Psychological capital was ranked as one of the factors that played a major role in safety performance in the construction industry. However, this factor has to do with individual growth in measuring performance, especially safety compliance (29). About the nature and development of psychological capital, if identified as a personal trait, a personality is also inherited and can be classified as neuroticism, extraversion, openness, agreeableness, or conscientiousness, as an individual's traits are the result of environmental influences (30). It is also expected that when workers possess a higher psychological capital (a personal cognitive resource, one's “positive developmental state”), they are expected to have a higher work-related engagement and enhanced level of personal safety (31). In correspondence to those above, the importance of one's psychological capital to eliminate stress and focus more on engaging behavior is a key personal resource, that can be leveraged (32). According to the (JD-R) Model, psychological capital influences workers' physical and social wellbeing, and work pressure, and excessive workload lead to the decreased safety performance in the construction industry (33). Additionally, individuals who use their psychological capital will attain good safety performance and perform effectively with respect to their work-related engagement (34). See Figure 1, based on the discussion, following hypotheses are formed to be tested. Psychological capital would positively impact Hypothesis 1: Work engagement, Hypothesis 2: Safety compliance, and Hypothesis 3: Safety participation.

Work engagement is an individual's motivating, a productive work-related condition that includes vigor, dedication, and absorption (35). Engaged personnel have greater enthusiasm and affection for their work, and they are usually so engrossed in their work that time rushes by unnoticeably (36, 37). Workers that are engaged in their work have a strong absorption in their work, like challenges, and have a high mental resilience, which allows them to overcome obstacles while still loving their work. According to the findings of the study, the medical staff was able to provide quality treatment to patients, and those patients reported feeling happier after having interactions with engaged professionals. This led to an increase in overall job effectiveness as well as an improvement in the quality of care provided (38). Work involvement also leads to better interpersonal ties among workers, which produces a healthy work environment.

Work engagement and enhanced interpersonal ties are believed to develop a proactive attitude among workers, leading to greater organizational performance. Work engagement has been classified into three subcategories such as trait engagement (positive view of life and work), state engagement (feeling of energy absorption and effectiveness), and behavioral engagement (extra-role behavior) (36). We will use a state and behavioral engagement viewpoint to link work engagement to safety performance because safety performance comprises safety compliance and participation. According to Idris and Dollard (39), the positive outcome of organizational citizenship behavior is the major basis for combining work engagement with safety behavior, which enhances the effectiveness of the organization. In addition, employees who have a higher level of interest with their work are more likely to engage in safe behaviors (40).

Additionally, engaged workers are more likely to participate in safety activities because they have higher levels of self-esteem, self-satisfaction, and well-connect with their job or role through vigor, dedication, and absorption. As a result, it is expected of them to go beyond the normal call of duty for the safety of the workplace and exhibit participatory behavior by encouraging others to participate in safety, giving suggestions and views to improve safety, and volunteering in safety-related programs (41). Furthermore, in the construction industry, some prominent acts by workers, such as highlighting possible hazards and reporting small injuries and unsafe conditions, are also classified as work engagement behavior (42). The higher the work engagement, will enhance job resources, such as autonomy, will modify safety compliance (in-role conduct) and safety participation (extra-role behavior). See hypotheses 4, 5, and 6 in Figure 1, work engagement would positively influence safety compliance, safety participation, and psychological capital in the construction industry.

Safety compliance and accident prevention strategies have been dealt with based on various theoretical approaches, and little consideration has been given to the workforce's potentially positive and proactive role in safety management (43). Traditionally, the worker has been viewed as a passive participant in the dynamics of organizational safety (44). On the other hand, more contemporary models of safety performance and involvement have recognized them as the main characteristic of organization safety performance (45). On the other hand, the term “safety participation” refers to a worker's capacity to participate in and comply with workplace safety activities while simultaneously carrying out their regular jobs. In addition, participating in safety means offering suggestions and receiving feedback, motivating others to learn, behave, and perform in a safe manner, actively learning and participating in safety training, bringing attention to potential safety-related problems, and serving as a steward in one's place of employment (46). Furthermore, it is embedded with the responsibilities and duties of a safety officer to encourage workers to participate and show their level of understanding about safety, which in turn will facilitate safety performance (15). Please refer to Figure 1, in which the role of safety participation and safety compliance for the construction industry has been hypothesized.

Work pressure was discovered to be an important factor influencing safety compliance in the research (47). Workplace stress is frequently mentioned in the context of workplace health issues (48, 49). Safety-related behavior, such as safety compliance, has been investigated using the model (50). According to Nahrgang et al. (51), high work demands and insufficient work resources were negatively associated with safety compliance. In a separate study (52), they discovered a positive relationship between routine violations and work demands (via work stress), as well as a negative relationship between work resources and frequent violations (via commitment). According to research, job resources and safety compliance are linked (53).

In addition, when it comes to estimating the available work resources, the results of work pressure have been inconsistent. In addition, the topic of work pressure has been extensively researched in the field of occupational safety and work-life balance, and its importance in the prevention of accidents cannot be overstated (16, 54). In addition, self-injury reports or workplace accidents and workers' attitudes toward workplace safety may impact safety compliance (55). In conclusion, it is imperative for every sector of the economy that functions in high-risk environments to strike a balance between job pressure and safety precautions in the workplace (56). See Figure 1, in hypotheses 7, 8, and 9, work pressure negatively affects safety compliance, safety participation, and work engagement in the study industry.

This study investigates factors that influence the construction industry's safety performance and explores the interrelationship and predictive performance of safety factors on safety compliance and participation. Data was gathered through a random sample of workers in the Malaysian construction industry, including administrative and field workers in various departments. However, questionnaires were used to collect information such as participants' age, years of experience, educational levels, and other pertinent details essential to the study. Experts in the field of research conducted a validation evaluation of the questionnaires to ensure that they covered or collected all the information needed for the study.

The final questionnaires were delivered to a random sample of 500 workers after all the appropriate assessments, with 28 ongoing construction projects in six provinces of Malaysia. A total of 345 questionnaires were completed and returned to the research team. Participants used a five-point scale to answer all the questions, ranging from strongly disagree to agree. For data analysis, SPSS-21 and SPSS AMOS-21 software were used to evaluate descriptive and reliability statistics, and confirmatory factor analysis (CFA) to examine the Validity of dimensions, including discriminant and convergent validity. Amos-21 was also used for structural equation modeling (SEM) and hypothesis testing.

This study used the measurement established by prior research (57) for the PsyCap measurement. There has been other research that has made use of this questionnaire, most notably in the context of the construction sector as well as some different industry contexts [(29), p. 20, (58–62)]. PsyCap consists of four separate sub-dimensions, which are hope (perseverance to reach the goal and alignment when necessary), self-efficacy (belief in one's abilities and expending efforts to succeed), resilience (tendency to bounce back after being adversely affected), and optimism (looking at the bright side through positive attributes). A total of 24 statements were included in the questionnaire, and six items were allotted to each of the sub-dimensions. An illustration of this would be the phrase, “I am comfortable analyzing a long-term problem in order to discover a solution.” “At work, I normally handle unpleasant situations in one way or another,” When things at work are uncertain for me, I typically expect the best,” “There are many routes that can be taken to avoid any difficulty.”

Prior researchers used a scale to measure work engagement (63). This scale is a condensed form of the work engagement scale, which consists of nine individual components. This one-dimensional scale illustrates three facets of employee engagement: vigor (a higher energy level combined with sufficient mental resilience), commitment (inspiration and excitement toward one's work), and absorption (being completely absorbed in one's work; well-connected and engrossed with work throughout). A 5-point Likert scale was utilized for the work engagement scale that we developed. Some of the comments that were used as questions were as follows: “When I am working, time seems to fly by,” “I take pride in the work that I perform,” and “At my job, I feel powerful and vibrant.”

One of the well-known instruments that Neal and Griffin had devised was used to conduct the safety behavior evaluation (5, 64). The cale consists of total six elements, representing different safety behaviors, that includes three statements for safety compliance (obligatory safety behaviors at the workplace, needed formal observation), and three statement for safety participation (extra-role behavior, not essential to perform, but self-generated behavior in the form of contextual performance). Statements such as “I utilize all of the necessary safety equipment to accomplish my job,” “I employ the correct safety procedures for carrying out my job,” and “I assure the highest levels of safety while I carry out my job” are examples of statements that demonstrate compliance with safety regulations. Remarks such as “I support the safety program within the organization” are examples of safety involvement statements. The phrases “I make an extra effort to improve workplace safety” and “I voluntarily carry out duties or activities that promote workplace safety” refer to the same action: making an extra effort to improve workplace safety.

The level of work pressure was determined using a scale with five components (55). The amount of pressure at work is shown through statements such as, “The primary focus of this organization is on production, everything else is secondary;” “If production demands are not met, I may be demoted or endure other unpleasant consequences relating to my employment.” A five-point Likert scale was used to grade everything on the list.

The respondents in this study were employed in the Malaysian construction industry. Because we do not possess a comprehensive list of the population being investigated, we resorted to the non-probability, convenience, and snowball sampling strategies (65). The data collection was finished after 4 months. Hard copies of the instrument were given to those who worked in the following trades: roofing workers, masons, plumbers, tile and brick installers, ironworkers, electricians, pipefitters, and concrete finishers. Responses were completed whenever workers were available, including during break intervals, meal breaks, and other pauses in the workday. During the process of providing their answers, respondents received assistance. The research team received the questionnaires once they had been filled out and submitted. The survey might be finished in 10–15 min. All ethical considerations were kept in mind while undertaking this study i.e., respondents were ensured of their privacy and anonymity, two of the most important aspects of any survey. Before the study began, respondents were asked to sign a consent form, and their participation in the research was optional. From the demographics point of view, 85% respondents were male, and 15% respondents were female. Participants ranged from 25 to 35 years old, with work experience ranging from 6 to 10 years. The participants' educational levels ranged from a professional degree to High school. See Table 1—demographic information of the participants.

A reliability test was carried out to verify the internal consistency of the questionnaire constructs. Cronbach's alpha must be sufficiently high to achieve acceptable response reliability (66). For all five constructs, Cronbach's alpha values varied from 0.868 to 0.949, showing that our findings are reliable. See Tables 2, 3 representing the results of Confirmatory Factor Analysis (CFA), convergent and discriminant validity. Convergent validity refers to the interpretive power of observable variables over latent variables. It might be evaluated using three common indicators, i.e., standardized factor loadings (SFL), construct reliability (CR), and average variance extracted (AVE). Factor loadings of all constructs were well above the cut-off limit, i.e., 0.70, where the standardized factor loading for psychological capital ranged from 0.70 to 0.756, work engagement from 0.70 to 0.76, work pressure from 0.84 to 0.88, and for safety performance 0.794–0.864, hence exhibiting a strong convergent validity.

Results of the concurrent validity indicators that demonstrated an acceptable convergent validity. See Table 2, to describe the discriminant validity of the constructs, there are common indicators like construct reliability (CR > 0.7), standardized factor loadings (SFL > 0.6), and average variance extracted (AVE > 0.5) that were used/compared (67). For discriminant validity, the square root of the average variance extracted (AVE) value was compared with the correlation coefficient of other variables (67). If the outcome value is greater than its correlation coefficient, then the discriminant validity was achieved, and all our constructs met this criterion. See Table 3.

To assess the validity of the measurement model for all variables, we used the subset-item parceling technique for psychological capital and work engagement, since they were taken as a higher-order construct in this study. Through this technique, the aggregation of individual items of a construct into one or more parcel(s) is performed to exhibit a latent variable (68, 69). Furthermore, the method allows researchers to form parcels by adding or averaging items in the desired parcel(s), e.g., if a scale carries eight items, the practitioner can aggregate all items to form one parcel or even two items to form four parcels. Our result indicated a good fit for measurement models (70).

A few of the indices like CMIN (chi-square X² /degree of freedom), chi-square X², comparative fit index (CFI), root-mean-square error of approximate (RMSEA), normed fit index (NFI), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), Tacker-Lewis index (TLI) were used (71–73). All the values for each index were well under the criteria. Results for five factors measurement model were (p-value = 0.183, RMSEA = 0.018, GFI = 0.958, NFI = 0.972, AGFI = 0.943, CFI = 0.997, TLI = 0.996, Chi-square = 139.123, and CMIN = 1.113, and DF = 125). To summarize the reliability, convergent Validity, and discriminant validity, our study findings support that the internal factor structure of the scales being tested is well-validated and reliable by meeting the convergent and discriminant criterion (74). These data corroborate the instruments' efficacy and predictability in this investigation.

The hypothesis model was created and tested using the SEM method to determine the Goodness-of-Fit; we examined whether the results fit the measurement and a structural model. For structural model fit criteria, the primary goal was to determine whether there were any abnormal variables, where all variances were significant with a value >0, standard errors were well under the limit, and all standardized factor loadings were substantial. Our results exhibit strong empirical evidence for the good primary fit of the data (i.e., p-value = 0.228, RMSEA = 0.009, GFI = 0.914, NFI = 0.922, AGFI = 0.902, CFI = 0.997, TLI = 0.997 and CMIN = 1.031). All the indexes met the criteria, demonstrating an acceptable overall model fit.

To test the hypothesized model, we employed a 95% confidence interval. We accepted or rejected research hypotheses based on the p-value and the nature of effect size (based on presupposed theoretical assumptions). As per the data outcomes, hypothesis−1, psychological capital, work engagement (β = 0.749, p < 0.001) was accepted. Unexpectedly, hypothesis−2, psychological capital, safety compliance (β = −0.096, p > 0.05), and hypothesis 3, psychological capital, safety performance (β = −0.05, p > 0.05), both were rejected based on their insignificant p-value. For hypothesis 4 and hypothesis−5, work engagement, safety compliance (β = 0.342, p < 0.05), and work engagement, safety participation (β = 0.153, p < 0.05), both were accepted. Also, as expected for hypotheses 7 and 8, Work pressure, safety compliance (β = −0.373, p < 0.05), and Work Pressure, safety participation (H6; β = −0.269, p < 0.05) both were accepted, unexpectedly, hypothesis−9 for Work pressure, work engagement (β = 0.052, p > 0.05) was rejected based on its insignificant p-value, regarding the overall variance caused in dependent variables by their independent variables; for work engagement, it was 0.549 (54%), for safety compliance 0.216 (21%), and safety participation 0.122 (12%; through their predictors).

Furthermore, analysis of mediation results indicates that work engagement played a partially mediating role between psychological capital and safety compliance (β = 0.203, p < 0.001), as well as between psychological capital and safety participation (β = 0.142, p < 0.001), hence hypothesis-6 was accepted. Nonetheless, there was an insignificant mediation effect observed between work pressure, work engagement, and safety performance objective indicators; safety compliance (β = 0.014, p > 0.05) and safety participation (β = 0.01, p > 0.05). See Table 4 for further explanation results of the hypotheses and correlation amongst all the variables.