contractor	Project Owner				43 Cases	100%
		32.6%	27.9%	39.5%
	Education Level	PhD	MSc	BSc	Associate degree	High School	Basic School	43 Cases	100%
			25.6%	67.4%	7.0%
	Professional Category	Project Director	Resident Engineer	Site Engineer	Supervisor	Labour		43 Cases	100%
		39.5%	16.3%	23.3%	7.0%	2.3%
	Years of experience	1–3	4–6	7–10	11–15	<15		43 Cases	100%
		11.6%	11.6%	23.3%	16.3%	37.2%

In the PB Group, which consists of 75 cases, most respondents were male (68.5%), primarily aged below 30 (39.2%). Representation-wise, there was a balanced presence of consultants (28.6%), contractors (40.5%), and project owners (31.0%). Educationally, most respondents held bachelor’s degrees (61.8%); in terms of professional category, resident engineers (41.2%) were predominant. Experience-wise, a significant portion had less than three years of experience (26.6%). The response rate for these frequent questions ranges from 56% (Subject’s representative side) to 98.7% (Subject’s age).

Comparatively, the GB Group, comprising 43 cases, had a less balanced gender distribution (27.9% female, 72.1% male). Age-wise, there was a varied distribution across different age brackets. Regarding representation, project owners were slightly more dominant (39.5%). Educationally, the majority had bachelor’s degrees (67.4%), and professional category-wise, project directors (39.5%) were prominent. Experience distribution was relatively even across different ranges. The subjects in this group answered all questions because it was mandatory in the Google form, the tool used in collecting the data for this group.

When comparing the two groups, it’s notable that the PB Group had a more diverse range of professional categories, with a substantial presence of resident engineers. In contrast, the GB Group had a higher percentage of project directors. Additionally, the PB Group showed a broader range of experience levels, including a significant portion of respondents with less than three years of experience, suggesting a mix of junior and more experienced professionals. The GB Group, on the other hand, had a more evenly spread experience distribution.

Overall, these distinctions highlight the varied composition of the two groups, which may reflect on the subject’s responses. The contingency table analysis was used to investigate if the sample distribution of each factor varies according to the sample group. The results show a statistically significant difference for age (χ² = 29.97, p < .001), experience (χ² = 11.20, p = .042), job (χ² = 22.52, p < .001), representative side (χ² = 27.93, p < .001), project location (χ² = 39.5, p < 0.001). The difference is insignificant for gender (χ2 = 0.146 p = 0.824) and education (χ² = 10.89, p = 0.095).

4.2. Safety measures application assessment

By analyzing the usage rates of safety measures among different responders, we can gain valuable insights into the strengths and areas for improvement in safety measures as perceived by PB and GB responders. GB group responders display an impeccable 100% response rate, surpassing PB responders’ rate of 90.5%. The overall response rate for the PB group reaches for the general traffic-Site premises 89.7% (90.8% and 88.61% for the GS and MM groups) and 91.2% for the within-site measures group (92.67%, 90.67%, 90.4%, and 92.67% for ST, TO, LU, and SA, respectively), indicating a high response level. In the General Traffic group, 59.6% of PB responders confirmed applying safety measures related to general site activities, but only 51.4% confirmed maintenance and management tasks adherence. On the other hand, GB responders rated the application of safety measures for tasks and activities related to maintenance and Management measures group that describes the supervision safety factor (64.1%). Generally, general site activities receive the least attention (59.1%). While both PB and GB responders in the General Traffic group show similar compliance rates in general site activities, GB responders exhibit a 12.7% higher application rate than PB responders in maintenance and Management tasks, indicating a notable difference between the two groups (Figure 5).

Figure 5. The application rate for measures of the investigated safety factors measures enhancing safety within the site of the project and its surrounding by respondent group.

Within the Site measures group, PB responders have the highest perception of adherence rate to traffic operation tasks (64.4%) and the lowest to site administration responsibilities (56.4%). Meanwhile, GB responders excel in loading and unloading activities, with an impressive 80.9% adherence rate but a slight lag in staff training activities (66.9%). GB responders consistently outperform PB responders in various safety factors within the Site group, including an 8.0% higher adherence rate in staff training responsibilities and a 9.7% higher compliance rate in traffic operation tasks. The most significant difference is seen in loading and unloading activities. GB responders display a remarkable 20.7% higher compliance rate, underscoring their strong commitment to safety measures in this factor group. Lastly, GB responders maintain a substantial 13.9% higher compliance rate in site administration tasks, emphasizing their superior commitment to site administration activities,

Overall, GB responders have a higher application rate of 61.5% in general traffic-related tasks than PB responders, who have a slightly lower average of 55.7%. For site-related tasks, GB responders exhibit a significantly higher perception of adherence rate of 75.0%, while PB responders demonstrate a 60.7% adherence rate. GB responders also have a 5.8% higher safety rating of measure application rate in general traffic-related tasks and a 14.3% higher rating within site-related tasks than PB responders. A further break down of the tested items describing each tested safety factor is illustrated in Figure 6.

Figure 6. The application rate for measures of the investigated measures of safety factors enhancing safety within the site of the project and its surrounding by respondent group.

The information presented provides insight into how responders in PB and GB settings view and implement safety protocols across various safety factors. The analysis demonstrates notable differences in adherence to safety measures between the two groups. Specifically, the group that assessed on-site perceptions had a more distinct divergence in how subjects perceived safety measures, with 50% of the items tested showing significant differences based on the respondent group. On the other hand, the group that evaluated general traffic measures found a difference in perception between the two groups in only 43.5% of the items tested (Figure 6, see Items indicated by yellow arrow).

For the General Traffic factor and within the General Site activities, the PB responders exhibited an application rate ranging from 41% to 79%, whereas GB responders demonstrated rates between 37% and 74%. The data indicates some significant differences; for instance, PB responders ranged from 38% to 71% of applications in traffic maintenance and management factor, while GB responders showed a narrower range of 49% to 79%. These figures reveal a mixed picture of adherence within the General Traffic Group, with varied application rates in different safety factor groups.

The disparities were more pronounced within the “Within-the-Site Group, particularly in Loading and Unloading Activities. PB responders showed rates between 51% and 74%, whereas GB responders demonstrated higher rates, ranging from 60% to 95%. Traffic Operation also exhibited a significant difference, where PB responders ranged from 49% to 74% application, while GB responders showed a more consistent range of 70% to 84%. In Site Administration, PB responders had rates between 47% and 64%, whereas GB responders demonstrated rates from 53% to 77%. These findings highlight substantial differences in safety measure application between PB and GB responders within specific safety factor groups.

The statistical significance of differences further emphasizes these disparities. For instance, in Traffic Operation, the application rates differed significantly between PB (56%) and GB (81%) responders. Similarly, in Loading and Unloading Activities, PB responders demonstrated a lower application rate of 51%, while GB responders had a significantly higher rate of 60%.

Statistical analyses, including ANOVA tests and t-tests, were employed to investigate the impact of various factors defining the sample characteristics on respondents’ perceptions regarding the application of safety measures. The results revealed significant differences in responses attributed to the tested factors. The results showed substantial variations in responses among different subject groups and genders. Statistical analyses, including ANOVA tests and t-tests, were employed to investigate the impact of various factors defining the sample characteristics on respondents’ perceptions regarding the application of safety measures. The results revealed significant differences in responses attributed to the tested factors. Specifically, the findings indicated statistical variances in responses concerning subjects’ group (t = 2.18, p = .016), gender (F = 5.516, P = 0.001), age (F = 3.15, P = 0.009), an education level (F = 5.59, P < .001), experience period (F = 17.34, P < 0.001), job category (F = 12.911, p < 0.001), representative (t = 5.4, p < 0.001), side (t = −3.89, p < 0.001), project location (F = 10.36, p < .001), and project site-contract (F = 12.911, p < 0.001). The results highlight how demographic and professional factors affect people’s views on safety measures.

4.3. Appropriateness of application safety measures

The following analysis will cover each area and discuss the responses related to the appropriateness of the application. Before conducting the study, it is necessary to investigate the reliability of the subject rating through the Cronbach Alpha test.

4.3.1. Reliability test

Table 3 shows Cronbach Alpha coefficients for various safety factors in General Traffic and Within the Site. The groups consist of PB and GB responders. The Cronbach Alpha values indicate the consistency and reliability of responses within each factor. In the General Traffic group, PB responders showed internal solid consistency across all safety factors (with values ranging from 0.83 to 0.97). The highest reliability was in Maintenance and Management (MM) activities (Alpha = 0.97) and the lowest in Loading and Unloading (LU) tasks (Alpha = 0.83).

Table 3. Reliability test result for each safety factor measures group by respondent group (Cronbach Alpha coefficients)

Group	GS		ST		TO		LU		SA		MM		General Traffic		Within the site		Overall
Group Based	PB	GB	PB	GB	PB	GB	PB	GB	PB	GB	PB	GB	PB	GB	PB	GB	PB	GB
	0.94	0.95	0.96	0.83	0.96	0.95	0.92	0.94	0.84	0.83	0.83	0.96	0.97	0.97	0.98	0.97	0.99	0.98
Both Groups	0.94		0.89		0.95		0.93		0.83		0.93		0.97		0.99		0.99

GB responders showed slightly lower but acceptable reliability, ranging from 0.89 to 0.99. Their highest reliability was in Maintenance and Management (MM) activities (Alpha = 0.99) and the lowest in Loading and Unloading (LU) tasks (Alpha = 0.83). Overall, GB responders demonstrated higher consistency across all safety factors. Within the Site group, PB responders showed similar patterns of internal consistency (ranging from 0.83 to 0.97), with the highest reliability in Traffic Operation (TO) activities (Alpha = 0.97) and the lowest in General Site (GS) tasks (Alpha = 0.83). GB responders exhibited higher reliability across the board (ranging from 0.83 to 0.99), with the highest in Maintenance and Management (MM) activities (Alpha = 0.99) and the lowest in Site Administration (SA) tasks (Alpha = 0.83).

GB responders consistently demonstrated higher internal consistency than PB responders across all safety factors. These Cronbach Alpha coefficients provide valuable insights into the reliability of responses within each safety factor, highlighting the importance of these factors in shaping respondents’ perceptions and experiences related to safety measures. The data presented response rates and ratings for PB, GB, and both groups on various safety factors in the General Site and Maintenance and Management categories. The response rates indicate how well safety measures have been followed, while the ratings evaluate the effectiveness of these responses.

4.3.2. Application appropriateness rating

4.3.2.1. General traffic conditions

In the General Site category, PB and GB responders show similar response rates, ranging from 57.3% to 82.7%. This suggests a consistent level of adherence across different safety measures. However, ratings reveal that there are some differences in perceived safety effectiveness. Although PB and GB responses are generally aligned, their evaluations have nuanced differences (Figure 7). For example, in Safety Factor GS1, GB responders give a lower rating (2.60) than PB responders (4.13), which suggests a difference in their perception of safety measures’ effectiveness in this area.

Figure 7. The perception of appropriateness of application for measures of the investigated measures of safety factors enhancing safety within the site of the project and its surrounding by respondent group.

The response rates in the Maintenance & Management category range from 33.3% to 73.3%. Even though there are differences in adherence levels, the ratings provided by both PB and GB responders are generally similar, indicating a consistent perception of safety effectiveness in this category. However, in Safety Factor MM8, there is a notable difference in ratings, with PB responders providing a lower score (2.86) than GB responders (3.44), which suggests a divergence in their views on the effectiveness of safety measures in this aspect.

The data highlight the importance of response rates and the qualitative evaluation of safety measures. Discrepancies in ratings between PB and GB responders suggest potential areas for further investigation or targeted interventions to ensure uniform understanding and adherence to safety protocols across all responders. This comprehensive analysis of both response rates and ratings provides valuable insights into different groups’ perceptions of safety measures, guiding efforts to enhance safety compliance uniformly.

On the other hand, in the Maintenance and Management safety group, both response rates and ratings differ significantly between PB and GB responders. Response rates fluctuate between 33.3% and 73.3%, indicating a wide range of adherence levels, and qualitative ratings also show discrepancies.

4.3.2.2. Within-the-site safety conditions

The data presents a detailed comparison of response rates and ratings across various safety factors within Staff Training, Traffic Operation, Loading and Unloading, and Site Administration categories for both PB and GB responders and the combined groups. In Staff Training, PB and GB responders show similar response rates, ranging from 62.7% to 73.3%. However, PB responders tend to rate safety measures higher (3.77) compared to GB responders (3.31), indicating a more positive perception of safety training effectiveness among PB responders.

In Traffic Operations, response rates vary from 57.3% to 82.7%. While PB and GB responders often share similar ratings, there are exceptions. For instance, in TO1, GB responders rate safety measures lower (3.47) than PB responders (3.87), highlighting a disparity in their views on the effectiveness of safety measures in traffic operations. In Loading and Unloading, response rates fluctuate between 54.7% and 80.0%. Interestingly, the ratings provided by PB and GB responders are almost identical (3.68), indicating a consistent perception of safety measures’ effectiveness in this area.

In Site Administration, PB and GB responders exhibit similar response rates (52.0% to 62.7%). However, in Safety Measure Item SA1, GB responders provide a lower rating (3.54) than PB responders (3.83), indicating a difference in their perception of safety effectiveness in site administration protocols. These nuanced differences highlight specific areas where targeted interventions and training programs are needed to address disparities in perception. A uniform understanding and adherence to safety protocols among all responders are crucial for fostering a safer working environment.

4.3.3. Impact of sociodemographic on the assessment

The study examined perceptions of road safety across different work zones and identified factors influencing safety. It utilized a diverse sample size to enhance the significance of observed differences. The analysis considered several sociodemographic factors such as age groups, experience levels, job roles, educational backgrounds, and genders (Table 4). The application of safety measures varies based on the site, a trend observed across all safety factors. Participants from site 4 perceive safety measures appropriateness as the highest, whereas site 2 participants have the lowest perception. This difference is significant except for maintenance and management safety factors. The assessment varies based on age groups. Young subjects (below 30 years old) rated most safety factors highly, whereas the age group of 30–39 gave almost the lowest ratings. However, the differences across all safety factors are insignificant. The assessment from the subjects’ experience perspectives revealed variations by age group across all safety factors. A noticeable trend emerged, indicating that participants with more than fifteen years of experience reported the highest ratings for appropriate application, while those with 7 to 10 years of experience provided the lowest ratings. The assessment of safety factors does not differ significantly due to the years of experience the subject has.

Table 4. Evaluation of road safety measures’ applicability in work zones, considering safety factors and subject sociodemographic variables (average of all responses for each factor)

Factors	Item and Statistical Test	Within the Site Safety Factor	General Traffic Safety Factor	Overall Safety Factor	General Site	Maintenance & Management	Staff Training	Traffic Operation	Loading & unloading	Site Administrat-ion
Site Number	1	3.826	3.584	3.776	3.835	3.456	3.83	3.902	3.00	3.607
	2	3.244	3.26	3.26	3.239	3.343	3.18	3.217	3.187	3.396
	3	3.876	3.462	3.733	3.807	3.198	3.8	3.916	4.011	3.875
	4	4.274	3.964	4.076	3.833	3.529	4.4	4.371	3.926	4.4
	5	3.498	3.34	3.416	3.13	3.372	3.31	3.468	3.677	3.541
	F-Value, Sig	3.71, <.001	2.74, 0.032	4.64, 0.002	4.58, 0.002	0.395,0.812	6.22, 0.00	5.76, <0.001	3.31, 0.013	4.48, 0.002
Age Group	<30	3.996	3.362	3.848	3.778	3.36	3.99	4.176	3.852	3.863
	30–39	3.595	3.59	3.558	3.547	3.44	3.50	3.46	3.597	3.632
	400–49	3.894	3.50	3.797	3.847	3.365	3.98	3.893	3.706	4.00
	50–70	3.718	3.67	3.675	3.638	3.464	3.94	3.789	4.034	3.95
	F-Value, Sig	1.93, 0.11	0.67, 0.618	1.68, 0.159	3.21, .016	0.04. 0.997	2.41, 0.054	3.06, 0.02	0.614. 0.64	0.79, 0.534
Experience Group	1–3	3.808	3.494	3.665	3.291	3.264	3.81	3.845	3.81	3.73
	4–6	3.653	3.304	3.536	3.515	3.079	3.46	3.734	3.683	3.5
	7–10	3.494	3.417	3.491	3.49	3.519	3.34	3.381	3.556	3.629
	11–15	3.755	3.612	3.654	3.435	3.51	3.69	3.768	3.752	3.685
	>15	3.80	3.636	3.704	3.457	3.44	3.85	3.883	3.803	3.827
	F-Value, Sig	0.51,0.769	0.54, 0.747	0.29, 0.92	0.18, 0.969	0.64, 0.669	1.055, 0.39	0.89, 0.49	0.357, 0.877	0.45, 0.814
Job on Site	Worker	2.988	2.885	2.998	2.978	3.087	2.52	3.214	3.003	4.25
	Forman	3.769	3.381	3.633	3.424	3.24	3.52	3.792	3.829	2.883
	Site Engineer	3.841	3.568	3.742	3.646	3.403	3.79	3.895	3.885	3.683
	Liaison Engineer	3.988	3.763	3.906	3.866	3.619	3.91	3.783	4.075	3.881
	Resident Engineer	3.525	3.38	3.434	3.248	3.319	3.46	3.491	3.556	4.185
	Project Director	3.73	3.565	3.626	3.439	3.454	3.68	3.771	3.675	3.657
	F-Value, Sig	2.56,0.024	1.88, 0.09	2.34, 0.037	1.57, 0.162	0.48, 0.824	2.91, 0.012	1.14, 0.435	2.61, 0.021	2.59, 0.023
Education Level	High School	2.167	2.636	2.254	2.083	2.772	2.5	2.625	1.667	1.667
	High School	3.177	3.076	3.331	3.458	3.818	2.33	3.679	3.406	3.406
	Associate	3.098	3.084	3.111	3.167	3.105	3.06	3.031	3.425	3.425
	BSc	3.751	3.538	3.665	3.55	3.415	3.7	3.758	3.781	3.781
	MSc	3.871	3.609	3.736	3.498	3.449	3.77	3.906	3.9	3.9
	PhD	3.613	3.2	3.452	3.9	2.4	4	3.75	2.95	2.95
	F-Value, Sig	2.33,0.037	0.954, 0.46	1.91, 0.086	1.12, 0.354	0.87, 0.517	1.653, 0.14	1.265, 0.28	4.09, 0.001	2.46, 0.03
Representative	Consultant	3.508	3.34	3.466	3.335	3.302	3.37	3.538	3.546	3.398
	Contractor	3.449	3.319	3.386	3.22	3.283	3.33	3.53	3.563	3.351
	Owner	3.72	3.619	3.623	3.448	3.48	3.76	3.703	3.706	3.821
	F-Value, Sig	4.65,0.004	1.53, 0.212	3.76, 0.013	3.46, 0.019	0.372, 0.773	3.08, 0.031	2.44, 0.069	2.25, 0.087	7.31, <.001
Group	PB	3.834	3.587	3.834	3.695	3.387	3.84	3.84	3.747	3.861
	GB	3.498	3.34	3.498	3.13	3.372	3.31	3.31	3.677	3.541
	t-value, Sig	2.29,0.012	1.51, 0.068	2.29, 0.012	3.45, <.001	−1.995, 0.027	2.79, 0.003	2.41, 0.009	0.47, 0.319	1.82, 0.036
Gender	Male	3.572	3.402	3.572	3.343	3.315	3.49	3.558	3.596	3.619
	Female	3.852	3.602	3.852	3.592	3.592	3.70	3.859	3.921	3.875
	t-value, Sig	−1.58, 0.059	−1.06, 0.145	−1.58, 0.059	−1.19, 0.118	−1.08, 0.143	−0.91, 0.183	−1.45, 0.076	−1.86, 0.033	−1.22, 0.112
Location	Urban	3.692	3.466	3.692	3.515	3.313	3.64	3.715	3.658	3.714
	Rural	3.8	3.642	3.8	3.329	3.718	3.57	3.796	4.021	3.816
	t-value, Sig	−0.499, 0.311	−0.88, 0.189	−0.55, 0.584	0.83, 0.204	−1.995, 0.027	0.298, 0.383	−0.53, 0.363	−1.92, 0.029	−0.45, 0.328

The ratings also vary based on the subjects’ job roles on site. Apart from site administration, the ratings consistently trend across safety factors. Workers on site consistently rate the appropriateness of application the lowest, while the liaison engineers give the highest ratings, with site engineers closely following. The ratings for Staff Training (ST), Loading and Unloading (LU), Site Administration (SA), and the general traffic and within-the-site factors significantly differ based on job roles on site but not for General Safety (GS), Materials Management (MM), and Traffic Operation (TO). The subjects’ education level appears to influence their ratings, with the lowest ratings reported by individuals with either a high school diploma or those who have not completed high school.

On the other hand, participants with post-graduate degrees gave the highest ratings, slightly surpassing those with bachelor’s degrees. However, except for the Loading and Unloading (LU) and Site Administration (SA) safety factors, the differences in their ratings are statistically insignificant. Table 4 highlights differences in assessments based on the representative side, with consistent trends across safety factors. Contract owners perceive appropriateness the highest, while contractor representatives rate it the lowest. Significant differences in ratings were noted for GS, ST, SA, measures within the site, and overall safety factors. Additionally, as mentioned earlier, the PB group’s assessment is higher than that of GB. Significant differences in ratings were observed for GS, ST, TO, within-the-site measures, and overall safety factors. Female respondents rated higher than male respondents, and subjects working in rural projects rated higher than those in urban areas. The analysis revealed no statistically significant differences across safety factors due to gender or project location.

4.4. Data reduction analysis

As previously illustrated, this study uses EFA and CFA to accommodate diverse research goals. EFA allows for exploring data structures and generating hypotheses, while CFA confirms theories and aligns models with the observed data.

4.4.1. Explanatory factor analysis

A summary of EFA showing the reduced factors’ Eigenvalues, Component Loadings, and the percentage of Variance explained for each factor is presented in Table 5. The results are based on the data irrespective of the type of respondent group.

Table 5a. Factor Analysis Displays Component Loading for each Safety Factor: General Traffic

Factor	Safety Item	Route & Stability (F1)	Control Device & Pedestrian (F2)
General Site	GS 1	0.872	0.256
	GS 2	0.805	0.276
	GS 3	0.815	0.359
	GS 4	0.817	0.354
	GS 5	0.763	0.236
	GS 6	0.846	0.181
	GS 7	0.672	0.532
	GS 8	0.233	0.861
	GS 9	0.176	0.870
	GS 10	0.270	0.884
	GS 11	0.441	0.713
	GS 12	0.412	0.666
	Eigenvalue	5.01	4.0
	Explained Variance (%)	41.77	33.32
	Cumulative Explained (%)	75.06
Maintenance and Management	Safety Item	Traffic Management (F1)	Control Device (F2)
	MM 1	0.864	0.292
	MM 2	0.809	0.367
	MM 3	0.885	0.185
	MM 4	0.931	0.091
	MM 5	0.517	0.627
	MM 6	0.911	0.174
	MM7	0.812	0.217
	MM8	0.166	0.900
	MM9	0.135	0.867
	MM10	0.156	0.841
	MM11	0.281	0.791
	Eigenvalue	4.96	3.63
	Explained Variance (%)	45.06	32.96
	Cumulative Explained (%)	78.02

The detailed breakdown of component loadings, eigenvalues, explained variance, and cumulative explained variance for safety items in different safety factors irrespective of subjects’ respondents’ group is provided in Table 5. In the General Site safety factor, two factors were extracted to describe the data. Route and Stability (F1) is characterized by strong positive loadings (ranging from 0.672 to 0.872) of General Site safety items. The second extracted factor is the “Control Device & Pedestrian factor that explains (F2)” with positive loadings (Ranging from 0.666 to 0.884). This factor accounts for 33.3% of the safety item variability, contributing to a cumulative explained variance of 75.06% for the two factors extracted. It mainly focuses on route stability and control-related variations (Table 5a).

Two factors were extracted to describe the safety items under the Maintenance and Management Safety Factor (MM 1 to MM 11), representing various safety aspects. Traffic Management Factors (F1) with strong positive loadings (ranging from 0.809 to 0.931) show significant connections with this factor, especially MM4, which has a loading of 0.931, signifying a strong relationship. This factor explains 42.9% of the variance in safety items. Control Device Factors include five safety items with positive loadings that noticeably vary (0.627–0.90) with an explanation power of 32.96%, leading to a cumulative explained variance of 75.84%, emphasizing its substantial coverage of maintenance and Management-related variability.

The extracted training factor consists of Staff Training items labeled ST 1 to ST 4. Positive loadings ranging from 0.794 to 0.922 demonstrate their strong association with this factor, especially ST 1, with a loading of 0.922, indicating a significant correlation between the factor (F1) and this safety item. This factor explains 75.68% of the variance in the observed safety items. Traffic Operations (F1) Safety Factor comprises safety items (TO 1 to TO8) reflecting traffic operations, efficiency, and control. These items exhibit strong positive loadings (ranging from 0.705 to 0.912), indicating their significance in efficient traffic operations. This factor explains 70.09% of the variance in safety items.

The Operational Control (F1) Safety Factor also comprises loading and unloading items (LU 1 to LU 5), while the Safe Operation Factors describe the safety items describing loading and unloading activities (LU6 to LU10). The positive loadings (ranging from 0.813 to 0.890 and 0.63 to 0.83 for 1st and 2nd factors, respectively) indicate their connections with these factors, especially LU 3, with a loading of 0.890, demonstrating a strong relationship. These factors explain 42.9% and 32.98% of the variance in safety items for 1st and 2nd factors, respectively, with a cumulative explained variance of 75.84%. The Site Administration Safety Factor encompasses Site Administration items (SA 1 to SA 4) focusing on PB site administration practices. These items exhibit positive loadings (ranging from 0.689 to 0.890), with SA 1 loading of 0.890, signifying a robust relationship. This factor explains 67.51% of the variance in safety items (Table 5b).

Table 5b. Factor Analysis Displays Component Loading for each Safety Factor: Within the Site

Factor	Safety Item	Staff Training (F1)	Factor	Safety Item	Operation Control (F1)	Safe Operations (F2)
Staff Training	ST 1	0.922	Loading and unloading	LU 1	0.813	0.432
	ST 2	0.896		LU 2	0.865	0.253
	ST 3	0.794		LU 3	0.890	0.280
	ST 4	0.862		LU 4	0.847	0.327
	Eigenvalue	3.03		LU 5	0.810	0.263
	Explained Variance (%)	75.68		LU 6	0.290	0.831
Traffic Operations	Safety Item	Traffic Operation (F1)		LU 7	0.622	0.613
	TO 1	0.751		LU 8	0.363	0.723
	TO 2	0.765		LU 9	0.269	0.765
	TO 3	0.896		LU 10	0.187	0.83
	TO 4	0.904		Eigenvalue	4.29	3.30
	TO 5	0.882		Explained Variance (%)	42.86	32.98
	TO 6	0.912		Cumulative Explained (%)	75.84
	TO 7	0.904
	TO 8	0.902
	Eigen value	6.0
	Explained Variance %	75.09
Site Administration	Safety Item	PB (F1)
	SA 1	0.890
	SA 2	0.876
	SA 3	0.689
	SA 4	0.816
	Eigen value	2.7
	Explained Variance (%)	67.51

Table 6a provides essential insights into the underlying structure of the data by presenting component loading values and eigenvalues and explaining variances for various safety factors across different respondent categories, namely PB and GB.

Table 6a. Explanatory Factor Analysis: Component Loading for each Factor by respondent Category and safety factor: General Traffic

Factor	Group	PB		GB
	Safety Item	Route & Stability (F1)	Control Device & Pedestrian (F2)	Control Device & Pedestrian (F1)	Control Device & Pedestrian (F1)
General Site	GS 1	0.886	0.120	0.422	0.833
	GS 2	0.844	0.267	0.295	0.851
	GS 3	0.952	0.138	0.756	0.455
	GS 4	0.960	0.145	0.772	0.428
	GS 5	0.941	0.196	0.383	0.566
	GS 6	0.910	0.237	0.169	0.872
	GS 7	0.578	0.642	0.506	0.737
	GS 8	0.111	0.951	0.659	0.417
	GS 9	0.078	0.967	0.672	0.358
	GS 10	0.065	0.950	0.870	0.311
	GS 11	0.481	0.713	0.891	0.206
	GS 12	0.478	0.659	0.759	0.223
	Eigenvalue	5.855	4.245	4.88	3.924
	Explained Variance (%)	48.8	35.38	40.62	32.7
	Cumulative Explained (%)	84.14		73.3
Maintenance and Management	Group	PB		GB
	Safety Item	Traffic Management (F1)	Control Device (F2)	Maintenance and Control (F1)	Traffic Management (F2)
	MM 1	0.861	0.165	0.776	0.54
	MM 2	0.734	0.494	0.506	0.744
	MM 3	0.954	0.09	0.502	0.735
	MM 4	0.898	-.004	0.619	0.633
	MM 5	0.348	0.797	0.861	0.311
	MM 6	0.952	0.184	0.562	0.665
	MM7	0.895	0.112	0.798	0.356
	MM8	-0.494	0.793	0.765	0.418
	MM9	-0.534	0.811	0.909	0.179
	MM10	-0.341	0.911	0.155	0.884
	MM11	-0.005	0.929	0.271	0.804
	Eigenvalue	5.38	4.03	4.70	4.09
	Explained Variance (%)	48.94	36.63	42.7	37.2
	Cumulative Explained (%)	85.57		79.9

The table delves into two key extracted factors: Route and Stability (F1), Safety Factor, and Control Device and Pedestrian Safety (F2) for the PB group, aligning with those of the GB group but in a different order. For the PB group, General Site variables (GS 1 to GS 6) exhibit robust positive loadings (ranging from 0.844 to 0.96) to the first factor and (0.642 to 0.967) to the second factor for safety items (GS 7 to GS 12), indicating their pivotal connection between observed safety items and the extracted factors related to route stability and control devices (Table 6a).

Notably, the first factor boasts a high eigenvalue of 5.855 and an explained variance of 48.80%, capturing a substantial portion of the variance in safety items. Simultaneously, the second factor, with an eigenvalue of 4.245 and an explained variance of 35.38%, contributes significantly to understanding the safety item variance. The cumulative explained variance of 84.14% highlights that Route and Stability and other factors collectively elucidate a significant part of the total variance in PB safety items. The explanation power of variation in the data extracted from the GB group perception (73.32%) is lower than that of the PB group. The component loading factors for the 1^st factor (Control Device and Pedestrian) range from 0.659 to 0.772, and the 2nd factor (Route Safety) range from 0.566 to 0.872. These values indicate moderate connections to the extracted factors, which are lower than the levels demonstrated for the PB group.

Likewise, two factors are identified for each description, but their sequence differs. In the data from the PB group, the first factor pertains to Traffic Management (F1), explaining 48.94% of the variation in the original data, whereas for the GB group, it constitutes the second factor, capturing 37.2% of the variation. Conversely, the GB control device factor (F1) accounts for 42.7% of the data, compared to 36.63% for PB (F2). The extracted factors for PB explain 85.57% of the variation, whereas for GB, it is 79.9%. Loading components range from 0.734 to 0.954 and 0.793 to 0.929 for PB’s first and second factors, indicating a strong association. For GB, these values range slightly lower at 0.765 to 0.909 and 0.633 to 0.884 for the first and second factors, respectively. Notably, the component loading variation is more minor for the control device factor and higher for the traffic management factor in the extracted data.

In the Staff Training (F1) Safety Factor, PB Staff Training variables (ST 1 to 4) exhibit robust positive loadings (ranging from 0.923 to 0.965), highlighting their substantial connection. This association is notably more robust than that observed in the GB category, where loadings range from 0.676 to 0.902, indicating significant variation. The high eigenvalues (3.57 in PB and 2.68 in GB) and explained variance (89.13% in PB and 66.7% in GB) further emphasize substantial differences between the two groups. Traffic Operation (F1) Safety Factor is marked by strong positive loadings of Traffic Operations variables (TO 1 to 8) in both PB (0.763–0.93) and GB data (0.718–0.919), indicating their significant correlation with this factor. High eigenvalues (6.46, 5.85 for PB and GB groups, respectively) and explained variances (80.76% in PB and 73.13% in GB data) highlight the central role of Traffic Management in understanding safety in both respondent categories.(Table 6b).

Table 6b. Explanatory Factor Analysis: Component Loading for each Factor by respondent Category and safety factor: Within-the-Site

Factor	Group		PB				GB
	Safety Item		Staff Training (F1)				Staff Training (F1)
Staff Training	ST 1		0.952				0.902
	ST 2		0.965				0.860
	ST 3		0.935				0.676
	ST 4		0.923				0.811
	Eigenvalue		3.57				2.68
	Explained Variance (%)		89.13				66.7
Traffic Operations			PB				GB
			Operation				Operation
	TO 1		0.763				0.756
	TO 2		0.836				0.718
	TO 3		0.925				0.885
	TO 4		0.929				0.897
	TO 5		0.963				0.846
	TO 6		0.907				0.919
	TO 7		0.919				0.907
	TO 8		0.930				0.890
	Eigen value		6.46				5.85
	Explained Variance (%)		80.76				73.13
Loading & Unloading Activities	Group	PB					GB
	Safety Item	Safe Operations (F1)			Operation Control (F2)	Safe Practice (F3)	Operation Control (F1)	Safe Operations (F2)
	LU 1	0.901			0.314	0.023	0.347	0.842
	LU 2	0.890			0.331	0.031	0.255	0.859
	LU 3	0.942			0.156	0.142	0.325	0.891
	LU 4	0.943			0.175	0.131	0.576	0.627
	LU 5	0.810			0.014	0.547	0.659	0.509
	LU 6	0.382			0.771	0.227	0.895	0.168
	LU 7	0.602			0.684	0.074	0.793	0.444
	LU 8	0.527			0.489	-0.290	0.823	0.267
	LU 9	0.121			0.287	0.885	0.776	0.422
	LU 10	-0.002			0.884	0.213	0.710	0.403
	Eigen value	4.84			2.43	1.31	4.27	3.53
	Explained Variance (%)	48.4			24.3	13.07	42.71	35.3
	Cumulative Explained (%)	85.8					78.01
Site Administration	Group			PB			GB
	Safety Item			Site Administration (F1)			Site Administration (F1)
	SA 1			0.945			0.839
	SA 2			0.924			0.853
	SA 3			0.617			0.734
	SA 4			0.765			0.846
	Eigen value			2.71			2.69
	Explained Variance (%)			67.80			67.18

In analyzing Loading and Unloading Activities, the PB group shows more robust associations with Safe Operations (F1) and Operation Control (F2) factors than the GB group. PB exhibits notably higher loadings across activities, emphasizing their significance in Safe Operations (LU 1 to LU 5) and Operation Control (LU 6 to LU 9). PB’s eigenvalues (4.84 for F1 and 2.43 for F2) and explained variances (48.4% for F1 and 24.3% for F2) surpass those of GB (4.27 for F1 and 3.53 for F2; 42.71% for F1 and 35.3% for F2), indicating PB’s more substantial influence on these factors. In both PB and GB groups, safety items SA 1 (0.945) and SA 2 (0.853) demonstrate strong associations with their respective F1 factors describing the Site Administration safety factor, indicated by high loadings (0.945, 0.924 in PB; 0.839, 0.853 in GB). While SA 3 and SA 4 exhibit moderate connections in both groups, with lower but notable loadings, the factors explain similar variances (67.80% in PB, 67.18% in GB) with comparable eigenvalues (2.71 in PB, 2.69 in GB). The PB group slightly outperforms, explaining a more significant portion of the total variance, showcasing its marginally stronger association in this context.

4.4.2. Impact of sociodemographic factors on extracted factors

The analysis of various sociodemographic factors within the given respondent groups (Site, Location, Gender, Age, Education, Experience, Representativeness, Job) has yielded crucial insights into the influence of these factors on different safety-related factors (Route and stability, Control Device & Pedestrian, Staff Training, Traffic Operation, Safe Operations, Operation Control, Site Administration, Traffic Management, Control Device). In the context of Route and stability (F1), the results indicated a significant impact of the Site factor (F = 3.08, p = 0.034), emphasizing the role of specific locations in shaping perceptions of route stability. Additionally, Staff Training (F1) demonstrated substantial differences based on multiple factors: Site (F = 3.37, p = 0.014), Experience (F = 1.84, p = 0.104), and Job (F = 3.47, p = 0.005). These findings highlight the nuanced influence of training methods on job roles and experience levels (Table 7).

Table 7. The impact of sociodemographic factors on the extracted factors: the entire data set

Factor	Site	Group	Location	Gender	Age	Education	Experience	Representativeness	Job
	F-value, sig	t-value, sig	t-value, sig	t-value, sig	F-value, sig	F-value, sig	F-value, sig	F-value, sig	F-value, sig
Route & Stability (F1)	3.08, 0.034	0.86, 0.396	−0.52, 0.605	−0.81, 0.423	0.64, 0.593	2.35, 0.043	1.50, 0.206	0.92, 0.438	1.18, 0.329
Control Device & Pedestrian (F2)	2.22, 0.095	0.24, 0.814	−0.42, 0.676	0.26, 0.795	0.26, 0.856	1.19, 0.325	0.99, 0.435	0.63, 0.598	0.65, 0.665
Staff Training (F1)	3.37, 0.014	1.33, 0.188	−0.53, 0.597	−0.40, 0.694	1.47, 0.229	1.84, 0.104	1.57, 0.179	1.65, 0.187	3.47, 0.005
Traffic Operation (F1)	0.79, 0.539	−0.46, 0.648	−1.94, 0.057	−0.94, 0.353	2.40, 0.078	0.91, 0.495	0.91, 0.483	0.16, 0.921	1.08, 0.79
Safe Operations (F1)	3.41, 0.024	−1.78, 0.093	−2.20, 0.032	−0.65, 0.520	1.48, 0.230	9.13, 0.000	0.79, 0.559	0.17, 0.918	5.84, 0.000
Operation Control (F2)	0.57, 0.638	1.10, 0.276	−1.29, 0.202	−1.23, 0.222	0.92, 0.435	2.60, 0.028	0.38, 0.863	0.22, 0.885	0.70, 0.625
Site Administration (F1)	1.63, 0.178	1.14, 0.260	−0.97, 0.336	−1.40, 0.168	0.60, 0.619	1.81, 0.112	0.14, 0.983	2.94, 0.040	2.01, 0.090
Traffic Management (F1)	5.99, 0.001	1.14, 0.246	−1.00, 0.321	−0.07, 0.943	2.60, 0.062	0.47, 0.757	2.62, 0.034	0.85, 0.472	0.996, 0.429
Control Device (F2)	15.47, 0.000	−0.51, 0.619	−1.69, 0.097	−0.96, 0.342	1.07, 0.371	0.51, 0.728	3.03, 0.018	1.46, 0.235	1.19, 0.325

Within Traffic Operation (F1), the results showed a significant impact of Representativeness (F = 2.40, p = 0.078), emphasizing the role of diverse representation in shaping perceptions of traffic operations. Notably, Traffic Management (F1) exhibited significant differences based on Site (F = 5.99, p = 0.001) and Location (F = 2.60, p = 0.062), indicating the influence of specific site contexts and locations on perceptions of traffic management practices. In the case of the Control Device (F2), the data revealed substantial disparities influenced by Site (F = 15.47, p = 0.000) and Experience (F = 1.07, p = 0.371), underscoring the critical role of site-specific conditions and experience levels in shaping perceptions related to control devices and pedestrian safety. Additionally, Safe Operations (F1) demonstrated considerable differences due to Location (F = 2.20, p = 0.032) and Site Administration (F = 2.94, p = 0.040), signifying the impact of location-specific factors and site administration practices on perceptions of safe operations (Table 7).

Several notable patterns emerged in the conducted analysis exploring the influence of sociodemographic factors on respondents’ perceptions within the PB and GB groups. The data revealed substantial disparities in safety perceptions attributed to specific factors within the PB group. Staff Training (F1) exhibited significant variance (F = 3.40, p = 0.031), emphasizing the impact of job roles on how individuals perceive training initiatives. Moreover, Safe Operations (F1) displayed considerable differences in experience levels, with a remarkably high F-value of 11.84 (p = 0.001), indicating a profound influence of experience on safety perceptions. Additionally, the Control Device (F2) perception was significantly affected by location, as denoted by the substantial F-value of 10.10 (p = 0.002), underlining the regional nuances in control device and pedestrian safety perceptions within this group. Meanwhile, in the GB group, distinct factors were found to shape safety perceptions. Safe Operations (F2) demonstrated significant variations attributed to age groups, with an F-value of 2.29 (p = 0.115), highlighting the need for age-specific safety interventions within this factor. Moreover, Site Administration (F1) was notably influenced by job roles, as evidenced by an F-value of 1.94 (p = 0.111), underscoring the specific job responsibilities’ impact on site administration perceptions (Table 8). Additionally, Traffic Management and maintenance (F1) perceptions were significantly associated with educational backgrounds, as indicated by an F-value of 2.59 (p = 0.066), emphasizing the influence of education on perceptions related to traffic management and maintenance practices within this group.

Table 8. The impact of sociodemographic factors on the extracted factors by respondent Group

Group	Reduced Factor	sociodemographic factors
		Site	Location	Gender	Age	Education	Experience	Representativeness	Job
		F-value, sig	t-value, sig	t-value, sig	F-value, sig	F-value, sig	F-value, sig	F-value, sig	F-value, sig
Project-Based	Route & Stability (F1)	2.52, 0.110		−0.96, 0.354	0.33, 0.805	1.95, 0.149	1.01, 0.450	0.25, 0.859	1.13, 0.388
	Control Device & Pedestrian (F2)	1.52, 0.247		0.85, 0.413	0.10, 0.960	0.35, 0.875	1.12, 0.395	0.51, 0.681	0.48, 0.783
	Staff Training (F1)	3.40, 0.031		−0.36, 0.725	0.70, 0.562	2.46, 0.060	1.51, 0.223	1.20, 0.326	3.24, 0.017
	Traffic Operation (F1)	0.80, 0.514		−1.76, 0.101	1.77, 0.198	0.44, 0.810	0.29, 0.908	0.09, 0.967	0.43, 0.819
	Safe Operations (F1)	0.61, 0.560		−2.03, 0.070	0.46, 0.640	11.84, 0.001	1.85, 0.191	0.30, 0.828	5.59, 0.010
	Operation Control (F2)	1.81, 0.202		−0.72, 0.489	2.33, 0.136	4.26, 0.025	0.40, 0.840	1.40, 0.291	0.29, 0.909
	Safe Practice (F1)	0.10, 0.908		−0.73, 0.481	0.92, 0.424	0.74, 0.61	0.77, 0.593	6.16, 0.009	0.48, 0.784
	Site Administration (F1)	1.54, 0.235		−1.50, 0.151	0.36, 0.782	2.18, 0.099	0.60, 0.699	1.97, 0.149	0.95, 0.469
	Traffic Management (F1)	2.38, 0.132		0.00, 0.998	1.12, 0.380	0.18, 0.906	3.58, 0.042	0.01, 0.998	1.74, 0.211
	Control Device (F2)	10.10, .002		−0.92, 0.379	1.25, 0.335	1.05, 0.407	1.29, 0.331	3.01, 0.072	0.11, 0.975
General-Based	Control Device & Pedestrian (F1)		−0.87, 0.389	−0.21, 0.832	0.51, 0.678	2.54, 0.091	0.10, 0.982	0.51, 0.604	0.61, 0.691
	Route (F2)		−0.84, 0.407	−0.35, 0.727	0.37, 0.772	0.28, 0.759	0.67, 0.615	0.20, 0.817	1.13, 0.362
	Staff Training (F1)		−1.56, 0.127	−0.19, 0.853	0.80, 0.502	1.17, 0.322	1.33, 0.276	0.70, 0.500	1.83, 0.130
	Traffic Operation (F1)		−2.03, 0.048	−0.09,0.93	1.01, 0.399	1.20, 0.311	0.82, 0.522	0.14, 0.872	2.03, 0.098
	Operation Control (F1)		−1.33, 0.190	−1.01, 0.319	1.01, 0.399	1.13, 0.334	0.40, 0.805	0.28, 0.755	2.02, 0.098
	Safe Operations (F2)		−2.15, 0.037	0.87, 0.388	1.73, 0.176	0.35, 0.709	0.49, 0.741	2.29, 0.115	0.79, 0.567
	Site Administration (F1)		−1.90, 0.065	−0.72, 0.476	0.57, 0.641	0.95, 0.397	0.32, 0.861	1.71, 0.195	1.94, 0.111
	Traffic Management & Maintenance (F1)		−1.48, 0.147	−0.99, 0.328	2.59, 0.066	0.20, 0.822	1.24, 0.312	1.27, 0.291	0.90, 0.492
	Safety Measures (F2)		−1.61, 0.114	0.53, 0.598	0.36, 0.782	2.38, 0.105	1.26, 0.304	0.10, 0.901	0.98, 0.446

4.4.3. Correlation analysis

The correlation matrix illustrates the relationships among observed safety factors in the specified groups. For the PB group, the measure perception in the general site is relatively similar for the Staff Training (ST) with a correlation coefficient of 0.698, the association with Traffic Operation (TOV), Loading and unloading (LU), and Site Administration (SA) were lower. Still, significance ranges from 0.469 to 0.521, indicating satisfactory interconnections between these factors and insignificant uncorrelated to the perception of maintenance and management factors (Table 9). Further, ST exhibits strong correlations with the perception of traffic operation (r = 0.707) and, to a lesser degree, with measures related to loading and unloading (r = 0.528), which is marginally different from those of site administrations (r = 0.522); all are significant but not association with maintenance measure (r = 0.196). The perception of traffic operation measures of this group is associated satisfactory with actions related to loading and site administration but, again, not with maintenance, which is not well-related to site administration, significant though. The GB group showed a higher correlation between the perception of measures describing different safety factors, which are all significant with a coefficient exceeding 0.728, reflecting that general practice is additional for specific practice.

Table 9. Correlation matrix for observed and extracted variables by respondent group

The extracted factors revealed robust significant positive correlations within the GB group, ranging from 0.424 to 0.851 for factors with significant associations. The average correlation in this group was 0.59, whereas for the PB group, it was 0.48, indicating a less consistent relationship. The significant correlation coefficients for the PB group, although fewer in number, ranged from 0.553 to 0.875. These findings underscore the stronger and more coherent connections in the GB group’s safety perceptions, highlighting their comprehensive understanding of safety factors compared to the PB group.

Upon analysing the relationship between the extracted factors and the observed variable, it was found that there is a perfect correlation for variables exclusively described by a single extracted factor, regardless of the respondent group. However, this ideal correlation does not apply to the variable explained by the two extracted factors. The general site’s safety factor hinges on site route stability (F1) and the presence of control devices and effective pedestrian management (F2), supported by strong correlations of 0.741 and 0.671. Similarly, loading & unloading safety relies on secure operations (F1) and efficient control mechanisms (F2), with correlations of 0.746 and 0.66. In maintenance & management, safety is linked to traffic management (F1) and control device implementation (F2), evident in correlations of 0.756 and 0.651. These findings highlight the importance of these primary factors in enhancing safety across site operations. The observed variables exhibit stronger correlations with the first principal extracted factors, consistently characterized by high eigenvalues. This pattern holds for all three safety factors and their corresponding six factors.

In comparing the observed data, both PB and GB groups displayed positive correlations for general traffic and within-site measures, ranging from 0.549 to 0.955. When examining the extracted factors, these associations increased slightly (0.744 to 0.96), indicating the robustness of these elements in explaining the observed patterns. The consistently high correlations underscore the reliability of the extracted factors across various analyses and groups. These findings provide insights into the intricate relationships among safety factors in different PB and GB contexts. The observed versus extracted data comparisons revealed consistent patterns, highlighting the complexity of interactions within safety factors and reinforcing the reliability of the extracted factors in explaining the observed data.

4.4.4. Confirmatory factor analysis

The results of the CFA for the first and second order as completed using IBM SPSS Amos 29 as presented in Table 10. The research investigated different model structures, starting with the results of the EFA, but the models were not well-fit the data. Relatively, the best fit models are still not satisfying the baseline comparisons criteria of 0.9 for the tested indices. The provided AMOS CFA results present a detailed picture of the relationships between latent variables and their respective observed items, along with various fit indices for model evaluation. The analysis was conducted using the entire dataset without distinguishing respondents’ groups. Attempts to divide the study into two groups did not result in a valid model, possibly due to limitations in sample size.

Table 10. CFA results for the 1^st and 2^nd order Models

Latent Variable	Observed/latent Variable	1^st Order Model		2^nd Order Model		Latent Variable Estimate	Observed Variable	1^st Order Model		2^nd Order Model
		Estimate	Sig (p-value)	Estimate	Sig (p-value)			Estimate	Sig (p-value)	Estimate	Sig (p-value)
General Traffic	General Site			0.958	<.0001	Staff Training	ST1	0.949	<.0001	0.953	<.0001
	Maintenance & Monitoring			0.967	<.0001		ST2	0.957	<.0001	0.962	<.0001
Within the Site	Staff Training			0.984	<.0001		ST3	0.840	<.0001	0.857	<.0001
	Traffic Operation			0.974	<.0001		ST4	0.917	<.0001	0.925	<.0001
	Loading & unloading			0.91	<.0001	Traffic Operation	TO1	0.872	<.0001	0.867	<.0001
	Site Administration			0.954	<.0001		T02	0.842	<.0001	0.839	<.0001
General Site	GS1	0.937	<.0001	0.935	<.0001		TO3	0.945	<.0001	0.942	<.0001
	GS2	0.884	<.0001	0.879	<.0001		TO4	0.951	<.0001	0.955	<.0001
	GS3	0.955	<.0001	0.952	<.0001		TO5	0.938	<.0001	0.942	<.0001
	GS4	0.935	<.0001	0.934	<.0001		TO6	0.953	<.0001	0.957	<.0001
	GS5	0.903	<.0001	0.897	<.0001		TO7	0.931	<.0001	0.936	<.0001
	GS6	0.898	<.0001	0.889	<.0001		TO8	0.938	<.0001	0.935	<.0001
	GS7	0.909	<.0001	0.899	<.0001	Loading & unloading	LU1	0.958	<.0001	0.959	<.0001
	GS11	0.899	<.0001	0.892	<.0001		LU2	0.924	<.0001	0.926	<.0001
	GS12	0.818	<.0001	0.799	<.0001		LU7	0.909	<.0001	0.785	<.0001
	MM1	0.949	<.0001	0.942	<.0001		LU3	0.770	<.0001	0.913	<.0001
Maintenance & Management	MM2	0.912	<.0001	0.901	<.0001	Site Administration	SA1	0.977	<.0001	0.978	<.0001
	MM3	0.949	<.0001	0.942	<.0001		SA2	0.937	<.0001	0.934	<.0001
	MM4	0.973	<.0001	0.968	<.0001		SA3	0.843	<.0001	0.842	<.0001
	MM5	0.854	<.0001	0.835	<.0001		SA4	0.946	<.0001	0.945	<.0001
	MM6	0.921	<.0001	0.912	<.0001
	MM7	0.831	<.0001	0.817	<.0001
Models’ Goodness of Fit
		1st Order Model		2nd Order Model				1st Order Model		2nd Order Model
CMIN	CMIN/DF	2.591		2.618		Baseline Comparisons	NFI	0.592		0.582
	Sig (p-value)	0		0			IFI	0.703		0.693
RMSEA		0.082		0.083			TLI	0.649		0.643
							CFI	0.694		0.685

4.4.4.1. 1^st order CFA model

According to the initial Confirmatory Factor Analysis (CFA) model, some of the safety variables associated with different safety factors do not meet the criteria for predicting the latent variables. This is particularly evident in the case of safety factors such as general site conditions, loading and unloading activities, and maintenance and management, each of which comprises more than ten safety variables. The coefficients of observed variables offer insights into the relationships between latent variables and their corresponding indicators within the Structural Equation Model.

In the case of the General Site (GS), observable variables (GS1 to GS12) demonstrate significant positive connections with the General Site latent variable, with coefficients ranging from 0.818 to 0.955. This indicates a strong influence of various general site conditions on the General Site latent construct. For Maintenance & Management (MM), observable variables (MM1 to MM7) exhibit substantial positive relationships, ranging from 0.831 to 0.973. This underscores the significant impact of maintenance and management activities on the MM latent variable.

Figure 8 demonstrates a strong correlation between the latent variable indicating general site conditions and those representing site operations, conditions, site management, and maintenance. Furthermore, site management and maintenance significantly correlate with site operation and condition, although the correlation is slightly weaker. On average, the correlations among latent variables describing perceptions of safety measures within the site are considerably robust but somewhat lower than their associations with variables describing general traffic conditions, though not significantly so.

Figure 8. SEM model: 1^st order model-six latent variables and thirty-six observed variables.

Model Fit Indices showed that the CMIN/DF ratio of 2.591 suggested a moderately acceptable fit of the model. However, further improvements might be needed for a better fit. The Comparative Fit Indices (NFI, RFI, IFI, TLI, CFI) ranged from 0.592 to 0.703, below the desired threshold of 0.9. These values indicated that the specified model did not fit well relative to a perfect model. The RMSEA value of 0.082 suggested marginal above the 0.08 threshold, indicating a mediocre fit but slightly better than the first-order CFA. However, further improvements could be made for a more precise.

4.4.4.2. 2^nd order CFA model

The coefficients provided in the 2nd-order CFA model offer precise quantitative insights into the relationships between latent variables and their observable counterparts. For example, General Traffic (GT), influenced by General Site and Maintenance and Management safety measure perception, exhibits robust positive connections ranging from 0.958 to 0.967. This underscores its significant impact on various site management and safety aspects.

The latent factor “Within the Site,” predicted by Staff Training, Traffic Operation, Loading and Unloading, and Site Administration latent variables, demonstrates strong associations with coefficients indicating their relationships with the 2nd order latent factor: 0.984, 0.974, 0.91, and 0.954, respectively. Notably, the connection between “General Traffic” and “Within the Site” is ideally linked, indicating a seamless relationship between these latent variables (Figure 9). The coefficients associated with General Site, ranging from 0.799 to 0.935, appear relatively lower than other constructed latent variables. In contrast, Maintenance & Management (MM) demonstrates a significant impact, with coefficients spanning from 0.817 to 0.968, highlighting MM4’s crucial role in overseeing maintenance activities

Figure 9. SEM model: 2^nd order model - eight latent variables and thirty-six observed variables.

Similarly, Staff Training (ST) exhibits notable associations, ranging from 0.857 to 0.962, with ST2 particularly influential, emphasizing its central role in shaping staff training dynamics. Traffic Operation (TO) reveals essential connections, varying from 0.839 to 0.957, where TO6 stands out, underscoring its critical importance in operational aspects. Loading & Unloading (LU) showcases substantial relationships, with coefficients ranging from 0.785 to 0.959 and LU1 being highly impactful, signifying its vital role in loading and unloading operations. Site Administration (SA) demonstrates pivotal connections, with coefficients from 0.842 to 0.978 and SA1 being central, indicating its fundamental role in administrative processes. The observed model coefficients are highly significant (p < 0.001), differing from zero.

Regarding model fit indices, the CMIN/DF ratio of 2.618 suggests a moderate fit, indicating potential areas for improvement to represent the data better. Comparative Fit Indices (NFI, RFI, IFI, TLI, CFI), ranging from 0.582 to 0.693, fall below the ideal threshold of 0.9, indicating that the model might not align well with the data. These indices imply that the specified model performs poorly compared to an ideal one. The RMSEA value of 0.083 suggests a mediocre fit; while it is generally acceptable below 0.08, it still implies room for adjustments to capture the data’s nuances better.

5. Discussion of results

5.1. Study scope

The literature review examined the safety implications of work zones and road-user interactions, with limited prior research critical success factors influencing safety and management for road construction projects (Keenan & Rostami, 2021; Reinaldo et al., 2021). This study highlights the necessity of enhancing traffic safety measures within work zones. It addresses a notable gap in road construction safety research by exploring the impact of safety measures on workers and their surrounding areas, a facet often overlooked in previous studies. Unlike prior research on supervision and management, this study uniquely evaluates their influence on the construction site and its immediate vicinity. The study relies on the safety climate concepts, dimensions, and factors introduced in the literature by Mosly and Makki (2020), Kajumulo et al. (2023), and Mohamed (2003) as its foundational framework.Delving into the perspectives of professionals and laborers enhances our comprehension of safety in road work zones, especially in Jordan. The research provides valuable insights into safety culture and climate in roadway construction projects. Socio-demographic factors such as gender, age, education, on-site role, and experience were analyzed to understand responses to various safety measures, adding depth to the findings. The study compares Project-Based Group and General-Based Group, revealing distinct profiles. The PB Group, comprising 75 cases, displays balanced gender ratios, younger age groups, and a mix of consultants, contractors, and project owners. Resident engineers are notably prevalent. In contrast, the 43-member GB Group has more males, diverse age ranges, and prominent project directors. Significant differences in age, experience, roles, representation, and project locations between the groups could impact their responses, while gender and education disparities were not statistically significant to the findings. The survey type could explain the difference; the GP was a face-to-face interview while the other was online, and filling in the form was a matter of choice. It was challenging to keep the balance, although many attempts were made to address different target groups.

5.2. Safety climate perception

The analysis meticulously dissects safety responses from two distinct groups, PB and GB, within construction sites, revealing subtle yet significant disparities in their perceptions. Across various contexts of safety climate factor, such as General Traffic Conditions, Maintenance, and Management; Within-the-Site Safety training, Traffic Operations, Loading and Unloading, and Site Administration, both groups demonstrate comparable response rates, ranging from 57.3% to 82.7%. However, distinctions arise when evaluating the effectiveness of safety measures within these categories. GB responders consistently demonstrate higher adherence rates in general traffic and site-related tasks. They show a superior commitment to safety measures when subjects state what they practice and indicate lower adherence when they reflect on what they do. The highlights a response rate of 65.74% for General Traffic conditions and 61.51% for Within-the-Site needs. Additionally, the PB group rated General Traffic safety at 3.54, while the GB group at 3.22. Within-the-site safety measures appropriate for application were rated at 3.55 by the PB group and 3.51 by the GB group. The average ratings increased to 3.40 for General Traffic and 3.53 for Within-the-Site conditions. These figures signify a balanced evaluation between the two groups, indicating a collective perception that influences construction site safety culture.

The ratings provide valuable insights into safety perceptions, regardless of the respondent group. General Site was rated at 3.43, indicating satisfactory safety levels, while Staff Training scored slightly higher at 3.55, reflecting positive training perceptions. Traffic Operation received a rating of 3.69, indicating a decent safety standard in traffic-related areas. Loading and Unloading, as well as Site Administration, both scored 3.68 and 3.69, respectively, indicating consistent safety perceptions. However, Maintenance and Management received a lower rating of 3.38, suggesting room for improvement. Specifically, measures addressing general traffic safety need enhancement to ensure a higher level of safety in this area. Sarireh and Tarawneh (2013) argue that companies with extensive experience are likely to possess enhanced safety protocols. They stress the vital importance of safety in mitigating expenses, preventing delays, and reducing injuries.

Cronbach Alpha coefficients confirm the reliability of responses within each safety factor. Discrepancies in ratings between PB and GB responders point to specific areas, such as general site activities and maintenance tasks, where targeted interventions are needed. For example, general Site had a 0.60 difference, highlighting varied safety understandings. Staff Training showed a 0.46 difference, indicating divergent views on training effectiveness. Traffic Operation displayed a 0.40 difference, emphasizing differing opinions on operational safety. Loading and unloading showed no difference, suggesting unanimous perceptions. Site Administration had a 0.29 difference, indicating minor disparities. Maintenance and Management showed a 0.02 difference, indicating relatively consistent understanding. These differences underscore intervention needs and emphasize the necessity for standardized safety measures across construction sites.

Investigating the differences between the ratings at measure level, in general, were similar, with some distinct differences. For example, for General Traffic Conditions, GB responders rate safety measures significantly lower (2.60) regarding the provision of suitable routes for traffic and pedestrians compared to PB counterparts (4.13), highlighting distinct beliefs about safety protocol efficacy. Similarly, in Maintenance and Management, a noticeable gap emerges in MM8, addressing whether traffic route marking is correctly maintained according to TMP, where PB scores lower (2.86) than GB (3.44), indicating varied perspectives on the effectiveness of Maintenance and Management-related safety measures.

Within-the-site safety Conditions reveal a contrast in Staff Training, where PB respondents rank safety higher (3.77) than GB participants (3.31). In Traffic Operations (TO1, assessing the provision of appropriate transport vehicles), GB responders assign a lower rating (3.47) than PB participants (3.87), indicating differing views on the effectiveness of Traffic Operations-related safety measures. However, both groups exhibit consistent rating, suggesting shared perceptions in Loading and Unloading conditions. The reflections on training show some differences but are still not very distinctive; the responses were slightly different when addressing induction and continuous staff training. In Site Administration, the safety measure SA1, evaluating whether workplace traffic rules are documented and distributed), PB respondents’ rates are higher (3.83) than GB counterparts (3.54), pinpointing specific areas requiring targeted interventions. The distinction in the responses is related to adherence to some measure describing the organization of the work-one site, operations, and procedures, and to some extent, training and education, reflecting different perceptions of safety. Analyzing adherence rates and qualitative evaluations across diverse safety groups is essential for establishing consistent understanding of safety protocols among workers, bridging the gap between stated intentions and actions. This insight allows tailored interventions and training programs to enforce standardized safety measures, ultimately improving overall safety compliance in the workplace. The literature underscores the critical importance of effective communication, standardized safety protocols, and positive safety climates in enhancing construction site safety and workers’ behaviours (Al-Bayati, 2021; Khoshnava et al., 2020; Sanni-Anibire et al., 2018; Zhang et al., 2020)

The study explores perceptions of road safety in diverse work zones, considering sociodemographic factors like age, experience, job roles, education, and gender Statistical analyses highlight the impact of demographics and professional factors on responders’ perceptions of safety measures. Safety measure applications vary across sites, with site 4 participants perceiving the highest appropriateness, while site 2 participants have the lowest perception, except for maintenance and management factors. Younger subjects rated safety factors highly, while the 30–39 age group gave lower ratings. Experience-wise, those with over fifteen years rated the application highly, whereas those with 7–10 years rated it lower. Job roles influence ratings, with workers consistently rating applications the lowest while liaison engineers rate them the highest. Education affects ratings, post-graduates’ rate higher than high school graduates. The representative side influences perceptions, with contract owners rating highest and contractor representatives lowest. Although GB rated adherence to safety measures, the assessments conducted by the PB group were significantly better than those of GB. It is worth noting that the PB group’s evaluation of General Site Safety, Staff Training, Traffic Operations, within-the-site measures, and overall safety factors were significantly different from GB’s evaluations, while other factors showed no such distinctions. No significant differences emerged due to gender or project location. The research emphasizes diverse perceptions influenced by sociodemographic factors, revealing the intricate relationship between safety assessments and individual traits, challenging the 1st hypothesis.

5.3. Exploratory factor analysis

The analysis uncovers vital insights through an Exploratory Factor Analysis (EFA) that yielded nine extracted factors. Two factors were identified for safety aspects involving more than 10 observed variables, while an additional factor was extracted to explain factors with fewer variables, offering a comprehensive understanding of the underlying patterns in the data. Two factors, “Route and Stability” and “Control Device and Pedestrian,” emerged in the General Site safety category, explaining 75.06% of the variability. Maintenance and Management Safety Factor displayed two factors, explaining 75.84% of the variance, offering insights into Traffic Management and Control Device Maintenance Factors. The Staff Training factor demonstrated strong associations (75.68% of the variance). Traffic Operations Safety Factor (70.09% variance) and Operational Control Safety Factor (75.84% cumulative explained variance) provided essential insights into traffic operations and loading/unloading activities. The explanation power is satisfactory for all safety factors. The aim was to explore the data to provide better insight into what it tells. Sociodemographic factors such as age, job roles, education, and representative side influenced perceptions. This further confirms the invalidity of the second hypothesis formulated in this study. These nuanced findings emphasize the need for tailored interventions to promote safer work environments based on diverse safety perceptions.

The analysis highlighted several crucial factors influencing safety perceptions within construction site environments. These factors include specific site locations significantly shaping Route and Stability (GS-F1) perceptions, emphasizing the impact of location-specific conditions on safety evaluations. Staff Training perceptions varied based on Site, Experience, and Job, showcasing nuanced effects of training methods, experience levels, and job roles on safety views. Traffic Operation perceptions were influenced by Representativeness, underlining the role of diverse representation in shaping safety perspectives. Traffic Management (MM-F1) was significantly affected by Site ot contract and project Location, indicating site-specific and location-related influences on safety perceptions. Control Device factor (MM-F2) perceptions were shaped by Site and marginally by Experience, emphasizing the impact of site-specific conditions and experience levels on safety evaluations. Safe Operations varied significantly due to Location and Site Administration, indicating the influence of location-specific conditions and site administration practices on safety perceptions. These factors provide valuable insights into the complexities of safety perceptions within the construction industry, guiding tailored interventions and strategies to enhance safety culture and practices on construction sites.

5.4. Correlation analysis

The analysis compared safety factor correlations in construction site respondents (PB and GB groups). The PB group showed moderate correlations between different observed factors (0.469 to 0.707), indicating satisfactory safety connections with weaker maintenance links. In contrast, the GB group exhibited stronger correlations (0.424 to 0.755) among extracted factors, indicating a better safety understanding (average correlation of 0.59) compared to the PB group (average correlation of 0.48). This highlights the need for focused interventions to enhance safety awareness in the PB group. The analysis revealed perfect correlations for variables explained by a single extracted factor, irrespective of respondent group. General site safety relied on route stability (F1) and control devices/pedestrian management (F2), with strong correlations (0.741 and 0.671). Loading & unloading safety depended on secure operations (F1) and control mechanisms (F2), correlating at 0.746 and 0.66. Maintenance & management linked to traffic management (F1) and control device implementation (F2), with correlations at 0.756 and 0.651. These findings highlight the vital role of these factors in enhancing site safety, supported by robust observed variables correlations with the first principal extracted factors. Both PB and GB groups showed positive correlations in observed data (0.625 to 0.955) for general traffic and within-site measures. Extracted factors strengthened these associations slightly (0.923 to 0.96), confirming their robustness in explaining patterns. High and consistent correlations underscored the reliability of these factors across analyses and groups, falsifying the third hypothesis of this study.

5.5. Confirmatory factor analysis

The literature showed that SEM models were developed mainly to explores the pivotal role of safety culture and attitudes in shaping construction safety behaviors among workers (Al Faqeeh et al., 2019; Newaz et al., 2019; Safeera & Bhavya, 2020; Seo et al., 2015). This work, include in addition to the behaviour, the management of work zone site and safety procedures as part of the modelling. The EFA analysis was the foundation for the 1^st order Confirmatory Factor Analysis (CFA). However, the CFA model did not confirm the grouping identified in the EFA concerning the relationship between observed and latent variables. For instance, only four variables related to loading and unloading safety factors were included in the model. At the same time, the rest were omitted due to their insignificant contribution to the latent variables. The same situation occurred in maintenance and management, where variables related to control device maintenance had minimal impact on the latent variable and were consequently excluded from the model. Numerous model structures were explored, and the results did not satisfy all the requirements of a well-fit model. The model revealed significant positive connections between observed and latent variables, and CMIN/DF was within the acceptable level for the good-fit model. However, the Comparative Fit Indices (NFI, RFI, IFI, TLI, CFI) ranged from 0.592 to 0.703, below the desired threshold of 0.9, indicating a less-than-ideal fit. The RMSEA value was 0.082, suggesting a mediocre fit but slightly better than the first-order CFA.

In the 2^nd order CFA model, General Traffic (GT) exhibited strong positive connections with various general traffic and within-site safety factors, confirming all previous findings on such an association and the validity of the alternative fourth hypothesis. The model fit indices, including NFI, RFI, IFI, TLI, and CFI, ranged from 0.582 to 0.693, below the desired threshold of 0.9, indicating a moderate fit and it was less than the 1^st order model as explained in the literature. The RMSEA value was 0.083, indicating a mediocre fit. These results underscore further model refinement necessary to better align with the data.

To improve the Structural Equation Model (SEM) fit, there is a need to reconsider the theoretical structure and sample size. Theoretical re-evaluation involves ensuring the model aligns with the underlying theory, simplifying complexities, exploring alternative pathways, and consulting experts and literature for insights. Simultaneously, increasing the sample size enhances stability and precision, reducing the random variability impact. A larger sample enhances statistical power, making genuine relationships easier to detect and ensuring the model better represents the population. Combining theoretical refinement with a substantial sample size is key to achieving a more accurate and reliable SEM.

To conclude, the study reveals the complex safety perceptions within construction sites, emphasizing the significance of sociodemographic factors and customized approaches to improve safety compliance. The robust correlations and reliable factors identified indicate their potential for enhancing future safety models. However, challenges in model confirmation persist, underscoring the necessity for further research and refinement in construction site safety.

6. Conclusions

• The analysis revealed nuanced differences in safety climate factor perceptions between PB and GB responders. Traffic Operation scored highest at 3.69, indicating decent safety standards, while Maintenance and Management rated lowest at 3.38, signalling areas for improvement. General Site received a rating of 3.43, showing marginal satisfaction and slight disparity with the lowest-rated aspects and slightly lower than Staff Training (3.55).

• Sociodemographic factors significantly influenced safety perceptions, including age, experience, job roles, education, and representative sides. These differences pinpoint specific areas in construction sites needing targeted interventions. Tailored strategies considering these factors should be implemented for uniform safety adherence.

• In the correlation analysis, the GB group demonstrated stronger positive correlations (0.424 to 0.851) among extracted factors, indicating a better understanding of safety factors compared to PB group (average correlation 0.48). The comparison of observed and extracted data underscored complex interactions within safety factors. Positive correlations (0.625 to 0.955) between general traffic and within-site factors were consistent for both groups, reflecting shared understanding. Extracted factors exhibited robust associations (0.923 to 0.96), confirming their reliability in explaining patterns.

• The Confirmatory Factor Analysis (CFA) faced challenges in confirming EFA groupings, with certain variables being excluded due to their insignificant contributions. Modes were developed shows strong association between the observed variables and the latent variables. However, some validation criteria were not fulfilled. Further research with expanded sample size and varied model structures is essential.

Ethical Statement

Verbal informed consent was obtained from all subjects before participating in this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the I personally funded my work.

Notes on contributors

Lina Shbeeb

Lina I. Shbeeb -Associate Professor in the civil engineering department at Hussein Technical University - was born in Nablus, Palestine, in 1962. She received a B.Sc. degree in civil engineering from the University of Jordan (1985) and an M.Sc. in traffic and highway engineering from the University of Jordan (1993). She had a Ph.D. in traffic planning and engineering from Lund University, Sweden (2000). She is a senior traffic and transport engineer with over 38 years of experience in traffic and transportation engineering, serving in public and private sectors, including 23 years in academia. She served as minister of transport (2013-2015) in the Jordan government and a dean of engineering at Al-Ahlyya Amman University (2017-2018) and Dean of basic and social sciences at Hussein Technical University (2019-2022). Research interests include traffic safety, traffic analysis and operation, pavement management, public transport planning and operation, transportation economics, and intelligent transportation systems. An active member in professional institutes. She is the co-founder and the CEO of Enrich, an engineering consultancy.