Safety Risks Analysis: Moderating Effect of Risk Level on Mitigation Measures Using PLS-SEM Technique

Abstract

The Malaysian construction sector registers higher fatal accidents than the manufacturing sector even though the latter has the highest cases of accidents. There is a need to implement effective safety risk management. The main objective of this study is to explore the moderating effect of risk level of accidents on mitigation measures implemented. For this purpose, the factors causing safety risks and the practical measures taken by contractors to mitigate these risks were identified, in addition to the operationalization of the likelihood and severity of accidents using suitable rating scales. Descriptive analysis shows that a fall-related accident is the most likely and the most severe safety risk at high risk level. Results from multivariate analysis using SmartPLS 4 show that safety risks have a significant positive relationship with mitigation measures, and risk level actually heightens this relationship. As a result, the practical measures implemented on construction sites to mitigate the impacts of accidents may be inadequate unless the moderating effect of risk level is considered during the planning, design, and management of construction safety. Therefore, mitigation measures taken by the contractors must take into account the types of factors causing safety risks, as well as the likelihood and severity of these factors.

1. Introduction

According to [1], risk is defined as the “combination of the likelihood of occurrence of a work-related hazardous event or exposure(s) and the severity of injury and ill-health that can be caused by the event or exposures”. In addition, [2] defines “a risk as the potential of a situation or event to impact on the achievement of specific objectives”. Risks are found in any business undertaking. As a result, incidents are bound to happen in any occupation of any sector which affects its smooth operation.

The construction industry is well-known for its complexity, dynamic nature, uniqueness, and diverse environments which could create uncertainty and challenges because the works involved nowadays could be high in the sky, deep underground, below water level, or across the sea which often involve adverse surroundings and situations. According to [3], risk may appear in any form and at any stage of the construction process. A construction site is thus full of hazards due to many people working in various activities and the use of heavy materials and moving machineries. Hence, the construction industry is highly prone to various factors which could cause safety risks. Therefore, implementing safety risk management in the industry is essential in anticipation of the unpredictable nature of safety risks with the objectives to mitigate or manage their impacts. Table 1 summarises the occupational accident statistics released by [4]. Among the industrial sectors listed, the construction sector registers the highest number of fatal accidents with an average of 89 cases per year over seven years compared to the manufacturing sector, even though the latter has the highest rate of accidents with an average of 3355 cases/year from 2015–2021, which is almost 14 times higher than the construction industry.

Table 1. Occupational accident statistics in Malaysia from 2015–2021.

Industrial Sector	Type	2015	2016	2017	2018	2019	2020	2021	Total from 2015–2021	Average over 7 Years
										Type (per Year)	Sector (per Year)
Manufacturing	Death	46	68	68	62	73	73	48	438	62.6	3355
	NPD	1906	2173	1985	2969	4661	4202	4015	21911	3130.1
	PD	89	74	125	197	214	231	206	1136	162.3
Mining and Quarrying	Death	4	4	8	4	5	3	8	36	5.1	43.6
	NPD	32	20	37	34	52	35	44	254	36.3
	PD	3	0	1	3	3	1	4	15	2.1
Construction	Death	88	91	111	118	84	66	65	623	89.0	240.0
	NPD	138	126	123	106	227	137	147	1004	143.4
	PD	11	5	6	8	15	3	5	53	7.6
Agriculture, Forestry, Logging and Fishery	Death	31	23	23	26	43	43	16	205	29.3	763.7
	NPD	440	435	488	709	1111	916	939	5038	719.7
	PD	9	9	11	14	22	20	18	103	14.7
Utility (Electricity, Gas, Water and Sanitary Services)	Death	6	2	10	5	9	3	8	43	6.1	161.7
	NPD	86	68	90	168	245	214	198	1069	152.7
	PD	4	4	4	NA	4	3	1	20	2.8
Transport, Storage and Communication	Death	22	12	16	12	21	11	6	100	14.3	215.6
	NPD	107	113	105	124	359	294	281	1383	197.6
	PD	2	2	1	1	9	6	5	26	3.7
Wholesale and Retail Trade	Death	3	0	10	1	0	1	2	17	2.4	113.0
	NPD	102	107	86	69	85	126	182	757	108.1
	PD	3	4	1	3	2	1	3	17	2.4
Hotel and Restaurants	Death	0	3	3	1	5	2	0	14	2.0	127.1
	NPD	62	85	110	120	227	137	125	866	123.7
	PD	0	2	1	2	3	1	1	10	1.4
Financial, Insurance, Real Estate and Business Services	Death	14	14	16	22	16	8	17	107	15.3	232.3
	NPD	105	101	124	190	384	312	264	1480	211.4
	PD	0	11	6	5	6	7	4	39	5.6
Public Services and Statutory Bodies/Authorities	Death	0	6	2	9	3	3	4	27	3.9	73.7
	NPD	31	101	64	48	93	73	68	478	68.3
	PD	1	3	0	1	3	1	2	11	1.6
All industrial sectors combined	Death	214	223	267	260	259	213	174	1610	230.0	5322.9
	NPD	3009	3329	3212	4537	7444	6446	6263	34240	4891.4
	PD	122	114	156	234	281	274	249	1430	204.3

Note: NPD = Non-Permanent Disabilities; PD = Permanent Disabilities; NA = Not Available.

Common safety risks in the construction industry, such as fall from height, being struck by a moving object, or workers being buried in a landslide can be significantly reduced if not eliminated by introducing safety management. Therefore, it should be taken as a critical element for creating value and thus increasing a project’s overall performance in terms of time, quality, and cost [5]. Realistically, achieving comprehensive and effective safety management is a challenge to all project managers because they must anticipate the risks that may occur and the resulting consequences. However, it has been found that the level of safety management practices in Malaysia construction companies is relatively low because of the lack of knowledge and understanding on the subject [5]. According to [6], most construction companies in Malaysia fail to implement a systematic process of risk management. It has been found that safety risks could be accepted, transferred, and mitigated by implementing a systematic process in safety management.

To address the lack of knowledge and understanding on safety management practices, this study set out to investigate the relationship between safety risks (SR) and the mitigation measures (MM) implemented to mitigate the impacts of these risks, as shown in Figure 1. It is hypothesised that there is a positive relationship between SR and MM. To achieve this purpose, this study aims to: (a) determine the likelihood and severity of commonly-occurred fatal construction accidents; (b) identify the factors causing safety risks; (c) identify the practical measures taken by contractors to mitigate or manage these safety risks; and (d) explore the moderating effect of likelihood and severity, in terms of risk level, on the mitigation measures implemented using the PLS-SEM technique.

Figure 1. Path model on the relationship between safety risks and mitigation measures.

Likelihood and severity are expected to affect the relationship depicted in Figure 1. Inherent in any risk are the likelihood of an accident to happen and the severity of its impact when it happens. According to [7], risk increases when the probability of an incident occurring increases or the severity of injury increases. The more likely it is for an accident to occur, and the more severe the accident, the higher the risk level. Hence, these twin characteristics of risks, working hand in hand, will determine the risk level of an accident which is given by “risk level = likelihood × severity”. The moderating effect of risk level on the relationship can be quantified by examining the R-squared value and path coefficient between these two constructs when risk level acts as a moderator between safety risks and mitigation measures.

2. Literature Review

Any industry which wants to succeed must operate safely, dependably, and on a long term basis [8]. Risks that have not been identified and managed will undoubtedly threaten a project’s objectives, resulting in high cost and schedule overruns [9]. To accomplish this goal, the industry must first identify the dangers and assess the risks connected with them. If an industry could identify and categorise risks before the commencement of a project, they would be able to improve risk management and avoid any potential losses.

2.1. Commonly-Occurred Fatal Construction Accidents

Ref. [10] conducted a study in the United States based on the OSHA fatalities data from 1980, 1985, and 1990. They concluded that fall-related, struck-by, electrocution, and being caught in-between are the most common forms of accidents. Table 2 shows the statistics from 2016 to 2020 on the various types of fatal construction accidents provided by [11], where nearly 50% of the fatal accidents that happened were due to workers falling from height. Other studies have also revealed that fall-related accidents are the most common fatal construction accidents [12,13,14,15], including in China [16].

Table 2. Commonly-occurred fatal construction accidents.

Item	Type of Accident	Period Unavailable in [12]		From 2010–2015 in [13]		From 2013–2018 in [14]		From 2010–2018 in [15]		From 2016–2020 for this Study [11]
		Cases	Percent	Cases	Percent	Cases	Percent	Cases	Percent	Cases	Percent
1	Falling from height	17	56.7%	56	43.4%	63	43.4%	304	38.2%	40	48.8%
2	Struck-by accident (e.g., moving object, moving vehicle, or by falling object)	4 + 2	20.0%	33	25.6%	49	33.8%	242	30.4%	25	30.5%
3	Fall into opening or drowning	2	6.7%	6	4.7%	8	5.5%	78	9.8%	5	6.1%
4	Buried	2	6.7%	8	6.2%	---	---	---	---	5	6.1%
5	Electrocution	3	10.0%	7	5.4%	7	4.8%	22	2.8%	3	3.7%
6	Road accident	---	---	---	---	---	---	---	---	1	1.2%
7	Caught in between	---	---	17	13.2%	11	7.6%	141	17.7%	1	1.2%
8	Fire or explosion	---	---	1	0.8%	2	1.4%	3	0.4%	1	1.2%
9	Insect pest	---	---	---	---	---	---	---	---	1	1.2%
10	Exposure to, or contact with, harmful substances	---	---	1	0.8%	1	0.7%	6	0.8%	---	---
11	Environmental factors	---	---	---	---	4	2.8%	---	---	---	---
Total		30	100%	129	100%	145	100%	796	100%	82	100%

2.1.1. Fall-Related Accident

Fall-related accidents are the most common type of safety risk not only in Malaysia but also in many other countries such as the United States, China, the United Kingdom, Spain, Korea, Singapore and Taiwan [15]. When compared to other forms of safety risks in the construction industry, fall-related accidents are believed to have the highest frequency of occurrence [17]. Any object that might cause a person to lose their balance and fall is considered to be a danger while working four feet or more above the ground. The majority of workplace accidents involve falling from a working platform, scaffolding, ladder, or structure. As a result, falls from height are still much more common in construction accidents than in other kinds of accidents [18].

2.1.2. Struck-by Accident

Being struck by any objects or equipment is known to be one of the factors that led to fatal injuries and deaths in the Malaysian construction industry from 2010 to 2018 [15]. A struck-by accident happens when a worker encounters any moving, dropping, or rolling material or object forcibly [19]. It shall also include incidents where the workers on-site or in public get hit by any falling material, moving vehicle, or machinery [20].

(a)

Struck by a Swinging or Slipping Object

When materials are mechanically raised, there is a possibility that they may swing and harm the employees below. As the weight is lifted, the materials may swing, twist, or spin in their respective positions. This movement has the potential to catch employees off guard, and they may be struck by the swinging load. Windy circumstances are particularly dangerous since the weight will swing more widely. If the worker is hit from behind and falls to another level, the worker may receive even more severe injuries. This is dependent on where the worker is positioned and the power behind the weight [21].

(b)

Struck by a Rolling Object

When an object is rolling, moving, or sliding on the same level as the worker, this is referred to as being struck by a rolling object. Incidents when the worker is hit or run over by a moving vehicle without being trapped beneath it, as well as incidents where the worker is struck by a sliding item or piece of equipment on the same level, are included under this category [22].

(c)

Struck by Falling Object

Injuries sustained as a result of being struck by a falling object or equipment occurred when the source of the injury is falling from a higher to a lower level. This includes instances in which the injured person is crushed, pinned, or caught under an object falling from a higher to a lower level [22].

2.1.3. Drowning and Asphyxiation

Drowning is considered as the world’s third highest factor causing fatal injury or death [15]. Drowning occurs when a person dies as a result of suffocation caused by a liquid that limits or blocks oxygen intake into the human body from the air, resulting in asphyxia [12]. Asphyxiation, on the other hand, is a situation comparable to drowning in which insufficient oxygen occurs in the human body as a result of poor breathing as a result of working in a confined space or drowning [23].

2.1.4. Buried

Accidents may happen when construction workers are found buried due to cave-in or collapse of earth during or after excavation work [24]. The author of [25] reported the occurrence of a gruesome work accident which led to the death of a construction worker after he was buried alive under a landslide.

2.1.5. Electrocution

Generally, an electrical hazard refers to the risk of getting burned, electrocution, shock, arc flash, or other injury due to exposure to a lethal amount of electrical energy. Burns could be defined as injuries due to contact or exposure to electricity, arc flash, or thermal contact, while shock often results when the human body reflex responds to the passage of electric current [26].

2.1.6. Road Accident

Road accident is one of the safety risks in the Malaysian construction industry. According to the Department of Occupational Safety and Health [11], a truck driver died in a road accident due to a malfunctioning blinker at a sharp bend. Road accidents could also happen due to the vehicle’s brake failure and hydroplaning. “Increasing number of highway construction zones” in highway construction projects have disrupted regular traffic flows which could cause traffic safety problems and accidents [27].

2.1.7. Caught in-between Accidents

Caught in-between accidents occurred when two or more objects or components of an object are caught, squeezed, compressed, crushed, or pinched between one’s body [13]. There are times when a construction worker is too focused on their own tasks and fails to see caught in-between hazards, such as standing between a heavy machine, such as a trailer and a forklift, or an immovable structure, such as a brick wall [28]. According to [29], incidents involving being squashed or crushed between rolling, sliding, or shifting things are also regarded as one of the most common forms of accidents in the construction industry.

2.1.8. Fire or Explosion

The potential danger of fire outbreak is particularly severe on many construction sites, especially during those high-risk activities such as hot work that generates heat, sparks or flame, or even overheating of the plants and equipment [30]. In fact, fire would easily break out with the presence of sufficient oxygen, fuel, and a source of ignition arising from hot work, overheating plant and equipment, smoking, faulty electrical installation, bonfires or arson [31].

The occurrence of explosions in construction sites, in fact, is not so frequent, but such risks will lead to significant consequences: not only defects on the structure but also the potential loss of a worker’s life. There was a case of explosion in 2017 at a Malaysian MRT construction site caused by an old bomb from the Second World War, which resulted in the death of one construction worker while another two were critically injured [32].

2.1.9. Insect Pest

According to [11], there was one worker who died after been stung by hornets at the Sarawak construction site. The employer was required to conduct HIRARC to identify such a risk and provide risk control measures such as destroying the honeycomb to prevent the safety risk from happening.

2.2. Factors Influencing Safety Risks

Accidents may happen on a construction site due to many reasons [33]. There are many heavy plants, heavy materials, rough terrains, and people working at high places. As a result, a construction site is a high-risk place.

2.2.1. Human Errors

Human errors, no matter how minor, may occasionally have a domino effect, resulting in enormous economic or life loss [34]. Human errors are often related with improper attitude, inadequate tools used, body effort and lack of experience [35]. Human errors are considered to be the main cause of fall-related accidents. The contributing factors to fall-related accidents include human errors and inappropriate use of a control [36]. Workers’ negligence in judgement accounts for approximately one-third of the fall accidents [37].

2.2.2. Failure to Use Personal Protective Equipment (PPE)

Every year, a large number of construction workers are killed or seriously harmed due to the improper usage and wearing of personal protective equipment (PPE) [38]. According to statistics from throughout the globe, 2 million individuals are predicted to be disabled each year as a result of work-related accidents, with 25% or more of those injuries occurring to the head, eyes, hands, and feet [39]. This is due to a lack of knowledge and use of safety equipment, such as hard safety helmets, which are only worn by 16% of those who have had occupational head injuries [40]. In addition, 23% of employees who had worn safety boots suffered from foot injuries. Moreover, 40 percent of those who had suffered from eye injuries had worn eye protection [39]. According to statistics, although there is no assurance that personal safety equipment can prevent incidents resulting in injuries from occurring, it may at least minimise the likelihood of such an incident occurring [41]. Ref. [42] believed that precise safety applications may help to minimise construction site accidents, as well as production costs, productivity development and profitability. Most significantly, he added, lives could be saved.

2.2.3. Unsafe Act and Site Condition

The major fundamental factors of accident cases are unsafe acts and site circumstances [43]. In total, 99% of construction safety risks are caused by either risky conduct or unsafe conditions, or both of these factors together [44]. These are regarded as the primary causes of all forms of construction safety risks. Unsafe activities are defined as the misuse of safety procedures, which increases the likelihood of an accident occurring on the construction site [13]. An unsafe site condition is a physical condition or environment that is surrounded by possible risks and might be the cause of a site accident [19]. The dangerous act mostly deals with hazardous equipment or unsafe methods, such as working without safety devices, equipment failure, inappropriate work process, worker knowledge level, and failure to follow work procedures [19]. Unsafe circumstances, on the other hand, include missing or inadequate guardrails on platforms, malfunctioning tools and equipment, fire dangers, a bad fire alarm system, a lack of housekeeping, poor climatic conditions, excessive noise, and insufficient light to operate.

2.2.4. Lack of Progressive Training

To prevent safety risks on construction sites, proper training is required. Safety risks sometimes occur when employers fail to provide sufficient training and knowledge on how to carry out the job. One of the problems in safety practices is the lack of budget allocation on safety management. The employers and workers need to attend safety training to improve their skills and enhance their safety awareness. However, the cost for attending the training course is high. Therefore, the company needs to allocate more budgets on safety to provide safety equipment, training, and other measures to enhance the safety awareness of the construction workers [45]. Safety training is a method of improving construction workers’ safety that focuses on the efficacy of the instructional delivery method. Effectiveness is connected to the level of understanding of instruction and may be enhanced by improving the instructional delivery method [46].

According to [47], most of the larger companies subcontract most of their work, which results in a lack of workforce development and training. However, safety risks may occur at any time. Employees bear the danger of being hurt while doing their jobs. A substantial amount of responsibility is placed on the skilled construction worker. As a consequence, the construction worker must be exceptionally brilliant and well-trained. Adequate safety training assists in improving proficiency and lowering the occurrence of safety risks [46]. In summary, employees involved in high-risk activity must have access to training content at all times.

2.2.5. Poor Communication

The term “communication” refers to the act of sending and receiving information from one person to another in a way that both parties can understand [48]. Some common poor communication examples on construction sites include language barrier, miscommunication and misunderstanding, and failure in conveying message. The construction industry relies heavily on communication, and there is a need for every firm or professional to get their messages through. Construction communication has gotten more difficult since the number of parties engaged has increased substantially, including developers, subcontractors, investors, members of the general public, and government organisations participating in the process [49]. In the construction sector, bad communication may occur on a big or small scale. In large-scale cases of bad communication, disagreements between construction partners lead to project failure, while small-scale cases of poor communication inside the company lead to delays, injuries, accidents and blunders [50]. A lack of project information, such as lack of timely information, poor project documentation, inaccessibility of project information, and unavailability of crucial information, could lead to performance deficiency and unproductive practitioners [51].

Most of the construction workers in Malaysia are from different ethnic backgrounds, as well as from different countries such as Indonesia, Myanmar, Thailand, Vietnam and Bangladesh. The majority of them do not speak or comprehend the language of the locals. This has made it difficult to communicate with each other. Messages may not be sent or received in a timely manner, which might lead to an increase in the number of deaths and injuries on the construction site [52]. For instance, there is a case in Malaysia where the workers were unable to speak English and their employer had to translate all of the information concerning the construction projects. Although the scaffolding at the building site was partly removed, the employer neglected to inform the workers that it could not be used due of the scaffolding’s dismantled status [53].

2.3. Practical Measures Taken by Contractors

When a risk event is identified and assessed, a decision must be made concerning which response is appropriate for the specific event. The risk responses can be considered in terms of elimination, control at source, minimization, and the use of appropriate personal protective equipment [54].

2.3.1. Personal Protective Equipment (PPE)

Death and injury always happen at the construction site due to failure to wear the PPE provided and ineffective usage of PPE. To mitigate the safety risk in the Malaysian construction industry, wearing PPE while working on the construction site is necessary. PPE serves to keep workers safe in the workplace by shielding them from possible dangers [55]. There are many types of PPE, such as safety helmets, ear protection, high visibility clothing, safety footwear, safety harnesses, etc. [56]. A severe accident can be avoided if the construction labourer is wearing PPE.

Aside from that, employers must consider the physical dimensions of individual employees, such as their body size and gender, while preparing PPE for them. PPE must be adjustable so that when problems emerge, the advice provided must take into consideration any medical conditions. The method, instruction, and training of PPE must be supplied by the employer to all personnel on the construction site in order to prevent accidents [45].

Ref. [57] stated that one of the fundamental steps or mandatory requirements that the construction company must provide for employees before beginning work is teaching them to use PPE at the construction site. Furthermore, training is provided to employees to ensure that they are well-equipped with the knowledge to carry out work on the construction site with minimum safety hazards [58]. Training would be effective if there were two main methods: informational-based training and a hands-on approach in which workers would have to try the PPE on their own in order for the workers to gain a better understanding and awareness of the PPE [59]. For example, the construction company would have to prepare a test or observe the use of the PPE at the construction site for a period of time before the workers are qualified in having full awareness of all of the aspects that are present in the PPE at the construction site [60].

Ref. [60] mentioned that PPE awareness involves choosing the appropriate and relevant PPE suitable of minimising the safety hazards that are threatening the employees’ health and safety. Safety masks, safety gloves, and protective gear must be provided for construction workers engaged in jobs such as welding in order to protect them from splatters of molten metal, as well as any other particles that may come into contact with their skin [61].

Maintenance and supervision of the PPE is also critical at construction sites. As a result, PPE must be of high quality and perform consistently in order to minimise the risks that construction workers face on-site [62]. Workers and their supervisors must continually inspect their PPE to verify that it is functioning properly. In order to keep the PPE in excellent working order and ready for use by the various site employees, workers must be aware of the various procedures for checking and maintaining the PPE.

2.3.2. Safety and Health Training

Safety and health training is essential in the construction industry’s safety management practices, which are commonly acknowledged as standard performance. Safety and health training in the construction site usually include the safety measures training, machinery operator training, working at height training, and the others. Aside from that, safety and health training are essential for occupational health and safety programmes in order to improve the attitudes, abilities, and knowledge of new construction employees and spot accidents on the construction site [63]. Ref. [64] found safety and health training is one of the four interrelated dimensions in a safety programme, in addition to management commitment and employee involvement, worksite analysis, hazard prevention and control systems.

One of the current challenges in the Malaysian construction industry is the lack of knowledge and skills of foreign workers, since most of these foreign workers originate from various countries with poor skilled labour and a lack of training. When foreign migrant workers came to Malaysia, they did not attend the safety and health training provided by the relevant government agencies, which led to an increase in accidents on the construction site [65]. There is thus a need for foreign migrant workers to attend safety training package in order to address the higher accident rates than the local skilled workers [66].

The majority of foreign workers lack knowledge and awareness since they did not attend safety training. It is critical that the content of training uses more illustrations to explain it in order to increase worker safety awareness [67]. Workers will be able to understand and know how to manage the machine more effectively if the training techniques use animation to display and explain the processes of operating machines [68].

2.3.3. Safety Meeting

A safety meeting is one of the ways that will be used to offer an opportunity for all parties participating in the construction team to introduce and discuss the precautionary safety concerns linked to safety and health on the construction site. Before beginning work, a safety meeting must be held to ensure that all personnel are on the same page and may review the previously provided information [69]. A safety meeting is an important aspect of developing a workers’ safety culture in order to reduce accidents on the construction site [45].

Before beginning a new project, kick-off meetings should be held to discuss the risks and hazards, how to select and utilise personal protective equipment (PPE), safety precautions, and safe work procedures that will be implemented at each stage of construction [70].

2.3.4. Proper Equipment

The construction company is responsible for supplying employees with suitable equipment and a safe working environment in order to properly implement the construction site safety culture among workers [40]. Poorly maintained equipment and machinery may result in significant injuries and fatalities. It is critical to offer suitable equipment and machinery that is in excellent working order. Machines must be serviced on a regular basis to guarantee proper operation. Even just a tiny piece of the tools also need to be handled well when carrying out the job in a construction site as it may extremely reduce the opportunity for injury or the fatality of a construction worker. Scaffolding, for example, must be built in the proper manner to provide construction workers safe access to the other level of the structure. As a result, the employer must provide enough equipment at all times while complying with OSHA’s safety regulations [28].

2.3.5. Promote Effective Communication

Promoting effective communication on-site by all construction parties is needed to prevent accidents from happening. In order to avoid workplace accidents, workers, supervisors, managers, contractors, and everyone on-site should be encouraged to communicate with each other and with the employer [71]. Good and concise communication emphasising safety issues shall be practiced among everyone in the construction site so that any misfortune may be avoided [72]. Ref. [73] found that it is important to promote safety communication among construction workers because this will encourage workers to participate actively in providing and receiving safety information.

Poor and ineffective communication can be due to many factors. Ref. [74] identified 33 factors which are responsible for poor communication in the construction industry. Of the 30 factors identified, [51] has categorized them under four dimensions, namely, organizational and management factors, behavioural and cultural factors, project information factors, and technology and method factors. A high accident rate has been found to be one of the impacts of poor and ineffective communication by [50,74].

3. Methodology

This study used the quantitative research approach for data collection and analyses.

3.1. Research Design

In this study, partial least squares structural equation modelling (PLS-SEM) using SmartPLS 4 software [75] was employed as the multivariate analysis technique to explore the moderating effect of risk level on mitigation measures taken. Hence, a survey questionnaire is suitable for data collection as long as the measurement scales are equidistant. The basic conceptual model used for this study is shown in Figure 1: the safety risks construct is conceptualized as a second order hierarchical latent construct consisting of five categories of factors causing safety risks, and the mitigation measures construct is conceptualized as a second order hierarchical latent construct also consisting of five categories of mitigation measures.

Sampling

The respondents, purposively selected for this study, comprised personnel working in the construction industry from the Klang Valley, Malaysia. Three hundred (300) copies of questionnaires prepared in Google Forms were distributed through emails and WhatsApp messenger to the respondents from June 2022 to August 2022. A total of 83 completed questionnaires were received with no missing data, giving a response rate of nearly 28%.

3.2. Research Instrument

The questionnaire consists of four main sections with closed-ended questions as explained below.

3.2.1. Demographic Information

This section is designed to collect the demographic information of respondents such as education level, current practice and the total number of years of working experience, types of projects involved in, and familiarity with management of safety risks.

3.2.2. Likelihood and Severity of Commonly-Encountered Accidents

This section consists of a list of nine commonly-occurred fatal construction accidents listed in Table 2. The respondents were requested to rate the likelihood and severity of these safety risks based on their opinions and experiences according to 5-point rating scales as shown in Table 3.

Table 3. Rating scales for likelihood and severity of commonly-occurred accidents.

Likelihood of Commonly-Occurred Accidents			Severity of Commonly-Occurred Accidents
Rating	Likelihood	Definition	Rating	Severity	Definition
1	Inconceivable	Has never occurred	1	Negligible	First aid, minor abrasions, cuts
2	Remote	Has not been known to occur after many years	2	Minor	Outpatient, medical leave not more than 4 days
3	Conceivable	Might occur sometimes in future	3	Serious	Hospitalized, medical leave 5 days or more
4	Possible	Chances to occur and not unusual	4	Major	Permanent disability, single fatality
5	Most likely	Happen extremely	5	Catastrophic	Numerous fatalities

3.2.3. Factors Influencing Safety Risks

This section contains 18 questions grouped under five categories of factors influencing safety risks. The respondents were requested to rate these factors measured on a 5-point Likert scale from ‘1 = strongly disagree’, ‘2 = disagree’, ‘3 = neutral’, ‘4 = agree’ to ‘5 = strongly agree’ based on their opinion and experiences.

3.2.4. Practical Measures Taken

This section contains 20 questions grouped under five categories of practical measures taken to mitigate safety risks. The respondents were requested to rate the importance of these practical measures measured on a 5-point Likert scale from ‘1 = not important at all’, ‘2 = slightly important’, ‘3 = moderately important’, ‘4 = important’ to ‘5 = very important’ based on their opinion and experiences.

4. Results

4.1. Descriptive Analysis

Table 4 presents the demographic information of the 83 respondents who participated in the questionnaire survey. Out of the 83 questionnaires received, 62 of the respondents are working in consultancy firms, while 21 respondents are from contractor companies. Of the 83 respondents, 62 of them have more than 2 years of working experience. In terms of educational background, 72 of them have at least a bachelor’s degree and above. Of the 83 respondents who participated in the questionnaire survey, 77 of them indicated they are familiar with safety risks. In terms of projects involved, 36 respondents mentioned they are involved with main building works, whereas 47 of the respondents mentioned they are involved in infrastructure works, including highway and railway projects.

Table 4. Demographic information of respondents.

Item	Response Category	Frequency	Percentage (%)	Total Percentage (%)
Current practice	Consultant	62	74.7	100
	Contractor	21	25.3
Types of projects involved	Highway project	17	20.5	100
	Infrastructure works	20	24.1
	Main building works	36	43.4
	Railway projects	10	12.0
Educational level	SPM	1	1.2	100
	Diploma	10	12.0
	Bachelor’s Degree	60	72.3
	Master’s Degree	10	12.0
	PhD	2	2.4
Working experience in the construction industry	2 years or less	21	25.3	100
	3–6 years	24	28.9
	7–10 years	24	28.9
	11–14 years	9	10.8
	15 years and above	5	6.0
Familiarity with safety risks	Yes	77	92.8	100
	No	6	7.2

4.1.1. Likelihood and Severity of Commonly-Occurred Fatal Construction Accidents

Table 5 displays the results for the likelihood of fatal construction accidents commonly happening in the Malaysian construction industry, with fall-related accidents having the highest mean value of 3.96, and insect pest as a safety risk having the lowest mean value of 3.18. The overall mean value is 3.57. In Table 5, the indicators for this construct have skewness values ranging from −0.108 to −1.072, and kurtosis values ranging from −1.322 to 0.292, showing that these indicators do not depart from the normality requirements according to Brown (cited in [76]).

Table 5. Likelihood of commonly-occurred fatal construction accidents (n = 83).

Construct	Indicator	Commonly-Occurred Accidents	Mean	Standard Deviation	Skewness	Kurtosis	Overall Mean
Likelihood	Likelihood1	Fall-related accident (human falling from height)	3.96	1.163	−1.072	0.292	3.57
	Likelihood2	Struck-by accident (struck by falling object, moving vehicle, rolling machinery)	3.54	0.979	−0.640	−0.179
	Likelihood3	Drowning and Asphyxiation (insufficient oxygen)	3.31	1.352	−0.108	−1.322
	Likelihood4	Buried (being buried under the landslide)	3.51	1.108	−0.511	−0.483
	Likelihood5	Electrocution (getting burn, electrocution, shock, arc flash)	3.71	1.099	−0.866	0.099
	Likelihood6	Road accident (hydroplaning, brake failure)	3.73	1.149	−0.746	−0.247
	Likelihood7	Caught in-between accidents (caught, crushed, squeezed between two or more objects on site)	3.51	1.173	−0.526	−0.637
	Likelihood8	Fire or explosion (fire outbreak, bomb explosion)	3.64	1.143	−0.652	−0.409
	Likelihood9	Insect pest (for example: stung by hornets)	3.18	1.354	−0.368	−1.159

Table 6 displays the results for the severity of the same fatal construction accidents commonly happening in the Malaysian construction industry, with fall-related accidents as the highest mean value of 4.18, and insect pest as a safety risk as the lowest mean value of 3.19. The overall mean value is 3.73. In Table 6, the indicators for this construct have skewness values ranging from −1.532 to −0.362, and kurtosis values ranging from −1.140 to 1.722, showing that these indicators, too, do not depart from the normality requirements according to Brown (cited in [76]).

Table 6. Severity of commonly-occurred fatal construction accidents (n = 83).

Construct	Indicator	Commonly-Occurred Accidents	Mean	Standard Deviation	Skewness	Kurtosis	Overall Mean
Severity	Severity1	Fall-related accident (human falling from height)	4.18	1.106	−1.532	1.722	3.73
	Severity2	Struck-by accident (struck by falling object, moving vehicle, rolling machinery)	3.72	0.860	−0.842	0.706
	Severity3	Drowning and Asphyxiation (insufficient of oxygen)	3.57	1.139	−0.446	−0.561
	Severity4	Buried (being buried under the landslide)	3.76	1.043	−0.886	0.583
	Severity5	Electrocution (getting burn, electrocution, shock, arc flash)	3.87	0.985	−0.748	−0.013
	Severity6	Road accident (hydroplaning, brake failure)	3.61	1.188	−0.817	−0.118
	Severity7	Caught in-between accidents (caught, crushed, squeezed between two or more objects on site)	3.83	1.069	−0.944	0.326
	Severity8	Fire or explosion (fire outbreak, bomb explosion)	3.82	1.038	−0.835	0.369
	Severity9	Insect pest (for example: stung by hornets)	3.19	1.339	−0.362	−1.140

4.1.2. Factors Influencing Safety Risks

Table 7 displays the results for the 18 indicators operationalizing the five categories of factors influencing safety risks, with unsafe act and site condition having the highest overall mean value of 4.31, and human error having the lowest overall mean value of 3.97. In Table 7, the indicators for this construct have skewness values ranging from −2.017 to 0.159, and kurtosis values ranging from −1.116 to 6.395, showing that these indicators do not depart from the normality requirements according to Brown (cited in [76]).

Table 7. Factors influencing safety risks (n = 83).

Construct	Indicator	Factors Influencing Safety Risks	Mean	Standard Deviation	Skewness	Kurtosis	Overall Mean
Human Error (HE)	HE1	Improper attitude	3.93	0.921	−0.238	−1.080	3.97
	HE2	Inadequate tools used	3.75	0.660	−0.199	0.097
	HE3	Excessive physical exertion	4.18	0.587	−0.421	1.551
	HE4	Lacks of experience	4.00	0.812	−0.140	−1.116
Failure to use PPE (FPPE)	FPPE1	Failure to use safety helmets	4.52	0.722	−1.560	2.276	4.25
	FPPE2	Failure to use face protection	4.07	0.640	−0.921	2.723
	FPPE3	Failure to use safety boots	4.23	0.831	−1.238	2.139
	FPPE4	Failure to use eye protection	4.17	0.730	−0.852	1.166
Unsafe act and site condition (UA)	UA1	Unsafe equipment	4.47	0.721	−1.190	0.746	4.31
	UA2	Unsafe methods	4.16	0.529	0.159	0.298
	UA3	Hazardous environment	4.25	0.622	−0.229	−0.570
	UA4	Improper work procedure	4.36	0.691	−0.847	0.469
Lack of progressive training (LackT)	LackT1	Employer fail to offer sufficient training	4.34	0.928	−1.481	1.821	4.13
	LackT2	Lack of budget allocation on safety management	3.93	0.729	−2.017	6.395
	LackT3	Lack of workforce due to subcontract work	4.12	0.929	−1.273	1.908
Poor Communication (PC)	PC1	Language barrier	4.22	1.048	−1.557	2.244	4.14
	PC2	Miscommunication and misunderstanding	4.02	0.796	−1.084	2.249
	PC3	Failure in conveying message	4.17	0.895	−1.284	2.260

Note: 1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree.

4.1.3. Mitigation Measures Taken

Table 8 displays the results for the 20 indicators operationalizing the five categories of mitigation measures, with proper equipment having the highest overall mean value of 4.47, and safety meeting having the lowest overall mean value of 4.30. In Table 8, the indicators for this construct have skewness values ranging from −3.752 to 1.025, and kurtosis values ranging from −0.993 to 19.0722, showing that only one indicator, that is PPE1, departs from the normality requirements according to Brown (cited in [76]).

Table 8. Practical measures taken to mitigate safety risks (n = 83).

Construct	Indicator	Practical Measures to Mitigate Safety Risks	Mean	Standard Deviation	Skewness	Kurtosis	Overall Mean
Personal Protective Equipment (PPE)	PPE1	Safety helmets	4.76	0.597	−3.752	19.072	4.45
	PPE2	Ear protection	4.16	0.689	−0.672	0.970
	PPE3	High visibility clothing	4.39	0.641	−0.841	1.047
	PPE4	Safety footwear	4.41	0.716	−1.612	4.911
	PPE5	Safety harnesses	4.53	0.591	−1.207	2.406
	PPE6	Training of PPE	4.45	0.590	−0.517	−0.632
Safety and health training (ST)	ST1	Safety measures training	4.55	0.737	−1.317	0.162	4.35
	ST2	Machinery operator training	4.20	0.435	1.025	0.337
	ST3	Working at height training	4.30	0.745	−0.555	−0.993
Safety Meeting (SM)	SM1	Discuss the precautionary safety concerns	4.48	0.755	−1.245	0.565	4.30
	SM2	Communication between job groups	4.20	0.488	0.462	0.227
	SM3	Report changes at the work site	4.36	0.531	0.081	−0.969
	SM4	Update the existing safety plan and procedure	4.17	0.640	−0.163	−0.579
Proper equipment (PE)	PE1	Supplying employees with suitable equipment	4.65	0.572	−1.419	1.081	4.47
	PE2	Safe working environment	4.30	0.535	0.124	−0.598
	PE3	Machines serviced regularly	4.46	0.591	−0.925	1.881
	PE4	Scaffolding with safe access	4.45	0.590	−0.517	−0.632
Promote Effective Communication (EC)	EC1	Employee pay attention for safety briefing	4.36	0.790	−0.897	−0.293	4.33
	EC2	Construction parties communicate with each other	4.14	0.665	−0.169	−0.717
	EC3	Rapid communication such as walkie-talkies	4.48	0.549	−0.380	−0.980

Note: 1 = not important at all, 2 = slightly important, 3 = moderately important, 4 = important and 5 = very important.

4.2. Structural Equation Modeling

The Mann–Whitney U tests carried out earlier showed that there were no significant differences between the two subgroups, namely 62 respondents from consultant practices and 21 respondents from contractor companies for all the indicators of the twelve constructs. The raw data for these two subgroups were then combined to test the conceptual model shown in Figure 1. SmartPLS 4 software [75] was employed for partial least squares structural equation modelling (PLS-SEM) purposes. The 2-step procedure recommended by [77] was adopted for assessments of the measurement models and structural model.

4.2.1. Assessment of Measurement Models

The following are the quality criteria adopted for assessment of the measurement models in Figure 1:

• Internal consistency reliability: A construct with high Cronbach’s alpha value indicates the indicators have similar range and meaning [78];
• Composite reliability (CR): Values greater than 0.60 are acceptable in exploratory study [79];
• Indicator reliability: Loading values equal to and greater than 0.4 are acceptable if the sum of loadings results in higher loading scores, contributing to AVE scores of greater than 0.5 [80];
• Convergent validity: In order to achieve adequate convergent validity, each construct should account for at least 50% of the average variance explained (AVE ≥ 0.50) [81,82,83];
• Rho_A: The reliability of rho_A usually lies between Cronbach’s alpha and composite reliability [84];
• Discriminant validity: The square root of AVE of a construct should be larger than the correlations between the construct and other constructs in the model [82]. According to [85], HTMT_.90 value of 0.90 indicates that there is a problem of discriminant validity. Using cross loadings to assess discriminant validity, each indicator should load high on its own construct but low on other constructs. Cross loadings of <0.1 should be deleted [86].

Figure 1 is a higher component model (HCM) with two second-order hierarchical latent constructs. Hence, two-stage HCM analysis, a combination of repeated indicators approach and the use of latent variable scores is needed [87]. First stage analysis shows the following two indicators have to be removed for the lower order component models to achieve quality criteria, namely:

(a)

Indicator PPE1 which has cross loadings of <0.10 needs to be removed for the PPE construct to achieve an AVE value > 0.50;

(b)

Indicator HE3 which has cross loadings < 0.10 needs to be removed even though the AVE value for the HE construct > 0.50.

In the second stage analysis, the latent variable scores from the first stage serve as the manifest variables in the higher order component models. The AVE values for safety risks construct and mitigation measures construct together with the outer loadings, path coefficient, and p values are displayed in Figure 2.

Figure 2. Higher order components with AVE values, path coefficients, and p values (n = 83).

The moderating effect of likelihood and severity is investigated with risk level as a second order hierarchical construct as shown in Figure 3, which gives the graphical output from first stage analysis. The AVE values of lower order constructs for safety risks, mitigation measures and risk level together with the outer loadings, path coefficient and p values are presented together.

Figure 3. AVE values, path coefficients, and p values from first stage analysis (n = 83).

The assessment results for lower order components from first stage analysis are summarised in Table 9. Based on the quality criteria given earlier, the lower order components achieve convergent validity with AVE > 0.50 for all the twelve constructs, indicator reliability with outer loadings ranging from 0.473 to 0.915, as well as construct reliability and validity with Cronbach’s alpha, Rho A and Rho C values well above 0.700.

Table 9. Assessment results for lower order components from first stage analysis (n = 83; CI = 95%).

Lower Order Component	Indicator	Outer Loadings	Construct Reliability and Validity
			Cronbach’s Alpha	Composite Reliability, Rho_A	Composite Reliability, Rho_C	AVE (≥0.50)
Likelihood of commonly-occurred accidents	Likehood1	0.776	0.919	0.923	0.933	0.610
	Likehood2	0.709
	Likehood3	0.753
	Likehood4	0.807
	Likehood5	0.851
	Likehood6	0.793
	Likehood7	0.850
	Likehood8	0.806
	Likehood9	0.663
Severity of commonly-occurred accidents	Severity1	0.758	0.907	0.915	0.925	0.582
	Severity2	0.737
	Severity3	0.730
	Severity4	0.823
	Severity5	0.791
	Severity6	0.803
	Severity7	0.855
	Severity8	0.829
	Severity9	0.473
Human Error (HE)	HE1	0.786	0.763	0.774	0.862	0.676
	HE2	0.810
	HE4	0.869
Failure to use Personal Protective Equipment (FPPE)	FPPE1	0.820	0.830	0.840	0.886	0.660
	FPPE2	0.790
	FPPE3	0.833
	FPPE4	0.806
Unsafe act and site condition (UA)	UA1	0.750	0.741	0.753	0.837	0.563
	UA2	0.698
	UA3	0.714
	UA4	0.831
Lack of progressive training (LackT)	LackT1	0.915	0.827	0.834	0.897	0.745
	LackT2	0.808
	LackT3	0.862
Poor Communication (PC)	PC1	0.911	0.880	0.880	0.926	0.807
	PC2	0.886
	PC3	0.897
Personal Protective Equipment (PPE)	PPE2	0.679	0.771	0.782	0.846	0.526
	PPE3	0.839
	PPE4	0.765
	PPE5	0.666
	PPE6	0.663
Safety and health training (ST)	ST1	0.806	0.711	0.734	0.841	0.641
	ST2	0.680
	ST3	0.900
Safety Meeting (SM)	SM1	0.781	0.726	0.726	0.830	0.551
	SM2	0.663
	SM3	0.747
	SM4	0.772
Proper equipment (PE)	PE1	0.741	0.735	0.740	0.835	0.559
	PE2	0.672
	PE3	0.788
	PE4	0.784
Promote effective communication (EC)	EC1	0.897	0.738	0.747	0.852	0.659
	EC2	0.727
	EC3	0.803

Table 10 shows the lower order components do not have any problem with discriminant validity because there is no HTMT_.90 value which is more than 0.90. In addition, Table 11 shows the lower order components achieve satisfactory discriminant validity too because the square root of AVE (along the diagonal) is larger than the correlation (off diagonal) for all the lower order components.

Table 10. Discriminant validity for lower order components using HTMT ratio of correlation.

Indicator	EC	FPPE	HE	LackT	Likelihood	PC	PE	PPE	SM	ST	Severity	UA
EC
FPPE	0.304
HE	0.168	0.486
LackT	0.485	0.517	0.440
Likelihood	0.217	0.399	0.698	0.351
PC	0.553	0.617	0.493	0.837	0.256
PE	0.659	0.357	0.309	0.258	0.313	0.454
PPE	0.783	0.286	0.335	0.406	0.350	0.494	0.865
SM	0.650	0.241	0.329	0.285	0.438	0.456	0.758	0.531
ST	0.451	0.406	0.436	0.506	0.330	0.704	0.751	0.507	0.684
Severity	0.221	0.391	0.476	0.442	0.700	0.483	0.322	0.441	0.413	0.376
UA	0.533	0.718	0.477	0.718	0.459	0.790	0.624	0.552	0.443	0.689	0.474

Table 11. Discriminant validity for lower order components using Fornell and Larcker criterion.

Indicator	EC	FPPE	HE	LackT	Likelihood	PC	PE	PPE	SM	ST	Severity	UA
EC	0.812
FPPE	0.247	0.812
HE	0.139	0.405	0.822
LackT	0.382	0.442	0.367	0.863
Likelihood	0.179	0.348	0.598	0.305	0.781
PC	0.450	0.541	0.416	0.716	0.229	0.898
PE	0.491	0.292	0.239	0.172	0.248	0.366	0.748
PPE	0.597	0.217	0.252	0.312	0.272	0.408	0.652	0.726
SM	0.477	0.136	0.252	0.221	0.355	0.365	0.561	0.406	0.742
ST	0.336	0.327	0.324	0.393	0.228	0.559	0.542	0.377	0.479	0.800
Severity	0.183	0.344	0.411	0.382	0.638	0.431	0.253	0.365	0.335	0.278	0.763
UA	0.399	0.577	0.376	0.574	0.355	0.648	0.452	0.406	0.317	0.504	0.381	0.750

In the second stage analysis for the moderating effect of risk level, the latent variable (LV) scores for safety risks, mitigation measures, and risk level from the first stage analysis serve as the manifest variables in the higher order components. The graphical output for second stage analysis is shown in Figure 4, where the AVE values for safety risks, mitigation measures, and risk level together with the outer loadings, path coefficient, and p values are presented.

Figure 4. AVE values, path coefficients and p values from second stage analysis (n = 83).

The assessment results for higher order components from second stage analysis are summarised in Table 12. Based on the quality criteria given earlier, the higher order components achieve convergent validity with AVE > 0.50, indicator reliability with outer loadings ranging from 0.731 to 0.916, as well as construct reliability and validity with Cronbach’s alpha, Rho A and Rho C values well above 0.700.

Table 12. Assessment results of higher order components (n = 83; CI = 95%).

Higher Order Component	Latent Variable Scores	Outer Loadings	R-Squared, R2	Construct Reliability and Validity
				Cronbach’s Alpha	Composite Reliability, Rho_A	Composite Reliability, Rho_C	AVE (≥0.50)
Risk Level	LV Scores—Likelihood	0.894	---	0.779	0.786	0.900	0.819
	LV Scores—Severity	0.916
Safety Risks	LV Scores—HE	0.606	---	0.837	0.882	0.885	0.609
	LV Scores—FPPE	0.731
	LV Scores—UA	0.846
	LV Scores—LackT	0.803
	LV Scores—PC	0.884
Mitigation Measures	LV Scores—PPE	0.782	0.368	0.829	0.834	0.879	0.593
	LV Scores—ST	0.745
	LV Scores—SM	0.741
	LV Scores—PE	0.837
	LV Scores—EC	0.741

Table 13 shows the higher order components have no problem with discriminant validity because there is no HTMT_.90 value which is more than 0.90. In addition, Table 14 shows the higher order components achieve satisfactory discriminant validity as well because the square root of AVE (along the diagonal) is larger than the correlation (off diagonal) for all the constructs.

Table 13. Discriminant validity of higher order components using HTMT ratio.

Higher Order Component	Mitigation Measures	Risk Level	Safety Risks	Risk Level × Safety Risks
Mitigation measures
Risk level	0.481
Safety risks	0.652	0.665
Risk level × Safety risks	0.165	0.114	0.470

Table 14. Discriminant validity of higher order components using Fornell and Larcker criterion.

Higher Order Component	Mitigation Measures	Risk Level	Safety Risks
Mitigation measures	0.770
Risk level	0.386	0.905
Safety risks	0.577	0.516	0.780

4.2.2. Assessment of Structural Model

The following are the criteria adopted to assess the higher component model shown in Figure 4:

Standardised root mean square residual (SRMR): A value less than 0.10 is considered a good fit [88].

Normed fit index (NFI): A value above 0.9 usually represents acceptable fit [89].

The assessment results for model fit of higher component model are summarised in Table 15, showing the higher component model has a good fit with SRMR = 0.094. However, the NFI value < 0.90.

Table 15. Assessment results of structural model (n = 83; CI = 95%).

Item	Saturated Model				Estimated Model
	Original Sample	Sample Mean	95%	99%	Estimated Sample	Sample Mean	95%	99%
SRMR (≤0.10)	0.094	0.071	0.088	0.099	0.095	0.073	0.093	0.104
d_ULS	0.696	0.403	0.604	0.760	0.700	0.429	0.681	0.840
d_G	0.328	0.258	0.380	0.452	0.328	0.328	0.385	0.458
Chi-square	153.132				153.258
NFI (≥0.90)	0.698				0.697

4.3. Moderating Effect of Risk Level

Table 16 summarises the results obtained in second stage analyses for Figure 2 and Figure 4. Without risk level as a moderator in Figure 2, the main effect between safety risks and mitigation measures is β = 0.580 (p < 0.001) as shown in Figure 2, with R² = 0.337 (p < 0.001) and effect size = 0.508 (p < 0.05). With risk level as a moderator in Figure 4, the simple effect is β = 0.611 (p < 0.001), with R² = 0.368 (p < 0.001) and effect size = 0.339 (p > 0.05). The path coefficient β increased to 0.611 from 0.580, which is an increase of 5.3% and the R² value increased to 0.368 from 0.337, which is an increase of 9.2%. The strength of the relationship between safety risks and mitigation measures increases when the risk level increases in the presence of risk level as a moderator. This is illustrated by the interaction plot shown in Figure 5. The effect size is 0.038 (p > 0.05), which is small according to [90].

Table 16. Moderating effect of risk level on mitigation measures (n =83, CI = 95%).

Case	Safety Risks	Mitigation Measures		Path	Path Coefficient	Total Effect	Effect Size, f-Square
	AVE	R2	AVE
Figure 2 Base model	0.609, p = 0.000	0.337, p = 0.000	0.592, p = 0.000	Safety risks → MM	0.580, p = 0.000.	0.580, p = 0.000.	0.508, p = 0.031.
Figure 4 Risk level as a moderating construct	0.609. p = 0.000	0.368, p = 0.000	0.593, p = 0.000	Safety risks → MM	0.611, p = 0.000.	0.611, p = 0.000.	0.339, p = 0.099
				Risk level → MM	0.090, p = 0.422	0.090, p = 0.422	0.009, p = 0.788
				Risk level × Safety risks → MM	0.170, p = 0.213	0.170, p = 0.213	0.038, p = 0.475

Figure 5. Interaction plot showing the moderating effect of risk level.

5. Discussion

Table 17 displays the results for composite reliability and validity of the initial conceptual model which consists of all the indicators identified for this research which are presented in Table 5, Table 6, Table 7 and Table 8. The results show that the measurement instrument used for data collection has a high internal consistency reliability with all the values well above 0.707. With the deletion of two indicators, namely HE3 and PPE1, the results in Table 9 show that the measurement instrument is further improved with good indicator reliability, adequate convergent validity, and adequate discriminant validity.

Table 17. Composite reliability and validity of lower order components.

Construct	Indicators in Construct		Cronbach’s Alpha	Composite Reliability, Rho_A	Composite Reliability, Rho_C(>0.50 but <0.90)	Average Variance Extracted Values, AVE
	Number	Reference
Likelihood of commonly-occurred accidents	9	Table 5	0.919	0.938	0.931	0.602
Severity of commonly-occurred accidents	9	Table 6	0.907	0.944	0.921	0.572
Human Error (HE)	4	Table 7	0.747	0.751	0.840	0.569
Failure to use Personal Protective Equipment (FPPE)	4		0.830	0.839	0.886	0.660
Unsafe act and site condition (UA)	4		0.741	0.754	0.837	0.563
Lack of progressive training (LackT)	3		0.827	0.833	0.897	0.745
Poor Communication (PC)	3		0.880	0.880	0.926	0.807
Personal Protective Equipment (PPE)	6	Table 8	0.776	0.781	0.843	0.475
Safety and health training (ST)	3		0.711	0.736	0.841	0.642
Safety Meeting (SM)	4		0.726	0.726	0.830	0.551
Proper equipment (PE)	4		0.735	0.741	0.835	0.559
Promote effective communication (EC)	3		0.738	0.750	0.852	0.659

The risk levels for the nine commonly-occurred fatal construction accidents presented in Table 5 and Table 6 were calculated and the values are presented in Table 18. The results show that fall-related accident is at a high risk level, confirming the finding from earlier studies which mentioned falling from height is the number one killer in the Malaysian construction industry [11,12,13,14,15,16]. All the other types of safety risks are in the medium risk levels, with insect pest as a safety risk having the lowest risk level score. Table 18 also summarises the ranking for all the nine commonly-occurred safety risks measured in terms of mean likelihood score, mean severity score and risk level. The results show that fall-related accident remains at the top, signifying that fall-related accidents are highly risky and the most likely and severe risk to happen at the construction site; fatal accidents due to insect pest is at the bottom of the ranking, confirming the data from [11].

Table 18. Risk levels of commonly-occurred fatal construction accidents.

Item	Commonly-Occurred Accidents	Likelihood		Severity		Risk Level (Mean Likelihood × Mean Severity)
		Mean Value	Rank	Mean Value	Rank	Score	Rank	Description
1	Fall-related accident (human falling from height)	3.96	1	4.18	1	16.6	1	High risk (≥15)
5	Electrocution (getting burn, electrocution, shock, arc flash)	3.71	3	3.87	2	14.4	2	4 < Medium risk < 15
7	Caught in-between accidents (caught, crushed, squeezed between two or more objects on site)	3.51	6	3.83	3	13.4	5	4 < Medium risk < 15
8	Fire or explosion (fire outbreak, bomb explosion)	3.64	4	3.82	4	13.9	3	4 < Medium risk < 15
4	Buried (being buried under the landslide)	3.51	6	3.76	5	13.2	6	4 < Medium risk < 15
2	Struck by accident (struck by falling object, moving vehicle, rolling machinery)	3.54	5	3.72	6	13.2	6	4 < Medium risk < 15
6	Road accident (hydroplaning, brake failure)	3.73	2	3.61	7	13.5	4	4 < Medium risk < 15
3	Drowning and Asphyxiation (insufficient of oxygen)	3.31	7	3.57	8	11.8	7	4 < Medium risk < 15
9	Insect pest (for example: stung by hornets)	3.18	8	3.19	9	10.1	8	4 < Medium risk < 15

Table 19 summarises that rankings for all the 18 indicators or factors of the five construction safety risks mitigation measures. The results show that 15 factors have mean scores well above 4, with the top 5 factors being failure to use safety helmets, unsafe equipment, improper work procedure, employers failing to provide sufficient training, and hazardous environment. Only three factors have mean scores slightly lower than 4, and they are improper attitude, lack of budget allocation on safety management, and inadequate tools used. In terms of overall mean value, the ranking for the five construction safety risks in descending order are: unsafe act and site condition, failure to use PPE, poor communication, lack of progressive training, and human error.

Table 19. Ranking of factors influencing safety risks.

Construct	Indicator	Factors Influencing Safety Risks	Mean		Overall Mean
			Value	Rank	Value	Rank
Human Error (HE)	HE1	Improper attitude	3.93	3	3.97	5
	HE2	Inadequate tools used	3.75	4
	HE3	Excessive physical exertion	4.18	1
	HE4	Lacks of experience	4.00	2
Failure to use PPE (FPPE)	FPPE1	Failure to use safety helmets	4.52	1	4.25	2
	FPPE2	Failure to use face protection	4.07	4
	FPPE3	Failure to use safety boots	4.23	2
	FPPE4	Failure to use eye protection	4.17	3
Unsafe act and site condition (UA)	UA1	Unsafe equipment	4.47	1	4.31	1
	UA2	Unsafe methods	4.16	4
	UA3	Hazardous environment	4.25	3
	UA4	Improper work procedure	4.36	2
Lack of progressive training (LackT)	LackT1	Employers fail to offer sufficient training	4.34	1	4.13	4
	LackT2	Lack of budget allocation on safety management	3.93	3
	LackT3	Lack of workforce due to subcontract work	4.12	2
Poor Communication (PC)	PC1	Language barrier	4.22	1	4.14	3
	PC2	Miscommunication and misunderstanding	4.02	3
	PC3	Failure in conveying message	4.17	2

Note: 1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree.

Table 20 summarises the rankings for all the 20 indicators or factors of the five construction mitigation measures. The results show that all the 20 factors have mean scores well above 4, with the top 6 factors being the use of safety helmet, supplying employees with suitable equipment, safety measures training, safety harnesses, discussion on the precautionary safety concerns, and rapid communication such as walkie-talkies. All the five construction mitigation measures have overall mean values between 4.30–4.47. The construct on proper equipment has the highest overall mean score, and the measurement indicators correspond to control the risks at the source and to design of safe work systems to minimise risks.

Table 20. Ranking of practical measures taken to mitigate safety risks.

Construct	Indicator	Practical Measures to Mitigate Safety Risks	Mean		Overall Mean
			Value	Rank	Value	Rank
Personal Protective Equipment (PPE)	PPE1	Safety helmets	4.76	1	4.45	2
	PPE2	Ear protection	4.16	6
	PPE3	High visibility clothing	4.39	5
	PPE4	Safety footwear	4.41	4
	PPE5	Safety harnesses	4.53	2
	PPE6	Training of PPE	4.45	3
Safety and health training(ST)	ST1	Safety measures training	4.55	1	4.35	3
	ST2	Machinery operator training	4.20	3
	ST3	Working at height training	4.30	2
Safety meeting (SM)	SM1	Discuss the precautionary safety concerns	4.48	1	4.30	5
	SM2	Communication between job groups	4.20	3
	SM3	Report changes at the work site	4.36	2
	SM4	Update the existing safety plan and procedure	4.17	4
Proper equipment (PE)	PE1	Supplying employees with suitable equipment	4.65	1	4.47	1
	PE2	Safe working environment	4.30	4
	PE3	Machines serviced regularly	4.46	2
	PE4	Scaffolding with safe access	4.45	3
Promote effective communication (EC)	EC1	Employee pay attention for safety briefing	4.36	3	4.33	4
	EC2	Construction parties communicate with each other	4.14	3
	EC3	Rapid communication such as walkie-talkies	4.48	1

Note: 1 = not important at all, 2 = slightly important, 3 = moderately important, 4 = important and 5 = very important.

The goodness of fit (GoF) for the path model can be determined manually by using the formula GoF = [(mean R²) × (mean AVE)]^1/2 [91]. Based on the R² value of 0.368 and mean AVE value of 0.674 for risk level, safety risks and mitigation measures constructs in Table 12, the GoF for the path model is found to be (0.368 × 0.674)^1/2 = 0.497, which is greater than 0.36 for large fit [92]. It can be concluded that the GoF for the model shown in Figure 4 is large for global PLS model validity.

6. Conclusions

This study investigated the moderating effect of risk level on mitigation measures implemented due to the numerous factors causing safety risks. In Table 16, the results for Figure 2 show that safety risks have a significant positive relationship with mitigation measures with β = 0.580, and p < 0.001. The effect size is large with f² = 0.508, and p < 0.05.

The following conclusions can be made from the results for Figure 4 in Table 16:

Safety risks has significant positive relationship with mitigation measures with β = 0.611, p < 0.001. The effect size is medium with f² = 0.339, p > 0.05.

Risk level has a positive but insignificant relationship with mitigation measures with β = 0.090, p > 0.05. The effect size is negligible with f² = 0.009, p > 0.05.

The interaction term, risk level × safety risks has a positive but insignificant relationship with mitigation measures with β = 0.170, p > 0.05. The effect size is small with f² = 0.038, p > 0.05.

The results for Figure 4 show that the relationship between safety risks and mitigation measures increases in the presence of risk level as a moderator with path coefficient β = 0.611, and p < 0.001. The interaction plot in Figure 5 actually illustrates that the relationship between safety risks and mitigation measures is further heightened in the presence of risk level as a moderator. Because of the positive moderating effect (β = 0.170), the relationship between safety risks and mitigation measures becomes stronger with higher levels of risk level. Even though the effect size of the interaction term (f² = 0.038) is small, under severe situations such as incidents that are categorized as ‘Acts of God’, the sudden surge in risk level would result in an immediate change in β value. Therefore, it is imperative to consider these extreme situations in the planning, design, and management of construction safety because the consequential impacts of these sudden and unexpected incidents could be disastrous, thereby disrupting the continuity of construction works.

It is important to note that uncertainty and severity are intrinsic/inherent properties of safety risks. Mitigation measures are put into place to eliminate the likelihood of safety risks from happening, and to reduce the severity and impacts of these safety risks when they actually happen, which could lead to the loss of lives and hence emotional sufferings, damage to property, disruptions to on-going works, stop-work orders, liquidated ascertained damages and litigation cases. The mitigation measures implemented should always consider the moderating effect of risk level of safety risks which may cause the practical measures implemented on construction sites to be inadequate. The effect of risk level is higher when either the likelihood or severity, or both, are higher. Therefore, mitigation measures taken by the contractors must always take into account the types of factors causing safety risks, as well as the uncertainty or likelihood and severity of these factors for the sustainability of development projects. According to [7], the likelihood of incidents and their severity could be reduced by conducting effective pre-job safety analyses.

The findings from this study have practical values in view of Section 15 in the Occupational Safety and Health Act 1994, which states “it shall be the duty of every employer and every self-employed person to ensure, so far as is practicable, the safety, health and welfare at work of all his employees”. The term ‘practicable’ should consider the following aspects, namely: “(a) the severity of the hazard or risk in question, (b) the state of knowledge about the hazard or risk and any way of removing or mitigating the hazard or risk, (c) the availability and suitability of ways to remove or mitigate the hazard or risk, and (d) the cost of removing or mitigating the hazard or risk” [93]. In this study, numerous factors which influence or cause safety risks were identified and presented in Table 7; some of the practical measures which can be implemented to mitigate or manage these safety risks in order to reduce their impacts were presented in Table 8. The author of [94] asserted that all the factors that influence safety on construction projects should be identified and categorized in order to prepare a construction accident causation framework which maps out these factors in terms of originating influences, shaping factors and immediate factors so that a comprehensive plan for training, awareness and monitoring can be prepared. The mitigation measures implemented should be able to manage or mitigate the impacts from accidents which are categorized as high-risks. This study also has academic value in applying the PLS-SEM method to analyse the data collected from the Malaysian construction industry. For generalization purposes, further research with larger sample size using the same technique should be replicated to provide additional evidence on the effects of likelihood and severity on mitigation measures taken for safety risks.